Claudio Nicolini PDF

Chromatin Structure
and Function
Molecular and Cellular
Biophysical Methods
Part A
NATO ADVANCED STUDY INSTITUTES SERIES
A series of edited volumes comprising multifaceted studies of contemporary

scientific issues by some of the best scientific minds in the world, as-
sembled in cooperation with NATO Scientific Mfairs Division.
Series A: Life Sciences
Recent Volumes in this Series
Volume 13 - Prostaglandins and Thromboxanes

edited by F. Berti, B. Samuelsson, and G. P. Velo
Volume 14 - Major Patterns in Vertebrate Evolution

edited by Max K. Hecht, Peter C. Goody, and Bessie M. Hecht
Volume 15 - The Lipoprotein Molecule
edited by Hubert Peeters
Volume 16 - Amino Acids as Chemical Transmitters
edited by Frode Fonnum
Volume 17 - DNA Synthesis: Present and Future
edited by Ian Molineux and Masamichi Kohiyama
Volume 18 - Sensory Ecology: Review and Perspectives
edited by M. A. Ali
Volume 19 - Animal Learning: Survey and Analysis
M. E. Bitterman, V. M. LoLordo, J. B. Overmier, and M. E. Rashotte
Volume 20 - Antiviral Mechanisms in the Control of Neoplasia
edited by P. Chandra
Volume 210 - Chromatin Structure and Function :

Molecular and Cellular Biophysical Methods
edited by Claudio A. Nicolini
Volume 21b - Chromatin Structure and Function:

Levels of Organization and Cell Function
edited by Claudio A. Nicolini
The series is published by an international board of publishers in con-

junction with NATO Scientific Affairs Division
A Life Sciences Plenum Publishing Corporation

B Physics New York and London
C Mathematical and D. Reidel Publishing Company
Physical Sciences Dordrecht and Boston
D Behavioral and Sijthoff International Publishing Company

Social Sciences Leiden
E Applied Sciences Noordhoff International Publishing

Leiden
Chromatin Structure
and Function
Molecular and Cellular
Biophysical Methods
PartA
Edited by
Claudio A. Nicolini
Temple University
Philadelphia, Pennsylvania
SPRINGER SCIENCE+BUSINESS MEDIA, LLC

Library of Congress Cataloging in Publication Data
Nato Advanced Study Institute, Erice, Italy, 1978.
Chromatin structure and function.
(NATO advanced study institutes series: Series A, Life sciences; v. 21)
lncludes bibliographical references and indexes.
CONTENTS: pt. A. Molecular and cellular biophysical methods. - pt. B. Levels
of organization and .cel! function.
l. Chromatin - Congresses. 2. Carcinogenesis- Congresses. 1. Nicolini, Claudio A.
Il. Title. III. Series.
QH599.N37 1978 574.8'732 78-24268
ISBN 978-1-4684-0975-8 ISBN 978-1-4684-0973-4 (eBook)
DOI 10.1007/978-1-4684-0973-4
First half of the Proceedings of the NATO Advanced Study

Institute, held at Erice, Italy, Apri112-26, 1978
© 1979 Springer Science+Business Media New York

Softcover reprint of the hardcover lst edition 1979
Originally published by Plenum Press, New York in 1979
AII righ ts reserved

No part of this book may be reproduced, stored in a retrieval system, or transmitted,
in any form or by any means, electronic, mechanical, photocopying, microftlming,
recording, or otherwise, without written permission from the Publisher
To
My Wife Julia
and
My Sons Davide and Christian
PREFACE
This book, divided in two volumes, is the result of a NATO-

Advanced Study Institute, held at Erice during 1978 and aims to
approach in a widely interdisciplinary fashion the problem of
chromatin structure, both at molecular and cellular levels.
It has been edited in an organic and tutorial format, with

the contributions of several leading scientists; to cite a few,
authors range from physical chemists (as K. F. Van Holde and
P. O. TS'o) biologists(as J. Bonner and J. Gilmour) molecular
biophysicists(as M. E. Bradbury and D. Olins) to cytologists
(as T. Casperson and M. Mendelsohn) and geneticists(as L. Sachs).
This first volume, following an introduction to the properties

of isolated chromatin (section I), contains basic chapters which
presents the theory and instrumentation of all techniques (section
II) applicable to the study of nucleosome and chromatin (as
electron microscopy, circular dichroism, hydrodynamics, Raman
Spectroscopy, nuclear magnetic resonance, neutron and x-ray
diffraction, equilibrium binding studies, flow birefringence,
transcriptional assays) and of nuclei and chromosome (cytochemistry,
automated image analysis, microfluorimetry, scanning and flow
cytometry).
In order to make this volume comprehensive and accessible to

a wider scientific community (particularly graduate and post-
graduate students in physical biosciences) equal emphasis is placed
in the presentation both for the breadth, going from molecular
biophysics to biophysical cytology~ and the depth, where the
numerous techniques are treated in detail.
Even if particular reference is made to the genetic apparatus

and its constituents, an attempt is made to present the relevant
advantages and disadvantages of each biophysical method in general
terms, as applicable also to the study of any large biomolecule;
towards this end, brief summaries of the relevant and fundamental
physical principles are frequently given.
Claudio Nicolini
CONTENTS OF PART A
Introduction xi
C. Nicolini
SECTION I: WHAT IS THE CHROMATIN?
Properties and Composition of

Isolat.ed Chromatin 3
J. Bonner
Expressed and Nonexpressed Portions of

the Genome: Their Separation and
Their Characterization • • • • • 15
J. Bonner
Discussion 25
SECTION II: PHYSICAL, CHEMICAL AND

BIOLOGICAL TECHNIQUES FOR STUDYInG NUCLEOSOME,
CHROMATIN, CHROMOSOME AND NUCLEI
Electron Microscopy: A Tool for

Visualizing Chromatin 31
A. L. Olins
Transcriptional Control of
Native Chromatin 41
R. S. Gilmour
Circular Dichroism of DNA, Protein

and Chromatin • • • • • • 67
G. D. Fasman
Important Hydrodynamic and Spectroscopic

Techniques in the Field of Chromatin
Structure • • • • • • • • . • • • 109
D. E. Olins
ix
x CONTENTS OF PART A
Preparation and Analysis of Core Particles

and Nuc1eosomes: A Conveinient Method
For Studying the Protein Composition
of Nuc1eosomes Using Protamine-Release
into Triton-Acid-Urea Gels • • • • • • 125
B. R. Shaw and R. G. Richards
The Interaction of Histones with

DNA: Equilibrium Binding Studies 137
D. R. Burton, M. J. Butler, J. E. Hyde,
D. Phillips, C. J. Skidmore and I. O. Walker
Nuc1eosome Shape and Structure in

Solution from Flow Birefringence 167
R. E. Harrington
Scattering and Diffraction by

Neutrons and X-rays in the
Study of Chromatin 187
J. F. Pardon
Nuclear Magnetic Resonance Studies

of Nucleic Acids and Proteins 217
P. O. P. Tslo and L.-S. Kan
Techniques for Cytochemical Studies

of the Nucleus and its
Substructures 251
T. Caspersson
Chromatin Study in Situ: I. Image Analysis 265

F. Kendall, F. Be1trame and C. Nicolini
Chromatin Study in Situ: II. Static

and Flow Microf1uorimetry 293
C. Nicolini, S. Parodi, S. Lessin,
A. Belmont, S. Abraham, S. Zietz and M. Grattaro1a
Chromatin Study in Situ: III. Differential

Effects of Feu1gen Hydrolysis • • 323
W. A. Linden, S. M. Fang, S. Zietz and
C. Nicolini
Scanning and Flow Photometry of Chromosomes • • • • • • •• 341

M. L. Mendelsohn
Discussion 357
Index • • • xxi
CONTENTS OF PART B
Introduction xiii
C. Nicolini
SECTION' III: VARIOUS LEVELS OF

CHROMATIN ORGANIZATION AND MECHANISMS
FOR TRANSCRIPTIONAL CONTROL
Histones Assembly and Their Structural

Role for Nucleosome Core 371
N. M. Maraldi, S. Capitani, L. Cocco and
F. A. Manzoli
Nuclease Digestion and the Structure

of Chromatin • • • • • • • 389
K. E. Van Holde, J. R. Allen, J. Corden, D. Lohr,
K. Tatchell and W. O. Weischet
Reconstitution of Nucleosomes • • • 413

K. Tatchell and K. E. Van Holde
Conformation of Polynucleosomes in Low

Ionic Strength Solution 427
B. R. Shaw and K. S. Schmitz
Chromatin Structure: Relation of Nucleosomes

of DNA Sequences • • • • • • • • • 441
A. Prunell
Histone Complexes, Nucleosomes, Chromatin and

Cell-Cycle Dependent Modification
of Histones • • • • • • • • • • • • 451
H. W. E•. Rattle, G. G. Kneale, J. P. Baldwin,
H. R. Matthews, C. Crane-Robinson, P. D. Cary,
B. G. Carpenter, P. Suau and E. M. Bradbury
xi
xii CONTENTS OF PART B
Evidence for Superstructures of Wet

Chromatin • • • • • • 515
S. Basu
Chromatin Fractionation and the Properties

of Transcriptionally Active Regions
of Chromatin • • • • • • • • • • • 541
J. Gottesfeld
Chromatin Reconstitution and Non-Histone

Proteins 561
R. S. Gilmour
Discussion 593
SECTION IV: STRUCTURE-FUNCTION OF THE

GENETIC APPARATUS AND CELL CYCLE,
AGING, NEOPLASTIC TRANSFORMATION,
DIFFERE1ITIATION, CHEMICAL CARCINOGENESIS
The Structure and Function of Chromatin

in Lower Eukaryotes 599
K. E. Davies and I. O. Walker
Chromatin Structure from Angstrom to Micorn

Levels, and Its Relationship to
Mammalian Cell Proliferation 613
C. Nicolini
Chromatin Pattern in Situ: Dependence upon

Cell Cycle, Preimplantation and
Development, and Cellular Aging
in Vitro .......... . 667
W. Sawicki
Neoplastic Transformation: The Relevance of

in Vitro Studies for the Understanding
of Tumor Pathenogenesis and Neoplastic
Growth • • • • • • • • • • . . • . . . 683
L. A. Smets
Cell Differentiation and Malignancy

in Leukemia • • • • • • • 705
L. Sachs
Cellular Morphometry in Transformation,

Differentiation and Aging • • 721
S. Parodi, G. Brambilla, F. Beltrame, S. Lessin
and C. A. Nicolini
CONTENTS OF PART B xiii
Basic Mechanisms in Chemical Carcinogenesis • • • • • • • • 751

P. O. P. Ts'o
Carcinogen Induced Alteration in Gene

Packing and Its Possible
Significance in Carcinogenesis 771
P. M. Rao, S. Rajalakshmi and D. S. R. Sarma
Covalent Binding of a Carcinogen to DNA

as a Probe of Chromatin Structure 781
F. X. Wilhelm, M. L. Wilhelm and G. Metzger
Carcinogenesis, DNA Repair and Chromatin 803

W. G. Verly and L. Thibodeau
Electromagnetic Induction of Electrochemical Information

at Cell Surfaces: Application to Chromatin
Structure Modification • • • • •• 811
A. Chiabrera, M. Hinsenkamp, A. A. Pilla and
C. Nicolini
Discussion 841
SECTION V: REVIEW AND SUMMARY OF

THE GENETIC APPARATUS
Session I: Basic Components of the

Genetic Apparatus • • • • • • • • • • • 849
E. M. Bradbury, S. Bram, G. Fasman, D. Olins,
J. Pardon, A. Prunell, R. Sperling, K. E. Van Holde
and I. Walker
Session II: The Second Level of Organization -

Chromatin . . . . • . . . . . • . . • 855
E. M. Bradbury, G. Fasman, S. Gilmour, J. Gottesfeld,
C. Nicolini, D. Olins, J. Pardon, B. Shaw and
F. X. Wilhelm
Session III: The Third Level of Organization 861

E. M. Bradbury, S. Bram, J. Gottesfeld,
F. Kendall, C. Nicolini and I. Walker
Session IV: Generalized Biological Effects • • • • 867

A. Chiabrera, W. Linden, C. Nicolini, S. Parodi
and W. Sawicki
Index •• 871
INTRODUCTION
During April 12-26, 1978, the eighth course of the International

School of Biophysics, a NATO - Advanced Study Institute, was held
at the "Ettore Majorana Center for Scientific Culture" in Erice,
Sicily, co-sponsored by the North Atlantic Treaty Organization,
National Science Foundation (USA), The Italian Government and the
European Molecular Biology Organization.
The subject of the course was "Chromatin Structure and Function"

with 91 participants (from 15 different countries) selected world-
wide.
The current high level of interest in the structure and function

of chromatin"is adequately testified by the thousands of manuscripts
which have appeared in the literature during the past five years
which have pertained to areas directly related to these subjects.
The scope and depth of knowledge and range of disciplines which have
been brought to bear in the study of chromatin structure and its
relation to cell function are indicated in several recent review
articles.
One of the objectives that the Erice course has successfully

accomplished has been to promote the close communication and colla-
boration among scientists active in this field of "chromatin" with
different backgrounds and expertise, such as: biologists,
physicists, biophysicists, biochemists, engineers, and physicians
toward an advancement of knowledge in this basic and interdiscipli-
nary field of life sciences.
The implications of a definite characterization of chromatin

structure and function are now obvious since they bear directly
on the mechanisms of cancer, aging, medical genetics, chemical
carcinogenesis, and cell proliferation.
During the Advanced Study Institute and consequent proceedings,

now published by PLENUM, we adopted a stpuctured, organic and
comprehensive appPOach to the pPObZem of chromatin structure and
function (both at the moZecutar and ceZZutar ZeveZ) with focus on
xv
xvi INTRODUCTION
the methodoZogies 3 techniques and on the various ZeveZs of chpomatin

opganization3 stpessing theip impZications fop ceZZ function.
Today new knowledge, not only in biophysics which is at the
cpossing of severuZ ''hapd'' and "soft" sciences, is frequently
produced by deeply interdisciplinary interactions among scientists
of different backgrounds. In this respect, chromatin constitutes a
unique example since we may identify at least three dimensions where
research is conducted: one (X-axis), al~ng the level of chromatin
organization studied from the Angstrom (histone protein octamers
and the nucleosome) through the multimeters and solenoid, up to the
micron level, i.e. intact interphase nuclei and metaphase chromo-
some; the second (Y-axis), along the methodology and technology
utilized, from biology through chemistry up to physics and engineer-
ing; the third one (Z-axis), along the specific biological system
or mechanism, approached from the concept of the cell cycle, through
aging and carcinogenesis, up to differentiation. Each investigator,
has his own X-Y-Z coordinates in such a '~hpee dimensionaZ configu-
pation" and frequently conducts his search in an isolated environ-
ment with occasional and superficial contacts with the remaining
"scientific space". As occurs also in all other human endeavors,
this frequently leads to an acritical intellectual inertia or at
best to self-perpetuating inner circles, whose primary functions
are to produce an avalanche of "papers", some of which do fulfill
a need for exchange of new findings, but some of which are generated
to satisfy personal, academic or economic imperatives. Looking at
the rate at which the scientific "literature" is growing, one
wonders whether knowledge is growing at the same rate, or whether
intellectual energy and economical resources (of finite amount in
any society) are wasted because proper "vaZue cpitepia and
channeZs of communications" are not open among scientists active
in parallel approaches toward the solution of the same problems.
Need exists, therefore, for the adoption of an absolute reference
system where findings and efforts are to be judged and/or compre-
hensive approaches developed. This should also help to decrease
the so frequently encountered intellectual arrogance (due to
cultural "isolation" or lack of sophistication) and increase the
sense of self-criticism and humility (in terms of a more open
attitude toward new technology or ideas) in studying the complex
mechanisms determining the structure and function of living systems.
In the twentieth century any significant conquest of the human race
(as splitting the atom or reaching the moon) has been the pigopous
(step-by-step, without mirucuZous shoptcuts, as attempted unsuccess-
fully over the past 20 years in cancer research) and anaZyticaZ
'WOpk of teams of scientists with diffepent ''hard science" backgpounds
and expeptise. Even if knowledge is transmitted to younger
generations (in the University) through traditionally separated
disciplines such as engineering, physics, chemistry, biology or
medicine, this surely does not correspond to the way new knowledge
is acquired in all fields of sciences, and particularly in life
science.
INTRODUCTION xvii
To contribute toward the filling of such gaps, participants

and lecturers of the Erice Advanced Study Institute and contribu-
tions to this book on chromatin have been chosen in such a manner
as to warrant spherically isotropic distribution in the three-
dimensional space outlined above.
The simultaneous contribution of several outstanding scientists,

each one a world-wide leader in his own specialization, has per-
mitted me to edit this comprehensive book, which hopefully respects
such interdisciplinary aims. Several books are available in the
area, but they usually cover specific topics, focusing mostly either
on a given technique, biological problem, chromatin constituents,
or level of organization, but few are covering the extremely broad
field in an organic and tutorial format (i.e. comprehensive and
accessible with profit to a wider scientific community) from histone
proteins to intact nuclei, from molecular to cytological approaches.
Within the inherent limitations of any conference proceeding

(such as this) I have attempted to structure the entire book in an
organic and tutorial format, such as to have not a scattered collec-
tion of research papers, incoherent and with frequent unnecessary
overlap, but a sequential series of chapters dealt in depth, from
the basic properties of chromatin throughout all the numerous
techniques employed (occasionally treated in details, including
a brief summary of their basic physical principles), through the
various levels of chromatin organization, up to their implications
for cell function.
The Institute's content did not reflect the volume of literature

pertaining to a particular technique or chromatin component, but
how they are uniquely useful in providing additional and complemen-
tary information on chromatin structure and its relation to cell
function.
Specifically the book consists of four parts, each one followed

by a chapter on the pertinent discussion which occurred at the time
of oral presentation.
I) an introduction to the physical, chemical and biological
properties of isolated chromatin and their relationship
to chromatin of living cells (Janes Bonner, USA).
II) basic chapters which present the theory and instrumenta-
tion of all the numerous physical, chemical, functional,
morphological techniques and methodology applicable to
the study of chromatin, both IN SITU and isolated from
living cells (Stuart Gilmour, UK; G. Harrington, USA;
Gerald Fasman, USA; Ada Olins, USA; Donald Olins, USA;
Ian Walker, UK; Frank Kendall, USA; John Pardon, UK;
Edwin M. Bradbury, UK; Tobjorn Caspersson, Sweden;
Claudio Nicolini, USA; Mortimer Mendelsohn, USA:
B. Shaw, USA; Paul Ts'o, USA).
xviii INTRODUCTION
III) various levels of chromatin organization as determined

by the above techniques, i.e. nucleosome, multimers,
chromatin, chromosomal proteins and their enzymatic
modifications, such as acetylation, methylation, and
phosphorylations in determining gene expression and
chromatin organization (Kensel Van Holde, USA; I. O. Walker,
UK; A. Prunell, USA; John Ploem, The Netherlands; Joel
Gottesfield, UK; S. Bram, France; G. Dixon, Canada;
Donald Olins, USA; B. Shaw, USA; S. Gilmour, UK).
IV) structure and function of the genetic apparatus in the

mammalian cell, stressing their relationship to neoplastic
transformation, aging, cell cycle, medical genetics
differentiation, and chemical carcinogenesis (Edwin
Bradbury, UK; Louis Smets, The Netherlands; Silvio Parodi,
Italy; W. Sawicki, Poland; Walfried Linden, West Germany;
Leo Sachs, Israel; Paul Ts'o, USA; Claudio Nicolini, USA;
D. S. Sarma, Canada; Ian Walker, UK; Ferruccio Ritossa,
Italy; F. X. Wilhelm, France; G. Verly, France).
At the end of the book, (part V) I have included a final review

and synthesis of the genetic apparatus dealing with clarifications
of specific topics, or focusing on controversial issues as models
for chromatin structure and in new avenues as biophysical cytology
or neutron diffraction. The course was of such interdisciplinary
nature that the scientists specialized in one field have been
teaching scientists highly qualified in a different area. The role
of lecturer and student was frequently interchanged during the
meeting as the theme of common interest (chromatin study) was
developed from the viewpoint of different sciences, in a beautiful
small town on top of a mountain overlooking the Mediterranean (that,
according to a legend, was founded by Erice, son of Venus, more
than three thousand years ago). In synthetic analytical terms we
could say, with L. Sachs, that se + AA = LE, that is Science in
Ch:r>omatin pZus Art in ArchaeoZogy equaZ Life in Erice. I t is not
paradox then to state that the Chromatin Institute was held in the
same geographical region where a few thousand years before the Greek
Leucippus and Democritus and Zater on the Roman Lucretius (in his
poem "De Rerum Natura") gave the foundation of biophysics, describing
how the atoms, after various interactions, acquire stabZe configura-
tions, cOI'I'esponding to the Ziving and inanimate worZds. This simple
and unitary theory, which brings Zife science into the reaZm of
physicaZ science, remarkably maintains its validity even after
several centuries of alternative vicissitudes.
To follow the evolution of such fundamental ideas in successive

steps, is quite impossible in such context: I like however only to
recall that the content of this Erice Institute (and therefore of
this book) which relates chromatin structure to cell function,
represents one of the most recent developments of that old idea.
INTRODUCTION xix
Following the earlier discovery of the direct reZationship

between spatiaZ structures of such moZecuZes such as methane and
benzene and chemicaZ activity, the discovery in 1953 of the struc-
ture of the doubZe heZix of DNA represents the turning point for a
simiZar reZationship between three dimensionaZ structure and bioZogy.
It is indeed this relationship that emerged as one of the most
intriguing "take home messages" from the institute: the reZation-
ship between ceZZ jUnction and tertiary (nucZeosomeJ and quaternary
(soZenoid or other form of superpackingJ structures of chromatin
DNA, as moduZated by interaction with histone and non-histone
proteins (and their enzymatic modifications) during the ceZZ cycZe,
ceZZ transformation, aging, and differentiation. In addition to
affirm a more dynamic view of DNA organization in isolated chromatin,
the Erice Institute raises the question as to whether tertiary -
quaternary structures are specificaZZy Zinked to a higher order
(quinternaryJ organization which can now be detected "In situ"
by means of recent technological advancements in the area of
biophysicaZ cytoZogy, to an extent up to now impossible to any
human observer or biochemical assay.
In conclusion, I hope that this book will constitute a useful

and stimulating guideline to doctoral and post-doctoral students
as well as to senior scientists, interested in the most recent
developments in the wide interdisciplinary approach to structure
and function of the genetic apparatus and its constituents and
their relationships to cell function.
Finally, I would like to express my graditude to Professor

Antonio Borsellino for giving me the opportunity to direct the
eighth course of the International School of Biophysics (which have
seen in previous years the active participation also of several
Noble-Prize winners, such as Wald, Eccles, Katz) and to Ms. Pinola and
Dr. Grabriele of the Majorana Centre for coupling high efficiency
and courtesy in a unique cultural setting. My last, but not least,
acknowledgement is to my wife Julia and my Uncle Luigi for their
constant advice and dedication, considering that to realize and
operate within a "three-dimensional scientific space" was a quite
difficult and absorbing experience, even if challenging, not only
in purely scientific terms, but also for its profound social
implications.
Claudio Nicolini
SECTION I:
WHAT IS THE CHROMATIN?
PROPERTIES AND COMPOSITION OF ISOLATED CHROMATIN
James Bonner
California Institute of Technology

Division of Biology
Pasadena, California 91125
Chromatin isolation and properties
We now know that the development of organisms depends on the

selective turning off and on of individual genes at particular
times during the development of the organism. The understanding
of development of higher creatures is therefore the understanding
of the control of gene expression. One way to approach the
control of gene expression is to study the control of gene expres-
sion with isolated interphase chromosomes, the state in which
chromosomes express themselves by transcription into messenger
RNAs or premessenger RNAs. During the last twenty years the study
of isolated interphase chromosomes, called chromatin, has become a
major subject of modern biology. We start herewith with the
isolation of interphase chromatin.
Chromatin isolation
Chromatin was first prepared in semipure form by Zubay ardDoty
(1959) who ground calf thymus tissue in low ionic strength buffer
and purified it by repeated centrifugation and resuspension of the
high molecular weight aggregate followed by solubilization by
shearing. Since thymus cells are almost completely composed of
nuclei this purification turned out to be sufficient for the
purposes of Zubay and Doty and for many purposes since then.
Later methods of preparation of chromatin are based mainly

upon the methods suggested by Huang et al. (1960), Huang and Bonner
(1962) and Marushige and Bonner (1966).--The crude chromatin pre-
pared by the method of Zubay and Doty is layered over 1.7 M
sucrose in 0.01 M Tris buffer, pH 8. The Chromatin is then
pelleted for 2 hours at 22 k rpm in a Spinco ik5 rotor.
3
4 J. BONNER
Membranes and ribosomes bound to membranes as well as adventitious

proteins remain in the gradient while the chromatin pellets. Most
chromatin preparations prepared as outlined above are of very high
molecular weight and form turbid solutions. To overcome this
problem it has been suggested by Marushige and Bonner (1966) that
the chromatin be sheared in the Virtis 45 homogenizer, 30 volts,
90 seconds. This reduces the DNA of the chromatin to about 10 kb
(1 kb = 1000 base pairs). The sheared chromatin is then pelleted
at 10 k rpm in the 8834 rotor of the 80rvall RC2B for 15 minutes.
The soluble supernatant is sheared chromatin.
Almost every investigator or group of investigators of chroma-

tin have produced variants on the above procedure. Two variants
are of particular importance. The first is the preparation of
chromatin from purified nuclei. It is not difficult to prepare
nuclei from liver, frozen liver, or from any other of a host of
animal cells and tissues by the method suggested by Wallace et al.
(1977). The nuclei are then lysed with detergent, the chromati;-
pelleted by centrifugation at 20 k rpm, and resuspended. The
chromatin thus prepared from isolated nuclei is virtually as pure
by criteria. to be outlined below, as that isolated by the method
of Marushige and Bonner, even without centrifugation through
sucrose. This method has the virtue that it eliminates the need
for the sucrose sedimentation step and the ultracentrifuge and
uses only lower speed preparative centrifuges such as the Sorvall
RC2B.
A second modification of procedures invented by protein

investigators is that intr0duced by Chong et al. (1974), Chae
(1975) and Douvas et al. (1975). It has beco;; apparent that
interphase chromatin contains a powerful serine protease and that
this protease degrades not only histones of chromatin but also non-
histone chromosomal proteins. It is well-known from earlier
chromatin studies that calf thymus chromatin stored in a refrig-
erator at 4°c degrades itself so rapidly that after one week as
chromatin it can no longer be considered to be of even semi native
composition. This is due to proteolytic degradation of chromo-
somal proteins. In order to avoid such proteolytic degradation
it is necessary to use serine protease inhibitors such as phenyl-
methanesulfo.nyl fluoride (PMSF) or di-isopropylfluor phosphate
(DFP). Use of these serine protease inhibitors from the very
beginning of the grinding process inhibits proteolytic degradation
of chromosomal proteins and results in chromatin from which pro-
teins can be extracted (myosin, for example) which are not
isolatable from chromatin prepared in the absence of protease
inhibitors.
One important problem and question in the isolation of chro-

tin has been the extent to which chromatin DNA and its associated
proteins is contaminated by nucleoplasmic or cytoplasmic proteins
ISOLATED CHROMATIN 5
which are only adventitiously associated with the isolated chro-

matin. This is particularly an important question since chromatin
isolated by the methods outlined above is isolated at very low
ionic strength, much lower than that of the nucleus which is
about 0.25 M. A considerable number of investigators have iso-
lated chromatin in the presence of separately labeled nucleo-
plasmic or cytoplasmic proteins. It has turned out that the
adventitiously added labeled proteins in such experiments consti-
tute about 5% of total chromosomal proteins and consist of course
of nonhistone chromosomal proteins only (Garrard et ~., 1974).
Furthermore some adventitiously bound proteins isolated with

chromatin, pelleted in low ionic strength buffer are proteins
which are very disadvantageous for the further study of chromatin,
for example, ribonuclease. We have shown that ribonuclease may be
largely removed from chromatin by the precipitation of the latter
(purified chromatin) from low ionic strength buffer, namely, 0.15
or even 0.3 M NaCl. This pellets the chromatin, leaves adventi-
tious proteins in solution and the pelleted chromatin when re-
suspended in low ionic strength buffer, proves to be greatly
depleted in the adventitiously bound ribonuclease. Three such
serial precipitations will substantially remove ribonuclease from
many kinds of chromatin.
Composition of chromatin
Chromatin as isolated from rat liver, which is a model for

chromatins in general, possesses a histone to DNA ratio of about
1:1. It also possesses a nonhistone chromosomal protein to DNA
ratio of 0.6:1. It possesses an RNA:DNA ratio of .1:1.
Chromatin which shows a composition substantially different

from the proportions outlined above may be suspected of contami-
nation. For example, in earlier days it was often found that the
RNA:DNA ratio of isolated chromatin would be of the order of 2 or
3:1. This almost surely indicated the contamination of the chro-
matin by ribosomes and therefore ribosomal RNA. Similarly a very
high protein to DNA ratio, higher than 2:1, should be suspected as
caused by contamination of nonchromosomal proteins adventitiously
adhered to chromatin. Criteria for purity of isolated chromatin
have been outlined in Methods in Enzymology, volume 12B, page 3
and further, 1968 (Bonner et al. 1968).
Histones and histone chemistry
In the early days of the study of isolated chromatin, let us

say, in the years 1950 to 1965, it was not known how many kinds of
histones there are even though histones had been discovered by
Miescher in 1871, the same year and the same Miescher who dis-
covered DNA. Early views concerning the composition of chromatin
6 J. BONNER
and the composition and number of histones are to be found in the

work The Nucleohistones by Bonner and Ts'o (1964). Serious work
on the separation of histones from one another and the study of
their chemistry was done by Rasmussen et al. in the laboratories
of J. Murray Luck in 1958 and continue~bY-Rasmussen and Murray in
1962. These workers found out how to separate three classes of
histones from one another. There were the lysine-rich, slightly
lysine-rich and the arginine-rich histones. (The three classes of
histones all possess approximately 24 mole %basic amino acids).
These studies were continued by Fambrough et al. (1966, 1968, 1969).
Fambrough's studies used the same kind of BioRex P70 column
chromatography with development by guanidine chloride as used by
Murray et al. Fambrough added preparative polyacrylamide gel
electrophoresis as a tool for the analysis of purity of individual
fractions. He showed that there are five histone classes: HI,
H2a, H2b, H3 and H4. These classes are found in all of the
higher eukaryotes and, with some slight alterations, in all lower
eukaryotes as well. Histones are absent from prokaryotes.
DeLan~, Fambrough and colleagues (1969a, b) found that the

primary structure of histone 4 is almost totally conserved as
between peas and cows, an evolutionary history of 600 million
years or more. This strict conservation of primary structure of
histone 4 awakened a great deal of chemical interest in the his-
tones and all of the five histones have now been sequenced from
one or more individual species of creatures. Histone 3 is also
quite conserved (DeIange et al., 1973; Brand and Van Holte, 1972)
Histones ·2a and 2b are somewhat less conserved (Iwai et al., 1970;
Pan yim and Chalkley, 1971; Sautiere et al., 1972). Histone HI,
which is the most different in its properties from the other his-
tones is the least conserved of the five and has the most variants
as has been shown most recently by Cole (1977). As we will see
below histones H4, H3, H2a and H2b playa cooperative role in the
production of a chromosomal unit. Histone HI, on the contrary,
plays a role as a companion of DNA on the spacer sequence between
adjacent chromosomal units.
It will be here noted in passing that the histone genes are

repetitive ones, that is, the gene for each histone is repeated
several times in the genome. This amounts to 20 or so times per
histone species in mammals to several hundred times per histone
species in other organisms (Kedes and Birnstiel, 1971). The
reason why the histone genes are repeated to this considerable
extent may well have to do with the fact that histones must be
rapidly synthesized during cell division time to be able to bind
with DNA produced during the replicative phase of the cell cycle.
The nonhistone chromosomal proteins

The histones associated with DNA and chromatin have been
studied extensively, particularly since their recognition by Huang

and Bonner (1962) as repressors of DNA transcription. The non-
histone chromosomal proteins have been less thoroughly studied
until recent years. One method of study of nonhistone chromosomal
proteins is to first remove histones from chromatin with 0.2 or
0.4 M H2S0 4 in which histones are soluble and in which DNA and
nonhistone chromosomal proteins are insoluble. Following this
treatment the DNA containing nonhistone chromosomal proteins can
be treated with SDS, sodium dodecyl sulfate, and the nonhistone
chromosomal protein then studied by SDS chromatography. This was
first done by Marushige et al. in 1968 and studied more extensively
by Elgin and Bonner (1970) and by Elgin et al. in later years. A
second method of attack on the nonhistone chromosomal proteins is
to remove most of the histones and nonhistone ~roteins from DNA by
dissociation of DNA from proteins with three of four molar sodium
chloride. The dissociated chromosomal proteins are then separated
from DNA either by exclusion chromatography on A 50 or by ultra-
centrifugation which pellets the DNA and separates it from proteins
(Douvas et al., 1975; Van den Broek et al., 1975). The proteins
thus separated from DNA (which do no~include all of the chromo-
somal proteins since H4 and some actin sticks particularly tightly
to DNA even in 4 M NaCl) are then subjected to dialysis to lower
the salt concentration to 0.4 M NaCl. The histones are then
separated from the nonhistones by ion exchange chromatography on
BioRex P70. At this salt concentration and on this weak anion
exchanger, histones are retained while most nonhistones are
allowed to pass through the column.
The SDS complexes of the nonhistone chromosomal proteins,

electrophoresed on polyacrylamide gels, exhibit a wide variety of
molecular weights, from about 225,000 down to the lower limit of
resolution of such polyacrylamide gels, namely, of molecular weight
of about 10 to 15,000. The major nonhistone chromosomal protein
components are similar in a wide variety of chromatins as those
of HeLa, rat liver, Drosophila, etc. (Elgin and Bonner, 1970).
The reason why the major molecular weight components, a dozen or
so, of the nonhistone chromosomal proteins of the chromatins of
different creatures are similar has been demonstrated now by
Martin et al. (1973) and Douvas et al. (1975). Martin et al.
(1973) ·~howed that two major protei~ of the nonhistone~ategory,
those of 38,000 and 40,000 molecular weight respectively, are the
hnRNA packaging proteins. Douvas et al. showed that a major
chromosomal protein of 45,000 molecular weight is actin, while the
the 50,000 and 55,000 molecular weight proteins associated with
all chromatins thus far studied are alpha and beta tubulin, the
components of microtubules. The 225,000 molecular weight protein
of chromatin is myosin. Other minor components of the nonhistone
chromosomal proteins are the serine protease alluded to above as
well as other components of the actomyosin system: actinin,
tropinin, tropomyosin, etc. Approximately half of the mass of all
8 J.BONNER
nonhistone chromosomal proteins are structural proteins as outlined

above.
There has been some discussion of whether or not the actomyo-

sin, tubulin, etc., found in chromatin are artifacts; that is,
artifactually isolated with chromatin, or whether they are de
facto components of interphase chromatin. The bulk of evidence
now suggests that actomyosin and the other structural proteins are
in fact components of chromatin. Thus actomyosin, etc., are not
found in nucleoplasma as free entities. The removal of the nuclear
membrane (and the actomyosin known to be associated with this
membrane) by detergents before the lysis of the nucleus does not
result in any lowered concentration of actomyosin in the subse-
quently isolated chromatin. It is quite probable therefore that
actin and myosin are in fact components of isolated interphase
chromatin. Speculations are principally concentrated on the
possibility that these components may have to do with chromosome
condensation and the subsequent act of mitosis or meiosis. It is
of interest that it is becoming increasingly clear that wherever
chemomechanical systems are found in living beings it turns out
that the chemomechanical system is an actomyosin powered one.
Fifty percent of the total nonhistone chromosomal protein

consists of structural components. What is the remaining 50%?
About 2 to 4% of the total mass of nonhistone chromosomal proteins
bind sequence specifically to homologous DNA (Sevall et al., 1975).
Further minor components are, in addition to the serine protease
alluded to above, acetyl transferase, RNA polymerase and probably
many other enzymes. We do not yet know how to account for all of
the nonhistone chromosomal proteins except that they are very
considerably in number. A 20 cm gel of the whole nonhistone
chromosomal proteins of rat liver chromatin reveals 115 components
which can be resolved ~arrard~ al., 1974). A 12 cm gel of the
nonhistone chromosomal proteins which bind sequence specifically
to homologous DNA reveals at least 100 components (Savage et al.,
unpublished results). There are apparently a vast number of--
chromosomal nonhistone proteins varying in abundance and in prop-
erties. It will still be a vast task to unscramble what they all
do.
Template activity of chromatin
It was found in 1962 by Huang and Bonner that when the acid
soluble proteins are removed from isolated chromatin the thus de-
proteinized chromatin becomes a 10- to 20-fold better template for
transcription by purified RNA polymerase than does chromatin it-
self. When the acid soluble proteins are reconstituted onto the
DNA, the DNA again becomes a poor template for RNA transcription
by purified RNA polymerase. This was in fact the first conclusive
evidence that the acid soluble proteins of chromatin act as regu-

lators of transcribability. On the basis of these findings
Marushige and Bonner (1966) have developed something called the
template activity assay of isolated chromatin. The way this assay
works is as follows: A series of reaction mixtures containing a
constant amount of purified RNA polymerase and constant amounts of
all the substances required for transcription of DNA by RNA poly-
merase is supplied with increasing amounts of either a) chromati~
as rat liver chromatin (or any other kind of chromatin), or b)
DNA prepared from the same chromatin preparation. In the case of
rat liver chromatin a given amount of DNA as pure DNA is approxi-
mately five to ten times more efficient as a template for RNA
synthesis than is chromatin of the same species. We say, there-
fore, that the template activity of the chromatin is 10-20% that
of pure DNA of the same kind. This is true for a wide variety of
chromatins. The template activity of chromatin assay as described
above and as described in detail by Marushige and Bonner (1966)
has been subsequently found to be independent of whether the RNA
polymerase used is homologous, that is, polymerase II of rat liver
chromatin (unpublished work) or heterologous polymerase, that of
purified!. coli polymerase.
The template activity of chromatin may also be determined by

transcription of chromatin with RNA polymerase and subsequent
hybridization of the transcripts to denatured deproteinized
chromosomal DNA by the methods of RNA-DNA hybridization. Alter-
natively, whole nuclear RNA may be hybridized to the DNA of iso-
lated chromatin and the template activity of chromatin in vivo
thus determined. This has been done by a variety of workers, for
example, by Holmes and Bonner. It has turned out that not only
the repetitive and single copy sequences of DNA are transcribed
in vivo in approximately the same proportion as they are repre-
sented in chromosomal DNA, but also, the extent of hybridization
of in yitro or in vivo transcribed RNA to DNA approximates the
template activity of the chromatin as measured by the template
activity assay.
Structure of chromatin
The structure of interphase chromatin has proved to be totally

resistant to study by optical microscopy and until recently,
resistant also to study by electron microscopy. Only during the
years 1974 to present has electron microscopy revealed that the
major component of interphase chromosomes, chromatin, is a beaded
strand consisting of subunits about 100 ~ across and separated
from one another by about 40 base pairs of DNA (Olins and Olins,
1974). The Olins and Olins electron micrographs provided convinc-
ing evidence that interphase chromatin is, at low ionic strength
at least, composed of a "beads on a string" structure; namely,
beads about 100 ~ in diameter separated from one another by short
10 J. BONNER
spacers. The whole structure was studied in more detail by

Griffith in 1975 who found that the viral sv40 minichromosomes,
that is, the sv40 DNA of HeLa cells, becomes complexed with his-
tones in their eukaryotic host cell and also assumes the beads on
the string structure. With this minichromosome Griffith was able
to show that at physiological ionic strength the beads are pressed
against one another so that linkers between beads are not apparent,
just as he has found previously in interphase chromosomes of
higher eukaryotes (Griffith, 1970). At very low ionic strength
such as 0.01 M buffer, the beads on a string separate and reveal
the DNA linker. The beads have been given the name "nubody" by
Olins and Olins (1974) and "nucleosome" by Oudet et al. (1975).
"Nucleosome" seems to be winning the struggle for---;~ival.
An important discovery concerning the structure of the nucleo-

some is that of Garrard et al. (1974) that the molecular stoichio-
metry of the histones, H2A,ih2B, H3, and H4 is approximately
1:1:1:1 with a mass ratio of .9 histone to 1.0 DNA. The mass
ratio requires two molecules of each of these four species of
histones to be present for each one mass of DNA. Our knowledge
of the stoichiometry of the histones in chromatin suggests that
the histones must interact with one another. This is suggested
also from our knowledge of the primary structure of the histone.
All of the four histones enumerated above, those present in the
nucleosomal structure, consist of a highly basic N-terminal
peptide and a highly hydrophobic C-terminal peptide. This struc-
ture suggests that the N-terminal peptide is for interaction with
nucleic acids, while the C-terminal peptides are for interaction
with other proteins. That this is so has been shown convincingly
by D'Anna and Isenberg beginning in 1974. These workers found
that histones do interact with one another in DNA-free solution.
The interactions are through the hydrophobic C-terminal tails and
the affinities are very great. The work of D'Anna and Isenberg
and also of Kornberg and Thomas and others (1974) has shown that
an octamer of eight histones, two of each species, appear to
constitute a core of the nucleosome structure and that the DNA is
bound around this core.
A particularly imaginative approach to the structure of chro-

matin was introduced by Hewish and Burgoyne (1973) and followed up
by Kornberg and Thomas in 1974 and by many others since that time.
It was revealed by Hewish and Burgoyne that interphase chromatin
when it is digested by micrococcal nuclease is cleaved into DNA
fragments of a standard length approximately 200 base pairs long.
Further digestion reduces the length of the unit fragment to about
160 base pairs, the so-called "core" fragments. The core fragment
is complexed to the histone octamer. The 40 base pairs which are
not in the core fragment is the linker and this linker is complex-
ed with histone 1. Histone 1 was earlier shown in 1970 by
Griffith and Bonner (unpublished results) to be stationed at
intervals about 160 base pairs apart along the DNA strands. This
was shown by making use of the method of Brutlag et al. (1970) by
which histone 1 is covalently attached to DNA by treatment for 1
hour with 1% formaldehyde at OOC. The other histones are not
covalently attached to DNA. By this treatment they can be removed
by banding the HI-DNA complex in cesium chloride of the appropriate
density. The resulting DNA-HI complex was subjected to high res-
olution electron microscopy and the HI molecules found to be
regularly distributed, as pointed out above, approximately 160 base
pairs apart. Many other pieces of evidence had previously suggesir
ed that HI does different things to chromatin than do the other
four histones. For example, the removal of HI has little effect
upon the biophysical properties of chromatin. It does not change
the melting properties of DNA or its hyperchromicity (Tuan and
Bonner, 1969) or the characteristic X~ray diffraction patterns of
chromatin fibers (Richards and Pardon, 1970).
The structure of interphase chromatin is therefore now

relatively clear. The bulk of the chromatin is composed of nucleo-
somes which are connected by linkers of DNA complexed with HI.
Each nucleosome is composed of an octamer of histone and the DNA
is wound around the outside of the octamer. A vast variety of
enzymological studies on the nucleosome have been carried out.
The DNA is arranged in such a fashion that it can be degraded by
micrococcal nuclease into 10 base pair long fragments by suffi-
ciently long degradative attack.
Nucleosomal structure appears to be a characteristic of

eukaryotic chromosomes. It is found not only in higher plants and
in animals but also in Dictyostelium, a lower protozoan (called a
slime mold), as well as in true fungi as in yeast and Neurospora.
In the lower organisms such as Dictyostelium and Tetrahymena,
histones are present and nucleosomes are found, but the histones
are not electrophoretically identical to the histones of higher
creatures. They differ slightly (or considerably in the case of
HI) in molecular weight and amino acid composition. Nonetheless,
nucleosomal structure and histones and their properties appear to
have been considerably conserved since the beginning of the
eukaryote kingdom.
Chromatin in the nucleus and in the test tube
Chromatin as isolated by the procedures outlined above,

is soluble in low ionic strength buffer and may be worked with as
a soluble solution of a nucleohistone. In 0.15 M to 0.3 M NaCl
or KCl solution, however, chromatin becomes almost completely in-
soluble. This is the range of ionic strength found in the nucleus
of higher organisms. Therefore, in the nucleus of higher organ-
isms the chromatin must be insoluble. We shall see below that the
12 J. BONNER
template expressed portion of chromatin is soluble in solutions of

such ionic strength and we must therefore envisage the chromatin
of interphase nuclei as consisting of an aggregated, precipitated
mass of nonexpressed nucleosomal chromatin from which protrude a
few loops of soluble chromatin which is in the expressable form,
that is, in a form transcribable by RNA polymerase. The differ-
ence between transcribable and nontranscribable chromatin will be
returned to below. It is important for our purposes, however to
note and even to stress that chromatin as we know it in the test
tube is not similar to chromatin as it is found in the nucleus in
life. In all probability chromatin as it is found in the nucleus
in life assumes higher orders of structure, the nucleosomal chain
coiling itself into structures of higher order.
REFERENCES
Zubay, G. & Doty, P. (1959) J. Mol. Biol. 1:1-20.

Huang, R. C. C., Maheshwari, N. & Bonner, J. (1960) Biochem.
Biophys. Res. Comm. 3:689-694.
Huang, R. C. C. & Bonner~ J. (1962) Proc. Nat. Acad. Sci. 48:
1216-1222.
Marushige, K. & Bonner, J. (1966) J. Mol. Biol. 15:160-174.
Wallace, R. B., Sargent, T., Murphy, R. & Bonner-,-J. (1977) Proc.
Nat. Acad. Sci. 74:3244-3248.
Chong, M.T., Garrard:-W. T. & Bonner, J. (1974) Biochemistry 13:
5128-5134.
Chae, C. B. (1975) Biochemistry 14:900-906.
Douvas, A. S., Harrington, C. & Bonner, J. (1975) Proc. Hat. Acad.
Sci. 72:3902-3906.
Garrard, ~ T. & Bonner, J. (1974) J. Biol. Chem. 249:5570-5579 ..
Bonner, J., Chalkley, G.R., Dahmus, M., Fambrough, D., Fujimura,
F., Huang, R. C. C., Huberman, J., Jensen, R., Marushige, K.,
Ohlenbusch, H., Olivera, B., & Widholm, J. (1968) Methods in
Enzymology, 12B:3-65.
Bonner, J. & Ts'~P. (Eds.) The Nucleohistones, Holden-Day, Inc.,
San Francisco, California.
Rasmussen, P., Murray, K. & Luck, M. M. (1962) Biochemistry l:
79 -89.
Luck, J. M., Rasmussen, P., Sutake, K. & Tsuetikova, A. (1958).
J. Biol. Chem. 233:1407.
Fambrough, D. & Bo~, J. (1966) Biochemistry 2:2563-2570.
Fambrough, D., Fujimura, F. & Bonner, J. (1968) Biochemistry I:
575-584.
Fambrough, D. & Bonner, J. (1968) J. BioI. Chem. 243:4434-4439.
Fambrough, D. & Bonner, J. (1969) Biochim. Biophys. Acta 175:
113-122.
DeLange, R., Fambrough, D., Smith, E. & Bonner, J. (19693.) J. Biol.
Chem. 244:319-334.
DeLange, R:-;-Fa.I!lbrough~ D., Smith, E. & Bonner, J. (1969b) J. Bic:iL.
Chem. 2~4:5669-567~.
DeLange, R., Hooper, J. & Smith, E. (1973) J. BioI. Chem. 248:

3261-3274.
Brand, T. & VanHolte, C. (1972) FEBS Letters 23:357-360.
Iwai, K., Ishikawa, K. & Hayashi, H. (1970) Nature 226:1056-1058.
Panyim, S. & Chalkley, R. (1971) J. BioI. Chem. 246:7557-7560.
Sautiere, P., Tyrou, D., Laine, B., Mizon, J., Lambelin-Breynaert,
M., Rufin, P. & Biserte,G. (1972) C. R. Acad. Sci., Paris
274:1422-1425.
Cole~. D. (1977) In: "The Molecular Biology of the Mammalian
Genetic Apparatus", P.O.P Ts'o (Ed.) Vol. I: pp.91-104,
North Holland, Amsterdam.
Kedes, L. & Birnstiel, M. (1971) Nature New BioI. 230:165-169.
Marushige, K., Brutlag, D. & Bonner, J. (1968) Biochemistry 1:
3149-3155.
Elgin, S. C. R. & Bonner, J. (1970) Biochemistry 9:4440-4447.
Douvas, A. S., Harrington, C. A. & Bonner, J. (1975) Proc. Nat.
Acad. Sci. 72:3902-3906.
Van den Broek, H., Nooden, L., Sevall, J. S. & Bonner, J. (1973)
Biochemistry 12:229-236.
Martin, T., Billings, P., Levey, A., Ozarslan, S., Quinlan, T.,
Swift, H. & Urber, L. (1973) Cold Spring Harbor Symp. Quant.
BioI. 39: 92l.
Sevall, J. S., Cockburn, A., Savage, M. & Bonner, J. (1975)
01ins, A. & 01ins-,-D. (1974) Science 183:320-332.
Griffith, J. (1975) Science 187:1202-1203.
Griffith, J. (1970) Ph.D. Thesis California Institute of
Technology.
Oudet, P., Gross-Bellard, M. & Chambon, P. (1975) Cell 4:281-300.
Garrard, W. T., Pearson, W., Wake, S. & Bonner, J. (1974) Biochem
Biophys. Res. Comm. 58:50-57.
D'Anna, J. & Isenberg, 1-.-(1974) Biochemistry 13:4992-4997.
Kornberg, R. & Thomas, J. (1974) Science 184:865-868.
Hewish, D. & Burgoyne, L. (1973) Biochem. Biophys. Res. Comm.
52:504-510.
Tuan-,-D. & Bonner, J. (1969) J. Mol. BioI. 45:59-76.
Richards, B. & Pardon, J. (1970) Exp. Cell Res. 62:184-196.
EXPRESSED AND NONEXPRESSED PORTIONS OF THE GENOME:
THEIR SEPARATION AND THEIR CHARACTERIZATION
James Bonner
California Institute of Technology

Division of Biology
Pasadena, California 91125
INTRODUCTION
We have considered the nature of the bulk of the eukaryotic

genome. It consists of DNA complexed with histones in nucleosomal
configuration. This bulk is not transcribed into RNA. What is
the nature and structure of that portion of the genome which is
transcribed? Several methods for separation of transcribed from
non-transcribed portions of the genome have been described. Some
of these are commented upon below.
METHODS
Sucrose density gradient centrifugation
It has been known for many years that when isolated chromatin
prepared as described in Lecture 1 is sheared by Virtis shearing
(30 volts/90 sec) and subjected to sucrose density centrifugation,
the resulting fragments exhibit a wide variety of sedimentation
coefficients. This range is from about 30 Svedberg units (S units)
to over 100 S units (Chalkley and Jensen, 1969). In more recent
work it has often been found that sucrose density gradient centri-
fugation of chromatin sheared as described by Marushige and Bonner
can be separated into two discrete components: A light component
of sedimentation coefficient about 30 S, and a heavy component of
sedimentation coefficient over 100 S, with little intervening
material. Chalkley and Jensen have shown that the heavy component
can be transformed to the light component by treatment with 5 M
urea and that this transition is not reversible. In addition, it
has been suggested from time to time that the light fraction
15
16 J. BONNER
represents the template active portion of chromatin and that the

heavy fraction represents the template inactive portion of chroma-
tin. To what extent is this in fact true? We have found (Savage
and Bonner, 1978) that the light fraction when it is studied by
the methods of renaturation kinetics contains the complexity of
the whole genome, that is, the light fraction which constitutes
about 10% of the whole genome for rat liver chromatin, is a
random set of the se~uences of the whole genome and, therefore,
no se~uence fractionation has been accomplished by separation of
light and heavy chromatin by sucrose density gradient centrifuga-
tion.
Other methods for separation of active from inactive chromatin

are reviewed by Savage and Bonner (1978). We have, in addition,
developed a method which appears to successfully separate express-
ed portions of the genome and to separate them in a ~uantitative
fashion from the nonexpressed portions of the genome. This method
relies upon attack of chromatin by the enzyme DNase II which makes
double stranded clips of the DNA of chromatin as well as of
purified DNA (Marushige and Bonner, 1971; Billing and Bonner, 1972;
Gottesfeld et al. 1974,1975). Because this nuclease makes double
stranded clips on chromatin DNA, it may be thought of as a shear-
ing agent, even though shearing is in principal a hydrodynamic
concept. Marushige and Bonner incubated unsheared chromatin with
DNase II in pH 6.6, which is rather far from the pH optimum of
the enzyme which is approximately 4.6. At the end of the incuba-
tion period the reaction is stopped by raising the pH to 7.5, even
further removed from the pH optimum of the enzyme. Unsheared
chromatin is separated by centrifugation at 15 kg. The rapidly
attacked material remains in the supernatant and is referred to as
supernatant 1, while the pellet is referred to as pellet 1. The
supernatant may be further fractionated by the addition of 0.15 M
NaCl or 2 mM MgC1 2 . Nucleosomal DNA, particularly that containing
histone 1, is extremely insoluble in NaCl or magnesium chloride at
these concentrations, and therefore precipitates. The resulting
precipitate, removed by centrifugation, is known as pellet 2 and
the remaining supernatant as supernatant 2. The kinetics of
digestion are clear-cut. With sufficient time as much as 80% of
the chromatin is solubilized into supernatant 1. By this time,
pellet 2 grows to about 50% of the total chromosomal DNA and the
supernatant S2, to about 20% of the total chromosomal DNA.
The final limiting amount of supernatant ~ S2, DNA obtainable

from a given chromatin varies according to the template activity of
the chromatin in ~uestion. Thus chromatin of Novikov cells has a
template activity 10% of that of its DNA. It yields an S2 e~ual
to 10% of total ascites chromosomal DNA. Rat liver chromatin with
a template activity of 20% with respect to its DN~yields 20% of
its DNA as supernatant 2.
EXPRESSED AND NONEXPRESSED GENOME PORTIONS 17
After 5 minutes of incubation with DNase II about 10% of rat

liver chromosomal DNA is found in the S2 fraction. After this
time of incubation the DNA has a double stranded length of about
700 base pairs and a single strand length of about 200-500 bases.
Characterization of the solubilized material has been done with
this 5 minute supernatant 2 fraction. That it is indeed the
transcriptionally active component in chromatin is shown by the
facts 1) that it bears nascent RNA, that is, RNA pulse labeled
in vivo with a radioactive tracer; 2) that it is hybridized to a
high level by whole cell RNA; and 3) that it is a subset of the
single copy se~uences of whole rat DNA as well as a subset of the
repetitive sequences. The sequences of rat liver DNA released by
DNase II are therefore a subset of all of the DNA sequences of the
whole rat genome, and these are the sequences which are hybridized
by the RNA sequences which are transcribed in rat liver.
Billing and Bonner (1972) found that over 70% of pulse

labeled nascent RNA is released from ascites cell chromatin when
less than 10% of the DNA has been solubilized by DNase I, that is,
the DNA of chromatin which bears nascent RNA is also readily
attacked by DNase I which makes single stranded nicks in DNA.
The same result has been obtained with DNase II.
The fact that DNase I selectively degrades DNA sequences which

are transcribed into RNA has been used extensively by Garel and
Axel (1976) and Weintraub and Groudine (1976). Garel and Axel
found that oviduct chromatin isolated from chicks induced to
produce ovalbumin lost its ability to react with ovalbumin messen-
ger cDNA following partial digestion with DNase I. Weintraub and
Groudine obtained similar results with the globin gene system. In
our laboratory we have found that if chromatin is first attacked
by DNase I the DNA isolated from that chromatin has lost its
ability to hybridize to S2 DNA released from chromatin by DNase II
digestion (Savage and Bonner, unpublished results). To understand
these results one must remember that DNase I rapidly degrades DNA
to acid soluble, very small oligonucleotides, while DNase II
releases large oligonucleotides and under the conditions used as
described above liberates single stranded fragments of an average
length of about 300 bases. One might say that DNase I destroys
active genes while DNase II releases active genes.
A third enzyme which has been used a great deal in the study
of chromosomal structure is Micrococcal nuclease, sometimes known
as Staphylococcal nuclease. This enzyme which produces single
stranded nicks in double stranded DNA is almost nonselective as
between expressed and nonexpressed portions of the genome. When
it is used to attack whole chromatin, it reduces approximately 50%
of the total exposed DNA to acid soluble fragments. Thus we have
an array of three enzymes: DNase II makes double stranded clips in
DNA and selectively clips out the expressed portion of the genome,
18 J.BONNER
DNase I by single stranded nicks selectively degrades to acid

soluble material the expressed portion of the genome, while
Micrococcal nuclease indiscriminately attacks all DNA of the
genome with the reservation that that portion most protected by
histones and nucleosomal structure is attacked much more slowly
than is the DNA of linkers, etc. Micrococcal nuclease does not
selectively release or degrade the expressed portions of the
genome.
FINDINGS
--------
Chemical composition of template active chromatin.
If rat liver chromatin is attacked by DNase II and the

template active segments released as described above, it becomes
possible to compare the chemical composition of the expressed
portion of the genome with that of the nonexpressed portion of the
genome. The histone-DNA ratio of the expressed portion is approx-
imately the same, perhaps slightly lower, than that of whole chro-
matin. The difference, if any, may be artifactual, due to protease
degradation of histones. The point to be emphasized is that
template active chromatin does contain histones and in nearly the
same proportion to DNA and in nearly the same proportion to one
another as the template active inactive portion of the genome or
as the whole chromatin.
When chromatin is attacked with DNase II, it is converted into

200 base pair long pieces of DNA. These fragments contain 8
molecules of the core histones and 1 molecule of HI. The template
active portion of the genome is also converted into 200 base pair
long units of DNA containing histones in the same proportion as
described above. Thus, histones do not disappear from DNA during
transcription. Therefore, the DNA of template active chromatin is
organized into subunits which are similar in size to those of
template inactive chromatin. In the limit digest of template
active chromatin about 50% of the DNA is converted to acid soluble
material, either by DNase II or by Micrococcal nuclease. This is
also true of whole chromatin. The physical structure of the
repeating units of template active chromatin is not yet clear.
What is known concerning this matter will be returned to below.
Properties of template active and template inactive chromatin
The first of these properties has to do with the melting

spectrum of the DNA of the several fractions of chromatin. When
rat liver chromatin is dialyzed to very low ionic strength (0.25
mM EDTA) , under which conditions pure rat liver DNA melts at 43°,
chromatin melts with three transitions. The first at about 56°;
the second at about 75°; and the third at about 85°. That the
latter two transitions are due to the complexing of histones to
EXPRESSED AND NON EXPRESSED GENOME PORTIONS 19
DNA is shown by the fact that when histones are selectively re-
moved by d~ociation of chromatin with increasing concentrations
of salt, these latter two transitions disappear while the transi-
tion at 43° progressively increases. The transition at 55° is not
affected, or at least minimally affected, by the removal of
histones.
The template active S2 portion of chromatin melts quite

differently. The template active portion shows a transition at
43°; that is, some of the DNA of the template active portion of
chromatin melts essentially as does pure DNA. A major portion
melts at 57° and little or none melts at the transitions associated
with histone complexed DNA. The melting properties of the template
inactive portion of chromatin are those of whole chromatin with
major transitions in the histone-DNA complex regions. These
facts imply that in the template active portion of chromatin,
histones do not interact with DNA as they do in template inactive,
or bulk, of chromatin.
Similarly, other types of biophysical investigations have

shown that the DNA of template active chromatin is more like pure
protein-free DNA than is the DNA of whole chromatin. Thus the
cDNA spectrum of template active chromosomal DNA is more like free
DNA than is that of template inactive chromosomal DNA (Gottesfeld
et al., 1974). It may be remembered also that electron microscopy
has shown that there are genes which are clearly expressed and
clearly recognizable as expressed by electron microscopy; namely,
the ribosomal genes of eukaryotes. It has been shown by Hamkalo
et al. (1973) that the contour length of the DNA undergoing tran-
scription by RNA polymerase to produce the ribosomal RNA precursors
is equal to the contour length of the RNA being produced, that is,
the DNA under transcription to form ribosomal RNA is completely
and fully extended. Its packing ratio is 1 as contrasted to the
packing ratio of DNA in whole (beads on a string) chromatin which
is 7 or greater. This finding also implies that the DNA of
expressed chromatin is relaxed, extended, and behaves with respect
to other physical properties as deproteinized DNA.
Transformation of inactive chromatin to active chromatin
We have seen that transcribable chromatin is characterized

by physical properties different from those of non transcribed
chromatin. 1) The DNA of transcribed chromatin exhibits a
relatively low T. 2) The packing ratio of transcribed chromatin
is close to 1. ~) the circular dichroism spectrum of transcribed
chromatin is that of B-form DNA. 4) Transcribed chromatin is
readily attacked by nucleases such as DNase II and DNase I. In
all of these respects the physical properties of isolated template
active chromatin approach the properties of free DNA. What kind
of modification of chromosomal proteins could bring about this
20 J. BONNER
alteration in protein-DNA interaction, an alteration which would

cause histones to cease to stabilize DNA against melting and to
cease to confine DNA to a packing ratio of 7 and to instead allow
the DNA to relax and extend? There has been a great deal of
speculation over the years that histone modification of some kind
might be responsible for the "activation of genes." Thus Allfrey
et al., 1964 have suggested that transcriptional control might be
achieved by histone modification as, for example, by acetylation of
the s-amino groups of lysines.
Marushige (1976) has in fact shown that chemical acetylation of

calf thymus chromatin increases its transcribability by purified
RNA polymerase and this without any removal of histones from
chromatin. Histones and chromatin remain associated. They are,
however, associated in someway different from that characteristic
of template inactive chromatin. We have studied this whole phe-
nomenon in considerable detail. We have found (Wallace et al.,
1977) that chemical acetylation of chromatin with acetic~hydride
as described by Marushige not only makes chromatin transcribable
by added RNA polymerase, but also causes the acetylated chromatin
to assume the physical properties of template active chromatin.
As a result of acetylation the DNA of chromatin becomes low melt-
ing. It also becomes sensitive to DNase I and to DNase II. It
does not become significantly more susceptible to Micrococcal
nuclease digestion than it was before acetylation. Further, it
has been shown by Marushige and by our group (unpublished) that
acetylation by acetic anhydride causes acetylation of the s-amino
groups of lysine and that the lysines which are acetylated are all
contained in the N-terminal peptides of the core histones, that is,
the acetylation occurs in those portions of the histones which
interact with DNA rather than in those portions of the histones
which interact with one another to form the nucleosomal core. We
interpret these findings to mean in a general way that acetylation
of chromatin causes the histones of the nucleosomal core to relax
their binding of N-terminal peptides to DNA and thus to allow the
DNA to escape, and to an extent that permits the DNA to relax to
a lower packing ratio. This results in turn in the DNA of the
nucleosome becoming transcribable. Thus acetylation is a possible
candidate for the role of transformation of template inactive
nucleosomes to template active chromosomal configuration.
We have no firm evidence yet whether acetylation is in fact the

primary cause of such transformation. We do know that pulse
labeling of chromatin with acetate causes template active chromatin
(82) to always contain a 2-4-fold higher specific activity of
acetyl groups per unit histone than does template inactive chroma·-
tin.
EXPRESSED AND NONEXPRESSED GENOME PORTIONS 21
There are, there~ore, still important unknowns to the under-

standing o~ the operation o~ chromatin and the activation o~
speci~ic genes. The ~irst is how sequence speci~icity is intro-
duced into the trans~ormation. I have described above how
acetylation could, ~or example, cause the observed changes and
alteration between template inactive and template active con~or
mation o~ genes. What is it that tells which gene is to be thus
trans~ormed, perhaps by the histone acetylase o~ chromatin (a chro-
mosomal enzyme)? It has been described in Lecture I how a portion
o~ the rat liver nonhistone chromosomal protein binds sequence
speci~ically to homologous DNA and, in ~act, to ~amilies o~
repetitive sequences. These proteins might be the agents which
introduce sequence speci~icity into histone acetylase-chromatin
interaction, or into some analogous chromosomal trans~ormation
interaction.
A second point o~ great interest and importance is whether or

not acetylation is in ~act the agent which trans~orms inactive into
active genes. Our two principal questions are,there~ore: 1) the
nature o~ trans~ormation and 2) the nature o~ the agent which
sequence speci~ically speci~ies that the trans~ormation shall take
place here.
All o~ the studies o~ how to trans~orm nontrans.cribable genes

into transcribable genes in the chromatin o~ higher organisms have
been carried out with whole chromatin; that is, genomes containing
many tens o~ thousands o~ genes capable o~ expression, a ~ew
thousands o~ which are expressed in any particular kind o~ chrom~in
and a ~ew tens o~ thousands o~ others which are expressed in the
chromatin o~ other organs o~ the same creature. All described
attempts to turn on particular genes (in the chromatin o~ higher
organisms) have been done using the methods o~ reconstitution o~
dissociated chromatin. Chromatin is dissolved in 2 M NaCl and
5 M urea. All histone and nonhistone proteins are dissociated ~rom
DNA. Next we gradually dialyze the NaCl away; next the urea.
Chromatin which looks like the original is reconstituted. The
reconstituted chromatin is now transcribed by RNA polymerase and
the resultant transcripts compared by hybridization-competition
with those o~ native chromatin.
Fidelity o~ reconstitution (sequence ~idelity) has been ~ound

by those whose hybridization studies included either (1) only
repetitive sequences (Bekhor, Kung and Bonner 1969; Huang and
Huang 1969), or (2) have used c-DNA probes to individual messages
(Paul et al., 1973; Stein et al., 1975. Both approaches are
~raught with arti~acts.
In addition, those who have sought to show that gene expression

is controlled by the complexing o~ particular nonhistone chromo-
somal proteins to DNA have even more problems. Their method is
22 J.BONNER
to dissociate, for example, brain chromatin, add erythropoetic

chromosomal nonhistone proteins, reconstitute the mixture and then
show that globin message is transcribed from the altered brain
chromatin. Problems of endogenous message, super sensitivity of
c-DNA probe, etc., abound. No single reconstitution study is
convincing.
It seems to me that the difficulty of working with a large

complex of genes simultaneously is an overwhelming one and makes
almost impossible pure and clean experiments. There are questions
of the residual messenger RNA contained in chromatin before it was
transformed and of the messenger RNA possibly contained in the non-
histone chromosomal portion allotted to the reconstituted chroma-
tin, as well as the fact that the gene to be probed as globig or
histone is one gene whi~h constitutes perhaps one part in 10 , or
at most, one part in 10 of the genome, or less. The signals
expected are very small. New methods are needed. They have
appeared in the nick of time.
Our current approach to the subject of the control of gene ex-

pression in eukaryote organisms is to isolate chromosomal frag-
ments 10-20 kb in length and containing particular expressed or
nonexpressed coding sequences from, for example, rat liver chroma-
tin. We clone these coding sequences by recombinant DNA technology.
We then reconstitute them with core histones into minichromosomes.
We now try to find what it is, for example, in liver nuclei which
causes the serum albumin gene of rat liver to become expressed,
transcribable, to produce the messenger RNA characteristic of
serum albumin. We have not yet found out what the element is in
rat liver nucleosomal material which when added to the fragment
of rat liver DNA containing this serum albumin coding sequence and
packaged as a mini chromosome causes this mini chromosome to become
transcribable. I am confident, however, that this technology, a
much simpler and surer one than any tried heretofore, will in the
long run lead us to discover both the nature of the elements which
confer sequence specificity upon the control of gene expression
and the nature of the alterations in histone-DNA interaction which
cause the alteration in physical conformation of nucleosomes so
that they become both extended, low melting and transcribable. To
my way of thinking, in the isolation and multiplication of individ-
ual specific coding sequences and their flanking regulatory se-
quences lies the secret of the unlocking of the mystery of the
control of gene expression.
EXPRESSED AND NON EXPRESSED GENOME PORTIONS 23
REFERENCES
Chalkley, R. & Jensen, R. (1969) Biochemistry I: 4380-4833.

Savage, M. & Bonner, J. (1978) In: "Methods in Cell Biology",
G. Stein and J. Stein (Eds.) Academic Press, New York, in pres~
Marushige, K. & Bonner, J. (1971) Proc. Nat. Acad. Sci. 68:
2941-2944. --
Billing, R. & Bonner, J. '(1972) Biochim. Biophys. Acta 281:453-
462.
Gottesfe1d, J., Garrard, W., Bagi, G., Wilson, R. & Bonner, J.
(1974) Proc. Nat. Acad. Sci. ]1:2193-2197.
Gottesfe1d, J., Bagi, G., Berg, B. & Bonner, J. (1976) Biochemistry
15 : 2742-2482 .
Gare1, A. & Axel, R. (1976) Proc. Nat. Acad. Sci. 11:3966-3970.
Weintraub, H. & Groudine, H. (1976) Science 193:848-856.
Gottesfe1d, J., Bonner, J., Radda, G. & Walker, I. o. (1974)
Hamka10, B., Miller, o. & Bakken, A. (1973) Cold Spring Harbor
Symp. on Quant. Bio1. 38:915-919.
A11frey, V., Faulkner, R.~ Mirsky, A. E. (1964) Proc. Nat. Acad.
Sci. 11:3937-3941.
Marushige, K. (1976) Proc. Nat. Acad. Sci. 73:3937-3941.
Wallace, R. B., Sargent, T., MUrphy, R. & Bonner, J. (1977) Proc.
Nat. Acad. Sci. 74:3244-3248.
Bekhor, I., Kung, G-.-& Bonner, J. (1969) J. Mol. Bio1. ~:351-364.
Huang, R. C. C. & Huang, P. C. (1969) J. Mol. Bio1. 39:365-378.
Stein, G., Maus, R., Gabby,E., Stein, J., Davis, J. & Adamadkan, P.
(1975) Biochemistry 14:1859-1866.
Paul, J., Gilmour, R., Affara, N., Birnie, G., Harrison, P., Hell,
A., Humphries, S., Windass, J. & Young, B. (1973) Cold Spring
Harbor Symp. Quant. Bio1. 38:885-890.
DISCUSSION (PART I)
DR. WILHELM: Did you try to digest the nuclei first with DNase I
and then with DNase II to see whether the active. fraction S2
disappeared?
DR. BONNER: We did digest chromatin with DNase I (~5-20% made

acid soZubZe). The remainder yieZded no 82 on DNase II digestion.
DR. WILHELM: Could you explain why you obtain only an enrichment
of 7 for the globin genes in the active fraction and why the
physical properties - thermal denaturation for example - of
the S2 fraction are so different from the bulk of chromatin
although there is a full complement of histone bound to the
DNA?
DR. BONNER: The 82 of ceU chromatin is about 15% of totaZ DNA.

More or Zess att 82 is removed from chromatin and aZZ gZobin
gene is in 82' so the enrichment in 82 over whoZe chromatin
wouZd be seven-fotd. This is what is found. The physicaZ
properties of 82' incZuding its thermaZ denaturation profite,
atZ indicate that the N terminat peptides of the his tones have
reZaxed their bindings to DNA.
DR. 8~: If you treat S2 DNA with DNase I to an extent of
1 to 10% do you digest all the transcribing genes?
DR. BONNER: Yes, because it is very rapidZy attacked by DNase I.

DR. 8ARMW: Does this mean that S2 DNA has lots of non-transcriba-
b1e genes?
DR. BONNER: I do not say this. It has both transcribabZe and

transcribing genes.
DR. 8~: What makes a transcribab1e gene different from
transcribing gene?
25
26 DISCUSSION
DR. BONNER: I do not know. Both ready to be transaribed and

transaribing genes seem to be in extended form.
DR. SARMW: What is the biological and functional advantage of
the nucleosomal structure?
DR. BONNER: Probably it represents the first step in aompaating

2 meters of DNA into the nualeus.
DR. SARMW: How are the two classes of transcribable genes (the
multicopy rRNA genes and tRNA genes synthesizing rRNA and
tRNA and unique copy gene synthesizing mRNA) distributed in
chromatin in terms of nuclease accessibility and conformation?
DR. BONNER: rRNA, tRNA and mRNA genes are aU in the S2 fraation.
However, eleatron miarosaopy evidenae has been aaaumulated whiah
indiaates that rRNA genes appear to laak nualeosomal partiales,
while mRNA genes possess nualeosomes.
DR. SARMW: DMSO treatment is generally coincident with the
release of viral sequences. Could this affect the transcription
of hemoglobin genes?
DR. BONNER: It is quite true that DMSO induaes expression of the

virus. However, we have not studied whether this aould affeat
the expression of hemoglobin genes.
DR. DIXON: You mentioned that in S2 there is almost three times
as much NHCP as in bulk chromatin. Is this NHCP enriched in any
particular NHCP species?
DR. BONNER: The non-histone ahT'omosomal proteins of the S2 fraation

aontains all of the Hn-RNA paakaging protein. Aatin, tubulin
and actomyosin are also present mainly in S2 fraation.
DR. SMETS: Does the S2 chromatin increase during DNA replication?
Are there cell cycle dependent changes in the percentage of
S2 DNA?
DR. BONNER: It has not been determined.

DR. VAN HOLDE: How much does the sedimentation coefficient drop
when histones are acetylated?
DR. BONNER: The sedimentation aoeffiaient drops to about 78.

DR. LANGLOIS: Do metaphase chromosomes have a different structure
at the nucleosome level than interphase chromatin?
DR. BONNER: I do not know; perhaps people working at the E. M.

level will be able to tell us.
DISCUSSION 27
DR. NICOLINI: I am disturbed by the lack of interest and precision

in defining the object of your analysis and how much it could be
different from the original object "in situ". Without such an
effort it is difficult to communicate. In studying higher
levels of organization such standardization is extremely important.
DR. BONNER: I think we are communicating here. MY view on

physiocochemistry of chromatin is that these types of studies
are useful only as they explain something. I am aware of the
effects of shearing. Chromatin prepared following method one3
that I described this morning3 is approximately 10 Kbases long.
DR. NICOLINI: You seem to feel that the main effect of shearing
is only the falling off of some terminal protein in the fragment.
It was however observed at the E.M. level that unsheared chroma-
tin displaysofibers 300A in diameter, while sheared chromatin
has only lOOA diameter fibers (Klug and co-workers).
DR. BONNER: I am aware of this difference in fiber diameter3 but

We had no occasion of discussing this subject.
DR. NICOLINI: Am I correct in summarizing your feelings, that you
think efforts in characterizing and standardizing our object
of investigation and their relationship with the original
object are of minor importance?
DR. BONNER: I have already stated that my 10 Kbases chromatin is

something that does not exist in the cell; however3 it is an
object simple enough for starting a study. You can do things
that you cannot do with an insoluble blob. I did not say that
you do not want to characterize the blob 3 but we have to study
the complex structure after we get to know something about the
elementary structure.
DR. NICOLINI: I think however that a strong word of caution
should be given on the possible artifacts, since too frequently
and superficially investigators did and still do extrapolate
their findings on sheared chromatin, to determine possible
mechanisms which control gene expression. Major efforts
should go both in physically standardizing the object of
investigation, and in improving the technology in order to
study chromatin as close as possible to its truly native
conditions. Higher order of chromatin organization (as
solenoid) could indeed be the key for an understanding of
control transcription and gene expression.
SECTION II:
PHYSICAL, CHEMICAL AND BIOLOGICAL TECHNIQUES
FOR STUDYING NUCLEOSOME, CHROMATIN,
CHROMOSOME AND NUCLEI
ELECTRON UICROSCOPY: A TOOL FOR VISUALIZING CHROMATIN1<
AnA L. OLINS
The University of Tennessee-Oak Ridge Graduate School

of Biomedical Sciences and Biology Division, Oak Ridge
National Laboratory, Oak Ridge, TN 37830
The molecular biologist is fortunate that many of the macro-

molecules of interest can be visualized by the electron microscope,
thus helping in the interpretation of biophysical and biochemical
information. The problem arises in finding objective criteria for
evaluating such images; gleaning the meaningful information and
discarding the chaff. It is the goal of this article to set forth
such guidlines to sharpen the student's awareness of the character-
istics of acceptable electron microscopic data. The evaluation of
a micrograph, as other biophysical data, can and should be objec-
tive and not based on artistic merit. Progress in our understanding
of chromatin structure has been accelerated by electron microscopy
and will serve as an example for this discussion.
CRITERIA FOR GOOD ELECTROU t1ICROGRAPHS
It requires a good deal of patience to take interesting, high

quality electron micrographs, yet only in very rare cases can I
recommend acceptance of a poor quality micrograph. Electron micros-
copy, like all other good data, must be reproducible, thus present-
ing the microscopist with many opportunities to improve the quality
of the micrograph. Some characteristics expected in good electron
micrographs are as follows:
*Research supported by the Uational Science Foundation Grants

PCM76-01490 and PCM77-21493 and by the Division of Biological and
Environmental Research, U. S. Department of Energy, under contract
W-7405-eng-26 with the Union Carbide Corporation.
31
32 A. L. DLiNS
1. A clean, empty background allows rapid identification of

the sample. Amorphous stain deposits, dirt and bacteria can be
confusing and certainly raise questions about the care exerted in
doing the experiments.
2. The contrast in the micrograph should be sufficient to

permit easy recognition of the structure as described by the micros-
copist. This does not, of course, apply to Xerox copies of micro-
graphs, transpariencies projected in an illuminated room or slides
viewed at a great distance.
Biological specimens are composed of small atoms which have

little intrinsic electron density, therefore most contrast is ob-
tained by the addition of electron dense molecules (stains).
Positive staining is achieved by incubating the specimen with

the stain followed by washing the specimen in order to remove all
unbound stain. This method requires a high affinity between stain
and specimen.
Negative staining is achieved by allowing an excess of stain

to surround the specimen producing a dark outline of a light macro-
molecule. The stain is not washed off after it is applied to the
grid. Since it does not (usually) have a high affinity for the
specimen, it generally surrounds the structures on the grid. The
goal is to obtain a very delicate outline of the structure. Often
the stain is too thick and obstructs the edges of the specimen.
Sometimes the stain dries in an uneven pattern leaving gaps which
could be misconstrued to represent the location of macromolecules.
In this case, reproducibility and correlation with other known
characteristics of the specimen must be considered by the micros-
copist. A simple control, rarely shown in the literature, is to
look at a grid covered with film and stain, but having no specimen
on it. Some stain irregularities can be eliminated by filtering
the stain before application to the specimen. Many different stains
should usually be tried, giving the microscopist a good opportunity
to become familiar with the sample structure and its variation with
different stains.
Shadowing a specimen with metal (e.g., Pt, Pd, Au) vapor in

vacuo produces very highly contrasted specimens, but is usuallY-done
in such massive quantities as to greatly increase the size of the
specimen. This method has been extremely useful for studying DNA,
spread in the presence of cytochrome c (Kleinschmidt, 1959), because
the length of these long molecules is not greatly changed by the
layers of cytochrome or metal. Single- and double-stranded DNA can
be distinguished, the contrast is very good, and biochemical data
is usually available to support the interpretation of micrographs
of shadowed specimens. However, this method cannot be reliably used
ELECTRON MICROSCOPY FOR VISUALIZING CHROMATIN 33
for high resolution ultrastructural analysis of macromolecules with-

out special precautions to reduce the amount of metal evaporated and
the granularity of the metal deposits. Shadowing can be done on a
rotating sample, or unidirectionally, the latter being especially
useful in determining the height of the molecular specimens.
Supporting film thickness and uniformity are extremely impor-

tant, especially for high resolution studies of stained macromole-
cules. Since the image is formed by absorption or scattering of
electrons, the larger the difference in electron density between
specimen and supporting film (background), the greater the contrast
in the final micrograph.
Recently, we have adopted the methods of Ottensmeyer et ale

(1975) for tilted beam, darkfield microscopy to achieve very mark-
edly improved contrast. In this method the image is formed by the
scattered electrons, eliminating the "noise" of the bright undeflec-
ted primary electron beam. For this method, more than any of the
others discussed, a thin uniform supporting film is essential for
improving the contrast.
30 Fine high resolution electron micrographs should be either

in focus or slightly underfocus. At focus, in bright field, the
grain is not visible and contrast is at a minimum; at an underfocus
setting, the grain is sharp or crisp; at an over focus setting, the
grain is blunt or rounded. It is important to set the correct focus
at a magnification higher than 40,OOOX so that the grain is visible
and adjustments in the objective lens current can be made accurately.
Proper focus cannot be achieved unless the astigmatism, manifest by
a directional orientation of the grain, of the objective lens is
corrected. The size of the background grain should be smaller than
the structure of the macromolecular detail of interest; i.e. a pat-
tern seen within the image of the macromolecule, which is also seen
in the background, cannot be interpreted to be macromolecular sub-
structure.
4. Magnification of an electron micrograph should be suffi-

ciently high so that the detail of interest is easily visible.
This magnification should be achieved by the electron microscope
and not a photographic procedure, so that the focus is correct and
a large grain pattern does not dominate the structural detail.
SPECIMEN PREPARATION
Life cannot proceed in the vacuum of the electron microscope.

At best, we can only hope to visualize an aspect of the structure
which reflects the true organization of molecules in their biolog-
ical roles. How much distortion can be tolerated? How little
distortion can be achieved? These are difficult questions, whose
34 A. L. DLiNS
answers change as new techniques are devised. At present, however,

we can suggest and justify criteria adopted within our laboratory
for preparing chromatin specimens:
1. Chromatin tends to form gelatinous fibrillar aggregates,

which should be avoided because the underlying substructure of the
fibers is obscured or altered. It is especially helpful to use
freshly isolated nuclei or chromatin. Frozen cells, frozen nuclei
or frozen chromatin all tend to form gelatinous aggregates. This
is not seen, however, in frozen nuclease digest fragments, especially
monomers (VI)' dimers (V2) and trimers (V3). We usually apply the
samples to the grids the same day that live cells are harvested.
The forces of air-water or air-buffer interphase spreading seem to
cause similar gelatinous fibrillar networks and this method is,
therefore, avoided in our laboratory.
20 Hany strange structures can be found on an electron micro-

scope grid. Uhich are artifactual and which reflect the structure
of chromatin? The obvious control of looking at stained grids with-
out sample has already been mentioned. Other correlations with the
known biochemical and biophysical properties of the sample must be
made. For example, if hydrodynamic data (such as analytical ultra-
centrifugation) and electrophoresis in DnA gels indicate that the
sample has the properties of VI but the sample on the grids consists
of large fibers or aggregates, it can be assumed that the prepara-
tion of the chromatin solution for microscopy should be improved.
If reconstructed samples are to be studied, close comparison

should be made to the original starting material; e.g., higher order
organization of nucleosomes in 20-30 nm fibers should be studied
before and after dissociation w·ith EDTA, if a "native" reorganizing
effect of l1g+t is to be demonstrated.
Another example of a useful comparison is the resemblance of

the ultrastructure of VI with nucleosomes on long chromatin fibers
spilling out of a freshly isolated nucleus, thus establishing their
identity with more certainty.
3. Is the structure of interest reproducible? This question

should have an affirmative answer for: a regular distribution of
the structure within a single sample; and for the reappearance of
the structure in samples prepared at different times from separate
sources. The rapid acceptance of the universality of the nucleosome
as the fundamental unit of chromatin structure depended, in large
measure, on its ubiquitous distribution in electron micrographs and
its correlation with biochemical data.
o
4. Organic solvents such as ethanol destroy the 100 A low-
angle x-ray reflections (Pooley et al., 1974) and the appearance of
nucleosomes in the electron microscope (\Voodcock et al. , 1976) • For

this reason, critical-point drying, and dehydration with organic
solvents, has been avoided in our laboratory. Salt and pH conditions
must also be carefully controlled and their effects understood. For
example, 0.6 If NaCl is known to remove the lysine-rich histones in
most eucaryotes; therefore, chromatin treated with this level of salt
cannot reflect the structural contribution of the HI histone class.
CURREUT OPTIHAL TECHHIQUES USED IN OUR LABORATORY
The methods used in our laboratory have been strongly influ-

enced by the important technical breakthrough made by Miller and
Beatty (1969), which permitted spreading of chromatin for visuali-
zation of "genes in action". This procedure, which involved isola-
tion of nuclei, centrifugation onto carbon-coated grids and drying
from dilute Kodak Photo-flo, displayed the close proximity of ru~A
polynerase molecules, the increasing lengths of ru~A transcripts
within a gene and the tandem arrays of ribosomal genes. Although
chromatin within the nucleus must be much more compact than seen in
such preparations, it demonstrates the primary level of association
of RNA, protein and DnA. Higher-order interactions, and internal
structure of DHA-protein and RNA-protein molecules were not devulged
in these early studies.
Our initial observations of nucleosomes (v bodies) were made on

fibers streaming out of isolated chicken erythrocyte nuclei. The
biochemical data showed that more than 95% of the proteins in these
nuclei are histones. The negative staining indicated that thicker
regions of the chromatin fibers were formed by close association of
several nucleosomes. The particles were visible everY\lhere. Further
rapid support for the particulate structure of chromatin came from
biochemical data reported in many laboratories (Felsenfeld, 1978).
Several samples of current chromatin micrographs are presented

in this section to illustrate the points raised earlier in this
manuscript. Figure 1 is an electron micrograph of a portion of a
chicken erythrocyte nucleus, with extended chromatin fibers stream-
ing out of the nuclear periphery.
This darkfield electron micrograph is in focus; the grain is

symmetric and the contrast is maximal. Although the stain was not
washed off the grid it seems to bind specifically to certain regions
of the chromatin. This is probably due to the high affinity of the
positively charged uranyl ions for the negatively charged phosphate
and carboxyl groups in the chromatin fibril. For thicker specimens
(see Figure 3) or at higher concentrations of uranyl acetate, the
stain would surround the specimen as a true "negative stain". A
bright field electron micrograph of a sir.lilar preparation ,.ould have
much less contrast (Olins et al., 1975).
Co)
0.
Figure 1. A darkfield electron micrograph of the edge of a chicken erythrocyte nucleus. The
freshly isolated nuclei were gently suspended in 0.2 M KC1, then diluted 1:100 with 0.2 ruM EDTA. l>
After 10 minutes of swelling the solution was made 0.9% HCHO, and centrifuged through 10% HCHO !
or
pH 7 onto a glowed carbon-coated grid. The grid was then washed in dilute Kodak Photo-flo,
drained on the edge of Bibulous paper, dried, stained for 30 seconds with 0.1% uranyl acetate, Z
en
drained and dried again. For details see Olins and Olins (1974).
Figure 2. A darkfield electron micrograph of isolated nucleosomes.

