Proteomics: Methods and Protocols
Proteomics: Methods and Protocols
Proteomics: Methods and Protocols
Proteomics
Methods and Protocols
Methods in Molecular Biology
Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK
Edited by
Lucio Comai
Keck School of Medicine, University of Southern California, Los Angeles, CA, USA
Jonathan E. Katz
Keck School of Medicine, University of Southern California, Los Angeles, CA, USA
Parag Mallick
Stanford University, Palo Alto, CA, USA
Editors
Lucio Comai Jonathan E. Katz
Keck School of Medicine Keck School of Medicine
University of Southern California University of Southern California
Los Angeles, CA, USA Los Angeles, CA, USA
Parag Mallick
Stanford University
Palo Alto, CA, USA
In the catalog of biochemical techniques, proteomics has barely reached its adolescent stage,
and a very immature adolescent at that. Like any teen, potential still overshadows realized
accomplishment, but the future is still bright with potential. This particular adolescent has
shown quite a level of promise. Indeed, for a number of tasks, proteomics is fully proficient—
determining the identity of a small number of proteins, providing absolute quantitation of a
similar number of proteins. For others, it is still testing its limits: How many proteins? How
many orders of magnitude of sensitivity? And we begin to doubt, but it is not impossible to
imagine the realization of the full parental dream—given a sample, what are the concentra-
tions and identity of every protein and every modification on those proteins. And then there
are the unexpected questions we can answer—our teenager has shown potential in areas we
never first imagined: What is the three-dimensional structure of a protein? How do proteins
interact? What is the turnover rate of various post-translational modifications?
We view proteomics as a pipeline with four discrete components: The isolation of material
from a biological specimen, sample preprocessing, sample analysis, and data interpretation.
Recognizing proteomic analysis almost always is a collaborative effort and that special-
ized analyses will always require domain-specific knowledge, our goals with this book are to
provide step-by-step protocols on a wide range of biochemical methods, analytical
approaches, and bioinformatics tools developed to analyze the proteome. Here are our
specific goals for this book:
1. Accessible. Most scientists in the life sciences will be able to employ the methods described
in this book. Aside from basic mass spectrometers, we have avoided unusual and/or
expensive equipment and reagents. (Specialists do not consult books as a primary
reference.)
2. Practical. The techniques herein described are broadly applicable, commonly employed
protocols.
3. Current. Mature well-established protocols will be referenced and briefly described.
“State-of-the-art” emerging standard protocols will be clearly and completely described—
common wisdom included at no extra charge!
4. One stop. Recognizing that proteomics is often a collaborative effort, this book shall
describe, as we see it, the complete proteomic pipeline, upfront biology through data
analysis. For analysis that has become or is emerging as routine, our hopes are for this to
be the “go to” reference.
v
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
vii
viii Contents
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Contributors
ix
x Contributors
Abstract
Proteins play a key role in all aspects of cellular homeostasis. Proteomics, the large-scale study of proteins,
provides in-depth data on protein properties, including abundances and post-translational modification
states, and as such provides a rich avenue for the investigation of biological and disease processes. While
proteomic tools such as mass spectrometry have enabled exquisitely sensitive sample analysis, sample prep-
aration remains a critical unstandardized variable that can have a significant impact on downstream data
readouts. Consistency in sample preparation and handling is therefore paramount in the collection and
analysis of proteomic data.
Here we describe methods for performing protein extraction from cell culture or tissues, digesting the
isolated protein into peptides via in-solution enzymatic digest, and peptide cleanup with final preparations
for analysis via liquid chromatography-mass spectrometry. These protocols have been optimized and stan-
dardized for maximum consistency and maintenance of sample integrity.
Key words Proteomics, Protein extraction, Acetone precipitation, Enzymatic solution digest, Liquid
chromatography-mass spectrometry
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_1, © Springer Science+Business Media LLC 2017
1
2 Michelle Atallah et al.
2 Materials
2.4 C18 Cleanup 1. Honeywell Brand Reverse Phase A (RP-A): LC-MS water with
0.1 % formic acid.
