119

Collagens and Gelatins

Dr. Barbara Brodsky1, Dr. Jerome A. Werkmeister2, Dr. John A. M. Ramshaw3
1
Department of Biochemistry, UMDNJ-Robert Wood Johnson Medical School,
Piscataway, NJ 08854, USA; Tel.: 1-732-235-4048; Fax: 1-732-235-4783;
E-mail: [email protected]
2
CSIRO, 343 Royal Parade, Parkville, VIC 3052, Australia; Tel.: +61-3-96627239;
Fax: +61-3-96627218; E-mail: [email protected]
3
CSIRO, 343 Royal Parade, Parkville, VIC 3052, Australia; Tel.: +61-3-96627217;
Fax: +61-3-96627218; E-mail: [email protected]
1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

120

2
2.1
2.2

121
121

2.3

Historical Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Early Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Golden Age of Collagen Discoveries: Composition, Conformation, and
Fibril Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Amino Acid Sequence, Genetic Diversity, and Precursors . . . . . . . . . . .

3
3.1
3.2

Chemical Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Unusual Amino Acid Features and Post-translational Modifications . . . . .
Cross-links Between Collagen Chains . . . . . . . . . . . . . . . . . . . . . . .

124
124
126

Structure of the Collagen Triple Helix . . . . . . . . . . . . . . . . . . . . . . . .

127

5
5.1
5.2

Occurrence and Function of Collagen . . . . . . . . . . . . . . . . . . . . . . .

Fibril-forming Collagens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Non-fibrillar Collagens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

129
129
131

Biosynthesis and Biochemistry of Collagen . . . . . . . . . . . . . . . . . . . .

131

Physiology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

133

Biodegradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

134

122
123

120

6 Collagens and Gelatins

Polypeptide Models of Collagen . . . . . . . . . . . . . . . . . . . . . . . . . . .

135

10

Molecular Genetics and Collagen Diseases . . . . . . . . . . . . . . . . . . . .

136

11

Chemical Analysis and Detection . . . . . . . . . . . . . . . . . . . . . . . . . .

137

12
12.1
12.2
12.3
12.4

Production and Applications . . . . . . . . . . . .

Biomedical Applications of Collagen . . . . . . .
Gelatin Production and Applications . . . . . . .
Recombinant Collagen and Gelatin . . . . . . .
Immunological Response to Collagen Products

.
.
.
.
.

138
139
140
140
141

13

Patents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

142

14

Outlook and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

146

15

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

147

Hyl
Hyp
P4H
MMP
OI
SLS

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

hydroxylysine
hydroxyproline
prolyl-4-hydroxylase
metalloproteinase
osteogenesis imperfecta
segment long spacing

Introduction

Collagens are the predominant fibrous

proteins in animals. They constitute a major
structural component in the extracellular
matrix of all multicellular animals, including sponges, invertebrates, and vertebrates.
The most abundant and well-characterized
collagens are found in axially banded fibrils
with a 670- repeat (Kielty et al., 1993).
These collagen fibrils make up as much as
30% of the total protein in humans and other
higher animals. The role of such fibrils is to
maintain tissue integrity and to provide
specific mechanical properties needed for
each connective tissue. For example, collagen fibrils form the basis for the strength of
tendons, the compressibility of cartilage, and

the flexibility of blood vessels and skin.

There are also other genetic types of collagen
that do not form fibrils and are found in
much smaller quantities in tissues. These
play critical structural roles, including formation of the framework and permeability
barrier of basement membranes, linking of
fibrils to other matrix components, and
anchoring of dermis to its associated basement membrane.
From a biomedical view, collagen has been
shown to play important roles in development, wound healing, platelet activation,
and angiogenesis. In addition, a growing
number of hereditary diseases have been
shown to be caused by mutations in collagen
genes. Perturbations in collagen synthesis,
degradation, or immunology also have been
noted in many common diseases of complex

2 Historical Outline

etiology, including arthritis, liver fibrosis,

diabetes, and cancer, as well as aging.
The commercial importance of collagen
relates to the wide application of gelatin as
well as leather production and biomedical
applications. Gelatin is derived from collagen in tissues by heat denaturation. On
cooling, the chains can rewind, but not in the
correct register, and the small triple-helical
segments formed may further aggregate
during gel formation. The uses of gelatin
in films and in food are widespread.
Collagen has a uniform triple-helical
conformation, which is determined by its
repeating (Gly-X-Y )n sequence pattern and
high imino acid content. Polypeptides with
these sequence features also adopt this
conformation (Heidemann and Roth,
1982). X-ray diffraction, electron microscopy, hydrodynamic methods, and a variety
of spectroscopy techniques have been used
extensively in characterizing collagen molecules, collagen fibrils, and gelatin. In this
chapter, collagen and gelatin will be reviewed in a biopolymer context in terms of
chemical structure, production, and methods of detection and analysis. The biologically unique features of collagen, including
biosynthesis, functions, and disease, also
will be presented. A number of excellent
books and chapters on collagen describe the
biochemical and biomedical aspects in
further detail (Nimni; 1988 1991; Royce
and Steinmann, 2002; Kadler, 1994; Bateman et al., 1995).