Chicken erythrocyte nuclei were digested with micrococcal nuclease
and fractionated by sucrose gradient centrifugation. A drop of vI
A260 = 1.0 was placed on a glowed carbon-coated grid for 30 seconds,
rinsed with dilute Kodak Photo-flo, drained, dried and stained as
in Figure 1.
In order to visualize nucleosomal ultrastructure, a micrograph

of isolated monomer nucleosomes (Figure 2) is presented. Again,
darkfield is used to achieve contrast. The clarity of detail is
due, in part, to the high magnification (200,000X) at which the
micrograph ,,,as made and, in part, to the low concentration of stain
used. The similarity of the particles in the chromosomal fibers of
Figure 1 and the nucleosomes (VI) isolated after micrococcal nucle-
ase digestion in Figure 2 is demonstrated.
The true negative staining results demonstrated in Figure 3

show the close packing of nucleosomes in the 20-30 urn fiber (Olins,
1977). Stain is clearly surrounding the particles rather than bind-
ing to them. These fibers are so thick that the internal structure,
i.e. the nucleosomes, cannot be clearly visualized in darkfield
microscopy. This is due, in part, to the extra electron density of
th e negative stain surrounding the fiber and, in part, to the super-
imposition of nucleosomes on opposite sides of the thick chromatin
fiber. This micrograph is slightly underfocus, as can be seen by
the increased contrast and the crisp symmetric background grain.
38 A. L. DLiNS
Figure 3. A bright field electron micrograph of a 20-30 nm chroma-

tin fiber at the periphery of a chicken erythrocyte nucleus. Freshly
isolated nuclei are suspended in 0.25 H KCl 0.2 mH HgC12 then diluted
1:100 in 0.2 H KCl. After 10 minutes the nuclei are fixed and pro-
cessed as described in the legend for Figure 1.
The final figure in this demonstration is a darkfield micro-

graph of part of an extrachromosomal nucleolar ribosomal gene as it
is being transcribed. Although this method of preparing the sample
preserves the nucleosomal structure of the inactive chromatin fibril,
it is not yet possible to describe the structure of either the ro~A
polymerase or the rolP fibrils. It is evident that new preparative
methods must be devised which preserve the structure of these impor-
tant chromatin components. The carbon film used in this preparation
is exceptionally uniform and thin, producing fine contrast of a del-
icate specimen.
The challenge of preparing high-quality micrographs was elo-

quently presented 300 years ago.
"Tis not unlikely, but that there may be yet invented

several other helps for the eye, as much exceeding
those already found, as those do the bare eye, such
as by which we may perhaps be able to discover living
Creatures in the Moon, or other Planets, the figures
of the compounding Particles of matter, and the partic-
ular Schematisms and Textures of Bodies." CR. Hooke,
Hicrographia, 1664.)
m
r
m
(')
--t
:0
o
Z
s:
(')
:0
oen
(')
o"'0
-<
."
o
:0
<
en
C
l>
r
N
z
Cl
(')
:I:
:0
o
s:
l>
--t
Z
Figure 4. A darkfield electron micrograph displaying part of a ribosomal gene in transcription

and an inactive chromatin fiber. Freshly dissected Triturus viridescens oocytes were opened in
0.15 Ii KCl to isolate the germinal vesicle (g.v.); in less than 2 minutes the g.v. was placed
in a drop of water pH 9, opened with forceps and allowed to disperse. The contents of a single
g.v. were centrifuged through 3.7% HCHO, 0.1 M sucrose pH 8.8 onto a glowed carbon-coated grid
W
and processed as in Figure 1. '<l
40 A. L. DUNS
ACKNOHLEDGEHENT
I thank Mayphoon H. Hsie and Elizabeth A. Wilkinson for ex-

cellent assistance, G. J. Bunick for vI and D. E. Olins for helpful
discussions.
REFERENCES
Felsenfeld, G. (1978) Nature 271, 115-121.
Hooke, R. (166 Lf) "Micrographia", Printers to the Royal Society,

London.
Kleinschmidt, A., and Zahn, R. K. (1959) Z. Naturforschg. l4b, no-

779.
Miller, O. L., and Beatty, B. R. (1969) Science 164, 955-957.
Olins, A. L. (1977) Cold Spring Harbor Symposium on Quantitative

Biology 42, in press.
Olins, A. L., Carlson, R. D., and Olins, E. E. (1975) J. Cell BioI.

~, 528-537.
Olins, A. L., and Olins, D. E. (1974) Science 183, 330-332.
Ottensmeyer, F. P., Hhiting, R. F., Schmidt, E. E., and Clemens,

R. S. (1975) J. Ultrastruct. Res. ~, 193-201.
Pooley, A. S., Pardon, J. F., and Richards, B. M. (1974) J. Mol.

BioI. ~, 533-549.
Woodcock, C. L. F., Safer, J. P., and Stanchfield, J. E. (1976)

Exp. Cell Res. 22, 101-110.
TRANSCRIPrIONAL CONTROL OF NATIVE CHROM.ATIN
R. Stewart Gilmour
Beatson Institute for Cancer Research
Wolfson Laboratory for Molecular Fathology
Garscube Estate, Switchback Road, Bearsden
Glase;ow, G6l JBD, Scotland
WHAT IS CHROMATIN?
Most work which has been done with chromatin refers to the
dispersed chromosomes of interphase cells. Isolated chromatin
consists of DNA, histones, non-histone proteins (NHP) and small
quantities of RNA. Numerous comparisons have been made of the
relative concentrations of these components in chromatins from a
wide variety of sources (see review by MacGillivray and Rickwood,
1974) • Although in higher eukaryotes histones and DNA maintain
an approximate one to one ratio, in lower eukaryotes this is not
the case. The levels of NHP vary considerably throughout the
species. There is some circumstantial evidence to suggest that
the amount of NHP is directly related to the amount of
transcriptionally active DNA in chromatin. The main definitional
problem is to decide which of the many types of molecule wit~in
the nucleus should be considered as a genuine component of
chromatin. The original preparative procedures simply involved
the extraction of whole cells with isotonic saline; however,
with the recognition of cytoplasmic proteins, membrane and
ribosomes, procedures were altered to exclude these contaminants.
Obviously a molecule which never associates at any time with the
DNA might not be considered as a p;!.rt of chromatin, for example,
RNA processing enzymes; but what about proteins which are
transiently associated with the DK~ and furthermore may shuttle
between nucleus and cytoplasm? Clearly chromatin is not a static
entity but can undergo a variety of chemical and structural
changes in response to the metabolic status of the cell. The
purpose of this chapter is to outline the in vitro experimental
approach that has led to the idea that chro~has built in
transcriptional controls which regulate the availabity of genetic
41
42 R. S. GILMOUR
information. In a subsequent chapter the structural features

responsible for this phenomenon will be considered.
EVIDENCE FOR DIFFERENTIAL GENE TRANSCRIPl'ION
One of the strongest arguments for differential gene activity

has 'come from the study of puffing in the polytene chromosomes of
larval Diptera. It has been firmly established that puffs are
chromosome sites undergoing active transcription and detailed
analysis of puffing patterns have revealed the presence of tissue
specific puffs as well as characteristic changes with development
(see reviews by Ashburner 1972 and Fanitz 1972). These results
which have been obtained from studies of several Chironomus and
Drosophila species suggest differential gene activity during
development and in functionally different cell types. Although
puffs have not been detected in the chromosomes of higher animals,
evidence suggests that chromatin can exist in different states of
condensation. Electron microscopy studies on interphase nuclei
of calf thymus lymphocytes described heterochromatin (condensed)
and euchromatin (diffuse) forms. Incorporation of radioactive
precursors in vitro showed that the bulk of the rapidly labelled
RNA was associAted mainly with euchromatin (Allfrey and Mirsky
1962; Allfrey ~., 1963; Frenster, 1963).
The application and development of DNA-RNA hybridization
matho'is to eukaryotic systems has contributed more than any
other single technique to our understanding of the relationships
between genomic DNA and its transcription product.
A demonstration of differential gene expression requires
evidence for a selection of DNA sequences transcribed into RNA.
Early investigations of the sequence homologies of total cellular
RNA isolated from a variety of species demonstrated that
qualitatively different RNA molecules are present in different
cell types of the same organism; for example in mouse tissues
(McCarthy and Hoyer, 1964), developmental stages of sea urchin
embryos (Glisen et al., 1966) and embryonic mouse liver (Church
and McCarthy 1967},'"Xenopus oocytes and blastulae (Davidson
~., 1968) and mouse liver uterus before and after oestrogen
stimulation (Church and McCarthy, 1970).
The finding that eukaryotic chromatin can support the
transcription of RNA in vitro in the presence of added RNA
polymerase, usually bacterial, suggested a cell-free approach to
the study of this phenomenon. Us ing the same hybridization
techniques to compare the homologies of in vitro transcribed RNA
and total genomic DNA it was concluded that~ability of chromatin
to support RNA synthesis is restricted compared with that of DNA.
This appeared to be due to a limitation in the availability of
sites in DNA which can be transcribed so that a restricted set of
TRANSCRIPTIONAL CONTROL OF NATIVE CHROMATIN 43
RNA sequences is produced (raul and Gilmour, 1966, 1968; Bekhor

et al., 1969; Huang and Huang, 1969; Smith et al., 1969,
Spelsberg and Hnilica, 1970; Sawada~., 1972). In some of
these experiments it was shown that in vitro RNA transcripts were
indistinguishable from the .!!2 !!!2 ~ 'OT'the organ from which the
chromatin was isolated, as judged by competitive hybridization.
(Faul and Gilmour, 1968; Smith~., 1969; Tan and Miyagi,
1970). This led to comparative analyses of chromatin transcripts
from different organs of the same animal and it was generally
concluded that while there were many common RNA sequences
transcribed tissue specifio RNA sequenoes were also synthesized.
However, with the discovery that eukaryotio DNA is composed

of both repetitive and unique nucleotide sequences it beoame
olear that in all these hybridization experiments the
ooncentrations of hybridizing DNA and RNA sequences wer~ such
that only the homologies between RNA and the reiterated sequences
in the DNA were detected. No information about unique sequence
transcripts can be obtained using these hybridization conditions.
There is now a growing body of evidenc~ which indicates that
structural gene sequences in eukaryotes probably ocCUr once per
haploid genome. It is likely that globin (Bishop and Freeman
1973; Bishop and Rosbash, 1973; Harrison et al., 1974),
ovalbumin (Harris ~., 1973) and fibroin (Suzuki ~., 1972)
are coded by Single genes although in the case of sea urchin
histones multiple gene copies are present (Weinberg et al., 1972).
That this conclusion can be extended to total cellulariiiRNA has
been determined by following the hybridization of labelled mRNA
or the cDNA derived from it by reverse transcription in the
presence of excess cellular DNA. Studies of this nature which
have been carried out on Friend cells (Birnie ~, 1974),
HeLe. cells (Spradling ~., 1974; Klein~., 1974; Bishop
!i..!!. 1974), rat myloblast (Campo and Bishop 1974), mouse L
cells tGreenberg and Perry, 1971) sea urchin embryos (Goldberg
~, 1973; Davidson~., 1975) and Dictyostelium (Lodish
et al., 1973) indicate that almost all mRNA sequences are
tranScribed from non-repetitive regions of the genome. For
further discussion the reader is referred to the review by
Lewin (1975).
Clearly this generalisation challenges the inference of the

original hybridisation experiments with in vivo made RNAs and
ohromatin primed RNAs. Where tissue spec~hybridisation was
equated with the presence of specific mRNA species. While it
is obvious that these would not have been detected, the fact
remains that these experiments show that broad tissue specificities
exist within the repetitive DNA transcripts and as yet remain
unexplained. The whole question of whether selecti.ve gene
expression can be seen in in vivo made RNA has been re-examined
from the point of view of messenger sequences. This has been
R. S. GILMOUR
done by hybridizing RNA to the unique DNA sequences isolated from

the total genomic DNA or to cDNA made to specific mUNA. Under
these hybridization conditions the concentrations of the unique
sequences is effectively concentrated and hence can undergo
hYbridization. For example, transcripts derived from unique DNA
sequences were analysed by hybridising isolated unique DNA to a
large excess of total cellular RNA or in some cases nuclear UNA.
There again distinctly different populations of unique sequences
are present in the RNA's of different tissues as demonstrated in
adult mouse tissues (Brown and Church, 1972; Grouse~.,
1972) ,mouse emb~~nic stages (Church and Brown, 1972),
Dictyostelium developmental stages (Firtel, 1972) and chick
oviduct before and after oestrogen stimulation (Liarakos ~.,
1973) •
The recent use of globin eDNA hybridization probes by a
number of workers has extended this conclusion for the particular
case of globin maNA in erythroid versus non-erythroid cellular
HNA (Axel ~., 1973; Leder et ale, 1973; Steggles.!L!!.,
1974; Humphries~., 1974; Gilmour.ll..!1:., 1974).
These results, from a wide variety of sources, attest to the
generality and significance of differential gene transcription.
Selective sequences of both repetitive and unique DNA are
transcribed ~ .!!!£ in a tissue specific fashion.
The problem of analysing in vitro synthesised chromatin
transcripts for specific mRN.As"1iaS'"been greatly facilitated by the
use of specific cDNA hybridization probes. In particular a
number of groups have reported the B1 ~ transcription of
globin mRNA from erythroid chromatin in a variety of tissues and
have contrasted this with a corresponding lack of these sequences
in transcripts from non-erythroid chromatin of the same species.
(Axel ~., 1973; Gilmour and Faul, 1973; Steggles~.,
1974; Barrett et al., 1974). Harris et al., (1975) has found a
SUbstantial concentration of ovalbumin mRNA sequences in the
transcripts from oestrogen stimulated chick oviduct chromatin
as compared with unstimulated oviduct chromatin. In HeLa cells
stein et ale (1975) have described the B1 ~ transcription of
histone genes in chromatin from S-phase cells but not in G1-phase
chromatin. Since then the validity of these and subsequent
results has been challenged by a number of criticisms applicable
to in vitro chromatin transcription in general. It seems
appropriate at this stage to consider the justification of these
criticisms and the attempts that have been made to answer them.
VALID~Y OF IN VITHO TRANSCHIPl'ION - (a) Endogenous RNA
The problems associated with endogenous RNA are best
illustrated by examining the author's results with mouse foetal
liver chromatin. This tissue attains maximum haemopoietic

activity afier 14 days .i!!. !i.E:2 at which time about 70~1, of the
cells are erythroid. Chromatin was prepared from 14 day foetal
livers and adult mouse brain and incubated with E. coli RNA
polymerase. Transcripts were isolated and hybridized to globin
cDNA to estimate the concentration of globin mRNA sequences.
The details of these steps are given below.
Chro~atin preparation: 1 g foetal livers are hand homogenized

(Dounce homogenizer, 0.001 inch pestel clearance) in 36 mls
2 mM Tris; Hel pH 7.5; 5 DIM MgC12; 1 roM Dithiothreitol (DTT);
3 roll CaC 12 and left on ice for 5 mins. After adding 4 mls of
2 M sucrose another 30 secs. additional homogenization is
carried out. 20 ml aliquots of homogenate are layered on 15 mls
of 2.(~i sucrose; 2 mM Tris pH 7.5; 1 mM DTT; 5 mM MgCl ;
0.28 m NaCl and centrifuged in a swing-out rotor at 30,006 g for
1 hr. at 4°C. The pellet is suspended in 20 mls 1 mM Tris;
0.1 roM EDTA; 0.28 m NaCl; 1~ Triton-X-100 and centrifuged at
12,000 g for 10 mins. The pellet is washed three times with
1 mM Tris; Hel pH 7.9; 0.1 roM EDTA. The viscous gel
constitutes chromatin,
Transcription: Reaction mixtUre (2 mls) contain 0.04 M Tris

pH 7.9; 2.5 mM MnC12j 0.1 roM EDTAj 0.1 mM DT'l'; 0.8 roM ATP,
CTP, GTF and UTP; 0.5 - 1 mg DNA as chromatin and 100 Burgess
tmits of E. coli RNA polymerase.
RNA isolation: Ai'ter incubation for 1 hr at 37°C the reaction

is cooled to 4°C and 20 J.l.g RNAse free DNAse added. Af'ter a
further hr. at 4 0 C sodium sarcosinate and EDTA are added to final
concentrations of 0.2% and 30 roM resp. and the mixture incubated
with 200 Ilg Proteinase K (International Enzymes) for 30 mins at
370 C. A single phenol:chloroform (50/50:v/v) extraction is
carried out and the aqueous phase passed over Sephadex G-50.
\'1here unmercurated UTP is used columns are run in 0.05 m NaCl;
10 IIlM Tris pH 7.6. The excluded material is precipitated with
2.5 vols. ethanol re-dissolved in 1 ml HEPES pH 7, 5 roM MgC12;
0.25 m NaCl; 3 111M Cael.:? and incubated with 20 Ilg RNAse free
for 15 mins. at 37°C. After re-extraction with phenol:
chloroform and chromatography on Sephadex G50 the RNA is
precipitated with ethanol; dried and hybridised.
Hybridization and estimation of globin sequenoes: In this 3 studY

we used a titration teohnique in which a fixed amount of [ H]
cDNA (1 ng) is annealed at 43°C with increasing amounts of RNA in
10cfl hybridization buffer (0.5 M NaCl; 25 mM HEPES; 1 mil EDTA;
5Cl' formamide, pH 6.7). The theoretical considerations of this
approach and the treatment of results are described elsewhere by
by Young~. (1974). Hybrid was measured by the sensitivity
of cDNA to single-stranded nuclease (S1) prepared from Takadiastase
46 R. S. GILMOUR
by the method of sutton (1971). The method is demonstrated in

Figure 1 whioh shows the titration of pure 93 globin mRNA and
reticulocyte polysomal RNA to 1 ng globin cDNA. Complete
hybridization of the pure globin mRNA is achieved at input ratios
of RNA: cDNA = 1.6: 1, consistent with the f:U1ding that cDNA
represents a 600~ copy of the mRNA. The input conoentration of
polysomal RNA required to achieve saturation is 50 times that
needed for pure mRNA, thus giving an estimate of 2% for globin
mRNA content. A small percentage of cDNA (5%) is normally
resistant to 3 nuolease in the absence of RNA while an
additional 10-15% does not appear to be homologous to 93 RNA and
is degraded.
An alternative method of estimating globin mRNA content is to

carry out a kinetio or Rot analysis. This method will be
demonstrated in later data.
Figure 2 shows the hybridization analysis of RNA transoribed

from foetal liver and adult mouse brain chromatins. There is an
apparent absolute difference between them; about 0.002% of the
foetal liver transoript is globin specific. In a control
incubation RNA polymerase was omitted and an appropriate amount
of t-RNA added. Re-isolation of the t-RNA was carried out in
parallel with the transcripts; significant hybridisation was seen
in this case. The globin RNA sequences deteoted in the oontrol
have come from the endogenous RNA that co-isolates with the
chromatin and oontributes an estimated 50% of the observed
hybridisation in aotively transcribing inoubations. The most
serious criticism for the present argument is that any in vivo
synthesized endogenous RNA isolated with the chromatin template
will be indistinguishable from in vitro RNA synthesized by the
exogenously added po~erase unl8ss-careful controls are
included.
There seems to be some varianoe in the literature as to

the amount of endogenous RNA found in the chromatin of different
systems, as judged by control incubations containing chromatin
but no polymerase. Nearly allthe data referred to are derived
from hybridizations with excess RNA, where even minute levels of
endogenous RNA contamination of the ~ ~ t~cripts will
produce misl~ng results. It is not oertain that actively
transcribing inoubations and oontrol incubations are exactly
eqUivalent in all other respects except for the presence of
RNA po~erase. In view of the discrepancies it is difficult
to assess the reliability of • enzl'IDe minus' oontrols l!!!: ~
unless they are supplemented by more rigorous data.
A more reliable oontrol has been enployed for measuring

both endogenous RNA and newly synthesised RNA simultaneously in
100
80
f-
0/-
0
.JP.--- -
0
,a
~
a
/
f
60
0
UJ
N
0
0:: 40
-
III
>-
:::c
«
:z
0
u 20
cw
o a
-
2 3 4
o 50 100 150 200

RNAIc DNA
.- .
Legend to Figure 1: Titration to 1 ng [3 H] globin cDNA of
pure 9S globin mRNA ( 0 -
( ).
0 ) amd reticulocyte polysomal RNA
48 R. S. GILMOUR
100r---------------------------__~
CI
LIJ
N
CI
~
al
>-
::t:
«z
CI
u
)(
O~ ______________ ~I ____________~
100 200
RNAIc DNA x 10- 3

Legend to Figure 2: Titration of RNA transcribed from chromatin
with E. Coli polymerase, against globin eDNA. Mouse foetal
liver chromatin, ( • - • ) i mouse foetal liver ohromatin
incubated without RNA polymerase, ( 0 - 0 ); mouse brain
chromatin ( x - x ). In incubations where no po4'merase was
present E. Coli tRNA was added and aliquots hYbridized in the
same proportions as for complete incubations.
the same incubation. RNA waS transcribed from foetal liver

3
chromatin incorporating highly label1 d [32p] AT? (1 - .? C:i/mM).
_~ter hybriiization as described to [ H] cDNA, the reaction was
diluted with 0.5 m1 0.2M NaCl; 0.05 M Tris-HGl pH 7.5 and
treated with 20 ~g/ml pancreatic and T1 ribonucleases at 300 C
for 2 hrs. The digest was passed through Sephadex G50 in 0.2 M
NaCl. Tho excluded material which contained less than 0.1'% of
the original 32p counts in RNA WaS centrifuged to equilibrium in
CsCl graiients according to the method of Szybalski (1968) to
separate hybrid~ed cDNA from unhybridized cDNA and [32p] RNA.
Figure 3 shows [ H] counts of cDNA appearing at CsCl densities
of 1.78 corresponding to RNA/DNA hybrid and 1.71 corresponding
to unhybridized cDNA. From the [32?] counts associated with
the hybridized cDNA it can be calculated that [32P] accounted for
60,;1 of the hybridized RNA. Similar experiments have been published
by Wilson et ale (1975&); it is also concluded that about 50;1 of the
hybridi?iD.i"'RN.A is derived from S!. ~ transcription of the
chromatin by the bacterial pol;ymerase and 50% of the hybrid arises
from contaminating endogenous RNA sequences.
It should be pointed out that these experiments suffer

trom the difficulty of recovering a relative13 small amount of
32P-label in hybrid from a huge background of unhybridized
32P-labelled RNA; however, when it can be demonstrated that
this background can be reducsd to an acceptable level the data
provide the strongest single argument for the ia ~
transcription of a specific gene.
(b) Transcription with mercurated nucleotides
The recent introduction of mercury-substituted ribonucleotide

triphosphates as substrates for RNA pol;ymerase offers in theory
a method for isolating newly synthesized ia ~ RNA frrom
endogenous RNA. In practice, chemically mercurated UTP (H~UTP)
is prepared by incubating UTP with mercuric acetate as described
by Dale ~., 1975. The acetate salt of H~UTP however cannot
be pol;ymerased in this form by E. Coli RNA polymerase, an enz;yme
sensitive to thiol modification. The meroaptoethanol derivative
on the other hand is an excellent analogy for UTP and exhibits a
Km value slightly higher than the natural substrate (Dale and Ward
1975). The inclusion of 15-20 111M mercaptoethanol in transcription
incubations containing Hg-UTP (acetate) ensures optimal
incorporation. After synthesis RNA purification is performed as
usual (details are given later) and the product passed over thiol
agarose prepared according to Cuatrecasas (1970). Only those RNA
molecules containing H~UTP are retained; only a few percent of
the U11P residues in the RNA need to be substituted for binding to
occur. After washing the column free of contaminating nucleic
acids, the purified H~RNA. is eluted with a buffer containing
mercaptoethanol. This approach has been applied to isolate
50 R. S. GILMOUR
3H eDNA 32 p RNA
epm epm
,
400~---------------------'~
1
X, I
/ x
\1
I
300~ X -lXX)
x.. x' "x \

, X-x x/./
/ x.....
200 - ! 'X-X"" 1\ ;
0
2000
X " .~
I '" '" '" ~. ..;.,.-/ \x,
100~ l ,..-/. ~
~.()-O -1000
\"S-'l-8-S- 0- x
I •
o 10
FRACTION NO.
Legend to Figure 3: Isopyonio bandillg in CsCl of the hybrid
formed between [3H] globin oDNA and [jZP] RNA transoribed from
mouse foetal liver ohromatin. COWlts in [3H], ( x --x );
cOWlts in [32P], ( • - • ); density of CsCl ( - - - - ).
A sample of [32P] RNA treated in an identical tashion but
without hybridization to oDNA was rWl in a parallel gradient,
( 0 - 0 ).
in vitro transoripts of chromatin from oontaminating endogenous

'RNA.:-;ith
subsequent hybridization analysis of the transoripts.
(Smith and Huang, 1976; Crouse et al., 1976; Beissmann!i..!!.,
1976; Beebee and Butterworth, 1976; Towle et al., 1977).
Reoently a number of serious oritioisms of this method have
been raised. In a oareful analysis of the transoription of duck
retioulooyte chromatin by E. ooli RNA polymerase Zasloft and
Felsenfeld (1977a,b) arrived at the conolusion summarised in the
following Iaragraphs.
a) Using Hg-UTP as the sole souroe of UTP, Hg-transoripts
from duok retioulooyte chromatin purified on thiol agarose was
found to oontain 0.02% of globin speoifio sequenoes on analysis
with 3H-globin eDNA.
b) In a oontrol experiment /IX 174 DNA was transoribed with ooli

polymerase in the presenoe of varying amounts of Hg-UTP and
unsUbstituted UTP. Transoripts containing up to 25% of their
UMP, residues merourated were found to hybridise baok to the
template DNA with identioal kinetios. Fully substituted RNA
however failed to hybridise to any great extent while 50%
substituted RNA hybridised only partially. This result raises
doUbts as to whether the globin sequences deteoted in fully
merourated chromatin transoripts are due to newly synthesised
RNA. It also confirms the observation of Beebee and
Butterworth (1976) who used fully substituted Hg-UTP for the
in vitro transoription of ribosomal RNA in rat liver nucleoli.
!he isolated RNA failed to hybridise to ribosomal DNA unless
it was first demerourated.
0) Zasloff and Felsenfeld turther demonstrated that the

hybridizing globin sequenoes arose not from newly synthesized
merourated RNA but from a carryover of endogenous globin
sequenoes present in duck retioulooyte chromatin. They propose
that the baoterial polymerase transoribes endogenous globin RNA
to form a hybrid which by virtue of its merourated oomplementary
strand is isolated by thiol agarose chromatography. Several
J.iDes of evidenoe support this oonolusion.
i. Artifioial chromatin templates reoonstituted from E. ooli

RNA and erythrooyte histones were supplemented with globin mRNA
to the same proportions found in duok retioulooyte chromatin.
A substantial fraotion of the mRNA could be bound to thiol
agarose after transoription with E. coli RNA polymerase in the
presence of Hg-UTP. It has been known for some time that RNA
{>l"imed synthesis of RNA by bacterial RNA pol.ymerases oan ooour
{Krakow and Ochoa, 1963; Fox!.i..!l-, 1964; Ka.itra ~, 1967;
Kelli and Pemberton, 1972; Wilson.!i..!:!., 1975b) and in a
52 R. S. GILMOUR
separate experiment Zasloff and Felsenfeld showed that under the

conditions used for chromatin transcription, E. Coli RNA
polymerase will copy globin mRNA directly to form a hybrid.
11. The recovery of globin sequences in transcripts from duck
reticulocyte chromatin is relatively insensitive to high levels
of actinomycin D. Even at concentrations sufficient to inhibit
95;'; of the total RNA synthesis, less than a 30';6 reduction in the
levels of globin sequences is observed. It is known that the
priming of RNA with single stranded RNA primers is relatively
insensitive to actinomycin D inhibition (Fox ~., 1964).
212. Yilien Hg-transcripts from duck reticulocyte chromatin are
heated to dissociate hybrids prior to thiol agarose chromatography,
over 90~~ of the Hg-RNA normally bound in the absence of heating is
still retained on the column. However the level of globin
sequences recovered in the bound fraction dropped to 5% of that
found with unheated RNA. This suggests that Hg-transcripts are
contaminated with unmercurated globin sequences in hybrid form.
This conclusion was further confirmed using anti-strand globin
cDNA as a probe. While little hybridization was observed
between heated Hg-transcripts and cDNA efficient annealing to
anti-strand cDNA occurred. The hybrids could also be bound to
thiol agarose in ~. Recently similar artefacts due to mRNA
copying by polymerase have also been reported by Giese.cke ~.
(1977) during the transcription of the ovalbumin gene in hen
oviduct chromatin.
These results raise a number of questions about chromatin
transcription data that have already been published. In the
studies of Biessmann et ale (1976) with Drosophila chromatin and
Smith and Huang (1976) with myeloma cell chromatin where fully
substituted Hg-UTP was incorporated into transcripts it is
difficult to escape the conclusion that the bulk of the observed
hybridization is due to co-purified endogenous sequences. In
the case of the data of Crouse~. (1976) with chicken
reticulocyte chromatin and Towle~. (1977) with oestrogen
stimulated chicken oviduct Chromatin where partial mercury
substitution is employed it is not clear whether the results are
due to de novo transcripts or contaminating endogenous RNA
sequenoes.---Clearly the findings of Zasloff and Felsenfeld
(1977a,b) demand that more stringent controls should be included
when using mercurated nucleotides. These authors suggest that
hybridisirtg sequences should be shown (a) to be actinomycjn
sensitive during synthesis; (b) to be retained on thiol agarose
after heat treatment and (c) to be retained on thiol agarose
after hybridization to cDNA.
In addition to the carryover of' endogenous RNA sequences by
hybrid formation, Crouse~. (1976) and Konkel and Ingram
(1977) have shown that mercurated RNA can aggregate non-

sIJecif'ically with endogenous RNA during certam isolation
procedures and thereby carry contaminating sequences through the
thiol agarose purification step. Several procedures are
described which minimise this effect. It is also pointed out
that some studies check for non-specific absorption of
unmercurated RNA to thiol agarose and others check for non-
sl~cific interactions between mercurated and unmercUl'ated B]U. by
mixing the purified components just before thiol agarose
purification. However, controls in which mixing occurs prior
to ?~A isolation should be carried out if the major source of
non-specific contamination is to be estimated.
Recently we have made a thorough examination of the ~ ~

transcription of the globin gene in foetal liver chromatin using
mercurated UTP. The procedures used are as described earlier
with the folloWing modifications: transcription react:i.ons contain
15 111M mercaptoethanol and 0.6 Dr UTF/O.2 mM Hg-UTP. Af'ter
deproteinisation the reaction mixture is passed through Sephadex
G-50 in CB (1% SDS; 0.2 m NaCl; 10 mM Tris: Hel pH 7.5) and
the excl.uded material precipitated with 2.5 volumes ethanol,
16 hrs. at _20°C. Felleted material is taken up in 1 ml 10 mM
Tris pH 7.5 and heated at 1000 C for 7 mins. After rapid cooling
an equal volume of 2 x CB and the sample run into a 3 ml column
of thiol sepharose 6B (8 fJ.mole thiol gpo per ml) previously
washed with CB + 0.3 K mercaptoethanol and re-equilibrated with
CB. The material is held in the column for 30 mins at room
temperature and then the cclumn washed with 100 mls CB. The
Hg-transcript is eluted with CB + 0.3 M mercaptoethanol,
precipitated with ethanol and finally desalted by passing
through Sephadex G-SO in distilled water.
Figure 4- shows the hybridization to globin eDNA of foetal

liver transcripts containing Hg-LTP which were purified with
and without prior heat denaturation. In these and subsequent
eXI19riments, hybridizations were carried out by the kinetic
or Rot analysis. Unlike the titration analysis described
earlier where all the available globin sequences in increasing
aliquots of RNA are hybridized to completion, in this method
all points have identical inputs of cDNA and RNA and the effect
of time on the extent of the reaotion is measured. This value
called Rot is expressed by the product of Ro (initial
concentration of RNA (moles nucleotide/l) x time , t (sees).
The value of Rot at 5OJ~ hybl'idisation of cDNA, Rot·h depends on
the concentration of available globin mRNA sequences. For
pure globin mRNA hybridised to globin eDNA under the conditions
described the Rot~ is 4- x W 3 ; the concentration of globin
sequences in an unknown RNA is proportional to the experimentally
determined Rott for the sample.
S4 R. S. GILMOUR
0r-------------------------------------------~
80
••...
A
40
...
...
A
iii
III
•
II:
A
80
80
Legend to Figure 4-: Hybridization of [3H] globin cDNA to liS-

transcripts from mouse foetal liver either with ( • - • ) or
without ( 0 - 0 ) heat treatment prior to thiol agarose
purification. Also included are Hs-transcripts from adult liver
chromatin, (D -1:1) and the hybridiza.tion obtained when mouse
foetal liver chromatin is incubated without po~rase and an
equivalent amount of "transcript" as carrier Hs-t-RNA added
(- -.).
It was found that both heated and unheated Hg-RNA transcripts

hybridized to globin eDNA; the effect of heating was to reduce the
level of hybridization to some extent. However in this case
0.004-% of the RNA was still found to represent globin mRNA
sequences. This result is not in agreement with the findings
of Zasloff and Felsenfeld (1977a, b) who found that wit...~ mercurated
transcripts of duck reticulocyte chromatin prior heating ~bolished
the ability of purified transcripts to hybridize. In control
experiments with adult liver chromatin no hybridization to globin
cDNA was seen. Similarly in the absence of polymerase no
hybridization oocurs with RNA isolated from inCUbations
containing foetal liver chromatin and carrier RNA. These results
indicate that the appearance of globin specifio m.RNA sequenoes in
tissue specific, enzyme dependent and not due to carryover of
endogenous RNA by an artefact of the method. In an additional
control experiment (Table 1) adult liver ohromatin was
contaminated with purified globin mRNA. Globin sequences
subsequently appeared in the thiol agarose purified Hg-transcript.
If the transcript is heated prior to the thiol agarose step or
if the mRNA is added during the phenol extraction very little
cross-contamination ocours. While this agrees with the findings
of Zasloff and Felsenfeld that globin m.."!lliA can be partially
transcribed by the polymerase and then co-isolated with the
transcript J it does not appear that endogenous RNA (i.e. chromatin
bound RNA) behaves in the same way. This was demonstrated 3
further in experiments where Friend oells were grown in [5 - H]
uridine for 4-8 hrs. to achieve uniform labelling of endogenous
RNA. The isolated chromatin was transcribed in the presence of
Hg - UTP and the purified transcripts applied to thiol agarose.
Table 1
Conditions %cDNA hybridised
(1 ) No additions 0
(2) 20 ng globin m.RNA at 0 time 89
(3) As above, with heat treatment 2
(4-) 20 ng globin m..'UiA after incubation 3
Legend to Table 1: Transcription reactions (2 ml) containing

adult mouse liver chromatin were inoubated with Hg-UTP. In
three oases globin mRNA was also added either before incubation
or after 1 hr incubation and cooling to 4-0 0. Hg-transcripts
were purified on thiol agarose without heat denaturation, except
for sample (3) which was denatured at 1000 C for 7 mins in 10 ml4
Tris : Hel pH 7.5.
56 R. S. GILMOUR
The results show (Table 2) that while substantial OD amounts

of Hg - RNA ~ere bound to the thiol agarose it contained virtually
none of the H-label. All of the applied radioactivity was
recovered in the unbound fraction suggesting that insignificant
amounts of complementary copying or chain elongation employing
Hg - UTP had occurred. This res ult is not in accord. with the
findings of Shih.!..'tJ!:!. (1977) who found that up to 20:% of the
endogenous RNA in a mouse cell line chromatin is elongated in the
presence of Hg - UTP. These authors did not carry out a control
for aggregation artefacts, already mentioned previously. It is
also apparent that after labelling cells for 18 hrs most of the
label is in completed RNA chains and very little in nascent
chains. That 20~'~ of the RNA label is still in nascent chains
after this time seems remarkable and may be explained more simply
by non-specific aggregation.
It is concluded that in the mouse foetal liver system globin

mRNA sequences are transcribed de novo and do not arise from
endogenous RNA sequences. It IS crear
that the effect of
exogenous RNA polymerase on added globin mRNA and on endogenous
RNA in this chromatin is totally different. Indeed the results
of Zasloff and Felsenfeld (1977a,b) could equally well be
explained by a sequestration of added polymerase on cytoplasmic
mRNA contaminating the reticulocyte chromatin. Indeed the
concentration of globin m.RNA in the reticulocyte cytoplasm is
extremely high compared with that of foetal mouse liver; if this
artefact is specific for globin mRNA then even very small levels
of cytoplasmic contamination in reticulocyte chromatin could
explain the findings.
Table 2
Thiol Agarose Total [3H] cpm %of original

fraction 3H RNA in chromatin
Bound 424- 0.002

Unbound 1.1 x 10 5 61
Legend to Table 2: Friend cells were grown for three

generations (48 hrs) in Ham's F12 medium containing 0.5 ~C~ml
~H-uridine. Isolated chromatin containing 3H-endogenous RNA
was transcribed with E. Coli RNA polymerase in the presence of
Hg-UTP. Transcripts were not heat-denatured prior to thiol
agarose purification. The [3H] RNA in bound and unbound
fractions was recovered by ethanol precipitation and oounted in
Triton-Toluene scintillation fluor.
Another factor which affects the ability of E. coli polymerase

to copy added globin mRNA is the ionio conditions of the
incubation. Table III shows the fate of 125r-labelled globin
mRN..1\. after addition to transoription incubation containing Hg-UTP
and a variety of divalent cation conditions.
Zasloff and Felsenfeld (1977a,b) found that the amount of
globin mRNA oopied by the polymerase du~ing the transoription of
duck reticulocyte chromatin was unaffected by changing the divalent
cation. Giesecke (1977) also found that the amo~t of mRNA
copying was the same whether 4 lIIM 1Ig2+ or 4 111M Ilg +/1 111M )4n2+ was
used. Table 3 suggests a different effect with the foetal liver
system. It would appear that replacement of Mn2+ ions with
Mg2+ ions reduces the degree of mRNA copying. This observation
has also been made by O'Malley and co-workers (personal
communication).
Clearly chromatin transcription systems vary considerably
and not all findings for a given system can be generalised. Each
system must therefore be analysed individually for its endogenous
RN.A levels, tendency for mRNA copying and sensitivity to divalent
cations. Perilaps the single most ~~eful experiment for testing
a system is shown in Table 4. Foetal liver transoripts
oontaining highly labelled with 32p..ATP were isolated as already
desc~ibed. Endogenous RNA is present but unlabelled. The RNA
Table 2
%Globin m...'lliA Bound RNA synthesised
Unheated Heated (!.1g)
1. 2.5 mI4 MnG12 18.3 7.1 20
(Gilmour et ale 1975)
2. 1 111M KnC12 50.1 19.2 t1
(Axel et ale 1973)
3. 5 m14 MgCl2 8.6 7.2 20
(Astrin, 1973)
4. 1 wI1I1nGb 8.5 1.'+ 15
5 111M MgCl2
(Biessmann et al. 1976)
Legend to Table}: Foetal liver chromatin (400 flg) was incubated
with 400 j.1g E. coli polymerase and 10 ng 125 Iodinated globin mRNA
under four different sets of incubation conditions. Purified
Hg-transcripts were purified on thiol agarose one half with a heat
denaturation step, the remainder without denaturation. The
fraction of the total 1251 counts bound to the thiol agarose
indicates the degree to which mRNA transcription has taken place.
58 R. S. GILMOUR
was hybridized to an excess of [3H]-globin eDNA which bad been

chemically mercurated according to the method of Dale .!l...!.!. (1975) •
Thia modification did not alter its hybridization properties and
conferred complete retention on thiol asarose columns. The
presence of excess eDNA ensured that all globin maNA sequences
whether 32p labelled or not were bybriaized. After ribonuclease
treatment to remove ~bridized RNA as described in legend,
the Hg-cDNA was bound to thiol agarose and the amount of 32~RNA
as hybrid determined. The background for the eJC;Periment was
determined by repeating the procedure with non-mercurated globin
oDNA. The results indicate the presenoe of de novo synthesised
globin maNA sequences in the transcription prOd~ amounts
comparable to that found for transcripts purified by the Hg-UTP
technique.
(0) Fidelity of in vitro transcription
The investigation of specific gene eJC;Pression by in vitro

transcription has been criticised on the grounds that bacterial
polymerases may transcribe chromatin less accurately than
homologous polymerases. Various findings have been cited as
demonstrating this.
1. Rat liver RNA poJ..ym.erase II transcribes chromatin at a

greater rate than M. luteus RNA polymerase (Butterworth et ale
( 1971;\ • -
Table J.,.
Reactants 32~RNA in Hybrid %globin sequence
4-0 !-Ig 32p_RNA (3.7 x 107 cts.)

+ 10 ng. Hg-cDNA 1850 cts = 2 ng
4-0 J.1g3~RNA (3.7 x 107 cts)
+ 10 ng. eDNA 306 cts = 0.}4 ng 0.0008
Legend to Table 4-: Foetal 1-tver transoript&; were synthesised in

the presenoe of 32~ATP to give RNA 5.A approx 10b clm/J.1g.
cDNA or Hg eDNA (both 3H-labelled) was hybridized in an excess of
at least 2O-fold over the estimated globin IIIRNA sequences rresent.
Hybridizations were carried out in 10 1J1 0.24- m sodium phospha.te
buffer pH 6.8, 0.1% SDSj 1 d4 EDT.! at 6000 for 5 hrs. The
incubations were diluted to 0.5 ml with 0.2 )( NaCl treated with
20 J.1g pancreatic ribonuclease (heated 1000 C for 5 mins) and
passed over Sephadex GSO in CB. The voided material was then run
on to thiol agarose ( 3 mls) and washed wirth 100 mls CB. A
oorrection was made for a small loss of Hg-cDNA (10%) binding
capacity as joo.ged by [3H] recovery. Retained material was
eluted with CB + 0.3 M mercaptoethanol, preoipitated with
1.5 mls of ethanol and radioactivity estimated.
II. Eukaryotic and bacterial RNA polYmerases bind to different

sites on the template as judged by the lack of competition
between the enzymes for the same sites (Butterworth et a 1.1971).
In the case of the animal polymerase these sites are""'iii'Oi;
specific on the basis of a comparison of initiating tripbospbates
(Keshgegian et ale 1973) or the kinetics of RNA synthesis
(Keshgegian and Furth, 1972).
III. RNA transcribed from rat liver chromatin by the homologous

pol,ymerase II has a larger molecular weight than E. Coli pol3merase
transcripts (II1a.rya.nka and Gould, 1973),
While these experiments demonstrate that differences between

the enz,yme can be detected they fail to demonstrate whether one
polymerase is more accurate in selecting for the transcription
of tissue specific sequences than the other. Accurate
comparisons of animal and bacterial polymerases have been made in
the part icular cases of ribosomal and 5s RNA transcription.
Reeder (1973) found that the bacterial enz,yme transcribed both
genes aberrantly in Xenopus liver chromatin; both strands and
spacer regions of each gene are expressed and in the case of
X. mullori. x laevis hybrids where virtually none of the
mulleri z-RNA genes are expressed ia D:!:2., the polymerase
transcribes actively from this gene in the hybrid chromatin.
In further stUdies Honjo and Reeder (1974) demonstrated that both
X. laevis polymerases I and II also transcribe aberrantly z-RNA
and 5s genes from chromatin. Both strands are transcribed
althougb ia ~ transcription is asymmetric and carried out by
polymerase III. In a more recent study Barker and Roeder (1977)
examined the transcription of 5s genes in Xenopus oocyte chromatin
by exogenous polymerase I and III from X. laevis ovaries,
polymerase II from mouse plasmacytoma cells and E. Coli polymerase.
In this system only polymerase I II is capable of stimulating
specifical.q the asymmetric synthesis of 5s RNA; the other
polYmerases stimulate total RNA synthesis but do not effect a
specific stimulation of 5s RNA transcription. When polymerases I
and III and Coli poJ.ymerases are caapared on a X. la.evis DNA
template not only is the synthesis of 5s RNA symmetric but the
specificity shown by polymerase III for 5s genes in chromatin
is lost. Taken together these findings suggest that chromatin
associated proteins are required for the selective and asymmetric
transcription of 5s genes in amphibians and that a specific RNA
polymerase is required to do this. Clearly the 5s gene represents
a special case. It is subject to fine ccntrols which are not
even recognised by the homologous type I and II animal polymerases
as well as E. Coli RNA polymerase. As yet no analogous
specificity has been shown between m-RNA genes and specific RNA
polymerase II species and hence to extrapolate from the aberrant
60 R. S. GILMOUR
transcription of 5s genes by Coli polymerase to an analogous

situation with m-RNA genes is not judtified.
At present it is difficult to aka an assessment of the

overall fidelity of chromatin transcription of chromatin by
bacterial polymerase, especially in light of the findings of
Zasloff and Felsenfeld (1977,a,b). In the case of globin gene
transcription in mouse foetal liver (Gilmour !.i..!:.!-, 1975) and
rabbit marrow (Wilson~., 1975a) chromatins where S!. n2!2.
synthesis of globin mRNA sequences is demonstrated, it can be
conoluded that there is some degree of tissue specific transcription
by the bacterial enzyme. However the sequences anal,ysed represent
a minute fraction of the total RNA. The question of fidelity
can only be resolved by carrying out a more extensive &ll&lysis of
the sequenoes transcribed in vitro from chromatin and comparing
them with the sequences prese~ vivo. Batchelor and Smith
(1976) compared mouse liver nuclear"iNA and E. Coli polymerase
transcripts from mouse liver chromat:m by hybridizing in RNA
excess to labelled mouse DNA fractions of high, intermediate
and low reiteration frequencies. The results suggest that all
the sequences transcribed in vivo are also present in chromatin
transcripts; however the 1it~contained additional sequences
transcribed from specific regions of high and low reiteration
sequences of the DN~ Unfortunately in this study no attempt was
made to isolatel:!! !!9:2. synthesised RNA from endogenous sequences
and in the absence of the appropriate controls it is not possible
to ascertain how ~ of the in vivo sequences detected were
actually synthesised .!!! vitro:- -
This question has also been studied by Biessmann ~.(1976)

by transcribing Drosophila. chromatin with l£. Coli po~rase and
Hg-UTP. The purified Hg-transcripts were analysed by
hybridisation to cDNA made to the po~denylated nuclear RNA of
the tissue. The results indicate that whila most of the
sequences represented in the cDNA probe were also found in the
tr~cript some aberrant, symmetrical transcription also occurred.
In these experiments no precautions were taken to exclude carry
over of endogenous RNA. in the event of RNA copying by the
polymerase. In addition, i f this occurred, it would also give
rise to merourated 'anti sense' strands. It is therefore
diffioult to say whether the results are due to symmetrical
de novo transcription of DNA in the chromatin or due to the
'(;O-purifioation of endogenous Rl:~ and its de novo sy.nthesised
oomplementary strand. It is olear that i fcopying of
endogenous RNA is generally prevalent it will also take place
where unsubstituted nucleotides are present. This possibility
suggests an alternative explanation for reports of poor strand
selection in l:!! ~ transoripts (Reeder, 1973; Honjo and
Reeder, 1974j ~trin, 1973; Wilson ~., 1975b).
In conolusion, it has to be admitted that the possibility

of transcriptional artefacts seriously oomplicates the
assessment of the overall fidelity of chromatin transoription.
Future studies will have to consider the possible presenoe of
artefacts in the particular system under investigation,
especially where strand seleotivity is considered. While the
incorporation of mercurated nucleotides into chromatin
transcripts can largely overcome the problem of endogenous RNA
contamination providing heat denaturation is carried out, this
teohnique will not eliminate mercurated copies of endogenous RNA
in the event of RNA copying by the polymerase.
Conclusions
The presenoe of emogenous RNA sequences in chromatin has
undoubtedly been responsible for a good. deal of spurious data.
The recent discovery revealed through the use of mercurated
nucleotides that baoterial polymerase can oopy m-P~A sequences
present as a contaminant in chromatin preparations provides an
additional caveat. The results of Zaslof'f' and Felsenfeld
(1977a,b) are particularly disconcerting in this respect since
they fail to show significant transcription from chromatin in a
system where the globin gene is known to be active (Fodor am
Doty, 1977). We have presented data to show that this is not the
case with the globin gene in mouse foetal liver chromatin.
However i.t is not clear how many of the other systems under
investigati.on are prone to this criticism. Transcriptional studies
with native chromatin require a more careful characterisation to
eliminate gross artefactual effects. Also it is still not
possible to assess how accurate an approximation to the in vivo
state the present methods give. It is quite possible that----
bacterial polymerase does not respond to the fine transcriptional
controls in chromatin but rather transcribes randomly from the
specific regions of chromatin normally active in vivo because they
are structually more accessible. Indeed it h&8 proved difficult
to construct a satisfactory in vitro transcription system which is
not open to criticism becauset'iiEi'iiiOde of action of the homologous
eukaryotic RNA polymerase II is so poorly understood. If
transcription with exogenous bacterial polymerase simply reflects
the accessibility of DNA sequences then this may put a limitation
on the infol'lD8.tion these systems can provide. For example, if'
chromosomal proteins bind to active DNA sequences causing a gross
structural alteration then bacterial polymerase may be able to
distinguish active chromatin on this basis. However, if the
primary regulatory event takes place round the promotor regions
and then requires correct initiation by the polymerase before
transcription can occur, the question of pol;ym.erase source
becomes crucial.
It is also important to bear in mi.nd that transcriptional

62 R. S. GILMOUR
control in chromatin in vivo may not involve a simple on-off

mechanism. Hybridization-Btudies on total messenger
populations shows that there exists a wide variation in the
numbers of copies of different messenger RNAs. This
distribution approximates three frequenoy classes, one
representing a few genes in high abundance, a second representing
several hundreds of genes in moderate abundance and a third class
where several thousand genes are represented by only a few copies
each. (Birnie.!i..!l., 1974; Bishop!i..!l., 1974; Young~.
1976). The inferenoe from these findings is that the phenotype
of eUkaryotic cells my be determined more by the relative
abundances of ~RNA sequences than by their absolute presence or
absence. The question of whether the low abundance, high
complexity class of maNA seen in most tissues is functionally
significant or whether it represents 'leaky' repressed genes
has been considered by Gelau.!iJ!:1. (1976) •
.An important generalisation which can be drawn from these
experiments is that the total fraction of the genome expressed
as maNA in the polysomes of a particular tissue is relatively
small, probably only a few percent. Only a small portion of the
total coding potential of the genome is utilised; however, it
is not clear how many of the remaining sequences actually represent
structural genes. Transcriptional modulation could involve
therefore either activation of previously silent genes or rate
limit ing controls on genes which are more or less permanently
active. Clearly it may not be possible to make this
distinction in all cases. However both must involve selection
mechanisms which influence the activity of specific genes.
References
Allfrey, V.G. and Mirsky, A.E. (1962) Froc. Nat, Acad. SCi. U.S.
~, 1590.
Allfrey, V.G., Littau, V.C. and Mirsky, A.E. (1963) Froc. Nat.
Acad. Sci. U.S. ~, 414.
Ashburner, M. (1972) In "Developmental stUdies on giant
Chromosomes" (W. Beermann, ed.) Vol. 4, p.101. Springer
Verlag, Berlin and New York.
Astrin, S.14. (1973) Proc. Nat. Acad. Sci. U.S. 1£,2,304-.
Axel, R., Cedar, H. and Felsenfeld, G. (1973) Proc. Nat. Acad.
Soi. U.S. 12, 2029.
Barrett, T., Maryanka, D., Hamlyn, P.H. and Gould, H.J. (1974)
Froc. Nat. Acad. Sci. U.S. 11. 5057.
Batcheler, L.T. and Smith, K.D. (f976) Biochemistry 15, 3281.
Beebee, T.J.C. and Butterworth, P.H.W. (1976) Eur. J:-Biochem •
.2,2, 543.
Bekhor, I., KUng, G.M. and Bonner, J. (1969) J. Mol. Biol. ~,351.
Biessmann, H., Gjerset, R.A., Levy, W.B. and McCarthy, B.J. (1976)
Biochemistry .1.2, 4356.
Birnie, G.D., 14a.cfhail, E., Young, B.D., Getz, M.J. and Paul, J.
(1974) Cell Differentiation l, 221.
Bishop, J.O. and Freeman, R.B. (1973) Cold Spring Barb. Symp.
Quant. BioI. 2,{!, 707.
Bishop, J.O. and Rosbash, M. (1973) Nature New Biol. 241, 204.
Bishop, J.O., Morton, J.G., Rosbash, Id. and Richardson, Id. (1974)
Nature 250, 199.
Brown, I.R. and Church, R.B. (1972) Develop. Biol. £2, 73.
Butterworth, P.H.W., Cox, R.F. and Chesterton, C.J.-(1971) BUr.
J. Biochem. ~, 229.
Campo, M.S. and Bishop, J.O. (1974) J. Mol. Biol. 2£, ~9.
Church, R.B. and McCarthy, B.J. (1970) Biochim. Biophys. Acta.
122., 103.
Church, R.B. and Brown, I.R. (1972) In "Results and Problems in
Differentiation" Vol. 3 (H. Urspring, ed.) p.11 Springer
Verlag, New York.
Crouse, G.F., Fodor, E.J. and Doty, P. (1976) Froc. Nat. Acad.
Sci. U.S. 12, 15~.
Cuatrecasas, P.-r1970) J. Biol. Chem. 245, 3059.
Dale, R.M.R., Martin, E., Livingston, D.C. and Ward, D.C. (1975)
Biochemistry 14, 2447.
Dale, R.M.R. a.Di Wara., D.C. (1975) Biochemistry 14, 2458.
Davidson, E.H., Hough, B.R., Klein, VI.H. and Britten, R.J. (1975)
Cell~, 217.
Firtel, R.A. (1972) J. Idol. Biol. 66, 363.
Fodor, E.J.B. and Doty, p. (1977) B'iochim. Biophys. Res. Comm.,
11., 1478.
Fox, C.F., Robinson, W.S., Haselkorn, R. and Weiss, S.B. (1974)
J. Biol. Chem. ~, 186.
Frenster, J.H., Allfrey, V.G. and Mirsky, A.E. (1963) Froc. Nat.
Acad. Sci. U.S. ~, 1026.
Galau, G.A., Klein, W.H., Davis, M.Id., Wold, B.J., Brutten, R.J.
and Davidaon, E.H. ~ 1976) Cell 1, 487.
Giesecke, K., Sippel, A.E., NgQyen-Hu, N.C., Groner, B., Hynes,
N.E., Wurtz, T. and Schutz, G. (1977) Nuc. Acids Res. ~,3943.
Gilmour, R.S. and laul, J. (1973) Proc. Nat. Acad. Sci. U.S. 12.,
344-0.
Gilmour, R.S., Harrison, P.R., Windass, J.D., Affara, N.A. and
Rl.ul, J. (1974) Cell Differentiation l, 9.
Glisen, U.R., Glisen, M.V. and Doty, p. (1966) Frac. Nat. Acad.
Sci. U.S. 22, 285.
Goldberg, R.B., Gallau, G.A., Britten, R.J. and Davidson, E.H.
(1973) Froc. Nat. Acad. Sci. U.S. lQ, 3516.
Greenberg, J.R. and Ferry, R.P. (1971) J. Cell Biol. ~, 774.
Grouse, L., Chilton, Id.D., McCarthy, B.J. (1972) Biochemistry ll,
798.
Harris, S.E., Means, A.R., Mitchell, W.M. and O'Malley, B.W.
(1973) Proc. Nat. Acad. Sci. U.S. lQ, 3776.
Harris, S.E., Schwartz, R.J., !sai, 1l.J., Roy, A.K. and 0 'Ma.:aley,
B.W. (1975) J. Biol. Chem. £21, 524.
64 R. S. GILMOUR
Harrison, P.R., Birnie, G.D., Hell, A" Humphries, S., YOWlg, B.D.
and Faul, J. (197~) J. Kol. Biol. ~, 539.
Honjo, T. and Reeder, R.H. (297~) Biochemistry .u, 1896.
Huang, R.C. and Huang, p.C. (1969) J. Mol. Biol. ~, 365.
Hum}ilries, s., Windass, J. and Williamson, R. (197~) Cell I, 267.
Keshgegian, A.A. and FUrth, J.T. (1972) Biochem. Biophys. Res.
Commun. ~, 757.
Keshgegian, A.A., Ge.ribian, G.S. and Furth, J. (1973) Biochemistry
g, ~337.
Klein, W.H., Kurpby, W., Attardi, G., Britten, R.J. and Davidson,
E.H. (197~) Proc. Nat. Acad. Soi. U.S. 11, 785.
Konkel, D.A. and Ingram, U.M. (19n) Hue. Acids. Res. ~ 1979.
Krakow, J.S. and Ochoa, S. (1963) Proc. Nat. Acad. Sci. U.S.
~, 88.
Leder, P., Ross, J., Gielen, J., Faokman, S., Ikawa, Y., Aviv, H.,
and Swan, D. (1973) Cold Spring Harbor Symp. Quant. Biol.
i!!, 753 •
.Lewin, B. (1975) Cell~, n.
Liarakos, C.D., Rosen, J.K. and O'Kalley, B.W. (1973) Biochemistry
g, 2809.
Lodish, H.F., Firtel, R.A. and Jacobson, A. (1973) Cold Spring
Harbor SymP. Quant. Biol. i!!, 899.
llaitra, U., Nakata, Y. and Hurwitz, J. (1967) J. Biol. Chem.
~, ~908.
McCarthy, B.J. and Hoyer, B.H. (2964) Proc. Nat. Acad. Sci. U.S.
~, 915.
MacGillivray, A.J. and Rickwood, D. (197~)!e Biochemistry of
Differentiation and Development, p.301, Ed. J. Faul. Oxford.
Kedical and Technical Publishing Co.
llaryanka, D. and. Gould, H. (1973) Free. Nat. Acad. Sci.• U.S. 12.,
1161.
Melli, M. and Pemberton, R.E. (1972) Nature 236, 172.
Fanitz, R. (1972)!!! "Developmental Studies on Giant Chromosomes"
(W. Beermann, ed.) Vol. ~, p.209, Springer Verlag, Berlin and
New York.
h.rker, C.S. and Roeder, R.G. (1977) Proc. Nat. Acad. Soi. u.s.
~44-.
Faul, J. and Gilmour, R.~. (1966) J. Mol.Biol. 12, 2~2 •
.Faul, J. and Gilmour, R.S. (2968) J. 1101. Biol. ~, 305.
Reeder, R.H. (1973) J. Kol. Biol. ~, 299.
Sawada, H., Crain, W.R. and Sanders, G.F. (1972) Biochim.
m,
BioJilys. Acta 643.
Steggles, A.W., Wilson, G.N., Kantor, .J.A.( Picoiano, D.J.,
Falvey, A.K. and Anderson, W.F. (197~J Proc. ltlat. Acad. Sci.
U.S. 11, 1219.
Stein, G., stein, J., Thrall, C. and Fark, w. (1975) !!!
"Chromosomal Proteins and their role in the Regulation of
Gene lbpression" (Stein and KJ.einsmitb, eds.) Academic Press,
New York and lDndon, p.1.
Shih, T.Y., Young, H.A., rarks, W.p. and Scolnick, E.M. (1977)
Biochemistry 16, 1795.
Smith, K.D., Church, R.B. and McCarthy, B.J. (1969) Biochemistry
8, 4271-
Smith: M.M. and Huang, R.C. (1976) Proc. Nat. Acad. Sci. U.S.
ll, 775.
Spradling, A., Penman, S., Campo, M.S. and Bishop, J.O. (1974)
Celll, 23.
Spelsberg, T.C. and linilica, L.S. (1969) Biochim. Biophys. Acta
ill, 63.
Sutton, W.D. (1971) Bioohim. Biophys. Acta 240, 522.
Suzuki, Y., Gage, L.P. and. Brown, D.B. (1972)J. Mol. Biol. JSl,
637.
Szybalski, w. (1968) l!! "Methods in Enzymology", (Colowick S.P.
and Kaplan, N.O. eds.) Academic Press, New York and London,
vol. 128, p.330.
Tan, C:.S. and Miyagi, M. (1970) J. Mol. Biol. 22" 64-1.
Towle, H.C. Tsai, M.J., Tsai, S.Y. and O'Malley, B.W. (1976)
J. Biol. Chem. £21, ~713.
Weinberg, E.S., Birnstiel, M.L., Purdom, I.F. and Williamson, R.
(1972) Nature 240, 225.
Wilson, G.N., steggles, A.W., Kantor, J.A., Nienhuis, A.W. and
Anderson, W.P. (1975&) Frec. Nat. Acad. Sci. U.S. ~, 86~.
Wilson, G.N., Steggles, A.M. and Neinhuis, A.W. (1975b) Frec.
Nat. .Acad. Sci. U.S. 11:., 4815.
Young, B.D., Harrison, P.R., Gilmour, R.S., Birnie, G.D., Hell, A.,
Hwnpt'.l. ies, S. and Faul, J. (197~) J. Mol. Biol. 84., 558.
1
Young, B.D., Birnie, G.D. and Faul, J. (1976) Bioohemistry.1.2.,

2823.
Zasloff, M. and Felsenfeld, G. (1977a) Biochim. Biophys. Res.
Commun. 12., 598.
Zasloff, M. and :telsenfeld, G. (1977b) Biochemistry 1.2..w. 5135.
CIRCULAR DICHROISM OF DNA, PROTEINS AND CHROMATIN
Gerald D. Fasman
Grad. Dept. Biochemistry, Brandeis University
415 South St., Waltham, Massachusetts 02154 U.S.A.
I. INTRODUCTION
Although great strides have been made towards the elucidation

of the structure of chromatin, many details remain unanswered as
to the manner in which DNA is folded and condensed in the nucleus.
The role played by the his tones in the condensation of DNA in the
nucleosome and the higher order tertiary structure still awaits
clarification. The state of condensation of DNA plays an important
role in genetic regulation as this conformational flexibility
determines which regions of the DNA are available for transcription.
The structure of chromatin has been probed by both biochemical

and biophysical techniques. Examination of nuclease digestion
products has yielded new insights into chromatin organization (see
recent reviews (1,2)). Physical-chemical methods have also been
widely used and one of these, circular dichroism spectroscopy, has
contributed significantly to our understanding of chromatin struc-
ture. Circular dichroism (CD) refers to the phenomenon of light
interaction with an asymmetric structure, when the wavelength
of incident light is within an absorption band of the molecule.
The asymmetry of proteins and nucleic acids is strongly dependent
upon the conformation of these structures, as well as their indiv-
idual asymmetric centers. Thus CD is an extremely sensitive probe
to follow conformational transitions in both components of chroma-
tin, the nucleic acid as well as the protein components. CD is
capable of delineating between the a-helical, 8-sheet and random
structure of proteins, alterations in the DNA secondary conforma-
tion due to variations in base stacking arrangements in the double
strand, as well as any tertiary organization (e.g., super helix)
of the DNA. Extensive reviews are available concerning the
67
68 G.D.FASMAN
application of CD to conformational studies (3-6).
This review is concerned with the application of CD to study

the DNA conformation in chromatin, as well as the conformation of
the proteins found in the nucleoprotein complex. The optical ac-
tivity of DNA arises from several sources. The nucleotides are
inherently optically active due to the presence of the asymmetric
pentose moiety. The major CD contribution arises from the asym-
metry of the base-base interactions due to the double helix forma-
tion. Any alterations in the base stacking which may be caused by
association with proteins (secondary structure) or due to asymmetric
condensation of the DNA in chromatin (tertiary structure) may be
detected by CD. Thus CD spectroscopy can serve as a sensitive
monitor of changes of either secondary or tertiary structure which
are concomitant with the condensation of DNA in chromatin.
II. CIRCULAR DICHROISM OF DNA
A. Conformation of DNA: Theoretical and Experimental Spectra
The chromophoric groups in nucleic acids with absorption

bands which contribute most significantly, in the instrumentally
accessible spectral region, are the pyrimidine and purine bases.
These bases exhibit TI-TI* and n-TI* electronic transitions in the
ultraviolet region. The major TI-TI* absorption band of DNA is
located near 260 nm. In the double helix, the stacking of the
bases leads to a strong interaction between the TI-TI* transition
dipole moments. According to the theory for CD of polynucleotides,
developed by Tinoco and coworkers (7-9), the interaction between
identical electronic transitions of the stacked bases gives rise
to a CD spectrum which resembles the first derivative of the ab-
sorption band. The crossover (zero ellipticity) point of this band
occurs at the absorption maximum. The spectrum has positive and
negative bands of equal magnitude, and is termed conservative.
In addition, there will be coupling between different electronic
transitions (e.g., TI-TI* transitions in the near ultraviolet
coupling with TI-TI* transitions in the far ultraviolet); this leads
to a normal gaussian CD band (nonconservative) centered at the
absorption maximum. The observed CD band contains contributions
from both of these interactions. The signs and magnitudes of the
separate contributions depend strongly on the relative orientations
and separation distances of the base transition moments. Thus, the
arrangement of the bases in a particular nucleic acid conformation
gives rise to a specific CD spectrum.
The above-described theory has met with considerable success

in prediction of nucleic acid CD spectra, because base-base inter-
actions provide the largest spectral contribution under normal
circumstances. It is not a simple matter, however, to resolve
CIRCULAR DICHROISM OF DNA, PROTEIN AND CHROMATIN 69
observed CD spectra into those two types of contributions. There

are additional minor contributions from the n-TI* transitions and
the intrinsic nucleotide optical activity. The determination of
DNA conformation therefore usually rests on comparison of observed
spectra with spectra recorded for known conformations.
X-ray diffraction analysis of DNA fibers formed in the pres-

ence of various salts and at differing relative humidities has
detected three canonical forms termed A, B, and C (10-12). These
forms of DNA differ with respect to the angle between the base
planes and a perpendicular to the helix axis; the tilt 1S small
for Band C forms (2° and 6°, respectively) but large for A form
(20°). The B form of DNA has exactly 10 base pairs per helical
turn, A form has 11, and C form has a non-integral 9-1/3 residues
per turn. The B form exhibits a 3.37 to 3.46 A rise per base
pair; the C form rise is 3.32 Aand the A form rise is 2.55 A.
It is clear that there should be differing base-base electronic
interactions for these structures, and consequently differing
CD spectra.
Tunis-Schneider and Maestre (13) studied films of DNA under

conditions essentially identical to those employed in the X-ray
studies, and observed characteristic CD spectra for the A, B, and
C forms of DNA (Figure 1) (147).
The B form spectrum observed was essentially identical to the

CD spectrum found for DNA in low ionic strength aqueous solution
(14). The positive ellipticity band maximum occurred at 275 nm,
the negative band at 245 nm, and the crossover point at 257 nm.
In low ionic strength solution, the positive band magnitude is
approximately 8400 deg. cm2 dmole- l . The spectral shape compares
favorably with that calculated by Johnson and Tinoco (9) for B form
DNA, although the measured ellipticity is much smaller than the
calculated value (15). Wide angle X-ray scattering data has pro-
vided some evidence that the structure of DNA in low ionic strength
solution, although still of the B-family, differs slightly from the
solid-state structure. The winding angle between base pairs differs,
so that there are closer to eleven base pairs per turn (16). This
conformational change will have a small effect on the CD spectrum.
In consideration of the above findings, it is now well es-

tablished that DNA structures of the B family exhibit nearly con-
servative CD spectra due to the small tilt of the bases relative
to a line perpendicular to the helix axis.
The CD spectrum assigned ~ Tunis-Schneider and Maestre (13)

to the A form of DNA is nonconservative, with a maximum near 260
nm. According to the Johnson and Tinoco theory (9), the large tilt
of the bases in A form DNA (and also in RNA) results in this non-
70 G.D.FASMAN
conservative CD characteristic. Experimental spectra for RNA in

aqueous solution and DNA in alcoholic solutions show the A type CD
spectrum (14).
Greater controversy exists concerning the spectral properties

of C form DNA. Utilizing a salt concentration and relative humidity
conditions similar to those employed in the original C form X-ray
structure determination (11), Tunis-Schneider and Maestre (13) ob-
tained the CD spectrum shown in Figure 1. The ellipticity above
270 nm is very low, while the negative 245 nm band is comparable
to that observed for B form DNA. Similar spectra may be obtained
for DNA in aqueous salt solutions (17,18). There are difficulties
with this assignment on both theoretical and experimental grounds.
Theoretically, the small conformational difference between Band C
form DNA (C form DNA may be regarded as part of the B family by
~
I
~
A
~~
~----~----~
O~--+---~
Figure 1. Schematized representation of the CD spectra for the A,

B, and C forms. The dotted lines are drawn through the absorption
maxima. Ivanov, V.I., et al. (147). Reprinted with permission of
John Wiley and Sons, In-c-.---
CIRUCLAR DICHROISM OF DNA, PROTEIN AND CHROMATIN 71
many criteria) should not result in the observation of a noncon-

servative CD spectrum, but rather should result in a low magnitude
conservative spectrum (9). Experimentally, Brunner and Maestre (19)
found difficulty in reproducing the original C form spectrum for
LiDNA films at low relative humidities. Since the original assign-
ment of the CD spectrum for the C-form DNA depended on the similarity
in the conditions of film preparation for the X-ray studies and the
CD studies, that assignment is now open to question.
B. 'I' DNA
In the presence of a neutral polymer, such as polyethylene

oxide, and salt, the structure of DNA is altered so that a unique
CD spectrum of large magnitude is observed. Lerman and coworkers
(20,21) have shown that under these conditions high molecular weight
DNA becomes intramolecularly condensed to an ordered tertiary struc-
ture, as a result of excluded volume effects. As a critical con-
centration of Eolymer and ~alt are required for this induced transi-
tion, they termed the structure PSI or 'I'-DNA. A series of CD spec-
tra of 'I'-DNA are seen in Figure 2, produced by various concentra-
tions of polyethylene oxide, after various times of mixing. Extremely
large negative ellipticity bands are observed. The Johnson and
Tinoco theory (9) for polynucleotide CD cannot account for this
spectrum. Lerman (20-22) believes this spectrum derives from a
super folding of the DNA and may be comparable to the condensed form
of DNA in the chromosome. X-ray scattering studies (22) have demon-
strated that 'I'-DNA is still in the B-form. Thus, it is possible
to condense DNA into an asymmetric tertiary structure, without de-
stroying the basic B-form structure of the Watson-Crick double
helix, and obtain a new type of CD spectrum. By analogy to studies
of cholesteric liquid crystals (23-27), the unusual CD spectrum
may be attributed to long range anisotropy in the arrangement of
the stacked DNA bases within the condensed structure. Destruction
of this ordered structure, by heating to a temperature below that
at which DNA denatures, results in a reversion of the CD spectrum
to the typical B type (28).
C. Salt Effects
The CD spectrum of DNA in aqueous solution can be altered by

the addition of salts. Permogorov et al. (29) and Fric and Sponar
(30) first demonstrated that in 2M NaCl the CD spectrum of DNA
shows a decrease and slight red shift in the positive band near
275 nm. Extensive investigations by other authors (17,31,32,147)
have shown that a number of monovalent and divalent salts produce
this effect (Figure 3). In all cases, the negative band near 245
nm shows relatively little change, while the positive band is
72 G. D. FASMAN
greatly reduced and may become negative. The greatest effect docu-
mented is that in 13 molal LiCl; DNA exhibits two large negative
CD bands centered near 285 nm and 240 nm (18,33).
One explanation of the salt effects relies upon the published

reference spectrum of Tunis-Schneider and Maestre (13) for C form
DNA. The DNA spectrum in 5.5M NH4Cl, for example, is nearly identi-
cal to the C form reference spectrum (Figure 3). Salt addition is
therefore postulated to cause a B~ conformational change in DNA.
The increase in winding angle between adjacent base pairs, which
would accompany this change, may be considered to arise from ef-
fective cation neutralization of the DNA phosphates (17,31,34,147
I. ••
... .. ... e•
O~~~~--------~.~-----------=~~~"~~HH~
•••• • •
•• •• •
••••
-30
••
> • •
::J
•• ••
--
C7
v ••••••
-0 •
'"E
u
J.
v
•
4.)
M •
•
4.)
"0
..,
I
• 5
0
X
•
~ -90
• •
-• •
-12
• -. .. --- •
280 300 320

" (nm J
Figure 2. Representative circular dichroism spectra of T7 DNA in

different concentrations of polyethylene oxide (PEO) and at different
times after mixing. The circles represent measured spectra in PEO
concentrations, and at the time after mixing as follows: 1, 80.9
mg/ml, 0.7 hr; 2, 98.1 mg/ml, 0.7 hr; 3, 126.5 mg/ml, 0.7 hr; 4,
90.1 mg/ml, 0.7 hr; 5, 126.5 mg/ml, 48 hr. DNA = 22 x 10-6M, Na+
= 0.26M. Jordan, CF., et al. (21). Reprinted with permission of
Macmillian Journals, Ltd.
35). The increase in winding angle need not stop at the canonical
C form, and thus conformations more tightly wound could occur and
give rise to further CD changes as was the case for high LiCl con-
centrations.
However, there is some evidence which is not consistent with

the above explanation. Wide angle X-ray scattering data obtained
for DNA in 6.0M LiCl (22), which exhibits the C-type CD spectrum,
have been interpreted as being consistent only with the B-form
DNA structure. As previously described (Section II, A), the refer-
ence spectrum for C-DNA may be questioned on several grounds. There
is, however, some evidence that the winding angle between bases in
NoCi
-2
-3
1 LJ.f,
Figure 3. Influence of the cations upon the CD spectrum of the

water solution of DNA. The DNA in concentration of lO-4M (P0 4 ) was
in the form of Na salt and initially contained 2.5 x lO-4M NaCl.
The curves are labelled with the molar concentration of the cations.
Ivanov, V.I., et al. (147). Reprinted with permission of John Wiley
and Sons, Inc.
74 G.D.FASMAN
DNA can vary with ionic strength (36) and yet remain in the B
family of conformations.
An alternate explanation for the effect of salt on the DNA CD

spectrum may be offered by analogy to the ~-DNA studies. Cation
binding by DNA in solution may allow an intramolecular condensation
into an ordered structure. Wolf et al. (33) have shown that upon
increasing the LiCl concentration, the sedimentation coefficient of
DNA becomes larger. At very high LiCl concentrations intermolecular
aggregation occurs, as well as intramolecular condensation. The
existence of a compact ordered form could give rise to a liquid
crystal-type CD contribution. In this regard, it is interesting
to note that the CD spectra for ~-DNA (21), DNA in 13 molal LiCl
solution (33), and LiDNA films at low relative humidities (19) are
quite similar in band shape and position, despite some light scat-
tering contributions. At lower salt concentrations, the CD spectra
of DNA may be resolved into contributions from the B-form DNA secon-
dary structure and the ~-DNA tertiary ordering.
D. Solvent Effects
The conformation of DNA can be perturbed by organic solvents

causing either an increase or decrease in the ellipticity, depend-
ing on solvent. Ethylene glycol or 95% methanol induce changes in
the CD spectral character, similar to those observed in aqueous
salt solutions, namely, a decrease in the ellipticity at 275 nm,
[6]Z75 (37-39). The effects of salt and these organic solvents are
addltlve (40,147). Upon increasing the concentrations of perturbing
solvent, the CD spectra tend to approach the C-form DNA reference
spectrum of Tunis-Schneider and Maestre (13). It is not known
whether these spectra contain any ~-like contributions.
Ethanol at low concentrations also produces the same effect on

the CD spectrum of DNA. As the ethanol content of the solvent is
increased from 0 to approximately 66% (v/v with water), the spectrum
changes from the B-type to the C-type (39), the magnitude being de-
pendent on the base composition of the DNA. Between 66% and 78%
ethanol a significant increase in the 275 nm ellipticity occurs in
the CD spectrum, producing spectra similar to the A-form DNA refer-
ence spectrum (147), which reaches a maximum value at 80% ethanol
(14). These effects are reversible upon addition of water. Once
again, however, the interpretation of such CD data, in terms of
DNA conformations, is not straightforward. Girod et al. (39) have
demonstrated that the DNA in 80% ethanol solution is condensed.
Furthermore, DNA has been crystallized from ethanol, and has been
found to be in the B conformation (41). The observed CD spectra
are thus likely to be complicated by light scattering and tertiary
ordering (liquid crystal) contributions, so that the A type spectra
obtained in this manner may not be simply due to the A-form con-
formation.
E. Temperature Effects
The secondary structure of DNA can be destroyed by heating,

and this helix-to-coil transition causes a loss of hypochromism in
the absorption band at 260 nm, as the bases unstack [Marmur and
Doty (42)]. Similarly, base unstacking is reflected in the CD
spectrum as a change in band intensities. Both the absorption and
CD changes are cooperative. The melting temperature of DNA is a
function of solvent and ionic strength, the helical form being
stabilized at higher salt concentrations (42).
The CD changes of DNA during heating are complex. Brahms and

Mommaerts (14) demonstrated that below the melting temperature, the
intensity of the 275 nm peak increases. This effect has been termed
a "premelt", and may correspond to an unwinding and/or base tilting
in the DNA conformation prior to denaturation (43). At the melting
temperature, the CD band intensity is decreased as base-base inter-
actions are destroyed (14). Denatured DNA retains the CD spectral
properties of the intrinsically optically active nucleotides, as
well as additional contributions from local interactions.
Destruction of the DNA tertiary structure (e.g., ~-DNA) by

heating can also lead to cooperative CD changes. The condensed
form of T2 phage DNA in 80% ethanol exhibits two CD melting bands
(39). The lower temperature band corresponds to destruction of
tertiary order, and the higher temperature band to the helix+coil
secondary structural change.
F. Interpretation of CD Spectra for Unknown DNA Conformations
It is evident from the discussion presented in the preceding

sections that the interpretation of a CD spectrum obtained for DNA
in an unusual conformation is not straightforward, as the spectrum
may contain both secondary and tertiary structure contributions.
In general, the latter contributions are to be suspected when band
magnitudes are extremely large and nonconservative, DNA is near
its solubility limit, or the thermal denaturation profile (CD) ex-
hibi£s more than one cooperative transition. The absence of these
indications, however, does not preclude a smaller tertiary structure
contribution, as was discussed for, the CD of DNA in salt solutions.
Light scattering artifacts provide an additional complication

to interpretation of spectra. Differential scattering of left and
right circularly polarized light from optically active particles
can result in band position and intensity changes (44-47). The
differential scattering contribution can, however, be estimated and
76 G.D.FASMAN
at least partially corrected by both theoretical (45, 148) and ex-

perimental means (48-50, 149). In scattering solutions there is also
an absorption flattening effect, not correctable by experimental
means, which generally results in reduced band ellipticities. The
sum of these two effects can provide a significant perturbation of
the CD spectrum for both DNA and proteins (47) ans should therefore
be investigated whenever absorption spectra of such solutions indi-
cate light scattering to be present.
The greatest utility of CD in conformational studies lies in

its extreme sensitivity to conformational changes. Thus when a
small change occurs in the base-base interaction geometry of DNA,
there can be a significant CD spectral alteration. CD may there-
fore be employed to monitor small secondary and tertiary conforma-
tional changes in DNA which may be observed with great difficulty
by X-ray scattering or other physical techniques.
III. CIRCULAR DICHROISM OF HISTONES
A. CD of Known Protein Conformations
The ultraviolet CD spectra of proteins arise from the n-n*

and n-n* electronic transitions of the peptide bond chromophores.
As in the case for polynucleotides, interactions between the transi-
tion dipole moments of these chromophores, in ordered conformations,
can lead to unique spectral effects (7).
Reference spectra for various protein backbone conformations

have been obtained from studies of synthetic polypeptides under
conditions shown to yield particular homogeneous conformations; e.g.,
100% a-helix. The three main conformations to be considered are
the a-helix, the 8-pleated sheet, and the random coil (or irregular
conformation). The CD reference spectra for these three conforma-
tions are seen in Figure 4 [Greenfield and Fasman (150)]. Curve
1, that observed for the a-helix, has two negative bands, at 222 and
208 nm (molar ellipticities are -35,700 and -32,600, respectively)
and a positive band at 191 nm (76,900). Curve 2 is that for the
8-pleated sheet, which has a negative band at 217 nm (-18,400) and
a positive band at 195 nm (31,900). The random coil conformation,
curve 3, shows a small peak at 217 nm (4,600) and a large trough
at 197 nm (-41,900). Thus, the basic conformations of proteins
are easily distinguished by their CD spectra. Other standard
curves (e.g., derived from proteins whose X-ray structure has been
determined) have been employed without significant improvement (51).
The 8-turn conformation, which often occurs at the point of

chain reversal at the surface of globular proteins, will also have
a CD contribution similar to that of the 8-pleated sheet. The
70
60
Curve
I 100 % a Helix
50 2 100% ~
3 100 % Random Chain
40
~
0
E
CJ
CD 30
0
......
N
E 20
CJ
CD
CD
~
0'
CD 10
0
If)
I
0 0
)(
~
Q)
........
-10
-20
-30
-40
190 200 210 220 230 240 250
X in mp.
Figure 4. Circular dichroism spectra of po1y-L-1ysine in the a, S

and random conformation. Greenfield, N., and Fasman, G.D. (150).
Reprinted with permission of the American Chemical Society.
78 G.D.FASMAN
exact parameters of the B-turn CD spectrum have not yet been deter-
mined.
B. Salt Effects on Histone Conformation
The CD spectra of histones dissolved in water or very low

ionic strength solution are essentially identical to the reference
spectrum for a random coil conformation (52-58,65). The spectrum ob-
served for H4 in 0.01 M Tris, pH 7.0, is that associated with the
random conformation. In 0.14 M NaF, the conformation of H4 is
markedly changed such that the CD spectrum can be resolved into
contributions of 24% a-helix, 36% B, and 40% random coil (53).
Similar changes due to variation in ionic strength can be observed
for the other histones. Thus in the presence of salts the histones
can adopt highly ordered secondary structures. This conclusion has
been confirmed by measurement of tyrosine fluorescence anisotropy,
which also indicates substantially increased structural rigidity
with increasing salt content (54-58). A review of literature on
this topic has recently been published by Fasman et al. (59).
Isenberg and coworkers (54-58) have thoroughly investigated

the salt-dependent histone conformational changes. Upon addition
of salt to solutions of low protein concentration, H3 and H4 undergo
both fast and slow conformational alterations. Only the fast con-
formational change can be observed for H2A and H2B. The fast change
was shown by CD studies to consist of a small increase in a-helical
structure. The slow step (the rate of which depends on histone
concentration, salt concentration, type of salt, and temperature)
consists of B-pleated sheet formation, concomitant with histone
aggregation.
When mixtures of the histones were similarly studied [D'Anna

and Isenberg (60-62)], significant histone-histone interactions
were observed. As noted above, H4 undergoes both fast and slow
conformational changes upon addition of salt, whereas H2B does not
exhibit a slow change. However, a mixture of these two his tones
did not yield intermediate CD changes, thus indicating interaction
between species. The CD curve for a 1:1 mixture of H2B and H4
shows no time-dependent slow conformational change. Similarly,
the CD spectrum of mixtures of H3 and H4 shows no slow changes, al-
though each of these his tones individually exhibits slow conforma-
tional changes. Sedimentation studies have demonstrated that these
different effects arise from specific histone-histone complexing
which inhibits histone self-aggregation. In the H3-H4 case, the
complex is a tetramer; for H2A-H2B and H2B-H4 the complexes are
dimers (61,63).
Such CD studies have demonstrated that specific histone-histone

complexes can form in aqueous salt solutions. Moreover, it has been
[NaCI]
-
0.01
-4 0.05
w 0.5
..J
0 2.0
~
0
LU
0
...... -8
N
~
0
I
(!)
-
W
0
-12
I t)
'0
)(
~ -16
-20
200 210 220 230 240 250 260
WAVELENGTH (NM)
Figure 5. Circular dichroism spectra of chicken erythrocyte his-
tones in solutions of 1 mM Tris-HCl (pH 7.5) and several concentra-
tions of NaCl. Bidney, D.L., and Reeck, C.R. (64). Reprinted with
permission of the American Chemical Society.
80 G.D.FASMAN
shown that the fast conformational change results in complexes with

substantially larger negative ellipticities at 220 nm than found
present for the individual his tones (60). The CD spectra indicate
that upon complexation a greater a-helical content resulted. The
a-helical content is dependent on salt concentration, higher salt
yielding higher a-helical contents. This pattern [Figure 5, Bidney
and Reeck (64)] is also observed for mixtures of total histone ex-
tracts (30,64). The CD spectrum, below 240 nm, in 2.0 M NaCl for
the solution of all the histones, [as well as the histone core
tetramer (66)] is quite similar to that obtained for chromatin at
low ionic strength. It appears that the DNA-histone complex inter-
action results in a salt-like neutralization effect on his tones so
that a compact conformation is adopted. Independent evidence (67,68)
indicates that the histone conformation in chromatin is like that
adopted by the four core histone mixture in 2M NaCl.
IV. CIRCULAR DICHROISM OF HIGH MOLECULAR WEIGHT CHROMATIN
A. Chromatin at Low Ionic Strength
It is now well established [recently reviewed by Kornberg (1)]

that chromatin exists as a repeating unit, in which specific histone
complexes, containing two each of the histones H4, H3, H2A, and H2B,
bind and condense a discrete length (about 140 base pairs) of DNA.
The repeating units are connected by a variable shorted length of
DNA which may be condensed or extended depending on solvent condi-
tions and interaction with HI (H5 in avian erythrocytes) (69,70).
The interaction of DNA with his tones could possibly result in a
conformational change in the DNA and/or histones which should be
amenable to investigation by CD.
The CD spectrum of calf thymus chromatin is shown in Figure 6

(71). Below 230 nm, chromosomal proteins provide the greatest spec-
tral contribution. The band shapes and magnitudes in this region
indicate substantial (~40%) a-helical character in the proteins.
Recent estimates of conformation gave 50% a, 5% Sand 51% random
(66). Above 250 nm, the protein contribution is negligible and
consequently the CD bands in this region provide information about
the DNA conformation. Whereas free DNA in solution exhibits a
positive band at 275 nm with [8]275 = 8,500 deg. cm 2 dmole- l , that
band is red-shifted and reduced to 4,000 for the protein-bound DNA
of chromatin. The altered DNA spectrum is a direct consequence of
protein binding, since dissociation of the proteins by addition of
sodium dodecylsulfate results in the observation of a normal B-form
DNA spectrum (71). This CD spectrum of chromatin has been amply
confirmed (29,30,72-74). It is thus apparent that the secondary
or tertiary ordering (or both) of DNA in chromatin is altered rela-
tive to B-DNA.
10
8. /"1.'\
I "
I \\
, I "
,
6
" \
I I '
,\ 1,( I "\
4 I ' \
, ~ DNA
.,
..... ! \ I \
2
, I \ ! \
o
E
'u.,
'III \ I "~
't)
......
o -\-j----\\ '
I~---------~~-
(II
E " \ I
-2
\ i
u
-4
\ i
If)
\ i
I
o \ i 10
-6 \ i o
\ , ii
-8 \ J;
Chromatin +SDS
,-'i'
-10
~/'
-12
200 240 280

-14
200 220 240 280 300 320
Figure 6. CD spectra of calf thymus chromatin, chromatin in 0.1%

sodium dodecylsulfate, and DNA. (---) calf thymus chromatin; ( .• • )
chromatin in the presence of 0.1% (w/w) sodium dodecylsulfate; (---)
pure calf thymus DNA. Solutions in 0.14M NaF, O.OlM Tris-HC1, pH
8.0. Cone. DNA = 1.6 x 10-4M (P04)' Mean residue ellipticity is
based on DNA residue concentration. Shih, T.Y., and Fasman, G.D.
(71). Reprinted with permission of Academic Press.
82 G.D.FASMAN
Calf thymus chromatin is a relatively simple system for CD

studies because it contains very little nonhistone protein or RNA.
Histones have few aromatic amino acids which could provide a CD
contribution above 230 nm. By contrast, the nonhistone proteins,
as a group, have a significant aromatic amino acid content. RNA,
as previously described (Section II, A), has a CD spectrum similar
to that of A-DNA. Thus the presence of nonhistone proteins and
RNA must be considered for the interpretation of the CD spectra ob-
tained for various chromatin samples from active tissues (75-78).
Thus it has been found that the CD spectra of chromatin samples
isolated from cells at different stages in the cell cycle are dis-
tinguishable; these differences are in part due to changes in the
nonhistone protein and RNA contents [reviewed by Baserga and Nico-
lini (79) and Baserga in a Chapter in this volumeJ. The discussion
herein will be limited to studies of chromatin samples for which
the non-histone protein and RNA contributions are negligible, thus
allowing a more detailed understanding of the DNA and histone con-
formations.
What can be said about the conformation of DNA in chromatin?