2.4.1 Solutions
and Reagents 2. Honeywell Brand Reverse Phase B (RP-B): acetonitrile with
0.1 % formic acid.
3. Wetting solution (50 % Honeywell RP-B in RP-A).
4. Equilibration/Wash solution (2 % RP-B in Honeywell RP-A),
see Note 3.
5. Elution solution (90 % Honeywell RP-B in RP-A).
3 Methods
3.1 Cell Lysis 1. Fill an ice bucket with wet ice and fill up to ice level with water.
and Acetone 2. Prepare 500 μL of cell lysis buffer for each 1e7 cells.
Precipitation
(a) Lysis buffer composition: 3 % SDS, 0.02 M DTT, 0.10 M
Tris–HCl pH 7.5, 1× Thermo Protease/Phosphatase
inhibitor, 1× PMSF. Make fresh and add protease/phos-
phatase inhibitors, especially PMSF that is active in aque-
ous solution for 30 min, just before use.
(b) Preheat lysis buffer at 95 °C. See Note 4.
3. Keep cell pellet tubes on dry ice. Add preheated lysis buffer to
frozen cell pellet; 500 μL for each ten million cells (see Notes
5 and 6). Pipette and/or vortex to mix. Critical point: the
frozen pellet must not be allowed to thaw until covered with
hot SDS lysis solution—thawing then should be rapid and
right into the concentrated surfactant.
4. Put cap lock on tube. Place tube in 95 °C heat block for 3 min.
Vortex every 15–20 s.
5. If necessary, split contents into tubes of ~500 μL each, see
Note 7.
6. Pour chilled ice water from ice bucket into clear plastic/acrylic
box (see Note 8), avoiding ice as much as possible. Wedge an
ice pack to the bottom of the acrylic box to keep the water
cool (see Note 9), and a foam tube holder to float the tubes at
the surface.
7. Vent the tubes by opening them briefly (with the opening cap
oriented away from you). Place them in the foam holder in the
acrylic box.
8. Sonicate samples while the tubes remain in ice water (see
Note 10). There are two critical points in this step, both
facilitated by the use of the clear acrylic box: first, the samples
must be kept cold to avoid degradation of proteins due to the
heat generated by the sonicator. We have found that keeping
Protein Extraction and Digestion 5
20. Resuspend the pellet from the 10 % (v/v) aliquot precipitation
from step 10a in 100–300 μL 3 % SDS and perform BCA (see
Note 24) quantification according to the manufacturer’s
instructions. From the results determine how much protein is
in each acetone-precipitated pellet based on the fraction of the
total sample that went into each tube. Critical point: ensure
complete resuspension and solubilization of the pellet. In addi-
tion to visually inspecting the pellet to confirm complete resus-
pension, vortexing the sample and/or heating at 37 °C for up
to 2 min can aid in resolubilization.
3.2 Peptide 1. Bring up dried acetone pellet in 50 μL 8 M urea, 100 mM Tris
Digestion pH 8 for a final concentration of 80 μg in 25 μL (see Note 25).
(a) Use manual pipetting and if necessary heat and/or sonica-
tion to break up pellet. Avoid any heating over 37 °C to
avoid carbamylation by urea.
2. Add TCEP to final concentration of 5 mM and incubate for
RT for 30 min (see Note 26). Mix well by vortexing, and then
knock down by pulse spinning in microfuge.
3. Add IAA to final concentration of 10 mM and incubate for RT
for 30 min in the dark (covered in foil, see Note 27). Mix well
by vortexing, and then knock down by pulse spinning in
microfuge.
4. Bring up to 250 μL with 100 mM Tris pH 8, reducing the
urea concentration to <1 M (see Note 28). Mix well by vortex-
ing, and then knock down by pulse spinning in microfuge.
5. Add 100 mM CaCl2 to a final concentration of 1 mM. Mix
well by vortexing, and then knock down by pulse spinning in
microfuge.
6. Add 8 μg trypsin (1:10 mass:mass) for a final concentration of
30 ng/μL. Mix well by vortexing, and then knock down by
pulse spinning in microfuge.