Historical Outline

The field of collagen is an old one that has

blossomed in multiple directions in recent
years. Further details on the history of
collagen and gelatin can be found in a
range other of texts (e.g., Gustavson, 1956;

Ramachandran, 1967a; Fraser and MacRae,

1973; Mayne and Burgesson, 1987).
2.1

Early Background

Collagen has been a useful material to

mankind for millennia. Indeed, it is not
clear when it was first recognized that hides
and skins could be preserved against decay,
by tanning the collagen, to give leather.
While materials such as hair probably first
were used as medical sutures circa 5000
years ago (see Breasted, 1930), collagen
catgut sutures were also developed and
used for repair of wounds, e.g., by Claudius
Galenus, circa 175, in treating wounded
gladiators (Mackenzie, 1971). The denatured
products of collagen, animal glues and
gelatin, were also in early use, with Pliny
the Elder noting that Boum coriis glutinum
excoquitor, taurorum praecipuum glue is
boiled from the hides of cattle, and the best
from those of bulls (Pliny, circa 50). It was
also recognized in historical times that
collagen, as gelatin- or glue-forming material, was found widely distributed throughout
all animal phyla.
The chemistry and other applications of
collagen, however, developed over the last
200 years. During the 1820s, studies on the
hydrolysis products of the natural substances that were later called proteins led to the
isolation of a new substance from gelatin,
sugar of gelatin, so named because of its
sweet taste (Braconnot, 1820). It was distinct
from sugar, and further analyses (Mulder,
1838; Gerhardt, 1846) led to a structure
proposal (Cahours, 1857) for what was by
then called glycine (Berzelius, 1848). Boiling
acid treatment of gelatin and other proteins,
initially by H2SO4 but later by HCl (Bopp,
1849; Hlasiwetz and Habermann, 1873), led
by 1900 to the isolation of 13 amino acids.
This led to the concept that proteins are

121

122

6 Collagens and Gelatins

polymers of amino acids (Fischer, 1902a;

Hofmeister, 1902). Further studies on gelatin led to the discovery of hydroxyproline
(Hyp) (Fischer 1902b) and of hydroxylysine
(Hyl) ( Van Slyke and Hiller, 1921; Van Slyke
et al., 1938). However, it was only realized
considerably later that these amino acids
were formed by secondary modification and
were not directly incorporated into protein
(Stetten, 1949; Sinex and Van Slyke, 1957).
Early histologists, when treating sections
with acid (e.g., Gillette, 1872), would have
been extracting soluble collagen; for example, tendons were observed to swell and
dissolve in dilute acetic acid (Zacchariades,
1900). The initial chemical studies on
soluble collagen were mainly by Nageotte
and colleagues in France from the mid-1920s
(Nageotte, 1927). Nageotte found that acetic
acid extracts from various tissues contained
soluble collagen and that this could be
precipitated by salt and re-aggregated into
fibrils by dialysis at neutral pH against water.
These and other studies on general solubility
properties under different conditions became the basis for collagen purification
(Tustanovskii, 1947; Orekhovich, et al.,
1948; Gross et al., 1955).
2.2

The Golden Age of Collagen Discoveries:

Composition, Conformation, and Fibril
Organization

In the 1940s and 1950s, a picture of the

unique features of the amino acid composition and conformation of the collagen
molecule emerged, together with an understanding of the collagen fibril structure. The
composition of collagen differed from any
known protein; its most striking features
were the high contents of glycine and of the
imino acid proline (see Eastoe, 1967). The
first experimental attempts to determine the
primary structure, using partial acid hydro-

lysates, gave indications of a regular structure, suggesting that glycine might be