Hanlon and coworkers (18, 80) have stated that a portion of the DNA
adopts the C canonical conformation, and the remainder is of the B
form. The CD spectra of chromatin isolated by two different pro-
cedures have been examined by Hanlon et al. (80) and compared to
the Band C-DNA reference spectra. It is, in fact, possible to
match the observed spectra for chromatin by a combination of the B
and C-DNA spectra. The C-DNA reference spectrum, however, corres-
ponds to no adequately established DNA structure (see Sections II,
A and II, C). In addition, the existence of C-DNA in chromatin is
not supported by X-ray scattering studies (16) which show that DNA
retains the B conformation.
If the secondary structure of DNA is not altered significantly

by histone complexation in chromatin, the CD change may be due to
the existence of an ordered tertiary arrangement of the condensed
DNA. Support for this hypothesis may be found in the calculated
difference spectrum between free DNA and chromatin DNA [Figure 7,
Shih and Lake (8l)J. The difference spectrum shows a negative
band centered at 275 nm, similar to the spectrum of ~-DNA. The
position of this band accounts for the lower [8J275/[8J284 ratio
in chromatin relative to free DNA. If this explanation 1S pursued,
any change in the extent of DNA condensation in chromatin will be
reflected in a change of the 275 nm ellipticity. Furthermore, the
changes at nearby wavelengths (e.g., the peak at 284 nm) will be
less extensive. Data which supports this explanation is discussed
in the following section.
10
0~~~-----~r---------------------~--------------------------=~------------1
'I
I
W
-l -10
I
o --~
~ /~
L)
w I
N
!2 -20 I
E
u I
I
CI I O~------------~r-----~
~ -30 I ':'Q
.,
I
o j ~-2
~-40 1/ ~
I: <l
1---..., !
Cb
'--' -4
-50 ...... / I "

, .... _--'
I
, -6
,I
I
I
260 280 300
-60 '.../
). ,nm
200 220 240 260 280 300 320

). ,nm
Figure 7. Circular dichroism spectra of metaphase and interphase
chromatin. (-- ----), metaphase chromatin; (---), interphase
chromatin. Ellipticity, [e], calculated on the basis of DNA residge
confiNRtration. Difference spectra shown in the inset (~[e] = [e]C r
[e] ) were calculated by subtracting CD of pure DNA (---) measured
under similar conditions. Solvent was O.OlM NaCl, O.OOIM Tris'HCl,
pH 7.6. Shih, T.Y., and Lake, R.S. (81). Reprinted with permission
of the American Chemical Society.
84 G. D. FASMAN
B. Tertiary Order in Chromatin
1. Thermal Denaturation of Chromatin. When DNA is condensed

into an ordered tertiary structure (see Section II, E), heating to
a temperature below the melting temperature of the DNA double helix
can sometimes lead to a reversion of the CD spectrum to that char-
acteristic of the secondary structure alone. Thus, the tertiary
structure can be destroyed at a lower temperature than the secondary
structure. An analogous situation is found to occur in chromatin
temperature melts. The DNA in chromatin melts out at a much higher
temperature than free DNA indicating that the double helix is
stabilized by bound protein. Chromatin appears to have two tempera-
ture transitions. At the first transition, the ellipticity at 280
nm shows a large increase (Figure 8), followed by a decrease at the
second temperature transition (73,82,83). The CD ellipticity at
227 nm, [8]227' observed at the first transition shows a reduction
in the histone a-helical content, simultaneously with an apparent
meltout of the tertiary order of the DNA (82). This is followed by
the meltout of the B form DNA (associated with histones) seen in
the second transition. Thus it would appear that the lower the
ellipticity at 275 nm, the higher the degree of tertiary structure
exists.
2. Shearing Effects on Chromatin. This topic will only be

briefly discussed. Very high molecular weight chromatin is not
easily solubilized for spectroscopic study. Since light scattering
contributions make interpretation of CD spectra difficult, many
laboratories have employed shearing methods to reduce the molecular
weight of DNA in chromatin and thus obtain soluble preparations.
If, however, the degree of DNA condensation in chromatin has an ef-
fect on the observed CD spectrum, then methods of isolation that
minimize the disruptive effects of shear forces should yield more
highly condensed chromatin, with a new characteristic CD spectrum
of lower magnitude. Thus if the negative ~-like spectral contribu-
tion centered at 275 nm plays a significant role, its contribution
should be larger for a more highly condensed and ordered chromatin
preparation. This explanation is consistent with data obtained by
several groups (49,75,80,84). When the CD light scattering con-
tribution is corrected for in a condensed chromatin sample, the CD
spectrum shows a positive band centered near 280 nm, but with a
nearly two-fold reduction in ellipticity relative to sheared
chromatin (49).
3. Specific Salt Effects on Chromatin. At concentrations

far below those necessary to cause histone dissociation, salts can
effect the CD spectrum of chromatin. It has been well established
that the extent of chromatin condensation as viewed by the electron
microscope is dependent on the concentration and type 01 salt pres-
ent [see for example Brasch (85); Finch and Klug (86)]. In general,
lower salt concentrations result in more extended chromatin. If
the positive CD ellipticity at 275 nm is a sensitive measure of
tertiary order in DNA, then a higher ellipticity is expected at

lower ionic strength. This has found to be true by a CD study which
was correlated with an electron microscope study to reveal the ac-
tual state of condensation (151). The electron micrographs, at
various concentrations of ammonium acetate, showed that the struc-
ture of chromatin is considerably more condensed at the higher salt
concentration. A parallel study of the CD properties, under the
same conditions, showed that the ellipticity at 278 nm was reduced
approximately 28% at the highest salt concentration, relative to
the lowest. Other reports, in which hydrodynamic measurements were
made to determine the extent of chromatin condensation, confirmed
the lowered CD ellipticity observed in solvents favoring a compact
structure for the chromatin (84,87).
8000 ",
.,'
6500
II)
I
0
:Ie
0
III
5000
N
,--,
Cb
L-I
3500
20 40 60 80 100
Temperature °C
Figure 8. Circular dichroism ellipticity, [9]280' versus tempera-

ture for chromatin (---); mononucleosomes (-- - -- -); and DNA ( •.• )
in 2.5 x 10-4M EDTA, pH 7.0. Mandel, R., and Fasman, G.D. (83).
Reprinted with permission of Information Retrieval Ltd.
86 G. D. FASMAN
4. Supercoiling. A supercoiled arrangement of DNA need not

lead to a negative ~-like spectral contribution. Maestre and Wang
(88) observed a slight increase of ellipticity at 280 nm, relative
to B-DNA, in closed circular supercoiled DNA from several sources.
The folded chromosome of E. coli has been reported (89) to exhibit
CD properties consistent with contributions only from RNA and simple
B-DNA. By contrast, intact chromosomal fibers from equine sperma-
tozoa do show a ~ DNA CD spectrum (90).
C. Perturbation of Chromatin Structure
1. Effect of Trypsin. Digestion of chromatin with trypsin

results in the cleavage of peptide bonds adjacent to lysyl and ar-
ginyl residues in the his tones which are exposed to the enzyme.
Limited tryptic digests result in cleavage of bonds adjacent to
basic residues which are not involved in ion-pair bonds with the
DNA phosphates (91). Thus protein segments not bound or only weakly
associated with the DNA backbone are digested. This results in
disruption of histone-histone interactions and an unfolding of the
DNA to an extended conformation (91). The CD changes with varying
levels of trypsin digestion (92) are shown in Figure 9. This data
illustrates the dependence of the depression of the positive CD
band of DNA on the maintenance of histone-histone interactions.
Unfolding of the condensed DNA structure results in a higher posi-
tive ellipticity.
2. Effect of Urea. Histone-histone interactions are disrupted

by the presence of urea. This disruption is a direct result of
histone denaturation, and is reflected in a loss of the a-helix
contribution as determined by the CD spectrum. There is a concomi-
tant increase in the ellipticity of the positive DNA CD band (73,
74,81,87,93). Hydrodynamic studies show that chromatin DNA is
more extended in urea solutions (87) in agreement with the above
observations.
3. Effect of Chemical Crosslinking. Formaldehyde, which can

form both histone-histone and histone-DNA linkages, causes a marked
decrease in the positive CD above 260 nm as a function of formalde-
hyde concentration (94). This reduction is most noticeable at 275
nm. Chemical crosslinking thus appears to enhance the ~-like CD
contribution in chromatin, probably by tightening the condensed
DNA structure.
4. Effect of Histone Chemical Modification. The lysine resi-

dues in the histones may be chemically altered by specific reagents
which may modify the DNA-histone interactions, leading to changes
in the CD spectrum. Mild treatment of chromatin with acetic an-
hydride results in the acetylation of approximately 25% of all
protein lysine residues (95). The extent of histone acetylation
.... -,
." .
:.
"
~
:.
..
2.0 "
~
1.0
:
...'...
"
. .
. ..
w 0
<l ··
·. 2.0
...........
o~--~~--------~~------~
-1.0 ··
··
..
-2.0
··
-3.0
-4.0
210 250 290 330

A(nm)
Figure 9. CD spectra of trypsin-treated nucleohistones. Trypsin
concentration in ~g/ml is 0 (---) , 10 (-- -- --), and 40 (_._);
DNA ("'). Contribution of trypsin to CD below 240 nm has been
corrected. It is noted that the total histone content, digested or
not, is the same for these samples with varied levels of trypsin.
Li, H.J. et al. (92). Reprinted with permission of John Wiley and
Sons, Inc.
88 G.D.FASMAN
is much lower than the value for total chromosomal protein, as

nonhistone proteins react more strongly. However, the CD spectrum
above 250 nm was reported to be unchanged by this low level of
modification (95).
Extensive modification of the histone lysine residues may be

achieved with ethyl acetimidate (96). The altered amino acids re-
tain a positive charge, and so may still interact with the DNA
phosphate groups. Up to 90% of the histone lysines may be so altered
without resulting in significant CD changes (96).
In both of the cases described above, the basic condensed

structure of DNA was shown, by other techniques, to be retained in
the chemically treated chromatin, which is consistent with the fact
that the CD spectra are not significantly changed.
5. Effect of Histone Dissociation. Essentially complete

dissociation of his tones from DNA by detergents (53) or high salt
concentration (29,30,74) results in the observation of a chromatin
CD spectrum for which the DNA contribution above 250 nm is the same
as that for free DNA in the same solvent. At intermediate salt
concentrations (0.6M to 1.6M NaCl) it is possible to selectively
dissociate particular histones from chromatin, and thus determine
the extent to which those his tones are responsible for the conden-
sation of DNA (97).
HI is effectively dissociated from chromatin with 0.6M NaCl

(73). This histone is thought to bind and at least partially con-
dense the spacer region of DNA between nucleosomes (69,98). As
shown in Figure 10, removal of HI from chromatin results in an in-
creased ellipticity due to the DNA, in accord with the expected
results for DNA unfolding (92). Several reports confirm this ob-
servation (72,87,99,100), but a decrease or lack of change in the
positive ellipticity of chromatin upon HI removal have also been
reported (73,101). Results obtained in the author's laboratory
support the observation of increased ellipticity in HI and H5-
depleted chicken erythrocyte chromatin.
H2A and H2B are removed from chromatin at NaCl concentrations

of 0.6M to 1.2M NaCl (97), whereas H3 and H4 are removed between
1.2M and 2.0M NaCl. Published reports differ in conclusions re-
garding the properties of HI, H2A, H2B-depleted chromatin. Thus
it has been observed that chromatin depleted of all his tones except
H3 and H4 exhibits a positive CD band of DNA similar to free DNA
(72,73); or that these histones do exert some conformation-perturb-
ing effect on DNA (87,92) as shown in Figure 10.
Removal of the different histone classes from chromatin thus

destroys the difference in CD properties between chromatin DNA and
free DNA. Thermal denaturation studies show a concomitant loss of
the higher temperature stabilization of the DNA (82). With increas-

ing temperature, free DNA shows one downward CD transition, centered
at about 44°C, whereas intact chromatin DNA shows an increase in
ellipticity around 60°C and a reduction in ellipticity transition
centered near 72°C. As the histones are progressively removed, by
extraction with higher NaCl concentrations, the higher temperature
transitions decrease in size and after the 2.0M NaCl extraction
the CD plot is similar to that of native DNA. Thus, the importance
of histone:DNA and histone:histone interactions is demonstrated for
maintenance of the chromatin structural stability .
...........
. ..: ... .
2.0 : .. -.......
.:/ .\"..
~
~
':.
.:/ .,-- ..... .\ :...
:
:........
til ':\\
. .ill \ .
....
0
w
<l 4.0
........
-2.0 0" '.
-4.0
-4.0
290
-6.0 L--_.L.U.."-----I~_~_---I"____ _"___---"_ _ _~
210 250 290 330

X(nm)
Figure 10. CD spectra of NaCl-treated nucleohistones. 0 M NaCl
( - ) ; 0.6M NaCl (---); 1.6M NaCl (-'-), DNA ( ... ). Li, H.J. et al.
(92). Reprinted with permission of John Wiley and Sons, Inc.
90 G. D. FASMAN
v. CIRCULAR DICHROISM OF CHROMATIN SUBUNITS (NUCLEOSOMES)
A. CD of Isolated Nucleosomes
The repeating subunit of chromatin structure is termed the

nucleosome, with a spacer of DNA between nucleosomes. The nucleo-
some core contains about 140 base pairs of DNA and 8 histone mole-
cules, two each of H2A, H2B, H3, and H4 [recently reviewed by
Kornberg (1), Felsenfeld (2)]. Nucleosomes (mono, di, tri, etc.)
are isolated from micrococcal nuclease digests of nuclei or chroma-
tin. Depending on the extent of nuclease digestion, the mononucleo-
some may also contain varying size "tails" of DNA to which HI is
bound (69,102,103); the "tails" originate from the approximately
60 base pair spacer DNA between nucleosome cores in chromatin and
are usually shorter than their original length in most large-scale
nucleosome preparations. Finch et al. (104) have recently proposed
a detailed model for the conformation of DNA within nucleosome
cores. In this model, the 140 base pairs of DNA form 1-3/4 super-
helica~ turns around a histone core. The pitch of the superhelix
is 28 A. Thus the adjacent coils of DNA are spatially close
enough for cation or histone salt bridges between phosphates.
Little information exists about the conformation of DNA "tails"
attached to the nucleosome cores.
The condensation of DNA into supercoils around a histone core

may provide the electronic interactions necessary to reduce the
ellipticity of chromatin compared to free DNA. The type of CD
spectrum observed above 250 nm for isolated PS particles which con-
tain in the order of 100 base pairs of DNA, i.e., subnucleosome
particles, in low ionic strength solution is shown in Figure 11
(105). The positive band exhibits a maximum at 284 nm, and a
shoulder at 275 nm. A new negative band is observed at 295 nm.
The positive CD band ellipticity is approximately one-half that of
chromatin, and the ratio of ellipticity at 275 nm to that at 284
nm is lower relative to chromatin. These changes are consistent
with the addition of a large negative CD band centered at 275 nm
to the CD band of B-form DNA; that is, more ~ character in the
spectrum. Shih and Lake (81) have shown that the calculated CD
difference spectrum (above 250 nm) between B-DNA and chromatin is
a single negative band centered at 275 nm. It is apparent from
the spectra in Figure 11 that the mononucleosome spectrum contains
a greater percentage contribution from the ~ band, than in whole
chromatin, which may be interpreted as arising from tertiary order
in DNA arrangement (supercoiling).
Mononucleosomes containing at least 140 base pairs of DNA

yield similar CD spectra (83,103,106-108). Ellipticity values re-
ported for the CD maximum near 284 nm range from approximately
1400 to 2200 deg. cm 2 dmole- l .
Figure 11. Circular dichroism spectra of whole calf thymus chromatin

(C), nuclease-resistant fragments (PS), trypsin-digested PS fragments
(TPS), and DNA. Solvent is 10 rnM Tris-HCl, pH 8.0, room temperature.
Sahasrabuddhe, C.G., and Van Holde, K.E. (105). Reprinted with per-
mission of the American Society of Biological Chemists.
92 G.D.FASMAN
There is additional evidence that the CD spectrum of mono-

nucleosomes contains a strong contribution from an ordered tertiary
structure of DNA. The temperature dependence of the 280 nm ellip-
ticity for DNA, chromatin, and mononucleosomes (83) is compared in
Figure 8. The mononucleosome meltout is very similar to chromatin:
there is a strong increase in ellipticity around 70°C followed by
a decrease near 80°C. The first transition may be interpreted as
destruction of tertiary order; the second transition corresponds
to denaturation of the double helix.
B. Perturbation of Mononucleosomes
1. Effect of Salts. Mononucleosomes may be precipitated

partially by KCl, and completely by a number of salts containing
divalent cations (109). Finch and Klug (86) demonstrated by elec-
tron microscopic evidence that aggregated mononucleosomes in water
form a superhelical fiber; this aggregate can be destroyed by
chelating agents which remove residual cations. Olins et al. (110)
reported a fibrous appearance for aggregated nucleosomes in 5 mM
MgC1 2 . Moreover, these authors found nucleosomes precipitated by
MgC1 2 to exhibit low-angle X-ray scattering with the same maxima
as chromatin.
It has not yet been possible to reproduce chromatin-like CD

spectra from associated mononucleosomes. Sahasrabuddhe and Saun-
ders (108) have studied mononucleosome aggregates (nucleosomes
precipitated with 10 mM MgC1 2 , then resolubilized by addition of
[NH4]2S04) and found that the CD spectrum of the initially solubi-
lized mononucleosome preparation exhibited an ellipticity at 282 nm
only one-half that for untreated mononucleosomes in EDTA at low
ionic strength. Additional features of the CD spectrum observed
were a lower 6275/6282 ratio and a larger negative CD band at 294
nm relative to the untreated nucleosomes. In the presence of ad-
ditional (NH4)2S04' beyond that necessary for solubilization of
the precipitated nucleosomes, the CD spectra showed a trend toward
the spectrum of chromatin, but did not yield a true chromatin-like
spectrum.
2. Effect of Urea. The effect of urea on mononucleosomes is

analogous to the effect on chromatin (see Section IV, C, 2). In
6M urea, nucleosomes exhibit a CD spectrum having a DNA contribu-
tion similar to that of free DNA in the same solvent (103,111). A
thorough study of the effect of increasing urea concentration on
mononucleosomes has been made by Olins and coworkers (112). The
CD data from that study is shown in Figure 12. The positive CD
band above 250 nm linearly increased in ellipticity with increasing
urea concentrations up to 8M. Parallel hydrodynamic studies found
that the viscosity and sedimentation properties of the mononucleo-
somes showed similar linear changes indicative of DNA uncoiling.
By contrast, the histone CD bands showed a cooperative loss of a-
helical structure between 4 and 7M urea. This data suggests that

the graduate unfolding of the nucleosome leads to a loss of the
spectral perturbation of protein-bound DNA relative to free DNA
and that the condensation of DNA is not directly related only to
the secondary structure of the histones, although histone-histone
interactions cannot be evaluated in this study.
3. Effect of Trypsin. The digestion of the histones, within

the mononucleosomes, with trypsin results in a change in the nucleo-
some shape from globular to extended (105,113). As a consequence
of the DNA unfolding, the CD contribution for DNA within the nucleo-
some changes towards that of free DNA (Figure 11). The CD spectral
change found for these subnucleosomal PS particles (105) is in
agreement with results found for mononucleosomes containing 140
and 160 base pairs (113). Thus the histone segments responsible
for DNA supercoiling are readily attacked by trypsin.
4. Effect of Histone Crosslinking. Linkages formed between

his tones in isolated mononucleosomes by dimethylsuberimidate (114)
result in only a small reduction in the DNA CD band at 275 nm; the
CD ellipticity decrease is much less at 284 nm. A slight tighten-
ing of the histone-histone interactions by chemical crosslinking
thus may cause a concomitant small increase in condensation of the
DNA. This effect is analogous to the observed changes in chromatin
CD after formaldehyde treatment (94) discussed in Section IV, C, 3.
VI. CIRCULAR DICHROISM OF MODEL CHROMATIN SYSTEMS
Significant information concerning the nature of protein:DNA

interactions in the complex structure of chromatin can be obtained
from the study of model protein:DNA systems. Firstly, it can be
established whether or not any single histone or model histone can
induce a conformational change in DNA similar to that found in
chromatin. Secondly, the stabilizing forces involved in histone:
DNA interactions can be elucidated by studying synthetic polypeptides
of different amino acid compositions. The relative importance of
electrostatic and hydrophobic interactions in the DNA:histone com-
plex may be evaluated from investigation of these different model
protein structures.
A. Polypeptide:DNA Complexes
The use of polypeptide models for his tones in the investiga-

tion of histone:DNA interactions allows the determination of the
degree to which electrostatic and hydrophobic forces determine
94 G. D. FASMAN
10
o -i-----\-'-':---
-10
r()
Q 1
x -20 J
Q..
Q)
.........
,
,
I
-30 • I
.'
, I
;J
-40 \'oJ!
200 240 280 320 360

WAVELENGTH
Figure 12. Circular dichroism of mononucleosomes at different urea
concentrations. Mononucleosomes in 0 (---), 2 (-.-), 4 (---) , 6
( - - ) M urea (thick lines), or 8 (----), 10 ( - - ) M urea (thin
lines). Also shown is DNA in O.lM NaCl ( .. • • ). Ellipticity cal-
culated on basis of DNA residue concentration. Olins, D.E. et al.
(112). Reprinted with permission of Information Retrieval Ltd.
specific aspects of the complexation. A model polypeptide for

investigation of predominately electrostatic protein:DNA binding
is poly-L-lysine. When poly-L-lysine is complexed to DNA, by dialy-
sis to 0.85-l.0M NaCl, the CD spectrum of DNA shows large changes
(115-117). A large negative ellipticity band centered at approxi-
mately 270 nm appears, the ellipticity of which is dependent on the
ratio of lysine:P0 4 , and can reach 25 times the normal range of
ellipticity values for DNA (Figure 13). X-ray diffraction studies
of the complexes (117) show that the DNA retains the B conformation.
Electron microscopic (117) and light scattering (116) investigations
found that the complexes are aggregated into highly solvated particles
of nearly uniform size with a doughnut-like shape. Lerman and co-
workers (21) attributed the CD spectrum observed for these poly-L-
,-..
I
I
X +10
I + 10 -.!.-
>
::I --->
-
cr :J
cu v
--
v
N O'~~~~------~-------7~~~~,~,~~O ~
E NE
Y
VI v
cu
C!..i
~
~ -10 -10 ~
"0
....I
'-'
v
to
'-'
o .q
I
x -20 -20 ~x
~
230 250 270 290

A( 11m)
Figure 13. The circular dichroism spectra of DNA in various com-
plexes or perturbing solvents. Curves are redrawn from published
data. 1, H4:DNA complex, r = 1.5; 2, DNA in ethylene glycol; 3,
Hl:DNA complex, r = 1.0; 4, T7 DNA in 0.2M NaCl and 126.5 mg/ml
PEO, 48 hr. after mixing; 5, poly-L-lysine:DNA complex, r = 1.1.
All data except curve 4 with calf thymus DNA. The ordinate on the
left applies to solid curves, while the ordinate on the right,
corresponding to a 10-fold larger value, applies to the dashed
curves. Jordan, C.F. et al. (21). Reprinted with permission of
Macmillian Journals, Ltd.
96 G.D.FASMAN
lysine:DNA complexes to electronic interaction between adjacent

DNA helices in a condensed DNA structure; that is, similar to ~
DNA. Thus, electrostatic interactions alone can playa significant
role in the condensation of DNA by proteins.
Synthetic polypeptides containing hydrophobic residues as well

as basic residues have been interacted with DNA, and their complexes
studied. The polypeptides studied include the random copolymers of
L-Lys and L-Ala (118-120), L-Lys and L-Leu (121-122), L-Lys and L-
Val (123-124), and L-Lys, L-Ala, and Gly (125). Sequential poly-
peptides which have been studied are: (L-Lys-L-Ala-Gly), (L-Ala-
L-Lys-L-Lys-L-Pro-L-Lys)n' (L-Ala-L-Lys-L-Pro) (119,126Y; (L-
Lys-GlY)n (127), (L-Lys-L-Ala) n (128), (L-Lys-f-Ala-Gly) n (125),
and (L-Lys-L-Ala-L-Pro) (125).
n
The inclusion of hydrophobic residues into the polypeptide
structure affects the manner in which the basic groups can change
the CD spectrum of DNA. Long runs or a large fraction of leucine
(hydrophobic groups) with lysine cause aggregation and a positive
contribution at 280 nm. Large runs of lysine with neutral amino
acids (Gly, Pro) cause ~-type DNA spectra. Lysine and valine random
copolymers cause little DNA CD change. Thus the nature of the hydro-
phobic side chain and its sequential order is important (124).
B. Histone:DNA Complexes
Histone H4, molecular weight ~11,000 has a high concentration

of basic amino acids near the N-terminal end of the molecule (129).
Upon binding H4 to DNA, by salt gradient dialysis, it is possible
to obtain two types of CD spectra (53). It is necessary to add
urea to the dialysis solutions to prevent H4 aggregation, and
these two spectra depend upon the salt concentration at which urea
is removed in the dialysis procedure. When the urea is removed at
a salt concentration of 0.14M NaCl, then altered CD spectra are
obtained. As the ratio of histone amino acid residues to DNA phos-
phates (r) is increased from 0 to 0.5, the CD spectra are essentially
unchanged; from r = 1.0 -1.5, a gradual blue shift and an increase
in amplitude of the 275 nm band is observed and a maximum value of
[9]Z70= 13,800 is obtained at r = 1.5. The 245 nm negative ellip-
tiClty band is concurrently decreased in amplitude and blue
shifted. A new negative band centered at 305 nm is generated.
When r becomes larger than 1.5, the magnitude of the CD change is
decreased, and at r = 2.5 the CD spectrum is close to that of B
form DNA. The spectra observed for H4:DNA complexes show no con-
centration dependence. The CD spectrum obtained at r = 1.5 (53)
looks similar to the CD spectra of RNA, the theoretically calculated
CD spectrum for A-form DNA (9), and the measured CD spectrum for
A-form DNA (13).
H4:DNA complexes give different CD spectra if urea is removed

at 0.015M NaCl (53). No changes in the CD spectra of these com-
plexes relative to B-DNA are observed. Thus, the exact environ-
mental media are important in formation of specific complexes.
Spectra of this latter type, where there is no DNA CD change, can
be utilized to evaluate the conformation of the bound histone. By
subtracting out the CD contribution of native DNA, the CD spectrum
of bound histone may be obtained. It was thus found that, upon
interaction with DNA, H4 assumes an ordered conformation with con-
siderable S structure. The sensitivity of H4 conformation to ionic
strength has already been discussed (Section III, B). Perhaps this
change of conformation with ionic strength is the reason for the
differences in the CD spectra when binding H4 to DNA under different
ionic conditions. These experiments illustrate the interdependence
of the conformations of both the DNA and protein.
Complexes of the other core his tones (H2A, H2B, H3) with DNA
have been studied in a similar fashion. These studies will be
described briefly. H2B, a slightly lysine-rich histone containing
125 amino acid residues, has a molecular weight of 13,800 and has
a lysine:arginine ratio of 2.5 (130). In 0.14M NaF, this histone
shows a CD spectrum indicative of about 30% a-helix (131). Upon
H2B complexation with DNA, at different histone:DNA ratios, the
CD spectra obtained (131) are similar to those observed for H4:DNA
complexes. At a ratio of r = 3.0, the H2B:DNA complex shows a
positive CD band in the 280 nm region with an ellipticity greater
than 40,000.
Histone H2A, a slightly lysine-rich protein, has 129 amino

acids, with the more basic region residing near the N-terminal end
(132,133). Complexes reconstituted from DNA and H2A show similar
CD spectra to those for H2B:DNA and H4:DNA complexes (134). Thus,
the positive CD band of the DNA is increased in magnitude and
slightly blue-shifted.
The arginine-rich histone H3 contains 135 residues, and has a

typical histone amino acid distribution with many basic residues
clustered in the N-terminal part of the protein (135). When H3 is
complexed with DNA, only slight alterations in the DNA-CD spectrum
are observed (134).
The perturbation of DNA structure caused by the core his tones

is apparently due primarily to interaction between DNA and the more
basic sections of the histone structure (131). This was demonstrated
in a study of H2B cleaved into two fragments of nearly equal size
but different charge densities. The more basic fragment caused a
conformational change in DNA similar to that described for intact
H2B (but lower in magnitude), whereas the less basic fragment did
not significantly change the DNA CD spectrum (131).
98 G.D.FASMAN
Further evidence that electrostatic interactions play a prim-

ary role in the histone:DNA complexation can be obtained from
studies of modified histones. Histone H4 is enzymatically acety-
lated (136) in vivo to form E-N-acetyl lysine residues at specific
sites. The CD spectra for complexes of DNA with H4 at various
specific levels of acetylation (137) show that modification of
even one lysine residue in the basic region of the histone moderates
the histone effect on DNA conformation. As it has been shown that
H4 is acetylated at the time of extensive gene activation in sea
urchin embryos (138), a possible mechanism is suggested by this CD
study.
CD studies of complexes formed between DNA and any of the

core his tones show that each of these his tones can perturb the con-
formation of DNA to some extent. This effect is dependent on strong
electrostatic interactions. Individual histones, however, do not
perturb the DNA structure in the same manner as the histone-histone
complex in nucleosomes. This is evident from the CD spectra, which
show a reduction in ellipticity and a red shift of the band for
DNA in chromatin relative to free DNA, but an increase in ellipticity
and a blue shift of the band position for DNA complexed by individual
core histones. These model studies illustrate the importance of
histone-histone interactions in chromatin.
The his tones HI and H5 (present in avian erythrocytes) are

thought to interact primarily with the spacer regions of DNA be-
tween nucleosomes (69,98,102). These his tones are very lysine-
rich, and higher in molecular weight than the core histones.
Histone HI has a molecular weight of 21,000. It contains 214

amino acids, 61 of which are lysine, 3 are arginine, and 16 are
acidic amino acids. The basic groups are not evenly distributed,
and 39 are found in the carboxyl end of the molecule (139). Upon
complexing HI with DNA, by gradient dialysis, an altered CD spec-
trum is obtained, as seen in Figure 15 (52). As increasing amounts
of histone are added, the peak at 275 nm decreases and red shifts,
being completely eliminated at a ratio (r, amino acid residues per
DNA phosphate) of 1.0. The extent of the CD change for any par-
ticular Hl:DNA ratio is concentration dependent, being greater at
higher concentrations. Thus aggregation between complexes plays
a role in the CD changes.
Histone H5, which largely replaces HI in avian erythrocyte

chromatin, has a molecular weight of 21,450. There are 197 resi-
dues, of which 49 are lysine, and 22-23 are arginine (140). H5
exerts a conformational effect on DNA similar to that of HI (141),
in that the positive CD band of DNA is reduced and shifted to
longer wavelengths.
HI and H5 are thus seen to perturb the DNA structure in a

10
2
I»
'0 0
E
u
0
I»
-2
......
(\j
E -4
u
CJ
-6
...
CJ
CI
CJ
0 -8
II')
I
0 -10
)(
,....... -12
Q)
1......0
-14
-16
-18
-20
-22
210 230 250 270 290 310

A (mp.)
Figure 14. Circular dichroism spectra of H1:DNA complexes as a
function of r, the histone residue : P04 ratio, DNA concentration
10- 3M (P04); solvent 0.14 M NaF, pH 7.0. 1, DNA (calf thymus);
2, r = 0.25; 3, r = 0.50; 4, r = 0.75; 5, r = 1.0. Error bars
represent reproducibility and noise dependence. Uncorrected for
histone contribution. Fasman, G.D. ~t a1. (52). Reprinted with
permission of the American Chemical Society.
100 G.D.FASMAN
manner similar to that found in chromatin, or in ~-DNA. This is

of great interest, since these histones are postulated to condense
the spacer regions of DNA (69,70) between nucleosomes. The indivi-
dual core histones, by contrast, do not perturb DNA in this same
manner, and must act as a specific histone-histone complex to
condense DNA into nucleosomes.
Enzymatic modification of HI (142) producing mono and di phos-

phorylated species, occurs in response to the pancreatic hormone
glucagon, upon initiation of protein synthesis. Thus it is postu-
lated that phosphorylation of HI results in derepression of the
template activity of the associated DNA. When phosphorylated HI
is complexed to DNA, the change in the CD spectrum found for Hl:DNA
complexes is completely reversed (143). Perhaps these phosphoryla-
tions cause a modification of the histone:histone and/or histone:
DNA interaction, reducing the state of aggregation in chromatin,
thus making DNA more accessible for transcription.
DNA may also be reconstituted with a mixture of the four core

histones. The extent to which such a complex resembles native
chromatin (depleted of HI) provides information about the mechanism
of chromatin assembly. That is, can the core histones alone organize
DNA into an ordered condensed state? Circular dichroism is an ex-
cellent sensitive probe for evaluating such condensation. By a
salt step gradient dialysis procedure, Garel et al. (144) showed
that an equimolar mixture of the four his tones H2A, H2B, H3, and
H4 could complex and condense DNA into a chromatin-like structure.
Both the CD spectra and CD melting profiles of the reconstituted
material were identical to HI-depleted chromatin. This report con-
firmed the observations of Stein et al. (145) that chromatin re-
constituted by a gradient dialysis procedure exhibits a CD spectrum
similar to that for native chromatin. An earlier review of this
subject matter should be consulted for further references (146).
VII. SUMMARY
The utilization of circular dichroism can be of great assistance

in defining the conformational states of both DNA and the proteins
found in chromatin.
The conformation of the his tones can be evaluated with a high

degree of confidence, especially when isolated from the nucleopro-
tein complex. A reasonable estimate of their conformation can be
made in chromatin using the CD bands below 240 nm, as the DNA con-
tribution in this region is minimal, compared to that of the protein
contribution. The CD spectrum has been interpreted to indicate
that the conformation of the his tones in chromatin are composed of
40-50% a-helix, 0-5% S-sheet and 51-60% random coil. A mixture of
all the his tones exhibits similar CD spectra in 2M NaCl as found

in chromatin, which suggests that the conformation of the histones
in chromatin is dependent in part on the electrostatic interaction
with DNA.
The conformation of DNA may be evaluated by examination of

the CD bands above 240 nm, where the his tones have a negligible
contribution. Upon binding of the his tones (and other proteins),
the CD spectrum of DNA becomes significantly altered, with the
largest change centered around 275 nm. This observation has been
interpreted as either 1) a change in the secondary structure of
DNA (e.g., B~ form) or 2) an ordering of DNA into a tertiary
structure. Concominant changes in other physical parameters are
known to be caused by the condensation of DNA, without necessarily
altering the secondary structure. Therefore, the formation of a
tertiary structure would be strongly implicated. Thus CD can be
used both to evaluate the degree of condensation, ie .• , tertiary
structure, as well as the secondary structure. CD spectra tem-
perature profiles can be used to study the stabilization of both
the secondary and tertiary structure of chromatin.
Great caution must be exercised in interpreting the CD spectra

of chromatin. Factors which must be evaluated are the composition
of the complex (proteins, DNA, RNA) and optical artifacts such as
scattering and absorption phenomena due to particle size.
As a probe for conformational changes in the structural or-

ganization of chromatin, circular dichroism offers a tool of great
sensitivity and ease of use.
ACKNOWLEDGMENTS
The writing of this review was generously supported in part

by grants from the U. S. Public Health GM 17533 and the National
Science Foundation (PCM76-2l856). This is Publication No. 1190
from the Graduate Department of Biochemistry, Brandeis University,
Waltham, Massachusetts 02154. The author wishes to thank Dr. M.
Cowman for her major contribution to the writing of this review.
REFERENCES
1. Kornberg, R.D. (1977). Ann. Rev. Biochem. 46, 931-954.

2. Felsenfeld, G. (1978). ~re 271, 115-122-.-
3. Tinoco, 1., Jr., and Cantor, C.R. (1970). In "Methods of Bio-
chemical Analysis" (D. Glick, ed.), Vol. 18, pp. 81-203. Wiley,
New York.
102 G. D. FASMAN
4. Brahms, J., and Brahms, S. (1970). In "Fine Structure of Pro-

teins and Nucleic Acids" (G.D. Fasman and S.N. Timasheff, eds.),
pp. 191-270. Marcel Dekker, Inc., New York.
5. Sears, D.W., and Beychok, S. (1973). In "Physical Principles
and Techniques of Protein Chemistry" (S.J. Leach, ed.), Part
C, pp. 445-593. Academic Press, New York.
6. Adler, A.J., Greenfield, N.J., and Fasman, G.D. (1973). Meth.
Enzymo1ogy~, Part D, 675-735.
7. Tinoco, 1., Jr. (1964). J. Am. Chem. Soc. 86, 297-298.
8. Tinoco, 1., Jr. (1968). I. Chim. Phys. 65,91-97.
9. Johnson, W.C., Jr., and Tinoco, I., Jr. (1969). Biopo1ymers l,
727-749.
10. Langridge, R., Marvin, D.A., Seeds, W.E., Wilson, H.R., Hooper,
C.W., Wilkins, M.H.F., and Hamilton, L.D. (1960). I. Mol. BioI.
1, 38-64.
11. Marvin, D.A., Spencer, M., Wilkins, M.H.F., and Hamilton, L.D.
(1961). I. Mol. BioI. 1, 547-565.
12. Fuller, W., Wilkins, M.H.F., Wilson, H.R., Hamilton, L.D., and
Arnott, S. (1965). I. Mol. BioI. 11, 60-80.
13. Tunis-Schneider, M.J.B., and Maestre, M.F. (1970).· J. Mol.
BioI. 52, 521-541. - --
14. Brahms, J., and Mommaerts, W.F.H.M. (1964). I. Mol. BioI. 10,
73-88.
15. Studdert, D.S., and Davis, R.C. (1974). Biopo1ymers 13, 1377-
l389.
16. Bram, S. (1971). I. Mol. BioI. 58, 277-288.
17. Studdert, D.S., Patroni, M., and Davis, R.C. (1972). Bio-
polymers 11, 761-779.
18. Hanlon, S., Johnson, R.S., Wolf, B., and Chan, A. (1972).
Proc. Nat. Acad. Sci. U.S.A. ~, 3263-3267.
19. Brunner, W.C., and Maestre, M.F. (1974). Biopo1ymers 13,
345-357.
20. Lerman, L.S. (1971). Proc. Nat. Acad. Sci. U.S.A. 68, 1886-1890.
21. Jordan, C.F., Lerman,L:S"., and Venab1;:-J.H., Jr.(1972).
Nature New BioI. 236, 67-70.
22. Maniatis, T., Venable, J.H., Jr., and Lerman, L.S. (1974). J.
Mol. BioI. 84, 37-64.
23. Saeva:--F:D.-,-and Wysocki, J. J. (1971). I. Am. Chem. Soc • .21,
5928-5929.
24. Saeva, F.D. (1972). J. Am. Chem. Soc. 94, 5135-5136.
25. Chabay, I. (1972). Chem. Phys. Lett. 17, 283-287.
26. Holzwarth, G., Chabay, I., and Holzwarth, N.A.W. (1973). J.
Chem. Phys, 58, 4816-4819.
27. Holzwarth, G., and Holzwarth, N.A.W. (1973). I. Q£!. Soc. Am.
63, 324-331.
28. Cheng, S.M., and Mohr, S.C. (1975). Biopo1ymers 14, 663-674.
29. Permogorov, V.I., Debabov, V.G., Sladkova, I.A., and Rebentish,
B.A. (1970). Biochim. Biophys. Acta 199, 556-558.
30. Fric, I., and Sponar, J. (1971). Biopo1ymers 10, 1525-1531.
31. Zimmer, C., and Luck, G. (1973). Biochim. Biophys. Acta 312,
215-227.
32. Hanlon, S., Brudno, S., Wu, T.T., and Wolf, B. (1975).
Biochemistry 14, 1648-1660.
33. Wolf, B., Berman, S., and Hanlon, S. (1977). Biochemistry 16,
3655-3662.
34. Zimmer, C., and Luck, G. (1974). Biochim. Biophys. Acta 361,
11-32.
35. Ivanov, V.I., Lysov, Yu.P., Ma1enkov, G.G., Minchenkova, L.E.,
Minyat, E.E., Schyo1kina, A.K., and Zhurkin, V.B. (1976).
Studia Biophysica 55, 5-13.
36. Bram, S., and Tougard, P. (1972). Nature New BioI. 239, 128-131.
37. Green, G., and Mahler, H.R. (1968). Biopo~e~, 1509-1514;
38. Nelson, R.G., and Johnson, W.C., Jr. (1970). Biochem. Biophys.
Res. Connn. 41, 211-216.
39. Girod, J.C., Johnson, W.C., Jr., Huntington, S.K., and Maestre,
M.F. (1973). Biochemistry~, 5092-5096.
40. Bronner, M., and Pysh, E.S. (1976). Biopo1ymers 12, 589-590.
41. Giannoni, G., Padden, F.J., Jr., and Keith, H.D. (1969).
Proc. Nat. Acad. Sci. U.S.A. 62, 964-971.
42. Marmur:-J., and Doty, P. (1962). J. Mol. BioI. l, 109-118.
43. Gennis, R.B., and Cantor, C.R. (1972~l. Mol. BioI. 65,
381-399.
44. Urry, D.W. (1972). Biochim. Biophys. Acta 265, 115-168.
45. Schneider, A.S. (1973). Meth. Enzymo1ogy~, Part D, 751-767.
46. Gordon, D.J., and Holzwarth, G. (1971). Proc. Nat. Acad. Sci.
U.S.A. 68, 2365-2369.
47. Gordon,-n.J. (1972). Biochemistry 11, 413-420.
48. Schneider, A.S., and Harmatz, D. (1976). Biochemistry 15,
4158-4162.
49. Nicolini, C., Baserga, R., and Kendall, F. (1976). Science 192,
796-798.
50. Bohren, C.F. (1977). J. Theor. BioI. 65, 755-767.
51. Chen, Y.-H., Yang, J.T., and Martinez-,-H.M. (1972). Biochemistry
11, 4120-4131.
52. Fasman, G.D., Schaffhausen, B., Goldsmith, L., and Adler, A.
(1970). Biochemistry~, 2814-2822.
53. Shih, T.Y., and Fasman, G.D. (1971). Biochemistry 10, 1675-
1683.
54. Li, H.J., Wickett, R., Craig, A.M., and Isenberg, I. (1972).
Biopo1ymers 11, 375-397.
55. Wickett, R.R., Li, H.J., and Isenberg, I. (1972). Biochemistry
11, 2952-2957.
56. DIAnna, J.A., Jr., and Isenberg, I. (1972). Biochemistry 11,
4017-4025.
57. D'Anna, J.A., Jr., and Isenberg, I. (1974). Biochemistry 13,
2093-2097.
4987-4992.
104 G. D. FASMAN
59. Fasman, G.D., Chou, P.Y., and Adler, A.J. (1977). In "The
Molecular Biology of the Mammalian Genetic Apparatus" (P.O.P.
Ts'o, ed.), Vol. 1, pp. 1-52. Elsevier/North-Holland Bio-
medical Press, Amsterdam.
1035-1043.
2098-2104.
4992-4997.
63. D'Anna, J.A., Jr., and Isenberg, I. (1974). Biochem. Biophys.
Res. Comm. 61, 343-347.
64. Bidney, D.L:: and Reeck, G.R. (1977). Biochemistry 16, 1844-
1849.
65. Bradbury, E.M., Cary, P.D., Crane-Robinson, C., Rattle, H.W.E.,
Boub1ik, M., and Sautiere, P. (1975). Biochemistry 14, 1876-
1885.
66. Thomas, G.J., Jr., Prescott, B., and 01ins, D.E. (1977).
Science 197, 385-388.
67. Weintrau~H., Palter, K., and Van Lente, F. (1975). Ce11~,
85-ll0.
68. Pardon, J.F., Cotter, R.I., Lilley, D.M.J., Worcester, D.L.,
Campbell, A.M., Wooley, J.C., and Richards, B.M. (1977).
Cold Spring Harbor Symposium, in press.
69. Noll, M., and Kornberg, R.D. (1977). J. Mol. BioI. 109, 393-404.
70. Renz, M., Nehls, P., and Hozier, J. (197~ p~ Nat. Acad.
Sci. U. S.A. 74, 1879-1883. -- --
71. Shih,~, and Fasman, G.D. (1970). ~. Mol. BioI. ~, 125-129.
72. Simpson, R.T., and Sober, H.A. (1970). Biochemistry~, 3103-
3109.
73. Henson, P., and Walker, 1.0. (1970). Eur. J. Biochem. 16, 524-
53l.
74. Mat suyama , A., Tagashira, Y., and Nagata, C. (1971). Biochim.
Biophys. Acta 240, 184-190.
75. Tashiro, T., and Kurokawa, M. (1975). FEBS Lett. 59, 250-253.
76. Hjelm, R.P., Jr., and Huang, R.C.C. (1975).~che;istry 14,
1682-1688.
77. Nicolini, C., and Baserga, R. (1975). Arch. Biochem. Biophys.
169, 678-685.
78. Huang, C.-H., and Baserga, R. (1976). Biochemistry 15, 2829-
2836.
79. Baserga, R., and Nicolini, C. (1976). Biochim. Biophys. Acta
458, 109-134.
80. Johnson, R.S., Chan, A., and Hanlon, S. (1972). Biochemistry
ll, 4347-4358.
81. Shih, T.Y., and Lake, R.S. (1972). Biochemistry 11, 4811-4817.
82. Wilhelm, F.X., DeMurcia, G.M., Champagne, M.H., and Daune,
M.P. (1974). Eur. J. Biochem. 45, 431-443.
83. Mandel, R., and Fasman, G.D. (1976). Nuc1. Acids Res. 1, 1839-
1855.
84. Rees, A.W., Dubuysere, M.S., and Lewis, E.A. (1974). Biochim.
Biophys. Acta 361, 97-108.
85. Brasch, K. (1976). Exp. Cell Res. 101, 396-410.
86. Finch, J.T., and K1ug, A. (1976). Proc. Nat. Acad. Sci. U.S.A.
n, 1897-190l.
87. Bartley, J., and Chalkley, R. (1973). Biochemistry 11, 468-474.
88. Maestre, M.F., and Wang, J.C. (1971). Biopo1ymers 10, 1021-
1030.
89. Baase, W.A., and Johnson, W.C., Jr. (1976). Nuc1. Acids Res.
1, 3123-313l.
90. Sipski, M.L., and Wagner, T.E. (1977). Biopo1ymers 16, 573-
582.
91. Simpson, R.T. (1972). Biochemistry 11, 2003-2008,
92. Li, H.J., Chang, C., Evage1inou, Z., and Weiskopf, M. (1975).
Biopo1ymers 14, 211-226.
93. Chang, C., and Li, H.J. (1974). Nuc1. Acids Res. 1, 945-958.
94. Senior, M.B., and 01ins, D.E. (1975). Biochemistry 14, 3332-
3337.
95. Simpson, R.T. (1971). Biochemistry 10, 4466-4470.
96. Tack, L.O., and Simpson, R.T. (1977). Biochemistry 16, 3746-
3753.
97. Oh1enbusch, H.H., Olivera, B.M., Tuan, D., and Davidson, N.
(1967).1. Mol. Bio1. 12, 299-315.
98. Whitlock, J.P., Jr., and Simpson, R.T. (1976). Biochemistry
15, 3307-3314.
99. Williams, R.E., Lurquin, P.F., and Se1igy, V.L. (1972). Eur.
J. Biochem. 29, 426-432.
100. Vengerov, Yu-.-Yu., and Popenko, V.I. (1977). Nuc1. Acids Res.
!!.., 3017-3027.
101. Hjelm, R.P., Jr., and Huang, R.C.C. (1974). Biochemistry 13,
5275-5283.
102. Varshavsky, A.J., Bakayev, V.V., and Georgiev, G.P. (1976).
Nuc1. Acids Res. 1, 477-492.
103. Whitlock, J.P., Jr., and Simpson, R.T. (1976). Nuc1. Acids
Res. 1, 2255-2266.
104. Finch, J.T., Lutter, L.C., Rhodes, D., Brown, R.S., Rushton,
B., Levitt, M., and K1ug, A. (1977). Nature 269, 29-36.
105. Sahasrabuddhe, C.G., and Van Holde, K.E. (1974).1. Bio1. Chern.
249, 152-156.
106. Ramsay Shaw, B., Corden, J.L., Sahasrabuddhe, C.G., and Van
Holde, K.E. (1974). Biochem. Biophys. Res. Cornrn. ~, 1193-
1198.
107. Lawrence, J.-J., Chan, D.C.F., and Piette, L.H. (1976). Nuc1.
Acids Res. 1, 2879- 2893.
108. Sahasrabuddhe, C.G., and Saunders, G.F. (1977). Nuc1. Acids
Res. !!.., 853-866.
109. 01ins, A.L., Carlson, R.D., Wright, E.B., and 01ins, D.E.
(1976). Nuc1. Acids Res. 1, 3271-3291.
106 G. D. FASMAN
110. 01ins, A.L., Brei11att, J.P., Carlson, R.D., Senior, M.B.,

Wright, E.B., and 01ins, D.E. (1977). In "The Molecular Biology
of the Mammalian Genetic Apparatus" (P.O.P. Ts'o, ed.), Vol. 1,
pp. 211-237. Elsevier/North Holland Biomedical Press,
Amsterdam.
111. Rill, R., and Van Holde, K.E. (1973). I. BioI. Chern. 248,
1080-1083.
112. 01ins, D.E., Bryan, P.N., Harrington, R.E., Hill, W.E., and
Olins, A.L. (1977). Nuc1. Acids Res. ~, 1911-1931.
113. Lilley, D.M.J., and Tatche1l, K. (1977). Nuc1. Acids Res. ~,
2039-2055.
114. Stein, A., Bina-Stein, M., and Simpson, R.T. (1977). Proc. Nat.
Acad. Sci. U.S.A. 74, 2780-2784. - - --
lIS. Cohen,~, and Kidson, C. (1968). J. Mol. BioI. 35, 241-245.
116. Shapiro, J.T., Leng, M., and Fe1se~fe1d, G~969). Biochemistry
~, 3219-3232.
117. Haynes, M., Garrett, R.A., and Gratzer, W.B. (1970). Biochem-
istry 2, 4410-4416.
118. Stokrova, S., Sponar, J., Havranek, M., Sedlacek, B., and
Blaha, K. (1975). Biopo1ymers~, 1231-1244.
119. Sponar, J., Blaha, K., and Stokrova, S. (1973). Stud. Biophys.
40, 125-133.
120. Pinkston, M.F., and Li, H.J. (1974). Biochemistry 12, 5227-
5234.
121. Ong, E.C., Snell, C. and Fasman, G.D. (1976). Biochemistry 15,
468-477 .
122. Ong, E.C., and Fasman, G.D. (1976). Biochemistry 15, 477-486.
123. Mandel, R., and Fasman, G.D. (1974). Biochem. Biophys. Res.
Comm. 59, 672-679.
124. Mandel, R., and Fasman, G.D. (1976). Biochemistry 15, 3122-
3130.
125. Schwartz, A.M., and Fasman, G.D. (1977). Biochemistry 1&,
2287-2299.
126. Sponar, J., Fric, I., and Blaha, K.B. (1975). Biophys. Chern.
1, 255-262.
127. Williams, R.F., and Kie11and, S.L. (1975). Can. I. Chern. 22,
542-548.
128. Privat, J.-P., Spach, G., and Leng, M. (1972). Eur. I. Biochem.
~, 90-95.
129. DeLange, R.J., Smith, E.L., Fambrough, D.M., and Bonner, J.
(1968). Proc. Nat. Acad. Sci. U.S.A. 61, 1145-1146.
130. Iwai, K.:-r;hikawa,~ and Hayashi, ~ (1970). Nature 226,
1056-1058.
131. Adler, A.J., Ross, D.G., Chen, K., Stafford, P.A., Woiszwi110,
M.J., and Fasman, G.D. (1974). Biochemistry 13, 616-623.
132. Yeoman, L.C., Olson, M.O.J., Sugano, N., Jordan, J.J., Taylor,
C.W., Starbuck, W.C., and Busch, H. (1972). J. BioI. Chern.
247, 6018-6023. - -- --
133. Sautiere, P., Tyrou, D., Laine, B., Mizon, J., Ruffin, P.,
and Biserte, G. (1974). Eur. I. Biochem. 41, 563-576.
134. Adler, A.J., Moran, E.C., and Fasman, G.D. (1975). Biochemistry
14, 4179-4185.
135. DeLange, R.J., Hooper, J.A., and Smith, E.L. (1972). Proc. Nat.
Acad. Sci. U.S.A. 69, 882-884.
136. Pogo, B.G.T., Pogo, A.O., A11frey, V.G., and Mirsky, A.E.
(1968). Proc. Nat. Acad. Sci. U.S.A. 59, 1337-1344.
137. Adler, A.J., Fasman, G.D., Wangh, L.J., and A11frey, V.G.
(1974). J. BioI. Cnem. 249, 2911-2914.
138. Wangh, L~, Ruiz-Carri11~A., and A11frey, V.G. (1972). Arch.
Biochem. Biophys. ISO, 44-56. --
139. RaIl, S.C., and Cole, R.D. (1971). ~. BioI. Chern. 246, 7175-
7190.
140. Sautiere, P., Briand, G., Kmiecik, D., Loy, 0., Biserte, G.
(1976). FEBS Lett. ~, 164-166.
141. Wagner, T.E., Hartford, J.B., Serra, M., Vandegrift, V., and
Sung, M.T. (1977). Biochemistry 16, 289-290.
142. Langan, T.A. (1969). J. BioI. Chern. 244, 5763-5765.
143. Adler, A.J., Langan, T.A~nd Fasma~G.D. (1972). Arch.
Biochem. Biophys. 153, 769-777.
144. Gare1, A., Kovacs, A.M., Champagne, M., and Daune, M. (1976).
Nuc1. Acids Res. 1, 2507-2519.
145. Stein, G.S., Mans, R.J., Gabbay, E.J., Stein, J.L., Davis, J.,
and Adawadkar, P.D. (1975). Biochemistry 14, 1859-1866.
146. Fasman, G.D. (1977). in "Chromatin and Chromosome Structure"
(Eds. Li, H.J. and Eckhardt, R.A.). Academic Press, pp. 71-142.
147. Ivanov, V.I., Minchenkova, L.E., Schyo1kina, A.K., and
Po1etayev, A.I. (1973). Biopo1ymers 12, 89-110.
148. Holzwarth, G., Gordon, D.G., McGinness, J.E., Dorman, B.P.,
and Maestre, M.F. (1974). Biochemistry 13, 126-132.
149. Dorman, B.P., and Maestre, M.F. (1973). Proc. Nat. Acad. Sci.
U.S.A. 70, 255-259.
150. Greenfield, N., and Fasman, G.D. (1969). Biochemistry~, 4108-
4116.
151. Slayter, H.S., Shih, T.Y., Adler, A.J., and Fasman, G.D.
(1972). Biochemistry II, 3044-3054.
U1PORTANT HYDRODYNAllIC AND SPECTROSCOPIC TECHNIQUES IN THE FIELD OF
CHROHATIN STRUCTURE
Donald E. Olins
University of Tennessee-Oak ~idge Graduate School of

Biomedical Sciences, Biology Division, Oak Ridge
National Laboratory, Oak Ridge, Tennessee 37830
INTRODUCTION
If chromatin were an enzyme or, at least if it bound reversi-

bly and tightly to a readily measurable ligand, we could directly
and meaningfully assay for chromatin functional states. We could
then search for conformational effector molecules, cooperative
structural transitions, and allosteric interactions. There are
reasons to. anticipate such conformational properties in chromatin.
Tn all probability, the nucleosome has several properties reminis-
cent of hemoglobin, the prototype of allosteric multisubunit pro-
teins. The nucleosome probably possesses a pairwise stoichiometry
of the constituent inner histones, an internal dyad axis with the
heterotypic tetramers acting as protamers, and a close-packing of
highly a-helical, globular polypeptide chainsl'2. Unfortunately,
simple functional assays are not yet the province of students of
chromatin structure. Save for complex transcriptional assays, con-
formational studies of chromatin and nucleosomes are largely con-
fined to enzyma~ic and biophysical probes.
The use of enzymatic probes has underscored the necessity of

identifying chromatin conformational states. The employment of
DNAse I, or of trypsin followed by micrococcal nuclease, has corre-
lated altered chromatin conformational states with transcriptional
activation 1 ,2; but these assays are destructive and clearly not re-
versible. ~1any biophysical techniques, on the other hand, are non-
destructive and conformational dynamics can be readily studied; but
the properties of molecules are necessarily averaged, and these
techniques are not easily focused upon a subset of nucleosomes with
conformations different than those of the bulk particles. !1ost bio-
physical studies to date are confined to "inactive" nucleosomes or
109
110 D. E. DLiNS
chromatin; amplification of the material, or the signal, is essential

for comparable studies on "active" states.
The purpose of this chapter is to illustrate the usefulness of

two broad classes of biophysical techniques - hydrodynamic and spec-
troscopic - emphasizing those currently in use in association with
this author's laboratory. Results of studies on the response of
mononucleosomes (VI) to various solvent parameters and perturbants
will document the contribution of the different techniques. Despite
the organization of this chapter around particular techniques, it is
important to recognize that different techniques must be used in
concert in order to develop a comprehensive view of nucleosome con-
formational states.
Hydrodynamic Techniques
Conventional hydrodynamic techniques (transport processes) have

the capability of yielding measurements of molecular weight, size and
shape, under varying solvent conditions 3 ,4. Analytical ultracentri-
fugation, combined with determinations of partial specific volume,
have permitted accurate estimation of the molecular weight of vI -
by equilibrium sedimentation, and by use of the Svedberg equation
(Table I).
Solvent-induced conformational changes that do not involve

changes in molecular weight are readily detected by monitoring sedi-
mentation (S), diffusion (D) or intrinsic viscosity [n]. Measure-
ments of S and molecular weight, or of D alone, permit estimation
of the frictional coefficient (f) which is a function of the shape
(asymmetry) and volume (hydration) of the macromolecules. The in-
trinsic viscosity of a molecule is also a function of particle
volume and asymmetry. Assumptions of particle volume (hydration)
are necessary to extract estimates of particle shape. It is impor-
tant to remember that changes in the S, D or [n] of vI with varying
solvent parameters can equally well be interpreted as arising from
changes in particle shape or volume, or a combination of both. The
Scheraga-}mndelkern factor (8) can be obtained by combining measure-
ments of Sand [n], or of D and [n]. 8 is a function only of par-
ticle shape (i.e., the axial ratio of an equivalent ellipsoid of
revolution). But, unfortunately, 8 does not effectively distin-
guish between oblate, spherical or slightly prolate objects. Low-
angle neutron scattering combined with the technique of "contrast-
variation" (i.e., employing buffers of varying H20/D20) undoubtedly
yields the most complete description of particle size and shape in
solution. Few of us, however, have ready access to a high flux
neutron source. Neutron scattering experiments are, therefore,
probably best employed after the conformational states are defined
by conventional hydrodynamic techniques.
HYDRODYNAMIC AND SPECTROSCOPIC TECHNIQUES 111
Examples will be presented on the influence of simple solvent

parameters on the size and shape of vI. The measurement of S, em-
ploying the analytical ultracentrifuge equipped with scanner optics,
is well-described in textbooks 3 ,4, and needs no further explanation.
Determination of D in the ultracentrifuge is based upon boundary-
spreading analysis of the solvent/solution interphase generated in
a synthetic boundary centerpiece, a technique of moderate accuracy
and lengthy computation. The advent of laser light scattering
techniques has permitted a rapid and accurate determination of D
by analysis of the time-decay of the intensity of the "twinkles"
of light scattered from groups of molecules, or the frequency-
broadening of scattered light due to Doppler shifts. 5
Urea Effects. Exposure of vI to urea (0-10 m

in buffers
containing 0.2 nL~ EDTA, pH 7.0, results in size and shape changes
without detectable dissociation of his tones and DNA6 • Figure 1
presents the results of Sand [n] measurements as a function of
urea. Combining S, [n] and partial specific volume (v) following
the theory of Scheraga and Mandelkern, permitted an estimation of
the dependence of S upon urea concentration. S was calculated to
increase monotonically between 0-8 M urea, indicating a transition
of vI toward a more prolate-type structure, from its slightly ob-
late shape in the absence of urea.
TABLE 1. Hydrodynamic parameters of v I in 0.1 t1 KCl
SO 11.11 + 0.30
20,w
DO 3.90 + 0.l3
20, ~..
v 0.670 + 0.005
~] (S & D) 209,547 ± 9,634

l\v (Equil Sed) 221,515 ± 2,143
[n] 5.628 + 0.003
S 2.119 + 0.058 x 10 6
112 D. E. OLiNS
Ionic Strength Effects. Isolated chicken erythrocyte vI has

been observed to undergo tno sharp conformational transitions
between 1. 0 and 100 mH ionic strengths 7 • Uolecular weight deter-
minations over the same range of ionic strength indicate that
dissociation is not taking place. Figure 2 contains a summary of
the hydrodynamic data (i.e., S20 and D20 ) over the ionic
strength range. Transitions are'~bservea a¥ approximately 1 and
10 mM, and are reversible.
In the absence of additional data it is not possible to ascribe

these changes in S, D and f to changes in particle shape and/or
particle volume (and hydration). As is usually done, one can make
certain reasonable assumptions and examine the consequences. We
10
~ ~
40-'~ , /
8
30
.~/
::£: ~~ ~
d
/ !-------- t----i
0 C\I
I 6cn
0
•I
•
,/
o
20
4
10
o o
o 2 4 6 8 10
UREA (M)
Figure 1. Hydrodynamic properties of vI as a function of urea mo-

larity. [n] units are cm 3 /g; sedimentation coefficients corrected
to S20,w'
:J:
-,
A B o-<
TRANSITION :JJ
2
o
o
• -<
z
3.'1 ••
tI __•__ _.... l>
e/- •• s:
3.2
.,. I
. C'5
l>
(/)
J, • z
19 TRANSITION .". o
a:: 0; 3.0 en
TRANSITION ,. I
w ~ •, •
III 11.2 2 ~~ U "m
("')
o ~~-.-. LL 28 -I
w :JJ
> / •
I 3
(/) lOS I./ I
1._._e __... .,; oen
(. ("')
I
3
TRANSITION I
•
I
0
~ 2.6
, I
•
. o
2104 I
(/) 1 I C'"5
24
.'~ ."'" \ -I
, m
100 •, ("')
I
1e/...-.--.- -.~ :J:
22 ~ Z
I I I
I • I
96~ / I o
I C
I .,./ -_/ m
2.0 .. en
92i-
• ••
-3 -2 -1 LSi -3 2 -I
Log (IONIC STRENGTH)
Figure 2. A. S20 w plotted as a function of the logarithm of the ionic strength. Initially the
monomer partic1es'were prepared in 0.2 mM EDTA (pH = 7.0), at an ionic strength of 1.0 mM. Then
the ionic strength was increased by additions of 0.25 M KC1 in 0.2 mM EDTA. The temperature was
regulated close to 20°C, and the speed was 48,000 rpm. B. D20,w plotted as a function of the log-
arithm of the ionic strength. The monomer particles were prepared in 0.2 mM EDTA (pH = 7.0) at an
ionic strength 1.0 mM. Then the ionic strength was increased by additions of 0.25 M KC1 in 0.2 mM
EDTA. The autocorrelation function was measured and fit to a single exponential plus baseline to
obtain D.
w
114 D. E. DUNS
TABLE 2. Estimated Shape Changes of vI at Different Ionic

Strengths
mM x 10 13 x 10 8 Oblate Prolate
Ionic Strength S f 20 ,wt f/fot
20,w alb alb
< 1.0 9.40 12.6 1. 22 4.9 4.S
1.0- 10.0 10.OS 11. 8 LIS 3.8 3.S
10.0 - 100.0 l1.0S 10.7 1.04 2.0 1.9
t
Calculated from:
S
Assuming: 1~ = 2.1 x 10 g/mole
v = 0.661 ml/g
p 0.998 g/ml
N 6.02 x 10 23 particles/mole
t Calculated from: f
o
6 TIn
o
(3
4TI
v) 1/ 3 = 6 TIn r
00
Where:
Assuming:
°1 1. 3 g H20/ g vI
0
v 1.00 ml/g
no 10- 2 g/sec-cm
o
Yields: r S4.6 A (radius of an
0
equivalent hydrated
sphere)
-8
f 10.29 x 10 g/sec
o
HYDRODYNAMIC AND SPECTROSCOPIC TECHNIQUES l1S
will assume that vI is hydrated to an extent similar to ribosomes

(i.e., 01 = 1.3 g H2 0/g vI) and that particle hydration does not
significantly change over the ionic strength rangeS. Table 2 is
a compliation of computations, interpreting the ionic-strength
dependence of sedimentation.
An oblate ellipsoid with axial ratio 2/1 would be in reasonable

agreement with current data on the shape of vI by low-angle neutron
scattering and x-ray crystallography 6. The particle could become
more oblate or more prolate with decreased ionic strength. Collabo-
rative studies are in progress with J. P. Baldwin and E. li. Brad-
bury, employing low-angle neutron scattering on vI at different
ionic strength, to permit a clearer selection of the many alterna-
tive models of particle conformational change.
pH Effects at Different Ionic Strengths. Spectroscopic studies

of vI as a function of pH (discussed below) have suggested that its
conformation is reasonably stable over a wide pH range (~5-9), with
a drastic and irreversible transition < pH 4.5. Parallel hydro-
dynamic studies 9 indicate that to some-extent, decreased pH can
stabilize VI against the conformational effects of low ion~c str-
ength. S20,w was measured on VI over the pH range (4.5-9) at three
different ionic strengths corresponding to the stable conformational
states at neutral pH (i.e., 1, 5.5 and 11 mM; see Figure 2). At
11 mM salt, S20,w remains constant at ~ 11.1 over the entire pH
range. At 5.5 nL~ salt, S20 w ~ 10.2 down to pH 5.4; a transition
toward a more compact conformation occurs between pH 5.4 - 5.0.
A similar sharp increase in S20,w between pH 5.4 - 5.0 is observed
in 1 mU ionic strength. Neither the exact conformational states
nor the mechanism of stabilization are understood. Weintraub 10
has reported that in 2 M NaCl buffers, the heterotypic tetramer
(H4, H3, H2A, H2B) converts to a mixture of the homotypic tetramer
(H3, H4h, H2A and H2B. It is conceivable that similar pH effects
occur within the nucleosome; stabilization of H4, H3 contacts at
low pH, tending to oppose the low-ionic strength unfolding of VI'
Spectroscopic Techniques
Our laboratory has utilized a number of spectroscopic techniques

to examine changes in conformational states of vl under varying sol-
vent conditions or perturbing treatments. A combination of circular
dichroism (CD) and laser Raman spectroscopy has been especially use-
ful in examining the secondary structure of the histones and the
DNAIl. The value of these methods arises from the considerable
background of reference spectra on model and on natural macro-
molecules. A detailed understanding of the theory of each method
is beyond the scope of this chapter, and unnecessary for the present
empirical level of data interpretation. A brief comparison of CD
and laser Raman is sufficient for the present purposes.
116 D. E. DLiNS
A 10 B
2
:0
....-0.
.: ....
0 o .. - - ..- - -
-2
-10
-4
rr0 r0
I
o
x -6 x
<{
<{
a.. -20
:§:
§: -8
-10 -30
-12
-40
-14
200 220 240 260 200 220 240 260 280 300 320
WAVELENGTH (nm)
Figure 3. (A) Circular dichroic spectrum at 25°C of inner histone

tetramer in 2.011 NaC1, 10 ~f Tris (pH 7), and 0.1 niH DTT. The
total concentration of amino acid residues was estimated by quan-
titative amino acid analysis of an HC1 hydrolyzate. Data are
expressed as molecular ellipticity per mole of amino acid residues,
[6]AA. (B) Circular dichroic spectra at 25°C of VI in 0.2 roM
EDTA (--) and in 2.0 H NaCl, 10 rolf Tris (pH 7), and 0.1 roM DTT
(----) compared to the spectrum of chicken DNA ( •..• ) in the same
2.0 11 NaC1 buffer. Data are expressed as molecular ellipticity
per mole of DNA phosphate [6]p' assuming the absorptivity per mole
of DNA phosphate (E p ,260) is equal to 6500 for DNA and for VI.
Circular dichroism measures the difference in absorbance

between left and right-handed circularly polarized light. Differ-
ence in absorbance arises from asymmetry within molecules, espe-
cially from the helical structures of DNA and protein. In chroma-
tin and nucleosomes, the spectral region 240 - 300 nm is dominated
by DNA helix conformation; between 200 - 240 nm, by the considerable
proportion of a-helix within the his tones (Figure 3).
Laser Raman spectroscopy consists of measuring the vibrational

spectrum of inelastically scattered light. The Raman frequencies
or lines have been assigned to particular subgroups. For example,
a strong and prominent Amide I (~1650 cm- I ; arising primarily from
carbonyl stretch) associated with a 1I1eak Amide III (~1230 cm- I ;
arising from C-N-H in plane bending and C-H stretching modes),
indicates the presence of considerable protein a-helix. A more
quantitative estimate of the amount of a-helix can be obtained by
comparing spectra in H20 and in D20. Laser Raman spectra can also
distinguish effectively between the B and A genus of DNA structure.
B is characterized by a strong dioxy symmetric stretch of the P02
group (~1094 cm- I ) and a weak phosphodiester backbone stretch of
O-P-O (~830 cm- I ). Laser Raman spectroscopy requires small volumes
«10~1) of material at concentration 10 2-10 3 times greater than
those required for CD studies, and clear spectra can be obtained
on precipitates. Laser Raman and CD complement each other when
different solvents or additives are employed l2 . For example, due
to the high ultraviolet absorbance of dimethylsulfoxide, CD measure-
ments are impossible; whereas, useful "windows" are available to
laser Raman. Vibrational spectra of DNA, inner histones, vI and
chromatin are presented in Figure 4.
From these CD and laser Raman studies we were able to estimate

the amount and nature of protein and DNA secondary structure ll .
We concluded that nucleosomal DNA is in the B-genus, and that asso-
ciated inner his tones consist of ~50% a-helix and negligible S-sheet
structure. We further suggested that the a-helix is primarily con-
fined to the apolar globular regions of the his tones, which would
consist of a localized a-helix content of 70-80%, comparable to
myoglobin and hemoglobin. Trypsin studies of isolated nucleosomes,
described below, are consistent with this conclusion.
The introduction of site-specific extrinsic fluorescent chromo-

phore into a macromolecule is a widely used technique for probing
a localized chemical environment. It is necessary, of course, that
the probe does not damage the native molecular structure. The
existence of a single cysteine residue at position 110 in chicken
erythrocyte H3 afforded the opportuni ty I3 to place a thiol-specific
fluorescent probe [i.e., NPM or N-(3-pyrene) maleimide] in a well-
defined site. Due to the lack of reactivity of the H3 thiol within
the intact nucleosome, NP11 was allowed to react with VI in high
118 D. E. OLiNS
>- c
r-
U)
zw
r-
z
3000 2000 2'500 1500 1000 500

FREQUENCY (em-I)
Figure 4. Laser Raman spectra at 32°C of (A) chicken DNA (4 per-
cent by weight) in 2.0 M NaCl; (B) inner histones (9 percent) in
2.0 M NaCl, 10 roM Tris (pH 7), and 0.1 roM dithiothreitol (DTT);
(C) vI (10 percent) in 0.2 m~ EDTA: (D) vI (10 percent) in 2.0 M
NaCl and 0.2 roM EDTA; and (E) chicken erythrocyte chromatin (26
percent) in 0.2 roM EDTA. Conditions: excitation wavelength, 488.0
nm; spectral slit width, 10 em-I; radiant power, 300 row; amplifi-
cation, A = 1 (300 to 1800 em-I), A = 3 (2500 to 2600 em-I).
Abbreviations: str, stretching; def, deformation; A, T, C and G,
adenine, thymidine, cytosine, and guanine; P, phosphate, phe,
phenylalanine; tyr, tyrosine; and Am, amide.
NaC1, followed by reassociation and purification of the resulting

NPM-vl' The emission spectrum of HPM-vl exhibited peaks at 374
and 392 nm, and a long wavelength fluorescence at 460 nm (excimer).
This "excimer" fluorescence, believed to arise from a pairwise
association of pyrene groups attached to the two thio1s of an in-
tact nuc1eosome, has proven to be remarkably sensitive to confor-
mation changes induced by various perturbants. Presumably this
sensitivity of the probe arises from the stringent geometric con-
straints required for energy transfer within the excited state.
Changes in the quantum yield of monomer fluorescence are also in-
formative of the extent of exposure and quenching of the NPM group.
The utility of these three spectroscopic techniques to examine

the conformational transitions of vI in response to various treat-
ments and perturbants will be illustrated with the following
examples.
Trypsin Digestion of v). Previous studies have indicated that

trypsin rapidly degrades the N-terminal basic portions of the inner
histones in chromatin or mononucleosomes 10 • Approximately 25% of
the total histone sequences are digested. On the basis of CD and
laser Raman studies we have suggested that very little a-helix is
contained within the N-terminal basic tails. In a recent study 14
we have confirmed this suggestion by monitoring the CD of VI during
the course of digestion with trypsin (Figure 5). It is evident
that by 60-90 minutes all of the native inner his tones have been
degraded, with virtually no decrease in a-helix content. On the
other hand, the "suppression" of the DNA spectrum is clearly de-
pendent upon the integrity of the histone basic tails.
Urea Effects. Hydrodynamic studies, described earlier, have

demonstrated that the addition of urea (0-10 M) to VI results in
a progressive non-cooperative increase in the frictional coeffi-
cient (f), consistent with progressive swelling or increase in
particle asymmetry. CD studies 6 , on the other hand, demonstrate
that the a-helix content of VI remains unchanged up to ~4~1 urea,
followed by a cooperative destabilization of a-helix between 4-7 M.
Spectrofluorometric studies of NPl1-vl as a function of urea concen-
tration substantiate the usefulness of this fluorescent probe
(Figure 6). Excimer fluorescence disappears at low urea concen-
trations during moderate and reversible swelling of VI' Uonomer
fluorescence is quenched simultaneously with the destruction of
a-helix (i.e., between 4-6 M urea), presumably due to destruction
of the solvent-protected environment around the pyrene chromophore.
120 D. E. OLINS
'Or-~r-~---.---.--~--~~
A '",
9
o
O~----------------------~
-10
- 10
,
'" -20
o
a:
'"
~ -30
o 60 120 180 240

TIME (min)
-40
-50
200 240 280 320

WAVELENGTH (nm)
H3
H2B~
H2A-
H4 ;,--
, )))) I I I ~l--.
CO 15306090120150180 240
c DIGESTION TIME (min)
Figure 5. (A) Circular dichroic spectra of trypsin-digested vI'
Samples: chicken DNA (_ •. _); vI (-); vI' digested 60 min (----);
vI' digested 120 min (---.---); trypsin + soybean trypsin inhibitor
(---------), at some concentration as present with digested vI'
Solvent systems: vI' 5 roM Tris (pH 8.0); DNA, 65 roM NaCl, 4.75 mM
Tris (pH 7.3). (B) Kinetics of changes in molecular ellipticities
at 281 nm (DNA conformation) and 223 nm (a-helix content), with time
of digestion by trypsin. (C) SDS gel electrophoresis of native and
trypsin digested vI' After varying periods of digestion with tryp-
sin at room temperature, soybean trypsin inhibitor was added.
Abbreviations: C, control vI; 0-240 min, digested vI; T, doublet
band of trypsin and trypsin-inhibitor; p, limit tryptic fragments
of the inner histones.
>-
f-- A B
(/)
Z
W
f--
Z
W
U
Z
W
U
(/)
w
a:
0
:::J
...J
u..
W
>
f=
<{
...J
W
a: - -= ~.:-:- .::.:--= ,- -
400 500 600 0 2 468 to
WAVE LENG TH ( nm) URE A ( M)
Figure 6 . (A) Fluorescence emission spectra of NPH-v 1 (G or good)

complexes at different urea concentrations . Urea concentrations:
o (--) ; 2 . 83 (----) and 6.71 (-_._-) M urea , containing 0.2
m!! EDTA (pH 7. 0). Spectra are presented at constant NMP-vl con-
centrations. Excitation wavelength : 340 nm . (B) Fluorescence
intensity changes as a function of urea molarity . Samples:
NPM-vl (G) at 374 nm (A---A) and 460 nm (t---e); NPM-vl (B or bad)
at 374 nm ( ~---~) and 460 nm (0---0); NPM-B-mercaptoethanol at
374 nm (0 - - 0) . G and B complexes are normalized to the same
DNA concentration; NMP-B-mercaptoethanol at the same NPM concen-
tration (based upon A340 nm) as the G fraction. All solvents
contained 0.2 m!1 EDTA (pH 7.0). Excitation wavelength: 340 nm.
Ionic Strength Effects. Spectrofluorometric and circular

dichroism studies have been conducted in parallel to the hydro-
dynamic data described above 9 '13. NPIf-vl exhibits a decrease in
excimer, and an increase in pyrene monomer fluorescence when the
ionic strength is decreased in the vicinity of transition 1 (i.e . ,
~10-3 If). Apparently unstacking of the pyrene groups is promoted
by the lowered ionic strength. A less well-defined fluorescence
changes is also observed in the vicinity of transitions 2 (i.e.,
~10-2 M), with a sharp decrease in both monomer and excimer fluo-
rescence progressing from low to high ionic strength. Thus, the
local environment of the pyrene chromophores reflects the larger
scale hydrodynamic conformational transitions.
Preliminary CD studies of v I over the range 10- 3-10- 1 M ionic

strength indicate essentially negligible change in a-helix content,
and small, but measureable effects on the DNA contribution to the
spectrum. The higher ionic strength conformations of v I exhibit
122 D. E. OLINS
0.8
,,, ....
(.
06
• --- - . - ---o--c~~(9 -- . e--.
0.4
C.2
9
B
8 .---. ---- .• ---- . ---. ---- . - -. ....
~.
,,
'~
. .,
7
.",. 4,t
6
,
, ;
I \
\
,
\
.\\ I •
~
5 f\1
4
I'•
\I
2
c
- to
'"
I
2
~ -20
'"
N
::: - 30
- 4C
- 50
9 8 7 6 5 4 3
pH
Figure 7. Absorption and circular dichroism of DNA and vl as a

function of pH. (A) A2 GO for DNA (I) and vl (0), A325 for DNA (A)
and vI CLl). (B) [eJ p 281 for DNA (0) and vI (0), [eJ p 223 for vI
(0). 4 . 75 and O. 5 m.~ ' buffers were used for DNA and vI: respectively.
Buffer solutions in the pH range 2-9 were prepared by using the
following systems: Tris-HCl; sodium cacodylate-cacodylic acid;
sodium acetate-acetic acid and glycine-HCl.
slightly greater "suppression" of the DNA signal, than is observed

at lower salts.
pH Effects. Figure 7 illustrates a summary of recent spectro-

scopic studies of vl conformation at different pH. Progressing from
neutral pH towards acid, several regions can be discerned: (A) pH
7.5 - 5.5, negligible changes in DNA conformation or a-helix content;
(B) 5.5 - 4.8, "suppression" of DNA ellipticity with no loss of
a-helix; (C) 4.6 - 4.2, sharp increase in DNA ellipticity towards
the conformation of naked DNA in solution, with slight loss of a-
helix; (D) pH 4-3.8, aggregation of vl. V. Gordon et a1 9 has care-
fully compared these CD observations with hydrodynamic properties
of vl. and concluded that region (B), the "suppression" of DNA
ellipticity, parallels the pH-induced increase in S at low ionic
strength. Laser Raman studies l4 demonstrate that protonation of
the bases C and A in DNA or vl does not occur significantly until
below pH "'4. The mechanism and reversibi1it) of these pH-induced
conformational transitions remains to be i11ucidated. There is,
however, a clear positive correlation between hydrodynamic "com-
pactness" of vl' and "suppression" of DNA conformation as assayed
by circular dichroism.
CONCLUSION
Combining hydrodynamic and spectroscopic techniques in the

study of conformational states of vl induced by a variety of
perturbants has led us to a general conception: the two structural
domains of vl (i.e., the DNA-rich outer shell and the a-he1ix-rich
apolar histone core) exhibit differential responsiveness. In
general, the a-helical regions are more resistant, than DNA con-
formation or vl size and shape, to the perturbing effects of urea,
decreased ionic strength and pH, trypsin treatment, or a variety
of water-miscible organic solvents. There are a number of reason-
able conceptual models to explain this differential responsiveness
of the structural domains of vl. A few of these models can be de-
noted with self-descriptive names: e.g., "shell-swell", "open-
clam", or "unravelled". Distinction between the various models
will require data from many other biophysical techniques, such as
neutron scattering studies. Whatever the exact geometry of the
transitions, it is tempting to speculate that the a-helical histone
core constitutes the "restoring force" of the nuc1eosome--returning
the conformationally perturbed nuc1eosome of its "inactive" compact
form.
ACKNOWLEDGEMENTS
The author is grateful to his many colleagues and collabora-
tors. Their names can be found, in detail, in the Bibliography.
Research sponsored jointly by research grants GMl9334 (DEO); PCM77-
124 D. E. OLiNS
21498 (ALO); AIl1855 (GJT); and GM13914 (VNS) and the Division of
Biological and Environmental Research, U.S. Department of Energy
under contract W-7405-eng-26 with the Union Carbide Corporation.
BIBLIOGRAPHY
1. Cold Spring Harbor Symposium of Quantitative Biology (1977)

XLII Cold Spring Harbor, NY.
2. Felsenfeld, G. (1978) Nature 271, 115-121.
3. Tanford, C. Physical Chemistry of ~fa.cromolecules (1959)
John Wiley & Sons, Inc., NY.
4. Van Holde, K. E. Physical Biochemistry (1971) Prentice-Hall,
Inc., Englewood, Cliff, NJ.
5. Ford, Jr., N. C. (1972) Chemica Scripta 1, 193-206.
6. Olins, D. E., Bryan, P. N., Harrington, R. E., Hill, W. E.
and Olins, A. L. (1977) Nucleic Acids Res. ~, 1911-1932.
7. Gordon, V. C., Knobler, C. :ti., Olins, D. E., and Schumaker,
V. N. (1978) Proc. Natl. Acad. Sci. USA, 75, 660.
8. 01 ins , A. L., Carlson, R. D., Wright, E. B.-,-and Olins, D. E.
(1976) Nucleic Acids Res. 3, 3271-3291.
9. Gordon, V. C., Knobler, C. M., Olins, D. E., and Schumaker,
V. N. (1978) Biophys. J. 21, 66a.
10. Weintraub, H., Palter, K., and Van Lente, F. (1975) Cell~,
85-110.
11. Thomas, Jr., G. J., Prescott, B. and Olins, D. E. (1977)
Science 197, 385-388.
12. Zama, M., Olins, D. E., Prescott, B., and Thomas, Jr., G. J.
(1978) Biophys. J. 21, 66a.
13. Zama, 11., Bryan, P. W:-, Harrington, R. E., Olins, A. L., and
Olins, D. E. (1977) Cold Spring Harbor Symp. Quant. BioI.
XLII, in press.
14. Zama, M., Olins, D. E., Prescott, B. and Thomas, Jr., G. J.
(1978) manuscript in preparation.
PREPARATION AND ANALYSIS OF CORE PARTICLES AND NUCLEOSOMES: A
CONVENIENT METHOD FOR STUDYING THE PROTEIN COMPOSITION OF
NUCLEOSOMES USING PROTAMINE-RELEASE INTO TRITON-ACID-UREA GELS
Barbara Ramsay Shaw and Randall G. Richards
Paul M. Gross Chemical Laboratory
Duke University, Durham, NC 27706
The nucleosomal repeat structure of chromatin consists of a

core of 140 base pairs of DNA wrapped around an octamer of eight
histones, and a spacer region which is variable in length and more
susceptible to nuclease attack than the core 1 - 10 • Histone Hl,
which exhibits tissue specific variation, has been postulated to
be associated with DNA in this spacer region 3 , 4 , l O - 1 3 . The
octameric histone complex in the core particle contains two copies
each of his tones H2A, H2B, H3 and H4. It is known that these core
his tones undergo postsynthetic covalent modifications such as
acetylation and phosphorylation at different times during the cell
cycle 14 • Recently, Cohen et al. 15 have shown in the developing
sea urchin and Franklin and Zweidler 16 have shown in the dif-
ferentiated tissues of mammals that there are also variations
within the primary sequence of his tones H2A, H2B, and H3. The
molecular differences between variant his tones arise from the
conservative substitution of one or two amino acid residues in the
polypeptide chain which can be detected on Triton/acid/urea gels.
In the sea urchin system, different H2A and H2B variants are
synthesized at different stages of embryonic development 15 • The
existence of multiple forms of his tones implies that there must be
multiple forms of nucleosomes.
Our laboratory has investigated methods of separating and

identifying different subsets of nucleosomes and 140 base pair
core particles. We report here the utility of polyacrylamide gel
electrophoresis in carrying out systematic characterization of the
homogeneity of such preparations. In particular, samples of
125
126 B. R. SHAW AND R. G. RICHARDS
nucleoprotein particles can be well resolved electrophoretically

into component particles that differ by their histone content as
well as the size of their DNA. Todd and Garrard 13 and Bakayev et
al. 14 have reported that nucleosome particles having discrete
sizes of DNA can be separated with polyacrylamide gel electro-
phoresis. We have used similar type gels to resolve the various
structural types of nucleoprotein momomers, dimers and higher
oligomers that result from digestion of chicken erythrocyte
chromatin. We have observed that the resolution of monodisperse
nucleoprotein particles obtained with nucleoprotein electropho-
resis is comparable to that obtained by isolating and electropho-
resing the component DNA. Such results suggest that nucleoprotein
particles obtained from the digestion of chromatin with micro-
coccal nuclease are produced only in discrete packages, wherein
the size of the DNA and the protein content of the products are
found in only certain allowable combinations.
Presented here is a high resolution two-dimensional gel

electrophoresis system which has been developed for rapid frac-
tionation and analysis of histone variants and other acidic
proteins found in the numerous products that result from digestion
of chromatin or nuclei with micrococcal nuclease. The fundamental
premise that underlies this procedure is that the complete removal
of histone proteins from DNA can be accomplished by treating
deoxyribonucleoprotein with a one-step protamine treatment dis-
cussed below. By combining gel electrophoresis of DNA-protein
complexes in low ionic strength buffer in the first dimension with
a protamine-released Triton X-IOO/acid/urea gel electrophoresis in
the second dimension, this procedure can readily screen for varia-
tions in the histone and other acidic protein components that
arise from discrete kinds of mononucleosomes, dinucleosomes and
higher oligonucleosomes found in nuclei of many cells. The tech-
nique is easier and potentially more powerful than most of the
current fractionation procedures, which employ columns or sucrose
gradients to fractionate chromatin digests. The time-consuming
collection, concentration and analysis steps required for.deter-
mining the histone components in each nucleosome fraction can be
avoided. More important, in one gel all the nucleoprotein products
from one time point of digestion can be analyzed.
METHODS AND MATERIALS
(A) Isolation of Erythrocyte Nuclei. Blood was obtained

from adult White Leghorn chickens by cardiac puncture in the
presence of 0.15 M NaCl-0.015 M Na citrate, pH 7.2 (saline/citrate).
After centrifugation at 3000 x g for 10 min at 4°C, the plasma and
buffy coat were removed. The erythrocytes were washed twice with
the isotonic saline solution and frozen at -60° until needed. The
frozen erythrocytes were thawed at 37° in an equal volume of
saline/citrate and centrifuged at 3000 x g for 10 min at 4°C;
CORE PARTICLES AND NUCLEOSOMES 127
the nuclear pellet was resuspended in 0.25% Nonidet P-40 in

saline/citrate. Throughout, 0.1 mM phenylmethane sulfonyl fluo-
ride (PMSF) was added to inhibit protease action. The nuclei were
repelleted, washed with saline-citrate and repelleted. A more
detailed procedure can be found on pp. 78-79 of Ref. 23.
(B) Digestion of whole chromatin. Freshly prepared nuclei

from 6 ml of packed frozen red blood cells were, after the above
treatment, resuspended in about 200 ml of digestion buffer (0.3 M
sucrose-0.75 mM CaC1 2 -lO mM TriseHCL, pH. 7.2) at a concentration
of 2 x 10 8 nuclei per mI. Digestion of the nuclei by micrococcal
nuclease (Worthington) was carried out at 37° with 125 units of
nuclease per ml of nuclei suspension. The digestion reaction was
terminated by making the solution 2 mM in EDTA.
(C) Isolation and digestion of Hl/H5-depleted chromatin.