7. Incubate overnight at 37 °C in static or shaking incubator.
3.3 C18 Cleanup The following is a modified protocol is based on the manufactur-
er’s recommendations for Pierce C18 tips (100 μL bed, Catalog
No. 87784).
Preloading tubes, or well plates, with each solution to be used
can increase throughput speed and minimizes downtime in which
the tip can dry. Each 100 μL aliquot of sample should be loaded
separately, and should be separately aliquoted into wells or tubes.
1. Wet C18 filter tip by aspirating 100 μL of wetting solution and
then discarding solvent (see Note 29).
2. Repeat wash step and discard solvent.
3. Equilibrate tip by aspirating 100 μL of equilibration/wash
solution and discarding the solvent.
Protein Extraction and Digestion 7
4 Notes
30. Aspirations up and down into the tip should be done slowly
(each single up/down cycle over 3–5 s, and multiple aspira-
tions up and down into the tip can be done to improve effi-
ciency. Recommended are 10 aspirations for each 100 μL
aliquot of sample loaded, and 5 for each 100 μL elution, and
3 aspirations at every other step. With experience and care,
implementation an 8-channel multichannel pipette and well
plates can improve throughput speed.
31. This requires careful monitoring of the drying, which can be
achieved with a strobe device (e.g. Labconco Centrizap) that
allows visualization of the sample without stopping the
Speed-Vac.
32. Avoid any pelleted material at bottom of tube. This final spin
step helps remove any particulates or debris that may clog the
mass spectrometer.
References
1. Mallick P, Kuster B (2010) Proteomics: a techniques for proteome analysis by mass spec-
pragmatic perspective. Nat Biotechnol 28:
trometry. J Chromatogr A 1418:158–166
695–709 3. Wiśniewski JR, Zougman A, Nagaraj N, Mann
2. Kachuk C, Stephen K, Doucette A (2015) Com M (2009) Universal sample preparation method
parison of sodium dodecyl sulfate depletion for proteome analysis. Nat Methods 6:359–362
Chapter 2
Abstract
Enhanced Filter Aided Sample Preparation (eFASP) incorporates plastics passivation and digestion-
enhancing surfactants into the traditional FASP workflow to reduce sample loss and increase hydrophobic
protein representation in qualitative and quantitative proteomics experiments. Resulting protein digests
are free of contaminants and can be analyzed directly by LC-MS.
Key words Enhanced filter aided sample preparation, Quantitative proteomics, Detergent,
Ammonium deoxycholate
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_2, © Springer Science+Business Media LLC 2017
11
12 Jonathan Erde et al.
this method can suffer sample losses near 50 % [5]. Enhanced FASP
(eFASP) addresses this challenge by incorporating deoxycholic
acid and, optionally, TWEEN® 20 into the FASP workflow.
Deoxycholic acid (DCA) is a secondary bile acid that, amongst
many other uses, is employed as a mild detergent for membrane
proteins. It increases the efficiency with which trypsin digests cyto-
solic and membrane proteins and is easily removed by acidification
and phase transfer (PT) to peptide-immiscible ethyl acetate (EA) in
liquid–liquid extraction [1, 6–11]. PT decreases EA-soluble con-
taminants, including SDS, n-octylglucoside, NP-40, and Triton
X-100 [12].
The eFASP protocol optionally uses the surfactant TWEEN®-
20 to passivate surfaces of Microcon® filter units and collection
tubes. TWEEN®-20, as well as SDS, are recognized choices for
minimizing protein binding to surfaces and have been recom-
mended by Amicon Centricon for use in filter units [13]. Passivation
of the filter units and collection tubes used in eFASP can reduce
peptide and protein loss due to nonspecific surface binding, but
care is needed to prevent TWEEN®-related ions from contaminat-
ing mass spectra.