present as every third residue along the
chain (Schroeder et al., 1953). Comparisons
between collagen and gelatin showed close
similarity in composition, but gelatin had a
lower ionic point as a result of amide
hydrolysis and showed partial conversion
of arginine to ornithine. Mature vertebrate
collagen also was found to have a low total
carbohydrate content, usually below 1%.
Later, in 1966, it was shown that this was
galactose or glucosylgalactose bound to the
hydroxyl groups of Hyl (Butler and Cunningham, 1966).
In the mid-1950s, the understanding of
the unique amino acid composition features
of collagen, together with fiber X-ray diffraction studies, led to the first proposal of a
triple-helical structure for collagen (Ramachandran and Kartha, 1954). The fiber
diffraction seen for collagen in tail tendons
was distinct from that seen for any other
fibrous protein (Astbury, 1933), showing a
strong axial reflection at 2.9 and a strong
equatorial reflection that varied between
11 and 16 , depending on the degree of
hydration (Rougvie and Bear, 1953). Improvement of the diffraction patterns by
stretching (see Cowan et al., 1955) showed
the transform of a rod-like structure with
helical symmetry. A three-stranded, triplehelical model for collagen comprising lefthanded helical chains related by a righthanded rope twist was suggested by Ramachandran and Kartha (1954) to account for
the X-ray pattern features and the unusual
amino acid features. This structure, in which
the three chains are arranged about a
common central axis, leaves no space in
the inner core for Cb atoms and therefore
must be occupied by Gly residues. Pro
residues could be accommodated in the
non-glycine positions without distortion.
Rich and Crick (1961) pointed out that the

2 Historical Outline

initial model violated stereochemical limitations, but that modifications to the basic
three-chain concept could be made to give a
model that was sterically satisfactory. The
modifications focused on the arrangement
of the hydrogen bonds, and the model
designated as RCII was preferred by virtue
of stereochemical considerations.
A detailed evaluation of the chain conformation in the collagen molecule using
quantitative X-ray diffraction data collected
from stretched kangaroo tail tendon (Fraser
et al., 1979) gave an estimated value for the
unit twist of the molecular helix of 107.18
 0.68, which is close to the value expected
(1088) for a helix with 10 units in 3 turns,
although there is no a priori reason why such
a simple relationship should exist. The best
solution very closely resembled the RCII
structure. The model was supported by
physicochemical investigations suggesting
that collagen molecules consisted of three
polypeptides each with length of about 1000
amino acids (Boedtker and Doty, 1956).
During this fertile time on collagen
research, the molecular organization in the
collagen fiber was also determined. Electron
micrographs of native collagen fibers or
fibrils reconstituted from soluble collagen
(Hall et al., 1942; Schmitt et al., 1942)
showed a simple banded appearance with
cross striations at 600 to 700 , similar to
the period of 670 observed in low angle Xray reflections carried out by Bear (1942).
This axial repeat is usually designated as the
D period. Two alternative forms of collagen
aggregates were also visualized: (1) the
centrosymmetric structure fibrous long
spacing (FLS ) formed when soluble collagen was mixed with highly negatively
charged compounds (Highberger et al.,
1950); and (2) segment long spacing (SLS ),
a parallel array of in-register molecules that
reflected the 300-nm molecular length of
collagen (Schmitt et al., 1953). Schmidt and

colleagues suggested, from comparison of

the cross-striation period along the collagen
fibril with the length of the collagen
molecule, that the collagen fibril was a
parallel array of molecules staggered by
one-quarter (Schmitt et al., 1955), and more
precise studies indicated a stagger of D
670 , where the molecule length is equal
to 4.4D periods (see Hodge, 1967). Early
studies using lathyrogens, which interfere
with cross-linking formation, led to the
conclusion that these D-periodic fibrils
contain intermolecular covalent cross-links
that lead to their high tensile strength and
mechanical stability.
2.3

Amino Acid Sequence, Genetic Diversity, and

Precursors

Introduction of enzyme methods (e.g., Kuhn

et al., 1963) gave a simple method for largescale extraction of collagen. Initial separations of collagen chains by ion exchange
chromatography (Kessler et al., 1959) led to
partial degradation, but the introduction of
milder conditions (Piez et al., 1960) led to
successful separation and characterization
of the intact subunit chains of soluble
collagen (Piez et al., 1963). These data
showed that the a components were single
polypeptide chains, with two types of these
chains in the majority of tissues studied. The
b and g components consist of two and three
polypeptide chains, respectively, joined together by covalent cross-links that survive
denaturation. Separation of the individual
chains of collagen by ion exchange chromatography (Piez et al., 1963) enabled peptides
obtained by chemical cleavage, e.g., by CNBr,
hydroxylamine, or enzymes (Bornstein and
Piez, 1966), to be separated and used for
sequence analysis. The first complete chain
sequence, for the bovine a1(I ) chain, was
achieved in 1972 (see Fietzek and Kuhn,