Following the method of Tatchell and Van Holde 24 , nuclei from 10
ml of packed red blood cells from part (A) were lysed with stir-
ring in 250 ml of 10 mM Tris-cacodylic acid - 0.1 mM EDTA - 0.1 mM
PMSF (pH 7.2) (lysis buffer). After several hours, the solution
was made 0.6 M NaCl with solid NaCl and gentle stirring, and
allowed to swell overnight at 4°C. The chromatin gel was centri-
fuged at 18,000 x g for 45 min; the pellet was resuspended with
gentle stirring in 400 ml of lysis buffer containing 0.65 M NaCl,
allowed to stand 24 hr. and pelleted at 18,000 x g for 15 min.
This last step was repeated one time. The pelleted chromatin
(~3000 A260 units) was brought up in 400 ml of lysis buffer and
stirred for 2 hrs. The chromatin, now in low ionic strength
buffer, was centrifuged at 18,000 x g for 30 min, brought up with
a Dounce homogenizer in 20-25 ml of lysis buffer containing 1 mM
CaC1 2 , and the 3000 A260 units digested at 37° with 125 units/ml
of micrococcal nuclease (Worthington). The reaction was termi-
nated after about 40 minutes by making the solution 10 mM EDTA and
cooling on ice.
(D) Fractionation of nucleosomes. The digested chromatin

from part (C) was pelleted at 10,000 x g for 10 min. Monomer and
dimer nucleosomes in the supernatant were isolated with gel chro-
matography on Bio-Gel A-5m by eluting with 10 mM Tris, 0.7 mM
EDTA, pH 7.2. 22 ,23 The core particles were obtained in a well
resolved peak in the included volume; however, some contamination
with Hl/HS-depleted dimer nucleosomes occurred. The peak fractions
containing about 300-400 A260 units of monomer were pooled, con-
centrated to 100 DD/ml, dialyzed in an Amicon concentrator and
fractionated on 10 mM Tris, 0.7 mM EDTA, pH 7.2 sucrose gradients
isopycnic for particle density 1.51 25 to yield mainly purified
monomer and some dimer nucleosomes.
(E) Electrophoresis of nucleoproteins. Nucleoprotein elec-

trophoresis was carried out at 25°C using 4.0% acrylamide
(acrylamide: N, N'methylene-bisacrylamide ratio, 25:1) tube gels
(12 x 0.3 cm) or slab gels (13 x 0.15 cm). We have found that
this acrylamide:bis ratio gives excellent resolution of nucleo-
somes in the monomer/dimer region. The buffer used to make the
gel and circulate in the electrophoresis apparatus was that used
by Todd and Garrard 13 , i.e. 6.4 roM Tris, 3.2 roM sodium acetate,
0.32 roM EDTA, pH 8.0. Samples of ionic strength less than 5 roM
were loaded in 10% glycerol, 2 roM EDTA, .025% bromphenol blue.
Sample loads contained on the order of 10-25 ~g each of DNA and
protein for one-dimensional slabs and 20-50 ~g each of DNA and
protein for one-dimensional tubes. Electrophoresis with recircu-
lating buffer was 15 min at 70 V followed by 4 1/2 hr at 90 V for
slab gels, and 15 min at 50 V followed by 5 1/2 hr at 60 V for
tube gels. Electrophoresis of DNA was carried out according to
Loening 18 • For visualization of DNA, gels were stained with 1
~g/ml of ethidium bromide in water for 1/2 hr and photographed on
a Model C-62 short wave ultraviolet light box (Chromato-Vue trans-
illuminator, UV Products) with Polaroid 107 or 55 film and a red
filter. For visualization of proteins, gels were stained in
Coomassie Blue R-250 following Fairbank's staining procedure 19 •
(F) Protamine-release method on Triton/Acid/Urea gels.

Electrophoresis of histone variants was performed at 25°C with
Triton/acid/urea 2o slab gels containing 12% acrylamide (acryl-
amide:bis ratio of 30:0.2), 8 M urea, 6 roM Triton X-IOO and 5%
acetic acid. The plates are sealed with 15% acrylamide instead of
agarose since the latter interferes in the staining of the gels.
Slab gels were 13 cm x 13 cm glass plates, and either 0.15 cm
thick if used for one-dimensional gels or 0.30 cm thick if used
for second-dimensional gels. Gels were pre-electrophoresed for 5
hr at 125 V with 5% acetic acid running buffer. The gel was then
scavenged for 1 1/2 hr at 125 V with 0.1 mls of a solution of 8 M
urea, 0.3 M cysteamine and 0.3 mg/ml protamine in each slot for one-
dimensional slabs or a total of 2 ml per gel for second-dimension
slabs. A second scavenge of 15-20 min duration was performed with
a similar volume of 0.6 M cysteamine. The remaining scavenging
solution was removed from the slots and the buffer replaced with
fresh 5% acetic acid.
Protamine release of his tones from nuclear digests and nucleo-

somes into Triton/acid/urea gels was accomplished as follows.
Samples of ionic strength less than 10 roM were made 8 M in urea-5%
in acetic acid-2 1/2% in thioglycolic acid-5% in B-mercaptoethanol
(sample buffer) and 0.1% in protamine sulfate from Salmon-sperm
(histone-free, Sigma) and applied directly to the gel. Electro-
phoresis was for 8 1/2 hr at 150 V. For two-dimensional electro-
phoresis to display histones, first-dimension nucleoprotein tube
gels (12 cm x 0.3 cm d.) were soaked for 1/2 hr in 5% acetic acid,
5% B-mercaptoethano1, 2 1/2% thiog1yco1ic acid and 8 M urea. The
tube was laid horizona11y across the top of a preformed and
prescavenged Triton slab (13 x 13 x 0.3 cm) which had a 10 cm long
middle slot to hold the tube gel, and a small slot at each end to
serve as sample well for a standard. The tube gel was carefully
over1ayered with 200 ~t of a solution of 8 M urea, 5% acetic acid
acid and 1% protamine sulfate. Electrophoresis was for 13 hr at
100 V.
RESULTS
Separation of nuc1eosomes. Our first dimensional nucleo-

protein gel is a 4% polyacrylamide gel system which utilizes the
low ionic strength buffer system described by Todd and Garrard 13 •
We have found that an acry1amide:bis ratio of 25:1 gives excellent
resolution of nuc1eosornes in the region of monomer and dimer sub-
units. In Fig. 1 we have electrophoresed samples of monomer (core
particles) and dimer nuc1eosomes obtained from column and sucrose
gradient fractionation of a micrococcal nuclease digest of H1/H5-
depleted chicken chromatin. On the left the samples were electro-
phoresed as DNA and the size calibrated with Hae III-PM2 rest ric-
tionfragments; on the right similar samples were electrophoresed
as nuc1eoproteins. Note the sharply defined monomer and dimer
nucleoprotein bands in slot B; these bands are as sharp or sharper
than the corresponding DNA bands in slot No. 2 on the left half of
Fig. 1. The dimer nuc1eosome seen in slots A and B has a DNA size
of 270 ± ~10 base pairs as measured in slots No. 1 and 2, and
corresponds to the "tight" (spacer1ess) dimer seen by Tatche11 and
Van Ho1de 24 .* Overloading of a core particle sample (slots C and
D) reveals the presence of a small amount of a second monomer
species that migrates slower than the core. The DNA in this minor
band migrates at about 160-170 base pairs (not shown). As can be
seen, when the gel is not overloaded, the nucleoprotein gel is
capable of separating nuc1eosomes with a resolution nearly com-
parable to that obtained with the component DNA. The sharpness of
our nucleoprotein bands implies that a broad band or series of
closely spaced bands within the monomer/dimer region would cor-
respond to different forms of nuc1eosomes.
*K1evin and Crothers have reported a dimer subset in H1-

depleted calf-thymus chromatin that has a DNA size of 240 base
pairs28.
DNA GEL NUCLEOPROTEIN

GEL
-trimer
332 - tight dimer
299
273
core
166_
IIfB-
1 2 3 4 ABC D E
Figure 1. (Right). Nucleoprotein gel showing chicken erythrocyte

(core particle) monomer, dimer and trimer nucleosomes. The
chromatin was HI and H5 depleted, digested with micrococcal
nuclease, and fractioned on an A-5m column followed by a 10 roM
Tris, 0.7 roM EDTA, pH 7.2 sucrose gradient. Samples were applied
directly to a 4% polyacrylamide (25:1) gel: A. Monomer, dimer
and trimer; B. Monomer and dimer; C. Overloaded monomer contain-
ing two species; D. Purer (core particle) monomer (overloaded
on this gel); E. Same sample as D. (Left). DNA extracted from
nucleosome samples of the same sucrose gradient as above and
electrophoresed on 4% polyacrylamide DNA gels. 4 ,18 1. Monomer,
dimer and trimer; 2. Monomer and dimer; 3. PM2-Hae III restric-
tion fragments; 4. Overloaded monomer sample.
Protamine Release of Proteins into Gels. In order to analyze

the histone content of nucleosomes, it has been necessary in the
past to either extract the histones from chromatin with acid 21 ,
or alternatively treat with SDS and electrophorese the proteins on
SDS gels 26 • The method of acid extraction is time consuming and
can result in selective extraction of certain histones 21 • The
SDS electrophoretic method is quite reliable; however, only the
five major histone classes can be resolved. We have developed a
method in which variant and covalently modified his tones can be
easily and quanitatively solubilized for analysis on Triton/acid/-
urea gels 1S which resolve at least 15 discrete forms of histones.
In order to release his tones from nucleosomes into Triton/-

acid/urea gels we have tried various solubilizing reagents,
including spermine, spermidine, cetyltriethylammonium bromide,
DNAase I (which Bafus et al have recently studied 27 ) , and protamine
sulfate from salmon sperm. Protamine sulfate is the best releas-
ing agent for our studies. Fig. 2 is a comparison of two of the
most effective protein-releasing reagents: spermine and prota-
mine. The sample applied to each slot was a multimeric subunit
digestion product from Hl/H5-depleted chicken chromatin. As seen
on the left side of Fig. 2, histone H4 is poorly released from the
spermine-solubilized sample at all concentrations of spermine
employed. On the right is the same sample using protamine as the
releasing agent. Note that concentrations as low as .05% prota-
mine are effective in displacing his tones and other proteins from
the DNA. Perhaps more important for electrophoretic analysis is
that the time needed for protein displacement is quite rapid using
protamine.
PROTEIN RELEASE
PROTAMINE
8 9 10 II
Figure 2. Triton-acid-urea gel illustrating protamine and

spermine release of protein from A-5m multimer fractions of a
micrococcal nuclease digestion (of HI and H5-depleted chicken
erythrocyte chromatin). Slots No.1 thru 7 correspond to samples
solubilized for 1 hr in 1, 2, 3, 4, 5, 7.5 and 10% spermine plus
sample buffer. In slots No.8-II the samples were solubilized in
sample buffer plus 0.05, 0.1, 0.1 and 1% protamine, respectively.
Whereas samples No. 8 and 9 were incubated at 25°C for 1 hour
prior to electrophoresis, samples No. 10 and 11 were applied to
the gel within 5 minutes after the addition of protamine.
Protamine is equal to or better than acid extraction, as can

be seen in Fig. 3. For comparison, four different nucleoprotein
samples are electrophoresed, in pairs. Acid extracted samples are
the left sample in each pair. In all cases, the protamine treat-
ment equaled or superceded the acid extracted samples in the
amount and number of proteins released from the chromatin.
Figure 3. Comparison of acid extracted (A) and protamine (P)

solubilized protein from various nucleoprotein samples. The (A)
samples were extracted from 1% SDS, 4M urea, 0.4N H2 S0 4 with
stirring, precipitated with ethano1 21 , and solubilized in sample
buffer (no protamine). The (P) samples were used directly. They
were either in column buffer (No. 1 and 2) or in 10 mM Tris-3 mM
EDTA-0.7S mM CaC1 2 - 0.3 M sucrose, pH 7.2 and all were adjusted
prior to electrophoresis with two volumes of 1 1/2 x sample
buffer plus O.lS% protamine. Pair 1: A-Sm multimer fraction of
Hl/HS-depleted chromatin; Pair 2: Degraded A-Sm monomer fraction
of a chicken erythrocyte nuclei digestion; Pair 3: Nucleoprotein
released from erythrocyte nuclei after a 4 min micrococcal
nuclease digestion; Pair 4: Erythrocyte nuclei digested four
minutes with nuclease.
Having shown that his tones as well as other proteins can be

displaced from chromatin with protamine for analysis on Triton/-
acid/urea gels, we have employed a variation of this method with
second-dimensional gels to screen for nucleosomes that differ in
protein composition . We have used this two-dimensional system to
analyze in Fig. 4 the chicken monomer and dimer sample shown in
Fig. 1, slot B. In the first dimension of Fig. 4 the nucleosomes
were electrophoresed as nucleoprotein. In the second dimension
the proteins were released with protamine from the nucleoprotein
gel and separated on a Triton/acid/urea gel as described in the
Methods. A nucleosome sample with protamine added is placed in
1° NUCLEOPROTEIN
,_. _ _ _ _ _ __ __ _ _...
) 4+1
--
. -
--
Sketcb
Figure 4. Two dimensional electrophoresis of monomer and dimer

from micrococcal nuclease digested Hl/H5-depleted chromatin from
chicken erythrocytes. (See figure 1, slot B) Nucleoprotein was
electrophoresed from left to right in a 0.3 cm diameter tube.
The tube gel was then soaked for 1/2 hr. in 8M urea, 5% acetic
acid, 5% 8-mercaptoethanol, and 2 1/2% thioglycolic acid and laid
horizontally across a preformed, prescavenged 0 . 3 cm thick Triton-
acid-urea slab gel. The tube was overlayered with 0.2 mls of 8M
urea, 5% acetic acid, and 1% protamine and electrophoresed for 13
hrs at 100V. Slots on the left and right correspond to protamine
released proteins from digested Hl/H5-depleted chromatin in
solution . (Arrows indicate migration of proteins which are not
visible in the picture.) Nomenclature follows Franklin and
Zweidler 16 •
the reference slots at the far left and far right of the gel. The
release of his tones from the first dimensional tube is comparable
to the protein in the reference sample. As can be seen, the histone
content of the chicken erythrocyte core monomer and "tight" 270
base pair dimer are practically identical (Fig. 4). This is not
unexpected since the erythrocyte is a quiescent cell. The simi-
larity in histone content of our core particle and "tight" dimer
suggest that the formation of this compact dimer is independent of
the nature of the component histone proteins.
Of special interest in Fig. 4 are the small amounts of non-

histones associated with the monomer and dimer species. These
proteins are difficult to visualize in the picture but are indi-
cated with arrows. Nonhistone proteins have been reported to be
associated with nucleosomes by several groups27,29,30.
DISCUSSION
In summary, the first dimension nucleoprotein gel electro-

phoresis system described here has been shown to separate monomer
and dimernucleosomes with high resolution. Moreover, the histone
as well as non-histone protein content of each class of these
nucleosomes can be readily examined using the second-dimension
Triton/acid/urea gel system described. The effectiveness of the
second dimension gel separation lies in the quantitative release
of proteins from the nucleoprotein complexes with protamine. We
have shown that protamine efficiently releases his tones and non-
histone proteins from nuclear digests and from chromatin subunits.
We have taken advantage of this protein solubilization to effect
second-dimension separation of his tones from various nucleosome
classes.
Using the method of protamine release to identify the histone

components of nucleosome classes, we have been able to observe
and compare different types of nucleosomes released from chromatin
by micrococcal nuclease. Certain mono- and di-nucleosome classes
contain no Hl or H5; some fractions contain either Hl or H5, and
most nucloesomes contain small amounts of non-histone proteins. 31
The nucleosomes that carry histones HI or H5, and nucleosomes that
contain core proteins plus certain non-histone protiens, migrate
slower than core particles on first-dimension nucleoprotein gels. 27 31
In the chicken erythrocyte chromatin, all classes of nucleo-

somes (within the limit of sensitivity of our first-dimensional
gels) appear to have similar amounts of variant core histones;
that is, the relative proportion of each core histone variant
found in total chromatin is similar to that found in mononucleo-
some, dinucleosome and multimer nucleosome classes. The
distinguishing features of chicken erythrocyte nucleosome classes

arise primarily from the size of the DNA and the relative amounts
of histones HI and H5. 31 In sea urchin embryo, however, we have
observed different types of nucleosome subsets that appear to be
derived from different variants of core his tones (to be submitted).
In summary, the two-dimensional gel method described here is

a very convenient method for the analysis of histone variants and
non-histone proteins in nucleosomes. The method has the potential
of easily screening numerous samples of nuclease digestion products
for the identification of modified and variant nucleosomes. Most
important here is the fact that with one gel all the variant and
modified histone proteins in nucleosome products produced at one
time of digestion can be analyzed, simultaneously, thus elimi-
nating the necessity of collecting and concentrating fractions as
one might do with sucrose gradient analysis.
ACKNOWLEDGMENTS
We wish to express our appreciation to Ms. Katherine Marsh

for excellent technical assistance. This research was supported
by grants from NIH (GM 23681), Duke University Biomedical Research
Science Support Grant, and by a Teacher-Scholar Award to B.R.S.
from the Camille and Henry Dreyfus Foundation.
REFERENCES
1. Van Holde, K. E., Sahasrabuddhe, C. G. and Shaw, B. Ramsay

(1974) Nucl. Acids Res. I, 1579-1586.
2. Van Holde, K. E. and Ise~berg, I., (1975) Accounts of Chemical
Research~, 327-335.
3. Van Holde, K. E., Shaw, B. Ramsay, Lohr, D., Herman T. M.
and Kovack, R. T. (1975) in Proceedings of the Tenth FEBS
Meeting, Vol. 38 Federation of European Biochemical
Societies, pp 57-72.
4. Shaw, B. Ramsay, Herman, T. M., Kovacic, R. T., Beaudreau,
G. S. and Van Holde, K. E. (1976) Proc. Nat. Acad. Sci. USA
73, 505-509.
5. Sollner-Webb, B. and Felsenfeld, G. (1975) Biochemistry 14,
2915-2920.
7. Axel, R. (1975) Biochemistry 14, 2921-2925.
8. Kornberg, R. D. (1974) Science 184, 868-871.
9. Oudet, P., Gross-Bellard, M., and Chambon, P. (1975) Cell
4, 281-300.
10. Shaw, B. Ramsay, Herman, T. M., Kovacic, R. T. and Van
Holde, K. E. (1976) in Molecular Mechanisms in the Control
of Gene Expression, ed. by D. P. Nierlich, W. J. Rutter and
C. F. Fox, Academic Press pp. 13-20.
11. Whitloch, J. P. and Simpson, R. T. (1976) Biochemistry 15,
3307-3314.
12. Varshavsky, A. J., Bakayev, V. V. and Georgiev, G. P. (1976)

Nuc1. Acids Res. 1, 477-492.
13. Todd, R. D. and Garrard, W. T. (1977) J. BioI. Chern. 252,
4729-4738.
14. See Elgin, S. C. and Weintraub, H. (1975) Ann. Rev. of
15. Cohen, L. H. , Newrock, K. M. and Zweid1er, A. (1975)
Science 190, 994-997.
16. Frank1in,~ G. and Zweid1er, A. (1977) Nature 266, 273-275.
17. Bakayev, V. V., Bakayeva, T. G. and Varshavsky,~ J. (1977)
Cell 11, 619-629.
18. Loening, U. E. (1967) Biochem. J. 102, 251-257.
19. Fairbanks, G. Steck, T. L. and Wallach, T., (1971)
20. Alfageme, C. ~, Zweid1er, A., Mahowald, A., and Cohen, L.
H. (1974) J. BioI. Chern. 249, 3729-3733.
21. Panyim, S. and Chalkley, ~(1969) Biochemistry ~, 3972-
3979.
22. Shaw, B. Ramsay, Corden, J. L., Sahasrabuddhe, C. G. and
Van Holde, K. E. (1974) Biochem. Biophys. Res. Commun. 61,
1193-1198.
23. Rill, R. L., Shaw, B. Ramsay and Van Holde, K. E. (1978) in
Methods in Cell Biology, Vol. 17, ed by G. S. Stein, J.
Stein and J. Kleinsmith" Academic Press, N. Y. pp. 69-101.
24. Tatche11, K. and Van Holde, K. E. (1977) Biochemistry 16,
5295-5303.
25. Noll, M. (1974) Nuc1. Acids. Res. 1, 1573-1578.
26. Laemm1i, U. K. (1971) Nature 227, 1-4.
27. Bafus, N. L., Albright, S. C., Todd, R. D. and Garrard, W.
T. (1978) J. BioI. Chern. 253, 2568-2574.
28. Klevin, L. and Crothers, ~M. (1977) Nuc1. Acids Res. i,
4077-4089.
29. Go1dknopf, I. L., French, M. F., Musso, R. and Busch, H.
(1977) Proc. Nat. Acad. Sci. USA 74, 5492-5495.
30. Levy Iv. B. and Dixon, G. H. (1978)Fed. Proceed 37, (6)
Abs. No. 2837 p. 1787. --
31. Richards, R. G. and Shaw, B. Ramsay (1978) Fed. Proceed
1I, (6) Abs. No. 2836 p. 1787.
THE INTERACTION OF HISTONES WITH DNA: EQUILIBRIU~ BINDING STUDIES
D.R. Burton, M.J. Butler, J.E. Hyde, D. Phillips,

C.J. Skidmore and 1.0. Walker
Department of Biochemistry, University of Oxford

South Parks Road, Oxford OX1 3QU
INTRODUCTION
Chromatin, the interphase form of the genetic material, is

made up of a linear array of repeating substructures called nucleo-
somes. Each nucleosome is composed of eight core histones, two each
of the histones H2a, H2b, H3 and H4 associated with about 200 base
pairs of DNA and two molecules of histone H1 [1]. The nucleosomes
interact with each other to form a linear fibre about 100 K in
diameter. Although a wealth of information has accumulated in the
last few years relating to the structure and function of chromatin
[2] there have been few quantitative studies on the binding of
histones to DNA. Such studies are important because they provide a
thermodynamic framework within which the properties of chromatin
can be discussed. Thermodynamic information on the interaction of
histones with DNA can be obtained from equilibrium binding curves
and their variation with temperature, pH, ionic strength and so on.
Histone-DNA interactions in chromatin are, however, highly complex
in the sense that five different proteins interact with each other
and with DNA to form a stoicheiometric complex in the nucleosome.
Thus the construction of binding curves for the binding of indivi-
dual histones or histone pairs to DNA may not be as revealing as
the binding of all five histones. On the other hand all the histones
bind very tightly to DNA at physiological ionic strengths which
makes the task of constructing equilibrium binding curves very dif-
ficult. It is known, however, that the histones may be dissociated
differentially from chromatin by increasing concentrations of
NaCI [3]. In the following the properties of the binding curves ob-
tained in this way are described. In these experiments a sheared
chromatin preparation has been used in which the higher-order
137
138 D. R. BU RTON ET AL.
structure has been destroyed since it does not give the character-
istic pattern of DNA fragments on digestion with micrococcal
nuclease [4,5]. However, these higher-order interactions appear to
involve only HI and not the core histones since the same limit
digest pattern is obtained on extensive digestion with micrococcal
nuclease whether the starting material is native chromatin or
sheared chromatin [6,7]. In sheared chromatin, therefore, the core
histone-DNA interactions are preserved whereas the HI-chromatin
interactions have been modified. In what follows the binding of
core histones to DNA in the presence of dissociated H1 are first
described. Secondly, the binding of HI to nucleosome core particles
is examined with particular reference to the role that HI plays in
the higher-order structure characteristic of native chromatin.
EXPERIMENTAL METHODS
Chromatin was prepared from calf thymus tissue by the method

of Zubay and Doty [8] and characterised as described previously
[3,9,10]. Circular dichroic and viscosity measurements were made as
previously described [9,10].
Digestion of chromatin with micrococcal nuclease (Worthington

Biochem. Corp. Ltd.) was carried out at 37°C in Tris-HCI (1 mM,
pH 7.0), CaCl2 (10-~ M). The reaction was terminated by adding
EDTA to 10 m1~ and cooling the reaction tubes in ice. The chromatin
was then incubated with proteinase K (Boehringer Biochem. Corp.),
100 Ug/ml, for 1 h at 37°C. The DNA was further deproteinised with
chloroform-isoamyl alcohol (24:1) and then dialysed against Tris-
HCI (40 mM) pH 8.0, sodium acetate (5 mM), EDTA (1 mM) for electro-
phoresis on agarose or polyacrylamide gels. After electrophoresis,
the DNA was visualised by staining with ethidium bromide. The size
of the DNA fragments was determined by comparison with fragments of
PM2 or polyoma DNA of known size digested with restriction nuclease
Endo R Hae III (gift from B. Ponder). Polyoma fragments were in
turn calibrated against sequenced ~ x 174 DNA fragments (B. Ponder,
private communication); PM2 fragment sizes were according to Noll
[11] .
The dissociation of core histones from DNA was induced by in-

creasing the concentration of NaCI over the range 0.7 to 2.0 Molar.
In all cases the salt solutions contained sodium dihydrogen phos-
phate, 10 mM and disodium hydrogen phosphate, 10 mM, so that the
final pH of the solutions was 6.2-6.4. The pH meter was calibrated
with sodium hydrogen phthalate, 0.1 M, pH 4.01 and sodium borate
decahydrate, 0.1 M, pH 9.18, all measurements being made at 18-20°C.
The construction of binding curves requires that the concentrations
of the products and reactants in the dissocation reaction be
measured without disturbing the equilibrium. This may be done by
true equilibrium methods which measure the concentrations of the
INTERACTION OF HISTONES WITH DNA 139
various species present in situ or by kinetic methods which separate

the products and reactants and allow their concentrations to be
measured separately. In the case of the system under study here the
application of equilibrium methods is not possible since there is
no easily measurable property of dissociated histone which distin-
guishes it from bound histone. A kinetic method has therefore been
used which separates dissociated histone from depleted chromatin
using gel filtration with Sepharose 4B. The method relies on the
fact that the rate of dissociation and reassociation is slow com-
pared to the time required to separate the products and reactants
during their passage down the column. The equilibrium is thereby
frozen. A typical elution profile is shown in Fig. I which shows
that dissociated histone can be completely separated from chromatin.
In high salt there is a tendency for the histone to dissociate from
the depleted chromatin peak which leads to a variation of histone
to DNA ratio across the depleted chromatin peak. However, the
histone-DNA ratio of the original depleted chromatin may be obtained
by pooling all fractions across the peak and measuring the average
histone-DNA content [12]. Chromatin (0.1-1.0 mg/ml) in 0.7 mM sodium
phosphate, pH 7.0, was dialysed for 3 hours against salt solutions
at either 4°C or 20°C. Aliquots (5 mls) were applied to a column
of Sepharose 4B (50 x 1.0 cm) equilibrated with the appropriate salt
3D
0.50
A260 a",
2D A230
0.25
1.0
o 0
tract ion no.

Fig. 1. Elution profile of chromatin in 1.2 M NaCI from Sepharose
4B column: (a) depleted chromatin; (c) dissociated histone; (b) the
variation of the protein:DNA ratio across the first peak (see inset
ordinate).
solution; fractions (2 ml) were collected and analysed.
H1 binding curves were obtained by dialysing chromatin to

equilibrium against solutions of NaCl, 0.7 mM sodium phosphate,
pH 6.8, at either 4° or 21°C. The solutions were centrifuged at 45 K
for 3 h to 6 h. After this time the depleted chromatin was com-
pletely removed to the bottom of the tube as a pellet. The amount
of H1 present in the upper two-thirds of the supernatant was deter-
mined chemically. The amount in the supernatant did not signifi-
cantly change with time over a 3 to 14 h period, indicating that
the dissociation had reached equilibrium and did not change during
the time-scale of the separation.
RESULTS
The dissociation curve of histones from DNA as a function of

NaCl concentration is shown in Fig. 2. Dissociation of histone takes
place in three distinct stages. Between 0.7 mM sodium phosphate and
0.7 M NaCl H1 is selectively dissociated as shown previously [13].
Between 0.7 and 1.2 M NaCl H2a and H2b are selectively removed from
the chromatin and between 1.2 and 2.0 M NaCl H3 and H4 are dissocia-
ted (Fig. 3 and refs. [3,12]). Dissociated H1 is always present in
-
"0
0
ell
·U
0
100
.18
"0
-
ell
C
0
en
:.c H2oH2b
ell
8'
C 25
tI
~ H1
0 0·5 1·0 1·5 2·0

[Noel] Molesllitre
Fig. 2. The dissociation of histone types from chromatin as a func-
tion of NaCl concentration.
a b c d e
Fig. 3. Polyacrylamide gel electrophoresis patterns of histone types

dissociated as a function of NaCI concentration: SDS-polyacrylamide
gels of (a) histones in chromatin; (b) histones remaining bound to
chromatin in 0.7 M NaCI; polyacrylamide-urea gels of (c) histones in
chromatin; (d) histones dissociated in 1 . 2 M NaCI ; (e) histones re-
maining bound to chromatin in 1.2 M NaCI.
the system during the dissociation and re-association of the other

four core histones.
(a) The Dissociation of H2a and H2b
The dissociation curve of H2a and H2b from chromatin is shown

in detail in Fig . 4 together with the re-association curve obtained
by dialysing samples of chromatin into 1 . 2 M NaCI, to dissociate
completely H2a and H2b, followed by further dialysis into salt solu-
tions of lower concentrations in the range 1 . 2 to 0.7 M. The two
sigmoid curves are equivalent showing that the dissociating species
are in reversible thermodynamic equilibrium .
The effect of chromatin concentration on the degree of disso-

ciation was investigated at a constant NaC1 concentration of 0 . 85 M
by decreasing the initial chromatin concentration . As would be ex-
pected from considerations of mass action effects on a dispropor-
tionation reaction , the degree of dissociation increases with
.£
g 100
..c
e
u
-e
E
.0
N 75
~
-
N
I
0
.§ 50
0
'g
.!a
"0
Q.I 25
~
C
Q.I
~
rf a
0.7 0·8 o·g lO II 1-2
[Noel] moles/litre
Fig. 4. The dissociation of H2a-H2b as a function of NaC1 concen-

tration: - . - forward dissociation curve at 4°C; - 0 - reverse
dissociation curve at 4°C; - 0 - forward dissociation curve at 20°C.
increasing dilution as shown in Table 1. The right hand column shows

the dissociation constant, Kapp, predicted by applying the simple
TABLE 1
Co a. Kapp
1010 0.26 9.2

725 0.28 8.2
520 0.29 6.1
250 0.38 5.9
Apparent dissociation constant,

Kapp = a. 2c o 2/(1-a.)co, as a func-
tion of degree of dilution of
chromatin at NaC1 = 0.85 M. Co =
initial concentration; a. = degree
of dissociation of H2a-H2b. Kapp
in arbitrary units; temp. = 4°C.
TABLE 2
[NaCl] ,M Ct. [H2a]/[H2b]
0.8 .25 1.04

0.9 .70 1.02
The ratio of H2a to H2b remaining

bound to chromatin at two different
degrees of dissociation, Ct.. Temp. =
4°C.
mass action law. Ct., the degree of dissociation, is defined in (d).

The relatively constant value of Kapp over the four-fold change in
initial concentration leads to the conclusion that H2a and H2b
interact with chromatin non-cooperatively, that is, they bind to
equivalent, independent, non-interacting sites. The effect of tem-
perature on the dissociation is shown in Fig. 4. Increasing the
temperature from 4°C to 20°C decreases the degree of dissociation
over the whole dissociation range although the two dissociation
curves a~e not parallel. The dissociation reaction is therefore
exothermic but the enthalpy change may vary with the salt concen-
tration possibly due to a contribution to the enthalpy change from
salt-induced conformational changes in the components of the system.
The amounts of H2a and H2b which remained bound to the chroma-
tin at various degrees of dissociation were monitored using poly-
acrylamide urea gel electrophoresis. The results are shown in Table
2 and suggest that these two histones appear to dissociate conco-
mitantly possibly as an equimolecular complex. This suggestion is
supported by crosslinking studies which show that between 0.7 and
1.2 M NaCl dissociated H2a and H2b may be covalently crosslinked by
formaldehyde to form dimers and higher polymers. No monomers were
detected [3].
The dissociation reaction is also reversible at the level of

the secondary structure of the histones and DNA in the chromatin
complex as judged by the CD spectra of partially dissociated and
reconstructed chromatin. Data were obtained for four different salt
concentrations between 0.7 and 1.2 M NaCl. The spectra of the disso-
ciated and reassociated chromatins were identical in all cases;
that for 0.85 M NaCl is shown in Fig. 5.
(b) The Dissociation of H3 and H4
The dissociation and reassociation curves of H3 and H4 from

DNA is shown in Fig. 6. As in the case of H2a and H2b the curve is
144 D. R. BURTON ET AL.
~--------------~----~~~------~-2
AE
-1
240 250
-I.
-6
-8
o~--------------~--------------~
Fig. 5. Circular dichroic spectra of partially dissociated and re-
associated chromatin in (a) 0.85 M NaCI and (b) 1.35 M NaCI:
---- dissociated species, ---- reassociated chromatin; ~E is the
difference in extinction per mole of nucleotide.
sigmoidal and thermodynamically reversible. It is important to

realise t~at the dissociation of H3 and H4 from DNA has here been
studied in the presence of unbound HI, H2a and H2b, which have com-
pletely dissociated before H3 and H4 begin to dissociate above
1.2 M NaCI (Fig. 2).
The effect of protein concentration on the degree of dissocia-

tion was investigated at 1.35 M NaCI, as before, by decreasing the
initial concentration. The equilibrium constants, calculated by
assuming a simple mass action effect on the reaction, are shown in
Table 3. a, the degree of dissociation is defined in (d). Over a
five-fold range of initial chromatin concentration, the value of
Kapp changes by two orders of magnitude, showing that Kapp is itself
a ~unction of concentration. This indicates that the binoIng of H3
and H4 to DNA is accompanied by cooperative interactions between
the reacting species.
Increasing the temperature from 4°C to 20°C provokes a decrease

in the degree of dissociation showing, as with H2a and H2b, that the
-
1J
II
0
'8III
CIt
=s
"::J:
M
I
::J: 40
II
8'
cII 20
~
II
Q.
1·3 1·4 1·5 2·0

[NoCij Molesllitre
Fig. 6. The dissociation of 83-84 as a function of NaCI concentra-
tion: - . - forward dissociation curve at 4°C; - 0 - reverse dis-
sociation curve at 4°C; -~ - forward dissociation curve at 20°C.
TABLE 3
Co a Kapp
106 0.10 l.17

97.5 0.16 3.02
76.1 0.42 23.0
66.7 0.52 37.5
46.9 0.70 76.4
Apparent dissociation constant,

Kapp = a2co2/(1-a)co, as a function
of degree of dilution of chromatin at
NaCI = 1.35 M. Co = initial concentra-
tion; a = degree of dissociation of
83-84. K in arbitrary units; temp.
= 40C. app
reaction is exothermic. The two dissociation curves are not paral-

lel which suggests that there may be a contribution to the overall
enthalpy change from salt-induced conformational changes.
An analysis of the histone types which are dissociated over

the range 1.2 to 2.0 M NaCl by gel electrophoresis has shown that
H3 and H4 are dissociated and reassociated in equimolar amounts,
probably as an equimolar complex, which must therefore be at least
a dimer [12]. This result is supported by crosslinking studies of
H3-H4 DNA in 2 M NaCl which show that the histones exist as hetero-
typic dimers, tetramers and higher polymers [3]. Circular dichroic
spectra of partially dissociated complexes and re-associated com-
plexes at the same NaCl concentration show that as regards the
secondary structure of the nucleoprotein the interaction is com-
pletely reversible (Fig. 5).
The intrinsic viscosity of the native, sheared chromatin pre-

paration was 20 dl/gm, in 0.7 mM sodium phosphate, pH 7.0. The
intrinsic viscosity of the DNA in the same solvent was 105 dl/gm.
A sample of chromatin was made 2 M in NaCl by the addition of solid
and the solution was then dialysed into 0.7 mM sodium phosphate. The
intrinsic viscosity of the reconstructed sample was 20 dl/gm. This
shows that dissociation and re-association of all five histones
takes place reversibly with respect to viscosity which is a measure
of the supercoiling and compaction of the molecule [14].
The intrinsic viscosity of H3-H4-DNA in sodium phosphate (0.7

mM, pH 7.0) was 42 dl/gm, indicating that the arginine-rich histone
DNA complex has a conformation intermediate between the fully super-
coiled native structure and free DNA. H3-H4-DNA which had been dis-
sociated in 2M NaCl and then re-associated by dialysis into 1.1 M
NaCl had an intrinsic viscosity of 51 dl/gm (measured in 0.7 mM
sodium phosphate). Thus the intermediate, partly supercoiled con-
formation is recovered after dissociation and re-association of the
arginine-rich histone complex.
(c) Interaction of HI with Chromatin Core Particles
The forward dissociation curve of HI as a function of NaCl is

shown in Fig. 7 together with the re-association curve. This latter
curve was obtained by completely dissociating HI by dialysis into
0.7 M NaCl. The NaCl concentration was then reduced to the required
value by further dialysis and the amount of dissociated HI estima-
ted as before. The coincidence of the two curves shows that the
binding process is reversible. The amount of HI dissociated in
0.7 M NaCl was 21% of the total histone present in the chromatin.
Since HI is completely and discretely dissociated under these con-
ditions, this represents the total amount present. It corresponds
to 1.5 moles of HI bound per mole of nucleosome core particle. The
binding curve is bi-phasic, indicating that two distinct modes of
binding are involved in the equilibrium process. Furthermore, each
phase of the binding curve is sigmoidal which suggests that the
binding processes are cooperative with respect to NaCl.
100
II
5
iii 75
1:
2
v:::>
c
E
£ 50
:I:
'0
c
.9
0
'v0
"'
0"'
~
0·2 (}3 0·4 0·5 0·6 (}7

M. Nael
Fig. 7. The dissociation of HI as a function of NaCl concentration:

- 0 - forward dissociation curve at 4°C; - . - reverse dissociation
curve at 4°C; - • - forward dissociation curve at 21°C .
The CD spectrum of chromatin, after HI had been completely dis-

sociated and re-associated, was identical to that of native chroma-
tin. Thus the binding of HI is reversible with respect to secondary
structure.
The effect of increasing the temperature on the binding curve

is shown in Fig. 7. At 21°C the relative proportion of HI dissociat-
ing in the first phase at lower [NaCl] increased from about 25% to
50%, whereas the proportion in the second phase was correspondingly
reduced .
The effect of decreasing the chromatin concentration on the

degree of dissociation of HI at constant [NaCI] was investigated by
diluting the chromatin with 0.7 mM sodium phosphate, pH 7 . 0, before
dialysis into the required NaCI solution. At each dilution the
amount of dissociated HI was estimated as before. By assuming that
the total number of binding sites on the chromatin for HI is repre-
sented by the number of molecules of HI bound to the native chroma-
tin preparation, i.e. all the sites are occupied in the native state,
the fraction of free and bound sites may be estimated at each
TABLE 4
0.2 M NaCl 0.3 M NaCl
Co (l Kapp Co (l
Kapp
870 .19 4.1 630 .40 1.7

500 .29 5.9 460 .49 2.2
250 .32 3.8 310 .50 1.6
160 .60 1.5
120 .64 1.3
0.4 M NaCl
Co (l
Kapp
986 .45 3.6
450 .68 6.5
230 .70 3.7
dilution as described in (d). The degree of dissociation increases

with increasing dilution as shown in Table 4. The apparent dissocia-
tion constant, Kapp' remains constant over a 4-5 fold change in
concentration at 3 different salt concentrations. This shows that
H1 binds non-cooperatively to equivalent, independent binding sites
on the core particles. The stoicheiometry of the binding suggests
that there are 1.5 H1 binding sites per nucleosome core particle.
Symmetry considerations imply that the number should be two [15]
but that not all these sites are filled in sheared chromatin
probably due to loss of H1 during preparation.
(d) Analysis of Binding Curves
The equation describing the equilibrium between dissociated

histone and chromatin as a function of NaCl may be written as:
DNH + nNaCl = DNH' + mH
where DNH is the molar concentration of undissociated chromatin,

H is the molar concentration of dissociated histone, DNH' is the
concentration of depleted chromatin, n is the number of moles of
NaCl bound to depleted chromatin when m moles of histone are dis-
sociated. The equilibrium constant, Keq, may then be written:
(DNH') (H)m
K (1)
eq (DNH) (NaCl)n
It follows that
log Keq = log [ (DNH') (H)m]

(DNH) - n log (NaCI).
For equivalent, independent, non-interacting sites for histone bind-

ing, m = 1; cooperativity is indicated if m > 1. It is assumed that
the number of binding sites for histone molecules in native chroma-
tin is equal, on a molar basis, to the number of moles of histone
bound. That is, in the native state, for core histones, the binding
sites are fully saturated and binding is stoicheiometric. If Co is
the total concentration of a given histone type bound in native
chromatin the concentration of dissociated histone is aco where a
is the degree of dissociation determined from the binding curves;
aco is the concentration of free binding sites on the chromatin and
(l-a)co is the concentration of bound sites on the crhomatin. It
follows from (1) that
( aco ) m+1
K =---- (2)
eq (l-a)co (NaCI)n .
Since it has already been shown that the binding of H2a-H2b to

DNA is non-cooperative (Table 1), m = 1 and a graph of log
(DNH')(H)/(DNH) against log (NaCI) will have a slope of n. Such a
graph is shown in Fig. 8, for the data at 4° and 20°C. The relation-
ship is linear in both cases with n = 30 ± 1 per dimer of H2a-H2b.
The intercept at 4°C gives Keq = .072 M and 6Go = 1.6 Kcals/mole at
1.0 M NaCI. The temperature-dependence of Keq yields values of 6Ho
and 6So equal to -10 Kcals/mole and -31 cals/mole/o respectively at
4°C and 1.0 M NaCI.
Thus the dissociation of one mole of H2a-H2b dimer results in

the net binding of 30 moles of NaCI. The standard free energy of
dissociation is positive which is to be expected since molar con-
centrations of NaCI are required to dissociate histones from chro-
matin in micromolar concentrations. Since the enthalpy change is
negative, the positive free energy term is dominated by the large
negative entropy change on dissociation. This presumably arises
primarily as a result of the ordering of solvent water molecules on
breaking electrostatic bonds and hydrophobic interactions in the
dissociation reaction. The number of sodium chloride molecules bound
as one molecule of histone dimer dissociates is very close to the
number expected (27) if all the lysine and arginine side chains in
the amino terminal domains of the histones defined as in ref. [15]
form salt links with phosphate groups on the DNA.
Since the dissociation of H2a-H2b is non-cooperative with

respect to histone the sigmoidal nature of the dissociation curve
(Fig. 4) shows that the dissociation with respect to NaCI is
o~---------------------------------------------,
-1
r
~ -3
~z
zc -4
c-
L-......I -5
....
o
CJI
..2 -6
-7
o -0·' -0·2 -0.3
10910 [NoCij
Fig. 8. The variation of loglO[(DNH')(H)/(DNH)] with lOglO (NaCl):

- . -,4°C; - 0 - 20°C.
cooperative. It is important to realise that n in Equation (1) is

not a Hill coefficient and is therefore not a measure of the degree
of cooperativity. It is not possible to construct Hill plots directly
from the data presented here because there is no direct measure of
the amount of sodium chloride bound as a function of the sodium
chloride concentration. However, by making the assumption that the
amount of sodium chloride bound (to either DNA or histone or both)
is directly measured by the degree of dissociation of the histone
it is possible to oonstruct a Hill plot with a slope of 11.5, which
indicates a high degree of cooperativity for sodium chloride binding.
The data of Table 3 show that the binding of H3-H4 to DNA is

cooperative and that m> 1 in Equation (1). It follows from (2)
that
a m 1
K
eq = i-a . (aco) .
(NaCl)n
and log Keq = log a/1-a + m log (aco) - n log (NaCI). At constant
(NaCI)
a
log K
eq
= log --- + m log (aco) + Constant.
1-a
Thus m may be obtained from the slope of a graph of log a/1-a ver-
sus log aco. This is shown on Fig. 9 and gives slope of m = 2.
2·0~r--------r---r----_
log r(a 3 ]
CO}
109(, ~a )
• l('- a }c o
'·0
'·0 '·5 0 0.5

log a Co log [NaCO
Fig. 9. Left: the variation of log (a/1-a) with log aco; right: the
variation of log. [(aco)3/(1-a)co] with log (NaCI).
From equation (2),
)m+j
log K
eq log ~ aco
(l-a)co - n log (NaCI) .
Thus a graph of log [(aco) m+1 .

/(l-a)co] w1th m = 2, versus log (NaCI)
gives a slope of n. This is shown in Fig. 8 and gives n = 14.0 per
mole of H3-H4 dimer. As before this nrmber represents the net number
of sodium chloride molecules bound when one mole of H3-H4 disso-
ciates. It may be compared to the number of basic amino acid side
chains (29) in the N-terminal domains of H3-H4 (see ref. [15]) and
indicates that, in contrast to H2a-H2b, displacement of the histone-
DNA salt linkages by bound counter ion is not stoicheiometric. The
apparent equilibrium constant at 1 M NaCI is 4.3 x 10- 6 M, corres-
ponding to ~Go = 7.4 Kcals/mole; ~Ho = -15 Kcals/mole at 1.4 M NaCI.
The data of Table 4 show that the binding of H1 is non-

cooperative and that m = 1 in equation (1). The variation of log

(DNH')(H)/(DNH) with log (NaC1) is shown in Fig. 10 at two dif-
ferent temperatures; the data refer to that shown in Fig. 7. It is
apparent that nand K are not constant across the dissociation
range. At 4°C n = 4 at low salt concentrations but from a point cor-
responding to a concentration of about 0.4 Molar the value of n
increases abruptly to 10. At 21°C, n = 1 at low salt and again in-
creases sharply to 10 at high salt with a transition point at about
0.4 Molar NaCI. Above this point the two curves are identical. This
indicates that the enthalpy of dissociation in the high salt range
(> 0.4 M) is zero. The value of Keq obtained by extrapolating this
part of the curve to 1 M NaCI is 1.0, giving 6Go = O.
In deriving the above equations the assumption has been made

that the activity coefficients of the macromolecular components
are unity. However, the ionic environment of these components and
in particular the dissociated HI, is varying across the dissocia-
tion range from one of very low ionic strength to one of relatively
very high ionic strength. If this change of environment induces con-
formational changes in HI or otherwise alters the activity coeffi-
cient, there will be a marked change to the activity of the dis-
sociated histone.
At low ionic strength the conformation of free histones is
-s " ' .................... ,

......~--.a..........._______
~~------------~----------~--------~~~~
o -().s -1·0 -1·5
Fig. 10. The variation of logIO[(DNH )(H1)/(DNH)] with loglo (NaCI

activity) .
largely random coil. Secondary structure, especially a-helical

structure, similar to that found in histone bound in the nucleo-
histone complex, is only apparent at high ionic strength. In the
case of HI there is a large decrease in ellipticity at 222 nm as
the ionic strength is reduced (data not shown), which indicates
that the molecule undergoes a conformational change leading to loss
of a-helical secondary structure. The mid-point of the change is
centred on 0.3 M NaCl. This evidence suggests that histone HI, dis-
sociated from chromatin below 0.4 NaCl, is in an unfolded, largely
unstructured conformation, whereas HI dissociated above 0.4 M NaCl
is in a more structured conformation similar to the conformation
of bound HI. If the structure of bound HI is considered the native
structure and the low-ionic-strength form a denatured structure,
then the results shown in Fig. 7 allow the equilibrium below 0.4 M
NaCl at 4°C to be represented by:
(Native Hl)b
oun
d + 4 NaCl * (Denatured H1)f
ree
'
and above 0.4 M NaCl by:
(Native H1)
bound
+ 10 NaCl * (Native H1)f
ree
.
From this it follows that
(Denatured H1)
free
+ 6 ~aCl * (Native H1)
free
.
Thus the refolding of the denatured to the native form is accom-

panied by the net binding of six sodium chloride molecules. These
changes are shown schematically in Fig. 11.
In order to investigate the binding of H1 to DNA H1 was
~__~~--~-COOH
106
Free 'denatured' Hl
~-4NaCI
+4NaCI~
.
DNA binds to +ve
-6Nac'Ji +6NaCI
..
regions of Hl
,I ,
,
NH2~COOH
-10Na~ONaCI nativeHl bound to DNA
NH2~OOH
Free 'native Hl
Fig. 11. Salt-dependent equilibria between native and denatured,
free and bound H1.
154 O. R. BURTON ET AL.
dissociated and separated from the chromatin by dialysis into 0.7 M

NaCI followed by gel filtration chromatography [13]. An amount of
DNA was then added to the solution of HI equivalent to the amount
of DNA present in the original chromatin. The mixture was then dia-
lysed into solutions of lower NaCI concentration in the range 10- 3
to 0.6 Molar. Samples were removed and applied to a Sepharose 4B
column. The DNA-HI complex eluted at the exclusion volume of the
column whereas free HI eluted as a separate, later fraction. The
amount of HI bound to DNA was determined, the amount of HI free
estimated by difference, and the association curve obtained. The
binding curves of HI to DNA and chromatin as a function of NaCI
were very similar. This leads to the conclusion that the binding
sites for HI on chromatin are primarily formed by the pattern of
free phosphate groups on the DNA created by the binding of the other
four histones [15] and that there is little contribution to the
binding free energy from interactions of HI with the other four
histones.
(e) Circular Dichroism Studies
It is well known that the circular dichroic spectrum of nucleo-

histone in the region 300-260 nm is very much reduced in intensity
compared to that of DNA [10,16]. This difference has been attribu-
ted to short range effects of the bound histones on the DNA con-
formation [10]. The separate contributions of the lysine rich and
the arginine rich pair to the decrease in dichroism at 280 nm is
shown as a function of the amount of each histone type bound in
Fig. 12. The removal of H2a-H2b caused a proportional increase in
the dichroism which is consistent with the non-cooperative mode of
binding revealed by the binding curves. By contrast the dissocia-
tion of H3-H4 caused a disproportionate change in the dichroism of
DNA showing that the interaction of H3 and H4 with DNA produces long
range cooperative changes in the conformation of the DNA. These con-
formational changes may be related to the cooperative binding curves
described earlier.
(f) Nuclease Digestion Studies
Nuclei, sheared chromatin, HI-depleted and H1-H2a- and H2b-

depleted chromatin were digested with micrococcal nuclease for
various times, the DNA extracted and the fragments separated and
sized on acrylamide gels (Fig. 13). The nucleosome repeat length for
native chromatin in nuclei was 200 base pairs, in good agreement
with previous estimates [1]. As others have found, sheared chromatin
and depleted chromatin did not give the regular ladder of base pair
repeats observed for the native material [5,6]. HI-depleted chroma-
tin showed discrete bands at 570, 420, 290 and 160 base pairs. The
larger fragments decreased in amount with time to produce a limit
0
CO
N
W
<l
OJ
0)
c 40
0
£.
U
0~
20
% Histone bound
Fig. 12. The percentage change in ~E at 280 nm for the dissociation
of H2a-H2b over the range 0.7 to 1.2 M NaCI and for the dissociation
of H3-H4 over the range 1.2 to 2.0 M NaCI. The latter spectra were
all measured at 1.30 M NaCI. ~E is the difference in extinction for
left and right circularly polarised light per mole nucleotide.
digest with bands at 145, 130, 108, 99, 77, 65, 55 and 45 base pairs.
At early stages in the digestion the oligonucleosome patterns indi-
cate an internucleosome spacing of 140-150 base pairs which is con-
siderably less than the repeat length in the native chromatin (200
base pairs). Thus, removing HI appears to have caused the nucleo-
some core particles to slide closer together. An alternative ex-
planation might be that the process of digestion and degradation of
the DNA itself promotes a rearrangement of the core particles so
that they move closer together. The oligomer bands are very broad
with a sharper leading edge than trailing edge. For example the
width of the trimer band corresponds to 50 base pairs and the mater-
ial in the band is clearly very heterogeneous arising from fragments
as large as 450 base pairs and as small as 390 base pairs. It is
possible that at very early times of digestion the repeat length
could approach that of the native chromatin. After longer digestion
times core particles are produced with maximum DNA lengths of 145
base pairs, as observed by others [6,7]. Extensive cutting of intra-
nucleosomal DNA has occurred to give the well-characterised sub-
nucleosomal lengths.
The gels of H3-H4 DNA showed a smear at early times of diges-

tion but later on discrete bands were observed at 260, 194, 145,
135, 129, 105, 73 and 50 base pairs. A metastable limit digest was
a b c d e f 9 h
Fig. 13. Digestion of H1-dep1eted chromatin and H3-H4 DNA with

micrococcal nuclease followed by electrophoresis on 5% polyacryla-
mide gels. a-c, time course of digestion of H1-depleted chromatin:
a, 60 mins (36% digestion); b, 15 mins (24% digestion); c, 10 mins
(14% digestion). d, partial digest on native chromatin; e,Polyoma
DNA digested with Hae III; f-g, time course of digestion of H3-H4
DNA: f, 7 mins (36% digestion); g, 10 mins (38% digestion); h,
40 mins (56% digestion). The extents of digestion were determined
in all cases from the absorbance of the sample at 260 nm, after
digestion and dialysis, relative to a zero-time control.
reached at about 60% digestion with major bands at 145, 135, 73 and
50 base pairs. It was not possible to estimate a repeat length for
the H3-H4 core particle from this data but it is clear that the
arginine-rich complex is capable of transiently protecting DNA
fragments which are longer than the DNA lengths associated with
complete nucleosome core particles (140 base pairs) and longer even
than the DNA repeat in native chromatin. From the size of the limit-
digest bands it is clear that H3-H4 alone is capable of protecting
DNA lengths the size of the complete nucleosome core particle.
DNA, reconstituted by salt dialysis with H2a-H2b or HI, H2a-

H2b, and then digested with micrococcal nuclease gave rise to smears
on gels; no discrete bands were seen in keeping with other reports
[17] .
(g) Separation of Limit-digest Products
H3-H4 DNA in 0.7 mM sodium phosphate, pH 7.0, was digested for

30 mins at 37°C with micrococcal nuclease, the reaction terminated
by the addition of 10 mM EDTA, pH 7.0, and the products applied to
a column of Sephadex G-200 equilibrated with 10 mM EDTA, pH 7.0 at
4°C. The elution profile showed that the column had resolved three
fractions of material absorbing at 260 nm. The protein and DNA
content of the three fractions were determined and the size of the
DNA fragments in each fraction were measured by electrophoresis on
polyacrylamide gels (Fig. 14). These experiments showed that the
first fraction which eluted from the column had a DNA protein ratio
of 1.9. The DNA fragments in this fraction were resolved into three
components of length 145, 125 and 104 base pairs present in approxi·
mately equal amounts. The amount of protein associated with these
fragments suggests that a tetramer of H3-H4 is associated with
104-145 base pairs of DNA. The sedimentation coefficient of frac-
tion I was 7.18 in 10 mM Tris, pH 7.0.
DISCUSSION
The results presented here show that the interaction of core

histones with DNA is a thermodynamically reversible process with
the respect to the amount and type of histone bound and with res-
pect to the secondary and tertiary structure of the nucleo-protein
complex. Other studies have shown that it is also reversible with
respect to the intimate association of the histones with DNA as
revealed by the protection to digestion of the DNA with micrococcal
nuclease [17]. Dissociation and re-association of core histones
takes place in two discrete stages. When the NaCI concentration is
increased above 0.7 M, the histones H2a and H2b dissociate as an
equimolar complex, probably a dimer, in a non-cooperative manner.
When the dissociation of this histone pair is complete, H3 and H4
a b c d e f 9 h
Fig. 14. Gel electrophoresis patterns of H3-H4 DNA digested with

micrococcal nuclease and then fractionated by gel filtration with
Sephadex G-200: a, unfractionated digestion mixture; b, c, d, frac-
tions across Fraction I (1st peak from column); e, f, fractions
across Fraction II (2nd peak from column) ; g, PM2 DNA digested with
Hae III; h, same as b.
dissociate cooperatively at higher NaCl concentrations as an equi-

molar complex. The cooperativity is manifested not only in the free
energy of interaction between the histones and DNA but also in the
conformational change produced in the DNA of the nucleoprotein com-
plex, which shows that the H3-H4 complex is capable of inducing
long range changes in the DNA to produce a conformation which is

intermediate between free DNA and the conformation it adopts in
tightly supercoiled chromatin. It is possible that DNA in 2 M NaCI
exists in a variety of conformational states some of which may
approximate to the supercoiled conformation found in chromatin. If
H3-H4 binds preferentially to supercoiled DNA, the arginine-rich
complex could stabilise such a conformation thus accounting for the
changes in circular dichroism. The binding would be further stabi-
lised by histone-histone interactions between bound tetramers. Such
a mechanism would also explain the decreases in viscosity of the
H3-H4 DNA which, on this view, result from the folding of the DNA
into a supercoiled conformation. Evidence in support of direct
histone-histone interactions has been obtained from crosslinking
studies on H3-H4 DNA at various ionic strengths [3]. It is inter-
esting to note in this context that Bina-Stein and Simpson [26]
have recently reported that H3-H4 histones can compact SV40 DNA by
2.6 fold into nucleosome-like particles. On extensive digestion with
micrococcal nuclease a tetramer of H3-H4 complexed to 104, 125 and
145 base pair lengths of DNA may be isolated. The larger of these
protected fragments is identical in size to the DNA protected in
nucleosome core particles (140 base pairs). This suggests that the
H3-H4 tetramer spans the nucleosome core particles, thus anchoring
the ends of the DNA. The time course of digestion with nuclease
suggests that lengths of DNA greater than the H3-H4 core particle,
140 base pairs, or even the complete nucleosome, 200 base pairs,
can be transiently protected from digestion. This implies that there
are long range interactions between H3-H4 tetramers and higher
polymers which may generate ordering or phasing of the H3-H4 core
along the DNA. The phasing interval cannot be less than 140 base
pairs and, on average, cannot be greater than 200 base pairs, the
repeat length in native calf thymus chromatin. It is tempting to
suggest that the cooperative binding of H3-H4 provides a mechanism
for phasing the completed nucleosomes at 200 base pair intervals
[12]. This argument is, however, difficult to sustain in the light
of the highly conserved amino acid sequences of H3 and H4, presum-
ably reflecting a highly conserved structure and function, and the
variability in the repeat length of nucleosome spacings in chroma-
tin from different sources [1]. Alternatively, it is possible that
H2a and H2b space out the final nucleosome repeat lengths when they
bind to the H3-H4-DNA complex. They probably have sufficient varia-
bility in structure to account for the variation in spacer lengths.
In correctly spacing the nucleosomes they would have to interact
with regularly spaced H3-H4-DNA tetramers, already bound to DNA,
and slide them along the DNA. In principle the change in free
energy for the sliding reaction is zero so that the binding of H2a-
H2b dimers would in itself provide sufficient free energy to drive
the reaction until H2a-H2b dimers made contact along the length of
the fibre to produce the correct spacing. Thus each nucleosome core
particle would be associated with 200 base pairs of DNA even though
the core histones in the limit only protected 140 base pairs from
nuclease digestion. However, to achieve this regular spacing re-

quires the H2a-H2b molecules to interact between nuc1eosomes and
the binding studies reveal no evidence for such interactions either
at the histone-histone level or between histone and DNA.
The proper binding of the H2a-H2b complex to DNA appears to

require the presence of H3-H4. The binding of H2a-H2b and/or H1 to
DNA by itself does not lead to the discrete protection of DNA
fragments from nuclease digestion, unlike H3-H4, as observed by
others [17]. This difference appears to be due to the differential
stability of the secondary structure in the two histone pairs.
Whereas, at low ionic strength, the secondary structure of the his-
tones in the H3-H4-DNA complex is retained, the H2a-H2b-DNA complex
loses structure at low ionic strength. Thus in the former case DNA
maintains the stability of the bound histones, whereas in the latter
it does not. On the other hand, when H2a-H2b binds to the H3-H4-DNA,
the secondary structure is stabilised at low ionic strength, presum-
ably due to histone-histone interactions between the arginine-rich
and the lysine-rich pair. This suggests that the binding site for
the H2a-H2b dimer is created by and formed from the H3-H4-DNA com-
plex, such that binding occurs not only to the histone but also to
the DNA. These conclusions confirm and support the observations of
Sollner-Webb et al. [17], Camerini-Otero et al. [18] and Oudet et
al. [19]on the central importance of the H3-H4 arginine-rich com-
plex in the structure of the nuc1eosome core particle. The thermo-
dynamic analysiS shows that the binding of H2a-H2b is non-
cooperative with respect to histone-histone interactions but highly
cooperative with respect to NaC1. A further indication of coopera-
tivity comes from considerations of the rates of dissociation and
association. The rate of dissociation of core histones from DNA is
a relatively slow process with a half time of at least one hour.
This property has been utilised in separating the products of the
reaction by gel exclusion chromatography in order to construct the
binding curves. Thus the dissociation rate constant is of the order
of 10 4 sec-I. From this and the equilibrium constant (in 1.0 M NaC1
equal to 0.072 M), we estimate that the association rate constant
must be in the range 10 5 -10 6 M sec-I. Both rate constants are small
compared to the values expected for simple ionic reactions involv-
ing unlike charges. These have very high frequency factors, low
activation energies and large negative activation free energies. By
contrast, the histone-DNA binding, which is dominated by ionic
interactions, appears to involve large free energies of activation.
This may be interpreted to mean that a large number of weak bonds
have to be made or broken simultaneously in the activated complex
for interaction to occur. In other words, the reaction is highly
cooperative.
It is generally assumed that the interaction between histone

and DNA involves electrostatic bonds between the positively charged
lysine and arginine side chains and the phosphate groups in the DNA.
The positive charges on the core histones are clustered mainly in

the N-terminal and C-terminal parts of the molecules [20]. It has
been suggested on the basis of nmr studies that these regions are
structureless and exist, in the uncombined histones, as random
coils which are thought to wrap around the grooves of the DNA when
the histones interact [21,22]. If the extent of the tails is as
defined by Hyde and Walker [18] then the number of basic side chains
varies between 12 in H2a and 15 in H2b and H4. If these basic side
chains neutralise adjacent phosphates on both chains of DNA, they
will interact with at least one half to three quarters of a turn of
DNA helix. Thus, on the assumption that the histone tails wrap into
the DNA grooves, the tails must be flexible enough to wrap around
at least half a turn of DNA and the binding mechanism must involve
a 'zipper' interaction in which successive segments of the molecule
bind consecutively. Such binding would involve fast 'on' and 'off'
rate constants and show a low degree of cooperativity [23]. Alter-
natively, binding may occur in a single step with the ligand (the
histone basic tale) rigidly and correctly positioned with respect
to the phosphate binding site. This mode of binding could occur if
binding takes place to one face of the DNA helix only so that the
tails do not wrap into and round the DNA grooves. Thus, interaction
takes place on that surface of the DNA which faces into the histone
core leaving the other, outer-facing half freely accessible to
enzymes and other small molecules. Such a binding process would be
slower than the 'zipper' mechanism and highly cooperative since all
the corresponding subsites on the macromolecules interact simul-
taneously. Both 'on' and 'off' rate constants would in this case be
small. The experimental data, such as it is, is more consistent
with the latter kinetic model.
THE MECHANISM OF CORE PARTICLE ASSEMBLY
It is evident from the data presented here that the formation

of core particles from free histone and DNA by dialysis from high
salt is an example of a self-assembling process. The key step in
this assembly appears to be the association of an equimolar H3-H4
complex with DNA; this is followed by the binding of H2a-H2b. Evi-
dence has been presented to show that the core histones in 2 M NaCl
exist in the form of a heterotypic tetramer or octamer [24,25]. If
such species do exist in high salt, it is clear that they are not
on the main thermodynamic pathway of assembly of core particles and
are not a necessary requirement for the in vitro assembly. The
situation in vivo will presumably depend very strongly on local con-
centrations of histones, DNA and counter ion at or close to the DNA
replication points.
Several models for chromatin structure have been described

which have as their basis the symmetry elements which arise as a
result of specific interactions between the eight histones in the
nucleosome core particle [15,27,28]. The model described by Hyde

and Walker [15] is based on a central core of histones composed of
two helical polymers of H3-H4 and H2a-H2b. The DNA is supercoiled
around the central core with a pitch determined by the pitch of
the helical polymers. The models described by Weintraub et al. [27]
and Worcel et al. [28] for the 100 A fibre are generally rather
similar to that described in [15] and will not be discussed further.
It is, however, instructive to see how the assembly of core particles
can be explained in terms of the structure described in [15]. In
high ionic strength solutions DNA may be visualised as an equili-
brium mixture of different conformations one of which may approxi-
mate to the supercoiled conformation found in chromatin. The core
histones also exist as an equilibrium mixture of heterotypic dimers,
tetramers and higher polymers in equilibrium with homotypic species.
As the salt concentration is decreased these equilibria are dis-
turbed by the cooperative formation of the H3-H4-DNA complex. In
1.2 M NaCl the H3-H4 dimers bound to DNA are in contact and may be
crosslinked with HCHO to form high polymers [3]. The core particles
are completed, on further lowering the salt concentration, by the
association of H2a-H2b oligomers (probably dimers) to specific non-
interacting sites formed by the H3-H4 DNA. This results in a tight-
ening of the structure, shown in a decrease in viscosity, a further
tightening of the supercoil which leads to changes in the circular
dichroic spectrum and the formation of an interacting, linear array
of nucleosome cores to form the 100 A fibre (Fig. 15).
Native H1 binds to nucleosome core particles reversibly and

non-cooperatively. The binding sites (probably two per core par-
ticle) are identical and are formed primarily by a specific pattern
of free DNA phosphate groups. Since the lengths of DNA protected
from micrococcal nuclease digestion increase from 140 to about 180
base pairs when H1 is bound [30], the binding sites are probably
situated at the extreme ends of the DNA in the core. The enthalpy
of H1 binding is zero, in contrast to the negative enthalpy changes
observed on binding the lysine-rich and arginine-rich pairs. This
difference may be due to the possibility that H1 is involved in
purely ionic interactions whereas the core histones bind via ionic
bonds with DNA but also via protein-protein bonds with each other.
The dissociation of H1 involves the net binding of 10 moles of
NaCl per mole of H1 dissociated. Clearly, salt binding is not
stoicheiometric since there are a total of about 60 positive charges
per molecule of H1.
The denatured form of H1 present at NaCl concentrations less

than 0.4 M does not form a complex with chromatin directly, although
it will bind to form a fully renatured complex. The greater extent
of dissociation at low ionic strengths is therefore caused by a
displacement of the equilibrium between bound and free native H1 by
the equilibrium between free native and free denatured H1. The salt-
induced renaturation of H1 involves the net binding of 6 moles of
}140 -200
base pairs
'8)140
& basepairs
Hl
Nucleosome core Complete nucleosome
particle
Fig. 15. The association of the arginine-rich and lysine-rich
histone pairs to form the 100 Ji fibre in terms of the chromatin
model described in ref. [15]. Bottom left: details of a nuc1eosome
core particle; bottom right: a complete nuc1eosome with molecules
of bound H1.
NaCl per mole of H1. This shows that electrostatic interactions

are important in maintaining the native tertiary structure of the
H1. Since the low-ionic-strength form of denatured H1 does not bind
to core particles whereas the native, structured form does, the
native conformation which includes a-helical structure, is essen-
tial for proper binding to occur. It has been suggested that the
primary site for H1-chromatin interaction is the c-termina1 half
of the molecule or the N-terminal, since these are the most basic
regions, whereas structure formation takes place in the central
region [29]. These results do not support this conclusion. They
show either that there must be structured regions in the COOH and
NH2 terminal ends, if these are the only sites of interaction, or
that the central structured region plays an important role in the
binding of H1.
164 O. R. BURTON ET AL.
Although HI binds reversibly to sheared chromatin it is clear

that the removal and rebinding of this histone to core particles
does not restore the 'native' properties of chromatin. Conditions
have not yet been found which will reverse the change which occurs
when native chromatin is rendered soluble by shearing. Until such
time as the native structure of chromatin can be fully restored
the role of HI in supercoiling the 100 K fibre must remain unclear.
Despite this drawback one possible function for HI becomes apparent
from a consideration of the way in which the 100 K fibre must fold
or coil to form the 300 K supercoil. Crosslinking studies using
very short crosslinking agents such as HCHO have shown that nucleo-
some core particles in solutions of moderate ionic strength make
contact through histone-histone interactions [3]. If it is assumed
that the core particles are identical in structure and that they
make contact through specific interactions then each core particle
is related to its neighbour along the fibre by symmetry rules
which give rise to a linear structure [15]. It is not possible to
coil the linear 100 K fibre so generated into a 300 K fibre without
breaking the symmetric interactions between nucleosomes despite
claims to the contrary [28]. It is suggested that the role of HI
is to break the symmetric contacts between the nucleosome cores by
binding to the DNA at the ends of the core particles and tightening
or loosening the DNA double helix (that is, inducing superturns).
This process would have the combined effect of breaking the inter-
nucleosome contacts and also providing sufficient free energy to
further coil the 100 K fibre into the 300 K fibre.
REFERENCES
1. Kornberg, R.D., Ann. Rev. Biochem. (1977), 931-954.

2. Felsenfeld, G., Nature (1978), 271, 115-122.
3. Hyde, J.E. and Walker, 1.0., FEBS Letts. (1975), 50, 150-154.
4. Noll, M., Nature (1974), 251, 249-
5. Noll, M., Thomas, J.O. and Kornberg, R.D., Science (1975), 187,
1203-1206.
6. Sollner-Webb, B. and Felsenfeld, G., Biochemistry (1975), 14,
2915-2920.
7. Axel, R., Biochemistry (1975), 14, 2921-2925.
8. Zubay, G. and Doty, P., J. Mol. Biol. (1959), 1, 1-21.
9. Henson, P. and Walker, 1.0., Eur. J. Biochem. (1970), 14, 345-
350.
10. Henson, P. and Walker, 1.0., Eur. J. Biochem. (1970), 16, 524.
11. Noll, M., Cell (1976),8, 349-355.
12. Burton, D.R., Hyde, J.E. and Walker, 1.0., FEBS Letts. (1975),
55, 77-80.
13. Skidmore, C.J., Walker, 1.0., Pardon, J.F. and Richards, B.M.,
FEBS Letts. (1973), 32, 175-178.
14. Henson, P. and Walker, 1.0., Eur. J. Biochem. (1971), 22, 1-4.
15. Hyde, J.E. and Walker, 1.0., Nucleic Acids Res. (1975), 2, 405-
421.
16. Wilhelm, X. and Champagne, M., Bur. J. Biochem. (1969), 10,

102.
17. Sollner-Webb, B., Camerini-Otero, R.D. and Fe1senfe1d, G.,
Cell (1976), 9, 179-193.
18. Camerini-Otero, R.D., Sollner-Webb, B. and Fe1senfe1d, G.,
Cell (1976), 8, 333-347.
19. Oudet, P., Germond, J.E., Sures, M., Ga11witz, D., Bellard, M.
and Chambon, P., Cold Spring Harbour Symp. Quant. BioI. (1977)
in press.
20. Elgin, S.R. and Weintraub, H., Ann. Rev. Biochem. (1975), 44,
725-794.
21. Bradbury, E.M. and Rattle, H.W.E., Bur. J. Biochem. (1972), 27,
270-281.
22. Lilley, D.M.J., Howarth, W., Clark, V.M., Pardon, J.F. and
Richards, B.M., Biochemistry (1975), 14, 4590-4600.
23. Burgen, A.S.V., Roberts, G.C.K. and Feeney, J., Nature (1975),
253, 753-755.
24. Weintraub, H., Palter, K. and Van Lente, F., Cell (1975), 6,
85-110.
25. Thomas, J.O. and Butler, P.J.G., J. Mol. BioI. (1977), 116,
769-781.
26. Bina-Stein, M. and Simpson, R.T., Cell (1977), 11, 609-618.
27. Weintraub, H., Worce1, A. and Alberts, B., Cell (1976), 9,
409-417.
28. Worce1, A. and Benyajati, C., Cell (1977), 12, 83-100.
29. Bradbury, E.M., Cary, P.O., Chapman, G.E., Crane-Robinson, C.,
Danby, S.E., Rattle, H.W.E., Bub1ik, M., Palau, J. and Avites,
F.J·., Bur. J. Biochem. (1975), 52, 605-613.
30. Kumar, H. and Walker, 1.0., unpublished observations.
NUCLEOSOME SHAPE AND STRUCTURE IN SOLUTION FROM FLOW
BIREFRINGENCE
R. E. Harrington
University of Nevada, Reno
Reno, Nevada 89557
As our knowledge of the general features of chromatin struc-

ture at all levels improves, we will necessarily wish to shift our
interest more and more toward an understanding of those dynamical
characteristics of chromatin structure that relate directl~ to
biological function. There is an increasing body of evidence, both
theoretical and experimental, that structural dynamics in chromatin
are to a degree controlled, and certainly modulated, by solvent
factors, partlcularly ionic strength and specific ion polyelectro-
lyte effects. Thus, an increasingly heavy experimental burden
will fall upon those biophysical methods capable of giving struc-
tural information in solution, and especially upon those capable of
interpretation under more-or-less arbitrary solvent conditions.
The number of biophysical methods which can provide this kind
of structural information on complex systems such as nucleosomes
or higher order chromatin chains in solution is not large. Spec-
troscopic and low angle scattering experiments involving X-rays and
neutrons have already made significant contributions to the struc-
tural problem. Hydrodynamic property measurements, particularly
sedimentation and translational diffusion coefficients from inelas-
tic light scattering, have provided good molecular weight data but
have been less successful, so far at least, in characterizing
particle shape and structure. t1uch of the latter difficulty comes
from the fact that the frictional coefficient must be related to an
assumed hydrodynamic model, and especially for more complex struc-
tures, the experimental data are only rarely consistent with a
single, unique structure. However, these methods can be exceedingly
useful for detecting stru~t~ral changes in nucleosomes and chro-
matin chains in solution. -
167
168 R. E. HARRINGTON
t10st of the hydrodynamic property work to date on chromatin

and its subunits has involved the complementary methods of sedi-
mentation and translational diffusion. For characterizing asym-
metric rigid particles and stiff chains, however, hydrodynamic
properties based upon rotational diffusion offer certain advan-
tages both in sensitivity and in clarity of interpretation. Two
such measurements are flow birefringence (or dichroism) and intrin-
sic viscosity. These methods are complementary in the sense that
they can be combined to remove much of the ambiguity associated
with the rotational frictional coefficient analogously to the
more usual combination of sedimentation and translational diffu-
sion.
Flow birefringence is an ancient and venerable method which
has received periodic attention by polymer physicists and fl~id
dynamicists since its simultaneous discovery in 1873 by Mach and
by Maxwell. 5 However, it was not until well into the present
century that the correct hydrodynami c basis of the so-called "1'1ax-
well effect" was elucidated, and a complete dynamical theory for
a suspension of rigid ellipsoids of revolution in 3-dimensional
velocity gradient flow based upon rotational diffusion in con-
tinuum fluids did not appear until 1939. 7 This theory, by Peter-
lin and Stuart, remains the accepted treatment for rigid particle
systems at the present time. f10re recent studies of the flow
birefringence phenomenon in 3-dimensions appears to have estab-
lished beyond question both the correctness and the completeness
of the Peterlin and Stuart hydrodynamic treatment. 8
The associated optical treatment, based upon the theory of
macroscopic continuum dielectrics, has been criticized in its
application to molecular systems, particularly in its ability to
apportion correctly the observed birefringence between the intrin-
sic and form or shape components. This question is of importance
because the form birefringence disappears only when the absolute
refractive indices of solvent and particles become equal, an
experimental condition not usually attainable with most biological
systems in physiological or quasi-physiological solvents. Experi-
mental studies have demonstrated that the Peterlin and Stuart
optical treatment evidently is satisfactory in its application to
DNA, however. 9 ,lD In any event, the interpretation of the extinc-
tion angle is based entirely upon hydrodynamic theory. Hence,
this quantity permits the rotational diffusion coefficient and
the size and shape parameters of the particle to be obtained
unambiguously from accurate experimental measurements, and thereby
provides, in favorable cases, an internal check upon the optical
theory of the birefringence.
The first really productive applications of flow birefrin-
gence to biological systems were the pioneering studies on fibrous
and gl obul ar proteins, fi rst by Edsall and subsequently by Scheraga
NUCLEOSOME SHAPE AND STRUCTURE IN SOLUTION 169
and others. These studies, along with others involving viruses and
other rigid systems, are documented ,1o?g with associated biblio-
graphy in several excellent reviews. - 6 The Peterlin and Stuart
rigid particle theory seems to describe remarkably well the flow'
birefringence of these systems when formal consideration of poly-
dispersity is included in the treatment. 17 Thus, a fairly exten-
sive precedent exists for using flow birefringence to characterize
the size and shape of rigid particle systems in solution through
the analysis of the rotational diffusion coefficient and the in-
trinsic and shape birefringence. However, the earlier studies were
limited, sometimes seriously, by available sensitivity in the mea-
surement of the flow birefringence and extinction angles, and the
earlier results may also have been complicated by the nonlinearity
inherent in the optical instruments used. Further application of
the method to small rigid macromolecules and structures of low
asymmetry in solution had to await the development of more sensi-
tive photoelectric instrumentation.
More recently, flow birefringence has provided considerable
information on DNA structure and conformation in aqueous solution.
DNA is a stiff-chain polymer under these conditions which assumes
an expanded coil conformation at high molecular weights. Its rota-
tional dynamics are therefore complicated and involve multiple
relaxation processes. The application of flow birefringence to
chain macromolecules has paralleled the development of chain dynam-
ical theory following the demonstration by Kuhn and Grun 18 that the
optical anisotropy of a chain element can be taken as proportional
to its mean square length. Current dynamical theory is based upon
the Gaussian subchain model as developed by Bueche, Rouse, Zimm,
Tschoegl and Peterlin. This model is equivalent in the high molec-
ular weight limit to the Kratky-Porod wormlike coil model which has
been used quite successfully to characterize many solution prop-
erties of DNA. These models and their applicability to biolorical
systems includ~Bg DNA have been reviewed by Bloomfield et ale 9 and
by Harrington.
Because of limitations inherent in the Gaussian subchain model,
real progress in the application of available theory to experimental
flow birefringence data on DNA could not occur until the development
of highly sensitive instrumentation capable of accurate measurement
near the limits of zero concentration and shear. In those labora-
tories where satisfactory instrumentation was developed, however,
progress was relatively rapid. Tsvetkov, Frisman and others in the
Soviet Union used flow birefringence to characterize a variety of
solution properties of DNA.14 Harrington in the United States has
determined the flow birefringence and extinction angles of DNA both
at very 10w9 ,lO and at very high 20 molecular weights. These inves-
tigations have provided considerable experimental information on
the structure of DNA in neutral aqueous buffers of near physiolog-
ical ionic strength including evidence in support of the Watson-
Crick B-form structure. Perhaps the most useful information to

come from flow birefringence studies upon high molecular weight
DNA was the characterization of chain stiffness. The Gaussian sub-
chain dynamical theory predicts that both the intrinsic birefrin-
gence 2l and the form birefringence 22 are related in a similar way
to the persistence length of the chain. These factors provide the
basis for Tsvetkov·s theoretical prediction that the ratio of the
Maxwell constant to the intrinsic viscosity should be directly pro-
portional to persistence length. 14 This prediction has been thor-
oughly substantiated for DNA.14,20
It is clear that flow birefringence has developed experimen-
tally and theoretically to the point where it offers real potential
to investigate chain stiffness and its relation to higher order
structure in chromatin fragments in dilute solution. Moreover,
since flow birefringence is a relatively simple yet highly sensi-
tive experiment using contemporary instrumentation and techniques,
it offers exciting possibilities for the extensive study of struc-
tural dynamics in chromatin as induced by solvent effects and the
binding of cationic proteins. Such investigations could provide
the conceptual precursors for understanding the true structure-
function relationships in this most labile and dynamic material.
THE NUCLEOSOME SHAPE IN SOLUTION: FLOW BIREFRINGENCE STUDIES

We have utilized the high available sensitivity of our photo-
electric flow birefringence apparatus to study the extinction angle
and flow birefringence of isolated, intact nucleosomes from chicken
erythrocytes. Because of the relatively small size and high sym-
metry of these particles, flow orientation is small and extremely
stable and sensitive equipment is required.
t1aterials
Isolated, fractionated nucleosome monomers from chicken eryth-
rocytes were prepared using standard methods 23 ,24 at the Division
of Biology, Oak Ridge National Laboratory, using facilities gen-
erously provided by Dr. Donald E. Olins. The whole nucleosome
preparations so obtained were dialyzed exhaustively against 0.1 M
KCl 0.2 mM EDTA solvent. The soluble nucleosome fraction obtained
from this treatment has been shown previously to contain negligible
Hl histone. 25
Samples were dialyzed against the KC1-EDTA solvent immediately
prior to flow birefringence runs. Final dialyzate was used as
reference solvent in all cases. In order to conserve material, a
total of 5 replicate runs were made at each sample concentration.
This latter procedure was required because of the relatively large
volume of the optical flow cell (-5ml). As an added precaution,

samples were dialyzed overnight against pure solvent between runs.
No changes in the flow birefringence, intrinsic viscosity, absor-
bance ratios at 230, 260 and 280 nm or circular dichroism were
noted with successive flow birefringence determinations. Hence,
the method as we employed it seems to be completely non-destructive.
General Methods
Concentrations of nucleosome solutions were determined on a
Zeiss PMQ-II UV-visible spectrop'hotometer using an extinction
coefficient for a 1% solution EJ% = 93.1. 25 The specific refrac-
tive index increment of KCl solu~ge nucleosomes in the 0.1 r'l K~l
0.2 mM EDTA solvent was determined as dn/dc = 0.184 ± 0.006 cm /gm
using a Phoenix differential refractometer with sample concentra-
tions determined spectrophotometrically as described above. The
solvent refractive index no = 1.33387 was determined on the same
differential refractometer using Phoenix standard KCl calibration
data.
The experimental intrinsic viscosity En] = 11.2 cm 3/gm was
measured in our laboratory at low shear rates using a cartesian
diver Couette-type viscometer. 20 The partial specific volume of
nucleosomes v = 0.661 cm 3/gm was measured in the laboratory of
Professor Walter E. Hill of the University of Montana on a Mettler-
Parr densitometer. Both of the above values have been reported
previously.23
Measurement of Flow Birefringence

In the classical method of measuring birefringence, the bire-
fringent medium is placed between crossed polarizers. The trans-
mittance of this system is related to the phase retardation along
the slow axis of the medium and hence to the birefringence, 6n,
given the optical path length, L, of the medium.
T = ~ sin 2 ~L6n (1 - cos 4S) (1)