Presented here is the eFASP approach, which utilizes 0.2 %
DCA and (optionally) TWEEN®-20 to quantitatively increase
recovery and proteomic coverage of hydrophilic and hydrophobic
proteins. An express eFASP method variant is also included, which
uses a one-step reduction/alkylation employing tris(2-
carboxyethyl)phosphine (TCEP) and 4-vinylpyridine (4-VP) prior
to deposition on the Microcon® filter, increasing alkylation speci-
ficity and speeding processing [1, 14–17].
2 Materials
3 Methods
3.1 eFASP: Standard 1. On a shaker, incubate filter units and collection tubes over-
night in Passivation Solution. Small batches of items may be
3.1.1 Surface
incubated in 50 ml Falcon centrifuge tubes.
Passivation (Optional)
2. With clean tweezers, remove each item and rinse its outer and
inner surfaces with MS-grade H2O dispensed from a squeeze
bottle.
3. Transfer items to a clean beaker containing a large volume of
MS-grade H2O; e.g., 250 ml or more. Incubate items for
30 min at room temperature, shaking at low speed.
4. Repeat step 3 two additional times with fresh MS-grade H2O.
5. Reserve the passivated collection tubes for peptide recovery
from ultrafiltration devices.
3.1.2 Sample Lysis 1. Wash and pellet cells according to established guidelines for
cell type.
2. Add sufficient Lysis Buffer to the pelleted cells such that a
25 μl aliquot of lysate will provide the quantity desired for
processing by eFASP (or a maximum protein concentration of
10 μg/μl). Ensure that the selected volume and tube size can
accommodate the sonicator probe.
3. Place the lysate into a 90 °C thermo-mixer and incubate for
10 min, shaking at 600 rpm. Remove lysate and decrease
thermo-mixer temperature to 37 °C for later use.
4. Sonicate lysate (employing sonicator/homogenization probe)
three times for 10 s each.
5. Centrifuge the lysate at 14,000 × g for 10 min.
6. Repeat sonication and centrifugation once (steps 4 and 5).
7. Sonicate the lysate (including any pelleted material), for 10 s.
Cool to 37 °C.
3.1.3 Sample Processing 1. Transfer 25 μl of lysate to a sample tube containing 200 μl of
Exchange Buffer. Vortex briefly to mix.
2. Place a (passivated) filter unit atop a non-passivated collection
tube.
3. Dispense the 225 μl lysate/Exchange Buffer sample to the fil-
ter unit and centrifuge at 14,000 × g for 10 min. Discard
filtrate.
4. Add 200 μl Exchange Buffer to the filter unit and centrifuge at
14,000 × g for 10 min. Discard filtrate.
5. Repeat step 4 two more times.
Enhanced FASP for Proteomic Experiments 15
3.1.4 Phase Transfer 1. To the collection tube with peptide-containing filtrate, add
200 μl of ethyl acetate and transfer to a 2 ml Eppendorf
LoBind® tube.
2. Add 2.5 μl TFA and quickly vortex. A white, thread-like pre-
cipitate may be visible if a large quantity of peptide is present.
3. Add ethyl acetate to nearly fill the tube, leaving only enough
space to agitate without losing liquid.
4. Agitate the mixture for 10 s in an ultrasonic bath and centri-
fuge at 16,000 × g for 10 min.
5. Carefully pipet most of the upper (organic) layer into a tube
for discard. Do not disturb the organic/aqueous boundary
layer.
6. Repeat steps 3–5 two times.
7. Place the uncapped sample tube in a 60 °C thermo-mixer, in a
fume hood, for 5 min to remove residual ethyl acetate.
8. Remove residual organic solvent and volatile salts by vacuum
drying in a SpeedVac®.
9.
Resuspend the dried sample in 50
% methanol and
vacuum-dry.
10. Repeat step 9 two times.
16 Jonathan Erde et al.
3.2 Express eFASP 1. Wash and pellet cells according to established guidelines for
cell type.
3.2.1 Surface
Passivation 2. Add sufficient Lysis Buffer to the pelleted cells such that a
(Optional) 25 μl aliquot of lysate will provide the quantity desired for
processing by eFASP (or a maximum protein concentration of
3.2.2 Sample Lysis 10 μg/μl). Ensure that the selected volume and tube size can
accommodate the sonicator probe.