123

124

6 Collagens and Gelatins

1976). These data showed an excellent

correlation between sequence and the
cross-striations seen in SLS aggregates and
intact fibrils. SLS structures prepared from a
wide range of different species indicated the
remarkable stability of the amino acid
sequence during evolution (Mathews,
1975). Automated protein-sequencing procedures accelerated the collection of sequence data, while in recent years, DNAsequencing technology has greatly increased
the available sequence database (see Kadler
1994). With recombinant technology, the
complexity of the collagen gene with a large
number of coding and non-coding sequences became apparent ( Wozney et al., 1981).
Until the 1970s, collagen had been considered as a single entity. Although Bazin
and Delauney (1964) first noticed that
granulation tissue appeared to contain a
type of collagen that was distinct from type I
collagen and similar to the type often
associated with embryonic collagens in
skin, an alternative collagen type was first
clearly identified in articular cartilage by
Miller (1971). Soon after, the distinct collagen of granulation tissue, skin, and other
tissues was characterized as collagen type III
(Epstein, 1974; Chung and Miller, 1974).
At least 20 genetically distinct collagen
types have now been described (Kielty and
Grant, 2002). A precursor form of collagen
was first observed in 1971, in biosynthetic
investigations of fibroblast cultures (Layman
et al., 1971) and in the skin of calves with
an inherited fragile skin disease (Lenaers
et al., 1971). These observations led to
significant, on-going efforts using cellular
and molecular biology approaches to understand the biosynthesis of collagens. This has
provided data on intracellular processes
including transcription, translation, chaperones, the variety of secondary modification
enzymes, chain recognition, and procollagen protease activities (see, for example,

Royce and Steinmann, 1993; Bateman et al.,

1995).

Chemical Structure

Collagen is a protein, synthesized from the

20 common amino acids, yet it is unique in
terms of its amino acid composition, repeating sequence pattern, high degree of posttranslational modification, and characteristic intermolecular cross-links (Kielty et al.,
1993).
3.1

Unusual Amino Acid Features and Posttranslational Modifications

The amino acid composition of collagen

shows an unusually high content of Gly, near
33%, and a high content of imino acids, near
20%. Examination of its sequence shows that
Gly is present as every third residue,
producing a (Gly-X-Y )n repeating pattern.
Collagen contains a considerable amount of
the unusual amino acid Hyp, which is rarely
found in other animal proteins, and significant amounts of Hyl, both formed by
enzymatic post-translational modification.
Hyl groups are often glycosylated, resulting
in Hyl-galactosyl and Hyl-galactosyl-glucosyl
moieties. The Hyp is almost exclusively
trans-4-Hyp, containing the OH group on
the g-carbon of the prolyl ring, although
small amounts of 3-Hyp have been reported
in invertebrates and basement membrane
collagen (Mathews, 1975). The most common tripeptide sequence found in collagen
is Gly-Pro-Hyp (Figure 1).
In contrast to the restrictive nature of every
third position being Gly, the other two
positions in each repeating unit, X and Y,
can accommodate any of the 20 amino acids,
including imino acids, without distortion.

3 Chemical Structure

Fig. 1

The chemical structure of collagen and its higher levels of organization.

This was shown first by model building

(Rich and Crick, 1961) and has been
confirmed recently by studies on hostguest
peptides (Persikov et al., 2000). These peptides of sequence (Gly-Pro-Hyp)3-Gly-X-Y(Gly-Pro-Hyp)4 indicate that all residues can
be incorporated in either the X or Ypositions,
but the stability of the triple helix depends on
the guest residue.
The restricted backbone of the cyclic imide
group results in phi,psi angles very close to
those found in native collagen; therefore,
Pro and Hyp groups are entropically favored

to adopt the triple-helix conformation (Josse

and Harrington, 1964). The high imino acid
content of collagen affects its folding as well
as its geometry. There is a significant
probability of cis peptide bond configurations in the unfolded state for both Gly-Pro
bonds ( ~ 14%) and Pro-Hyp bonds ( ~ 6%)
(Sarkar et al., 1984). Since only trans peptide
bonds are found in native triple-helical
collagen, cis bonds in the unfolded chains
need to undergo isomerization, an inherently slow process that limits the rate of
propagation in vitro.