X
In this expression, B is the angle between the slow axis of the

birefringent medium and the plane of polarization of the polarizer,
and X is the wavelength of the light. This equation shows that
elliptically polarized light is transmitted except when the slow
axis of the medium is along either polarization axis. Under these
conditions, a four-fold symmetric region of extinction is observed
corresponding to the symmetry in a. If the birefringence is due to
the hydrodynamic orientation of suspended anisotropic particles, as
it is in flow birefringence. it is clear that if the principal

optical and geometrical axes are coincident. the extinction angle.
X. is also the mean orientation angle of the particles with respect
to the streamlines of flow. These relations are illustrated for
the case of Couette (concentric cylinder) hydrodynamic flow in
Fi gure 1.
i-r------"-'+-::::----+--+-- Analyzer
Polarizer
Figure 1. Mean molecular orientation in the velocity gradient of

a Couette device (outer cylinder rotating clockwise as shown) for
rodlike particles. Relationships are indicated for the extinction
angle. X. with respect to the flow streamlines and for the cross
of isocline with respect to the polarizer plane.
In the optical system just described. the birefringence has a

quadratic dependence upon the transmittance (for small birefrin-
gence) so that different sources of birefringence such as that due
to the solute in solution. the solvent alone and various contribu-
tions from the optical components of the instrument itself mix
nonlinearly and cannot in general be separated. Such an optical
system cannot be used. therefore. to determine accurately very
small amounts of solute flow birefringence due to inherent noise
levels in any flow birefringence instrument. This problem affects
all optical flow birefringence equipment presently in use and
severely limits their usefulness in the study of small biological
macromolecules and aggregate structures.
The transmittance of the optical system just described can be
linearized if the polarizer and analyzer are at an angle somewhat
different from 90 0 and if a large amount of fixed birefringence is
deliberately introduced into the system. Under these conditions,

the transmittance is given by three terms: a large term of con-
stant transmittance independent of the birefringence; a term linear
in the birefringence; and a term quadratic in the birefringence
with a four-fold symmetry in S as in the crossed polaroid case above.
We have developed a highly sensitive photoelectric flow bire-
fringence instrument based upon the general principles just
described. In our instrument, the entire annulus of a Couette-
type flow cell is scanned by two spots of constant intensity light
located 180 0 opposite to one another. The transmitted intensity
is detected photoelectrically, and the linear term is isolated
electronically. A schematic representation of the instrument is
given in Figure 2.
d
~-7
I
o ~ (top view)
Figure 2. Schematic representation of the flow birefringence

instrument: (a,b,c) light source; (d) projection lens; (e) 546 nm
sharp cut-off filter; (f) light field compensation plates; (g) scan-
ning plate; (h) polarizer; (i) cross bars; (j) concentric cylinder
flow cell (rotating outer cylinder); (k) calibration plate;
(l) quarter wave plate; (m) analyzer; (n,o) photocell and optical
system; (q) angular scales; (p,r-b') electronic Fourier analyzer
Figure 2. (continued)
and time averaging system. Items (m,n,o) are rotated together to
set the angular variable y. Items (i,p) are rotated together to
set 6.
The intensity of light due to the linear term in the trans-

mittance is obtained by integrating the transmittance over the
dimensions of the scanning light beams.
8+CP
Ilinear =~ Tlineard8 = ± D sin 27T~L'ln sin 2y sin 2cp sin 2(8-X) (2)
8-Cp
In this expression, D. is the annular gap width, 2CP is the annular
region defined by the width of the scanning slits (g in Figure 2),
X is the extinction angle, y is the angle between the planes of the
polarizers, 8 is the annular position relative to the plane of the
first polarizer (h in Figure 2) and the positive and negative signs
are associated with clockwise and anticlockwise rotation respec-
tively. Thus, scanning the annulus produces a photomultiplier
signal whose intensity as a function of angle is proportional to
the flow birefringence (for small birefringence such that sin~n=~n)
and whose phase is proportional to the extinction angle.
It is clear from equation (2) that both extinction angle and
flow birefringence can be determined from independent measurements
at two values of 8. However, data reduction and time averaging
procedures are simplified if measurements are made at exactly
e = 00 and 8 r 45°. Under these conditions, equation (2) gives the
following for the photomultiplier output.
EC (45) = k(45)L'ln c cos 2xc
Ec(O) = -k(O)~nc sin 2Xc
(3)
Ea (45) = -k(45)~na cos 2Xa
Ea(O) k(O)~na sin 2Xa
In these equations, the subscript c refers to clockwise rotation of
the flow cell and a to anticlockwise rotation, and k is an instru-
ment constant determined by calibration against an optical standard
of known birefringence and phase (k in Figure 2).
Equations (3) will in general contain both the flow birefrin-
gence of the solution and extraneous birefringence ~n2 with phase
cos 2x2 due to the instrument itself. Since the instrument responds
only to the linear term in the birefringence, these effects are
additive. However, the flow birefringence changes sign with rota-

tional sense of the flow cell while the extraneous signal is con-
stant. Thus, the photomultiplier outputs of equation (3) can be
combined to isolate the extraneous term.
2A = EC (45) + Ea (45) = k(45)[lm c cos 2Xc - ~na cos 2Xa ]
+ 2k'~n2 cos 2X2
= -2k(45)~n cos 2X + 2k'~n2 cos 2X2 (4)
2B = Ec(O) - Ea(O) = k(O)[-M c sin 2Xc - Ma sin 2Xa]
= -2k(O)~n sin 2x
~n and X are now the true birefringence and extinction angle
obtained, in effect, as the average of clockwise and anticlockwise
flow cell rotation.
Since B in equation (4) contains no extraneous birefringence,
its intercept at zero velocity gradient, G, will be due to minor
errors in setting e or in measurement of flow cell angular velocity.
Such intercepts can always be removed adequately by fitting B to a
quadratic polynomial in G and subtracting the zero shear intercept.
However, particularly at extinction angles near 45°, the A function
in equation (4) may be quite nonlinear and determination of the
zero shear intercept may be difficult.
A satisfactory procedure for obtaining 2k' ~n2 cos 2X in
equation (4) has recently been described by Barrett and Hafringtonf 6
Since the extinction angle is particularly sensitive to the extra-
neous term in A, the procedure, in effect, maximizes the fit of
extinction angle versus shear rate data to the best smooth curve.
Although the method was developed for large chain macromolecules,
it is equally applicable to rigid particle systems such as
nucleosomes.
Most theories for the extinction angle predict a dependence
upon shear of the form
(5)
where the subscript ~ refers to an intercept-corrected value

and c is the residual correction to A. The constants kl and k2
are functions of shear dependent reduced viscosity and particle
shape parameters only, and are very nearly independent of shear.
To a first approximation, therefore,
_A_ = k 0 + k 1G + k G2 +
I I I (6)
Bcorr 2
Subtracting equation (6) from equation (5) and rearranging with
neglect of terms in G higher than the first power, one obtains
_ k' 0 c (
BcorrG - kl - k'l Bcorr + kl - k'l 7)
Thus, a plot of A/Bcorr versus G gives k'O as the intercept. This

value is then used along with a plot of BcorrG versus Bcorr in a
quadratic or higher order regression analysis to obtain the residual
correction.
c = ( cons tant coeffi ci ent ) kI
(8)
linear term coefficient 0
This procedure is performed by computer and reiterated until c is
a constant to the desired degree of accuracy. When a satisfactory
value of Acorr is obtained, the true extinction angle is calculated
from equation (5) and the flow birefringence from the relation
An = (A~orr + B~orr)~ (9)
This data reduction procedure results in an exceedingly high level

of experimental precision when the instrument functions A and B
are obtained from effectively time averaged values of E in
equation (3).
Because of the photoelectric detection feature and the fact
that the entire annulus is utilized in the detection of flow bire-
fringence, the instrument described is capable of extremely high
theoretical sensitivity. The principal limitation in this respect
is the available signal-to-noise ratio. Instrumental noise ranges
from the steady extraneous instrumental contribution and very slow
drift due to thermal changes in the optical bench components to
rapid optical noise in the flow cell and electrical noise in the
photomultiplier housing. Most of the noise is optical, however,
and comes from light reflections in the flow cell, scattering due
to particulate material in the flowing solution, and most impor-
tantly, refractive gradients in the flowing solution due to heating
at high shear rates. The signal-to-noise ratio is significantly
improved by time averaging the instrument outputs in equation (3)
over the same integral number of flow cell revolutions. In our
instrument, this is done automatically with an integrating recorder
as shown in Figure 2.
NUCLEOSOME SHAPE AND STRUCTURE IN SOLUTION
The instrumental factors just described become particularly

acute in studies on isolated nuc1eosomes in solution at low con-
centrations. Because of the small size and relatively low
asymmetry of these particles, both the flow orientation and the
flow birefringence are very small even at high velocity gradients.
Thus, the extraneous instrumental birefringence and the optical
noise peaks at the higher shear rates may be many times larger than
the true solution flow birefringence, and effective noise averaging
and correction to the A function becomes exceedingly critical.
This is especially true for the extinction angles; equation (5)
shows that if the extinction angle is very near 45°, the theoret-
ical limit for dominant rotation diffusion, the B function becomes
vanishingly small and determining its magnitude in the presence of
high levels of optical noise becomes quite difficult.
Flow Birefringence Run Protocol

We have determined the flow birefringence and extinction
angles of n~c1eosomes in the velocity gradient range 1,000 to
16,000 sec- (the present non-turbulent upper limit of our appara-
tus with these solutions). Although our flow cell components
including the support bearings are effectively thermostatted,
appreciable sample heating always occurred during runs due to
shearing at the higher shear rates. After a minute or more of
such shearing, thermal gradients were produced leading to severe
optical noise and drift. To minimize such effects, a standard run
protocol was developed: (1) the flow cell was brou~ht to speed
and equilibrated for 10 sec; (2) the output signal (equation (3))
was then time averaged for 25 sec; (3) the system was allowed to
re-equi1ibrate thermally at a low flow cell speed for 5 min or
longer. Under these conditions, sample heating was kept to less
than ~C and thermal noise and drift were held to tolerable levels.
Extinction Angle Data

Extinction angles extrapolated to zero concentration are shown
in Figure 3 as a function of shear rate. These data are remarkable
for their precision at very low particle orientation. The data are
compared to calculated lines for ellipsoids of axial ratios ~ (ob-
late) and 2 (prolate) using rigid particle theory.
The rotational diffusion coefficient for a rigid ellipsoid of
revolution is given as 10,27
(10)
in which o=Jv is a relatively insensitive function Of axial ratio,

p, and J and v are as defined by Perrin2tl and Simha 29 respectively.
The extinction angle is calculated from the rotational diffusion
coefficient using the theory of Peterlin and Stuart. 7
(G/D r ~ 1.5) X= ~
4
- _G_ [1 _ G2 (1
12D r 180D 2
+.2!c@..)+
35
.. :l(l1)
J
r
d = (p2 - 1)/(p2 +1)
X(deg)
p =1/2
p=2
2 4 6 8 10 12 14 16
G x 10- 3 (sec- 1)
Figure 3. Experimental extinction angle versus velocity ~radient

data. Solid lines: theoretical trends for oblate (p = ~) and
prolate (p = 2) ellipsoids of revolution.
The solid lines in Figure 3 are obtained from equation (11)

and rotational diffusion coefficients as given by equation (10).
Table I
oblate ellipsoid prolate ellipsoid
p = ~ p =2
o = 2.52 o = 1.93
Dr = 4. 49 x 10 5 Dr = 3.44 x 10 5
It is evident that the extinction angle data are most con-

sistent with the oblate model; the disagreement with the prolate
model is outside experimental error except at the lowest shear
rates. However, the scatter in the data are still large on the
scale required, and although the extinction angles definitely
imply an oblate model with axial ratio approximately ~, they are
not sufficiently precise to establish the axial ratio at a useful
level of precision.
Flow Birefringence Data

Flow birefringence data are shown in Figure 4 as functions of
shear rate. Each data point in this case represents the average
of 5 replicate determinations. The precision of the data is excel-
lent in comparison to previous flow birefringence work at high
shear rates, and is particularly so in view of the relatively small
birefringences measured.
The shear dependence of the flow birefringence was f~tted by
least squares to a quadratic polynomial: 6n = a + bG + cG. The
linear term coefficients are given in Table II.
Table II
c (ugm/ml) b = (M/G)G -+- 0
100 -9.028 x 10- 14
300 - 2•709 x 10 -13
500 -4.514 x 10- 13
700 -6.320 x 10- 13
The data in Table II are linear with zero intercept within exper-
imental error. A linear least squares analysis gives the Maxwell
constant from the slope.
En] = (c~n \ = -1.010 x 10- 7 (cgs units)
\1 nO)c,G-+-Q
E~1 = -8.418 x 10-9 (cgs units)
9
b..n x 10
10
2 4 6 8 10 12 14 16
G x 1O- 3 (sec- 1 )
Figure 4. Experimental flow birefringence versus velocity gradient
data. Sample concentrations are upper line to lower line respec-
tively: 700 ugm/ml; 500 ugm/ml; 300 ugm/ml; 100 ugm/ml.
The experimental data can be interpreted using the theory of

rigid particles. 10 The Maxwell constant is related to the an-
isotropy in polarizability per unit volume
[n] -_(L\n
~
~ _ 27Tvd
- 15nn 0 (gl - g2
) ( 12)
o c,G-+{) 0 r
where V is the partial specific volume of the particles, and nand
no are respectively the refractive index and absolute viscosity of
the solvent. Certain specific details of the theoretical model
are removed by replacing the rotational diffusion coefficient by
the experimental intrinsic viscosity.
~ = ~ iiM (g - g2) (l3)
[n] 5nkT aNA 1
In equation (13), Mis the particle molecular weight, NA is

Avogadro's number, k is the Boltzmann constant and T is absolute
temperature.
The individual polarizability components can also be deter-
mined from the experimental refractive index increment.
(14 )
Thus, the principal components of the absolute real refractive

index ellipsoid can be calculated from the relations
(y 1 - Y2 ) = Mv (g _ 9 ). (15 )
NA 1 2 1 nt
Yi = ~ (ni 2 - n2) (16)

4TI 1 + [(ni 2 - n2)/n2)Li
Ll = (l - e 2) [_1 ln~-ll
e2 2e 1- e J
Using equations (13)-(16), the absolute optical parameters

for nucleosomes are computed from the experimental data and
specific refractive index increment dn/dc = 0.184. Such calcula-
tions are shown for oblate and prolate models in Table III. In
both cases, axis 1 is the "symmetry" axis.
Table III
oblate prolate
(gl - g2) 3.26 x 10- 3 -2.49 x 10- 3
gl 6.12 x 10- 2 5.73 x 10- 2
g2 5.79 x 10- 2 5.98 x 10- 2
nl 1.665 1.598
n2 1.608 1.640
(n l - n2) 5.70 x 10-2 -4.20 x 10- 2
Comparison of the Data with Various Nucleosome Models

Several models for nucleosome shape and structure have recently
emerged based upon low angle neutron 30 ,3l and3~-ray25 scattering
data and upon X-ray crystallographic results. All these models
are consistent with an oblate ellipsoidal shape having an axial
ratio of roughly ~ and with the DNA wound in an overall equivalent
circular fashion around the minor (symmetry) axis. However, none
of these studies have yet been performed at high enough resolution
to determine the nucleosome structure unambiguously. Low angle
neutron scattering with H20/D20 contrast matching definitely in-
dicates a concentric structur.e with DNA on the outside of an inner
histone core. Furthermore, this core consists of an octamer con-
taining 2 each of the inner histones H2a, H2b, H3 and H4. 33 Some
evidence is available that this octamer may consist of two hetero-
typic tetramers as basic subunits. 34 This concept, combined with
the DNA structural model of Pardon et al. and Klug et a1. in which
140 nucleotide pairs of DNA wind in a superhelical trajectory of
1-3/4 turns around the histone core, offers an attractive hypoth-
esis for the first level of structural transformation in nucleo-
somes during the dynamic phases of transcription and replication:
a transition from a closed nuc1eosome to a more open structure of
some kind. Some experimental evidence for such a transition
already has been reported. 2 ,35
In view of the uncertainty still surrounding the nucleosome
structure, it will be of interest to compare the flow birefringence
results to current nucleosome models. If we model the nuc1eosome
after Pardon et a1. and Klug et al., we may define principal
refractive indices as shown in Figure 5.
Figure 5. DNA trajectory in the nucleosome based ~~on the oblate

ellipsoid model of Pardon et a1.30 and Klug et al. The core
histone octamer is not shown, but is presumed to be optically
isotropic.
The absolute refractive indices of DNA have been determined

previously from flow birefringence data as nl = nIl = 1.622 and
n2 = n. = 1.748' where the subscripts are with respect to the
helicaT axis. l If a constant 0.084 is added to the oblate refrac-
tive indices in Table III, the following relationships are obtained.
nl + 0.084 = 1.749 = n2 (DNA) (17)
_ nl (DNA) + n2 (DNA)
n2 + 0.084 = 1.692 = 2 (=1.686)
This degree of agreement with the oblate model may be in part for-
tuitous, although it must be remembered that the same methods and
same theory were used to obtain the refractive indices of both
nuc1eosomes and DNA. The results are certainly highly suggestive
of an oblate model with the DNA wound superhe1ical1y on the outside
of an optically isotropic inner histone core. The equivalent pro-
late model would show birefringence and anistropy of opposite sign
to that actually observed.
An alternative model consistent with the algebraic sign of
the observed flow birefringence would be a prolate model with the
DNA wound along the symmetry axis in some fashion. A possible
schematic representation for such a model is given in Figure 6.
(
~-~ 1
)
nz'nJ
Figure 6. Possible DNA trajectory in a nuc1eosome model based

upon a prolate ellipsoid with the DNA wound along the major (sym-
metry) axis. The core histone octamer is not shown.
For this
n1 + 0.109 = 1.707 ( 18)
n2 + 0.109 = 1.749 = n2 (DNA)
In this case, the sum n + 0.109 = 1.707 is quite different from
nl (DNA) = 1.622 and th~s prolate model would therefore be diffi-
cult to justify unless the histone core were highly optically
anisotropic.
The above results are suggestive of an oblate model for the

nucleosome in 0.1 MKCl along the lines described by Pardon, Klug
and others. In spite of the fact that the optical theory employed
is based upon macroscopic continuum dielectrics, we believe the
results have significance since the optical theory seems to pro-
vide a high degree of consistency between nucleosomes and DNA, and
the experimental and theoretical methods applied to both systems
are similar. Although the conclusions are highly preliminary, we
are encouraged by the degree of internal consistency between the
extinction angle and flow birefringence results and by the agree-
ment with other experimental methods, especially neutron scattering
from solutions and X-ray crystallography. In spite of the par-
ticular experimental difficulties in flow birefringence studies
upon such small, relatively symmetric particles such as nucleosomes,
the method is still a rapid and simple one compared to many other
structural techniques. It shows considerable promise, we believe,
in making a significant contribution to the vast array of struc-
tural problems in chromatin which must be solved if we are Ulti-
mately to understand in depth the dynamic character of chromatin
structure.
REFERENCES
1. G. S. Manning (1978), Quart. Rev. Biophys. (in press)
2. V. C. Gordon, C. M. Knobler, D. E. Olins and V. N. Schumaker
(1978), Proc. Natl. Acad. Sci., U.S.A., 75,660
3. K. S. Schmitz and B. Ramsay-Shaw (1977), Biopolymers, ~, 2619
4. K. S. Schmitz and B. Ramsay-Shaw (1978), Biophys. J., 21, 96a
5. M. E. Mach (1873), Optische-Akutische Versuche, Calve, Prague
6. J. C. Maxwell (1873), Proc. Roy. Soc. (London), ~-22, 46
7. A. Peterlin and H. A. Stuart (1939), Z. Physik, 112, 1, 129
8. P. J. Oriel and J. A. Schellman (1966), Biopolymers, 4, 469
9. J. L. Sarquis and R. E. Harrington (1969), J. Phys. Chern., 73
1685
10. R. E. Harrington (1970), J. Amer. Chern. Soc., 92, 6957
11. J. T. Edsall (1942), Advan. Colloid Sci., 1, 269
12. R. Cerf and H. A. Scheraga (1952), Chern. Rev., §I, 185
13. H. A. Scheraga and R. Signer (1960) in Physical Methods of
Organic Chemistry, Vol. I, Part III (A. Weissberger, Ed.),
Interscience, New York, Chapter 35
14. V. N. Tsvetkov (1963) in Newer Methods of Polymer Character-

ization (B. Ke, Ed.), Interscience, New York, Chapter 14
15. R. E. Harrington (1967) in Encyclopedia of Polymer Science and
Technology (N. M. Bika1es, H. F. Mark and N. G. Gaylord, Eds.)
Interscience, New York, Vol. 7, p. 100
16. A. Peter1in and P. Munk (1972) in Physical Methods of Chemistry,
Part IIIC (A. Weissberger and B. W. Rossiter, Eds.), Inter-
science, New York
17. E. D. Alcock and C. Sad ron (1935), Physics, ,2,.,92
18. W. Kuhn and F. Grun (1942), Ko110id-Z., 101,248
19. V. A. Bloomfield, D. M. Crothers and I. Tinoco, Jr. (1974) in
Physical Chemistry of Nucleic Acids, Harper and Row, New York,
Chapter 5
20. R. E. Harrington (1970), Biopo1ymers, 9, 159
21. B. H. Zimm (1956), J. Chern. Phys., 24, 269
22. R. Koyama (1961), J. Phys. Soc. Japan, l§.., 1366
23. D. E. 01ins, P. N. Bryan, R. E. Harrington, W. E. Hill and
A. L. 01ins (1977), Nucleic Acids Res., 1, 1911
24. A. L. 01ins, J. P. Brei11att, R. D. Carlson, M. B. Senior,
E. B. Wright and D. E. 01ins (1977) in The Molecular Biology
of the Mammalian Genetic Apparatus (P.O.P Tlso, Ed.),
Elsevier/North Holland, Amsterdam
25. A. L. Olins, R. D. Carlson, E. B. Wright and D. E. 01ins (1976),
Nucleic Acids Res., 1, 3271
26. T. W. Barrett and R. E. Harrington (1977), Biopo1ymers, l§.., 2167
27. H. A. Scheraga and L. Mande1kern (1953), J. Amer. Chern. Soc.,
75, 129
28. F. Perrin (1936), J. Phys. Radium, Z, 1
29. R. Simha (1940), J. Phys. Chern., 44, 25
30. J. F. Pardon, D. L. Worcester, J. C. Wooley, R. I. Cotter,
D.M.J. Lilley and B. M. Richards (1977), Nucleic Acids Res., 1,
3199
31. R. P. Hjelm, G. G. Kneale, P. Suau, J. P. Baldwin and
E. M. Bradbury (1977), Cell, lQ, 139
32. J. T. Finch, L. C. Cutter, D. Rhodes, R. S. Brown, B. Rushton,
M. Levitt and A. K1ug (1977), Nature, 269, 29
33. K. E. Van Holde, C. G. Sahasrabudde and B. Ramsay-Shaw (1974),
Nucleic Acids Res., 1, 1579
34. H. Weintraub, K. Palter and F. Van Lente (1975), Cell, ,2,.,85
35. P. Oudet, C. Spadafora and P. Chambon (1977), XLII Cold Spring
Harbor Symposium on Quantitative Biology
SCATTERING AND DIFFRACTION BY NEUTRONS AND X-RAYS IN THE STUDY
OF CHROMATIN
J.F. Pardon
Searle Research Laboratories
Lane End Road, HIGH WYCOMBE, Bucks, HP12 4HL, England
INTRODUCTION
X-ray scattering (1,2) and diffraction (3,4) have been used
to study the structure of chromatin since methods for the routine
preparation of histone-DNA (nucleohistone) complexes were devel-
oped (5). In comparison with the earlier X-ray diffraction studies
from fibres of DNA (6), the amount of information coming from these
studies has until recently been relatively small. The slow progress
was largely a result of the lack of detail available in the fibre
diffraction patterns. Whilst it was possible to examine a variety
of different models which might explain the limited data it was
not possible to derive a rigorous model. Data from other physical
and biochemical studies were, until recently, equally uninformative
and offered little assistance in the interpretation of the diffrac-
tion results, largely, as we now know, due to the nature of the
specimen preparation.
Initial scattering measurements from chromatin solutions (1,2)
and diffraction (3,4) from fibres showed that the DNA is far less
crystalline when combined with histones than when drawn into fibres
in the absence of protein. In patterns from chromatin, the dif-
fracted intensities corresponding to the layer lines of the DNA
double helix are diffuse and only poorly oriented.
The early X-ray studies of chromatin demonstrated the presence
of a series of low angle maxima (1,3) which are present both from
gels and fibres of chromatin. These maxima are present at spacings
greater than the dimensions of the double-helix and demonstrate the
presence of regular tertiary structure. They produced the first
evidence for a regular structure in chromatin other than the DNA
187
188 J. F. PARDON
double-helix. Studies of chromatin gels over a wide range of con-

centrations showed that the intensities of the diffraction maxima
vary with concentration (1). The series of low angle rings, the
poor orientation of the DNA and the concentration dependence of
the diffracted intensities were compatible with an extendable struc-
ture such as a super-helix (4,7). It was not possible to estimate
the possible extent of superhelical structure from the limited data
available. The regularity in the structure was attributed to the
DNA and the contribution made to the scattering intensities by the
histones was largely ignored. A less regular helical structure
was proposed to explain measurements made on dilute solutions of
chromatin (2) using X-ray scattering techniques.
A subsequent re-evaluation of the diffraction from concen-
trated gels included an analysis of the continuous scattering
profile upon which the maxima are superimposed (8). A model to
explain the complete scattering/diffraction data included short
regions of super-helix containing a few turns -1-3 -of helix inter-
dispersed with non-regular super-helical DNA. This approximates to
our present concept of chromatin structure although it did not
include a histone core.
The discovery of the sub-unit structure (9,10) resulted both
from bi~chemical and electron microscopy studies (11,12,13), although
the 110A X-ray spacing featured in early discussions of the spacing
between sub-units (9). A re-appraisal of the X-ray data had to
incorporate the concept of a central histone core around which the
DNA is wound. The contribution of the histones to the diffraction/
scattering maxima could no longer be neglected. When methods were
developed for isolating discrete chromatin particles (14) it was
possible to make the first unambiguous determinations of some of
the parameters of the sub-unit (15,16); also to discriminate between
those maxima which are part of the Fourier transform of the sub-unit,
as opposed to interference maxima arising from the assembly of sub-
units into either regular arrays or tertiary structures.
The availability of neutron scattering facilities (17,18) and
the development of the theory of contrast variation enabled a new
series of scattering and diffraction measurements to be made. These
methods, together with the availability of monodisperse solutions
of sub-units, rapidly led to the validation of the original prop-
osals for the structure. They were able to show unambiguously that
there is a protein core with DNA on the outside (15,16), and more
recently provided a model for the subunit structure to a resolution
of 25A (19). This model has subsequently been supported by X-ray
crystallography (20). It is now possible to re-interpret the orig-
inal data from gels and fibres of chromatin and make more detailed
discussions of models for the assembly of the sub-units into regular
tertiary configurations.
NEUTRONS AND X-RAYS IN THE STUDY OF CHROMATIN 189
In this chapter the recent developments in the study of chrom-

atin sub-units will be emphasised and the contribution made by
neutron techniques explained. In so far as it is possible, the
data from fibres and concentrated gels of chromatin will be re-
interpreted to include conclusions that can be drawn from the
structure of the sub-unit itself. Whilst much progress has been
made in studies of the sub-unit it will be apparent that the know-
ledge of higher levels of structure in chromatin is less complete.
The extent of regular high ordered structure in chromatin remains
to be determined.
THE LIMITATIONS OF THE EARLY X-RAY DIFFRACTION DATA
An indication of the amounts of information available in X-ray
diffraction patterns from fibres and gels of chromatin is shown in
figure 1. X-ray cameras with a low angle resolution of about 400~
have been used to study both fibres of chromatin maintained at
various relative humidities, to give concentrations in the range
60-100% w/w, and gels at lower concentrations. Oriented gels
produced by swelling fibres in suitable buffers show that the
lowest angle maximum at about 110,11. is meridiona11y oriented (8).
The degree of orientation obtained does not permit any distinction
between true meridional orientation and an arc apparently centred
on the meridian but resulting from two broad maxima which are
slightly off the meridian. The latter could arise from a helical
structure where the lowest angle maximum arises from the Bessel
function on the first layer line of the helix transform. Fibre
diffraction patterns demonstrate slight meridional orientation in
the 55~ and 35-38A maxima (3) and include equatorial arcs which
vary in position in the range 35-27A, depending upon the concen-
tration of the chromatin in the fibre. There have been no reports
of any pronouQced X-ray diffraction ~axima with spacings in the
range 150-400,11., consequently the 110,11. maximum is taken as the first
layer line repeat in disc~ssions of superhelical models (4,7).
Higher angle maxima, present at spacings at submultiples of 110A,
have relative intensities co~patible with the transform from a
regular helix with pitch 120,11. and di~meter 110A. At higher con-
centrations the intensity of the 110,11. reflection decreases and at
60% w/w this reflection is no longer present. This variation in
the relative intensities was originally thought to be due to the
interference between helices as they pack together to form either
paranemic or plectanemic bundles (7). The 110A maximum is not
observed from solutions and dilute gels where the maxima are super-
imposed on a strong central scattering profile (1,2,8,21,22). The
continuous central scatter was originally thought to arise either
from a non-regular helical structure (2) or from short lengths of
regular helix interdispersed with non-helical DNA (8).
With the limited data available, there was clearly no justifi-
190 J. F. PARDON
Fisure 1: X-ray diffraction pattern from a fibre of chromatin

malntained at 98% relative humidity. Inner diffraction rings at
approximatelx 55, 37, 27 and 22A. An equatorial reflection in the
region 27-3TA..
Inset: The c~ntra] region ftom a less concentrated gel showing
maximi at l08A., 55A, 37A, 27A., 22A. Equatorial maxima at 60A. and
27-37A.
cation for more complex calculations than were made to determine

the feasibility of simple models. The assumption was made that the
diffraction was predominantly arising from a regular DNA structure
with the histone evenly distributed along the double helix (7),
possibly in the large groove (23). Contrast variation studies with
X-rays could provide no new information since it was not possible
to adjust the scattering density of the solvent to give conditions
where the histone is the dominant component. Using sucrose as the
medium to generate different contrast conditions it was possible to
obtain conditions where the solvent scattering density equalled
that of the histone. Under these conditions, where the DNA domi-
nates, the X-ray scattering from gels is not dissimilar to that
from chromatin in water. It was not possible using sucrose to
obtain conditions where the histones dominate the scattering. Salt
solutions are inappropriate for contrast variation studies on
chromatin since histones are removed from DNA in the ionic strength
range O.S-2.0M.
The availability of neutron beams (17,18) and the development
of the theory for analysing neutron scattering from large molecules
(reviewed in 24), especially molecules with large internal fluc-
tuations in scattering density, provided new opportunities to put
the assumptions made in the original X-ray studies to test. Thus
it became possible to ask the following questions:-
(a) Do the scattering maxima all arise from a common struc-
ture?
(b) Do the scattering maxima predominantly arise from regular
DNA structures or possibly from histone complexes?
(c) How are the histones arranged relative to the DNA?
The small angle scattering camera at Harwell developed by
Haywood and Worcester (17) enabled measurements to be made covering
a similar range of spacings to those studied using X-rays. Although
more material was required, exposure times using the medium flux
source at Harwell were for fibres of chromatin similar to those used
in previous X-ray measurements. Much shorter times were required
using the small angle scattering camera Dll at Institute Laue-
Langevin at Grenoble (18).
The combination of X-ray and neutron scattering studies on
solutions of isolated subunits have provided a model for the subunit
and these data together with studies of crystals of core particles
permit a more detailed interpretation of the original X-ray data
from fibres and gels.
192 J. F. PARDON
CONTRAST VARIATION WITH NEUTRONS

This technique utilises the large difference between the
coherent scattering amplitude of hydrogen (-3.74 x 10-13cm) and that
of deuterium (6.67 x 10-13cm ). It is possible (25) to adjust the
solvent scattering density using mixtures of D20 and H20 such that
DNA and histone each have scattering densities greater than or less
than the solvent or alternatively such that either the histone
(60% D20) or the DNA dominate the scattering (40% D20) - see Table
1. It is important to note that the scattering density of the
chromatin varies with the D20/H20 ratio of the solvent since there
is exchange of labile proteins.
In addition to varying the scattering density of the solvent
it is also possible to vary the internal fluctuations in scattering
density by deuterium labelling (26,27). This can be achieved by
extracting either histones or DNA from cell cultures maintained in
a medium rich in D20 and subsequently reconstituting chromatin using
a mixture of deuterated and protonated components '(28). The fol-
lowing discussion relates to the former, namely the influence on
the scattering profile of changing the solvent scattering density.
a. The Scattering Profile I(k)

Ibel and Stuhrmann (29) showed that a convenient way of
expressing the excess scattering density p(r) of dissolved particles
is to divide it into two parts such that:-
(1 )
TABLE 1
Coherent Scattering Amplitude Densities
H2O -0.0056
D20 0.0638
Histone * 0.021
(39% D2O)
DNA* 0.036
(60.5% D2O)
*Data taken from sludies of isolated chromatin core particles (15) .
Units of 10- 12 cm/ 3.
where p is the difference between the mean scattering density of

the protein and the scattering density of the solvent. PF(r) is
a dimensionless function that is unity inside regions of the particle
inaccessible to solvent molecules and zerQ elsewhere. ps(r) des-
cribes the variation of scattering density inside the particles
around the mean value. The scattering amplitude can therefore be
written:-
(2)
where the momentum transfer k = 4n sin a/A for scattering angle 2a

and wavelength A. Stuhrmann and Kirste (31) showed that the scat-
tering intensity can be divided into three basic scattering func-
tions:-
I(k) = <A(k)2) = p? IF(k) + p. IFs(k) + Is(k) (3)
and lIFs(k)\ ~ 2..f{ IF(k) x Is(k) } (4)

-+ -+
Is(k) and IFs(k) -+ °for k °
-+
Is(k) is observed under conditions of zero contrast, i.e.

the mean scattering density of the particle equals the scattering
density of the solvent. IF(k) is the intensity under infinite
contrast conditions, i.e. only the shape of the particle contributes
to the scattering.
This theory is only true where the particles are not affected
by the exchange of atoms with the ~olvent. Thus for the neutron
scattering of particles using H20/ H20 mixtures:-
-+
p(r)=p. [- + -+] +ps(r)
PF(r)-PE(r) -+ (5)
where PE(r) defines the posit~ons and density of exchangeable

atoms. With (5) I (k) = p Ic(k) + p1cs(k) + Is(k) (6)
By measuring the distribution of intensities as a function of k
for at least three different contrast conditions it is therefore
possible to solve 6 and obtain the three functions Ic(k), Ics(k)
and I (k). Is(k) represents the internal structure function
proviaing information of the distribution of scattering intensities
about the mean. Ics(k) is a function of both the shape and the
internal structure of the particle. Ic(k) is a function mainly due
to the overall shape of the particle with a small component arising
from the distribution of exchangeable sites.
194 J. F. PARDON
b. The Guinier Region

For very small scattering angles:-
I(k) = -RG 2k2

1(0) exp _ _ (7)
3
RG is the radius of gyration where:-
RG 2 = p~ J V r 2.( p(r) - p(s)) d3r
for solvent scattering density p(s). This is Guinier's Law (30).
This law is extremely useful but can only be used for a monodis-
2
pers solution of particles where log I(k) plotted as a function
of k is strictly linear. It is important to collect data within
the Guinier region to enable radii of gyration to be determined.
Thus in general the maximum value of k for a spherical particle
is given by kRg = 1.3 and for a rod-like particle by kRg = 0.7.
In most of the neutron studies of chromatin and chromatin particles
the Guinier conditions have been met, in previous X-ray scattering
this has not always been so.
The measured radius of gyration is a function of the difference
between the mean scattering density of the particle and the density
of the solvent (29). It can be expressed in the following way:-
Rg2 = Rc 2 + a _ (3 (8)
~ ~2
Rc represents the radius of gyration at infinite contrast. a and (3
depend both on the solvent and internal fluctuations in the scat-
tering density. a has the significance of a mean square distance
of fluctuations from the centre of shape and (3 is the square of the
mean distance of the fluctuations.
In practice, the sign and magnitude of a determines whether
the response of higher scattering density are towards the outside
or the inside of the particle. (3 is useful in two or more component
systems since it can be used to determine whether the centres of
(3 = 0 and a plot of Rg
2
shape are coincident O not. If the centres of shape are coincident
versus /~ is linear. Since the chromatin
subunit contains two components, DNA and protein, with different
scattering densities an analysis of the variation in Rg with con-
trast provides a measure of the average distribution of these com-
ponents in the particle. Subsequent analysis of the entire scat-
tering curve can be used in two ways:-
(a) By collecting data for at least three different solvent
scattering densities and calculating the three functions
Ic(k), Ics(k) and Is(k).

(b) By collecting data under high contrast conditions and
calculating the scattering profile for various models
which are constrained to comply with the known variation
in radius of gyration for different solvent contrast con-
ditions. It is important to include realistic variations
in the internal scattering densities for any model con-
sidered. Since both the structure of the DNA and the
composition of the subunit are known, the number of models
which can be constructed, subject to the constraints
imposed by the radius of gyration data, is limited. Thus
this second approach was the first to be used to investi-
gate the structure of the core particle and is described
in some detail in the following two sections.
RADIUS OF GYRATION MEASUREMENTS ON CORE PARTICLES AND HISTONES

Chromatin core particles (14) were originally studied (15)
using the small angle scattering camera at Harwell (17). Measure-
ments were made on 11 mg/ml solutions in 020' 75% 020, 25% 020 and
H20. Additional data were collected for 5 mg/ml and 2.5 mg/ml solu-
tlons in 020. In each instance the Guinier region was linear allow-
ing the radius of gyration to be determined and the zero angle
scattering intensity 10 to be obtained by extrapolation. The plot
of II.: as a function of the mole fraction of 020 in the buffer
provi8ed a mean scattering density for the particle equivalent to
the scattering density of 49% 020 (figure 2). This is the value
predicted from the composition of the particle and the mean scat-
tering length densities of the components with 76% of the labile
protons exchanging with the solvent.
As mentioned in the previous section, the radius of gyration of
an object with internal variation of scattering density is a function
of the scattering density and hence isotopic composition of the
solvent. H~ving measured the mean scattering density of the particle
from the 10 2 versus % 020 plot the relationship between Rg2 and
lip (equation 8) was investigated as shown in figure 3. The size of
the error bars for the data points under low contrast conditions
illustrates the limitations of using a medium flux reactor for
scattering studies of this type. From these limited data it was
concluded that B has a value close to zero. Thus the regions of
high scattering density, within the accuracy of these data, are con-
centric with the regions of lower scattering density. The slope of
the line in figure 3 shows that the particles are composed of regions
of very different scattering den 6ity. The value of a for the chrom-
atin particles, 508 (± 80) x 10- (15), is much smaller than found
for lipoprotein (2800 x 10- 6 ) (32) but significantly greater than the
value found for globular proteins (35 x 10- 6 ) (29). The sign and
196 J. F. PARDON
150
100
50
I~
o
50
100
150
Figure 2: Plot of square root of zero angle scattering intensity

(Io~) versus mole fraction of D20 in the buffer for a 11 mg/ml
solution of chromatin core partlcles.
slope demonstrate that the material with higher scattering density

(the DNA) is at a greater radius than regions of lowrr scattering
density. By extrapolating from the graph of Rg vs /p it is pos-
sible to obtain radii of gyration where the DNA dominates the scat-
tering and conversely where the protein dominates. Since the DNA
is towards the outside and 76% of labile protons exchange with the
solvent it was assumed that complete exchange occurs for labile
protons in the DNA. Thus the mean scattering density of the DNA is
equal to that of the solvent when it is 60.5% D20. Similarly
allowing for complete exchange of the protons in the DNA only 68%
of the labile protons of the histones exchange with the solvent
giving equality between the mean scattering density of the protein
and the water for 39.0% D20.
Using values of p for 60.5% and 39% D20 and assuming 8 = 0, the
radii of gyration under conditions where the protein 9 dominates and
°
where the D~A dominates are 30.6 ± 2A and 50.5 ± 1.4A respectively.
Also where /p = (conditions of infinite contrast) the radius of
gyration is that of the particle with the same shape and dimensions
but no internal variation of scattering intensity. Under this
3000
2-0 1-0 o 1-0

1jp
Figur~: The variation of the radius of gyratio~2of isolated core

particles as a function of lip.
conditi?n the radius of gyr~tion extrapolated from the plot of

Rg2 vs /~ is 41.1 (± 0.4) A. More precise measurements using the
higher flux neutron facility at Grenoble (33) have produced slightly
smaller values for the radii of gyration but their conclusions are
similar. These studies unambiguously showed that the DNA is on the
outside of the core particle, surrounding a protein core, with a
small amount of overlap between the DNA and the histone.
Additional studies of a complex of histones H2A, H2B, H3 and
H4 isolated from non-Sheal"ed chromatin using 2.0M NaCl at pH 9.0
were made over a concentration range 2.5-40 mg/ml. The radius of
gyration of this complex was not dependent on the conkentration or
the D20/H20 ratio. The value obtained was 30.1 ± 0.3A, similar to
the value for the intact core particle under conditions where the
protein dominates the scattering (34).
198 J. F. PARDON
A MODEL FOR THE CHROMATIN CORE PARTICLE DERIVED FROM

SCATTERING CURVES FROM CORE PARTICLES AND ISOLATED HISTONE COMPLEXES
The values of the radii of gyration for the core particle
provide information of the gross fluctuations in internal scattering
density. However, a variety of different models can account for
these radii. This choice of models is limited to those where the
biochemical data dictated that the,DNA is constrained to ~ 140 base
pair segment with length about 476A and diameter about 20A. The
number of models was further reduced by a consideration of the scat-
tering data from the isolated core protein. Since the 'native'
X-ray diffraction pattern for chromatin is obtained when DNA is
added to isolated core protein this complex almost certainly retains
its native conformation after isolation. Using haemoglobin, oval-
bumin and human albumin as protein standards the molecular weight
of isolated core protein was determined from the intensities for
zero angle neutron scattering. The value obtained, 60,500 ± 5,000,
is roughly equivalent to a tetramer of histones H2A, H2B, H3 and
H4. The small difference between the measured radii of gyration
between the isolated protein (a tetramer) and the histone core (an
octamer) in the intact particle suggests that in the particle two
tetrameric histone complexes stack directly on top of each other.
If they are laterally spaced with respect to each other a larger
variation in the radius of gyration would have been observed.
Furthermore, the high angle scattering curve from isolated core
protein is unlike that calculated for a spherical protein (34) but
more like profiles calculated for a variety of disc-like structures
(figure 4).
The maximum dimensions of the core particle and core protein
were determined to within an accuracy of lOA by calculating pair
probability functions. The scattering intensities can be used to
generate a function, similar to the Patterson function used in
crystallography, which gives the probability for vector distances
between volume elements as a function of radius. To generate this
function the scattering curves have to be extended in two directions:
(a) To zero scattering angle. This can be achieved using
the Guinier Law (30);
(b) To high angles using the Porod Law I(Q) ~ Q-4 (35).
From the data shown in figure 5, and using the rules of Guinier
and of Porod to extrapolate the data, the pair probability functions
were calculated using the expression:
j[ Q.I(Q) Sin (Q.r)dQ

l1li
P(r) = __1
__
21T2r 0
1·0
0 ·1
0·01
Inl.nall,
0 ·001
Figure 4: Experimental and calculated neutron scattering profiles

for histone core protein in D20.
A Spherical models. Uniform sphere radius 35A (a) and spheres
with compact centres and less dense outer domains (b) and (c).
B More squat structures with geight in the range 25-50A and
diameter in the range 82-92A (d)-(f). More details of models
in reference (34).
C Scattering profile obtained from a solution of core protein in
D20, 2.0M NaCl (g).
where the scattering vector Q has magnitude 4n sin e/A for scattering
angle 2e and neutron wavelength A. The curv~s obtained are shown
°
in figure 6. Maximum dimensions of 118 ± lOA and 80 ± lOA can be
determined from the p(r) = intercepts for the core particle and
core protein respectively (36).
These combined data therefore restricted the possible models
for the core particle to include:-
200 J. F. PARDON
' ,0
Intensity
0 ,'
0 ·2 0 ·3
Figure 5: Neutron scattering profiles from:-

a Chicken erythrocyte core protein at 12 mg/ml in D20, 2.0M NaCl;
b Chromatin core particles from chicken erythrocyte nuclei at
9 mg/ml in D20, 10mM Tris 0.7mM EDTA at pD 7.1.
Data collected using the small angle scattering camera at AERE
Harwell (17).p Q = 4~ sin 8/ A where scattering angles 28 and wave-
length A (4.7A). The wavelength spread ~ A / A = 2% FWHM.
(a) An internal protein core diameter about 80~ composed of

two tetrameric disc-shaped protein complexes stacked
directly on top of each other;
o
(b) A ~ylindrical DNA molecule of length 476A and diameter
20A wound or folded on the outside of the protein with a
small degree of protein/DNA overlap;
(c) Radii of gyration of 30.6 and 50 . 5A tor the histone and
DNA dominating respectively and 4l.1A under infinite
contrast conditions. A radius of gyration of 41 .6A was
obtained from X-ray scattering measurements using the
position sensitive detector in the laboratory of Dr. D.
Sadler;
1-0
p( r)
0 -6
cor. palrlicle
0 -2 <MO\\ ,,10 "10
40 80 120
Fi gure 6: Probabil ity functions deri ved from the sca tteri ng
profiles shown in figure 5 (36).
(d) A maximum dimension for the core particle of 118 ± lOA,

suggesting a diameter of about 110A.
These factors severly limit the choice of model. Additional data
were also available from electron microscopy in at least three dif-
ferent laboratories . This can be summarised as follows:-
1. The DNA tends to enter and leave the particle on one
side (37);
2. The DNA is concentrated at the top and bottom of the
particle (38);
3. The particle is disc-shaped (39) .
Initial calculations indicated thatothe relative neutron scattering
intensities in the region of the 35A maximum and the shape of the
scattering profile were inconsistent with a roughly spherical par-
ticle (40). Therefore cylindrical structures approximating to both
prolate and oblate ellipsoids were considered.
Scattering profiles were calculated using the Debye equation
202 J. F. PARDON
(41):-
I(Q) =l:'. L:. p.,

sin (Q.rjj )
p.J
, J Q. rij
Model structures were generated by dividing the model into a number,
typically 700, of scattering volumes of equal size separated by
distances rij . These were assigned scattering densities (p) equiv-
alent to those for histone, DNA, nucleoprotein or a mixture of one
of these components and solvent. The scattering density of DNA was
given the value 1.0 relative to the solvent. Calculations were
also made for X-ray scattering profiles since data was also obtained
using X-ray techniques. The subsequent calculations (40) can be
summarised as fo11ows:-
o /)
(a) Neither cylindrical particles 110A high with 110A diameter
nor spherical particles with diameter 110A explain the
neutron scattering data. A series of ~axima are obtained
rather than the single rather broad 35A maximum (figure 7a);
(b) Prolate structures do not explain the neutron data;
(c) Structures with a hole through their centre are unsuitable;
(d) Oblate structures with height 50-55A and diameter 110~ give
the correct type of scattering profile only when the DNA
is confined to two loops at the top and bottom of the
particle (figure 7b);
(e) The refined model shown in figure 7 fits the neutron data
well (figure 7c) and also provides the following features
of the X-ray scattering profile obtained from solutions
of core particles (42):-
o
1. A shoulder at ~ 60A;
/) /)
2. Poorly defined maxima at about 38A and 28A;

o
3. A pronQunced minimum between the 38A maximum and
the 60A shoulder (figure 7d).
These are also the main features of the scattering pro-
file from dilute solutions of intact chromatin at low
ionic strength.
This final model was subsequently validated by X-ray crystal-
lographic work by Finch et £1. (20). More recently Suau et £1. (33)
concluded that it can account for the shape function Ic(Q) obtained
from solutions of core particles - see equation 6. Thus studies of
core particles both in solution and in cry~ta1s have provided very
similar models to a resolution of about 25A.
TABLE II
Calculated Separation
Assemblies with
all particles
Experimental Linear equidistant from
Oligomer size Radius of Gyration assemblies centre of mass
0 0
Dimer with 51 ± 2A 66 ± 6ft. 66 ± 6A
H1 and H5
0 0 0
Dimer without 72 ± 2A 121 ± 6A 121 ± 6A
H1 and H5
5A•
0 0
Trimer wi th 84 ± 1A 91 ± 128 ± 4A
H1 and H5
2A• 3A•
0
Trimer without 93 ± 103 ± 145 ± 6A
H1 and H5
Hexamer with 125 1A• 70
. .
± ± 2A 119 ± 1A
H1 and H5
For point~ lying on a helix with ~itch 220ft. and radius 110A: trimer
has Rg 95A and hexamer has Rg 125A. Partic1e.separation along helix
107A.
RADIUS OF GYRATION MEASUREMENTS FROM SMALL CHROMATIN OLIGOMERS

It is not obvious from the structure of the core particle how
the chromatin subunits assemble themselves to form the unit chroma-
tin thread. The electron micrographs of Finch et £1. (20) show
that isolated core particles can, under some conditions, stack
a1~ost directly on top of each other to give a separation of about
55A. However in chromatin these particles are connected by a
length of up to 60 base pairs of DNA which may have histone H1
attached in some way that has an influence on th~ structure of the
chromatin thread. Thus there could be up to 300A between the
centres of the particles.
Sperling and Tardieu (21) have shown from small angle X-ray
scattering studies that in 0.2mM EDTA the electronic mass per unit
length of chromatin is 1240 eLA, corresponding to a separation
between subunits Qf about 110J1.. They do not record interference
maxima in the 100A region and therefore conclude that the particles
are contiguous to form a more or less uniform thread.
204 J . F. PARDON
b
, ,0
Intensity
0,'
0-0,
0 ,' 0 ·2 0'3 0-, 0·2 0 ·3

Q (.0.-', h (... - ' )
C d
Figure 7
Figure 7:
a Neutron scattering profile from a 5 mg/ml solution of core
particles in D20, 10mM Tris 0_7mM EDTA (1) and the calculated
scattering profile for a spherical model a~proximating to an
inner sphere of protein (radius 39.5A) surrounded by a shell
containing DNA (radius 53~) (2);
b Calculated neutroQ scattering for oblate structures (height
50A, diameter 110A) with the DNA located in domains with
scattering density 1.0 and the histone at the centre (scat-
tering density 1.54). Cross-sections of the models are
shown in the inserts. Experimental scattering profile (D);
c Calculated and experimental neutron profiles for the core
particle. The dimensions and distribution of scattering
intensities are shown in the schematic diagram which represents
a section through the model along the cylindrical axis. The
domain with scattering density 1.0 corresponds to a short
segment of DNA joining two three-quarter filled annuli con-
taining DNA and histone tails (scattering density 1.02). The
protein centre has a dense centre (scattering density 1.54)
surrounded above and below by more hydrated regions (1.08);
experimental profile
o
••••• model with h 50A
o
-- .--.-. model with h = 53A
d The calculated X-ray scattering profile for the model for the
core particle showing the major features of the experimental
X-ray scattering profile from a solution of particles .
..
A somewhat different approach was to isolate small oligomers

containing a limited number of subunits and to measure their radii
of gyration using the small angle neutron scattering camera 011
(18) at the Institute Laue-Langevin at Grenoble. A summary of the
results from oligomers purified by Ti14 zonal ultracentrifugation
and containing two, three and six subunits respectively is shown
in Table II. The radii of gyration were calculated from the slopes
of the Guinier plots shown in figure 8. The calculated separation
between particles is shown both for linear strings of subunits and
also for more compact clusters similar to those observed in the
electron microscope. These data indicate that the particles are
not widely spaced but have centre-to~entre separations in the range
206 J. F. PARDON
0-6 a
04
0-2 b
0-'
'-0 c
0-6
Intensity
0-2 d
0-'
O-S e
0-6
30
Figure 8: Guinier plots from solutions of chromatin oligomers iso-

lated from chicken erythrocytes in 020, 10mM Tris, 0.7 EOTA pO 8.6.
a 11 mg/ml solution of di-nucleosomes containing histones Hl
and H5;
b 5 mg/ml solution of di-nucleosomes with histones Hl and H5
removed;
c 10 mg/ml solution of tri-nucleosomes with histones Hl and H5;
d 4 mg/ml solution of tri-nucleosomes with histones Hl and H5
removed;
e 6 mg/ml solution of oligomers each containing six nucleosomes
with histones Hl and H5.
Oata obtained using the 011 instrument at Grenoble (18).
o
66-145A.
The higher angle scattering profiles of both dimers and trimers
(small oligomers containing two and three nucleosomes) have a single
o
broad maximum at 35A similar to that obtained for single core par-
ticles. This suggests that under the low ionic strength conditions
(lOmM Tris, 0.7mM EOTA) used for these studies the individual par-
ticles behave as if they are scattering independently of each other.
This would suggest a rather flexible link between particles as was
originally proposed by Kornberg (9) rather than a rigid arrangement
whereby the particles either stack directly on top of each other or
alternatively assemble themselves edge-to-edge. This may not be
the case for larger oligomers where there might be additional stab-
ilising forces which could constrain the subunits to adopt a more
regular tertiary structure. It is int~resting to note that the
radius of gyration of a hexamer is 125A. This is the value obtained
for one turn of helix containing six subunits and having a pitch of
220~ and radius 100A. Such a helix is also able to account for
radii of gyration of larger oligomers at the same ionic strength
but studied by laser light scattering (43).
NEUTRON SCATTERING FROM FIBRES OF INTACT CHROMATIN

The position and intensity of the lowest angle diffraction maxi-
mum in X-ray patterns obtained from fibres of chromatin in the con-
centration range 40-70% w/w is dependent on the concentrati2n.
Thus for a 45% w/w gel there is a strong maximum at 100-110~ (8).
At 60% w/w there is no maximum observed in the spacing range 70-120~
(1,3,8) and in more con~entrated gels a strong intensity maximum is
observed at about 75-80~ (3). An early attempt to explain the reduc-
tion in intensity and subsequent loss of the 110~ maximum in terms
of interdigitating super-helices was based on the assumption that
it arises largely from superhelical DNA (7). Studies of chromatin
fibres using neutrons were able to put this assumption to the test
and prove it to be invalid (25,44).
°,
For chromatin in O2 in the concentration range 40-70% w/w, the
neutron scattering maXlmum at 80-110~ is present at all concen-
trations (25), the spacing decreasing gradually with increase in
chromatin concentration. Thus under conditions where a maximum is
not recorded with X-rays (fibres maintained at 98% relative humidity)
a strong maximum is recorded with neutrons (figure 9a). Furthermore
the variation in intensity of this maximum with solvent contrast
provides information relating to the distribution of DNA and histone
along the direction of the fibre axis. In general terms, for the
maximum to originate from a largely DNA repeat the 020 percentage
at which its intensity goes through a minimum has to be about 60%.
For a protein repeat the 020 percentage for a minimum intensity is
nearer 40%. For a nucleoprotein structurei such as a superhelix
evenly coated with an equal mass of histone, an intermediate value
would be expected. The minimum intensity for the 82~ maximum from
fibres at 98% relative humidity occurs at neither of these values
but at about 10% 020 (figure 9a).
208 J. F. PARDON
Int....ity
3 6 9
28 (degrees)
5000
01000
Intensity
(counts)
B
3000
200
1000 L-----,..----.-_-,.-_-.-_......_--.•....---'
28·
Figure 9
Figure 9: Neutron diffraction from fibres of chromatin. Data

obtained by scanning along the meridian using the small angle scat-
tering camera at Harwell (17).
(A) The effect of varying the D20/H20 ratio in solutions of
°,
potassium chlorate used to equilibrate the fibres to a
relative humidity of 98%; (a) 020' (b) 75% O2 (c) 50% 020'
(d) 25% 020 and (e) H20;
(8) The effect of reducing the relative humidity of the fibres
using three different salt solutions each dissolved in 100%
O2°.
Under these contrast conditions (10% O2°), domains containing
hydrated DNA will scatter with an average scattering density equal
to relatively unhydrated protein domains. The two types of domain
will not be resolved and any interference maxima arising from alter-
nating protein rich and DNA rich regions will not be present in the
scattering profiles. For neutron scattering in 020 solutions the
protein domains have a higher contrast than the DNA (relative to
solvent scattering) and the interference maxima will be present at
all chromatin concentrations (figure 9a). The densities of histone
and DNA for X-ray scattering are approximately the same as those Q
for neutron scattering in 10% 020. Thus the absence of the 80-110A
X-ray reflection from fibres of chromatin at 60% w/w and its pres-
ence at all concentrations of chromatin for neutron scattering in
DZO indicates a regular protein/DNA repeat along the fibre axis
(l.e. the data were collected along the meridian).
o
The intensity of the neutron 82A maximum from fibres of chromatin
(figure 9b) decreases as the material becomes more concentrated (44).
This again supports a model where there are alternating relatively
unhydrated protein cores contrasting with hydrated DNA. As water is
removed from the DNA containing regions the contrast between those
regions and the protein decreases and the scattering maximum becomes
less intense. With X-rays the opposite occurs and the low angle
maximum becomes the most intense and easily recorded of all the
scattering/diffraction maxima from fibres of chromatin.
For chromatin concentrations below 50% w/w unoriented gels have 0
been examined. The position of the low angle maximum ~hanges to 108A
for gels with concentrations 30 and 38% w/w and to 200A for ~% gels
(44). Furthermore the variation in the intensity of the 200~ maximum
with 020 percentage has a minimum at ibout 55% 020. This indicates
that within the 6% gel there is a 200A nucleoprotein repeat. Such
a repeat might represent the separation of chromatin threads or
alternatively arise from a higher order of structure within the indi-
vidual thread, such as a helix. Additional studies are required to
distinguish between these two possibilities.
210 J. F. PARDON
X-RAY FIBRE DIFFRACTION RE-EVALUATED

Recent progress resulting from the use of neutrons and studies
of isolated chromatin subunits demands a review of the long estab-
lished X-ray fibre diffraction data. The intensity variation with
concentration of the 80-110~ maximum has already been re-interpreted
in the previous section. In this section we consider the origins
of the diffraction maxima and in doing so invoke information from
the recent studies of crystals of core particles.
Despite early expectations, contrast variation methods in
neutron diffraction studies of fibres of chromatin were unable to
distinguish regorously between the various possible origins of the
strongest of the diffraction maxima (25). X-ra~ solution scattering
studies of core particles indicated that the 37~, 27~ and 22~ maxima
are present in the Fourier transform of the.isolated subunit (42),
which also contains a shoulder in the 55-60A region.
o
The 110A reflection is not present in profiles from dilute solu-
tions (~20% w/w) of either isolated core particles (42) or intact
chromatin (1,2,8,21,22). At higher concentrations nucleosomes
(containing 165-190 base pairs of DNA and histone Hl) produce dif-
fraction patterns very similar to those recorded in diffraction
patterns from intact chromatin. For instance a gel (centrifuge
pelletl with concentration about 35% w/w produces maxima at 124, 62
and 38~ (34), similar to those recorded from intact chromatin. When
the concentration is increased and the material maintained at 99%
relative humidity the nucleosomesoagain behave like intact chromatin
providing matima at 55, 35 and 27A but no maximum in the spacing
range 80-120 (figure 10). Dry samples of nucleosomes produce a
strong 75-82 maximum, again similar to intact chromatin. Thus if
a higher order structure such as a helix (or solenoid) does exist
in more concentrated gels of chromatin it does not depend upon the
continuity of the DNA between subunits for its existence.
Studies of crystals of core particles by Finch et~. (20) vali-
dated the model for the core particle derived from neutron and X-ray
scattering studies. In addition the crystal data can be used to
consider the origin of some of the maxima observed in the original
X-ray fibre diffraction studies. By combining this new information
with the older data describing the relative orientations of the
maxima in the original X-ray fibre diffraction patterns we may con-
sider the way in which the subunits are assembled in the chromatin
thread. Considering each reflection in turn the following consid-
erations can be made from an analysis of the various reflections
obtained from the crystals:
o
(a) 55A maximum. The neutron fibre studies suggested (25)
that this reflection arises from the transform of the DNA.
However, the crystal studies show that it arises from two
NEUTRONS AND X·RAYS IN THE STUDY OF CHROMATIN 211
a b
Figure 10: X-ray scattering from concentrated gels of nucleosomes
(containing 165-190 base pairs of DNA and histones Hl and H5) .
(a) Centrifuge pellet giving a concentration of ~35 % w/w. Rings
are present at 124, 62 and 38~ .
(b) Gel equilibrated to give a relative humidity of 99% (~60 % w/w).
Rings are present at 54, 35 and 27~ .
components related to the shape of the core particle.

These are the height of the particle and its diameter .
Since these two components are at right angles it is not
possible to derive any information relating to the orien-
tation of the subunit in chromatin based on the meridional
orientation observed in X-ray fibre diffraction studies (3).
o
(b) 37A reflection. This arises from spacings in the plane of
the disc-like structure of the core particle most probably
with components from both the DNA (20) and the histones
(34,42). The slight meridional orientation of this reflec-
tion (3) would suggest that the subunits assemble themselves
in the fibre with their axes aligned perpendicular to the
fibre axis. However, orientation in fibres is slight and
might arise from shearing the chromatin during the prep-
aration of the sample . This maximum is still present in
212 J. F. PARDON
,::110";'
(2x55";')
1
z76(2x38)";'
27";'
+
1
A B
Figure 11: Models for the arrangement of the core particles in the
chromatin unit thread.
A Core particles stackin~ in a top-to-bottom arrangement to give
a spacing of about 110A along the fibre axis.
B Core particles assembled edge-to-edg~ to give a contiguous
distribution of particles spaced 110A apart. Two chains are
included parallel to each other, separated by about 6oA. With
this arrangement an individual chromatin thread might form a
coil-bound helix or solenoid in which the particles in success-
ive gyres stack directly on top of each other. Such a helix
would account for the contrast variation behaviour of the 82~
maximum described in figure 9 since there are alternating gyres
rich in DNA and histone respectively.·
diffraction from highly stretched fibres of chromatin (4).

o
(c) The 27A reflection arising from the spacing between the
loops or gyres of DNA in the core particle is oriented in
a direction perpendicular to the plane of the particle.
Fibre diffraction ~atterns do include an equatorial com-
ponent at about 27~, however this might arise from
'stretched' chromatin molecules which pa~k together with a
regular side-by-side spacing of about 27~.
(d) The '110~' reflection. The two possible orlglns for this
reflection are illustrated in figure 11. These are:-
i. repeat distance between subunits in a linear assembly;
ii. the pitch of a regular helical structure such as the
solenoid proposed by Finch and K1ug (45) and Carpenter
et~. (46).
The neutron diffraction studies of Baldwin et~. (25)0

showed that this reflection moves progressively to~80A as
water is removed from the fibre. Also as described above,
this maximum arises from a structure which contains a regu-
lar repeat along the fibre axis arising from the spacial
separation of DNA and histone. Such a separation would
clearly be provided by model A in figure 11 where the core
particles tend to stack on top of each other in the chroma-
tin thread. However, it could equally well be explained
by model B (figure 11) in which the unit thread consists
of a string of subunits assembled edge-to-edge and where
the unit thread forms a helix in which particles in
adjacent gyres stack more or. less on top of each other.
The latter type of structure is supported both by the
electron microscopy of Finch and K1ug (45) and also by the
off meridional orientation of the 80A reflection observed
by Carpenter et a1. (46) from concentrated samples of
chromatin. ----
Further evidence supporting a solenoid type of structure
has been obtained by Campbell et a1. (43) who studied
mu1timer fragments containing Q1screte numbers of subunits.
For molecules having between 50 and 150 nuc1eosomes the
radius of gyration was found to vary linearly with the
number of subunits. The slopes of the distributions
decreased with increase in ionic strength showing that the
chromatin has a more compact structure in the presence of
salt. Whilst the results at low ionic strength could be
explain~d by a series of helices with pitch in the range
210-440A and outer diameter in the range 300-500~, the
214 J. F. PARDON
higher ionic strength data were supported only by a more

compact structure of the type proposed by Finch and Klug
(45).
FUTURE POSSIBILITIES OF USING DIFFRACTION/SCATTERING

STUDIES TO STUDY CHROMATIN
A combination of several techniques has clearly made possible
the determination of the structure of,the chromatin subunit (core
particle) to a resolution of about 20A. Details of the precise
structure of this particle will emerge during the next few years
as arrangement of the histones within the particle (H2A, H2B, H3
and H4) and relative to the DNA will then be known in detail.
Interest is already beginning to be placed towards the study of
subunits modified to allow transcription. This is an area where
X-ray and neutron scattering methods are likely to provide a sig-
nificant contribution.
The study of the structure of the higher order structures in
chromatin has been hampered by lack of detail in the X-ray and
neutron diffraction patterns which indicate the lack of regularity
in the structure. Whether this is inherent in the structure of
chromatin itself or a result of harsh methods used in isolating
chromatin from the nucleus is still not known. Solution scattering
from multimers of chromatin has produced a variety of different
results for the mass per unit length of the chromatin thread. An
extension of the type of experiments descPibed by Sperling and
Tardieu (21) including a wider range of concentrations and ionic
strengths might produce a better definition of the gross features
of the chromatin thread. Any such study should take advantage of
the very low angle resolution now possible with neutron scattering
cameras such as Dll at Grenoble. It is perhaps not encouraging
that the one study of metaphase chromosomes (47), which might have
been expected to contain more regularity in their structure,
provided diffraction patterns no better than those more recently
acquired from relatively short lengths of so-called 'native' chroma-
tin (22).
ACKNOWLEDGEMENTS
I would like to thank Dr. Brian Richards and Dr. David Lilley
for their comments on the manuscript, John Hobbs for preparing the
figures and Pat Campbell for carefully editing the final version of
the text.
NEUTRONS AND X·RAYS IN THE STUDY OF CHROMATIN 215
REFERENCES
1. Luzzati, V. and Nicolaieff, A. (1963) J. Mol. Biol. 7, 142
2. Bram, S. and Ris, H. (1971) J. Mol. Biol. 55, 325
3. Wilkins, M.H.F., Zubay, G. and Wilson, H.R. (1959) J. Mol. Biol.
1, 179
4. Pardon, J.F., Wilkins, M.H.F. and Richards, B.M. (1967) Nature
215, 508
5. Zubay, G. and Doty, P. (1959) J. Mol. Biol. 1, 1
6. Langridge, R., Wilson, H.R., Cooper, C.W., Wilkins, M.H.F. and
Hamilton, L.D. (1960) J. Mol. Biol. 2, 19
7. Pardon, J.F. and Wilkins, M.H.F. (1972) J. Mol. Biol. 68, 115
8. Pardon, J.F., Richards, B.M. and Cotter, R.I. (1974) Cold
Spring Harbor Symposia of Quantitative Biology 38, 75
9. Kornberg, R.D. (1974) Science 184, 868
10. Van Holde, K.E., Sahasrabuddhe, C.G. and Shaw, B.R. (1974)
Nucleic Acids Res. 1, 1579
11. Olins, A.L. and Olins, D.E. (1974) Science 183, 330
12. Woodcock, C.L.F., Safer, J.P. and Stanchfield, J. (1976) Exp.
Cell Res. 97, 101
13. Woodcock, C.L.F., Sweetman, H.E. and Frado, L.L. (1976) Exp.
Cell Res. 97, 111
14. Shaw, B.R., Corden, J.L., Sahasrabuddhe, C.G. and Van Holde,
K.E. (1974) Biochem. Biophys. Res. Comm. 61, 1193
15. Pardon, J.F., Worcester, D.L., Wooley, J.C., Tatchell, K.,
Van Holde, K.E. and Richards, B.M. (1975) Nucleic Acids Res. 2,
2163
16. Hjelm, R.P., Kneale, G.G., Suau, P., Baldwin, J.P., Bradbury,
E.M. and Ibel, K. (1977) Cell 10, 139
17. Haywood, B.C.G. and Worcester, D.L. (1973) J. Phys. E. 6, 568
18. Ibel, K. (1976) J. Appl. Cryst. 9, 296
19. Richards, B.M., Pardon, J.F., Lilley, D.M.J., Cotter, R.I.,
Wooley, J.C. and Worcester, D.L. (1977) Cell Biol. Int. Rep. 1,
107
20. Finch, J.T., Lutter, L.C., Rhodes, D., Brown, R.S., Rushton, B.,
Levitt, M. and Klug, A. (1977) Nature 269, 29
21. Sperling, L. and Tardieu, A. (1976) FEBS Lett. 64, 89
22. Sperling, L. and Klug, A. (1977) J. Mol. Biol. 112, 253
23. Pardon, J.F. and Richards, B.M. (1973) In Subunits in Biological
Systems Vol. 6, 1, Ed. Fasman and Timasheff, Marcel Dekker
24. Jacrot, B. (1976) Rep. Prog. Phys. 39, 911
25. Baldwin, J.P., Boseley, P.G. and Bradbury, E.M. (1975) Nature
253, 245
26. Engelman, D.M. and Moore, P.B. (1972) Proc. Natl. Acad. Sci.
USA 69, 1997
27. Hoppe, W. (1972) Israel J. Chern. 10, 321
28. Bradbury, E.M., Hjelm, R.P., Carpenter, B.G., Baldwin, J.P.,
Kneale, G.G. and Hancock, R. (1977) In The Molecular Biology of
the Mammalian Genetic Apparatus (Ed. P. Ts'o), p. 53, Elsevier/
North Holland
216 J. F. PARDON
29. Ibel, K. and Stuhrmann, H.B. (1975) J. Mol. Biol. 93, 255
30. Guinier, A. and Fournet, G. (1955) In Small-Angle Scattering
of X-rays. John Wiley, New York
31. Stuhrmann, H.B. and Kirste, R.G. (1965) J. Phys. Chern. 46, 247
32. Stuhrmann, H.B., Tardieu, A., Mateu, L., Sardet, C., Luzzati,
V., Aggerbeck, L. and Scanu, A.M. (1975) Proc. Natl. Acad. Sci.
USA 72, 2270
33. Suau, P., Kneale, G.G., Braddock, G.W., Baldwin, J.P. and
Bradbury, E.M. (1977) Nucleic Acids Res. 4, 3739
34. Pardon, J.F., Cotter, R.I., Lilley, D.M.J., Worcester, D.L.,
Campbell, A.M., Wooley, J.C. and Richards, B.M. (1978) Cold
Spring Harbor Symposium of Quantitative Biology, Vol. 42, In
Press
35. Porod, G. (1951) Kolloid Z. 2,83
36. Lilley, D.M.J., Richards, B.M., Pardon, J.F., Cotter, R.I.
and Worcester, D.L. (1978) Proc. FEBS 11th Meeting Copenhagen
1977, Vol. 43, Ed. Clark. Pergamon.
37. Olins, A.L., Breillatt, J.P., Carlson, R.D., Senior, M.B.,
Wright, E.B. and Olins, D.E. (1977) In The Molecular Biology of
the Mammalian Genetic Apparatus (Ed. P. Ts'o) p. 211, Elsevier/
North Holland
38. Varshavsky, A.J. and Bakayev, V.V.(1975) Mol. Biol. Rep. 2,
247
39. Langmore, J.P. and Wooley, J.C. (1975) Proc. Natl. Acad. Sci.
USA 72, 269 1
40. Pardon, J.F., Worcester, D.L., Wooley, J.C., Cotter, R.I.,
Lilley, D.M.J. and Richards, B.M. (1977) Nucleic Acids Res. 9,
3199
41. Debye, P. (1915) Zerstreuung von Rontgenstrahlen, Ann. Physik.
46, 809
42. Richards, B.M., Cotter, R.I., Lilley, D.M.J., Pardon, J.F.,
Wooley, J.C. and Worcester, D.L. (1976) In Current Chromosome
Research (Ed. Jones & Brandham) p. 7, Elsevier/North Holland
43. Campbell, A.M., Cotter, R.I. and Pardon, J.F. (1978) Nucleic
Acids Res., In Press
44. Pardon, J.F., Worcester, D.L., Richards, B.M. and Wooley, J.C.
(1976) Harwell Report No. MPD/NBS/29
45. Finch, J.T. and Klug, A. (1976) Proc. Natl. Acad. Sci. USA 73,
1897
46. Carpenter, B.G., Baldwin, J.P., Bradbury, E.M., Ibel, K. (1976)
Nucleic Acids Res. 3, 1739
47. Pardon, J.F., Richards, B.M., Skinner, L.G. and Ockey, C.H.
(1973) J. Mol. Biol. 76, 267
NUCLEAR MAGNETIC RESONANCE STUDIES OF NUCLEIC ACIDS AND PROTEINS
PaulO. P. Ts'o and Lou-Sing Kan

Division of Biophysics, School of Hygiene and Public
Health, The Johns Hopkins University, Baltimore, Md. 21205
With the advances in spectrometer instrumentation and computer

technology, NMR has become a very powerful technique in the study
of biopolymers, such as nucleic acids and proteins. This approach
can potentially provide information about the properties and inter-
actions of biopolymers at the atomic level through investigation of
the magnetic properties of many nuclei, particularly those having
spins of 1/2, such as lH, l3C, 19F, and 3lp. To most effectively
utilize this approach, however, we must define (1) the type of in-
formation which can be gained using NMR, (2) the type of questions
to be posed about the characteristics of prot~ins and nucleic acids
at the atomic level, and finally (3) the kinds of improvements
needed in NMR studies for the future. The biochemist is mostly
concerned not only with the nuclei which were mentioned previously,
but also with lIB, l5N, and l4N•
The possession of both spin and charge confers on a nucleus a

magnetic moment PN which is proportional to the magnitude of the
spin,
(1)
where YN is the magnetogyric ratio of the given nucleus and is

measured in radian' sec-1'gauss- 1 • Quantum theory demands that the
allowable nuclear spin states are quantized; the component mI, the
nuclear spin quantum number, in any given direction can take up
only one set of discrete values which are +1, (I-I), ••• , -I. For
the nuclei to have a nuclear spin =1/2, mI may only take the values
1/2 and -1/2; for the nuclei to have a nuclear spin =1,m1 may take
three values, i.e., 1,0, -1. If a steady magnetic field ~ is
applied on the nuclei, there is an interaction between the field and
217
218 P. O. P. TS'O AND L.·S. KAN
+
the magnetic moment ~N' which may be represented in terms of a
Hamiltonian
(2)
The energy corresponding to each splitting level when a magnetic

field is applied will be
(3)
The selection rule for transitions among the energy levels is that
mI changes by ±l; therefore
(4)
In order to induce transition between the two nuclear spin levels,

an oscillating electromagnetic field must be applied to the system
and the frequency V of the oscillating field must satisfy the
resonance condition hv=bE; therefore
YN H
v = 27T (5)
This result clearly implies the following:
(1) For each nucleus (Y N is a constant), the re~onance fre-

quency is directly proportional to the applied field H.
(2) For a given field, nuclei with a larger YN will resonate
at smaller magnetic fields.
GENERAL INTRODUCTION
Here, we shall first mention the principal difference existing

between the NMR spectroscopy and other types of spectroscopy, such
as UV absorption spectroscopy or infrared spectroscopy which is
familiar to most readers. For NMR, the different states of the
nucleus having varying energy levels are manifested by the applied
magnetic field, but are detected by the absorption of the radiation
at radio frequency. Therefore, the energy differences between
these states are directly related to the strength of the applied
magnetic field as revealed by changes in the frequency of radiation
absorbed by the nucleus at resonance. In other words, the resolution
of resonances, i.e. absorption of radiation at different frequencies,
is directly proportional to the strength of the applied magnetic
field. The second major difference between NMR and other conven-
tional spectroscopy is the relaxation process. Since the energy
differences between various states are relatively small, unless a
proper process exists to dissipate the absorbed energy, the system
will soon be "saturated" in reaching equilibrium, and hence will
show no signals. This problem will be returned to briefly in a
STUDIES OF NUCLEIC ACIDS AND PROTEINS 219
later section.
As a general introduction, the following four parameters can

be derived from the NMR studies:
(1) Chemical shifts (0, ppm)
When a molecule is placed in a magnetic field Ho ' orbital

currents are induced in the electron clouds. Therefore, each
nucleus is, in effect, partially shielded from Ho by the electrons,
and the local magnetic field strength will be
Hloc = Ho(l - a) (6)
where cr is the so-called screening constant (expressing the

change of field). Sigma is independent of Ho but highly dependent
upon the chemical structure of the molecule and can be either posi-
tive or negative. Therefore, the resonance frequencies have to be
21f\) = YN HI oc = YNHo(l - a) (7)
When an NMR experiment is carried out with a group or a mixture of

molecules at a given magnetic field, the signals from the various
nuclei are spread out in a spectrum according to their nuclear en-
vironments. Since v is proportional to the applied magnetic field,
the spacing between NMR signals corresponding to different types of
nuclei is also proportional to magnetic field. Thus, in NMR spectro-
scopy, in contrast to optical spectroscopy, there is no absolute
zero or standard reference. Hence, the chemical shift between two
sets of nuclei is defined as the difference in their resonance fre-
quencies measured at constant field. It is conveniently expressed
in a field independent unit as part per million (ppm) of the con-
stant field or frequency;
o= v2 - VI X 10 6 (8)
vI
where 0 is chemical shift in ppm, v2 is the measured frequency and
vI is the reference frequency. A standard reference substance for
proton and carbon-13 spectra, tetramethylsaline (TMS), has been
proposed and widely accepted.
In case of a complex molecule, such as nucleic acids or pro-

teins, in addition to the influence of the electron clouds through
covalent binding, the neighboring groups can also exert a magnetic
or electric field to the atom of interest. This through-space
field effect is reflected in the measurement of the.chemical shifts
which can now provide an indication of the spatial relationship of
these groups to the atom under measurement. This is a very powerful
tool in studying the conformation and interaction of biopolymers.
It should be cautioned, however, that the geometrical relationships
of these field effects are not simple and are often anisotropic, as
shown in later paragraphs.
(2) Coupling constants (J, H2)
In addition to the lines which had different chemical shifts,

the high resolution NMR spectra of many compounds contain patterns
which reveal the interactions of neighboring magnetic dipoles.
Magnetic nuclei may transmit influence to each other indirectly
through the intervening chemical bonds. This interaction occurs by
the slight polarizations of the spins and orbital motions of the
valence electrons, and the magnitude of the interaction is expressed
in terms of a coupling constant J which is not affected by the
tumbling of the molecules and is independent of Ho. Usually one
neighboring spin (1/2) would split the resonance of a single nucleus
into a doublet with intensity 1:1; and two equivalent neighboring
spins (1/2) would split the resonance of a single nucleus into a
triplet with intensity 1:2:1. In general, if a nucleus of spin 1/2
has n equivalently coupled neighbors of spin 1/2, its resonance will
be split into n+l peaks corresponding to the n+l spin states.
The coupling constant for two spins is usually large when sep-
arated by one bond (lJ), smaller when separated by two bonds (2J)
and, is even smaller when separated by three bonds (3J). However,
the vicinal coupling constants (3J) have been shown to be related
to the dihydral angle of a A-X-Y-B system, where the coupling is
between the spin A and spin B (3 JAB ) and the dihedral angle describes
the rotation around the X-Y bond. In the case of the couplings of
the H-C-C'-H' system in the furanose ring, the following relationship
was found by Lemieux.
3JHH' = J o cos 2 ~ - 0.28 Hz
where 3JHH' is the vicinal coupling constant, ~ is the dihedral

angle between H-C-C' and the C-C'-H planes in the fragment of
H-C-C'-H in the furanose. From the experimental results, J o = 9.27Hz
for ~ below 900 and J o = 10.36 Hz for ~ above 900 were calibrated.
Similarly, our laboratory has adopted the equation proposed by

Govil and Smith (1973) for the coupling constant relationship in the
conformational analysis of the 3'-C-OP bond and 5'-C-OP bond in the
nucleotides, dinucleoside monophosphates, and polynucleotides,
3JC,p ~ 9.5 cos 2e - 0.6 cos e

where e is the dihedral angle between the planes l3C CO and C0 3lp,
and the values of 3Jtrans = 10.1 Hz and 3J gauc he = 2.1 Hz are cal-
culated.
(3) The area of the resonance line or the amount of radiation

energy absorbed (relative units).
The areas of the resonance lines which represent the amount

of radiation absorbed, reflect proportionally the number of nuclei
contributing to these signals, subjected to the modification of the
relaxation process. Therefore, if the relaxation process is prop-
erly understood, such as in a small molecule, not involving chemical
exchange, spin-spin interaction, etc., then the areas of a resonance
line provide the vital information as to the number of nuclei in
that signal.
(4) The relaxation times (Tl, T2' in seconds)
In the previous sections, the behavior of an isolated, spinning

nucleus has been examined. When NMR is actually observed in bulk
matter, the observed signal represents a large number of identical
nuclei. These nuclei may interact among themselves and with their
surroundings. These interactions provide the means for the dissi-
pation of the absorbed energy. Consider an assembly of identical
nuclei experiencing the same magnetic field; such an assembly con-
stitutes a magnetically equivalent set. For a spin =1/2 nucleus,
there are only two magnetic energy levels which correspond to the
two alignments of the nuclear magnetic moment, i.e., either along
or against the magnetic field. At equilibrium the nuclei are dis-
tributed between the two energy levels and the ratio of the number
of spins in each level as given as follows
n+
-= e
-(~E/kT) (9)
n-
where n+ and n- are the populations of the upper and lower spin
states respectively, ~E is defined by Eq. (4), k is the Boltzman
constant, and T is the absolute temperature. For a given ~E, the
number of lower state spins will always be larger than that of upper
state spins. If the system is irradiated at a frequency V=~E/h, the
system absorbs energy from the radiation field with a consequent
increase in the n+/n- ratio. When the system absorbs sufficient
energy to equilize the population of the two states, it is said to
be "saturated". A saturated or partially saturated spin system it-
self will tend to return to a thermal equilibrium when the radiation
field is lifted. Two simultaneous processes are involved in the
return of a saturated system to equilibrium: (1) The absorbed energy
is given up from the spin system to the lattice; this process is
called spin-lattice relaxation. A time period is required to accom-
plish this relaxation and is denoted by Tl, the spin-lattice relaxa-
tion time; (2) Redistribution of the absorbed energy among the nuclei
by mutual exchange of nuclei between the higher and lower states;
this process is called spin-spin relaxation. T2 represents the spin-
spin relaxation time. These two processes, having different mecha-
nisms, do not necessarily occur at the same rate.
In general, the relaxation processes in biopolymers are related

to their freedom of motion and rotation as a means for dissipation
of energy and to their interaction of neighboring spins. As an
example for the first case, large and rigid DNA molecules do not
show narrow resonance lines because of the lack of motion, and as
an example for the latter case, 13C resonances of 13C bonded to
proton (i.e. 13C-H) are effectively relaxed by the dipole-dipole
interaction.
Application
With the instrumentation currently available, under optimal

conditons the individual nuclei in lH NMR can be detected at
10-4M (or 30 ~mole in 0.3 ml), which is about the same for 19F NMR,
while 3lp NMR can be detected at 10-3M and l3C NMR at 10-2M (at
natural abundance, l3C-enriched compounds can be detected with
proportionally greater sensitivity). The sensitivity of detection
is also dependent on the various relaxation processes of the sample.
Owing to the intrinsic capabilities and limitations of this

technique, the NMR investigation of the biopolymers can be formu-
lated from two different vantage points: (1) for static informa-
tion, and (2) for dynamic information.
In the first type of inquiry, the investigator wishes to

obtain the description of a three-dimensional structure at the
atomic level. Such a study is not unlike a study using the x-ray
diffraction technique, except that it is carried out in solution.
The information is provided mostly from the chemical shifts and
the coupling constants. The molecule under investigation is often
assumed to have only one predominant conformation; occasionally a
dynamic distribution between two conformational states can be
ascertained with some confidence. When the conformation of the
molecule is distributed dynamically into three or more major states,
then the population of the conformational states of this molecule
cannot be readily described. While such a study mayor may not
produce a precise description of the conformation, it has been
most useful in detecting changes in conformation due to interaction
or perturbation. Considerable success in this approach was
achieved in the conformational studies on small molecules (mol.
wt. less than 5000), including cyclic oligopeptides and nucleic
acid short helices. Often a comparison is made between the con-
formation in the crystalline state determined by x-ray diffraction
and that in solution determined by NMR. The most important quan-
titative equations in this study are the calculation on ring-
current magnetic anisotropy (particularly for the bases in nucleic
acids) made by B. Pullman, C. Giessner-Prettre and coworkers con-
cerning the through-space effect on chemical shifts of lH, and the
relationship between the dihedral angle and the three-bond coupling
3 1
constant, J, particularly between H and other nuclei, as proposed
by M. Karplus. Examples for the application of these theoretical
considerations to the experimental data will be presented.
Currently, the effects of the neighboring magnetic environment and
the conformation on the chemical shifts of l3C, 3lp , and 19F are
not understood in quantitative terms as well as the interpretation
of the coupling constants of these nuclei. Both theoretical and
experimental work are urgently needed in this area. In order to
facilitate this investigation, a computer graphics program has
been constructed in our laboratory so that the three-dimensional
coordinates of every atom of nucleic acid of any sequence and
conformation can be displayed numerically and graphically. The
main thrust of the NMR-computer technology is to extend the inves-
tigation on nucleic acids and possibly protein-nucleic acid inter-
action to the atomic level. The present challenge is to focus on
tRNA and interaction of tRNA with its cognate synthetase.
1. A Study of Short Nucleic Acid Helices in Solution. In

1975, we reported the comprehensive study of the proton magnetic
resonance on the non-exchangeable protons and on the hydrogen-
bonded NH-N protons of the ribosyl ApApGpCpUpU helix in solution
(Borer et al., 1975, Kan et al., 1975). All the resonances of the
base proto~ ribose-HI, and the hydrogen-bonded NH-N protons have
been assigned with a great deal of certainty. The effects of helix-
coil transition on these twenty resonances have been carefully
recorded and analyzed in terms of changes in chemical shifts, coup-
ling constants, and line width. The effect of salt concentration
and oligonucleotide concentration on the helical coil transition
process has also been investigated.
Confronting the challenging task of making spectral assignments

for the 17 C-H resonances and the three hydrogen-bonded NH-N
resonances, we have adopted three major procedures, in addition to
the conventional approach of relying on chemical shifts, coupling
constants, and the spin lattice relaxation time, Tl.
The first procedure compares the spectra of a sequence-related

series of oligonucleotides in which each member of the series is
incremented one nucleotide unit from its predecessor. Hence, this
aspect is termed "incremental assignment". The shortest member of
this series must be previously assigned by other standard methods
and the longest member is obviously the molecule of interest. In
the present case, ApA, A G, A2GC, A2GCU, and A2GCU 2 comprise the
series. The spectra of ~he series of oligomers are usually record-
ed in low salt, at high temperature, and, if possible, at low
strand concentration. In this environment the inter- and intra-
strand interactions are greatly reduced as indicated by the
similarity of the oligomer and monomer spectra, and also the
resonances are narrow and usually better resolved. Each of the
oligomers in the series exists as a substantially unstacked single
224 P. O. P. TS'O AND L.-S. KAN
strand; therefore, any major change in the resonance pattern from

an oligomer to the one incrementally longer is due to the reson-
ances of the new nucleotide and its shielding effects on the
previously present protons. The magnetic field effect of the new-
ly added nucleotide acts principally on its immediate 5' neighbor.
The procedure is illustrated in Figure 1.
The second assignment procedure compared the effect of temper-

ature variations as demonstrated in Figure 2. The entire spectrum
of the hexamer and those of many oligomers in the series have been
recorded over a 0-900 C range. The interval between temperature
points was 2-40C in regions where 0 changes rapidly with tempera-
ture or several protons resonate very close to each other. The
interval was ~lOoC in temperature regimes of little overlap or
change in o. From such data the spectral assignment established
at one temperature can be transferred to that at another temperature.
The third procedure involves the assignment of NH-N resonances

which depends on the line-width measurements. The line widths of
the NH-N resonances of the helical duplex are related to the rates
of the following exchange reactions:
k hc
... NH Col..1 (1)
NHhelix
" kch
k ,
cw
NH (2)
coil
" kwc
Based on various analyses of the known rate of exchange and of

helix-coil transition, we reached the conclusion that the line-
widths of the NH resonances in the current experiment are primarily
determined by the lifetime of the helix.
The above conclusions readily lead to the assignment of the

NH resonances based on the linewidth data. The NH resonance which
has the largest linewidth and the highest sensitivity to thermal
effects on the line-broadening and line-shifting is assigned to
the NH-N of the two terminal A(1)·U(6) pairs; the NH resonance
which has the smallest linewidth and the lowest sensitivity to
thermal effects is assigned to the NH-N of the two middle G(3)·C(4)
pairs; and the NH resonance which has an intermediate linewidth
and sensitivity is therefore assigned to the two interior A(2).U(5)
pairs.
In analyzing the chemical shift data, a helical duplex of

Kendrew's molecular model of A2GCU 2 was constructed from the
coordinates of A'-RNA and from the coordinates of B-DNA. Based on
CJ)
-t
C
o
(0) m
( b) (c) CJ)
ACZ~I A«tlt!, A(~~ ~«t)H2 A(~)H,. A(.')H,.

o
'T1
Z
!.,l ····l ···I.i. ( ' C
(')
to, AO ~,)M,. I(.~)H,. r
m
AJp~ \j ;)\li I " iI n
'.
.. : \
' 1 i... »
(')
\ i \~ .. : \
j \11 I"" , oCJ)
,!plp~p~ II il ! ! III 1 »
::\ ........ z
I.' \i \) o
,!p!p~~p~ II i i i II 1/ 'j I "Il ::0
o"
-t
.Ii "':f'lt":'::fi::-i::' ·" "1, m
I l • 8 • jIll Z
~
ApApGp pUpU .... ..>. .: ....
II Ii : .,::.::::.: ..... ....,..::::1.. . 11::::L::::.......... CJ)
ACZIM, oI('lHa AC2IHz ""1Hz GlSIHa UlS)Hs Ul!ilH s '~(4)H6 A(IIM;; Al'JH. U{e)"a U(SIHa" C(4)HS · ·.(S).... U(.~ ·~i; . .,Ail~:"~~~~,::~~~;;:~; ~;; ~ ~ ~~~~.;;. :~(5iH; 'c (4)M,
c 4""~
~~
1 .4 11:2 I.D T.I 7.' u 1.0 u 11.1 U -- O 11.4 u
CHEMICAL SHIFT, p.p.m.