3. Place the lysate into a 90 °C thermo-mixer and incubate for
10 min, shaking at 600 rpm. Remove lysate and decrease
thermo-mixer temperature to 37 °C for later use.
4. Sonicate lysate (employing sonicator/homogenization probe)
three times for 10 s each.
5. Centrifuge the lysate at 14,000 × g for 10 min.
6. Repeat sonication and centrifugation once (steps 4 and 5).
7. Sonicate the lysate (including any pelleted material), for 10 s.
Cool to 37 °C.
8. Add Alkylation Stock to the lysate to a final concentration of
25 mM 4-VP, and place the sample tube into a 37 °C thermo-
mixer for 1 h at 300 rpm.
9. Add Quench Buffer to the lysate to a final concentration of
40 mM DTT.
3.2.3 Sample Processing 1. Transfer 25 μl of lysate to a sample tube containing 200 μl of
Exchange Buffer. Vortex briefly to mix.
2. Place a (passivated) filter unit atop a non-passivated collection
tube.
3. Dispense the 225 μl lysate/Exchange Buffer sample to the fil-
ter unit and centrifuge at 14,000 × g for 10 min. Discard
filtrate.
4. Add 200 μl eFASP Digestion Buffer to the filter unit and cen-
trifuge at 14,000 × g for 10 min. Discard filtrate.
5. Repeat step 4 two more times.
6. Detach the filter unit from the non-passivated collection tube,
and place it on top of a passivated tube.
7. Add 100 μl eFASP Digestion Buffer to the filter unit.
8. Calculate the volume of Trypsin Buffer to dispense in order to
achieve the desired enzyme-to-substrate ratio; e.g., 1:50 w:w.
9. Deposit the calculated volume of Trypsin Buffer to the filter
unit, and move the filter/collection tube assembly to a 37 °C
thermo-mixer for 12 h of shaking at low speed. Cap the filter
unit to reduce evaporation.
10. Remove the filter/collection tube assembly from the thermo-
mixer and centrifuge at 14,000 × g for 10 min. Retain the
peptide-containing filtrate.
Enhanced FASP for Proteomic Experiments 17
3.2.4 Phase Transfer 1. To the collection tube with peptide-containing filtrate, add
200 μl of ethyl acetate and transfer to a 2 ml Eppendorf
LoBind® tube.
2. Add 2.5 μl TFA and quickly vortex. A white, thread-like pre-
cipitate may be visible if a large quantity of peptide is present.
3. Add ethyl acetate to nearly fill the tube, leaving only enough
space to agitate without losing liquid.
4. Agitate the mixture for 10 s in an ultrasonic bath and centri-
fuge at 16,000 × g for 10 min.
5. Carefully pipet most of the upper (organic) layer into a tube
for discard. Do not disturb the organic/aqueous boundary
layer.
6. Repeat steps 3–5 two times.
7. Place the uncapped sample tube in a 60 °C thermo-mixer, in a
fume hood, for 5 min to remove residual ethyl acetate.
8. Remove residual organic solvent and volatile salts by vacuum
drying in a SpeedVac®.
9. Resuspend the dried sample in 50 % methanol and
vacuum-dry.
10. Repeat step 9 two times.
4 Notes
References
1. Erde J, Loo RRO, Loo JA (2014) Enhanced 10. Lin Y, Zhou J, Bi D, Chen P, Wang X, Liang
FASP (eFASP) to increase proteome coverage S (2008) Sodium-deoxycholate-assisted tryp-
and sample recovery for quantitative proteomic tic digestion and identification of proteolyti-
experiments. J Proteome Res 13(4):1885–95 cally resistant proteins. Anal Biochem
2. Erde J (2012) High throughput analysis of 377(2):259–66
proteome perturbations induced by radiation, 11. Lin Y, Liu Y, Li J, Zhao Y, He Q, Han W et al
radiomitigators and chemotherapeutics. (2010) Evaluation and optimization of
University of California, Los Angeles removal of an acid-insoluble surfactant for