125

126

6 Collagens and Gelatins

3.2

Cross-links Between Collagen Chains

The formation of specific covalent crosslinks between collagen molecules stabilizes

collagen fibrils in tissues (Yamauchi and
Mechanic, 1988; Bailey, 2001; Brady and
Robins, 2001). Lysyl oxidase initiates collagen cross-linking by catalyzing the formation of Lys- and Hyl-derived aldehydes, i.e.,
allysine and hydroxyallysine, at specific
residues in the telopeptides. These aldehydes then undergo a series of condensation
reactions with adjacent Lys residues from
the telopeptide or with specific lysine
residues from the triple-helical domain to
provide the initial cross-links (Figure 2).
These cross-links, the reducible cross-links,
can be stabilized for analysis by reduction
with sodium borohydride. Further spon-

Fig. 2

taneous reactions result in a variety of other

multifunctional cross-links that characterize
mature tissues (Figure 2), although only a
limited number have been characterized.
The number and proportions of the various
cross-links show tissue specificity, and it is
likely that this is regulated by the steric
relationship between molecules, the type of
collagens copolymerized, and glycosylation
and hydroxylation of the participating amino
acid residues. In mature tissues, non-enzymatic glycosylation can further cross-link the
collagen through non-specific cross-links at
various locations, leading to decreased
solubility and a characteristic fluorescence
(Monnier and Cerami, 1982). This additional cross-linking can make tissues too stiff
to function properly and can affect the
filtration properties of the glomerular basement membrane. Because cross-linking may

The chemical structures of examples of collagen cross-links.

4 Structure of the Collagen Triple Helix

3 Fig. 3 The triple-helix conformation from a peptide
crystal structure (Kramer et al., 2000).

be increased in diabetic patients, it is a

potential cause of the tissue-based problems
that emerge in this disease.

Structure of the Collagen Triple Helix

The collagen family represents a group of

diverse molecular structures, yet all are
linked by a common structural element
the collagen triple-helix structure (Rich and
Crick, 1961; Bella et al., 1994; Brodsky and
Ramshaw, 1997). The collagen triple helix is
comprised of three polypeptide chains, each
in a left-handed extended polyproline II-like
helix (Figure 3). The three chains, which are
parallel and staggered by one residue with
respect to each other, are supercoiled about a
common axis in a right-handed manner.
Interchain peptide hydrogen bonds link the
three chains. The close packing of the three
chains in the triple-helix requires that every
third residue in the sequence be a Gly
residue, while the extended conformation of
the individual chains is stabilized by a high
content of imino acids.
The original triple-helix model was constructed and refined on the basis of fiber
diffraction studies of tendons, but more
recently, crystal structures of triple-helical
peptides at resolutions higher than 2 have
allowed elucidation of the details of the
triple-helical structure at a molecular level
(Bella et al., 1994, Kramer et al., 1998, 1999,
2000, 2001; Vitagliano et al., 2001; Berisio
et al., 2001). These studies confirm the
overall features proposed earlier. The one
direct NH (Gly)CO (X position) hydrogen

127

128

6 Collagens and Gelatins

Fig. 4 A schematic illustrating interchain hydrogen bonding and water-mediated hydrogen bonds in the
collagen triple helix.

bond between chains can be visualized, and

when Pro does not occupy the X position, an
additional interchain hydrogen bond,
NH (X )CO (Gly), is formed through one
mediating water molecule. In contrast to
alpha-helices and beta-sheets, where all NH
and C O groups of the peptide backbone
are involved in hydrogen bonds, the collagen
triple-helix always has available backbone
C O groups and, in some cases, NH
groups. An extensive hydration network is
part of the response to satisfy available
backbone groups, and this is visualized in
the crystal structures (Figure 4). The role of
Hyp in collagen stability may be related to
this hydration network (Privalov, 1982) and/
or to inductive effects that influence the
puckering of the imide ring structure
(Bretscher et al., 2001). X-ray crystallography studies indicate that imino-acid-rich
sequences, such as repeating Gly-Pro-Hyp
units, adopt a helical symmetry of 7/2, while
imino-acid-poor regions have a symmetry
closer to 10/3, the average seen in collagen
fiber diffraction patterns (Kramer et al.,

1999). Such sequence-dependent variation

in superhelix pitch could play a role in
recognition and intermolecular interactions.
The high-resolution structures also show the
conformers and interactions of various side
chains in the triple helix (Kramer et al., 2000,
2001). Some side chains hydrogen bond
directly to available backbone carbonyls,
while many interactions are mediated by
water.
NMR studies on peptides with specifically
labeled residues have confirmed the oneresidue stagger and close packing of the
three chains in solution (Li et al., 1993). It
also has been possible to monitor the
dynamics and folding of individual residues
in the triple helix. Computational modeling
has been used to clarify the features and
interactions of the triple helix and, more
recently, to clarify the thermodynamic features of the helix (Mooney et al., 2001).