Fig. 1. The incremental assignment scheme for all the base and HI' resonances of A2GCUZ' The
thickness of the lines represents the number of signals contained in a resonance envelope. (a) The
assignment of H2. H6. and H8 resonances at ~650C. (b) The assignment of HI. and H5 resonances at
~650c. Four resonances are clustered at this temperature. (c) The assignment of HI. and H5
resonances at ~350C. The four resonances clustered at ~650 are resolved at this temperature and can
now be assigned. [Na+)=O.02 M and Cs=l roM for each of the spectra except for the hexamer spectrum
shown in the 5-6 ppm region where Cs=lO roM and [Na+)=O.07 M. Chemical shifts are expressed in
reference to DSS. (From Ref. 4).
~
~
O!
226 P. O. P. TS'O AND L. ·S. KAN
O(3IMe
!l
I 2 3 ~ 5 •
ApApGpCpUpU
0 All)
°
0
A(21
0(3)
•
..
C(41
E 0 U(!!I
6 U(61
Q.
Q.
l-
LL.
:I:
Vl
...J
<l: 6.0
U "'121M,.
:::c
UJ .::~
:I: - U(~
U
G(3IM,.
TEMPERATURE °c
o 10 20 30 40 50 60 70 80 90
Fig. 2. The plot of the chemical shifts of base and HI' protons
of A2GCU2 in D20 (10 mM in strand concentration, 0.01 M sodium
phosphate buffer, pD = 7.0, 0.07 M Na+) versus temperature. All
chemical shifts are expressed in reference to DSS. The solid
symbols represent the data from the 100 MHz spectrometer and the
open symbols represent the data from the 220 MHz spectrometer.
(From Ref. 4).
these models, projections of neighboring protons onto the plane of

each base were made from the Kendrew mode~ such as that for the
A2GCU2 helix in A'-RNA geometry in Fig. 3. These projections now
permit determination of the distance of a proton from the base ring.
Additional and valuable information can also be obtained from

the coupling constants data on the HI' resonances during the helix
coil transition and the temperature variation profile. The low
Jl,_Z' values of the HI' resonances in the helical conformation
indicate that the furanose of the helical duplex is most likely in
the C3'-endo conformation. It is known from the x-ray diffraction
studies that the C3'-endo conformation belongs to the A and A'-RNA,
while the CZ'-endo conformation belongs to the B form of nucleic
acids, such as B-DNA. Therefore, the data based on Jl,_Z' clearly
favors the notion that the AZGCU Z duplex in solution assumes an A
or A' conformation.
A comparison between the predicted chemical shifts of these

twenty proton resonances and the observed resonances were made.
The predicted chemical shifts of formation for the AZGCU 2 sequence
in both the A'-RNA and B-DNA geometries are reported in Table IA.
The computation used ring current isoshielding contours which were
mapped in planes parallel to the bases at the distances shown in
Figure 3. In a similar manner, the measured chemical shifts of the
three NH-N resonances are compared with the computed values based
on A'-RNA and B-DNA model, as shown in Table lB.
In summary, conformational details of the AZGCU Z helix in

solution can now be ascertained by NMR on the basis of the follow-
ing self-consistent information: (i) The Jl,_Z' values of all
the residues in the helix indicate that each furanose is in a 3'-
endo conformation. Model building based on the x-ray diffraction
data of nucleic acid fibers reveals that the furanose conformation
in the A'-RNA (or A-RNA) is 3'-endo, while the furanose in B-DNA
is 3'-exo or 2'-endo. (ii) The chemical shifts of 17 C-H reson-
ances ~the AZGCU Z helix in solution agree with the computed
values based on the geometry of the A'-RNA much better than those
based on the geometry of B-DNA. (iii) The chemical shifts of
three sets of NH-N resonances representing six base pairs of the
A2GCU 2 helix also agree· with the computed values based on the
geometry of A'-RNA significantly better than those based on the
geometry of B-DNA. Thus, these data strongly support the conclusion
that the A2GCUZ helix in solution must assume a conformation closer
to that of the A'-RNA than to that of B-DNA. It would be of great
interest to synthesize a short DNA helix containing the A2GCU 2
sequence and to study the conformation of this short DNA nelix in
solution by PMR following the approach outlined above.
The NMR data on the helix-coil transition of this short helix
(AZGCUZ)2 contains both thermodynamic as well as structural infor-
~
~
co
Table lA. The Chemical Shifts of Formation of 17 Non-Exchangeable Protons in (1) The A GCU 2
Duplex as Compared with Calculated Ring Current Shifts from Two X-Ray Crysta!lographic
Models and (2) The A2GCU 2 Coils at 90 0 C (8 in ppm from DSS)
DUPLEX FORM COIL FORM
8a M ob s b M ca1cc M -M 8d Me
obs obs calc
A' B A' B
G(3)H 8 7.20 0.93 0.85 0.33 0.08 0.60 7.91 0.13
C(4)H 6 7.56 0.48 0.42 0.21 0.06 0.27 7.77 0.15
C(4)H 5 5.14 0.97 1.07 0.66 -0.10 0.31 5.89 0.19
A(2)H 2 7.50 0.76 0.64 0.44 0.12 0.32 8.16 0.13

A(2)H 8 7.84 0.67 0.79 0.25 -0.12 0.45 8.28 0.27
U(5)H 6 7.95 0.07 0.15 0.18 -0.08 -0.11 7.79 0.11
U(5)H 5 5.43 0.51 0.46 0.31 0.05 0.20 5.87 0.06
f f f f
A(1)H 2 6.89 1.36 1.05 f 1.10 f 0.31 f 0.26 f 8.16 0.12
A(1)H 8 8.16 0.22 0.00 0.02 0.22 0.20 8.20 0.16
U(6)H 6 7.95 0.07 0.03 0.03 0.04 0.04 7.81 0.09
U(6)H 5 5.69 0.25 0.20 0.09 0.05 0.16 5.90 0.03
:-c
G(3)H 1 ' 5.59 0.35 0.08 0.17 0.27 0.18 5.77 0.13 0
C(4)H1 , 5.48 0.50 0.04 0.07 0.46 0.43 5.92 0.04 :-c
--I
en
A(2)H 1 , 5.45 0.69 0.10 0.27 0.59 0.42 5.98 0.15 d
U(5)H 1 , 5.60 0.38 0.02 0.05 0.36 0.33 5.92 0.08 »
z
0
A(1)H 1 , 5.86 0.27 0.00 0.22 0.27 0.05 5.92 0.19 !
U(6)H 1 , 5.86 0.12 0.00 0.00 0.12 0.12 5.92 0.08 0
A
»
z
Table lAo Footnotes ~
c
o
m
CJ)
a These are the low temperature plateau values of 0 at 10 roM strand concentration, pD 7.0, 1.07 o
M Na+. If the 0.17 or 0.07 M Na+ 0 vS. T(OC) profiles level off at low temperature, their "z
plateau values were averaged with the 1.07 M Na+ number to generate the 0duplex value. c(")
r
b m
~OduPlex=OduPlex-OO;~~onucleotide is the chemical shift of formation for a proton in the duplex. o
Appropriate mononucleotide values were selected from the following list (0 in ppm from DSS): »(")
pG-H8, Hl" 8.126, 5.936, respectively; pC-H6, H5, Hl', 8.038, 6.110, 5.979; pA-H2, H8,Hl', 8.257, oCJ)
8.512, 6.141; pU-H6, H5, Hl" 8.015, 5.935, 5.982; and Ap-H2, H8, Hl" 8.252, 8.381, 6.132,
measured in 0.01 M sodium cacodylate buffer, pD 5.9±0.1 in D20, 1 mM in the appropriate 5'- »z
mononucleotide (a 2.0 roM sample of 3'-AMP was also measured). Little or no association of the o
."
mononucleotides is expected at these low concentrations. Temperatures were 0-5 0 • :xl
o
-t
c m
Based on the A'-RNA and B-DNA geometries (Arnott et al., 1972; Arnott and Hukins, 1972) and ring Z
current isoshielding contours provided by B. Pullman~private communication, see discussion). CJ)
d
These are averages of the 90 0 C chemical shifts of protons in 0.07, 0.17, and 1.07 M Na+ at
10 roM strand concentration, pD 7.0.
e
~Ocoil=Ocoil-o~g~onucleotide is the chemical shift of formation for a proton in the high tempera-
ture coil form. These mononucleotide values were used (see note b for buffer and nucleotide
concentrations): pG-H8, Hl', 8.041, 5.890; pC-H6, H5, Hl', 7.921, 6.084, 5.960; pA-H2, H8, Hl',
8.287, 8.547, 6.130; pU-H6,H5, Hl', 7.905, 5.930, 6.001; and Ap-H2' H8, Hl', 8.282, 8.359, 6.109.
f
These calculated values are subject to a correction of ~0.15 ppm due to shielding in end-to-end
aggregates.
t-.)
~
Table lB. The Chemical Shifts (in ppm) and the Linewidth at Half-
Height (in Hz) of the Hydrogen Bonded NH-N Resonances
of Guanine and Uracil in {A2GCU2)2 at lOC. a
0
calcd
b 0
calcd c
oobsd 0
- calcd linewidth
observed (AI-RNA) (B-DNA) AI-RNA B-DNA (uh)
G(3)NH 13.S d 13.3 12.8 0.2 0.7 30
U{S)NH 14.2 14.1 14.1 0.1 0.0 44
U(6)NH 13.2 13.S 14.3 -0.3 -1.1 80
a
The intrinsic values of the chemical shifts of the NH resonance
in the A.U pair and in the G·C pair is taken to be 14.7 and
13.6 ppm respectively (Kearns and Shulman, 1973), with the A·U
pair in the B-DNA geometry given the same value, 14.6 ppm, as
derived for the A·T pair (Patel and Tonelli, 1974).
b
Calculated values based on the geometry of the A'-RNA helix
(Borer, et al., 1975).
c Calculated values based on the geometry of the B-DNA helix

(Borer, et al., 1975).
d
All negative signs are omitted.
en
-t
C
C
A(I ) m
C(4)
en
o
'T1
Z
G131H 8.2 7 A . C
A{Z)H 2 ,28A (")
X
CI4)H ·,40A
·0 •
1 r
UI5JHr' ,2
o
6A
U'51H"'~A ·
0
N
• G(3)~.36A
o·~·o. o
GBlHa,68A
• 0 0
A(2)Hg.33A "
m
n
U(5)H6,3 Iii "
o A(2lHr, 29 A »(")
A(2)
C
U(5) A(i)He,2 7 A en
C(4)~,30A
.. ·. »z
((4)H&29A . c
.."
A(IIH(,40A
U.'H,,'O¢'
·0 •
· • :D
o
X GI3)H1',35A -t
U(6)H 1',3 5 A N • C14lK( ,40.6. Ul6»;'.40A
m
o U(6)H~4A III X • Z
o GC3IIt.2 9A en
U(6) G(3)
U(5)~.2 SA
.. •
AIZlHe,2 SA
UIS)H 6 ,2 4 A
A!2lHr,38A
·. •
lJI51t;'.39A
• U(51H 1,,37.6. X
o ((4)H(.30";'
""", *
Fig. 3. The projections of neighboring protons onto the plane of each base of the A2GCU2 helix
(A'-RNA geometry). Projections and distances were taken directly from the Kendrew model. Open
circles (0) indicate the protons positioned below the base plane; filled circles (I) indicate the
protons positioned above the plane; broken circles (~) indicate protons from a distant neighbor
about 6.8 A away; and crosses (X) indicate HI' atoms in the other strand. The views are normal to
the base planes with the free 5'-OH terminus above and the 3'-terminus below the base planes.
Vertical distances of the protons to the base planes are indicated in parentheses. (From Ref. 4).
t-.)
~
mation. These helix-coil transition profiles are reports of 20

atoms at separate locations within the helix (including the three
NH-N profiles) about their magnetic properties during the thermal
transition; therefore, in principle, they can provide useful data
to assign statistical weights to the various partially bonded
states. However, each transition curve is a reflection of the
changes in the local magnetic environment of each proton rather
than a direct report on the helix-coil populations. In the
partially formed duplex the 0 value of a proton is determined by
its 0 value in each microstate, weighted according to the popula-
tion of the state. Moreover, the value in a particular microstate
is determined by shielding influences which are anisotropically
distributed through space.
Thus, it is a very challenging task to disentangle the various

factors in order to obtain meaningful information without more
accurate knowledge of the application of NMR theory to nucleic acid
research. However, two general conclusions can be reached at
present. First, the NMR data follow the expected patterns de-
rived from the optical studies carried out at lower concentrations
concerning the effect of concentration and ionic strength on the
helix-coil transition of this short helix (Borer et al., 1975).
The "average T " increases with increasing concentration, showing
a linear relat!onship in a liT vs. -log conc. (strand) plot,
and also with an increase in ignic strength. Second, the melting
of this short helix clearly does not reflect an all-or-nothing
pattern. Both C-H resonances and NH-N resonances reveal that,
with respect to temperature change, the G3 'C 4 pair in the center
is more stable than the A2 'U 5 pair, which in turn is more stable
than the Al ·U 6 pair at the end of the helix. These results are
equivalent to the fraying of the ends of this short helix.
It should be noted that valuable studies on short helices have

also been made by others (Patel and Tonelli, 1974, 1975; Arter
et al., 1974, Patel, 1976, and Early ~ al., 1977).
2. Theoretical computation of proton chemical shifts. It is

self-evident that a quantitative application of the NMR theory is
urgently needed. We have taken two approaches in our research.
In the first approach, the spatial dependence of the ring-current
magnetic anisotrophy of the nucleic acid bases has been calculated
over a cylindrical domain of 10 i radius, which extends 8 i above
and below each ring of these bases (Giessner-Prettre, et al., 1976).
As shown in Fig. 4, separating the bicyclic purine base-into two
individual cyclic rings allows this spatial dependence to be
presented in a series of graphs in cylindrical domains. An exam-
ple of this graphic approach for the adenine ring is shown in
Fig. 5. By this approach, the through-space ring-current effects
on the chemical shifts of any resonance for atoms of nucleic acids
with known and fixed conformatioIB can be calculated by this first
,
-"
.............
""-
..... "- \.
\
\. \
\ \
\\ \
\
\ \
l~~H81 I
I ,
....... I I
. . . .f.. I
J . . . . -,......
/ Ps
/
/
/
/
",-
..... /
---",-
Fig. 4. Concentric contours for each ring of adenine base

spaced at 1.0 R intervals. The large dot in the center of each
ring is the origin of the cylindrical coordinate system for that
ring; PS( ~ ) represents the radius for the five-membered ring
and P6( ~ ) represents the radius from the six-membered ring.
(From Ref. 9).
,..p.""
o
p.p.m.
1.5
-0·1
•• !.
-11-2
0
p =5.0 5 6 7
:5 8 z(1)
(a)
p.p.m.
2 :5 z(1)
0
p. p.m.
p=8.0
1·0
-O.OS
6 7
(b)
Fig. 5. Adenine, (a) ~06 vs. Izl, the vertical distance from the
base plane and P6, the radius from the center of the six-membered
ring, (b) ~05 vs. Izl and P5, the radius from the center of the
five-membered ring. (From Ref. 9).
approximation and by the independently determined atomic coordinates.

While this graphic approach does not replace the more precise
computer-computation approach, it does provide a simple method for
manipulation of geometric parameters in order to predict the
probable mode of base-base or even base-drug interactions.
In the second approach (Kan, Kast, Ts'o, and Ts'o, unpublished

data), we have just completed a computer program using Fortran-lO
(a DEC-lO computer) which will print out the precise Cartesian
coordinates and cylindrical coordinates of all atoms of nucleic
acid duplexes having three helical structures (A form, A' form,
and B form) of any specified sequences up to 30 base pairs in
length. This program can also calculate the distance between
atoms. Programs for other helical structures can be readily added.
Thus, the precise coordinates of all atoms of any nucleic acid
duplex up to 30 base pairs (such as that shown in Fig. 3) can be
ascertained in three or more conformations for use in NMR calcula-
tions. We hope now we will be able to use the computation program
established by Professor B. Pullman and Dr. C. Giessner-Prettre in
conjunction with this computer program for coordinates for the
calculation of chemical shifts of all of the atoms of known coor-
dinates. Various through-space, electromagnetic effects exerted
on the atoms, such as ring current anisotrophic effect, the hydrogen-
bonding effect, and the electrostatic effect of the charged group
(such as the phosphate), can now be considered individually and
collectively in various combinations (Giessner-Prettre et al., 1977)
in a quantitative fashion.
3. Quantitative NMR Studies on the Yeast Transfer RNAPhe •

We have attempted to apply the above systematic approach at least
in a trial fashion to the study of yeast tRNAphe in solution (Kan
et al., 1977). The two-dimensional structure of yeast tRNAphe in
the]proposed clover-leaf model is shown in Fig. 6.
Recently, the structure of yeast phenylalanine transfer

ribonucleic acid (tRNAphe) in crystalline state has been clearly
elucidated by x-ray diffraction studies. Furthermore, the three-
dimensional coordinates of all atoms (except hydrogen) in this
tRNA molecule have been reported by several laboratories (Quigley
et al., 1975; Ladner, et al., 1975b; Sussman and Kim, 1976; Stout
et al., 1976; See revi;; by Rich and RajBhandary, 1976). The
tRNAPhe structures determined from these two crystal forms are
very similar to each other. Therefore, a very crucial question
arises: Does the common conformation of tRNAphe determined in the
crystalline state also exist in aqueous solution? At present,
information obtained from nuclear magnetic resonance (nmr) studies
can provide a direct and defined answer to this question.
Two spectral regions in the lH NMR spectrum of tRNAphe can be
investigated for quantitative conformational information. The
236 p. O. P. TS'O AND L.·S. KAN
YEAST t RNA PHE A 3'

e 75
e
A
5' pG e
e G a.a. STEM
G e 70
G U
5A U
U A He LOOP
60
U A 65 e u
D LOOP U G A e A e mlA
G
D 1~ A 10, A mSe
~O G U G T 1jI e
D e u e rn 2G e 55
U
GG GAG A G ,."e rn7G
25 rn~G EXTRA
20 e GAG 45 LOOP
e G
A U
30 G rnse 40
A 1jI
ern A a.c. LOOP
U Y
Gm AA
35
Fig. 6. The cloverleaf structure of Baker's yeast tRNAphe.
first region, which will be discussed later, contains the very low
field NH-N hydrogen-bonded proton resonances in H20. The second
region is described in this section, in which the high field (1-4
ppm) resonances of the methyl/methylene protons from the minor
nucleosides in yeast tRNAphe in D20 are investigated. The emphasis
in this study is not on base pair regions which can be studied by
the NH-N hydrogen bonded proton resonances but rather on the loop
regions and tertiary structure of tRNA molecule (Kan, et al., 1974;
Kastrup and Schmidt, 1975; Reid and Robillard, 1975; Daniel and
Cohn, 1975 and 1976), especially the D, anticodon, and l¢C loops
on tRNAphe.
In addition to the intact tRNAphe molecule, four important

fragments have also been investigated. These fragments are not
only very useful in the assignment of the methyl/methylene reson-
ances of the intact tRNAphe, but also provide information on the
conformation of the whole molecule of tRNAphe.
First, by taking advantage of recent developments in the

distance dependence of the ring-current magnetic effect of the
bases described above (Giessner-Prettre, et al., 1976) and in the
refined atomic coordinates of yeast tRNApne in the crystalline
state (A. Rich, private communication), a quantitative comparison
between the calculated shielding effects (~o) and the observed
shielding effects was made. This study suggests that the conforma-
tion of yeast tRNAphe in aqueous solution is grossly, but not
totally, identical to that determined in the crystalline state,
especially in the ~C and D regions. The initial step for assign-
ment of these high field resonances from tRNAphe and its fragments
is to compare their spectra to those from monomers at high temper-
ature. All modified mononucleosides (-nucleotides) found in
tRNAphe molecules were investigated separately at high temperatures.
This knowledge from the mononucleosides (-nucleotides) together
with the sequence data provides the basis for the unamb~@uous
assignment of the methyl, methylene resonances in tRNAP fragments,
which were then used for the assignment of the spectra for the
whole tRNA molecule. The intact tRNAphe in the presence of Mg++
was investigated at 360 MHz frequency, as shown in Fig. 7.
The calculated ring-current shielding/deshielding effects of

the methyl or methylene groups from their neighboring bases in
yeast tRNAphe has been made (Kan et al., 1977). These calculations
of shielding or deshielding effects ;;re based on the structure of
yeast tRNAphe in orthorhombic crystal as described in atomic
coordinates (A. Rich, private communication) and the graphic
approach to the computation of the ring current effects (Giessner-
Prettre, et al., 1976). A comparison between the calculated ring-
current effect (in terms of ppm) is made to the observed shielding/
deshielding effects which are defined by the differences in ppm of
the chemical shift values from the intact tRNA versus those from
.'c t
,
• , • I • ••
y T
(a)8!5'C
Fig. 7. The 360 MHz nmr spectra of high field proton resonances,
region of Baker's yeast tRNAphe in a temperature range of 50-85 0 C.
These spectra were taken under the following conditions: acquisi-
tion time, 1 second, 512 transients for spectrum a, 1024 transients
for spectra b, c, d, and e. (From Ref. 16).
the mononucleotides (-sides)' determined at 40-S00C, i.e. the lower

plateau region with respect to the temperature perturbation.
The results in this comparison can be classified into four

categories. The first category contains 7 resonances, Cm32 , Gm34 ,
Y3 (all four methyl resonances) and mSC40 • The observed and tfie
calculated 60 values of these seven resonances are in agreement
with each other within 0.1 ppm. Since these four modified residues
are either in anticodon loops or in the anticodon stem, the results
suggest no difference in the conformations of anticodon stem and the
anticodon loop of this tRNA in aqueous solution versus that in
the crystalline state can be found by this approach. In addition,
this categpry also contains one unusual case which is m~G26' There
are two ring-current effects predicted for the two methyl groups
in m~G2 (0.82 and 0.13 ppm) based on a static str~cture of tRNA
in crys~al, but only one broad signal was seen on H nmr spectrum
at ~6SoC (Fig. 6) having an observed 60 value of 0.46 ppm. This
observed value of 60 is very close to the average of the two
calculated values. As expected, the C?-N(CH3)2 bond in m~G rotates
at a sufficiently fast rate so that only one broad peak is
observed. Therefore, it is possible that the average conformation
of the two methyl groups from m~G in aqueous solution may also be
no diffirent from the average positions from its crystalline form.
Since m2G26 is located on top of the anticodon stem, this reasoning
again reinforces the preceding conclusion, i.e. the conformation of
the anticodon region of yeast tRNAphe in aqueous solution is similar
to that in the crystalline state.
SThe second category contains m2GlO ' 016 l7(C S)' 016 17(C6)
and m C"9' The observed ~d of these resonanc~s in this cAtegory
are much higher (more than 0.1 ppm) than the calculated ~o, indi-
cating these protons are more shielded than the predicted values
based on crystal structure and ring-current effects. While
there exists some doubt about the assignment of 016 17(Cn) resonance
in the native tRNA spectrum, this conclusion is mo~t liRely correct,
since it is supported by the results on 016,17(CS),
7 1
The third category contains m G46 and m ASS' The observed
~o of these two resonances are much lower (0.lS-0.20 ppm) than the
calculated ~o, indicating these protons are less shielded than the
predicted values based on crystal structure and ring-current effects.
Finally, the fourth category contains a T residue. The

methyl resonance from T has only one predicted ring-current effect
value but two 60 values are observed (Kan et al., 1977).However, both
experimental values are not in good agreem~t-;{th the predicted
value (interestingly, one was too high and the other too low).
In summary, the comparison indicates that no differences

between observed and calculated ~o values from resonances in the
anticodon stem and loop can be found, but differences in the T~C
stem/loop and D stem/loop have been uncovered. The nature of the
results, with both agreement and disagreement involving values that
are too high or too low, suggests that the difference may indeed be
due to the difference in conformation. It should be noted that the
regions showing the differences in conformation are the regions of
the tertiary structure of the molecule which are more readily
influenced by packing and also are the areas which are less defined
by the x-ray diffraction data. The segment involving m~G26 residue,
which connects the D stem and the anticodon stem may have the same
conformation in aqueous soluti02 as in the crystalline state except
that the Cf2 )-N(CH 3 )Z bond in m2G is rotatable in aqueous solution
but fixed n the sol~d state.
A similar approach is bfiing used for the hydrogen-bonded

NH-N resonances of yeast tRNAP e (Kan and Ts'o, 1977). The differ-
ence in this case is that the resonances have not been assigned
with certainty.
Figure 8 shows the 360 MHz spectrum of the lH nmr resonances

of the hydrogen-bonded NH from yeast tRNAphe sample at 23 0 C. Under
this condition, the spectrum is essentially insensitive to tempera-
ture variation within +lOoC and can be considered as a reliable
representation of the hydrogen-bonded NH resonances of yeast tRNAphe
in native conformation. This spectrum has a good signal-to-noise
ratio, and contains 15 well-resolved peaks plus a shoulder (k') in
the region of 11 to 15 ppm from DSS. This observed spectrum closely
resembles the published and unpublished spectra of the same tRNA
obtained under slightly different conditions (Kearns et al., 1974;
Robillard ~ al., 1976). - -
Based on the experimental spectrum in Figure 8, the estima-

tion that the spectrum contains 25 NH's, a simulated spectrum was
constructed (Fig. 9b). This simulated spectrum is considered to
be the represegtative spectrum of the hydrogen-bonded NH resonances
of yeast tRNAP e in the native state as observed in Fig. 8. This
simulated spectrum is used subsequently in a comparison with the
computed spectrum (Fig. 9).
In the calculation of the chemical shifts of these hydrogen-

bonded proton resonances, two parameters are needed: First, the
shielding (or deshielding) effect on the resonance, and second, the
intrinsic chemical shift of the resonance of this proton in differ-
ent base pairs. For the NH-N resonance in base pairs of Watson-
Crick type, we adopted the values of Kearns and Shulman (1974),
i.e., 14.7, 13.4, and 13.6 ppm for AU, A~, and GC base pairs, res-
pectively. As for the tertiary hydrogen bonds, there are many
types, most of which are not Watson-Crick base pairs such as Gl~C48
(Table IIIB). Recently, Kallenbach et al. (1976) have examinea
the (U)N 3H-(A)N 7 hydrogen-bonded proton resonance of reverse
CJ)
-t
C
0
k m
CJ)
0
"T1
Z
C
("')
r
m
c=;
l>
("')
0
CJ)
l>
z
0
c "'C
e I\ \ . :0
0
h I m -t
m
Z
CJ)
CHEMICAL p.p.m.
Fig. 8. A 360 MHz spectrum of the yeast tRNAphe at the 11.5-14.5 ppm region from DSS showing the
hydrogen-bonded NH resonances at 23°C. The sample contained 25 mg/ml of tRNAphe, dissolved in
0.01 M MgC12. 0.15 M NaCl, 0.002 M EDTA and 0.01 M potassium phosphate buffer, pH 7.0. (From Ref. l3) •
~
~
10,51
(01 I
II I I II III I I II II I III
•
..1--_ I I
14 IS 12
.,
., .,
....... . -
.,
. • -, • ,-,
~
. .... .,-...... ....... 0----- ....
14 13 12
CHEMICAL SHIFT, p_p,m,
Fig. 9. (a) The computed spectrum of NH-N hydrogen-bonded proton

resonances in yeast tRNAphe based on calculated chemical shifts
(froD refer. 13). The number of these NH resonances represent
the fOilowing base pairs:
6: U6 1\7 , 12: U12 l\3' 5: As U6 B" 54: '15 .. mlAse, 7: U7Ass, 8: USA l .. ,
52: US2AS2, 50: UsoAs.. , 29: A29U .. l, 11: Cll G2.. , 53: GS3 C Sl ,
19: Gl 9CS S, 2: C2G7l, 10: m2GlOC2S' 31: A3l~39' 1: Gl C72 ,
13: Cl3G22, 51: GSICS3, 27: C27G .. 3, 30: G30 ms C.. O, 3: G3C70 ,
49: mSC"9G~, and 28: C2eG .. 2. Every computed peak has an equal
linewidth at half-height; 36 Hz in 360 MHz scale.
(b) A simulated spectrum for the observed spectrum shown
in Fig. 8. Every peak has a 36 Hz linewidth at half-height in
360 MHz scale. The symbols between (a) and (b) represent the
adjustments needed to transform the computed spectrum (a) to the
simulated spectrum (b); (-) represents removal of resonance peaks,
(+) represents addition of resonance peaks, and (~or +) represent
moving the chemical shifts to high field or low field, respectively.
Hoogsteen type in an U·A·U triple stranded helix by lH nmr and

recommended an intrinsic chemical shift of 14.1 ppm. From their
data, together with a careful evaluation of the geometry and
shLelding effect, we estimated a value of 14.3 ppm as the intrinsic
chemical shift of a reverse Hoogsteen (U)N3H-(A)N7 hydrogen-bonded
proton resonance in an A'-RNA conformation (Arnott, et al., 1973).
As for the resonances of hydrogen-bonded NH-O in G.U-,-G~, or G·C
base pairs, as well as that of NH-N in G·A base pairs, no intrinsic
values of their chemical shifts have been determined experimentally.
Therefore, no calculations of the resonances of these five base
pairs were made.
The interatomic/intermolecular magnetic field experienced by

the NH-N resonances in yeast tRNAphe was then calculated based on
the ring current effect, as evaluated from the coordinates derived
from the x-ray diffraction data (Kan and Ts'o, 1977). This result
was plotted by the PDP-10 computer as shown in Figure ~ The
adjustments needed to transform the computed spectrum to the
simulated experimental spectrum are shown between Figs. 9A and 9B.
This computed spectrum contains 23 resonances, 6 resonances less
than the total recommended by the three dimensional structure of
tRNA determined by x-ray diffraction.
Calculation of hydrogen-bonded NH resonances have been made by

others (Robillard, et al., 1976, Kearns, 1976, and Geerdes and Hil-
bers, 1977). A dis~ssion has been made in comparison with these
calculations (Kan and Ts'o, 1977). Special attention should be
given to the interesting results by Geerdes and Hilbers (1977).
Their work showed that the calculated NMR spectra are very sensitive
to slight changes in structure.
In conclusion, despite some uncertainties in_the theoretical

treatment, this quantitative comparison between the simulated
experimental spectrum and the calculated spectrum based on the
atomic coordinates of the tRNA in crystal and on ring-current
effects clearly indicates that the native conformation of yeast
tRNAphe in solution is fundamentally similar to that in the
crystalline state. The minor difference is probably in the tertiary
structure involving the folding of the T$C loop and stem to the D
loop and stem. This conclusion is reinforced by the lH nmr studies
on the methyl/methylene resonances of the minor bases reported in
the preceding section. In addition, some of the hydrogen-bonded
base pairs existing in the tRNA in the crystalline state may not
be detectable in solution.
4. Concluding Remarks. The above description indicates that

the lH-nmr research on nucleic acids has been highly informative
and sophisticated. In comparison, the research on l3C-NMR and 3lp~
NMR properties of nucleic acid is still in its infancy but has a
promising future (see General References for publications on this
subject). However, so far the investigation has been focused

mainly on the static conformational properties of nucleic acids,
which was treated as the first type of inquiry in this section.
In the second type of inquiry, the investigator wishes to

obtain information about the kinetic parameters of the relaxation
process. From this information, the specific and general inter-
action of these nuclei with the environment may be known, parti-
cularly about their freedom in motion. The studies of histone
interaction are mostly in this category. Specific information on
conformation can also be obtained if the relaxation process is
predominantly influenced by a paramagnetic center near the nuclei
of interest. The interatomic distance between the paramagnetic
source and the relaxed nuclei can be deduced from the relaxation
data. Also, soon the relaxation process will be more readily
investigated by spectrometers of different frequencies. The
frequency-dependent data provide a better estimate of the mechanism
of the relaxation process. The ability to study the relaxation
process of these nuclei is an unique feature of NMR. In order to
utilize the investigative power of NMR fully, both static and
dynamic data about the sample should be obtained and integrated
into a coherent conclusion.
Looking into the future, in order to derive more quantitative

information from the NMR data, we have an urgent need to develop
theory, followed by experimental verification. Also, NMR studies
should be extended to complex mixtures, including possibly living
cells. This can be achieved by the use of NMR label containing
19F or enriched 13C, nuclei normally not found in the biological
system. Under special conditions, the hydrogen-bonded NH---N
resonances and the 3lp resonances can also serve this purpose.
ACKNOWLEDGMENTS
This research was supported in part by grants from the

National Science Foundation and from tne National Institute of
General Sciences as well as by various fellowships granted to the
individual investigators over the past 10 years. The valuable
assistance of Cathryn Alden and Christine Dreon in the preparation
of the manuscript is gratefully acknowledged.
GENERAL REFERENCES ON NUCLEAR MAGNETIC RESONANCE AND ITS

APPLICATION TO RESEARCH ON NUCLEIC ACIDS AND CHROMATIN
1. J.D. Roberts, ed. Nuclear Magnetic Resonance: Applications

to Organic Chemistry, McGraw-Hill Book Co., 1959.
Z. J.A. Pople, W.G. Schneider, and H.J. Bernstein, eds., High-
Resolution Nuclear Magnetic Resonance, McGraw-Hill Book Co.,
1959.
3. F. Bovey, ed. Nuclear Magnetic Resonance Spectroscopy, Academic
Press, 1969.
4. M. Boublik, E.M. Bradbury, C. Crane-Robinson, and E.W. Johns.
"An Investigation of the Conformational Changes of Histone FZb
by High Resolution Nuclear Magnetic Resonance", Eur. J.
Biochem. 17, 151 (1970).
5. V.M. Clark, D.M.J. Lilley, O.W. Howarth, B.M. Richards and
J.F. Pardon, "The Structure and Properties of Histone FZa 13
Comprising the Heterologous Group FZal and FZaZ Studied by C
Nuclear Magnetic Resonance", Nucleic Acids Research, Vol. I,
No.7, 865 (1974).
6. P.o.P. Ts'o, ed., Basic Principles in Nucleic Acid Chemistry,
Vols. I and II, Academic Press, 1974.
7. L. Jackman and F.A. Cotton, eds. Dynamic Nuclear Magnetic
Resonance Spectroscopy, Academic Press, 1975.
8. T.L. James, ed. Nuclear Magnetic Resonance in Biochemistry:
Principles and Applications, Academic Press, 1975.
9. D. Fitzsimons and G. Wolstenholme, The Structure and Function
of Chromatin, Ciba Foundation Symposium Z8, London, on
April 3-5, 1974, Elsevier/North-Holland Biomedical Press, 1975.
10. E.M. Bradbury, P.D. Cary, C. Crane-Robinson, H.W.E. Rattle,
M. Boublik and P. Sautiere, Biochemistry 14, 1876 (1975).
11. P.N. Lewis, E.M. Bradbury, and C. Crane-Robinson, "Ionic
Strength Induced Structure in Histone H4 and its Fragments",
Biochemistry 14, 3391 (1975).
lZ. D.G. Gorenstein, "Dependence of 3lp Chemical Shifts on Oxygen-
Phosphorous-Oxygen Bond Angles in Phosphate Esters", JACS, 97,
4, 898 (1975). .--
13. D.G. Gorenstein and D. Kar, ,,3l p Chemical Shifts in Phosphate
Diester Monoanions. Bond Angle and Torsional Angle Effects",
BBRC 65, 3 (1975).
14. D.J. Patel, "Proton Nuclear Magnetic Resonance Studies of the
Helix-Coil Transition of d-ApTpGpCpApT in DZO Solution",
246 p. O. P. TS'O AND L.-S. KAN
15. D.J. Patel and A.E. Tonelli, "Nuclear Magnetic Resonance

Investigations of the Structure of the Self-Complementary
Duplex of d-ApTpGpCpApT in Aqueous Solution", Biochemistry
14, 3990 (1975).
16. M.A. Young and T. Krugh, "Proton Magnetic Resonance Studies
of Double Helical Oligonucleotides. The Effect of Base
Sequence on the Stability of Deoxydinucleotide Dimers",
17. C.W. Hilbers and D.J. Patel, "Proton Nuclear Magnetic Reson-
ance Investigations of the Nucleation and Propagation React-
ions Associated with the Helix-Coil Transition of d-ApTpGpCp-
ApT in H20 Solution", Biochemistry, 14, 2656 (1975).
18. D.M.J. Lilley, O.W. Howarth, V.M. Clark, J.F. Pardon, and B.M.
Richards, "An Investigation of the Conformational and Self-
Aggregational Processes of Histones Using lH and l3C Nuclear
Magnetic Resonance", Biochemistry, 14, 4590 (1975).
19. A.E. Pekary, S.l. Chan, C.-J. Hsu, and T.E. Wagner, "Nuclear
Magnetic Resonance Studies on the Solution Conformation of
Histone IV Fragments Obtained by Cyanogen Bromide Cleavage",
Biochemistry, 14, 1184 (1975).
20. T. Moss, P. Cary, C. Crane-Robinson, and E.M. Bradbury,
"Physical Studies on the H3/H4 Histone Tetramer", Biochemistry
15, 2261 (1976).
21. P.D. Cary, C. Crane-Robinson, E.M. Bradbury, K. Javaherian,
G.H. Goodwin and E.W. Johns, "Conformational Studies of Two
Non-Histone Chromosomal Proteins and Their Interactions with
DNA", Eur. J. Biochem. 62, 583 (1976).
22. D.G. Gorenstein, J.B. Findlay, R.K. Momii, B.A. Luxon, and
D. Kar, "Temperature Dependence of the 3lp Chemical Shifts of
Nucleic Acids. A Probe of Phosphate Ester Torsional Confor-
mations", Biochemistry 15, 17 (1976).
23. P.J. Cozzone and O. Jardetzky, "Phosphorus-3l Fourier Trans-
form Nuclear Magnetic Resonance Study of Mononucleotides and
Dinucleotides. I. Chemical Shifts", Biochemistry 15, 22
(1976).
24. P.J. Cozzone and O. Jardetzky, "Phosphorus-3l Fourier Trans-
form Nuclear Magnetic Resonance Study of Mononucleotides and
Dinucleotides. II. Coupling Constants, Biochem. 15,4860 (1976).
25. D.J. Patel, "Proton and Phosphorus NMR Studies of d-CpG(pCpG)
Duplexes in Solution. Helix-Coil Transition and Complex n
Formation with Actinomycin-D", Biopolymers 15, 3 (1976).
26. J.E. Coleman, R.A. Anderson, R.G. Ratcliffe and I.M. Armitage,
"Structure of Gene 5 Protein-Oligodeoxynucleotide Complexes as
Determined by lH, 19F, and 3lp Nuclear Magnetic Resonance",
Biochemistry IS, 25 (1976).

27. D.J. Patel, "Proton and Phosphorus NMR Studies of d-CpG(pCpG)
Duplexes in Solution. Helix-Coil Transition and Complex n
Formation with Actinomycin-D", Biopolymers IS, 533 (1976).
28. T.R. Krugh, J.W. Laing, and M.A. Young, "Hydrogen-Bonded
Complexes of the Ribodinucleoside Monophosphates in Aqueous
Solution. Proton Magnetic Resonance Studies, Biochemistry,
IS, 1224 (1976).
29. N.R. Kallenbach, W.E. Daniel, and M.A. Kaminker, "Nuclear
Magnetic Resonance Study of Hydrogen-Bonded Ring Protons in
Oligonucleotide Helices Involving Classical and Nonclassical
Base Pairs", Biochemistry 15, 1218 (1976).
30. D.J. Patel and L. Canuel, "Nuclear Magnetic Resonance Studies
of the Helix-Coil Transition of Poly (dA-dT) in Aqueous Solu-
tion", PNAS 73, 3 (1976).
31. The Molecular Biology of the Mammalian Genetic Apparatus,
Vol. I, ed. by P.O.P. Ts'o, Elsevier/North Holland Publishing
Co., 1977.
32. W. Egan, H. Shindo, and J.S. Cohen, "Carbon-13 Nuclear Mag-
netic Resonance Studies of Proteins", Ann. Rev. Biophys.
Bioeng. 6, 383 (1977).
33. D.G. Gorenstein, B.A. Luxon, and J.B. Findlay, "The Torsional
Potential for Phosphate Diesters. The Effect of Geometry
Optimization in CNDO and AB Initio Molecular Orbital Calcula-
tions, BBA 475, 184 (1977).
34. D.R. Kearns, "High-Resolution Nuclear Magnetic Resonance
Studies of Double Helical Po1ynucleotides", Ann. Rev. Biophys.
Bioeng. 6, 477 (1977).
35. D.M.J. Lilley, J.F. Pardon and B.M. Richards, "Structural
Investigations of Chromatin Core Protein by Nuclear Magnetic
Resonance", Biochemistry 16, 2853 (1977).
36. P.G. Hartman, G.E. Chapman, T. Moss, and E.M. Bradbury, Studies
on the Role and Mode of Operation of the Very-Lysine-Rich
Histone H1 in Eukaryote Chromatin. The Three Struc·tural
Regions of the Histone HI Molecule, Eur. J. Biochem. 77, 45
(1977) •
SPECIFIC REFERENCES FOR THE MATERIALS IN THIS CHAPTER
1. J.L. Alderfer and P.O.P. Ts'o, Biochemistry 16, 2410 (1977).

2. S. Arnott, D.W.L. Hukins, S.D. Dover, W. Fuller, and A.R.
Hodgson, J. Mol. Bio1. 81, 107 (1973).
3. D.B. Arter, G.C. Walker, O.C. Uhlenbeck, and P.G. Schmidt,
Biochem. Biophys. Res. Commun. 61, 1089 (1974).

4. P.N. Borer, L.S. Kan, and p.a.p. Ts'o, Biochemistry 14, 4847,
(1975).
5. W.E. Daniel, Jr., and M. Cohn, Proc. Natl. Acad. Sci. USA 72
2582 (1975).
6. W.E. Daniel, Jr., and M. Cohn, Biochemistry 15, 3917 (1976).
7. T.A. Early, D.R. Kearns, J.F. Burd, J.E. Larson, and R.D. Wells,
8. H.A.M. Geerdes, and C.W. Hilbers, Nucleic Acids Res. 4, 207
(1977) •
9. C. Giessner-Prettre, B. Pullman, P.N. Borer, L.S. Kan, and
p.a.p. Ts'o, Biopolymers 15, 2277 (1976).
10. C. Giessner-Prettre, B. Pullman, and J. Caillet, Nucleic Acid
Res. 4, 99 (1977).
11. G. Govil and I.C.P. Smith, Biopolymers 12, 2589 (1973).
12. N.R. Kallenbach, W.E. Daniel, Jr., and M.A. Kaminker, Biochem-
istry 15, 1218 (1976).
13. L.S. Kan and p.a.p. Ts'o, Nucleic Acids Res. 4, 1633 (1977).
14. L.S. Kan, p.a.p. Ts'o, F. von der Haar, M. Sprinzl, and F.
Cramer, Biochemistry 16, 3143 (1977).
15. L.S. Kan, P.N. Borer, and p.a.p. Ts'o, Biochemistry 14, 4864,
(1975) •
16. L.S. Kan, p.a.p. Tsro, M. Sprinzl, F. von der Haar, and F.
Cramer, Biochemistry 16, 3143 (1977).
17. R.V. Kastrup and P.G. Schmidt, Biochemistry 14, 3612 (1975).
18. D.R. Kearns. Prog. Nucleic Acids Res. Mol. BioI. 18. 91 (1976).
19. D.R. Kearns and R.G. Shulman, Acc. Chern. Res. 7, 33 (1974).
20. J.E. Ladner, A. Jack, J.D. Robertus, R.S. Brown, D. Rhodes.
B.F.C. Clark, and A. Klug. Nucleic Acids Res. 2, 1629 (1975).
21. D.J. Patel. Biopolymers 15. 533 (1976).
22. D.J. Patel and A. Tonelli. Biopolymers 13. 1943 (1974).
23. D.J. Patel and A. Tonelli. Biochemistry 14. 3990 (1975).
24. G.J. Quigley. N.C. Seeman. A.H.-J. Wang, F.L. Suddath, and
A. Rich. Nucleic Acids Res. 2. 2329 (1975).
25. B.R. Reid and G. Robillard. Nature (London) 257. 287 (1975).
26. A. Rich and U.L. RajBhandary, Annu. Rev. Biochem. 45. 805
(1976).
27. G.T. Robillard, C.E. Tarr, F. Vosman, and H.J.C. Berendsen,

Nature (London) 262, 363 (1976).
28. C.D. Stout, H. Mizuno, J. Rubin, T. Brennan, S.T. Rao, and
M. Sundara1ingam, Nucleic Acid Res. 3, 1111 (1976).
29. J.L. Sussman and S.-H. Kim, Biochem. Biophys. Res. Commun.
68, 89 (1976).
TECHNIQUES FOR CYTOCHEMICAL STUDIES OF THE NUCLEUS AND ITS
SUBSTRUCTURES
Torbjorn Caspersson
Karolinska Institutet, Inst.for Tumor Pathology

Karolinska Hospital
S 104 01 Stockholm 60, Sweden
Qualitative methods for determining the chemical compos~t~on

of cellular structures go quite far back in time. Even in the
early days the main interest was the cell nucleus. Overenthusias-
tic use of these staining techniques brought, however, the field
into disrepute especially in the 1920-ies and this came to be one
of the main reasons for developing quantitative optical microsco-
py for cytochemical research.
With passing years and the rapid development of modern biology

and medicine, the main fields of interest in experimental cell
biology have changed so that a considerable number of different
types of instruments are available to the cytochemist of today.
The first quantitative optical cytochemical work concerned

nuclear structures. The central problem at that time was to loca-
lize and determine the amounts of nucleic acids by ultramicro-
spectrography (UMSP). This photometric work was done in the ultra-
violet using UV-microscope objectives developed already at the
turn of the century by von Rohr and Kohler. In the very beginning,
photographic photometry was used, but this was soon replaced by
photoelectric spectrophotometry.
To your eyes, these early instruments may look primitive but

they were quite powerful in precision and resolution.
251
252 T. CASPERSSON
.15 1----+----:::b"F'<-----+ .=--*--11--\-+---+--1
z
o
i=
lJ
Z
i= .10 \--+-+-+-I--+---t----"R'<-.....:hl--\---t---j
x
w
OL---~--~~2~60~O~-L~2~8tO~O--~~~~~
WAVE LENGTH
Fig 1. Absorption spectra from different parts of a Drosophila

salivary gland nucleus: (1) chromocenter, (2) chromosome insertion
on chromocenter, (3 and 4) euchromatic bands, (5) nucleolus, (6)
interband space.
Figure 1 gives examples of ultraviolet absorption spectra

of small nuclear structures measured in 1940 by the first photo-
electric UMSP. The size of the measuring spot is 0.3 f. These
UMSPs also gave semiquantitative information on the amount of
protein in cellular structures. The data won in the early forties
indicated a clear relationship between protein synthesis and nuc-
leic acids.
Later, improved models of the original UMSP were built. Figure

2 shows an instrument built for spot measurements as well as inte-
grating measurements, the "Universal UMSP". Together with Carl
Zeiss Company of Oberkochen this model was further developed
(figure 3) and Zeiss began commercial production of that instru-
ment in the early sixties.
This instrument is still the most advanced instrument avail-

able for high resolution spectrophotometric analysis of nuclear
structures and other small elements inside the cell and because
of that, a short description of its function will be given. Basi-
cally it is a double beam instrument with identical pathways for
CYTOCHEMICAL STUDIES OF THE NUCLEUS 253
Fig 2. Universal-ultramicrospectrophotometer (UMSP) for spot

measurements and for integrating work.
the measuring beam and the comparison beam. It works in the wave-
length range from 220 to about 700 nanometers. The special "Ultra-
fluar" microscope objectives developed by Zeiss for the instrument
work well over this entire range. Recordings of absorption spectra
can be made of spots within the cells, with the smallest spot size
determined only by the resolution of the optics (up to numerical
aperture 1.25, even in UV). In addition automated scanning
measurements can be made across an object. During the scan the
running transmission values taken as the running integral of the
extinction values are recorded on the same paper. The distance be-
tween each of the subsequent scanning lines can be chosen arbitrar-
ily down to 0.5 ~.
For instance, if a whole free lying cell is scanned, the end-

point of the extinction-integral curve gives the total extinction
of the whole cell directly at the wavelength used one can correlate
different points in the object with the corresponding points on the
recorded curves and it is thus possible to determine the total ex-
tinction of any intracellular element, such as nucleoli or chro-
matin. The instrument is also equipped for photography and with an
254 T. CASPERSSON
". . ..
.. .......'
.
Fig 3. The UMSP built by C. Zeiss.
image converter which visual observation in the UV is quite easy.

This instrument has been used and is being used by many groups for
different kinds of studies of cells and their internal structures
during the course of growth and differentiation and during differ-
ent types of functions.
The instrument permits work with extreme optical resolution

and is suitable for work on very small structures. For example,
Figure 4 shows recording UV-measurements along rye metapahse chro-
mosomes. Rya has seven chromosomes of practically the same length
which are almost indistinguishable from each other in the micro-
scope. The recordings show fine patterns in the DNA-distribution,
the details of which lie very close to the theoretical limits for
the UV microscope resolution. The reproducibility of the measure-
ments is so good that all seven of the individual chromosome types
can be recognized through their individual absorption profiles.
In the late forties and early fifties several UMSP-model were

CHROMOSOME 5
Figure 4. High resolution recordings of DNA-distribution along

four rye chromosomes. The absolute length is 7-9 ~.
built in different laboratories, a few for UV but most for use in

the visible region. The main aim was DNA determination by Feulgen-
UMSP and this work contributed greatly to our present view of the
DNA constancy in the nucleus. Most of the instruments permitted
only "plug" - measurements and not high resolution integration
scanning work. This gives low accuracy in the determinations, even
if such "tricks" as the "two-wavelength procedure" are used. Simi-
lar instruments are now available from several of the large opti-
cal companies.
Several rather complex instruments have been built combining

microspectrophotometers with computers and are now commercially
available. Some of these incorporate scanning stages, permitting
steps down to 0.5 p range. One use for such instruments is to scan
slides for selected cell types in studies aiming at "automated cy-
tology", which, however, falls outside the frame of this presenta-
tion. A field of application which is of considerable interest for
chromatin analysis, is the determination of the substance distri-
bution within an area, for instance the nucleic acid component in
a cell nucleus.
256 T.CASPERSSON
Fig 5. A recording and integrating ultramicrointerferometer.
Out of our early cytochemical work on the role of the cell

nucleus in cell growth and on its interplay with the cytoplasm, a
desire ensued for better methods for protein and/or cellular dry
mass determinations. In 1948 Engstrom and Lindstrom developed a
method based on X-ray absorption, a technique which Leon Carlson
then carried to a resolution comparable to that of microscopy in
the visible range. The technique works very well, but was eventual-
ly replaced by the more convenient interferometric techniques. The
first dry mass determinations by interference microscopy were made
in 1950 by Davies and Maurice Wilkins with the Dyson interference
microscope. Ever better interference optics became available
during the following years.
In our group we have built several models of microinterferome-

ters for different purposes. Figure 5 shows a scanning-recording
instrument using the Jamin,Lebedeff type of optics. Its main use
is to measure the distribution of cell dry mass in different com-
partments of the cell, but it can also perform integrating measure-
ments, even of intracellular details. Interferometry has not
proved to be very useful for analysis of small intranuclear details
because of certain limitations in resolution when these optics are
applied to conventional biological specimens. On the other hand, it
has proved to be an important tool in studies of the nucleus-cyto-
plasm interaction.
It should be mentioned here that the limitations of interfero-

metry at high resolution led to the development in 1959 of a method
for dry mass determination in small cellular elements by measure-
ment of electron scattering in the electron microscope (Bahr, Bloom,
Carlson, Zeitler). This method has later on by Bahr and Zeitler
been proved to be a very valuable tool for the analysis of nuclear
details in general, and especially of chromosomes.
Microfluorometry, a logical outgrowth, developed rather slow-

ly despite efforts in many laboratories. This was due mainly to
lack of suitable optics and adequate light sources. Around 1960, the
situation in this field changed. Figure 6 shows an example of an
instrument for the study of nuclear elements in which both fluore-
scence, excitation spectra and absorption spectra can be measured
one aiter the other without the object having to be moved out of
the field of vision of the microscope system.
The instruments referred to as yet were all the result of

efforts to analyze fine nuclear details and reached their present
stage of development in the early sixties. They are quite a potent
array of instruments for cytochemical chromatin studies of differ-
ent kinds. These instruments are fast, efficient and are easy to
use accurately on a routine basis. As to further improvements,
they have shown little reason for future development.
In the early sixties as a result of the interest in cancer

research there was a desire for instruments more suitable for kine-
tic studies of protein and nucleic acid synthetic processes during
growth. At that time, tissue culture material became more easily
available for studies of both normal and pathological growth. In
such studies, whole cells must be analyzed and measurements made
on large numbers of cells. Tissue culture offered good opportuni-
ties for preparation of cell suspensions in which entire cells
could be measured directly without the complications of tissue em-
bedding and sectioning. The instruments described above, in spite
258 T. CASPERSSON
Fig 6. Fluorometer for high resolution work. In the instrument re-

cordings can be made of fluorescence and excitation spectra and
also of the absorption spectrum of small cell details .
of their having been built primarily for analysis of fine structu-

ral details, also permit whole cell measurements, although they
are hopelessly slow in kinetic studies where one must analyze po-
pulations of hundreds of cells in each experiment.
This led us to the development of a series of instruments for

whole cell measurements by UMSP, interferometry and fluorometry
all working many times faster than was possible in the instruments
predecessors. Different instrument-models were built between 1960-
1970 and were used by several research groups for growth and dif-
rerentiation studies of many different kinds. As most of these
studies did not specifically concern chromatin questions, I need
not describe them here. Figure 7 illustrates the type of data ob-
tainable with such instruments for cell population studies. The
object is a rapidly growing tissue culture and the histograms de-
scribe the distribution of DNA, RNA and total dry mass in the cell
population.
Nuclear cytochemistry can be said to have gotten. an extra sti-

mulus around 1970 from the general work in carcinogenesis. A need
was felt for better ways to study the interplay between nucleus
and cytoplasm during cell growth and transformation processes.
Furthermore, considerable attention was being paid to chromosomal
changes during "transformation" and there was a need for better
ways to analyze metaphase chromosomes. Both these fields needed
considerable improvement of the then available equipment and pro-
cedures. During the last few years our group has tried to develop
better techniques for study of nuclear-cytoplasmic interaction as
well as for chromosome analysis. New equipment for high resolution
scanning and integrating UMSP and interferometry was built in
such a way that the nucleus could be measured separately from the
cytoplasm. For fluorometry, an instrument was built with which one
can, with great precision, outline the field of the cell to be
measured. Figure 8 shows the appearance of an UMSP equipped with
an arrangement for "preselection of the measuring field". The in-
strument works in the 220 - 700 nanometers range and permits work
15
10
N: 660
5
DRY MASS ( OM )
15
~ 10
N= 437
II:
II! 5
~
Z
...J
...J
34.4
RNA
1&1
u
IJ
20
15
10
N=594
0 3.0
DNA
Fig 7. Histograms of the distribution of dry cellular mass (large-
ly proteins), total cellular RNA and DNA in a rapidly growing nor-
mal tissue culture.
T.CASPERSSON
Fig 8. Rapid scanning integrating UMSP for UV and the visible re-
gions, equipped with arrangements for preselection of field.
close to the theoretical limit given by the lens resolution for

the wavelength used. The scanning is by mechanical movement of a
high precision mechanical stage. Long experience has shown that
this type of scanning gives much higher precision than scanning
by movement of optical elements, especially in the ultraviolet.
The underlying reasons are the difficulties in producing an ab-
solutely homogeneous illumination of a large enough field for the
specimen. These difficulties are especially great in the UV. The
chosen mode of scanning makes the arrangement for preselection of
field somewhat complex mechanically, but the design chosen has
proved to work in a satisfactory fashion even for large scale rou-
tine work. The instrument is built so as to permit measurements
of selected fields in the cell. In routine work, the size of the
measuring spot is generally set at 0.25 p and the distance between
the scanning lines at 0.5 p. For an average mammalian cell this
gives a measuring time of about 5 seconds. This value has been
chosen as it is about one magnitude less than the average time
needed for a trained operator to go from one cell to the next.
Provided the nuclear elements can be spread in the specimen

so that there is no overlying material, the instrument can be used
for measurements of nucleoli, chromosomes and chromosome details.
Large series of measurements of individual mammalian chromosomes
have, for instance, been made in conventional metaphase spreads.
Fig 9. Rapid integrating interferometer with preselection of

field.
Figure 9 shows a scanning, integrating interferometer with an

arrangement for field preselection. In this case it is not necessa-
ry to use stage scanning, which somewhat simplifies the situation.
In routine work, a series of interchangeable diaphragms of differ-
ent shapes are used as field limiters. This instrument has also
stood the test of large scale routine work well. The main value of
such an instrument in chromatin work is the possibility it gives
for measuring changes in nuclear mass (largely proteins), which
can be very great during different stages of cell growth, espe-
cially during blastogenesis.
262 T. CASPERSSON
Fig 10. High resolution fluorometer.
Fig 11. Arrangement for recording photographic fluorometry in

chromosomes.
A fluorometer with arrangements for preselection of field has

also been constructed (figure 10). It is built so as to permit
choice of the cell to be measured and setting of the preselection
device without any UV-irradiation of the specimen, which is thus
exposed to short wave irradiation only during the time of measure-
ment.
During the last few years, growing interest in making several

different kinds of measurements in one and the same cell has led
to development of systems for determination of the location of in-
dividual cells on a slide so that one cell can later be measured
in different instruments. In our laboratory a computer driven
stepping microscope stage is used for cell search. By pushing a
button the observer, after having selected a cell, automatically
transfers its coordinates to the memory of a PDP 8 computer to-
gether with a number, defining the cell. After a suitable number
of cells has been selected, the teletype writes out the coordinates
in millimeters as a list or presents them as a punched tape. The
individual cells can then be found and measured in the different
measuring instruments, either by aid of digital gauges directly
connected to the microscope stage (visible in figures 8 and 10)
or by aid of a teletype regulated stepping stage identical with
that in the search microscope.
It is often desireable to use both phase microscopy as well

as staining procedures for the determination of the cell type in
complex cell popUlations such as those met in cytopathology. Most
conventional nuclear stains greatly impair subsequent spectro-
photometric and interferometric work because of the high light
absorptions in the stained specimens and it is rarely possible to
extract the stains without losing cellular substances. Some help
in these matters can be obtained by the use of extremely weak
staining. By combining TV-microscopy with electronic image enhan-
cement the microscope image can be "lifted up" to a clearly ob-
servable level on the TV monitor.
The second field referred to above for which interest has

grown conspicuously during the last few years concerns the endo-
nuclear structures, particularly the chromosomes. Ever since the
beginning of quantitative cytochemistry, chromosomes have been
under study, but these studies have mainly concerned species
with very large chromosomes or with peculiar chromosome organiza-
tion. The rapid progress of cancer research, however, has put the
very small metaphase chromosomes from higher animals and plants
in the center of interest. The length of such chromosomes, for in-
stance those of man, lies around 3 - 8 p and the thickness of the
individual chromosome is around 1 p, that is about two wavelengths
of green light. In spite of their small size they can be worked
on in the UMSP and microfluorometric instruments described in the
beginning of this article. This is illustrated in figure 4, where
T.CASPERSSON
the size of chromosomes corresponds to that of the larger human

chromosomes.
However, before any reasonable cytochemical analytic work on
mammalian karyotypes could be done it was necessary to have proce-
dures for identifying individual chromosomes and preferably also
regions within these chromosomes. In our laboratory this led to
the quinacrine banding procedure for metaphase chromosomes. I men-
tion that here mainly because it is a direct derivative of quan-
titative cytochemistry in that the whole work establishing the con-
stancy of the chromosome patterns and the differences between chro-
mosomes rested on f1uorometric measurements of fluorescence pro-
files. These were first recorded in the instrument depicted in
figure 10. Later supplemented by a faster working arrangement with
photographic fluorometry (figure 11), With the latter such a large
number of fluorescence profile measurements were collected that
a comprehensive statistical analysis could be made by computer.
This showed banding pattern analysis to be a reliable tool for
chromosome identification.
For most routine applications of the chromosome banding tech-
niques as yet there is no need for cytochemical measuring devices.
They were, however, necessary requisites for the development of
the method. In the work aiming at automation of the chromosome ana-
lytic work, measurements of banding patterns is an important link.
In summary it can be said that q~antitative optical cytoche-
mistry offers a considerable number of tools for study of chroma-
tin structure. Microspectrophotometers for ultraviolet and the
visible, as well as microf1uorometers permit analysis of nuclear
structures down to the size range of 0.5 p (in ultraviolet 0.3 p).
Integrating measurements can routinely be made of nuclear details,
whole nuclei and/or whole cells. Microinterferometry, used for dry
mass determinations can be applied to objects of over 1 p in size.
Microinterferometric mass determinations - as spot measurements
or integrating measurements - can also be applied to cellular ele-
ments larger than 1 p.
For the study of chromatin function, the new instruments with

preselection of measuring field can give information about the ex-
change between nucleus and cytoplasm. In this connection it is al-
so an advantage to have access to the arrangements for mu1tipara-
meter measurements which have recently been developed.
CHROMATIN STUDY IN SITU:
I - IMAGE ANALYSIS
F. Kendall, F. Beltrame and C. Nicolini

Division of Biophysics
Temple University Graduate School
Philadelphia, PA 19128
INTRODUCTION
In all physico-chemical and functional characterization of
isolated chromatin, there is of course a large degree of artifact
introduced during the isolation process. In this respect abundant
evidence exists in the literature that the structural and function-
al properties of native chromatin are modified by physical altera-
tions introduced during chromatin preparation (1,2,3). In order
to bypass such limitations, one approach has been to analyze the
geometry and densitometric textures of the nuclear images produced
by differential staining of chromatin-DNA in situ (such as by a
Feulgen reaction). For a detailed account of the most recent find-
ings during image analysis on the study of chromatin during cell
cycle, visus transformation and cell proliferation, the readers
should refer to recent publications (4-7). Here we intend only to
outline a critical overview of the state of the art with advan-
tages and limitations.
AUTOMATED IMAGE ANALYZER

An image analyzer for densitometric-geometric studies is
usually equipped with a plumbicon scanner (that, having a light
transfer characteristic of 0.99, warrants an overall linear cali-
bration between 0.0 OD and 2.00D) and densitometer module (see
Fig. 1) on line with a digital computer. The scanned area is
divided into 880 x 588 picture elements, each of which can be
digitized into 64 gray levels where the area of each picture
point is (0.09 x 0.09) ~m2 under the given optical conditions
(4-7). Fig. 2 shows how, by applying the proper optical threshold
265
266 F. KENDALL ET AL.
Automated Image Ana lysis

Fi.ld Specific
Data Output Serial S.ri.1 Puis ••
Thr•• hold, Puis. I.pr.senting
1o--=8;.;.in;;.;a;.;.r..:..y......~Chorcl-S i •• , ,..,..F.:.:i.:.:.:ld=---_ _.TI>
Video Connecting Count.,
Lo ic
D
;:
L-_+~ Hardwar.
I-......:LJ---7~ Featur.
Optical Input
TTY Ou,pu,~------I
Disc Storage ~----I

L...-_ _---I
Fig. 1. Block diagram of inage analysis system which utilizes

hardware real-time processing of morphometric and densitometric
data in accordance with chord-size and programmable optical density
criteria. The system can be expanded to four feature parameter
computers for simultaneous real-time computation.
to eliminate background, the nuclear image can be properly filter-

ed for subsequent geometric-densitometric analysis.
Nuclear morphometry itself is a technique that readily

exploits existing hardware to preprocess image data. Thus thres-
hold and minimum chord size criteria can be used in real time to
define the border of a feature during scanning, and such measures
as integrated optical density (proportional to DNA content for
Feulgen staining) and nuclear area can be output at the end of a
scan through a feature. Distributional error is negligible using
oil-magnification. This technique places a premium on such global
descriptors as the area and perimeter which are independent of the
state or orientation of the cell. Such measures as Horizontal and
Vertical Feret diameters (shadow projections) and complex projec-
tions (sums of shadow projections of lagging edges) can be made
substantially independent by employing the Euclidean norms of
orthogonal projections, thus yielding global descriptors that are
sensitive to both shadow diameters and re-entrance. Table I shows
the dramatic improvements in coefficient of variation and total
range which are realized when the Euclidean norms, and horizontal
and vertical complex projections are averaged for various re-
entrant figures as they are rotated 90 degrees.
CHROMATIN STUDY IN SITU: I 267
..· :... ..... ...--...........

~ - ...... .... .:
··:~,
··: ................
,
, ............ ...... .
.................
.................
................
....... ........ .
..,:
··.'·
:'
...............
.................
... ...........
... . ..........
.· ,
'
. .............
. .......... .
-:::::.:.: .
,
.:
................
...............
.. . ..................
.. ............. ..
··
·
,
··:::
'
,
'
· ·· ... ..... .. .. .... .... ....... ..•

. :,
'
-----
Fig. 2. The left panel shows a line printer listing of discretized
optical density values unloaded point-by-point from a given area of
a microscopic field into the core of a minicomputer. Application
of the detection logic using analog thresholding and chord-sizing
results in the "cleaned" image which appears in the right panel.
Speed can be enhanced by parallel info~ation processing.

This becomes obvious to the users of analyzers which are capable
of providing a point-by-point image transfer into a computer. A
considerable amount of subsequent computational time is required
to establish connectivity in order to measure the perimeter of the
feature. In contrast, the perimeters of 10 or 20 objects can be
computed and released with existing hardware in 200 msec. A block
diagram of such a system is shown in Fig. i. This system computes
one geometric parameter (e.g., area) for all the features (e.g.,
nuclear images) in the field per feature parameter computer per
field scan. Two different parameters for each feature can be com-
puted simultaneously in a single field scan if 2 computers are
used. Permissible regions are established by analog optical den-
sity settings of the threshold detector and chord-sizing logic
which establishes connectivity of the resulting border and the
x-y coordinates of the lowest point (feature flag). Feature para-
meter data are only computed by the feature parameter computer on
the basis of picture elements which fall within the permissive
TABLE I
GLOBAL INVARIANCE OF THE EUCLIDEAN NORM
OF THE COMPLEX PROJECTIONS OF THREE RE-ENTRANT FIGURES
Co-efficient of Percent (1) Percent (2)
Letter Parameter N Variation Percent Spread Range
Y Horizontal 10 15 41 53
Vertical 10 15 39 47
Norm 10 1.2 3.9 4.0
X Horizontal 10 3.8 10 11
Vertical 10 3.2 8.7 9.1
Norm 10 0.58 1.5 1.5
S Horizontal 10 8.6 25 28
Vertical 10 6.8 21 24
Norm 10 0.54 1.8 1.8
(1) Percent spread is defined by the range divided by the mean.

(2) Percent Range is defined by the range divided by the minimum
value.
The test figures used were opaque letter "Y", "X", and "S" which
were affixed to a glass slide, magnified 30x, and rotated through
90°.
regions and whose discrete optical density values equal or exceed

those which are established by the densitometer. This configura-
tion permits disconnected geometric data (such as the total area
of several separate islands of high optical density) to be associ-
ated with the feature to which they belong, rather than generating
additional features as functions of optical density threshold.
The feature parameter computer and the multiplexer/interface asso-
ciate the geometric data for each feature with the corresponding
feature flags. When a scan line crosses a feature flag, the geo-
metric and coordinate data associated with the flag are unloaded
directly into the core of a minicomputer. The densitometer can
transform analog picture element luminances into logarithmic opti-
cal density values on a scale of 63 discrete digital values. The
feature parameter computer can thus compute integrated optical
density (lOD-proportional to DNA content in Feulgen-stained nuclei)
for individual features. It is also possible to load the digitized

optical density values for each element in a feature directly into
the computer core memory. With parallel processing this system
can compute up to 4 different parameters of many features simul-
taneously. It is also possible to automatically program optical
density threshold, which may be programmed in equal density steps
at least .01 OD to obtain 7 or more thresholds.
We have found that a satisfying compromise can be made between

the desirable speed of parallel processing (and its attendant hard-
ware costs), the necessity of tnsuring that only desired images
are collected, and the necessity that resulting data sets are as
free of artifact as possible. The greatest saving in time is
effected by on-line computation of any particular parameter for as
many features as may exist in a field. Restricting the number of
parallel processors to one increases the number of scans required
to· compute the specified number of parameters at each specified op-
tical density threshold. This increase in the time necessary to
make all of the necessary measurements is advantageous because
artifacts which depend on electrical noise in the image circ·uits
and in the analog threshold circuitry will be afforded opportuni-
ties to add or delete features to the field and to shift the posi-
tions of others. Software coordinate matching at each optical
density threshold deletes features which cannot be matched on suc-
cessive scans. Other software checks the total number of features
at the completion of all scans at each optical density threshold.
If the total number of features at all thresholds is invariant, a
software check then uses two orientationally independent global
descriptors, to compute a form factor (area divided by perimeter
squared) which cannot exceed ~n.
As an added precaution the numerical index and integrated

optical density (IOD) value of each image are displayed next to
the image, and the parameters acquired are printed as hard copy
in a procedure which was designed to slow the acquisition process
to a point where a human observer would be able to manually select
a single feature for deletion from the acquired data set. As a
practical matter, this complimentary use of hardware and software
has routinely resulted in the lowest coefficients of variation com-
patible with various parameters of images of nuclei of cells in such
functionally homogeneous states as may be expected several hours
after synchronization by selective mitotic detachment. Inasmuch
as the total acquisition cycle for an artifact-free field of 5
images is of the order of 30 seconds, the additional 2 second (7
percent) burden imposed by lack of parallel processing is a rela-
tively small price to pay for the resulting quality of data.
The absence of serious distributional error leaves the ques-

tions of gray scale size, stability (repeatability) and field uni-
formity (including distortion and shading error) to be answered.
The first case represents a practical tradeoff between various

costs and usefulness. There is little merit in achieving even 6
or 7 Bit gray level resolution unless it can be justified for a
specific application. State of the art equipment may approach
this resolution capability in terms of signal-to-noise ratio if
signal-averaging is employed. Stabilities of scanners, -analog cir-
cuits, and illumination systems can be made acceptable for periods
of one hour between recalibration. We have made 110 measurements
of integrated optical density (lOD) and area over a one hour period.
The coefficient of variation of nuclear lOD was 1.8%, while the co-
efficient of variation for area was 0.9%. Some of this variation
was induced by the bleaching of the Feulgen-stained specimen. We
routinely achieve field uniformities (estimated by measuring the
same image at nine locations through the top, center, and lower
portions of the scanned area) sufficient to yield a coefficient
of variation less than 3% for nuclear IOD and less than 1% for area.
Cell populations which are in relatively homogeneous functional

conditions (shortly after synchronization by selective mitotic de-
tachment or in confluency) will easily produce larger coefficients
of variation for these parameters. The only instrumental error
directly comparable to flow systems performance is lOD and the
validity of comparison can be questioned on the basis of the physi-
cal nature of the analog quantities measured and the inherent in-
fluences of random artifacts common to each system. As with most
exotic systems with many optical and electronic degree of freedom,
error is also a function of methodology or lack of it.
GENERAL CONSIDERATIONS
State of the art hardware processing adds desirable speed to
the necessary transformation of monumental volumes of data in
pattern space (which can be acquired with great ease by image scan-
ning systems) to feature space so that the result can be applied to
description space. This is also a necessary process of image
classification (8) because it is required in order to produce re-
sults which are physically comprehensible to humans. The point of
departure between the classical use of image analysis for purposes
of pattern recognition and classification and its uses for an ana-
lytical determination of the structure-function relationships of
chromatin resides in the viewpoints of the two approaches: in the
former the image is to be described in a fashion whereby it can
be recognized and/or classified; in the latter case the image
is treated as an incidental necessity in the same manner as a
hologram or other optical transform.
A frequent complaint that was voiced during Automated Cytology

V in 1976 was that financial support for image analysis investiga-
tions was small compared to that available for investigations in-
volving flow systems. This state of affairs is readily understand-
able when the literature of biological (specifically medical) image
analysis is reviewed: generally the approach taken is that of

image classification (9,10). The overriding purpose is to sup-
plant slow, error-prone and expensive human classifiers with fast
and efficient automatons that have suitably low error rates. In
short, the objective is all too frequently to achieve sophisti-
cated, ~pid and economical quantitative evaluations of geometric
and densitometric data in order to duplicate uncertain, qualita-
tive, and frequently subjective visual value judgements that rely
heavily upon experience, less upon stoichiometry, and least upon
precise geometric or densitometric measurements. Although the
basis of human discrimination of texture has been modeled by
Markovian processes and satisfying agreement with human judgement
has been achieved by image analyses based on this model, the ac-
tual process is probably not fully described (11), and no systema-
tic body of quantifiable data exists which relates this optical
property in rigorous fashion to either chromatin structure or
function. The apparent morphological integrity of fixed and
stained cells has subtly beguiled countless workers into a re-
duced awareness of the fact that they are observing a standardized
artifact which in both microscopic and molecular definitions is
not synonymous with the original living system.
Contrary to most workers, we attempt to assign physical

boundary conditions to image analysis in the manner of flow sys-
tems where we made every attempt to relate stoichiometry and
fluorescence as well as interference to structure and structural
interactions. Perhaps accurate geometric and densitometric
measurements are such exotic entities that habit does not suggest
that they be systematically applied to understanding the physical
differences which exist in "artifacts" obtained from well defined
relatively homogeneous cell populations in different functional
states.
CHROMATIN CHARACTERIZATION
Recent interest in nuclear morphometry based on the Feulgen
reaction seeks to exploit biological image analysis as a tool to
quantitatively measure differences in the resulting organization
of nuclear D~ spaces from cells in different functional states.
These studies were prospectively designed to determine if physi-
cal differences in DNA spaces probably reflected structural dif-
ferences which had been observed to exist at the cellular and
molecular levels. The advantage of this approach when contrasted
to the classification approach, is that the former method usually
makes it easier to design experiments in accordance with testable
hypotheses. Even within its limitations, the Feulgen approach
may permit highly specific and stochiometrically accurate (12,13)
(under proper optimal acid hydrolysis (14» identification of
chromatin DNA organization in situ. Indeed, convincing evidence
exists that such functional properties as template activity of
isolated chromatin and such physical properties (which are rela-

ted to the tertiary-quaternary structure of chromatin) as
ethidium bromide binding sites are directly related to the morpho-
metric and densitometric properties of the DNA space of Feulgen-
stained nuclei.
In order to focalize on the meaning of "form factor" in the

study of chromatin in situ~ we may analyze its relationship to
average optical density (Fig. 3) and to the number of primary
binding sites of chromatin isolated from the same cell lines
(Fig. 4 bottom): these data clearly indicate that increased chro-
matin convolution (lower form factor, accompanied by a decreased
mean bound path in the face of increased area) corresponds to
an increased chromatin condensation (higher degree of DNA super-
packing) in situ and to an increased number of sites for inter-
NUCLEAR MORPHOMETY HE LA
8
I"N"1
o
~
6
III:
o
....
u
c(
I&.
6 12 18
AVERAGE OPTICAL DENSITY
Fig. 3. Relationship between mean values of form factor and

average optical density of nuclear images of Feulgen-stained
HeLa cells synchronized by selective mitotic detachment and
observed at 3 hours (early Gl-eG), 5 hours (mid Gl-mGl), 12 hours
(mid-S-mS) and 15 hours (G2) under overall magnification of
lOOOX at a wavelength of 540nm.
~ 4
~
...>-
.•
lit
Z
.
III
!
...
Q
~
<C ~
...~
A.
lit
_ ..a
0
III
c
" .E
<C
III
0
~
:;I
>
<C
TEMPLATE ACTIVITY %
8
f'N'"1 • HELA SYNCHRONIZED
o ..
..
- :;I
L.......!!J ':
:;I
-6
o ..
_tit
... >-
V a
<c f
...
~.;
.
-4
...0
0
50 1 0
BINDING SITES or CD(%)
T.rtlary - Quat.rnary Structur.
Fig. 4. (top panel) Relationship of Average Optical Density of

nuclear images of Feulgen-stained 7-day confluent WI-38 cells
(WGO) and stationery 2RA cells (RA) and of 7-day confluent WI-38
cells three hours after receipt of a nutritional stimulus (WGI) to
the percentage of template activity of chromatin extracted from
similar cell populations.
(bottom panel) Relationship between measured form fac-
tor and percentages of available binding sites of chromatins from
synchronized BeLa cells in various stages of the cell cycle
(early and mid GI, mid S, and G2).
calating dyes or macromolecules such as R~ polymerase. Separate

experiments have shown that an inverse relationship exists between
the average optical density and the measured template activities
of cells in different functional states (Fig. 4 top).
The interpretation of chromatin morphometry, and its modulation
during various functional states of the cells is confirmed by the
threshold dependence of the geometric parameters (see Fig. 5 for
synchronized HeLa nuclei) which for instance may uniquely prove the
existence of a drastic increase of chromatin dispersion during late
GI (5 hours) - early S (8 hours). Indeed, while the geometric
parameters of Feulgen-stained HeLa cells are quite similar between
5 and 8 hours after selective detachment, the nuclear area (upper
panel, Fig. 5) or perimeter(lower panel go to zero for early S
(8 hours) with increasing optical density threshold, but maintain
a large positive value for late GI (5 hours). This is compatible
with an increased chromatin dispersion permitting objective ident-
ification of the two distinct subphases even if geometric and den-
sitometric (DNA) parameters are similar at the base threshold.
Similar objective identifications on the basis of threshold depend-
ence, are possible on other cell cycle phases and subphases (see
ref. 4-5 and Fig. 5) to a level up to now impossible to any human
observer.
Even a conservative a priori binary classification of functional
state should be based on some physical evidence that at least one
discrete state is involved. The study of the morphometry of the
DNA space has revealed that, in those instances where a binary state
classification is warranted, the geometric and densitometric changes
in the DNA space are consistent with alterations in chromatin struct-
ure which have been observed by other means (6,7). And the specifi-
cation of functional state of a cell population in accordance with
a 'magic recipe" will probably prove to be inadequate to the task
of definition. Definition of functional state in terms of the
physico-chemical analysis of bulk samples of chromatin may itself
be misleading inasmuch as all distributional information is lost;
measurable differences in physical properties can be due simply to
changes in percentage composition of the contributors. This was
the case in the particular instance of the conformational change
of chromatin associated with the GO-GI transition. The discrete
nature of the transition could only be reasonably inferred from
flow system measurements which yielded enough distributional data
to warrant the assumption that conformation-dependent changes in
dye content were not part of a continuum. In this instance the flow
system was indispensible, but subsequent image analysis studies have
provided support for the physical interpretation of the results of
the previous work (6,7). Similarly, geometric-densitometric analysis
has itself provided sufficient distributional information to show
that it is possible to unequivocably differentiate populations of
Feulgen-stained confluent WI-38 cells and
60
0.04 0.12 0.20 0.28 THRIlSII(U) 0.04 0.12 0.20 (l.28

OD UN ITS oD U NIT S
60
B
1211
0 .04 0.12 0. 20 0. 28 Tll<g<;H(lJ)

OD UN ITS oD UN ITS
Fig . 5. (upper panel) Mean values of the area (~m2) of Feulgen-

stained nuclei of 53 He La cells at seven different times after mi-
totic detachment (1,3,5,8,12,15, and l8-h levels) versus threshold
from 0.04 aD (defining the nuclear border of each image) to a max-
imum value of 0.32 aD with equal spacing of 0 . 04. The average
points versus threshold are obtained on the same constant number
of cells (about 100), including zero values at high threshold .
(lower panel) Perimeter mean values (~m) of Feulgen-stained nuclei
from HeLa cells at 1,3,5,8,12,15 and 18 h after mitoses, measured
at eight different optical density levels (0.04 OD-0.32 OD) (ref.
10).
their stationary SV-40 transformed counterparts (2RA ceZZs) on the

basis of average optical density. The IOD (Fig. 6 top) and area
(Fig. 6 center) of the 2RA nuclear images show increased disper-
sions, which are not surprising inasmuch as chromosome numbers can
vary considerably from cell to cell with reported values of 68% of
metaphases containing 61-80 chromosomes and 19% exceeding 120
chromosomes. A correlation between IOD and area of the trans-
formed cells is to be expected on the elementary basis that nuclei
with more chromosome will probably be larger. The striking degree
of correlation (r = 0.98) suggests that a very systematic rela-
tionship may exist in the particular case of quiescent 2RA cells
which is sufficient to be clearly seen through the biological and
instrumental "noise" which are causes of dispersion of measured
values of IOD and nuclear area of the WI-38 cells (the majority
of which have 46 chromosomes). Indeed, while the frequency dis-
tributions of directly measured parameters, such as area and IOD,
show a substantial overlap between WI-38 and 2RA cells, a derived
parameter (IOD per unit area) yields two distinctly separated
distributions, which permits a unique objective identification of
individual SV-40 transformed cells regardless of heterogeneity of
chromosome numbers (Figure 6 bottom).
For the present, the full realization of the usefulness of

rigorous multiparametric analysis of image data for analytical
purposes is hampered by a lack of adequate physical definition of
the functional state of the target population that can never be
overcome by the percentage of agreement among the subjective
visual judgements of a panel of I~xperts. II N-dimensional cluster
analysis will not in itself provide any physically meaningful in-
formation relating structure to function in the absence of ade-
quate experiments and models to explain the possibly non-linear
connection.
FUTURE DEVELOPMENTS
In spite of the high hardware costs involved it is our
opinion that biological image analysis systems will evolve (with
certain specialized exceptions) in favor of image-plane scanning
systems employing an amalgam of analog video technology and
digital logic and which will be interfaced to small computers for
purposes of control and data acquisition. The current state of
the art is such that major advances in technology are not required
to provide the demand necessary to generate sales volume that will
lower with costs a la pocket calculators. Demand is predicated
on the acceptance of utility and acceptance is the factor that is
lacking even because of the primitive suspicion or superficial
attitude of most life scientists toward advanced automated tech-
nology.
LLI
...J60 _1·31 A
U
::J
:z
LL..
o
LLI
Cl
~
:z
LLI
U ..
n
a:::
LLI
a. ..; :
.'
::
:1 n
20
- r' "·-1 r- f1
40
J
ISO
: ..
to
INTEGRATED OPTICAL DENSITY