3. Manza LL, Stamer SL, Ham A-JL, Codreanu shotgun analysis of membrane proteome.
SG, Liebler DC (2005) Sample preparation Electrophoresis 31(16):2705–13
and digestion for proteomic analyses using 12. Yeung Y-G, Nieves E, Angeletti RH, Stanley
spin filters. Proteomics 5(7):1742–5 ER (2008) Removal of detergents from pro-
4. Wisniewski JR, Mann M (2009) Spin filter– tein digests for mass spectrometry analysis.
based sample preparation for shotgun pro- Anal Biochem 382(2):135–7
teomics. Nat Methods 6(11):785–6 13. Passivation of Amicon Microcon Concentrators
5. Wisniewski JR, Zielinska DF, Mann M (2011) for Improved Recovery (1999). Bedford, MA:
Comparison of ultrafiltration units for pro- Millipore Corporation, Technical Note
teomic and N-glycoproteomic analysis by the PC1001EN00
filter-aided sample preparation method. Anal 14. Sebastiano R, Citterio A, Lapadula M, Righetti
Biochem 410(2):307–9 PG (2003) A new deuterated alkylating agent
6. Masuda T, Tomita M, Ishihama Y (2008 Feb) for quantitative proteomics. Rapid Commun
Phase transfer surfactant-aided trypsin diges- Mass Spectrom 17(21):2380–6
tion for membrane proteome analysis. 15. Bai F, Liu S, Witzmann FA (2005) A “de-
J Proteome Res 7(2):731–40 streaking” method for two-dimensional elec-
7. Masuda T, Sugiyama N, Tomita M, Ishihama trophoresis using the reducing agent
Y (2011) Microscale phosphoproteome analy- tris(2-carboxyethyl)-phosphine hydrochloride
sis of 10,000 cells from human cancer cell and alkylating agent vinylpyridine. Proteomics
lines. Anal Chem 83(20):7698–703 5(8):2043–7
8. Masuda T, Saito N, Tomita M, Ishihama Y 16. Liu S, Bai F, Witzmann F (2006) Destreaking
(2009) Unbiased quantitation of escherichia coli strategies for two-dimensional electrophoresis.
membrane proteome using phase transfer sur- In: Separation methods in proteomics. Eds.:
factants. Mol Cell Proteomics 8(12):2770–7 Smejkal GB, Lazarev A, CRC Press/Taylor &
9. Zhou J, Zhou T, Cao R, Liu Z, Shen J, Chen Francis, Boca Raton, pp. 207–17
P et al (2006) Evaluation of the application of 17. Righetti PG (2006 Sep) Real and imaginary arte-
sodium deoxycholate to proteomic analysis of facts in proteome analysis via two-dimensional
rat hippocampal plasma membrane. maps. J Chromatogr B 841(1–2):14–22
J Proteome Res 5(10):2547–53
Chapter 3
Abstract
The biological functions of given genomic regions are ruled by the local chromatin composition. The
Proteomics of Isolated Chromatin segments approach (PICh) is a powerful and unbiased method to ana-
lyze the composition of chosen chromatin segments, provided they are abundant (repeated) or that the
organism studied has a small genome. PICh can be used to identify novel and unexpected regulatory fac-
tors, or when combined with quantitative mass spectrometric approaches, to characterize the function of
a defined factor at the chosen locus, by quantifying composition changes at the locus upon removal/addi-
tion of that factor.
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_3, © Springer Science+Business Media LLC 2017
19
20 Sophie L. Kan et al.
Preclearing Telomeric
chromatin
Denaturation
Hybridization
LNA sequence
Hybrids capture
Spacer
Desthibiotin
Streptavidin
dynabead Protein
precipitation
Washes Elution Crosslink reversal
Western-blot
Mass-spectrometry
2 Materials
3 Methods
3.1 Tissue Culture 1. Plate cells on enough 150 mm plates to harvest 109 cells per
condition (see Note 1).