LLJ (a. u. )
<340
::J
:z WI-38
LL.. B
o
LLI
Cl20
n
~
:z
LLI
,,
,
,
.
I
,, I,
U
a::: I
I
I
I
LLI ,," ,,
a.
~~~--~~~~~~
AREA (micron 2)
, ,
...,
In
....I
....I
'11'1-:'
c
LLJ
U
LL..
0
LLI
Cl
'"
~ [_.I
:z
LLJ
u :-1: "'- .. ,.-,
a::: : l •
LLI i :".J: I
a. ~..
AVERAGE 0.0.
Fig. 6. Histograms of Integrated Optical Density (top), Area
(center) and Average Optical Density (bottom panel) of nuclear
images of Feulgen-stained confluent WI-38 cells (solid lines) and
stationary 2RA cells (broken lines).
The recent systematic application of morphometric analysis

to cell population will receive added impetus from recent advances
in multiparameter cell sorting. Under proper conditions (acridine
orange concentration of 10-4M and low molar ratio R (dye/DNA)
fluorescence-"sorted" cells have sufficient quantum yields to be
detected by a plumbicon scanner and consequently permit the anal-
sis of ahromatin differentiaZZy stained by a supravital dye. It
is then realistic to assume that this technological marriage will
eventually lead to a systematically developed body of knowledge
which will better connect the morphometry of DNA space to the
physical reality from which it was created.
In order to reconstruct the three-dimensional organization

of chromatin-DNA at the miaron level (quinternary structure), we
are presently transferring the nuclear image point by point into
the core of a mini computer by direct memory access under software
control. The image is then stored on a mass storage device and is
available for display on any display device connected to the
Unibus. (See Appendix A) The image (Fig. 7) is presented in
terms of discrete optical density values in a properly calibrated
scale of 63 possible levels for analysis of the patterns of
optical density distribution within each nucleus relative to its
center of mass. By means of one and two dimensional Fast Fourier
transforms, we are searching for eventual periodicities in the
geometric-densitometric properties of chromatin-DNA in situ which
are not detectable by other means and that would be indicative of
an ordered structure at the higher "quinternary" level of chromatin
organization. Preliminary results (Figs. 7-8) indicate indeed
that reproducible periodicity, a unique function of cell cycle
phases, does exist around the nuclear DNA border within a
SOOO-8000Ao band (readily detectable only in mid-GI but not in
S or M phases) and on the central sections of the same nuclei,
(for every cell cycle phase) compatible with a model previously
outlined for the "quinternary" organization of chromatin-DNA
in situ (see pg. 660). Any periodic variations in optical
density due to a drapery-like periodic structure around the
nuclear DNA boundary (as detected in GI) would tend to be
obscured in most central areas by squashing the nucleus and
would most likely be enhanced in the region lying along the image
border (where absorbance would be larger). Analysis of several
nuclei in the same cell cycle phase indicate the existence
of a highly reproducible chromatin-pattern and DNA periodicity,
suggesting that interphase chromatin is organized in a quite
ordered and isotropic '~uinternary" structure (see Appendix A)
which modulates along the cycle. Preliminary studies using cross-
correlation of parallel averaged scans obtained at intervals
well spaced with respect to optical resolving power indicate
that DNA periodicities can be detected even in central sections
of synchronized HeLa nuclei, in early Sand mid-GI phases
(Fig. 9).
A.
_ B.
• c.
Fig. 7. Perspective optical density displays of three images

chosen at random of Feulgen-stained HeLa nuclei from a smear made
of cells selectively detached 3 hours earlier. An annular group-
ing of higher optical density values is apparent from the displays
but is not obvious in the photographs. (middle GI nuclei)
Fourier Analysis of Nuclea r Section

from 104 E LA 01 Cells
N-'
X(k)= ~ x(n) W!n
n=O
Y-Sectlons X- Sections
Fig. 8. Power (top panels) and phase (lower panels) spectra of

optical density values acquired along the borders and through the
center of the nuclear image of middle Gl Feulgen-stained BeLa
cells. Comparison between data obtained in two orthogonal axes
indicates that the similarity of results is dependent upon the
portion of the nucleus scanned rather than the direction of scan.
Identical results are obtained in other nuclei from middle Gl HeLa
cells.
Auto Correlation of Cross Correlation Between Two

Two Periodic Functions Centra I Sections of Nuclea r Images
(256 p.p./sec t ion)
Sin ... t
Sin3 ... t
Early S
Fig. 9 (left panels) 256-points auto correlations of two 256-

points periodic functions, with the period being three times
smaller in the bottom respect to the top function.
(Right panels) Cross correlations between monodimensional parallel

scans (averaged over 1000 times to enhance signal to noise ratio)
of a synchronized HeLa nuclei, at early S (bottom) and mid-Gl (top)
phase.
In these studies we have utilized linear cross correlation

algorithm (see Appendix).
Identical cross correlations were found, when parallel scans

were separated by either 10, 20, 40 or 60 picture points.
These data show the repeatibility of the same optical pattern

along one direction, as function cell cycle phase.
The correlation was computed always via Fast Fourier Transform

(see Appendix A) .
ACKNOWLEDGEMENT
This work was supported by Grant CA20034 from the National Cancer
Institute.
REFERENCES
1. Nicolini, C., Baserga, R. and Kendall, F., Science, 192, 796

(1976) .
2. Nicolini, C., and Kendall, F., Physiol. Chern. and Phys. (1977).
3. Noll, M., Thomas, T. and Kornberg, A., Science, 187, 796
(1975).
4. Nicolini, C., Kendall, F., and Giaretti, W., Biophys. J., 19,
163 (1977).
5. Kendall, F., Swenson, R., Borun, T., Rowinski, R. and
Nicolini, C., Science, 196, 1106 (1977).
6. Nicolini, C., Kendall, F., Desaive, C . , and Giaretti, W.
Exp. Cell. Res., 106, 119, (1977).
Nicolini, C., Linden, W., Zietz, S. and Wu, S., Nature,
270, 607 (1977).
7. Kendall, F., Wu, S., Giaretti, W., and Nicolini, C., J.
Histochem. Cytochem., 25, 724 (1977).
8. Andrews, H., Introduction to mathematical techniques in
Pattern recognition, Wiley-Intersci., New York, Ch. 1-2
(1972) •
9. Bartels, P.H., Bahr, G.F., Jeter, W.S., Olson, G.B.,
Taylor, T., Jr., and Wied, G.L., J. Histochem. Cytochem.,
22, 69 (1974).
10. Bacus, J.W., IEEE Trans. Sys. Man. Cyber, SMC-2, 2, 513 (1972)
11. Pickett, R.M., J. Exp. Psychol., 68, 13, (1961).
12. Garcia, A.M. Stoichiometry in Dye Binding versus Degree of
Chromatin Coiling, Wied, G.L. and Bahr, G.F. (eds) ,
Introduction to Quantitative Cytochemistry-II, Academic
Press, New York, 153 (1970).
13. Dijndarn, W.A.L. and Van Duijn, P. The influence of chroma-
tin compactness on the stoichiometry of the Feulgen-Schiff
procedure studied in model films II. Investigations on films
containing condensed or swollen chicken erythrocyte nuclei.
J. Histochem. Cytochem., 23:891 (1975).
14. Linden, W., Zietz, S., Fang, S. and Nicolini, C., in "Chroma-
tin Structure and Function," ed. Nicolini, C., Plenum Co.,
pubs. (1978).
APPENDIX
Although the pattern of chromatin morphology studied by image

analysis is essentially a two-dimensional one, much information
can be obtained by analyzing one dimensional cross sections of
the image. In particular, such monodimensional techniques can
help in identifying isotropy and repeatibility of certain optical
density patterns acquired column by column.
One of our approaches to trying to discover periodicities

in chromatin morphology is to use cross correlation techniques
between different scan lines of the image, either parallel or
orthogonal to each other. To enhance the signal to noise ratio,
we first average each scan line of 256 pp 100 times. We then
attempt to find existing periodicities in the 256 pp average
lines.
To aid in this search as well as to calculate certain

correlation functions, it is extremely useful to utilize
techniques from Fourier analysis. For the reader, we will
review some of its theory below. Although our problem is
inherently discrete, for sake of compactness of presentation,
we will use the continuous analog where no confusion can
develop.
If f(x) is a real function of position, the Fourier transform

F(w) of f(x) is defined as:
F(w) = where j = (1)
It is a basic fact that the original function f(x) can be obtained

from its Fourier transform by the following inverse formula
1 00 •
f(x) = --2
1T
f F(w)eJwxdw
-00
(2)
The precise hypothesis that permits us to define (I) and

(2) and show their equivalence can be found in reference 1.
What is to be noted is that although f(x) is a real function,

F(w) is a complex valued function. We take advantage of this by
writing F(w) in its polar form.
F(w) = A(w)eH(w) (3)
where A(w) and ~(w) are real functions. A(w) is known as the
amplitude and (w) is the phase.
With this representation, we then can interpret equation (2)

as representing f(~) by a sum of elementary periodic functions
(in sine and cosine), the contribution of the period w to the
function f(x) is given by the amplitude function A(w).
CORRELATION FUNCTION
If two functions flex) and f 2 (x) are given, this cross correZation
R(t) is defined as
R(t) (4)
In the case that fl and f2 are periodic functions of period T

the cross correlation becomes
T/2
R(t) = -T1 J (5)
-T/2
If fl and f2 are the same function, then the function R is called

the autocorTeZation of f.
RELATIONSHIP BETWEEN FOURIER TRANSFORM AND CORRELATION FUNCTIONS

The correlation function R(t) can be expressed in terms of
the Fourier transform. Let us define the p01J}er spect:l'WTl Sew) of
a function f(x) as the square of the Amplitude function A(w) Le.,
S(w) = lA(w)1 2
Then the autocorrelation function R(t) can be shown to be equal

to the inverse Fourier transformation of the power spectrum, i.e.,
T
R(t) =....!..
2lf
J
-T
This relationship is useful in computing correlation functions.
APPLICATION OF FOURIER TECHNIQUES TO DISCRETE MEASUREMENTS

In applying the above theory to help search for regular
structures in the chromatin geometry, we must first consider how
the discrete nature of the data affects the problem. Recall that
a row of the nuclear image consists of 256 picture points. We
wish to ascertain whether there are periodic regularities to such

a column. However, it is impossible to ask whether a finite
record is periodic. Instead, what is usually done is to arbitrarily
make a periodic function by repeating the record. We then search
for periodicities in the new function, remembering that we have
arbitrarily introduced a period T = length of record = 256 pp.
An important result of the Fourier analysis of discrete

signals is the Nyquist Sampling theorem which states that if
f(x) is a function such that its Fourier transform F(w) = 0
for Iwl>wc then f is uniquely determined from its values
fn = f(nT)
Wc
at a sequence of equidistant points.
In the case of time varying signals this condition requires

that the sampling frequency be at least twice as great as the
highest frequency which is present in the signal which is
being sampled in order to avoid the phenomenon of aliasing
(in which the high frequency components become "folded" into the
lower frequency components by the sampling process itself). In
practice, the sampling frequency is usually set to about three
times the highest significant frequency in the incoming signals.
The sampling process itself is entirely oblivious to the nature
of the incoming "signal". If the "signal" contains high-frequency
noise components of greater than one half the sampling frequency,
these components will be aliased and will appear as lower-frequency
spectral amplitudes which are intermixed with those of the true
"signal". Consequently, most sampling systems contain one or
more low pass filters which attenuate frequencies higher than a
value determined by the sampling frequency.
In the case of image-plane scanning systems, the picture

information is "read" by an electron beam which is swept across
the target face at constant velocity. The signal is sampled at
equal discrete distance intervals by means of clock circuitry.
This space-domain signals are mapped into time domain signals and
the corresponding frequency domain representation of these signals
has its counterpart in the "spatial frequency" (or inverse distance)
domain of the image itself. For any scanning velocity the maximum
spatial frequency is dependent on both amplifier band width and
optical resolution. Several techniques are utilized to reduce
noise, and prevent aliasing.
Amplifier noise can be reduced by reducing amplifier band

width. Spatial resolution can be preserved if slow-speed
scanning techniques are employed to keep the highest frequency
components of the signal within the reduced band width. The
optical system itself can be used as an effective low-pass filter.

Low distributional error of sampled optical density values
requires that the dimensions of a single scanner picture element
correspond to dimensions of the external object which are below
the limits of resolution of the optical system. Inasmuch as
the limit of optical resolution of a lOOx, 1.25 NA objective is
about O.3~ at 540 nm and the dimensions represented by an individual
picture element in our system is about 0.7~, the minimal resolvable
distance is about 1/3 that which can be resolved by the optics.
In other words, the spatial sampling frequency is above three times
the optical band width, so aliasing will not be a problem.
However, discretized luminance signals contain random electrical
noise which can contain some components corresponding to high
spatial frequencies. Inasmuch as this noise is random, it can
be removed by the simple process of sampling the same field 100
times and averaging each individual sampled value.
Making the computation more explicit and keeping in mind the

relationship between "time" and "frequency", for example for a
periodic signal of period T, i.e.
T l/f = 2n w = 2n f
w o
o
we can interpret the transformed domain as an inverse-distance
space, and attribute_Go w the dimensionality of distance- l with
unit [l/~] (l~ = 10 m)
Optical, Density. or Transmittance Val,ues

f(y)
0-63
0 I
I
I
I
t f
\F(w) \
I I
N
*T
I I
"2- 1
Y
2n
[rad/~] or
NT
Suppose T = .05~, N = 256

Resolution in frequency = 256 x1 .05
'" .078 [l/~]
Maximum frequency = 128 x 78 x 10- 3 [1/~] '"= 9.984 [1/~]
.
l.e. '"
= • 1~ (because of the symmetry).
This is below the resolution of the microscope (d '" .26~)
/I of sample + 1 16 128
I
frequency N
value + 1. 25 2.5
I
0 5 9.98 w,f[l/fl]
I I I I
I I I I
~-value + .8~ .4fl .2~ !l~
n
I I I I
I I I I
I I
I I
I I
I I
I I
Border
Line
Region not detectable

because below the
resolution.
(*) For example this region could be the frequency window in which
are present the so called "higher ordered fibers" of the chromatin
(0.16 f 0.3fl) detected by freeze-fraction (S. Bram) and wet
replica (S. Basu) techniques.
288 F. KENDALL ET AL
It can be shown (see Reference 3) and 4)) how the spectrum

estimate obtained either through the indirect (through the auto-
correlation function of the signal and then taking the DFT Of the
result) or the direct (taking the DFT of the original signal and
squaring it) method has some undesirable properties such as a very
high variance, leakages, i.e. contributions to a given frequency
estimate from distant parts of the spectrum. To reduce or minimize
the most disturbing effect we intend also to weight the autocorrela-
tion estimate or the raw data with a suitable function (the
"window") chosen in a proper way.
COMPUTATIONAL CONSIDERATIONS AND THE FAST FOURIER TRANSFORM (FFT)
In order to implement the Fourier techniques, it is convenient

to utilize a computational method known as the Fast Fourier
Transform. These methods allow one to greatly decrease the
computational time necessary to petform Fourier calculations.
These methods usually reguire that the number of points be a
power of 2 (note 256 = 2). They are now standard on almost
all computing systems. The routine we utilized was developed for
the PDPll/40 by Professor F. Bertora (reference 4).
APPLICATION OF THE FFT TO HIGH SPEED CORRELATION
Given 2 finite sequences of the same length N
their linear correlation is defined as:
N-I
Z(m) ~n xl(n) x 2 (m+n), 0 < m < 2N-I
o
Z is sum of lagged products.
Correlation, within the classical Fourier theory, is

easily evaluated in the frequency domain. The same property
is true in the DFT case with gain in speed compared with
direct computation. The product of the suitably computed
DFT's of Xl and x2, one of which must be conjugated, is the
DFT of their linear correlation. It has to be pointed out
that we cannot take simply the DFT's of xI(n) and x2(n),
one of which conjugated, to obtain the DFT of their linear
correlation. The result of this procedure would be, in fact,
the circular and not the linear correlation of Xl and x 2 .
(See Reference 2). DFT = Discrete Fourier Transform.
In order to obtain correct result by still operating in

the frequency domain we must force a periodic correlation to
give a linear correlation. This can be achieved by generating
two new sequences xi(n) and xi(n) of length 2N that coincide
with the original sequences for the first half points and are
zero elsewhere:
xi(n) xi(n) x2(n) o< n < N-l
o N < n < 2N-l
The IDFT of Xi(k) • Xi(k)* (* means conjugate) is the periodic

or circular correlation of xi(n) and xi(n) , i.e. the correlation
of the periodicized sequences Xpl and x~2 of period 2N, i.e. the
linear correlation of xl(n) and x2(n).
ALGORITHM FOR CROSSCORRELATION COMPUTATION
N-l
Z(m)~ E x 2 (m+n), 0 < m < 2N-l
On
1. Perform
xi(n) xl(n) xi(n) x 2 (n) o < n < N-l
" 0 " 0 N < n < 2N-l

3. Take
2N-l
E -j(2n/2N)kn
Xi (k) xi (n) e
on
2N-l
-kn o < k < 2N-l
En xi(n)W 2N
0
N-l
Xi(k) En xi(n)W 2N 0 < k < 2N-l
0
3. Take
X'*(k) (complete conjugate of Xi(k»
2
4. X3 (k) = Xi(k) X'*(k)
2
5. Inverse Transform:
1-- 1 {X3 (k)~ 1

= Z(m) = 2N
2N-l
Ek
0
km
X3 (k)W 2N 0 < m < 2N-l
Figure I - Linear cross correlation between two orthogonal

(A) and parallel (B) central scan lines 10 pp apart, each
averaged 100 times for image III (mitotic cell). Similar results
are found if parallel lines are 30 or 50 pp apart.
SIGNIFICANCE OF CORRELATION AND SEARCH FOR PERIODICITIES OF

CHROMATIN MORPHOLOGY
In using the above results, we have taken two approaches.

The first involves taking scans across either the center or
borders of the nuclei and comparing the Fourier Transform of
the resulting arrays of points (Fig. 8 of the text)
The second involves taking cross correlations of scans

either parallel to each other or perpendicular to each other
(Fig. I). The reason behind this approach is as follows:
given that any two dimensional pattern is obtained from the actual
three dimensional chromatin pattern by "squashing" the nucleus,
any periodicity found should be independent of the orientation of
the cell (and therefore the scanned lines).
Preliminary results indicate that either the linear (Fig. I)
or the circular (not shown) cross correlation yield similar
results either between two parallel (Fig. IB) or orthogonal
(Fig. IA) scan lines. Furthermore, if cells of different
phases have different configurations, we might expect to
observe different correlational patterns between these cells
: : . .• ~....••••.
• I " • _ •
....
" . : ~ •
'.:'\." c
..• -:-. ;." ..".,....~~ .•~:"":.~
,""t-l
. ' - ..'
' " ..," •
. .:"" :'.~' . ,::."".... : .. -;.
.... ;."'" :. ;'t-' '.
Figure II - Circular cross correlation between two parallel lines

10 pp apart (100 times averaged) central scan lines of mitotic (A),
middle Gl (B), and early S (C) He-La nuclear images.
as we do for early S, middle Gl and M phases (Fig. II). Indeed

if the correlational patterns were substantially different within
a given cell, such differences would have thrown doubt on the
cycle phase differences just discussed.
Of course, we should not be surprised that peripheral and central

scans of the same cell will yield different results, but they
should be uniform with respect to cells in the same phase, as
they are in the study so far conducted. For instance, middle-Gl
cells do consistently show periodic spikes along the nuclear
border while mitotic cells do not (Fig. III) .
Figure III - Optical Density - overall display of mitotic (M)

HeLa nuclear image (64 x 64).
REFERENCES
1) A. Papoulis, The Fourier Integral and Its Application,

McGraw Hill, 1962.
2) A. V. Oppenheim & R. W. Shafer, Digital Signal Processing,

Prentice Hall, 1975.
3) G. D. Bergland, A Guided Tour of the Fast Fourier Transform,

IEEE Spectrum, July 1969, Volume 6, #7 p. 41-52.
4) F. Bertora, C. Braccini, G. Gambardella and Musso,

Study on the use of the Fast Fourier Transform in
spectral analysis, ESOC Contract n. 486/73/T,
Instituto di Electtrotecnica Universita di Genova - Italy.
CHROMATIN STUDY IN SITU: II. STATIC AND FLOW MICROFLUORIMETRY
C. Nicolini, S. Parodi, S. Lessin, A. Belmont,

S. Abraham, S. Zeitz and M. Grattarola
Division of Biophysics
Temple University Graduate School
Philadelphia, Pa. 19128
I. INTRODUCTION
A recent review (1,2) indicates that significant correlation
exist between the structure of chromatin and the extent of cell
proliferation (aging, serum stimulation, virus transformation,
cell-cycle phases, etc.). Chromatin, of course, refers to the
diffuse interphase form of chromosome isolated from eukaryotic
cells and the question remains open as to whether observed chroma-
tin changes are reflected in the intact cell.
The techniques described in this report, by studying the dye-

concentration dependence of mean fluorescence per cell, is intend-
ed to detect these differences as fast and as physiologically as
possible, opening new approaches to the utilization of microfluor-
imetry in the characterization of chromatin, in situ.
In recent years, microfluorimetry has become extremely popu-

lar among cell biologists causing a chain reaction of scattered
papers in a variety of related fields. Regardless of these ef-
forts, a review of all pertinent literature of both static and
!tow microfluorimetry reveals little awareness of: a) the optimal
staining conditions (which are frequently presented with slightly
more dignity than a "cooking recipe"); b) the physico-chemical
mechanisms and macromolecular organization, in situ, determining
the spectral emission and fluorescence quantum yield; c) the dif-
ferent binding processes between the various dyes commonly utili-
zed and cell components, as RNA, chromatin-DNA, and protein;
d) the unique spectrofluorimetric properties of each dye when in-
teracting with double or single stranded nucleic acid and pro-
teins, as functions of ionic strength.
293
294 C. NICOLINI ET AL
This review is aimed to address the above points in a

coherent and comprehensive fashion, in order to prove that, undep
ppopep staining aondition, it is possibte to diffepentiatty study
ahPomatin aonfoPmation in situ with dyes such as ethidium bromide
(EB) and acridine orange (AD). Specifically, extending our origi-
nal work with ethidium bromide (3), in this overview we have the
goal of establishing a bPidge between the quantitative wopk done
in sotution with EB (op AO) and potynuateotides, and the aett
system.
II. BIOPHYSICAL-CYTOCHEMICAL CONFIGURATION
A. FLOW SYSTEMS
Cell fluorescence (green, red, and total) and low angle

forward light scatter are measured on a flow microfluorimeter us-
ing an ion laser of various powers (between 35 mW and 4 watt) at
various wavelengths, ranging between the far-ultraviolet up to the
visible. The most commonly utilized. in the study related to
chromatin-DNA, is an argon-ion laser with optimal excitation wave-
length of 488 nm. Usually a sample of fluorescently tagged cells
may be injected into the center of a flowing stream. The stream
after emerging from the nozzle tip, is about 50 microns in diame-
ter. As shown in Fig. 1, as cells pass through the laser beam,
different cells will give different but characteristic signals,
both in fluorescence and in scattering. These differences mani-
fest themselves in variations in the intensity of the light from
each cell, which, in turn, are converted into electrical pulses
of varying amplitudes by the photodetectors. The scatter signal
is detected by observing the illumination in the forward direc-
tion. The scatter signal can be observed over a wide angle (up to
about 10 degrees on either side of the laser beam) or over a nar-
row angle (down to 2 or 3 degrees on either side of the beam).
Wide-angle scattering is comparatively insensitive to the effects
of flow and instrument artifacts. Narrow-angle detection is
dependent primarily upon cell or particle cross-sectional area,
and not upon its refractive index relative to the medium in which
it is immersed, nor its probable lack of optical density - a fea-
ture that is sometimes of importance in the observation of un-
stained biological materials.
The fluorescence characteristics of the cell are determined

simultaneously with the scatter signal. Cells that have been
tagged by means of a suitable fluorescent material can be detected
by means of their emission, when excited by the high-intensity
light from the laser. As shown in Fig. 1, in order to perform
multiparameter analysis with our flow systems (either an Drtho cyto-
fluorograf or a B-D cell sorter) on-line with our PDPll/40 compu-
ter, various electronic modifications to allow multiple fluores-
cent and/or scatter measurement on each cell were made (4). After
CHROMATIN STUDY IN SITU: II 295
r-----
I CYTOFLOORnrET[R
,
I
I
I
I ,
I
,'" , '"
- - -- - - --"
I
L_ '" (' lPS,II - S
I
I
---- , J
I
~-
-' ,
•
I
- - - - "in- -house"
I
LV.!2E~
- - - - ~
Fig. 1. Schematic diagram of our own

automated multiparameter flow cytofluorimeter
configuration (4), from data acquisition (cyto-
fluorimeter) up to display (video).
collection, the data are sorted and stored in a two-dimensional

50x50 array on a Deck Disc for further analysis. To process the
acquired data, various software routines have been written which
allow for easy manipulation of the data (4). This routine in-
cludes the capability of plotting the two-dimensional cytogram of
any specified window on the data as well as the simultaneous plot-
ting of the 2-one dimensional histograms obtained by projecting
the cytogram on each axis. The statistics on each window are
available along wi th the capability of storing projection for the
different windows, to allow the analysis to be performed by our
computer, in real time. Unfixed cell suspensions either from in~_
vivo or in vitro systems were stained with EB or AO at 1-4 x 10~-M
either directly at the final R (pM dye/pM DNA) = 1-5 (3,5) or, in
order to prove the fallicy of certain arguments, after pretreat-
ment with Triton XlOO and chelating agents (6). Occasionally, in
both stainings, cells were incubated with RNAse A (10 units/ml;
Freehold, New Jersey) for 30 minutes at 35 0 C.
296 C. NICOLINI ET AL.
B. STATIC SYSTEMS
Epi-illumination is employed for quantitative fluores-

cence measurements as it reproduced the fluorescence excitation
conditions automatically. The epifluorescence is measured with a
Zeiss Ultraphot microscope, equipped with 510 nm reflector.
Fluorescence-exciting radiation is provided by a 100 watt high
pressure mercury arc lamp, with proper filter to isolate the blue
line (450-485 nm). Color estimate of the nucleus and the cyto-
plasm is carried out by direct comparison of the fluorescence
emission against the color diagrams of the Commission Internation-
al de l'Eclairage. A fluorescence photograph was taken immediate-
ly after each experiment to serve as a permanent record. Our
color assignment was constantly cross-checked, in terms of wave-
length fluorescence emission, against an automated recording
Zeiss Microscope Photometer 03. Quantitative corrected fluores-
cence emission spectra are obtained using interference contrast
optics and blue light for illumination: a cell is selected and
the variable photometric aperture is adjusted to encompass a fixed
area either in the nucleus or the cytoplasm of the individual
cell. The cells grown in vitro are stained either directly on the
coverslips before and after fixation or as smears after detachment
by trypsin or scraping. The cells grown in vivo are stained
either directly after a mechanical-trypsin dissociation procedure
(1) or after subsequent fixation. The mildest fixation procedure
is with 2% glutaraldehyde in Krebs-Ringer Phosphate Buffer (pH 7)
at 37 0 C for 1 hour; which gives completely permeable cells.
Static fluorescence study was conducted in a 5 x 35 x 16 pm cham-
ber as described elsewhere. (15)
III. ETHIDIUM BROMIDE STAINING

Ethidium bromide is a specific intercalating dye for nucleic
acid, both RNA and DNA, which has been extensively studied both in
solution and in situ.
A. IN SOLUTION
The ethidium bromide is important for at least two

reasons: 1) the ethidium bromide molecule can be used as a probe
of the DNA conformation in such structures as the closed circular
DNA of phages and viruses and the chromatin of eukaryotic cells;
and 2) the ethidium bromide interaction with DNA, probably an in-
tercalation between two couples of paired bases, can be considered a
typical model for a certain class of intercalating agents (7).
The ethidium bromide molecule is an interesting probe also for the
structure of RNA molecules. Important steps in the understanding
of the ethidium bromide-DNA interaction are the discovery of a
metachromatic shift in the absorption spectrum near 500 mr (8),
the observed large increase in the fluorescence quantum effi-

ciency of intercalated ethidium bromide (7), the observation that
the induced optical activity near 308 m~ and its direct effect on
the DNA spectrum is simply linearly proportional to the amount of
intercalated dye (9). Scatchard plot analysis (10) of DNA - EB
interaction at low ionic strength and 26 0 C yields an association
constant for the primary sites of 1.3 x 10 6 (moles/l), which
decreased by about a factor 10 at higher ionic strength (0.2M).
EB may intercalate with double stranded RNA, only with slightly
different association constants. secondary sites 3 weakly bound
outside the nucleic acid, may also yield fluorescence, but with a
quantum yield equal or only slightly larger than free dye, while
primary sites do have a fluorescence enhancement of about factor
10, as can be easily verified by spectrofluorimetric studies (7).
B. IN SITU
In order to discriminate between the binding process

(primary or secondary) of either DNA or RNA, we evaluate the
mean fluorescence per cell as a function of added dye. This could
allow us to determine the amount of bound and free dye at any
concentration of ethidium bromide. By analogy with the Scatchard
plot analysis (10) we can then determine the association constant
and the number of primary binding sites in the intact cell.
Fig. 2 shows the mean fluorescence per cell as a function of the
ratio R of added dye per unit of DNA from WI-38 human fibroblasts
confluent (Go)' mature duck erythrocytes (Go)' and HeLa S3 cells
in Gl and G2 phases. These curves show that, at saturation, the
mean fluorescence per cell reaches a plateau value which is di-
rectly related to the amount of DNA per cell; 7.0 pg for WI-38
human fibroblasts, 10.3 pg for WI-38 human fibroblasts, 10.3 pg
for Gl HeLa cells, and 20.6 pg for G2 HeLa cells. The exception
is constituted by mature duck erythrocytes (2.65 pg/cell) which
do not have any sizable amount of RNA: (a) the saturation occurs
at a lower R (0.5-0.6), and (b) the ratio of the mean fluores-
cence per WI-38/erythrocytes (or HeLa/erythrocytes) is, at
saturation (R>3.0), larger than what would be expected from a
pure DNA content ratio. In all other cell lines, the ratio of
observed mean fluorescence, at saturation, corresponds closely to
the expected value from pure DNA content per cell, perhaps because
in these examples the increase in RNA closely parallels the in-
crease in DNA content per cell. Under our experimental condi-
tions, the chromosomes maintain their conformation intact, such
that mainly at low R «1.0), a variation in conformation could be
reflected in variation of fluorescence per cell. This becomes
apparent, when by analogy with Scat chard plot analysis, we eval-
uate the association constant between the dye and the intact cell,
and the relative number of binding sites available to the dye in
the intact cell (both primary and secondary with DNA and RNA).
Utilizing the observed intensity of mean fluorescence at very low
dye concentrations (R<0.20) where all ethidium bromide added can
----.".--------------~_.~~,~I--------- WI-lifo,) •
•
~ _ _ _._..- - - _ _ . - - - - - e IErythrocyt.
2 3 6
R focW.d d,JDNA I
6 '\\
\
\
\
\
, \
\HG2,,
'0 ,\
,,
\
\
\,
.,
\
\
\
\
,,
\
\ \
4,, \
o\
.
\
\
\
0...,
' .... ,
-
\
\- "
0.1 0,2 \
0,3 O.S
r (BOlN> C1I'E/CEU ONA)
Fig. Z. (above) Mean fluorescence/cell as a function of R

(added dye/DNA) from duck erythrocytes (e --- e) and WI-38 human
fibroblasts confluent (.___.). HeLa cells in GI (X---X) and GZ
phase (0 --- 0) . The assignment of cell-cycle phase for HeLa
cells was obtained from the frequency distribution of fluores-
cence/cell from a log phase growing in suspension. The so-called
"GZ" cell population also presumably contains M-phase cells.
(below) r/CF vs r (bound dye/DNA) for the binding
process between the intact living cell and the ethidium bromide
from mature duck erythrocytes (ME,4t---e), WI-38 confluent
(GO,D ---D), HeLa in GI phase (HGI, ~---~) and GZ phase
(HGZ , 0 --- 0) .
CHROMATIN STUDY IN SITU; " 299
be considered bound, we obtained for each given cell a function

which gives the expected value (E) for the mean fluorescence per
cell versus the pmols of added dye/umols of DNA (in a given cell),
if all dyes added should bind to the nucleic acid. These func-
tions are:
EE 100 R for duck mature erythrocytes

EW = 240 R for WI-38 human fibroblasts
EHGI 331 R for HeLa cells in Gl phase
EHG2 = 652 R for HeLa cells in G2 phase
The ratio (r) of ~mols of bound dye/pmols of DNA (in a given cell)
is obtained by
r = R (OB/E)
where OB'and E are respectively the observed and expected mean

fluorescence per cell at a given R (added dye/DNA). The amount
Uwmols) of free dye (CF) per DNA, normalized to 100 pmols of DNA.
for a given cell population is then given by
(E-OB) x 100 R
E
The lower panel of Fig. 2 shows the plot of r/C F vs r(bound dye/
DNA), computes as previously indicated for the same cell lines.
From this, it appears (see Table I) that mature duck erythroaytes
and aonfZuent WI-3B human fibrobZasts have approximateZy the same
number of primary binding sites avaiZabZe (even if the amount of
DNA per aeZZ differs by a faator of 2.6~ whiZe the number of
primary binding sites for HeLa Gl or WI-3B Gl (3~5) and HeLa G2
aeZZs aPe aZso very simiZaP (regaPdZess of the tUJo-faator differ-
enae in DNA aontent)~ but aPe substantiaZZy inereased with respeat
to the WI-3B (Go) and mature duak erythroaytes. The relative
association constant K can be obtained (see Table I) by
I) r =K (n-r)
CF
where (n) is the abscissa intercept and represents the number of

binding sites (in the intact cell) per DNA phosphate, (r) is the
ratio of bound dye per DNA phosphate, and CF is the concentration
of free dye in umols. The 'primary sites" assoaiation aonstant
for EB is of the same order of magnitude (3.2 x 105)aomputed (see
following pages) for AO primary sites in situ (2.6xZ05)~ and in
perfeat agreement with EB assoaiation aonstants in soZution~ at
0.2 M NaCl. Parallel studies, conducted with RNA-ase treated
cells, seems to indicate that the primary sites aPe reZated to
an aZteration in ahPomatin DNA aonfo~ation~ as during GO-Gl
transition of several in vitro (3,5,11) and in vivo (12) systems
TABLE I
Association constants (i-mol- I ) and numbers of primary sites for
chromatin-DNA IN SITU, for ethidium bromide primary (KP) and
secondary (KS) binding sites. Data were obtained from a plot of
r/Cf versus r obtained by staining unfixed cells with increasing
concentration of E.B.
n' nil
0.38 0.47 3.2xl0 5 6.8xl0 4
Mature
Duck 0.25 5.7xl0 5
Erythrocytes
WI-38 GO 0.27 0.44 4xl0 5 7xl04

WI-38 Gl 0.37 0.43 3.0xl0 5 6.3xl0 4
ReLa G2 0.40 0.48 3.3xl0 5 6.5xl0 4
TABLE II
Time (T) necessary for EB to enter 100% of CRO cells versus abso-
lute cell concentration expressed in M DNA. CRO cells were stain-
ed with EB=10-3M directly on the coverslip. Cell viability was
monitored by 0.2% Trypan Blue exclusion, after EB entered all
cells.
T Percentage of
DNA (Minutes) Viable Cell
4.2xlO- 5 2 90%
3.2xlO- 6 3 90%
2.llxlO- 6 5 50%
1.05xlO- 6 3 90%
1.0x10- 6 4 80%
5.3xI0- 7 6 80%
5.3xI0- 7 5 80%
The orange-red fluorescence intensity is weak at very early time

(1-2 minutes), constantly increasing until reaching a plateau
level around 8-10 minutes corresponding to the emission obtain-
able after fixation. During the same lag period, Trypan Blue
uptake does not change appreciably. The orange emission remains
unchanged between 10 and 60 minutes when most cells loose viabil-
ity (Trypan Blue, TB, enters). The same cell, which do uptake TB
instantaneously, do also uptake EB instantaneously and intensely.
A B D
23
1 • •
12
•
10
•
•
~
cZs
~
o~ 6
<i.
•
~
~ INA .. ONA ••
01: ..
04---------~~--~------~~--~~--_r--------~~----
10-7 4xlO- 6 10 -~ 2xlO-~ 10-·
A.O. TOTAL CONCENTRATION (M )
C
Fig. Sa. WE-38 cell, stained with Ethidium Bromide at R=4 and
10-5M absolute dye concentration. As evident, cell maintains its
original morphology and viability (as determined by Trypan-blue
exclusion).
Fig. 5b. Static fluorescence emission of Melanoma B16 tumor cells,
stained with Acridine Orange at 2.5 x 10-5M and R=0.16 (below);
R=4 (middle); R=lO (above).
Fig. 5c. Phase diagram of color emission for nuclei and cytoplasm
from the same melanoma cells, stained with Acridine Orange at var-
ious AO total concentration and various molar ratios (~M AO/ ~M DNA).
Fig. 5d. Effect of DNA-ase (center) and RNA-ase (above) digestions
on the same melanoma cells, stained at R=4 and AO-2.S x 10-SM (below,
for untreated cell). Cells were unfixed and immediately AO stained.
DNA digestion was conducted for 1 hour at 37°C, with 3160 U/ml of
DNA-ase, RNA digestion for 1 hour at 37°C using 7000 U/ml of RNA-ase.
CHROMATIN STUDY IN SITU; II 301
(see Fig. 3), and compatible with findings on chromatin isolated

from the same cell lines (2) (see also chapter by C. Nicolini on
chromatin structure, pp. 613). Static fluorescence studies (see
Fig. 4) indicate that the EB fluorescence is mostly confined in
the nuclei (both at R = 0.16 and R = 4) while the cytoplasm is
weak (at any R)which may be due to the low number of primary
sites in cytoplasmic RNA and/or to different quantum yield arising
when EB intercalates different base pairs or base sequence (13).
C. CELL VIABILITY AND EB UPTAKE
Frequently cell death has been correlated with EB

uptake. We now address such correlations, even if cell death
(eventually occurring in unfixed cells during the procedure) is
not at issue here, since it does not alter DNA organization and
subsequent fluorescence emission, within the short period of time
between cell staining and analysis. A careful time sequence on
CHD line in vitro cells, directly stained on their coverslip with
EB at saturation, show that within 2-6 minutes (depending on cell
concentration) all viable cells exhibit clear orange fluorescent
nuclei while preserving their morphology (see Fig. Sa, Table II).
Furthermore, when stained at R = 4 and absolute EB concentration
of 4 x 10- sM, not only do viable cells pick up the dye, but they
(both CHD and M3 cells) maintain their capability to grow after
quick dye removal (Table III).
IV. ACRIDINE ORANGE STAINING

Acridine orange (AD) is widely used for flow microf1uori-
metric (FMF) studies and for static fluorescence (SF) studies of
the intact cells. However, until recently (15) we were lacking a
rigorous approach to the AD staining, where, even if we are deal-
ing with cells and not with nuc1eotides in solution, we have a
comparably good control and understanding of the entire dye-cell
interaction and subsequent fluorescence emission. The basic
approaches to AD staining in SF are essentially of two types:
overstaining-differentiation and equilibrium staining methods.
A typical example of the first type is the widely used method of
Rigler (16), where the cells are overstained with 10- 4M AD and
then differentiated in a citric-acid-phosphate buffer. The over-
staining-differentiation approach is essentially the classical
histological approach. Other authors have used in SF equi1ibrium-
staining methods (17) which are indeed the only one possible for
FMF studies. In the latter cases, both for FMF and SF, we have
found in the literature either empirical protocols (12,13,15) or
quantitative but misleading approaches (6,8). For this reason,
we have explored (15) a large range of R (total dye) ratios and
(total DNA)
final AD concentrations, and found that the results in situ even
if more approximate, are essentially in agreement with the studies
in solution, both in SF and FMF studies.
Fig. 3. Green (or total) fluorescence emission of

various cell lines, stained with EB at lO-5M and R=O.75 during
GO-Gl transition in vitro (WI-38 serum-stimulated; AF-8
temperature sensitive mutants at the permissive temper-
ature, 33 0 C and non-permissive temperature, 39 0 C, at 23
hours after stimulation) and in vivo (rat liver after
partial hepatectomy). In all three systems "GO and GI"
cell population have the same amount of DNA, and DNA
synthesis, i.e., 3H-thymadine incorporation begins
several hours later.
B 16 STATIC FLUORESENCE
R=4
IE 81 =10-5
-Nucleus
17 • ···C y toplasm
15
13
11
r;t
1liJ9
."
"i
>= 7
-
0
E
:;)
c
0
:;) 5 I \
I \
•
~
3 \
'.,
O~~~~~--~~~--~~--~~~~
480 600 720

Wavelength nm.
Fig. 4. "Static" Spectral- fluorescence emission

of the same WI-38 cell shown in Fig. 4a, stained
with EB EB=IO-5M and R=4,as determined by a Zeiss
microfluorimeter.
TABLE III
CHO cells were incubated at EB = 4 x lO-5M and R = 4, then 10
minutes later (with 100% of cells showing orange emission but
0% Trypan blue uptake) the EB was removed by 2 consecutive washes.
The same cells (who received EB treatment) were then incubated
at 37°C for 3 days. A control of untreated CHO cell was monitored
at the same time.
Number of Cell
Number of Cell Plated 3 Days After Staining
EB treated 45,000 322,000
Control 45,000 315,000
TABLE IV
Binding constants (~.mol-l)for AO-DNA and AO-RNA interaction in
solution (at high ionic strength, i.e., 0.2M) and in situ. Pri-
mary sites refer to the green emission, while secondary sites re-
fer to dimer and higher aggregate formation outside nucleic acids.
AO-DNA AO-RNA
Primary Secondary Primary Secondary
SOLUTION
(at high ionic 1. 3xl05 0.8xl0 4 1.3xl0 5

strength
IN SITU 2.6xl0 5 0.72xl0 4 2.6xl0 5 0.43xl0 5

A. AO-NUCLEIC ACID INTERACTION IN SOLUTION

From the experimental work done with AO-DNA, the associ-
ation constants K1 for primary sites and K2 (secondary sites) in
solution can be computed using a model proposed by Armstrong,
et. a1. (18), with
2 (Sl - S2) (1-2S )

1
K1 = ----~--~------=-
eM (1-4S1)2
where ~, Sl and S2 are respectively the concentration of free dye,
fraction of available primary and secondary sites occupied per unit
DNA (or RNA). As summarized by Table IV, it appears that: 1) at
very low primary sites, i.e., S<O.l, we have essentially monomers
of AO intercalated between base pairs, the quantum yield of the
540 nm fluorescence being approximately 3 times the quantum yield
of the free AO monomers (19) and at high ionic strength (0.2M) the
association constant of this monomer for native DNA is of the
order of 1.3xl0 5 (18). 2) In the range of R ; 0.1 - 0.3, another
molecular species appears, tentatively a dimer. The formation of
the dimer induces a red shift, with a progressive decrease in the
quantum yield of the 540 nm fluorescence and probably the dimer is
practically unf1uorescent at 540 nm. At high ionic strength
(0.2M) the association constant for these second binding is of the
order of 0.8 x 10 4 (18). For R>0.3 higher orders of aggregation
can be found (see in situ studies) responsible for the red fluo-
rescence at 640-670 nm. The studies in solution at R>0.3 are,
however, made difficult by the tendency of the complex to pre-
cipitate (18).
B. STATIC MICROFLUORIMETRY
In order to evaluate the AO stainability properties of
nucleus and cytoplasm in presence or absence of cell pretreat-
ments, we explore a large range of R (pMAO!pMDNA-P) from 0.05 to
100 with special attention to the range 0.1-10. This range of R
was explored, by varying also the final AO concentrations between
10-7M to 10-3M. Fig. 5b shows a typical result obtained in melan-
oma B16 tumor cells at final AO concentration of 2.5x10- 5M, by
varying the molar ratios R: the nucleus and the cytoplasm are
both green at R = 0.16, respectively green and reddish orange at
R = 4, and respectively yellow-orange and red at R = 10. These
data are confirmed by the quantitative spectral analysis of these
cells' fluorescence by means of static microf1uorophotometry in
terms of actual emission wavelength (Fig. 6). Similar dependence
upon molar ratios R(ranging between 0.1 and 23) are found in a
wide range of total AO concentrations (10- 4 - 10- 7M), as shown in
Fig.5c where a color code (see Fig. 6 for corresponding emission
wavelength) is utilized to visualize in a comprehensive and com-
pact fashion the differential staining of nuclei and cytoplasm.
8-16 STATIC FLUORESCENCE

"'0 -Nucleus
CU 7
---Cytop I asm
> 5
E
...c::
:J 3
~ 0
a ~~++++~~----
480 SSS 630 70S
11
Wavelength nm. 11
-s
.
1=10.0.
-.
550
1=4.0 ~.o] =2.4X 10
~o.] =2.4Xl0-S
-
• 9 ~ 9
:J III
III
7
"'0
"'0 CU
CU >
;:5 E
...c::
5 \
:J \
...c::
E
:J
3 ~ 3
I
\
\
,,
III
:J
a
a
'., ,
0-+----- ---- o ,
711 612 744 822
Wavelength nm. Wavelength nm.
Fig. 6. Static spectraZ fluorescence emission for

the same cell, as Fig. 5, at R=0.16, 4 and 10, as
determined by a Zeiss micro fluorimeter (Acridine
Orange staining)
CHROMATIN STUDY IN SITU: " 307
In these experiments we used amelanotic melanoma B16 cells disso-

ciated by trypsinization and then fixed with 2% glutaraldehyde:
identical results have been obtained in "unfixed" cells, with the
exception that the cytoplasm stains slightly more intensely and
homogeneously after fixation. In order to determine what macro-
molecular species are predominant in the nucleus and the cytoplasm
and how they are related to differential spectral emission, we
have carried out RNAse and DNAse digestion in the same "unfixed"
B16 cells, at AO concentration of 2.5 x 10-5M and R = 4: under
these conditions, while the nuclear green fluorescence is selec-
tively affected (80-90% reduction in intensity) by DNAse~ the
cytoplasmic reddish-orange fluorescence is drastically reduced by
RNAse digestion (see Fig. 5d). These studies, confirmed in other
cell lines (as CHO and M3), gave identical results whether with or
without cell pretreatment by Triton XIOO and chelating agents at
various pH (6). It is apparent that the differential spectral
emission~ i.e., nuclear DNA versus cytoplasmic RNA~ strongly
depends from the dye/DNA ratio contrary to previous statements
(6). Recent optical studies (18) of AO-DNA interaction in solu-
tion show, indeed, that denatured and native DNA have the same
association constant and if RNA was selectively denatured should
yield red cytoplasm even at R = 0.16. All these findings indicate
that the claim (12) for differential RNA and DNA denaturation by
chelating agents are quite erroneous, in agreement with the follow-
ing FMF studies conducted in parallel (Fig. 9-12). Our results
indicate that at high ionic strength and AO total concentration in
the range 5 x 10- 7 - 10-6M, even at high R both nucleus and cyto-
plasm are green. This suggests that the association constants of
the dimer and higher aggregation sites are too low to generate
significant secondary binding (longer wavelength emission) even in
the cytoplasm. Actually, the nucleus becomes yellow (i.e., appear-
ance of secondary sites) only at R = 10 and total AO equal
2.5 x 10-~, remaining green even at 1 x 10-5M; this yields (see
Table V) an association constant for chromatin red fluorescence in
the order of 0.72 x 10 4 in striking agreement with previous deter-
mination of DNA secondary sites in solutions (18). In the cyto-
plasm, however, the red fluorescence begins to appear for R around
4, at AO concentration about 10 times lower (4 x 10-5M), suggest-
ing an affinity 10 times larger (0.5 x 105). Above concentra-
tions of 4 x 10-5M even the nucleus can become red, at high R.
Table V summarizes the relevant parameters and computed associa-
tions constants for the primary and secondary sites of nuclear DNA
and cytoplasmic RNA in situ. Our static fluorescence data (Table
V and Fig.5c) indicate that the association constants for chroma-
tin-DNA green fluorescence is 2.6 x 105, while for the chromatin-
DNA red fluorescence (secondary sites) and for the dimer formation
in solution, are identical (0.8 x 10 4 ), suggesting that we may
assume the same association constant for all secondary sites
(dimer or higher aggregate). In summary, the fluorescence data in
TABLE V
Stumnary values of fractions of AO primary (131), secondary sites
(132), for the molar ratio of total dye/DNA (R) and absolute dye
concentration (AO) , at which green or red fluorescence originally
arise in nuclear DNA or cytoplasmic RNA (see also Fig. 3).
NUCLEAR DNA CYTOPLASMIC RNA

Green Red Green Red
(AO (M) 10- 7 2.SxlO- S 10- 7 4xlO- 6
R 0.1 10 0.11 11
13 1 0.01 0.21 0.01 .1S
132 0.00 0.031 0.00 0.022
~(M) .8xlO- 7 2.4xlO- S .8xlO- 7 3.80xlO- 6
Kl 2.6xlO S 2.6xlO S
K2 0.72xl0 4 0.43110 S
Association constants (t.mol- l ) are computed for each binding

process according to Kl = 2 (13 1-13 2 ) (1-2131); K2 = 13 2
~ (1-4 131)2 Cm (13 1-13 2 )
where free AO concentration Cm can be directly computed by
subtracting the bound from the total dyes, i.e.,
~ = AO~- (131 S2)DNA _ (131 S2)RNAl

~ RDNA 1.IRDNA:J
and for primary sites, 13 1= .01 corresponds to appearance of pri-
mary sites (green flourescence); 13 2 = .lSa1 corresponds to
appearance of secondary sites (yellow) t17hen '13'1= 0.21 , i.e.,
saturation primary sites.
Note We calculated K2 as follows: We assumed % red fluo-

rescence 3.333132 which gives yellow color. Then from
=131+3 . 333 132= .2S
Kl we compute 13 1 and therefore e2=0.lSSl. Using those values, we
compute K2 •
solution and in situ are in striking agreement (Table IV). We

may then assume for the AO staining of intact cell that both
double stranded DNA (B form) and RNA (A form) possess primary
(green) and secondary (red, related to multimer AO formation)
binding sites, with different association constants (see Table IV):
at Zow R, onZy p~imary sites with both DNA and RNA ~e oaaupied;
at inte~ediate R, onZy RNA exhibits seaondary site binding; at
high R, DNA aZso saturates its p~im~y sites and exhibits seaondary
sites.
To obtain an independent estimate of the association con-
stant, at high ionic strength, related to green nuclear DNA fluo-
rescence, we utilized data obtained by others (19) in mouse leuko-
cytes (see also ref. 15). Our computations tend to suggest that
at low S and high ionic strength, the association constant for AO
primary sites with DNA in situ is about 1.4 - 1.8 x 10 5M in agree-
ment with our own data (2.6 x 105M): this number is furthermore
of the same magnitude reported in solution. It is curious to note
that the same author (17) obtained the same number, throughout a
series of compensating errors; for instance, West computed the
molarity of bound AO by assuming (17) a nuclear volume so small
(5 nm 3 ) that it would give an intracellular DNA concentration of
4.2 M (!?).
C. PHYSICO-CHEMICAL MODEL FOR AO-BINDING IN SITU

This qualitative description of AO binding can be
formulated in a more quantitative framework using a model proposed
by Armstrong, et. a1. (18). Assuming two types of binding proces-
ses, p~im~y consisting of dye intercalation and secondary con-
sisting of formation of a dimer by the binding of an external dye
molecule with the intercalated molecule, and assuming that the
binding process does not change the electrostatic potential of the
nucleic acid polymer, Armstrong, et. a1., arrive at the following
relationships:
2 (Sl-S2) (1-2S i )
Kl
~ (1-4S l)2
S2
K2
~ (Sl-S2)
where Si' S2 and ~ are fractions of available primary and second-

ary sites per unit DNA, and free AO concentration. These two equa-
tions, in addition to the requirement (AO)T = (AO)bound + eM'
allow for the solution of 81 and 82 as function of AO T and R, given

constants Kl and K2 .
In the case of the intact cell, assuming that for every pM

of DNA, we have 1.1 RNA, for the nucleus (n)
(I)
(II)
8 cyt + 8 cyt (III)

AO .AO+ 1 2 .AO
l.lxR
where 8 1cyt and 8 2cyt refer to the primary and secondary binding
in the cytoplasm, computed as (I-II). Using these equations we
arrived at the following values for e (monomer) = 81 - 82 and
8 (dimer) = (2 x 82 ), shown in Fig.7.
We can now relate these numbers to the percentage of red

fluorescence, under the premise that the monomer emits in the
green and the dimer in the red, with their relative quantum yield
(actually quantum yield X absorption). Actually in the real
situation, the species responsible for red fluorescence is most
likely a large aggregate while in the dimer, fluorescence is prob-
ably quenched. The fact that the equilibrium conditions for dimer
formation appear to be identical to those for appearance of red
fluorescence is probably a reflection of the critical step in
aggregate formation, being dimer formation. If we incorporate
this change in our model by assuming 80 = 8AG and including this
change in our expression for eM as a function of AOT, we obtain
the results shown in Fig. 8, for aggregate sizes of 20 (which
yields the optimal fit to the experimental data). Of course, what
we have presented above a "yes or no" phenomenon(dimers -+ aggre-
gate, green fl. -+ red fl.) seems to be really a oontinuous process
(as indicated by solution studies-our own data), whereby there is
a progression in aggregate size accompanied by progressions in
quantum yield and fluorescence emission spectra. Thus taking
these factors into account, our present treatment, although prim-
itive, does yield the basic description of the binding phenomena
and supports our present understanding of the equilibrium binding
process applied to intact cells. In the model, we have assumed
A.O. Dimer Formatio.n
•• nu..c1eUI
o-cpoplalm
0.5
5 10
.
II
E
o
~
5 10
5 10
Fig. 7. Acridine Orange dimer formation

according with the model described in the text,
as function of molar ratios R, at 3 different
AO final concentration.
Static Fluoresence Emission of Nuclei C-L and CytoplasmC- --)

versus Molar Ratio!R) for Di"e~t Absolute ConcenJ,rations of
Acridine Orange: TheoreticalR vs Experi mental(~
__ ,.. _ "'0-
,.,.-
- ~-----
.IO()..&
.._ ..._ -_ _ _ _ _ _ _ ~y ______" ~~: 4 X 10- 6 ",
+---~--------~--------~----~~,
2 6 10 23
z
Q
VI , .. - 0- - - "b - - - - - - 0- - - - - -0-
VI
'0
~
~
I
l
I x
~
V I
Z I x
~
V I
~.9: 2xI0-5",
VI 00
~ x
III:
o +----r-------,-------~~----~~
....u..
;:)
2 6 10 23
,'---0--- - (j - - - _
_ - - -0 - -~
~
--T'"-~x
I
~
I x
x ~~ :lxl0- 4
",
+----r-~~--,---~--~~--~~~
o 2 4 6 8 10 12 23
R (MmA.O·/MmDNA)
Fig. 8. Experimental color fluorescence emission

of AO-stained melanoma cells, for nuclei (X) and
cytoplasm (0) as function of molar ratio, at 3 dif-
ferent final AO concentration. The dotted (cytoplasm)
and continuous (nucleus) lines represent the theo-
retical prediction of the physico-chemical model
outlined in the text.
that "primaI'Y sites" (bound AO monomer) yield an enhancement (19)

of quantum yield of 3 with respect to free AO monomer, with the
"seconda::t>y sites" (bound mu1timers) yielding 10 (see Table V).
D. FLOW MICROFLUORIMETRY
Frequently in cell biology, the cell populations to be

characterized are quite heterogeneous, which when combined with
the macromolecular species' heterogeneity (as previously shown)
require automated muZtipaI'ameter data acquisition for a meaning-
ful characterization. The heterogeneity of our melanoma cell
suspensions can be made apparent, using a compressed logarithmic
scale for the scatter, by the computer two-parameter histogram of
scatter vs. fluorescence shown in Fig. 9. Cells were directly
stained with acridine orange at R = 4 with or without RNAse treat-
ment. This compressed scatter scale allows simultaneous identifi-
cation of debris (low scatter and low fluorescence); necrotic
melanoma cells (low fluorescence and intermediate scatter); lym-
phocytes (low scatter and intermediate fluorescence), and viable
melanoma cells (large scatter and various levels of fluorescence
intensity). This assignment to various cell types has been
achieved by analyzing the morphology of the Wright-stained
"sorted" cell sub-populations (using our B-D cell sorter) (15).
As shown also for cells pretreated with Triton-X100 and che1ating
agents (15), after direct staining with acridine orange, the green
fluorescence distribution of cells of any size is unaffected (or
slightly affected) by the RNAse treatment. On the contrary, RNAse
treatments do cause a four-fold decrease on the mean red fluores-
cence of melanoma cells (largest scatter). We should remember
that the gain on our photomultiplier was greater for the red than
for the green signal since we only intend to determine relative
effect of RNAse on the two spectral emissions. Moreover, Zet us
cZaI'ify a semantic probZem: due to the photocathodes-filters
utilized in the cytof1uorograf the actual spectral response of the
FMF configurations are quite broad, both in the so-called "green/l
(between 515 and 585 nm) and "red" (between 520 and 750 nm):
i.e., we may have a Gaussian orange emission (as for ethidium
bromide) centered around 600 nm, being detected both as a "green"
and "red" in FMF Zanguage (Fig. 10) (1,3,5,6). Considering the
spectral emission of single and double stranded nucleic acid, at
R = 0.1 - 1.0 molar ratios most signals are detected in the "green"
FMF region (15), but even in the presence of mostly DNA primary
sites, a signal properly magnified may be detectable by the "red"
FMF configuration (Fig. 10). At AO concentration 1-3 x 10- 5M and
R = 1-10, while the green emission cZearZy refZects onZy vaia-
tions in chromatin-DNA primaI'y dinding sites with AO in situ the
red emission paI'tiaZZy relates to the amount of RNA present in the
intact cell (either as single or double strand in the A form) but
reflects aZso vaI'iations in seconda::t>y weak-binding sites with DNA
(at higher R) and in the formation of AO mu1timers bound to acid
BEFORE RNAse
AFTER RNAse
BEFORE RNAse AFTER RNAse
Fig. 9. Computer-drawn two parameters hystograms

of scatter versus red and scatter versus green fluores-
cence, for melanoma B-16 cells, stained with AO at R=4
and final AD concentration 2.SxIO- SM. To display the
effect of selective removal of RNA, the same cytograms
are shown before and after RNase digestion at 37 0 C for
I hour .
EMISSION
RED
"'. (X50)
\"
490 600 700 790
RED
(X500)
490 600 ',' 00 790
Fig.lO. Spectral emission (----) of DNA-bound

ethidium bromide (above) and Acridine Orange (below)
both determined by a Spectrofluorimeter at molar
ratio R=l and dye final concentrations la-SM. The
dotted line represent the quantum efficiency for the
emitted light, of the "red" and "green" standard
cytofluorograph configuration, as determined by the
filters and photocathodes utilized.
316 c. NICOLINI ET AL.
macromolecules as polysaccharides, and is consequently strongly

dependent upon the dye concentration. If we now trigger, using
our simultaneous three parameter acquisition, only on large
scatter, we may obtain the red versus green distribution only for
melanoma B16 cells (15). If we now subdivide the "pretreated"
melanoma cell population into subpopulations, in terms of their
red fluorescence (Fig.ll ), we found that the cells with Gl fluo-
rescence or larger seems to have green fluorescence distribution
typical of a log phase population with the presence of a small
peak (about 6% of the total) with green j1uorescence lower than
Gl and a red j1uorescence equal or larger than G2 cells. Sorting
this cell population by means of our fluorescence-activated cell
sorter appears to confirm this assignment as mitotic cells in
agreement with the recent findings on PHA-stimulated lympho-
cytes (6). The most striking example, which confirm both the
existence of a differential emission for RNA versus DNA at the
proper AO molar ratio and our previous EB data on the GO-Gl
transition of WI-DB fibroblast, is shown in Fig. 12. WI-38
human diploid fibroblasts have been grown up to 23 days into
"deep" confluency with weekly changes of the medium. Two days
after plating, WI-38 cells show a log-phase distribution for the
green fluorescence with a population of cells having the same
scatter (respect to Gl) but quite lower chromatin primary sites
(green) and lower amount of RNA (red): This subpopulation, likely
relating to non-cycling cells (that, after plating, even if
adherent to the plastic, did not start growing), drastically
reduces at 5 days (5%) and then progressively increases at 12
days (~30%) and 23 days (up to 95%). This reduced lack of pro-
liferation is apparent also by the red vs green cytograms, where
the red fluorescence (RNA), under the same staining and instru-
ment conditions, progressively decreases with time after plat-
ing, going below the minimum threshold at 23 days. It is comfort-
ing that at 2 days G2 cells with twice as much DNA (green) show
also twice as much RNA (red) in respect to G1. At the same time, the
green fluorescence distribution show the constant presence of a
Gl peak (around channel 20) progressively decreasing in amplitude
at the expense of a "quite lower fluorescence GO peak (con-
stantly around #5). These findings are compatable with previous
observation, using EB, that the transition between a proliferating
Gl and non-proliferating GO cells is not a continuum, but rather a
quantum jump (5). An intermediate chromatin AO-uptake, appears at
12 hours, where WI-38 cells already reached confluency: (This
data, to be yet confirmed by more observations.) This could refer
to a chromatin difference between a readily reversible (GO) and a
deeper GO (or Q) non-cycling cells. Occasionally most WI-38 are
in the "so-called" deep GO, already at 12 days, under similar
nutritional conditions. We have to stress, indeed, that a large
variability does exist in WI-38 cells stimulated to proliferate
(regardless of previous "superficial" optimistic statements), that
only proper automated multiparamenter analysis of cells stained
(A)
(B)
Fig . 11. Green fluorescence distribution of melanoma

cells with small (A) and large (B) red fluorescence.
Cells are stained with AD at R=4.
Scatter vs Green Fluorescence

WI38 Fibrabla.t
at Varlou. Interval. After Plating
Red VI Green Fluore.cence
2 Day.
5 Day.
12Dayl
23 Days
Fig. 12. Two parameter hystograms of WI-38 at 2, 5, 12 and 23 days

after plating (Phase II, grown in parallel) . Cells were stained
with AO = 3xlO- 5M (final) and R = 4-5. All cytograms were obtained
with green PMT = 4.6 and red PMT = 4.8 Scatter=Medium in a cyto-
fluorograf on line with PDPll/40. Cytogram A was obtained when
channel 3 on the scatter axis was isolated from the 2 days after
plating cytogram. The medium was changed weekly - 95-98% of the
WI-38 cells are viable, as shown by Trypan blue exclusion.
remembering the mass action-Law~ may quantitatively monitor.
Sorting, by means of a fluorescence activated cell sorter,

of the GO and Gl fluorescence peak either in vitro or in vivo
(Fig. 13), shows that the two cell subpopulations are indistin-
guishable by light microscopy, compatible with the fact that the
larger dye uptake relates to molecular alterations in the tertiary-
quaternary chromatin structure (and RNA synthesis). In this
respect is significant a recent comment of the Editors of an interna-
tional Journal to the effect that "we cannot understand the authors'
cLain that they can distinguish proLiferating and non-proLiferating
ceUs by machine in the face of nothing distinguishabLe by eye."
This reflects a common empirical attitude among life scientists,
where an automated image analyzer or flow microfluorometer
("machine") is approached with a suspicion typical of a fifteenth
century inhabitant of a South Pacific island approaching an auto-
mobile.
We should remember that experiments based on subjective

hands and/or eyes left most problems of life science yet unresolved
(actually the literature is "floated" with controversial empiri-
cal observation); while hard physical scientists were able to
split the atom (without ever seeing a proton or a neutron with
their eyes, being less than 10-1 4 cm), produce electricity and
the most sophisticated electronic devices, as computer without
ever seeing an electron), predict the presence and exact location
of a planet by a series of equations (without ever dreaming of
one).
v. CONCLUSION
In conclusion, AO allows us to selectively discriminate
between chromatin-DNA and cytoplasmic RNA but only at proper R
and absolute AO concentration and without need of any cell pre-
treatment. This is based on the differentiaL spectraL emission
of AO primary (green) and secondary (red) binding sites which are
on the contrary onLy quantitativeLy (quantum yieLd) but not qual-
itativeLy (same orange emission) different for EB. However, we
have to stress that both for AO and EB the same mechanism is
mostly responsibLe for the Larger dye uptake in Gl with respect
to GO (or Q) cells: namely, the aZteration in chromatin structure
i.e., its two-order superhelical configuration as originally and
frequently proved by our laboratory (see chapter by C. Nicolini
on Chromatin, pp. 613). The same conformational changes during
the cell cycle originally monitored in isolated chromatin from
M, Gl and S phases (2) are responsible for the reported decreased
primary binding sites for metaphase ceLLs. These studies, both
by static and flow microfluorometry gave identicaZ results,
either before ("unfixed") or after fixation with glutaraldehyde
(8), either with or without pretreatment by Triton-XlOO and
Go+q
Fig. 13. Light microphotograph of Wright stained

B-16 melanoma cells. GO + Q cells (A) and G1 cells
(B) were sorted from GO + Q and Gl peaks of green
fluorescence vs. scatter histogram (bottom) obtained
from B-D FACSII cell sorter.
chelating agents, at various pH. Investigators should pay more

attention to the "mass-action l,auJ" instead of randomly exploring
all possible combinations of chemical environment and pH. Unfor-
tunatel,y~ the mass-action l,auJ and other basic physico-chemical,
Za:uJs are usual,l,y ignored by most investigators working in either
cl,assical, or automated cytol,ogy~ who use intact cel,l, staining
with sUghtl,y more dignity than a "cooking" recipe. For what
concerns the static fluorescence, the method of overstaining and
differentiation (5,7) is widely used. However, this method appears
quite l,imited when quantitative measurements are desired, consid-
ering that the rediffusion process of the excess AO is difficul,t
to control. Within a slide, variation in the local concentration
of cells can be reflected in variation of the diffusion gradient
and consequently of the destaining process; indeed, this has been
frequently reported, even without a clear understanding of the
reasons. Also, different degrees of convection in the solutions
are very likely from experiment to experiment. In our opinion
this procedure of overstaining and differentiation with reversible
dyes, quite popular in classical histology, has such a degree of
empiricism and variabil,ity that it should be substituted with
equil,ibrium staining. As above reported, a difference between AO
and EB can be expected in their kinetic interaction with native
highly superhelical chromatin-DNA in viable cells, since the
directional entrance into DNA appears to be from the narrow groove
with EB intercalating and from the wide groove with AO (14). EB
may bind first to the kink in DNA and then subsequently "sl,ip
into ll the interior of the double-helix; AO may first bind to the
sugar-phosphate chain at the kink and then intercalate as DNA
straightens and base-pairs separate. If we may speculate, these
differential binding processes already detectable in dinucleotide
monophosphate crystalline complexes (14), can be further enhanced
with consequent kinetic alterations by a highly ''kinked'' super-
coiled configuration at the levels of tertiary~ quaternary and
quinternary structure in native chromatin-DNA in the intact cell.
A final reflection: biol,ogical, phenomena are of such compl,exity
that empirical recipes and qualitative analysis onl,y contribute
to further confusion.
ACKNOWLEDGEMENT
This work Was supported by Grants CA18258 and CA20034 from the
National Institutes of Heal,th.
REFERENCES
1. Nicolini, C., Biophys. Biochem. Acta, 458, 243-282 (1976)
2. Baserga R., and Nicolini, C., Biophys. Biochem. Acta, 458
109-134 (1976)
3. Nicolini, C., Desaive, C., Kendall, F. and Fried, J., Canc.
Treatment Rep., 60, 1819-1826 (1976)
4. Wu, S., Toton, S., Zietz, S., Kendall, F. and Nicolini, C.,
Pulse Cytophotom., III, 57-76 (1977)
5. Nicolini, C., Kendall, F., DeSaive, C., Clarkson, B.,
Fried, J., Exp. Cell Res., 106, 111 (1977)
6. Dar zynkiewicz, T., Traganos, F., Sharpless, T. and
Melamed, M., Canc. Res., 37, 4635 (1977)
7. LePecq, J. and Paoletti, C., J. Mol. BioI., 87-105 (1967)
8. Waring, M., J. Mol. BioI., 27, 87-109 (1965)
9. Parodi, S., Kendall, F. and Nicolini, C., Nucleic Acid Res.,
2, 477-486 (1975)
10. Scatchard, G., Ann. N.Y. Acad. Sci., 51, 660-671 (1949)
11. Linden, W., et. al., Pulse Cytophotom., 277-289 (1977)
12. Nicolini, C., Linden, W., Zietz, S. and Wu, S. Nature, 270,
607-609 (1977)
13. Krugh, T. and Beinhardt, C., J. Mol. BioI., 97, 133-162,
(1975)
14. Sobell, H., et. al., J. Mol. BioI., 119, 333-365 (1977)
15. Nicolini, C., Parodi, S., Lessin, S., Zietz, S. and Belmont A.
submitted for publication
Parodi, S., Lessin, S., Fang, M., Zietz, S. and Nicolini, C.,
Biophys. J., 24, 97a (1978)
16. Rigler, R., Acta Physiol. Scand., 67, suppl., 267, 1-291
(1977)
17. West, S., in Physical Techniques in Biological Research,
Pollister, A., ed., Vol. III part, C., 1 Academic Press,
N.Y.-London (1969)
18. Armstrong, R., Kurmsev, R. and Strauss, U., J. Am. Chem. Soc.,
92, 3174-3181 (1970)
19. Kubota, Y. and Steiner, R.F., Biophys. Chem., 6, 279-289
(1977)
CHROMATIN STUDY IN SITU: III. DIFFERENTIAL EFFECTS OF
FEULGEN HYDROLYSIS
W. A. Linden, S. M. Fang, S. Zietz and C. Nicolini
Department of Physiology and Biophysics

Division of Biophysics, Temple University
Health Sciences Center, Philadelphia, Pa. USA
ABSTRACT
Smears of synchronized HeLa S3 cells,at 1, 3, 5, 8 and 12

hours after mitosis were Feulgen-stained with lN HCl at 60 0 C for
15, 60 and 120 minutes. For each cell population the integrated
optical density, nuclear perimeter and area, measured by using the
image analyzer Quantimet 720-D on line with a PDPll/40 computer,
show a striking differential dependence upon hydrolysis time. IOD
frequency distributions of log-phase HeLa cells are statistically
significant at the three hydrolysis times; specifically, both the
coefficient of variation of "Gl" peak and the fraction of cells
with "Gl DNA content" increase with hydrolysis time. Laser flow
microfluorimetric studies were also conducted on the same log-
phase HeLa cells, Feulgen-stained with acriflavine, respectively
at lN HCl at 60 0 C for 5, 10, 20, 60, and 120 minutes; 4N HCl at 250 C
for 15, 60 and 120 minutes and 5N HCl at 37 0 C for 15, 60 and 120
minutes. The decompositon of the DNA histograms yields dramatic
increases of "Gl" fractions, accompanied by increases of the co-
efficients of variation of the "Gl" peaks, whenever the modal
fluorescence per cell decreases as a result of variation in HCl
normality, temperature or time of hydrolysis. This differential
hydrolysis dependence of the Feulgen reaction is compatible with
the difference in chromatin supercoiling for cells in different
phases and subphases of the cycle. Optimal hydrolysis conditions
are outlined, stressing the limitations of utilization of Feulgen
reactions for a quantative assay of DNA content and for FMF cell
cycle analysis.
323
324 w. A. LINDEN ET AL.
INTRODUCTION
The Feulgen reaction is generally considered as a reasonably

specific and quantitative method for measuring DNA content (1,2).
However, there are indications that the amount of Feulgen stain is
not only depending on the actual amount of DNA present in the nuc-
leus but that it is reflecting the state of the deoxynucleoprotein
complex and the degree of chromatin condensation (1). Thus, dif-
ferent somatic cells require different time of hydrolysis (3).
Still more important are differences in the hydrolysis time de-
pendence reported by Bohm and Sandritter for normal cells and
mouse ascites tumor cells (4). Comparative studies of mesothelial
cells, lymphocytes and tumor cells performed by these authors
showed that, always in respect to the diploid value of the lympho-
cytes, the staining of the tumor cell population varied during dif-
ferent times of hydrolysis from triploid to hypotetraploid values
(4). The same holds true for granulocytes and lymphocytes, which
may yield DNA values 5-15% under the expected diploid value (5,
6, 7).
With the advent of flow microfluorimetry the fluorescent

Feulgen method has been widely used for the estimation of the
fraction of cells in each phase of cell cycle (8,9). This is
accomplished by computer evaluation of the cytophotometric DNA
distributions (10,12,11). Previous physico-chemical studies con-
ducted either in isolated (16) and "in situ" (15) chromatin from
synchronized HeLa cells indicate drastic changes in chromatin con-
formation during the entire cell cycle. A differential staining
effect, for each cell cycle phase, of variation in temperature and
acid concentration during the Feulgen hydrolysis reaction, due to
the difference in chromatin structure, would have significant im-
plications for these studies. Thus, we tried to evaluate the in-
fluence of hydrolysis time, temperature and acid concentration on
Feulgen stained HeLa cells in different phases of the cycle using
automated image analysis and flow microfluorimetry.
MATERIALS AND METHODS
Cell Culture
Logarithmically growing HeLa S-3 cells were maintained in

suspension culture in Joklik - modified Eagles Minimum Essential
Spinner Medium supplemented with 3.5% each of the calf serum and
fetal calf serum (13).
CHROMATIN STUDY IN SITU: III 325
Automated Image Analysis
The cells were synchronized by selective mitotic detachment,

as described previously (13). Approximately 90% of the detached
cells were observed by phase-contrast microscopy to be in mitosis
immediately following harvesting. The detached cells were main-
tained in culture at 37 0 C for 18 hours. During this period,
smears were prepared in triplicate from the same culture at 1, 3,
5, 8, 12, 15 and 18 hours after synchronization. The triplicate
smears for each post-detachment time were dried for one hour and
then fixed for 30 minutes in a mixture containing 85% methyl
alcohol, 10% formalin and 5% glacial acetic acid. Hydrolysis
treatment was performed in lN HCl at 60 0 C for 15, 60 and 120
minutes respectively. The smears were stained with the Schiff
reagent for 60 minutes following the procedure of DeCosse and
Aiello (14). After staining, the samples were mounted in Canadian
Balsam. The smears were prepared with particular attention since
prolonged hydrolysis time could lead to solubilization of DNA,
quite sensitive to fixation and air drying. Furthermore, in order
to avoid any systematic differences due to reagents, all fixatives,
hydrolysis, and staining of triplicate specimens were physically
conducted in parallel in the same reagents, and the same reagent
stocks were used for triplicate specimens sampled at other times.
Subsequent analysis proved that the geometric and densitometric
data, obtained from the three smears for each sample, were highly
reproducible.
Illumination was provided by a 100w tungsten halogen lamp and

a highly regulated direct current power supply using a 546nm
filter (40nm half-backwidth, Fish-Schurman, New Rochelle, N.Y.).
Shading error was minimized by use of a shade corrector. A flat
field was produced by use of a 100x planar achromatic oil object-
ive of 1.25 n.a. and open iris with internal magnification of lOx
produced by a Reichlert high-quality magnification changer. The
coefficients of variation for IOD (integrated optical density) and
area measured for a single nucleus positioned at 6 locations
around the border of the field and in the center were 1.0 per cent
for I.O.D. and 2.5 per cent for area. Each slide was used as its
own blank to define 0.0 O.D. and densitometer calibration was
then checked by means of a neutral density filter. Both the gain
of the video amplifier and the densitometer zero level were con-
tinuously referred to peak white during measurement by means of a
Servo system which can typically maintain measured values within
1% over a twenty-four hour period. Besides IOD several geomet-
trical parameters were measured (15): nuclear area and perimeter
will be presented in this paper. At least 100 cells were eval-
uated per slide. The raw data were loaded on a PDP11/40 computer
for further analysis. A detailed description of our image analysis
procedures as well as the data processing had been given elsewhere
(15, 16, 21).
326 W. A. LINDEN ET AL.
Flow Microfluorimetry (FMF)
HeLa S-3 cells logarithmically growing in suspension culture

were prepared for measurement in the Cytofluorograph (Bio/Physics
Systems Inc., Mahopac, N.Y.) using the fluorescent Feulgen reac-
tion adapted for flow microfluorimetry by Truillo et. al. (8).
Basically, the cells were removed from culture medium by centri-
fuging at 250xg for 4 minutes. The pellet was washed in cold
saline, the cells were resuspended in saline containing EDTA
(0.5 roM) and trypsin (0.1 mg/l.0 ml saline) and fixed for 18 hours
at OoC. Similar results were found using either pure alcoholic
fixative (80% methanol) or 20% formaldehyde. Several hydrolysis
treatments were applied, specificall~ IN HCl at 60 0 C for 5, 10,
20, 60 and 120 minutes, 4N HCl at 25 C for 15, 60, 120 minutes and
5N HCl at 37 0 C for 15, 60, and 120 minutes respectively. The
cells were stained for 20 minutes at room temperature with acrifla-
vine HCl (500 mg K2S205, 10 ml 0.5 N HCl 90m! 0.02% aqueous
acriflavine HCl Allied Chemical). Non specifically bound dye was
removed afterwards by washing three times with acid-alcohol (lml
12N HCl in 100 ml 70% ethanol). The performance of the FMF was
checked before and after each series of measurements by means of
acridine orange-stained calf thymocytes. This procedure was used
to insure that no detectable instrumental drift had occurred
during the run and to insure that the instrument itself produced
a constant coefficient of variation of approximately 5.5% for all
of the experiments herein reported. Each hydrolysis treatment was
conducted in parallel on 3 different samples: measurements on
e.ach stained sample were then repeated 3 times at various time
intervals, yielding a peak position and coefficient of variation
quite reproducible (with an overall variation of 0.5%). Changes
in fluorescence intensity due to internal staining variability,
were therefore minimal.
Computer Determination of Cell Cycle Parameters
In order to decompose the histogram to determine the fraction

of Gl , Sand G2+M cells, we took our basic method from the work
of J. Fried (12). Basically the method involves fitting the
experimental histogram by a sum of Gaussians, one for Gl cells,
one for G2+M cells, and a predetermined number for cells in S
phase. The problem becomes one of estimating the means, standard
deviations, and amplitudes of the Gaussians. As previously
described (12), we assume that the ratio of the mean G2+M (DNA)
channel position to the mean Gl position is 2, and that the co-
efficient of variation (C.V.) of all the Gaussians is constant.
After determining the C.V. by taking ~ of the full width of 0.67
of the height of the Gl p"eak, the heights of the Gaussians are
determined by utilizing a global optimization routine, which
allows the entire method to be programmed on our PDPll/40 computer.

More details are presented in other communications (18). Since the
basic mathematical model is ill-conditioned (18) we can show that
equally good fits, both by visual observation and chi-square an-
alysis, can be obtained by changing the initial values of the
parameters (Gl peak position and functions). Furthermore, math-
ematical analysis of the problem shows that as the C.V. of the
Gaussians increases, the variance of the estimated parameters
drastically increases (18). Thus, several runs have been con-
ducted both by varying the fit of the model to the same set of
data or by the same fit to different samples from the same Feulgen-
stained population, and report the mean and standard deviation
of the parameters obtained by the method (See Tables I-II).
RESULTS
Image Analysis
The dependence of integrated optical density on hydrolysis

time using lN HCl at 60 0 C is demonstrated in Fig. 1. The IOD
values are means of approximately 100 cells. The error bars
designate 95% confidence intervals. The 4 curves correspond to
HeLa cells at 1 hour, 5 hours, 8 hours and 12 hours after selec-
tive mitotic detachment. At these times the main fractions of the
cell population were respectively in early Gl , late Gl, early S
and middle-late S phase respectively (15,17). The curves show a
monotonic decrease of mean IOD with increasing hydrolysis time
but reveal a statistically different significant dependence from
hydrolysis time for different times after mitosis, suggesting that
cells in different phases (or sub-phases) of the cell cycle are
differentially affected by Feulgen hydrolysis, perhaps related to
a specific difference in chromatin morphometry (15,16).
Figures 2 and 3 demonstrate that both the geometric param-

eters, nuclear area and perimeter also show a unique differential
hydrolysis dependence for each time interval elapsed after mitosis.
These two geometric parameters are differently affected by in-
creasing hydrolysis time: i.e., at the same time after mitosis,
the area of cells in early S-phase (8 hours) decreases slightly
with increasing hydrolysis time (Fig. 2) while perimeter shows a
marked increase (Fig. 3).
These differential effects of hydrolysis time on cells in

different phases of the cell cycle have an impact on quantitative
evaluations of integrated optical density (DNA content) as it
appears evident in Fig. 1 and 4.
t,
1.5
'. ,
" ,
1.25 '.,
", '.t
lib...
I<
. .'..,
..,
..,
"'\. \
.,
~
.. --
\
'.
t······ "'~
1.0 '.
...r:: \
; - .......... '.
\
,
.\>~,
\.\
...~ ""."'" ,
\ \ ,\ , --"'\
\.
~ ,
\t
\ \
I...
0.75 \ " ,, \
\ .. 12 hr
,
,,
,
,
~
"f
.. '.
\. 8 hr
\.
0.5
\. 5 hr
\.
\.
\.~ 1 1\r
o 30 60 90 120
HYlIUl..YSIS Tum (minI
Fig. 1. Integrated optical density (means ± 95% confidence

intervals) of He La cells versus hydrolysis time at different
times after mitotic detachment: 1 hour ~---~, 5 hours ~---_),
8 hours (~---~), 12 hours (i---i). Hydrolysis was performed in
IN Hcl at 60°C.
80
-~--- ..-;
Ihr
_..__._-_..- ..5hr
----,
HYmOLYSIS TH£ Cmln)
Fig . 2 Nuclear area of HeLa cells (means ±9S% confidence intervals)
versus hydrolysis time in IN HCI at 60 0 C at different time intervals
after mitosis o (0- - 0 ) 1 hour, ( 1---1) 5 hour s , (6-- - ~) 8 hours ,
00
and (! _ o_ o_! ) 12 hours after mitosis o
60 60
-----~
20
o~ ~- -----· ~6~
0 ----~90
~----~1~20
~~ O----~~
~----~60~----~
90~--~
120
HYmOLYSIS Tn£ IItl n)
Fig . 3 Nuclear perimeter (means ± 95% confidence intervals) of HeLa

cells versus hydrolysis time at dif ferent times after mitotic
detachment, as Figure 2 .
Figure 4 presents IOD (DNA) distributions of HeLa cells 12

hours after mitosis using 15, 60 and 120 minutes of Feulgen hydro-
lysis. The number of cells is plotted versus the IOD (expressed
in arbitrary units). A Kolmogorov-Smirnov statistical test carried
out on the three frequency distribution proved that significant
differences exist between these IOD histograms obtained at 3 dif-
ferent hydrolysis times. The first peak of each histogram cor-
responds to GI cells (DNA content X), the second peak to (G 2+M)
cells (DNA content 2X). The cells recorded between the two peaks
are S cells with variable DNA content between X and 2X. Figure
4A gives the IOD histogram in the same scale showing the decrease
of IOD with increasing hydrolysis time, while in Figure 4B the
IOD scale is normalized in order to display the GI peaks at the
same vertical position (unit 15). As is well known (17), at 12
hours after mitosis the degree of synchrony has decreased such
that a major fraction of the cells is in S phase, but there are
considerable fractions of cells also in the Gl - and (G 2+M)-phases.
Figure 4B reveals that with increasing hydrolysis time the IOD
distribution is changed drastically; the fraction of cells with
higher IOD (DNA content 2X) decreasing with increasingly hydroly-
sis time (see Table I). The implications of this finding for a
quantitative estimation of DNA content, using Feulgen staining on
cells with the same ploidy level, but different degrees of chroma-
tin supercoiling, are evident. In this communication, we further
explore the implication of this finding on the quantitative cell
cycle analysis from cytophotometry of Feulgen-stained cells.
Laser Flow Microfluorimetry
The performance of the FMF was checked before and after each
series of measurements using acridine orange-stained calf thymo-
cytes in order to maintain a constant coefficient of variation
(C.V.) of approximately 5.5% throughout all experiments herein
reported.
Decomposition of the DNA histograms (See Materials and

Methods) yields the fraction of cells in each phase of cell cycle:
regardless of both the low X2 and the perfect overlap of theoret-
ical and experimental frequency distribution of fluorescence per
cell (18), these model's estimates are mostly given to infer trends
in the data. Table II shows that C.V. of the Gl peak for log-
phase HeLa cells Feulgen-stained in 4N HCl at 25 0 C for 5 minutes
is about 11.3%, compatible with the 10-11% previously reported
under the same conditions with a similar cytofluorograph (9).
Better C.V. can be obtained using our B-D Cell Sorter but the
FMF distributions show a similar dependence from Feulgen hydro-
lysis. Figure 5 demonstrates that in agreement with our image
analysis data and previous findings (19), the modal fluorescence
CHROMATIN STUDY IN SITU: III
331
..................... __ ..... --.---_.... ---'! .... _------_ ... -.... __ .. ---------;_ .. _. __ . _.... ;
! . !
15 _In . 8 .. 19 .tn !
..
J, It !
U ,
U
U ,_
i
-"
10
! )I; lUI X"
tdI
X II 1I1IIX lII. A)QC
JlX" XXlIIXIUI X 111 X XIII JOI X
•••• O •• X"II lnCIcv J: .'C1'JlItU
I ""......... 'Ita lIXXX :QXU X xu lI'l ••••
;.................. __ ............. _-- ...-........ --.---..-----------. -------...,

(/) ;............ --.....
u .. Q,..~.JIa
1.:llJUU!. .xKnlunr.UUXXlD&
a
)I Jl !
JltlUUI'XXIO( "IlXXxxPNCX ..
:aur..X)I.'UUUlxX~x)ll'X ,., II
...J
......
...J
;: II 60 .J" ~x 60 _In
~ 10 n
II. "
x
1t
p
xx ,
)1,111. X P
III X 11. xlii' )UI
•)IX. 1t'
- ~u
X1l.at
".lA~ .U.I .au.
•• u. !Ul'XXICXX )I; XIUllCX1C JOI ""XIOl)l
1f.....X~ ,bUOC"'. X IIXJQCK1Ill"• •1QnI:xlOnr: 11
:-...--.. --.!!!~=~!..!----~-----!.. .. --.--.----~~~~~~--~-- .. ~!

!
I ,. lIill
J 'II .. JI
r
! J,JI JlII:
10 ~.::
I )l'X1l: !CIIM: II:
,t I
=~ :::::=
xr.XXI{XXXU
: ;~ xx~ x::
II'UJI)oc I(XX'If WII.
I XXIC.)Cxx!ur;.c. 1C1UUUUC ltINXlIC 1tft1C
, liXXXXX10UIXX 'JIUC~ :IUlJlDX10'lC" •
t :t"")(XJ"XXXXY. ! X'rJUUlUX.XXKJlX"S)iXftx.!
.--···i'O···-ai····-·3l;·······.········-I6-······16-··-··:\6-·····-40
INTEGRATED OPTICAL DENSITY (a.u.)
Fig. 4 IOD (DNA) frequency distributions of HeLa cells 12 hours
after mitosis: A) IOD scale in absolute value (a.u.), B) different
IOD scales in order to have the Gl peaks lined vertically.
:zoo 5 N HCL 370c
~
(/)
z ISO
U.I
I- I N HCL 600C
z &.--~
,,
U.I ,,
U ,,
Z
U.I
U 100
(/)
U.I 4 N HCL 2SOc \
a:: .... -... - .-. - -_.-'1(::-
o
;:)
..J
.....
...J
c(
C
o
:::E:
o 30 60 90 120
HYDROLYSIS TIME (min)
Fig. 5 Modal fluorescence intensity of log-phase HeLa cells versus
hydrolysis time for IN HCl at 600 C (6.----/::;), 4N HCl at 2S o c (_---.),
and SN CHI at 37 0 C (A---&). The ordinate gives the channel number
(a.u.) of the Gl peak in the FMF fluorescence (DNA) histograms).
intensity of the acriflavine Feulgen-stained HeLa cells is also

strongly dependent on HCl normality and hydrolysis time. The
values were taken from FMF DNA histograms of the HeLa cells in
log-phase. The ordinate gives the channel number on the multi-
channel analyzer of the FMF for the Gl peak of the fluorescence
intensity (DNA) histogram, the abscissa gives the hydrolysis time.
For IN HC\ 60 0 C we find a rapid decrease of fluorescence intensity
corresponding to the rapid decrease in IOD shown in Figure 1.
After hydrolysis of 4N HCl at 25 0 C and 5N HCl at 37 o C, the fluor-
escence intensity is respectively constant or slowly decreasing
during the time interval studie~ the hydrolysis in 5N HCl at
37 0 C yields approximately double fluorescence intensity, com-
patible with previous absorbance measurements (19).
Figure 6 shows few representative histograms of log-phase

HeLa cells after acriflavine Feulgen-staining using different
HCl normality and hydrolysis time; respectively, in order of de-
creasing fluorescence intensity: A) 5N HCl at 37 0 C for 5 minutes,
B) IN HCl at 60 0 C for 10 minutes, C) 4N HCl at 25 0 C for 60 minutes,
D) IN HCl at 60 0 C for 120 minutes.
Visual analysis indicates that a decrease in the relative

intensity of the Gl peak, with increasing HCl normality and/or
hydrolysis time is accompani€d by a relative decrease of the
fraction of cells in the second (G2+M) peak with corresponding
increasing of the fraction of cells in the Gl peak using the same
log-phase HeLa cell popUlation. Simultaneously, the coefficient
of variation of the Gl peak has increased considerably, with
decreasing modal fluorescence. All these empirical observations
are confirmed by a rigorous computer analysis of the same Feulgen-
stained HeLa cell population, hydrolyzed at different temperatures,
HCl normalities and times. The mean values for each fraction of
cell cycle phases are obtained by a least square fit of the DNA
histograms to the mathematical model previously described (see
Table II); they are intended mostly to infer trends in the data,
since occasionally the C.V. is larger than 15% (not due to a
reduced performance of the cytofluorograph, which was constantly
monitored against the standard calf thymocytes). The same trends
are indeed substantiated by automated image analysis, which does
not suffer the same pitfalls.
All these studies have been carried out in the wide range of
temperature, time and acid concentration, commonly utilized;
particular attention was given to the short time scale (5 - 10
minutes), changing temperature and HCl normality.
Figure 7 summarizes the findings obtained, either by laser

flow microfluorimetry (upper panel) or automated image analysis
(lower panel), on the effect of varying Feulgen hydrolysis
(either time, temperature or HCl normality) on the same log-phase
()
:I:
:0
o
s:
»-I
(AI (el z
5N HCL 4N HeL CJl
-I
37°C 25°C C
5 MIN 10 "fIN o
-<
~
~
S
(BI 101
IN HCL IN HCL
600 e 60°C
10 MIN 120 "fIN
Fig. 6. Fluorescence frequency distributions of

log-phase HeLa cells Feulgen stained with acriflavine
with different hydrolysis : (Al SN HCl at 37°C for
S min . , (B) IN HCl at 60°C for 10 min. , (el 4N HCl
at 2SoC for 60 min . , (D) IN HCl at 60°C for 120 min. Co)
Co)
Co)
.... 301
' ...., 1
".
....'. -'-.
........'"
G2+S
-----.-- - - ---- ..
..
,..-- ---- ---,.. . ~:".:.~-
~
---
.... >
...
'- .............
.... -.... c.v . o
.................... _
10
o 100 200
tIlDAL FWCIU!8CIiNCE Ca. u. ,
o 100 2110
IN'1'8BAftI) CPl'ICAL IBBlft Kl03Ca.u.,
Fig. 7 Upper: Percentage· of cells in Gl (. ___ .), G2iB. (x----x),

and coefficient of variation of Gl peak (0-.-) versus modal fluores-
cence of the Gl peak, obtained by the same log-phase HeLa cell popu-
lation hydrolized at different extents (Fig. 5 and Table II using
laser flow microfluorimetry).
Lower: Percentage of cells in Gl (. ___ .) and G2+S (x----x) versus

integrated optical density, determined on the same HeLa cell pop-
ulation 12 hours after mitosis, using automated image analysis.
The fraction of cells in mitosis in both cases is less than 1%.
HeLa cell population, containing a constant mixture of cells in

different phases of the cycle.
A decrease of the modal fluorescence per cell measured by

laser flow microfluorimetry (Fig. 7, upper panel), is accompanied
by an increase in the "computed" fraction of Gl cells at the
expense of a significant decrease in the computed "G 2+S" fraction.
Identical phenomena is observed by the independent method of
automated image analysis (Fig. 7, lower panel), which measures
integrated optical density, i.e. DNA absorbance.
Another common feature of these two independent measurements

is the dramatic increase of the coefficient of variation of the
Gl population: (Fig. 7 upper panel) for laser flow microfluorim-
etry, and Table III for image analysis.
DISCUSSION
The idea that Feulgen-staining may be influenced by chromatin

has been around for a long time as it can be seen from recent
excellent reviews on the subject (1,2). This combined study of
Feulgen-stained densitometric texture analysis (15,16,21) and
laser flow microfluorimetry (8,9), specifically characterizes the
differential staining effect of Feulgen hydrolysis on the various
compartments of the cell cycle. Several profound implications
can be drawn from these findings. First, the Feulgen stain,
which also reflects the state of chromatin (1), cannot be con-
sidered a quantative method for measuring DNA content when cells
with different metabolic activity and extent of proliferation are
compared. G2 cells, which have twice the amount of DNA with
respect to Gl cells, show a decreased ratio in their relative
absorbance, with increasing hydrolysis (Fig. 4), or fluorescence
(Fig. 6), as indicated by the more pronounced decrease in "G 2 "
peak and consequent apparently increased fraction of Gl cells in
a frequency distribution of a log-phase population (Table I-II):
this is compatible with the differences in chromatin morphometry
between G2 and Gl cells, which have been previously reported
(15,21). Similarily, cells with the same Gl DNA content show a
differential hydrolysis dependence of densitometric (Fig. 1) and
geometric (Fig. 2 and Fig. 3) parameters of Feulgen-stained
synchronized HeLa cells at 1 hour (early S) after mitosis: this
differential effect is responsible for the increase in the coef-
ficient of variation (Table II-III) of the Gl population and is
compatible with a significant variation of chromatin morphometry
and supercoiling (23) at the late Gl- early S transition and
within the Gl phase, which can now be subdivided into three sub-
phases (15,21). This indicates that the Feulgen stain is related
not only to the amount of DNA, as previously suggested, but also,
in a specific manner to DNA conformation within the intact nucleus.

Indeed, this property can be properly utilized to objectively
characterize, at the optimal hydrolysis time and acid concentration,
cells with the same DNA content, but different chromatin function
and structure, i.e., proliferating (Gl) and non-proliferating
(GO) cells (16). Our findings suggest reasons for the strong
hydrolysis dependence reported for tumor cells with respect to
lymphocytes (4) and the lower yield of DNA absorbance of granulo-
cytes and lymphocytes (up to 15% under the expected diploid value
(5-7)).
Finally, the differential hydrolysis dependence of Feulgen

reaction, due to differences in chromatin super coiling within the
cell cycle phases and subphases (15-17) has implications for flow
microfluorimetric (FMF) analysis of cell kinetics. As can be
visualized by Figures 5-7, not only the relative fluorescence in-
tensity of the Gl peak (Fig. 5), but also the frequency distribu-
tion of fluorescence intensities (Fig. 6) and the computed cell
cycle parameters (Table II) drastically change by varying either
the HCl normality, the temperature or the time of hydrolysis on
the same HeLa cell population.
This implies that cell kinetic data collected in various

laboratories on similar biological systems, by FMF analysis on
acriflavine Feulgen-stained cells, can be compatible only if ob-
tained under the same hydrolysis conditions: the optimal condition
is achieved with 5N HCl at 37 0 C; where A) the fluorescence in-
tensity per cell is larger and slightly dependent on hydrolysis
time; B) the coefficient of variation of the Gl peak is minimum
(extremely important for an accurate computation of cell cycle
parameters). The 5 N HCl at 37 0 C also gives the closest correspond-
ence between autoradiographic (not shown) and FMF estimates of
the cell cycle parameters.
A general trend can be found in both laser flow microfluori-

metric (Fig. 5-6, Table II) and densitometric image analysis
(Fig. 1-4, Table I and III), regardless of the normality, temp-
erature and time: the decrease in relative intensity in either
fluorescence (Fig. 7, upper panel) or absorbance (Fig. 7, lower
panel) is systematically accompanied by a significant decrease of
the computed G2+S fraction, with consequent increase of computed
Gl fraction from the same HeLa cell population. In both cases
the spread of the Gl population significantly increases, compatible
with the existence of at least 3 subcompartments in Gl phase with
different chromatin structure and therefore with different hydroly-
sis dependence (15,21). Furthermore, the most commonly utilized
Feulgen-staining, i.e., using 4N HCl at 25 0 C (in FMF analysis) and
1N HCl at 60 0 C (in cytopathology), seem to be quite limiting be-
cause of the lower yield of fluorescence (or absorbance), the
higher coefficient of variation and (for iN HCI) the critical de-

pendence on hydrolysis time. Since identical data were obtained
in alcoholic fixatives, either with or without formalin, this
conclusion is compatible with recent studies (20), using alcoholic
fixatives containing no formalin, on the exposure and removal of
aldehyde groups during Feulgen acid hydrolysis, which seems optimal
at temperatures slightly above room temperature in combination with
high acid concentration (20). Finally, maybe as a consequence of
the significant variations in chromatin supercoiling (23) and
nuclear morphometry (15,21) between and within cell cycle phases
the inherent limitation of densitometric and FMF determination of
cell cycle parameters appears evident in Feulgen-stained cells.
In that this limitation, accompanied by the pitfalls of the various
mathematical models (10-12, 18) (including ours which is used only
to infer trends in the data), makes imperative the need for altern-
ative approaches to cell kinetic studies, particularly by means of
physical techniques which provide more direct physical observables,
by mathematical models which more faithfully reflect the charac-
teristics of the various subphases of cell cycles, and by the mathe-
matical analysis of time sequence of FMF distribution in perturbed
and unperturbed cell populations (22, 24, 25).
TABLE I
Percentage of cells by automated image analysis in a given

phase (G l ,S,G 2+M) of the cell cycle from the same HeLa cell pop-
ulation 12 hours after selective mitotic detachment (17), for
each hydrolysis time in iN HCl at 60 0 C (Fig. 4).
The mean value and standard deviations are obtained as des-

cribed in Materials and Methods.
HYDROLYSIS TIME Gl (%) S(%) G2+M(%)
15 minutes 25 ± 3 40 ± 3 35 ± 6
60 minutes 28 ± 1 58 ± 3 14 ± 2
120 minutes 35 ± 2 62 ± 3 3 ± 1
TABLE II
Percentage of Cells in Gl , S, G2+M phases by a least square fit

to a mathematical model (see methods) of various Feulgen stained
HeLa - S3 cells in log phase (laser microfluorimetry).
GIS G2 + M C. V.
5N HCl-37oC 49.9 36.2 13.9 10.3

5 minutes (± 4.1) (± 4.0) (± 1. 3)
5N HCl-37oC 50.9 37.1 11.9 12.0

60 minutes (± 0.8) (± 2.1) (± 1.8)
5N HCl-37oC 61.7 24.0 14.3 11.9

120 minutes (± 2.9) (± 2.0) (± 1.5)
iN HCl-60 o C 57.7 24.5 17.8 12.3

10 minutes (± 2.4) (± 1.5) (± 0.8)
IN HCl-60oC 67 26.5 6.5 31.5

120 minutes (± 3.5) (± 1.5) (± 1.5)
4N HCl-250C 52.5 30.8 16.7 11.3

5 minutes (± 2.0) (± 7.5) (± 1.7)
4N HCl-250C 56.0 36.9 7.1 21.8

60 minutes (± 3.0) (± 4.9) (± 1.1)
TABLE III
Time hydrolysis dependence of mean value and standard deviation
(in percentage of the mean) of integrated optical density (a.u.)
from a population obtained by combining HeLa cells with similar
Gl DNA content but taken 1 hour (early Gl ), 3 hours (middle Gl ),
5 hours (late Gl) and 8 hours (early S) after mitosis. The cell
population was hydrolized in IN HCl at 60°C for 15, 60, and 120
minutes. The data were obtained by automated image analysis (see
fig. 1 and reference 15).
Hydrolysis
Time Mean (a.u. ) S.D. (%)
15 minutes 110 x 10 3 13.5
60 minutes 88 x 10 3 12.5
120 minutes 52 x 10 3 22.8

REFERENCES
1. Ringertz, N.R. (1969) in Handbook of Molecular Cytology, ed.

A. Lima-De-Faria (North Holland Publishing Co. Amsterdam),
p. 658.
2. Lillie, R.D., Fullmer, H.M. (1976) Histopathologic Technic and
Practical Histochemistry (McGraw Hill Book Co., N.Y.) P. 17.
3. Sprenger, E. (1974) in Impu1scytophotometric, ed. M. Andreeff
(Springer-Verlag, Berlin), p. 5.
4. Bohm, N. and Sandritter, W. (1966) J. Cell BioI. 28, 1-7.
5. Bohm, N. and Sandritter, W. (1975) in Current Topics in
Pathology, 60, ed. E. Grundmann and W.H. Kirsten (Springer-Verlag,
Berlin) p. 156.
6. Garcia, A.M. (1964) Acta Histochem. 17, 249-257.
7. Mayall, B.H. (1969) Histochem. Cytochem. 17, 249-257
8. Trujillo, T.T. and Van Di11a, M.A. (1972) Acta Cytologic 16,
26-30.
9. Nicolini, C., Kendall, F., Desaive, C., Baserga, R., Clarkson, B.,
and Fried, J., Cancer Treatment Rep. 60, 1818-27 (1977).
10. Dean, P.N. and Jett, J.H. (1974) J. Cell Bio1. 60, 523-527.
11. Baisch, H., Gohde, W., and Linden, W.A. (1975) Rad. Environ.
Biophysics, 12, 31-39.
12. Fried, J., (1976) "Computers and Biomedical Research", 9, 263-76.
13. Stein, G. and Borun, T. (1972) J. Cell BioI. 52, 292-307.
14. DeCosse, J.J. and Aiello, N. (1966) J. Histochem. Cytochem. 14,
601-604.
15. Kendall, F., Swenson, R., Borun, T., Rowinski, J., and
Nicolini, C. (1977) Science, 196, 1096-1108
16. Nicolini, C., Giaretti, W., Desaive, C., and Kendall, F.
(1977) Experimental Cell Research 106, 119-125.
17. Nicolini, C., Kozu, A., Borun, T., and Baserga, R. "J. Biochem."
(1975) 250, 3381-3385.
18. Wu, C.T., Toton, S., Kendall, F., Zietz, S., Linden, W., Eisen, M.
and Nicolini, C. (1977) Pulse Cytophotometry III, 51-62.
19. Fand, S.B., "Introduction to Quantitative Cytochemistry",
(1970), ed. Wied, G.L. and Bohr. Vol. 2, Academic Press, N.Y.
p. 209.
20. Kje1lstrand, P.T.T. (1977), The J. Histochem. Cytochem. 25,
p. 129-134.
21. Nicolini, C., Kendall, F., and Giaretti, W., 1977 Biophysical
Journal, 19, 163-176.
22. Scherr, L. and Zietz, S., Radiation Research 67(3), 585, 1976.
23. Nicolini, C. and Kendall, F., Differential light scattering
in native chromatin. Corrections and Inferences, combining
melting and dye binding studies. A two-order superhelical
model. Physio1. Chern. & Phy. 9(3) 265-83
24. Zietz, S. and C. Nicolini Cell Tissue Kinet. (submitted, 1978)
25. Zietz, S. and C. Nicolini in Biomathematics and Cell Kinetics
(1978) ed. A.J. Va11eron, Elsevier.
SCANNING AND FLOW PHOTOMETRY OF CHROMOSOMES
Mortimer L. Mendelsohn
Biomedical Sciences Division
Lawrence Livermore Laboratory
University of California
Livermore, California 94550, U.S.A.
The chromosome deserves a prominent place in a book on

chromatin structure and function because in its condensed
metaphase form the chromosome achieves the highest order of
chromatin structure. Of course chromosomes are interesting for
many other reasons. The formation and behavior of chromosomes in
mitosis and meiosis continues to be a fascinating problem in
biology, chromosomes provide a clinically useful window into
human genetics, and chromosomal vulnerability to breakage and
subsequent functional disruption by clastogens defines a type of
genetic injury which requires careful study and control
This chapter describes two approaches toward biophysical

photometric study of chromosomes: one based on scanning
microscopy and one based on flow cytometry. They stem from two
major achievements of Tobjorn Caspersson, the development of high
resolution microscopic cytophotometryl and the discovery of
chromosome banding with the associated insights into DNA
fl uoroch romes. 2
BACKGROUND
An average human metaphase chromosome contains several

hundred million base pairs of DNA, enough for approximately a
million nucleosomes divided equally between two identical
chromatids. These paired 4 cm lengths of DNA compact into a 5 by
1 by ~ micron, dense, refractile, highly stainable, resilient
341
342 M.L.MENDELSOHN
object. Lower orders of structure within the chromatids, such as

bands and major coils, are sometimes visible in the light
microscope.
Many properties of the chromosome can be studied optically.

Basic staining is perhaps the oldest of these and takes advantage
of electrostatic binding to DNA phosphate. UV absorbance is a
property of the bases and when carefully applied can be an
excellent measure of DNA content. Deoxyribose is selectively and
quantitatively stained by the Feulgen reaction. In this
half-century-old method, the DNA is depurinated by strong or hot
hydrochloric acid and the uncovered sugar aldehydes are stained
by one of many chromophoric Schiff reagants. The Feulgen
reaction was the standard method for cytophotometric DNA
measurement until its recent displacement by DNA fluorochromes.
Some DNA fluorochromes, such as ethidium bromide and acridine
orange, intercalate into DNA; others, such as chromomycin A3 and
Hoechst 33258, bind to the grooves. The fluorochromes are
comparatively easy to use, their few constraints allow them to be
combined readily with other cytochemical methods, and as will be
shown below, they have useful and interesting differences in
their affinities to various DNA's. Fluorescent antibodies to (a)
single stranded DNA, (b) BrdU in DNA and (c) thymine dimers and
other photoproducts of DNA can be used to stain these DNA
components specifically. Repetitive DNA sequences can be marked
by RNA or DNA complementation using stains or autoradiography.
At an even higher level of organization, chromosomal bands are
elicited by quinacrine staining and by several absorbent or
fluorescent stains combined with enzymatic or chemical
degradation of the chromatin. Finally, the chromosome itself can
be measured for length and area and can be classified to varying
degree by its size and the location of its centromere. Size
properties unfortunately are highly relative because chromosomes
continue to compact (i.e., become shorter and denser) as cells
are held in metaphase.
Over the past 15 years, first at the University of

Pennsylvania and now at the Lawrence Livermore Laboratory, the
approach my colleagues and I have taken to chromosome measurement
is to seek relatively invariant chromosomal properties,
particularly properties based on DNA content, and to use these to
identify chromosomes and to study their stability under normal
and abnormal conditions. We began with image-analytic,
absorbance measurement of chromosomes on slides.
CHROMOSOMES ON SLIDES
Analysis of DNA content of individual chromosomes of a well

flattened, properly stained, metaphase cell involves application
SCANNING AND FLOW PHOTOMETRY OF CHROMOSOMES 343
of standard photometric principles to objects which have been

morphologically isolated by image ana1ysis. 3
The Beer-Lambert Law relates intensities (1 0 = incident

intensity, I = transmitted intensity) of monochromatic light to
the concentration (c), path1ength (1) and absorptivity (k) of a
uniformly distributed chromophore.
I -kcl
e
I
o
The equation is often used in its logarithmic form
I
- log OD .. kcl
I
o
where OD is the optical density or absorbance. For
cytophotometry it is convenient to restate the Law in terms of
the mass of chromophore (m) in the measuring field and b the area
of the field.
OD = k~
b
b
m OD k
Obviously objects such as chromosomes do not have uniformly

distributed chromophore. The presence of resolvable differences
of grayness within a measuring window is equivalent to averaging
intensities; because of the exponential relationship in the Beer
Lambert Law, such averaging results in a negative error in
measured mass of chromophore. This distributional error
increases with heterogeneity and optical density and can well be
30% for deeply stained chromosomes. Scanning microscopy
essentially eliminates distributional error by making each
measurement of optical density in a spot or window which is at
the limit of resolution of the optics and hence shows no
resolvable internal differences in grayness. Many such
measurements cover the entire object in a regular raster and the
sum of the optical densities gives a true measure of stain
content (M) of the object
M .. ~ 1: OD
k
where r corrects for overlaps or gaps in the scanning raster. 4
Image analysis is required in this process to limit the

sU1lllllation of optical densities to the region representing the
single chromosome or chromosome part. The image in this case is
344 M. L. MENDELSOHN
a high resolution digital image of a metaphase cell. In our work

it consists of a 200 by 200 raster of 40,000 points; each point
is separated by 0.25J;l and contains up to 8 bits (256 levels) of
optical density information. We obtain the image from a
microscope slide in six seconds using a flying-spot cathode-ray
scanner operating near the limit of optical resolution in the
blue region of the visible spectrum. The high quality of such a
digital ~age is shown in Fig. 1.
A well prepared metaphase cell is a compromise between

keeping all the chromosomes sufficiently close to establish that
they are from the same cell and sufficiently separated to
minimize overlapping and touching chromosomes. Because
chromosomes are inherently unbounded objects and are seen through
that are roughly Gaussian.

S
diffraction l~ited optics they present broad grayness profiles
The tails of these profiles extend
Fig. 1. A digital ~age reconstituted from a partial scan of a

human metaphase cell. The chromosomes are stained with
gallocyanin-chrome alum and their grayness represents
DNA content. This image is near the limit of optical
resolution of the light microscope.
a micron beyond what appears to the eye to be the chromosome core

and actually contain a third of the chromosome's optical
density. We program a general-purpose digital computer to core
the chromosomal images by finding the grayness level that
corresponds to the inflection point of each chromosome's grayness
profile. At this point, human editing is used interactively with
the computer to correct occasional touching or fragmented cores.
A one-micron region around the core is then added to each
chromosome, and points claimed by more than one core are assigned
to the nearest core. Finally a sampling of unclaimed points in
the vicinity of each chromosome is used to estimate the
background. The computer is now ready to sum the optical
densities, subtract out the background and store the definitive
estimate of relative mass of chromophore.
The centromere of a chromosome can be identified either as

the region of narrowest width or by tracking the chromatids back
to their intersection. We prefer yet a third method based on the
relative paucity of DNA in the centromeric region. The computer
finds the axis of the chromosome, divides the chromosome into
strips perpendicular to the axis, sums the optical density in
each strip, searches for the dip in optical density in the
centromeric region, fits a least-squares quadratic to the dip,
and uses the minimum of the function to define the centromere to
a resolution of 0.1 micron. 6
Typically, we analyze ten metaphase cells (460 chromosomes)

from an individual. The cells are obtained from blood cultures
and prepared by conventional methods. They are first stained
with quinacrine, photographed and analyzed by eye using banding
patterns to classify each chromosome. They are then destained,
treated with ribonuclease and restained with gallocyanin-chrome
alum, a basic stain with good stoichiometry and reasonable
specificity for DNA phosphate. The scanning and computer
analysis follow, and finally the data are stored and manipulated
in a large computer file. The throughput is slow: about one
individual can be studied per week. 3
A typical result is shown in Fig. 2. By combining DNA

content and centromeric index (the ratio of DNA contents of the
long arm and the total chromosome), the computer can characterize
the chromosomes well enough to classify them into 20 groups.
This performance is better than the eye can do on non-banded
chromosomes, but is not competitive with banding which generally
allows classification of each chromosome. This is a secondary
issue in our studies because the chromosomes are already
classified and the main point is to cumulate and interpret the
quantitative data on DNA content.
0.9
-.J
~
~
~
"- 0.8
~
~
0:::
<[
~
19
0:::
<[
-.J 0.7
X 17
W
0
Z 9
~ (0
U
~
0:::
W 0.6
a0
~
0
0:::
I-
z
w
u "b
0.5
1.0 2.0 3.0 4.0 5.0
DNA STAIN CONTENT. percent

Fig. 2 The DNA content and centromeric index of the chromosomes
from 10 metaphase cells of a normal man. Each
pre identified chromosome type is shown as the 50%
confidence ellipse centered on the mean for the type.
This general pattern is highly stable among normal
humans. On the abscissa 100% is the sum of DNA contents
of the 44 autosomes.
How good are these measurements? In replicate scans, made

after repositioning the cell, mUltiple measurements of the same
human chromosome give an average standard deviation of
replication of 2.5% for DNA content and 2.3% for centromeric
index. Standard deviations between homo logs or among chromosomes
of the same type are 4.4% for DNA content and 3.5% for
centromeric index. Standard errors for means are correspondingly
smaller. Thus we estimate that in a sample of 10 cells we can
detect mean deviations in DNA content of average chromosomes of
three-ten thousandths of a genome or 3 million base pairs. By

molecular-genetic standards we are still talking about a gross
property, but by cytological standards this is extraordinary
resolution. To achieve this error rate we must normalize the
measurements within each cell to the cumulative stain content of
all 46 chromosomes. This is because cellular DNA stain content
varies by 6.9%. Such large variation is typical of cytochemical
measurements on slides; it is poorly understood but perhaps
reflects the subtle effects on stoichiometry of drying,
attachment to the glass, surrounding cytoplasm and fixation of
cells and chromosomes.
The table shows the means and the standard deviation of the
means within individuals and among individuals from our standard
human data set. 7 The means differ only slightly from
corresponding data on chromosome length, but the error bars for
photometry are much smaller than for length measurements.
Perhaps the most intriguing and perplexing result of the
photometry is the appearance of several outliers in the
chromosomes from each individual. About half of the outliers are
associated with a morphologically identifiable change, such as in
banding or in centromeric heterochromatin. For example, in one
normal individual a large, brightly fluorescent satellite is
STANDARD VALUES FOR DNA-BASED MEASURES OF THE HUMAN

KARYOTYPE
Normalized optical density "dex
Centromere In
1
Mean
4.295
I SO
within
0.141
J SO
among
0.198
Mean
0.518
I SO
within
0.012
J
SO
among
0.016
2 4.190 0.129 0.163 0.612 0.016 0.022
3 3.482 0.116 0.128 0.541 0.018 0.022
4 3.336 0.131 0.121 0.727 0.019 0.017
5 3.183 0.130 0.089 0.730 0.019 0.018
6 2.984 0.101 0.161 0.647 0.018 0.021
7 2.769 0.105 0.114 0.625 0.019 0.016
8 2.515 0.120 0.132 0.687 0.025 0.021
9 2.371 0.108 0.173 0.651 0.024 0.023
10 2.355 0.112 0.065 0.696 0.023 0.034
11 2.335 0.079 0.080 0.600 0.023 0.025
12 2.319 0.096 0.080 0.731 0.021 0.033
13 1.896 0.031 0.156 0.857 0.024 0.048
14 1.781 0.090 0.139 0.852 0.025 0.039
15 1.728 0.077 0.010 0.843 0.024 0.052
16 1.608 0.070 0.125 0.599 0.028 0.041
17 1.471 0.064 0.080 0.695 0.031 0.049
18 1.395 0.061 0.061 0.765 0.030 0.033
19 1.082 0.053 0.087 0.554 0.032 0.041
20 1.160 0.062 0.091 0.581 0.042 0.046
21 0.830 0.056 0.074 0.772 0.036 0.039
22 0.888 0.053 0.116 0.771 0.038 0.046
X 2.659 0.114 0.154 0.624 0.033 0.041
Y 0.918 0.056 0.086 0.766 0.055 0.018
present in every cell on one chromosome 21 and associates with a

20% increase in DNA content of that chromosome. But for the
other half of the deviants, we see no change in the chromosome
and have only the DNA measurement to go on. These variants occur
at or below the 0.01 level of significance at a frequency of 1%
and hence could be statistical in nature. However, preliminary
studies in families suggest that over 90% of the outliers in
children can be traced back to one or the other of the parents,
indicating the high heritability and the possible reality of most
of the deviant chromosomes. We presume the deviations are due to
repetitive and other non-informational DNA. Whatever their
cause, they provide both an opportunity and a difficult challenge
to our approach. The opportunity is to understand more about the
phenomenon of DNA constancy and its exceptions, and the challenge
is to learn how to set normal limits and apply chromosomal DNA
measurements in the face of such stable, apparently normal
outliers.
CHROMOSOMES IN SUSPENSION
Beginning four years agoB, we have been exploring a new and

dramatically different approach to cytogenetics based on flow
cytometry and sorting of fluorescently stained chromosomes.
Fluorescence is a more complex and information-rich process

than absorbance. Its measurement is affected by the phenomena of
fading, quenching and energy transfer, all of which have no
counterpart in absorbance; but because fluorescence provides a
positive signal against a near-zero background it can be much
easier to measure than absorbance. In fluorescence, the
absorbance of exciting light follows the Beer-Lambert Law and is
potentially a source of distributional error; fortunately, good
measurements with many flu oro chromes can be made at such low
optical densities of excitation that distributional error is
essentially non-existent. In flow cytometry one can assume that
excitation is proportional to stain content and fluorescence is
proportional to excitation.
A simplified schematic of a flow cytometer is shown in Fig.

3. Fluorescently stained objects in suspension flow rapidly
through an exciting beam of light and the resulting burst of
fluorescence is collected by a lens, detected by a
photomultiplier and measured by a pulse-height analyzer. Signals
are proportional to stain content and typical rates of
measurement are 1000 objects per second.
In flow sorting (Fig. 4), objects are measured the same way,
but those with predefined fluorescence (or other) values can be
separated from the rest of the popUlation. The preferred way to
accomplish this is to have the flow stream break up into droplets
Prolit.r.ting cells
with
f l u _ t DNA stain
Photomultiplier
PuIM
height
-Iyzer
'~I'lL
DNA/cell
Exciting
1_ be.."
Fig. 3. The basic flow cytometer. For flow cytogenetics replace

proliferating cells with disrupted metaphase cells and
DNA/cell with DNA/chromosome. From Van Dilla and
Mendelsohn 9 •
soon after passing through the measuring beam. The few droplets
likely to contain the desired object are electrically charged,
causing them to be displaced as they subsequently fall between a
pair of highly and stably charged deflection plates. Using
positive and negative charging of droplets, two classes of
objects can be sorted at rates of 1000 measured objects per
second.
Flow cytometry and sorting of chromosomes requires abundant

mitotic cells, but is otherwise remarkably straightforward.
Cultured cell lines are harvested after blocking the cells in
metaphase with colcemid. The cells are treated with hypotonic
buffer and are briefly sheared to disrupt the cells and suspend
the chromosomes. A DNA fluorochrome is added and the suspension
is entered into the flow system without purification or
manipulation. 10
Flow cytometric results with one line of Chinese hamster

cells are shown in Fig. 5. 11 Each peak represents a single
chromosome or homologous pair of chromosomes. The location of
Proliferating cells
with
fluorescent DNA stain
Photomultiplier
Pulse
processing
electronics
Pulse
height
analyzer
Decision logic,
droplet
charging
#~·lL
DNA/cell
o
o ~-
r=
+\i _
-~~I-
Remaining
o
I (;) I~'- Sorted cells
~.~
Fig.4. The basic flow sorter. From Van Dilla and Mendelsohn 9 •
the peak indicates mean stain content, the area of the peak
indicates relative abundance of the constituents, and the shape
of the peak indicates variability of the measurement.
Corroborative experiments have confirmed the identities shown in
the figure; such experiments include sorting the chromosomes from
individual peaks, followed by visual identification, as well as
measuring corresponding metaphases by scanning photometry to
Peak Mun Are. Chrom
A 1.00 0.96 1
B 0.95 0.97 1
1.92
1110 C
0
0 .84
0.53 2 .06
2.2
4.4
1110 E 0.52 1.03 t(X.5)
..., ON F
G
0.45
0.38
0.97
2.22
5
(6.6)
CD
E H 0.36
0.33
1.99
0.97
(7 .71
(V)
...,0 t(X,5)
I
J 0.26 2.09 8,8
0 8 K 0.26 0.96 Ml
E M1 8 7 E L
M
0.22
0.21
1.08
0.89
9
9
0~
76
K~ J
N 0.16 2.09 (10.10)
~
H6 4 0 0.15 2.00 (11 .11)
....U0 G 4
P 0.13 0.85
23.0
(M2)
M2 99
P
~
CD
J:I ML 0 2
E Y 2
:I I C
z
Chromosomal fluorescence
Fig.5. A flow karyotype of Chinese hamster MJ-l line

chromosomes stained with Hoechst 33258. The scale of
the ordinate is such that the peak point J represents
thousands of chromosomes. From Gray, et a1.ll.
relate stain content with chromosome type. The areas of the

peaks correspond to expectation, indicating that all chromosomes
have an equal likelihood of entering suspension and being sampled
by the measurement. The coefficients of variation of the peaks
are around 2% and are decidedly smaller than the variability
observed with scanning measurements.
Comparison of several DNA f1uorochromes in Chinese hamster

and other chromosomes has shown interesting differences. 12
Ethidium bromide, a non-banding interca1ator gives flow results
which closely mirror the ga11ocyanin-chrome alum measurements of
metaphases on slides. Hoechst 33258, a groove-binder with
preference for AT regions, produces quinacrine-like bands and in
flow measurements gives different peak means and smaller
coefficients of variation than ethidium bromide. Chromomycin A3,
another groove-binder but with GC preference, produces reverse
banding and has yet another pattern of peak means in flow
measurements.
Ethidium bromide Chromomycin

CA3
en
CD
E
5lo
E
~
...o
....o
(,)
... CYDAC simulation Hoechst 33258

~
E
:;,
z
Relative fluorescence
Fig. 6. Distributions for human chromosomes stained by four

different methods. Metaphase chromosomes were prepared
from human foreskin strain No. 706. In the CYDAC
simulation, the chromosomes are stained with
gallocyanin-chrome alum and measured on slides by
scanning photometry. In the other three panels, DNA
fluorochromes and flow cytometry are used. The two
panels on the left are almost identical suggesting that
ethidium bromide and gallocyanin-chrome alum reflect
similar properties of DNA, whereas the panels on the
right differ dramatically indicating that chromomycin A3
and Hoechst 33258 each reflect distinctive properties of
chromosomal DNA.
These differences are particularly dramatic for human

chromosomes as shown in Fig. 6. 11 Differences among several
human cell lines are also seen with anyone fluorochrome and are
akin to the differences we generally see among normal people with
scanning measurements.
For diagnostic and mechanistic studies it is highly

advantageous to measure two fluorochromes in the same
chromosome. Studies show that stain pairs such as Hoechst 33258

and chromomycin A3 compete little for DNA sites and behave with
good stoichiometry when present simultaneously.ll However,
they cannot be measured by taking advantage of their differing
emission spectra because energy transfer from Hoechst 33258 to
chromomycin A3 causes almost all emission to follow the
chromomycin A3 spectrum. Fortunately these stains can be
separated by their excitation spectra using a dual laser flow
cytometer. In this device the chromosomes flow first through a
laser beam at 351 and 364 nm and give a fluorescent signal
representing Hoechst 33258; 20 microseconds later they flow
through the second beam at 454 nm and the fluorescence from
chromomycin A3 is recorded. 12
A dual laser result with human chromosomes is shown in Fig.

7. The mountain-range effect is a dramatic indication of the
resolution achieved by the complementary stoichiometry of Hoechst
33258 and chromomycin A3. It is our current working hypothesis
that the relative contents of these two stains reflects the AT-Gt
ratio of specific chromosomes.
9- 12
~
c
·u
>
E
o
E
e
6
Hoechst 33258
"IoU. A U I' II Ii Ie AI~' un (C )1(1 I U)))lII)IU"'(

2221 Y 20 19 18 17 16 15 14 13 12 11 10 9 X 8 7 6 5 4 3 2 1
Fig. 7. Dual laser and dual fluorochrome flow cytometry of human

chromosomes. Below are shown the 24 types of human
chromosomes. On the upper left is a two-dimensional
distribution of Hoechst 33258 and chromomycin A3
fluorescence of human chromosomes. The tendency for
these stains to show some negative correlation is seen
best in the upper right where a topographic
representation of the dual fluorescence is shown.
354 M.L.MENDELSOHN
The ability to sort chromosomes can be of enormous

importance. Purified chromosomes are suitable for biochemical
analysis, template activity, antigenicity and biological
transduction. They make possible critical studies on linkage,
chromosome structure, mutagenicity, gene control, evolution and
genetic engineering in a variety of species including man. For
well separated peaks, purities of 95% are available, and with
present rates of flow analysis, an 8-hour sort of Chinese hamster
chromosomes can produce 2.5 x 10 6 purified chromosomes. For
the average-sized chromosome, this amounts to a microgram of DNA,
a quantity that requires the utmost care in handling and is
suitable for only the most sensitive techniques of analysis.
Ongoing studies on sorted chromosomes include comparison of
histones on different chromosomes, localization of viruses or
genes to a particular chromosome and attempts to use purified
chromosomes as antigens.
THE FUTURE
Future studies in quantitative cytogenetics will be oriented

toward improving and applying the scanning and flow analysis of
chromosomes. In scanning, a major thrust will be the development
of fluorescence scanning to take advantage of banding and of the
interesting affinities of fluorochromes for DNA. The nature of
chromosomal deviants and their heritability needs further study,
and attempts will be made to apply quantitative analysis of
metaphase cells to relevant clinical problems.
In flow, we will continue the exploration of DNA

fluorochromes, alone and in combination. The related
cytochemistry of nuclear proteins will also be pursued. Methods
are sorely needed to permit flow cytogenetics in other than
cultured cell lines; we particularly need methods for cultured
blood and tissue biopsies in the human. An active program is
underway to use slit scanning to locate and count chromosomal
centromeres. Finally, there are many potential and exciting
applications of flow sorting of chromosomes to important problems
in biology and medicine.
Work performed under the auspices of the

U.s. Department of Energy by the Lawrence
Livermore Laboratory under contract number
W-7405-ENG-48.
This report was prepared as an account of

work sponsored by the United States
Government. Neither the United States nor the
United States Department of Energy, nor any of
their employees, nor any of their contractors,
subcontractors, or their employees, makes any

warranty, express or implied, or assumes any
legal liability or responsibility for the
accuracy, completeness or usefulness of any
information, apparatus, product or process
disco1osed, or represents that its use would
not infringe privately-owned rights.
REFERENCES
1. Caspersson, T. Cell Growth and Cell Function, Norton, New

York, 1950.
2. Caspersson, T., Zech, L., Johansson, C., and Modest, E.J.

Identification of Human Chromosomes by DNA-Binding
Fluorescent Agents. Chromosoma 30, 215-227, 1970.
3. Mendelsohn, M.L., and Mayall, B.H. Chromosome Identification

by Image Analysis and Quantitative Cytochemistry. (In) Human
Chromosome Methodology (J.J. Yunis, ed.), Academic Press-,----
311-346, 1974.
4. Mayall, B.H., and Mendelsohn, M.L. Errors in Absorption

Cytophotometry: Some Theoretical and Practical
Considerations. (In) Introduction to Quantitative
Cytochemistry, Volume 2, (G.L. Wied and G.F. Bahr, eds.),
Academic Press, 171-197, 1970.
5. Mendelsohn, M.L., Mayall, B.H., and Perry, B.H. Generalized

Grayness Profiles as Applied to Edge Detection and the
Organization of Chromosome Images. (In) Advances in Medical
Physics, Second International Conference on Medical Physics,
Inc., Boston, 327-341, 1971. .
6. Mendelsohn, M.L., Bennett, D.E., Bogart, E., and Mayall,

B.H. Computer-oriented Analysis of Human Chromosomes. IV
Deoxyribonucleic Acid-Based Centromeric Index. J. Histochem.
Cytochem. 22, 554-560, 1974.
7. Mayall, B.H., Carrano, A.V., Moore II, D.H., Ashworth, L.K.,

Bennett, D.E., Bogart, E., Littlepage, J.L., Minkler, J.L.,
Pi1uso, D.L., and Mendelsohn, M.L. Cytophotometric Analysis
of Human Chromosomes. (In) Automation of Cytogenetics (M.L.
Mendelsohn, ed.) Asilomar Workshop, 30 Nov. - 2 Dec., 1975,
ERDA Conf-751158, NTIS, Pacific Grove, Ca., 135-144 1976.
8. Gray, J.W., Carrano, A.V., Steinmetz, L.L., Van Di11a, M.A.,

Moore II, D.H., Mayall, B.H., and Mendelsohn, M.L.
Chromosome Measurement and Sorting by Flow Systems. Proc.
Nat. Acad. Sci. U.S.A. 72, 1231-1234, 1975.
9. Van Dilla, M.A., and Mendelsohn, M.L. Resume (In) Flow

Cytometry and Sorting (M.R. Melamed, P. Mullaney and M.L.
Mendelsohn, eds), Wiley & Sons, expected 1978.
10. Carrano, A.V., Van Dilla, M.A., and Gray, J.W. Flow
Cytogenetics: A New Approach to Chromosome Analysis. (In)
Flow Cytometry and Sorting (M.R. Melamed, P. Mullaney and
M.L. Mendelsohn, eds), Wiley & Sons, expected 1978.
11. Gray, J.W., Langlois, R.G., Carrano, A.V:, and Van Dilla,
M.A. High Resolution Chromosome Analysis: One and Two
Parameter Flow Cytometry. (in preparation).
12. Jensen, R.H., Langlois, R.G., and Mayall, B.H. Strategies

for Choosing a Deoxyribonucleic Acid Stain for Flow Cytometry
of Metaphase Chromosomes. J. Histochem. Cytochem. 25,
954-964, 1977.
13. Dean, P.N., and Pinkel, D. Dual Laser Flow Cytometry. J.

Histochem. Cytochem. 1n press, 1978.
DISCUSSION (PART II)
DR. YAGIL: Why is it necessary to add~. coli polymerase. Could

not endogenous polymerase be responsible for transcription?
Also, can one obtain specific hybridizable 32p labelled
globulin mRNA in intact foetal liver cells?
DR. GILMOUR: With endogenous pol,ymerase we obtain amounts of

messenger too smal,l, to be used for our hybridization experiments.
There is evidence that~. col,i pol,ymerase transcribes sequences
that are not transcribed in intact nucl,ei from endogenous
pol,ymerase, but it remains a probl,em to make hybridization
experiments starting with this system.
DR. SARMA: Anti-globin RNA can arise by copying the contaminating
endogenous globin mRNA or anti-globin gene. What evidence
exists that globin mRNA serves as the template and not anti-
globin gene? Can Hg-UTP by reacting with major groove (where
his tones are believed to occur) enable copying anti-globin
gene which is not normally copied?
DR. GILMOUR: The copy of the message is not a compl,ete copy. It

"looks l,ike what attracts the pol,ymerase is the pol,yA 3' end,
but it does not go very far in copying the structural, RNA.
DR. WLNICOV: In view of your model which invokes an imbalance of
two or more chromosomes in expression of a malignant state,
would you comment on the cell fusion experiments from the
laboratory of Dr. Carlo Croce, which seem to propose the
involvement of a single chromosome and also show that in each
case it does not have to be the same chromosome?
DR. SACHS: I do not think that such resuZts invaZidate the

resuZts of our experiments. In fusion experiments between
maZignant and nonmaZignant ceZZs you come out with maZignant
or nonmal,ignant progeny apparentl,y depending upon which
chromosomes are sel,ected for.
357
358 DISCUSSION
DR. PARODI: Has somebody tried to hybridize cells which had

incorporated large amounts of BUdR into their DNA with an
untreated, different (as in differentiation) cell and there-
after tried to destroy the BUdR-DNA with U.V.? If the nucelo-
proteins alone maintain differentiative potentialities we
could perhaps detect this fact.
DR. SACHS: That is a good experiment; at the best of my kYwwledge

I do not know anybody who has done it.
DR. NICOLINI: Do you not think that in the complex mammalian

chromatin "in situ" we should think not only in terms of
static genes but also in terms of the dynamic chemical-
electromagnetic environment?
DR. SACHS: You aPe asking for differentiating what is due to

genes and what is not due to genes. It is important -
however, it is difficult to do that.
DR. DIXON: An apparently clear finding of the physical techniques

is that in 2M salt when the DNA is dissociated from the histone
core there is very little change in CD. This seems difficult
to understand if the N-terminal tails of the his tones have
any sort of ordered structure in the nucleosome.
Would you like to comment on which sort of structure you

might visualize the N-terminal tail as having in the intact
nucleosome and if the structure reverts to a random coil in
2M salt, why is there no reflection in the CD spectra?
DR. FASMAN: The conformation of the N-terrmnal tait either

in complex with DNA in the nucleosome or after dissociation
is not known.
DR. BRAM: Dr. Fasman, your binding studies with acetylated

his tones are very interesting. On the other hand, you were
careful to point out that changes in CD could arise either
from a higher order "tertiary fold" or from changes in
secondary structure. However, you presented the classical
canonical A, B, C, scheme. Unfortunately DNA structure
has been shown to be far more complicated than A, B, C. The
A conformation is the only form of DNA which is more or less
unique. For B my X-ray studies in solution and in chromatin
show that they are families of structures where the helical
parameters vary by about 15%. That means the angle between
base pairs can change by up to 50% per b.p. In fact, Ivanov
et al (Mol. BioI. 58, 277 1971) have shown that most changes
in CD can be explained by a variation in angles between base
pairs. Consequently it seems to me that all changes in CD
DISCUSSION 359
you report could well be interpreted by a modification of DNA

secondary structure.
DR. FASMAN: I agpee on what you said.

DR. YAGIL: You demonstrated that two different condensed forms of
DNA both verified by X-Ray as being in the B conformation have
very different C.D. spectra. This indeed takes the basis for
C.D. to be indicative of conformation in the condensed state.
How can be extrapolate from the condensed to the dissolved
state, when C.D. is not unique in condensed phase?
DR. FASMAN: I think thepe ape two types of aondensation: one

that invoZves ahanges in the C.D. speatpa ppobabZy affeats the
seaondaPy stpuoturoe and otheps, opepating on a Zapgep saaZe
(soZenoid,supepheZix), that pephaps do not affeat signifiaantZy
the seaondary strouaturoe.
DR. SA~: Chemical acetylation does not alter the C.D. spectra
of chromatin, whereas such acetylation alters the transcription
ability and DNase I digestibility. Do you want to comment
on this?
DR. FASMAN: It is a disapepanay. ObviousZy, the two teahniques

ape measuroing diffepent aspeats aonaeroning the stabiZity
of the strouaturoes, but I have no expZanation of this obserovation.
DR. HARRINGTON: Are any other HI modifications observable in
the histone from C.D. spectra?
DR. FASMAN: Yes. AaetyZation shows simiZaro effeats to that

obseroved with phosphopyZation.
DR. BETTECKEN: Are the holes in the nucleosomes as visible in
E.M. micrographs or are they artifacts?
DR. OLINS: stain binds to negative ahapges, thepefope you see

negUgibZe stpuaturoe. Thepe might be a deppession, but not
weU visibZe.
DR. BETTECKEN: Did you try E.M. microsamples of histone-enriched
nucleosomes?
DR. OLINS: No.

DR. MARLADI: Do you mean that the dense spot inside the nucleo-
some you see in your picture is a real depression as indicated
by scanning - tranmission E.M. (quoted) and that this depression
could be interpreted as the presence of a low density protein
core?
360 DISCUSSION
DR. OLINS: Yes~ it may be interpreted as a Zow density region

due to the ZocaUzation of histone just in the center of the
nucZeosome particZe.
DR. VAN HOLDE: Is there evidence that ribosomal genes that show
chromatin structure in the E.M. also digest to give the 200
base pair repeat?
DR. OLINS: No.

DR. TRIFONOV: Are the rope like helical structures that you
showed in some sperm chromatin made of some material different
than histones?
DR. OLINS: I honestZy don't know what the protein is in the sZides
I showed. You are right in that the protein need not be
histone; our exampZe of this is the sea urchin spePm.
However~ I presented this data primariZy to show how another
method of eZectron microscopy can handZe higher order structures.
DR. BONNER: What is structure of the chromatin strand in that
portion of Bombix chromosome which is being transcribed into
fibrin message? (the EM picture presented was by Steve McKnight)
DR. OLINS: The poZymerases are cZoseZy packed and it is

impossibZe to identify nucZeosomes in the transcription units.
DR. DIXON: You mentioned that in transcribing regions of
Oncopeltus chromatin nucleosomes could not be seen, but in
Drosophlia they could and you suggested that even where they
could be seen where there were still 200 base pair repeats
to micrococoal nuclease as a probe as well as sensitivity to
DNA-ase I (in the core region presumably). How were these
experiments done? Is it possible to assess whether the
nucleosome structure is absent or very unstable? Is it
possible to stabilize the nucleosomes that are transcrible
or committed to transcription by chemical crosslinking before
EM?
DR. OLINS: The chromatin is aZready fixed in formaZdehyde but

the Oscar MiZZer technique does tend to extend chromatin
and it couZd be particuZarZy ZiabZe to extend especiaZZy
ZabiZized nucZeosomes in transcribing regions. It wouZd be
nice however to find a technique for stabiZizing transcribing
regions of chromatin.
DR. TRIFONOV: The change of sedimentaiton coefficient of
nucleosome in acidic media is it not just a result of denaturation
and collapse of DNA?
DISCUSSION 361
DR. D. OLINS: The JJompaPative study on DNA that lUaS done lUaS uYith
ZaPge size DNA and not 140 base size DNA. However it is
known that acidic denaturation of DNA takes pZace at considerabZy
Zower pH and aZthough we have to maintain reservation in
extrapoZating to 140 base size, I think that probabZy DNA is
not denaturated.
DR. SHAW: You indicated that the nucleosome cores that you used
for your physical studies were enriched in 140 base pair DNA
by precipitatation with KCl to remove 160 base pair particles.
In our laboratory we are not able to remove all 160 base pair
particles using similar salt precipitation methods. There
still remain considerable amounts of 160 base pair particles
that lack HI and H5. Exactly how pure were the cores that you
used? If they contained 160 base pair contaminants could
that account for the changes in sedimentation coefficient you
observed by varying the ionic strength since there were extra
tails of DNA?
DR. D. OLINS: Yes there were considerabZe amounts of 160 base pair
particZes in the prepaPations used, as seen by geZ eZectrophore-
sis. We now use the method of H1 removed first (as adapted
by TatcheeZ and Van HoZde) to obtain cZean 140 base pair
cores. We have not used these in the studies above. It is
quite possibZe that the extra 20 base pairs in the cores
couZd infZuence the sedimentation changes with vapying ionic
strength.
DR. PRUNELL: Was HI present in your nucleosomes?
DR. HARRINGTON: No. We worked uYith core paPticZes.

DR. BRAM: Your calculations suggest that banding of DNA is
energetically favorable. Some time ago J. Brahms and I
built CPK models with a flexible center support and found
that the force of gravity was sufficient to band DNA into a
conformatioy with an equivalent radius of curvature equal
to about 30A.
DR. HARRINGTON: I perfectZy agree if the charges are neutraZized

in the histones.
DR. BETTECKEN: What evidence does exist for binding of H2A and
H2B to undenatured portions of chromatin during thermal
denaturation?
DR. WALKER: H2A and H2B do not bind to denatured DNA. During the
first haZf of the ther.maZ transition, H2A and H2B dissociate
and rove onto the undenatured haZf of chromatin. Denaturation
of chrorratin is therefore "aZZ-or-none". The binding of H2A
362 DISCUSSION
and H2B to the native ahromatin inareases the 1m of the seaond

step. On aooling from 1m the transition is not reversible but
the fuZZ hypoahromism is X'egained. The denatured fraation rrny be
separated from the native fraation and is found to aontain only
H3 and H4.
DR. GIARETTI: The use of the parallel axis theorem relationship
to calculate the distances of subunits in a complex has been
criticized (Jacrot; Rep. Progr. Phys. 39 (1976) 911). Could
you please comment on this?
DR. PARDON: I am not aware of this aritiaism.

DR. PIXON: Your model showed bands of protein crossing over the
DNA gyres in eight places - do you consider these to be the
N-terminal fingers?
DR. PARDON: Yes, but not definitely - this is only one possible
way that the N-terminal fingers aould interaat with the DNA.
They aould also be between the gyres of DNA in a somewhat
indefinite position.
DR. WILHELM: How did you obtain the H2a, H2b, H3 , H4 tetramer?
Have you been able to study the octamer? and the (H3H4)2
tetramer?
DR. PARDON: By extraating ahrorrntin with 20M NaCl, pH9.0

essentially using the methods originally desaribed by both
Weintraub et al and Thomas and Kornberg. We have examined
an oatamer formed from tetramers by akemiaal aross linking.
The radius of gyration (Rg) of this oatamer is very similar
to both the (Rg of the isolated histone) tetramer and the
aore partiale under aonditions where the protein dominates.
We have not studied the (H 3H4 ) 2 tetramer.
DR. T~IFONOV: What kind of further detail could give maximum of
35A in scattering curve of histone core (at 65% D20)?
DR. PARDON: Possibly some distribution of a-heliaes in the

histone aore.
DR. HARRINGTON: If you have a polydisperse system in your

neutron scattering work on core histone complexes, what
kind of average Rg does the neutron scattering method yield?
A z-average?
DR. PARDON: Probably a weighted average, but the Rg value ahanges

very little between oatamer and tetramer and also is independent
of aonaentration.
DISCUSSION 363
DR. FASMAN: Have you any evidence that crosslinking doesn't

change the conformation of the his tones (in the octamer)?
DR. PARDON: No evidence - S values are approximately J.? to 4.2.
DR. HARRINGTON: Are you aware of any applications of

"magic angle" spinning techniques for nucleic acids
or nucleoproteins?
DR. TS'O: These are not yet sufficiently well developed to

be applicable here.
DR. BETTECKEN: What solvent conditions did you use for
your nucleosome examinations?
DR. TS'O: What is the ultimate limit of resolution of

the cytochemical analyses by light microscope?
DR. CASPERSON: O. J micron.

DR. LINDEN: What is the resolution in the TV-system for
fluorimetry?
DR. CASPERSON: Up to now we have been 1,'I()rking with photographs

from chromosomes using the quinacrine fluorescent binding
technique. These photographs are viewed by the TV system.
We hope to be successful in developing a TV system for
fluorometry with good resolution.
DR. LINDEN: You mentioned the use of quantitative inter-
ferometry. Would you comment on the value of this technique
compared to the use of cytochemical stains with absorption
or fluorimetric techniques?
DR. CASPERSON: This method still has its value because of

its higher accuracy and reproducibility. We just did
some studies on the protein content of paramecium with
a reproducibility of 2 - J% which is higher than what you
get in normal cytophotometry.
DR. DIXON: I was intrigued by your finding that you could
separate the two homologues of chromosome I and 9 in
your line of CHO cells. Would it be correct to assume
that if you looked at other lines you might see for
example, two identical I chromosomes of one class or
two of the other class assuming that the differences
are of paternal or maternal origin.
364 DISCUSSION
DR. ~NDELSOHN: Yes, one would expect such a situation but it

has not been systematically looked for. The CHO line (3-1)
is about ? years old and may be from the same source as the
CHO line of Stubblefield in which he has shown a 6%
difference in the chromosome 2 homologues by 3H hymidine
incorporation.
DR. PARODI: Could the fact that you can distinguish between
chromosomes of the same couple be related to the Lyon
phenomenon?
DR. ~NDELSOHN: It is a possibility that we have to keep in

mind.
DR. KOVAL: Do you think a potential exists for eventually
detecting damaged areas of DNA or chromatin with your
machine as you have been able to resolve 10 6 base pairs
already?
DR. ~NDELSOHN: Perhaps. I don't know. We would have to get

the damages to fluoresce.
DR. SAWICKI: What is the contribution of cytoplasmic DNA
for chromosome measurement by pulse-cytophotometry.
DR. MENDELSOHN: It may cause a very slight elevation of the

background on which the chromosome peaks are sitting. The
effect would not produce peaks in the chromosome region.
DR. BAISCH: There have been problems in flow systems with
particles deviating from spherical shape. Chromosomes are
considerably different from spheres. Do you have
problems from orientation of chromosomes while being
measured?
DR. ~NDELSOHN: This refers to experiments that have been done

with several cell types. We have done it with sperm and the
Hertzenberg group has done it with red cells. When orthogonal
flow systems are looking at flat cells the signal depends on
how the cell is oriented. The operation one gets with a
flat spep,m cell is the most extreme degree of non-randomness
that we have available. It can be as much as twofold. It is
an interesting artifact and involves both the excitation and
environmental parts of the phenomenon. With chromosomes we
have information from high speed photographs that they line
up in the flow direction. So We are talking about whether the
two chromatids are in line as they look at the laser beam or
whether they are at right angles. And this is probably going
DISCUSSION 365
to be a very small effect. It is down to the resolution of the

beam and we don't see the effect in the histogram, so my guess
is that it is not present.
DR. KOVAL: Concerning Janet Rowley's data, is it possible that
a chromatin structural deficiency exists such that the
chromosome is susceptible to breakage at a certain point?
DR. MENDELSOHN: Perhaps but it could also be that this certain

phenotype results in the leukemia rather than vice-versa.
DR. BETTECKEN: Does it make sense to continue working with cells
characterized and sorted in your flow cytometer, since the
handling inside the instrument is very rough (staining, shear
forces, exposure to high intensity light and high voltage)
and might destroy the cells?
DR. MENDELSOHN: Experiments with stain Hoechst 33342 show that

cells sorted by the flow-cytometer grow as well as untreated
cells.
DR. NICOLINI: We have done the same in Melanoma B16, with low
concentration of acridine orange. It works fine.
DR. DIXON: When you see that single red cells "light up" with a
specific antibody to HbS, do you really know that the red
cells contain HbS or merely that it contains a mutant HbS
which cross-reacts with an anti-HbS antibody? In other words
the mutation need not giving rise to identical HbS which
alterates B chain Glu + Val but merely gives rise to a
structure cross-reacting with anti-HbS antibody.
DR. MENDELSOHN: Yes, there could be more distal interactions

which effectively produce a mutant Hb which could cross-
react with anti-So
DR. LINDEN: You mentioned your method of screening for carcinogens.
I suppose you use fluorescence activated chromosome sorting.
Would you please comment on that in some more detail.
DR. MENDELSOHN: We are looking for trans locations. But we hope

to apply the same technique for use in sickle cell anemias and
hope to apply it to humans in the fUture.
DR. SA~: Do you have any data in identifying preneoplastic
cells in a normal population of cells.
DR. MENDELSOHN: We are working on these problems. For instance,

we are working on Farber's model of inducing nodules and looking
for cells by gamma-glutamyl transpeptidase activity.
366 DISCUSSION
DR. SARMA: Today many speakers were using the term "chromatin
in situ". In fact in some of their studies the chromatin was
hydrolyzed, possibly depurinated and been molested, if I may
be permitted to say so. I think this term "chromatin in situ"
is inappropriate to these studies.
DR. NICOLINI: This is true only in oup image analysis studies. But
this is merely the first step. For what concerns the semantic,
I think that the term "Chromatin in situ" aan be considered
sufficiently aa~te since Feulgen's reaction reflects DNA
space and location, in situ.
DR. DIXON: Since the Heisenberg uncertainty principle it has

always been a problem for scientists that the method of obser-
vation may perturb the system during the observation. Worcel
observed that ethidium bromide could dissociate his tones from
chromating by intercalating and unwinding the DNA. Do you
feel that your method of binding ethidium bromide and acridine
orange might not perturb the chromating being observed in situ?
DR. NICOLINI: We don't think that the unwinding is a serious

problem at the moment although we are aware that intercalation
of ethidium bromide does induce unwinding. We must start
somewhere and later we will explore this problem.
DR. SARMA: You correlated some changes observed by image
analysis to increased transcriptional ability of the cells.
This maybe true. My question is: What type of changes
would you observe when the cells have been treated with
carcinogens of a type that actually inhibits transcriptional
ability? Maybe you will find similar changes. The reasons
for asking this question is that carcinogen treated rat liver
chromatin exhibits similar CD spectra to those of replicating
rat liver chromatin-DNA - even though the carcinogen-treated
cell has less transcription when contrasted to the replicating
cell.
DR. NICOLINI: In fact we have correlated differences in average

optical density with both increased transcriptional activity
(as in WI~38 cells stimulated to proliferate) and decreased
transcriptional activity (as in stationary 2RA cells when
compared to quiescent WI38 cells). Experiments of the natupe
that you suggested have yet to be completed but these are
some of the directions in which we are working.
DR. ~NDELSOHN: In the two-dimensional density reconstruction
of nuclei, there were spikes that looked to be one picture
point wide. At a sample spacing of O.l~m, the modulation
transfer function of the microscope would not permit optical
responses of such high frequency. Can you comment?
DISCUSSION 367
DR. NICOLINI: Coroe size constmints imposed by OU1' minicomputero

roestroicts us to a maximum dispZay ar>r>ay size of 64 x 64.
Consequently the dispZays ~eroe obtained ~th much lo~ero
magnifications than those ~hich ~e employ foro analytical ~orok
(~hen a single pictU1'e point roeproesents an interval less than
0.1 jJ1Il). This lo~ magnification data is subject to senous
distroibutional er>r>oro. We use it only foro display purrposes.
DR. ~NDELSOHN: What was the period length in microns of the
sinusoidal correlations seen in GI and S cells? What was the
relation to 60 cycles per second?
DR. NICOLINI: The peroiodicity of the mid-Gl nuclearo image

cororoesponded to 13 microa and in the S-phase image to above
4.2 microa. A single column scan roeproesents about 4Msec.
Inasmuch as the system clock was independent of the po~ero line
and the roesulting frome roate was not integroally roeZated by
multiple oro submultiple of the po~ero line, any signal
produced by this SOU1'ce ~ould tend to droift thr'ough the image
on roepeated scans. Each aroroay was in fact scanned 100 times
and the roesults avemged, so such effects ~ould be negligible.
It ~uld be faro moroe likely that such arotifacts ~uld roesult
from mechanisms controolled by the system clock. The fact that
the peroiodicities observed ~eroe consistently dependent upon
cell cycle phase roathero than equipment set-up on a paroticularo
day indicates that they aroe probably not arotifactual.
DR. BETTECKEN: Did you consider the elctric field caused by the
power lines? Did you shield?
DR. CHIABRERA: The wave foram s~ in the slides is the meaSU1'e-

ment of the rootational component of the electnc field,
~hich prooduces the cUr'r'ent in the solution. Then the contnbu-
tion of the po~ro line is negligible. In any case, by comparoing
the behavioro of control cells ~ith the ir>r>adidated ones
changing the amplitude and the roepetition roate of the signal,
any 60 Hz effect can be excluded. We checked the effects of
a shield, and ~e concluded that it was not necessaPy noro
useful. But in fact, eddy CU1'roents may be induced in the
shield by the signal itself, thus affecting the actual
electroic field ~veforam in the solution.
DR. TS'O: Who makes the final decisions about the positive
or negative diagnoses?
DR. PLOEM: The pathologist.

SUB J E C T I NDEX
AAF, 783-798 Blastular, 42

Absorption Boltzmann Constant, 181
flattening ~ 76 Bromodeoxyuridine, 756
spectra~ 252, 631 B16 Melanoma Tumor, 320
Acetylation, 125, 458, 492, 556, Butyric Acid, 727
563,625,627
Acridine Orange, 301-321, 641 Calcium, 502, 522, 814
Actinomycin D, 52, 712 Calf Lymphocytes, 467
Activation, 503 Cancer
Active Fibers, 662 initiation~ 752
Activity Coefficient, 152 research~ 257
Adsorption, 812 Carcinogen, 727, 781, 803
AF-8, 632, 641, 647 Carcinogenesis, 644, 722, 751,
Affinity Region, 6~6 762, 771, 803
Aging, 642, 673, 730 CD Spectra, 70, 74, 76, 146-154
Alkylation, 752, 775 628, 630
Anchorage Independence, 760 eDNA, 43, 49
Angstrom to Micron Level, 613 Cell
Annuli, 661 cycle, 324, 492, 622, 669,
Area (Nuclei) Versus OD Threshold, 736
275, 674 cycle phases, 634
Argenine, 146-160 migration, 709
Aromatic Amino Acids, 82 morphology, 760
Asymmetric Bindings, 422 proliferation, 635
ATPase, 757 sorter, 320, 350
Automated Image Analysis, 265, surface, 811
343, 620, 668 Cellular Aging, 667, 675, 678-
Autoradiography, 620 679
Average Optical Density, 273-274, Center of Mass, 278, 743
621, 634, 659, 677 Centromeric
index, 345
Bacteriophage regions, 531
T2~ 516 Chain Stiffness, 170
T?~ 615 Chemical Shifts, 228, 452
BALB/3T3 Cells, 725 Chicken Erythrocyte
Banding Pattern Analysis, 264 core protein, 200
Beads-on-a-string, 10, 390, 564, nuclei, 522, 524
613 Chinese Hamster, 351, 531
Beer-Lambert Law, 343 Chironomous, 42
Benzopyrene. 758 CHO Cells, 736
@-PiLeated Sheet, 76 Chromatid, 345
~Turn Conformation, 76 Chromatin
Bessel Function, 189 composition, 5, 645
Binding, 137 fractionation, 15, 541
curves~ 140, 148 melting, 631
equilibrium~ 137 nuclear, 265
sites~ 149, 299, 309, 619, 645, preparation, 3, 41, 517, 544
774 reconstitution, 561, 628
xxi
xxii INDEX
Chromocenter, 252 Crystallography, 188

Chromosomes, 341-356, 520 Csoionic Point, 423
abnormaUty, 709 C-terminal, 159-161
banding, 264, 351 Cyanosin Bromide, 460
insertion, 252 Cyclic AMP, 493, 499, 503, 733
Chymotrypsin, 460 Cylindrical Structure Particles,
Circular Dichroism, 67, 107, 115- 201
117 Cytochemical Studies of Nucleus,
aeZZ ayaZe ahromatin, 622 251, 294, 324
DNA, 68 Cytochromes, 32
high moZ. wt. ahromatin, 80 Cytometry (see mi~ofZuorometry
of his tones , 76-77 and image anaZysis)
RNA, 77, 628 Cytopathology, 263
saZt effeats, 6, 89 Cytoplasm, 261
shearing effeats, 83, 630 Cytosine Abrabinoside, 712
speatra, 70, 74, 76, 116, 146-
154, 637 D20, 195, 196
Clonal Morphology, 758 Debye Equation, 201, 478
Clone, 723 Decondensation of Chromosome, 496,
.Cloning Efficiency, 731 522
Coefficient of Variation, 270 Dedifferentiation, 732, 818
Coherent Scattering, 192 Densitometer, 265, 620
Colony, 723, 740, 760 Densitometry, 265, 734, 740
Commitment Theory, 727 Derivative Plot, 629
Computer Deuterium, 192
anaZysis of aeZZ ayaZe, 226-227 Development, 42, 672-673, 752
dPiven miarosaope stage, 263 Dibutyrl-cAMP, 736
Computers, 255, 265, 294 Dictyosteliom, 43
Coomassie Blue-R-250, 128 Differential Light Scattering,
Cooperative 75, 617-618, 633, 649
binding, 150 Differentiation, 705, 732, 772
dissoaiation, 158 Diffraction, 187, 210
Concanavalin A, 710 rna:r:ima, 210
Confluent WI-38 in Phase I, II, pattern, 189
III, 647, 675, 678 Digestion, 389
Contrast Variation, 191-192, 473 DNAse I, 402, 551
Core Particle, 195, 197, 212 DNAse II, 549
asserribZy, 161, 399, [:·82 miaPOaoaaaZ, 390, 544
arystaZs, 210 Dimer, 78, 157, 385
preparation, 125 H2A-H2A, 374
sizes, 391-392 HJ-H4, 376
thermaZ denaturation, 400 Dipole Moments, 76
COT, 765 Dish-like Structures, 198, 201
anaZysis, 46 Dispersion (Chromatin), 274
aovette-type /Zow aeZZ, 173 Dissociation of H3 & H4, 143
Covalent Binding, 781 Distributional Error, 269
Cross Divalent Cation, 57
aorreZation, 278, 288 DMN, 645, 771
Zinking study, 143, 164 DMSO, 712
INDE.X xxiii
DNA Elongation, 599

A foY'/7/, 69 Endogenous RNA, 44, 46
B foY'/7/, 69, 81 foetal livers, 45, 48
C foY'/7/, 70 Enthalpy, 143-152
lji, 71, 74 Entropy, 149
CD spectrwn, 80 Epigenetic, 188, 752
distribution, 254-255 Epithelial Cells, 755
fragments and reconstitution, Erythrocyte, 134
419, 422 histones, 51
phosphates, 72-73 Erythroid, 44
polyoma, 138 Equilibrium
repair, 803-808 binding stUdies~ 137-166, 297-
replication, 663 299
RNA hybridization, 42 Kapp~ dissociation constant~
sequences, 441 142-148
spacers, 397, 649 Keg~ constant, 144, 148
synthesis, 635 Ethanol, 74
to H3 binding, 143, 155 Ethidium Bromide, 128, 296-304,
DNAse I, 129, 402-407, 416, 524, 351, 608, 619, 646
552-554 chromatin complex~ 622
Double Helix, 187 Ethylene Glycol, 74
Drapery-like Chromatin Structure, Euchromatic, 42, 865
278, 660 bands~ 252
Drosophi1ia, 42, 52 S-phase~ 44
Drying (effect of), 533 Exclusion Chromotography, 458
Dry-mass Determination, 256-257 Excitation Spectra, 257
Duck Reticulocyte, 51 Exothermic Reaction, 145
Dyad axis, 109 Expression, Gene, 15, 561
Extinction, 253
Early G1, 272, 634 angle~ 168, 174-175, 177
EB Binding Sites, 632, 636, 646 Extraneous Birefringence, 174
Ehrlich Ascites
ceUs, 502 Feret diameter, 266
kinase, 499 Feu1gen, 265, 323, 342, 620,
Electrical Noise, 269 670, 732
Electrochemistry, 811 acrifiavin~ 329
Electromagnetic Wave, 641, 811 optimal hydrolysis, 323, 330
Electron Microscopy, 31, 621 UMSP, 255
effect of drying, 523 Fibre a 34, 187-188, 207, 213
effect of fixation, 522 100A, 162, 164
general criteria~ 3l-40 300~, 519, 652
wet replica, 516-517 1~500A, 287, 520
Electron scattering, 251 effect of shearing, 632
Electrophoresis, 138 predicted width, 632, 655-656
polyacrylamid gel, 125 Fibrinolytic Activity, 758
two dimensional, 128 Fibroblast, 755, 633, 706
Electrostatic Bonds, 160 Fibroia, 43
Elliptically Polarized Light, 67~ Filtration Chromatography, 154
171 Fixation, 522
xxiv INDEX
Flow Granulocytes, 707

birefringence, 167-179 Gray Level, 270
cell sorting, 328 Groove, 191
cy tome try , 313-321, 323
341, Grm.,th, 254, 258
621 '
associated kinase, 499
Fluorescence, 257, 264, 293, 640, Guinier
731
lClhJ, 198, 205
Fluorescent Probes, 293 region, 194-195
F1urochromes, 351
Flux Reactor, 195 HI, 203, 341, 409
Form Factor, 267-273, 621, 633, binding, 163, 466
733 binding enthalpy, 162
Fourier Transform, 188, 210, 281 phosphorylation, 493, 627
fast, 278, 289 removal, 646
Fractionation (Chromatin), 15, 541 H2A-H2B Dimer, 159-160, 462
Free H2A, H2B, H3, 78, 373
binding, 149 during cell cycle, 627
energy, 149, 151-152 160, 164 H3-H4
histone conformation' 152 dimer, 151, 158-159
Frictional '
DnA interaction, 155-156, 375
coefficient, 167 H4, 78, 377
ratios, 460, 466 H5, 469-472
Friend Cells, 43
H615 Cells, 630
Haploid Genome 43
GO Cells, 628 Hardware, 270 '
GO-G1 Transition, 635-636, 641
Harwell, 190
G1 Phase, 44
Heat Shocks, 500
G2 Phase, 267, 499 HeLa .
Ga11ocyanin-chrome Alum 351
cells, 43, 323, 502
Gel Electrophoresis, 125
chromatin, 622, 632
DNA, 394, 397
nuclei, 633
proteins, 627, 638 Helix, 76, 428
Gene
coil transition 625, 629
activity, 42 Hemisome, 424 '
packing, 771-773 Hen Oviduct, 52
ribosomal, 49 Hepatectomized Rat Liver 639
Genome
Heterochromatin, 42, 506'
expressed, 18, 555 Heterotypic Tetramer, 161
non-expressed, 19, 545 Hexamer, 207
Genetic, 752
HGPRT Locus, 757
apparatus, 753 High Resolution
Geometric-densitometric Image
Analysis, 265, 62U
integrating, 260
Globin
scanning, 260
Higher Order Structure, 137,
cDNA hybridization, 44, 55 209, 455-456, 485, 639, 655
gene, 53 Hill
mRNA, 44, 46
sequences, 52 coefficient, 150
plot, 150
Globular Proteins, 168, 456
INDEX xxv
Histone Interfiber Distance, 656

bound, 139 Internuc1eosome Spacing, 155,
cell cycle modification, 492, 291, 656
627 Interphase, 41, 633
core, 160, 162, 183, 188, 199 phosphorylation, 497
depleted chromatin, 154, 156 Intrinsic
dissociated, 134, 147 birefringence, 170
DllA inteructions, 458 viscosity, 168, 636
genes, 44 In Vitro vs. In Vivo, 683-699
ionic strength, 455 Ion-exchange Chromatography,
phosphatase, 502 46
Histones, 78, 393, 408, 563, 781 Irradiation, 752
chemistry, 5 Isobutyrate, 493
H1, 125, 137-164 Isopycnic Centrifugation, 547
H2A, 137-164, 465, 488 Isotopes, 756
H2B, 137-164, 456
H3, 137-164, 458 Jamin-Lebedeff Optics, 257
H4, 78, 137-164, 45C
H5, 469 Karyotype, 711, 764
radius of gyrution, 195 by c£nsitometry, 347
self-assembly, 372 by fluorescence, 351
HMG Proteins, 493 Kirkwood Theory, 431, 566
Hoechst 33258 , 351
Homotypic, 162 Lac Operon, 601
Hybridization, 43-62 Larval Diptera, 42
Hydrated, 482 Laser
Hydrodynamic, 109-113, 431-436 dual, 353
interaction tensor, 431 flow microfluorimetry, 295,
Hydrogen, 192 350
Hydrolysis, 271 Raman spectroscopy, 114-119,
Hyperchromicity, 400, 630 123
Latex Spheres, 454
Image Lectins, 712
analysis, 265, 343, 620 Leukemia, 705
analyzer, 266, 323 Li C1, 72
classification, 270 Light Scattering Artifacts, 75
In Situ, 265, 272, 622 630
Inelastic Light Scattering, 167 Limit Digest, 157
Information Processing, 267 Linker DNA, 395, 486
Insulin, 493 Lipoprotein, 195
Integrated Optical Density (IOD) , Liposomes, 767
266 Liver
Interband Space, 252 chromatin, 624, 639, 774-775
Intercalating Agents~ 728 lower' euka:Pyotes, 599
Interface, 188 uterus, 42
Interference, 209 Lymphocytes, 636, 638
microscopy, 256 Lysiae, 149, 154, 160
Interferometric Techniques, 256,
259
xxvi INDEX
M Phase, 622 Molecular Wt. of Core Protein,

Macrophages, 706 198
Magnesium, 502, 522 Morphometry
Malignancy, 705, 722 aeZZ, 739
Markovian Processes, 271 ahromtin, 615
Mass Action Law aoZony, 741, 743
aeZZ studies, 299, 306 nuaZei, 327, 642
soZution studies, H3, 305 ~~ouse
Maxwell brain, 45, 48
aonstant, 179 embPyo fibpobZasts, 670
effeat, 168 Uvep, 42
Mean Dispersion Path (MDP), 621 tissues, 42
Melting, 75 mIDIA, 43, 563
absopbanae, 400, 631 Multimers, 212
aipau~ diahroism, 631, 401 Multiparameter, 278
Uembrane, 814 FMF, 621
l1ercaptoethanol, 49 Multiprobe Cytopho·.tometry, 668
Mercurated UTP, 53, 56 Mutagens, 726
Mercury Substituted Ribonucleo- Mutants, 302, 630, 639
tides, 49 Mutation, 772
Meridian, 189 fpequenay, 726, 757
Metaphase Chromosomes, 341-351, Myeloid Cells, 705, 715
487, 531, 635 peguZatop-I1GI, 706
Methanol, 74 Myeloma, 52
Methylation, 625, 775 Myoblast, 43
Micrococcal nuclease, 17, 138,
154, 156, 389, 414, 422, NaCl, 140
544, 603 Native Chromatin, 629-630, 646
Microfluorimetry N-Bromosucinimide, 460
aeZZ, 257 neoplastic Transformation, 683-
ahrmnatin, 293-322 699
ahPOmosome, 349-355 Neutron
Zasep fw'W, 293 beam, 191, 199
statia, 296 diffpaation, 187, 472
Microinterferometer, 257 saattePing, 192, 428
Microtubules, 522 saattePing ppofiZe, 200, 208
Mid Nicked Circles of SV40, 615
Gl, 272, 626, 634 Nitrosoguanidine, 712
S, 272, 626, 634 Non-cooperative Dissociation,
Mitotic Detachment, 624, 737 157
Mixed Reconstitution, 625 Non-histone, 6, 561
MNNG, 758 gene ~eguZation, 567
MNU, 644, 771 pPOteins, 6, 82, 397, 487,
Models of Chromatin Structure 561, 625
teptiary, 408, 410, 1~8l synthesis, 41, 636
quate~, 486, 652 Non-proliferating Cells, 635
quintePnaPff, 660 N-terminal, 161, 163, 456
pevie'W, 849-870
Molar Ellipticity, 622-647
INDEX xxvii
Nu-body (see nucleosome) Optical

Nuclear activity, 68, 619
ch:r>orrntin, 467 anisotropy, 169
cytochemistry, 243, 259, 323 density levels, 275
geometry, 463 noise, 177
histone, 472 Organic Solvents, 74
images, 265 Orientation Index, 735
IOD, 270 Ouabain, 757
morphometry, 266, 271, 633 Ovalbumin, 43
nucleosome, 653 gene, 52
stains, 293, 323
Nuclear ~~gnetic Resonance ~- electron, 453
chromatin, 467 ~-~* and n-~* Transitions, 68
DNA, 473 Packing Ratio (Chromatin), 20
general introduction, 217, 455 Pair Prob. Function, 198
his tones , 161, 457-459 Paramagnetic Ions, 454
polynucleotides, 223·235 Partial Hepatectomy, 622
t-RNA, 235-244 Patterson Function, 198
Nuclease PDP Computer, 263, 294
and chromatin fractionation, Peptide Bonds, 76
548-551 Periodicity (higher order), 278
dtaphyloccal, 17 Permissive vs. Non-Permissive
Nuclei, 154 Temperature Mutants, 302
image analysis, 265 Peter lin and Stuart Hydrodyna-
electron microscopy, 456 mic Treatment, 168
microfluorimetry, 297 PH Effects, 115
Nucleofilaments, 491, 525 PHA Stimulated, 635
Nucleolar DNA, 604-605 Phase Microscopy, 263
Nucleolus, 252 Phase I, II, III WI-38, 642
Nucleosome, 31, 33, 37, 67, 137, Phenotypic Changes of Tumors,
167, 177, 182, 613 683-697
circular dich:r>oism, 85 Phosphate Groups, 149
core, 399 Phosphorylation, 492, 563, 626,
dimers, 447 627
DNA sequences, 441 Photoelectric, 170
hydrodynamic parameters, III Photographs Fluorometry, 264
interactions between, 436 Physarum, 494, 603
meUing, 400 Pitch, 189, 212, 433, 651
preparation, 125 Ploidy State, 457
radius of gyration, 170, 176, Plug l1easurements, 264
210, 389, 437 Plumbicon Scanner, 265, 278
reconstitution, 413 Polarized Light, 68, 617
shape, 114 scattering, 618
Polarizer, 172
Oblate Structures, 183, 202 Poly A, 765
Octamer, 163, 406 Polyacrylamide, 129
Oestrogen, 42 gels, 7, 141
Oligomers, 203
Oligonucleosome, 155
Oocytes, 42
xxviii INDEX
Polymers Rat
H2A-H2B. 162 Kidney, 638
H3-H4. 162 Uver, 638
Polymerase prostate, 638
RNA. 58 uterus, 638
Polypeptides. 76 Rate of
Polynucleosome. 427. 486. 519 dissociation, 139, 141, 143, 160
Polysomal RNA. 46 reassociation, 139, 160
Polytene Chromosomes. 42 Rayleigh, 616-618
Porod Law. 198 Reconstitution
Premelting. 75. 625 ch:Pomatin. 578, 628
Prereplicative Phase. 635 nucleosome, ,413
Prokaryotes. 600 Reflections, 212, 632
Prolate Structures. 181. 202 Refractive Index, 168, 181, 183
Proliferating Cells. 635-638 Refractometer, 171
Pronase Digestion. 415 Relaxation Processes, 169
Prophase. 529 Renaturation, 162, 627
Protamine Sulfate. 129 Repeat Sizes, 391
Protease. 4. 767 Repeating Unit, 80, 395
Protein Repetitive Nucleotide Sequences,
kinase. 493. 499 43
modification (cell cycle). 492. Resting Tissue, 635
627 Restriction Enzymes, 553-556
Proteinase K. 138 Ribonuclease, 767
Proton Exchange. 196 Ribosomal
Puffing, 42 gene, 571
Pulse-Cytophotometry (See Flow RllA synthesis. 606
GYtometpY and/or Laser Flow Ribosylation, 625
PncrofZuorimetry) Ring Current Shift, 453
RNA
Quantum Yield. 310. 639 cistrons, 755
Quasielastic Light Scattering. polymerase, 58. 273, 599
428 transcription, 43, 546-550
Quaternery Structure. l137. 522. RNAse, 45, 313, 552, 766
487 Rod-Like Particle, 194
Quiescent Roots of Lens Culinaris, 638
ceU. 297. 654 Rotation Diffusion Coefficient,
Quinacrine Banding. 264 177
Quinternary Structure. 278, 525. Rotational Diffusion, 168
660 Rye Chromosomes, 255
Radioactive Tracer. 451. 623 S Phase, 497, 626-627

Radius of Gyration. 194, 195, 197, Sl Nuclease, 45
205, 207, 477 Salivary Gland, 638
Raman Salt Effects, 71. 524
ch:Pomatin spectra, 118 Salt Wash, 645
Zaser spectroscopy, 117 Saturation Density, 724. 728
Random Coil, 76-77, 428 S-bromodeoxyuridine. 712
INDEX xxix
Scanning Stiff Chain Polymer, 169

Zines~ 263 Stimulation
photometry~ 341-349 fibrobZasts~ 635
Scattering Zympliooytes., 641
density, 191-192, 195,202,209 Stoichiometry, 149, 162
Zength, 475 Strict Light Scattering, 168
voZurnes, 202 Subphases, 273, 634
Scheraga r~nde1kern Factor, 32 Subunit, 783
Sea Urchin, 42-43 Superbeads, 515, 521, 522
Secondary Structure, 143 Superhe1ix, 188-189, 207, 649,
DNA, 69 652
protein, 76 density~ 649
Sedimentation heZi:r: transition~ 625-; 629
coefficient, 112 reZaxing~ 555
veZocity, 422 Supernuc1eosomes~ 522
Semiconductivity, 835 Superpacking~ 272
Senescence, 759, 642, 673 Superstructures., 519
Separation of Active and Inactive Supercoi1ed Chromatin, 159
Genome, 545-548 SV-40
Sepharose, 139, 154 mini chromo some , 424
Shading, 269 transformed ceZZs, 277, 643
Shear Rate, 175, 177 Svedberg Equation, 431
Sheared Chromatin, 613, 630 Synchronized HeLa, 650
Shearing, 137, 154, E30, 632
effect of, 632 Template, 8, 16-17-, 41-66
Side-chains, 161 active vs. inactive, 18-22
Sigma Factor, 566 activity, 273, 614
Sodium~ 757 specificity, 563
Software.. 262 Terminal Differentiation, 727
Solenoid, 210, 212, 428, 487, Tertiary Structure, 75, 163,
515, 520, 657 204, 406, 483, 652
superheZicaZ turn, (51 Tetramer, 80, 386, 458, 464
Sonication, 544, 630 Texture, Chromatin, 271-281,
Spacer 670-679
DNA (See aZso Zinker DNA)~ Thermal
395-397 denaturation, 400, 620, 626
Zength, 159, 188, 654, 656 noise, 177
Spacing of DNAse I sites, 448 transitions, 629
Spectropolarimeter, E19, 630 Thermodynamics, 137, 151
Spermidine, 129 Thio1 Agarose Chromatography,
Spherical Particle, 194 52
Spin, 217 Threshold Dependence, 274
Spin Decoup1ing, 455 Titration Shift, 454
Spin-spin Splitting, 455 Transcription, 53, 492, 494,
Spot Size, 253 543, 764
Staining, 265 Transcriptional
Staphylococcal, 17 assays, 9, 45, 49
Stationary 2RA, 642 controZ, 41-67, 560, 599
Stem Cells, 730
xxx INDEX
Transcriptionally Active Regions, X-irradiation, 712

549, 641 X-ray
Transformation, 259, 683, 695- absorrption, 256
698 diffpootion, 69, 189, 210,
fpequency, 726, 757 488, 491, 630
Transmission, 253 seattePing, 187, 211
Transmittance, 175
Trimer Band, 155 Zero Angle Scatter, 477
Tritiated Thymidine Incorporation, Zipper Binding Meachnism, 161
624
Triton-Acid Urea Gels, 125-136
t-IDIA, 46
Tumor Progression, 696-697
Tumorigenicity, 725, 758
Tumors, 684
TV-microscopy, 263
Two-order Superhe1ica1 Hodel,
652
2RA, 642 PART A - Pages 1 - 370
eells~ 273-274, 276
Tyrosine, 78 PART B - Pages 371 - 870
Ultra
eentpifugation, 32
miePospeetrogpaphy, 251
Unique Nucleotide Sequences, 43
Unit Thread, 212
Unstacking, 75
UV Microscopy, 254
Virus
'eRNA, 710
SV40, 276, 424, 623
tpansfozrmation, 642
Viscometer, 171
Viscosity
sheaPed ehPomatin, 146
unsheaPed ehPomatin, 637
Water Content, 622
Wet Replica, 517
WI-38, 318, 632-642
aftep sePUm stimulation, 635
eells, 302, 638, 641
ehPomatin, 636
nuelei, 273, 671
Winding Angle, 73

Claudio Nicolini PDF

Uploaded by

Copyright:

Available Formats

Claudio Nicolini PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Claudio Nicolini PDF

Uploaded by

Copyright:

Available Formats

Chromatin Structure

A series of edited volumes comprising multifaceted studies of contemporary

Volume 13 - Prostaglandins and Thromboxanes

Volume 14 - Major Patterns in Vertebrate Evolution

Volume 210 - Chromatin Structure and Function :

Volume 21b - Chromatin Structure and Function:

The series is published by an international board of publishers in con-

A Life Sciences Plenum Publishing Corporation

D Behavioral and Sijthoff International Publishing Company

E Applied Sciences Noordhoff International Publishing

SPRINGER SCIENCE+BUSINESS MEDIA, LLC

First half of the Proceedings of the NATO Advanced Study

© 1979 Springer Science+Business Media New York

AII righ ts reserved

This book, divided in two volumes, is the result of a NATO-

It has been edited in an organic and tutorial format, with

This first volume, following an introduction to the properties

In order to make this volume comprehensive and accessible to

Even if particular reference is made to the genetic apparatus

SECTION I: WHAT IS THE CHROMATIN?

Properties and Composition of

Expressed and Nonexpressed Portions of

SECTION II: PHYSICAL, CHEMICAL AND

Electron Microscopy: A Tool for

Circular Dichroism of DNA, Protein

Important Hydrodynamic and Spectroscopic

Preparation and Analysis of Core Particles

The Interaction of Histones with

Nuc1eosome Shape and Structure in

Scattering and Diffraction by

Nuclear Magnetic Resonance Studies

Techniques for Cytochemical Studies

Chromatin Study in Situ: I. Image Analysis 265

Chromatin Study in Situ: II. Static

Chromatin Study in Situ: III. Differential

Scanning and Flow Photometry of Chromosomes • • • • • • •• 341

SECTION' III: VARIOUS LEVELS OF

Histones Assembly and Their Structural

Nuclease Digestion and the Structure

Reconstitution of Nucleosomes • • • 413

Conformation of Polynucleosomes in Low

Chromatin Structure: Relation of Nucleosomes

Histone Complexes, Nucleosomes, Chromatin and

Evidence for Superstructures of Wet

Chromatin Fractionation and the Properties

Chromatin Reconstitution and Non-Histone

SECTION IV: STRUCTURE-FUNCTION OF THE

The Structure and Function of Chromatin

Chromatin Structure from Angstrom to Micorn

Chromatin Pattern in Situ: Dependence upon

Neoplastic Transformation: The Relevance of

Cell Differentiation and Malignancy

Cellular Morphometry in Transformation,

Basic Mechanisms in Chemical Carcinogenesis • • • • • • • • 751

Carcinogen Induced Alteration in Gene

Covalent Binding of a Carcinogen to DNA

Carcinogenesis, DNA Repair and Chromatin 803

Electromagnetic Induction of Electrochemical Information

SECTION V: REVIEW AND SUMMARY OF