2. Incorporation of isotopically stable amino acids into cellular
proteins for qPICh. Grow the two cell lines (e.g., WT and
mutant for a gene of interest) in culture media with isotopi-
cally distinct amino acids (see Note 2).
3. Count cells of one or two representative plate for each condi-
tion. For qPICh this step is extremely important.
4. FACS analyze the cell cycle profile of the cell populations to
account for potential differences cell cycle distributions among
the distinct cell populations.
3.5 Hybridization 1. Split the chromatin in two equal volumes, one for the experi-
and Chromatin ment and the other for the scrambled control (see Note 17).
Capture 2. Add the LNA probe to a final concentration of 0.25 μM
(see Note 18).
3. Aliquot 150 μL of the chromatin-LNA mixture into PCR
tubes.
26 Sophie L. Kan et al.
3.6 Washes 1. Wash the beads five times with lysis buffer. Resuspend the beads
gently between the washes, avoid vortexing (see Note 21).
Perform the additional wash with low salt buffer.
2. Resuspend the beads in 1.2 mL lysis buffer per tube and trans-
fer to low binding 1.5 mL microtubes (see Note 22).
3. Immobilize the beads on the magnetic stand.
4. Discard the supernatant.
5. Resuspend in 1 mL of lysis buffer.
6. Immobilize the beads on the magnetic stand. A “halo” should
be apparent on the sides of the wall of the microcentrifuge
tube (see Note 23).
7. Discard the supernatant.
Proteome Characterization of a Chromatin Locus… 27
3.8 Protein 1. Prepare the inputs: add 890 μL of lysis buffer to the 10 μL of
Precipitation input saved at step 37.
2. Precipitate proteins by adding 200 μL of 100 % trichloroacetic
acid to the 900 μL of eluates (the final concentration of tri-
chloroacetic acid should be in the range of 15–20 %).
3. Incubate on ice for 10 min.
4. Centrifuge at 16,000 × g for 10 min at 4 °C.
5. Carefully remove the supernatant with a pipette and leave
about 50 μL (see Note 26).
6. Wash the pellet twice with 1 mL of ice-cold acetone 100 %.
7. Vortex for 10 s.
8. Centrifuge at 16,000 × g for 10 min at 4 °C.
9. Carefully remove the supernatant and leave about 50 μL.
10. Evaporate the acetone by heating the tube at 100 °C for 10 s.
Repeat this step if the acetone is not entirely evaporated.
11. Resuspend in 70 μL of cross-linking reversal solution if the
sample will be analyzed by mass spectrometry or in 100 μL of
reversal cross-linking solution if the sample is used for western-
blot analysis. In both cases, resuspend the input in 400 μL of
cross-linking reversal solution.
12. Incubate for 24 min at 99 °C, after 12 min vortex briefly and
centrifuge to retrieve condensation from the cap back to the
bottom of the tube and incubate another 12 min at 99 °C.
28 Sophie L. Kan et al.
3.9 Control for PICh 1. Separate proteins on a 12 % Bis-tris gel run with MOPS buffer.
Efficiency Load 2 μL of input and 5–10 μL of scrambled and PICh
extracts (see Note 27).
2. Silver stain the gel to reveal the proteins (see Note 28).
If 1/6th of eluted material was used, there should be visible
proteins bands after 3 min of developing. If the bands do
not appear after 5 min it means there are not enough pro-
teins in the remaining PICh extracts for analysis by mass
spectrometry (Fig. 2).
3. The PICh extracts can be used for western-blot analysis
(Fig. 3), mass spectrometry analysis or stored at −80 °C.
3.10 Protein Analysis 1. Resolve proteins (usually up to 40 % of the total, ~30 μL) on a
by Mass 12 % Bis-Tris gel.
Spectrometric 2. Cut each lane into 5–10 bands.
Approaches
3. Process each band using standard mass spectrometric identifi-
cation protocols.
191 kDa
97 kDa
64 kDa
51 kDa
39 kDa
28 kDa
19 kDa
1 2 1 2 1 2
TRF1
H3
4 Notes
10 kb
8 kb
6 kb
5 kb
4 kb
3 kb
2 kb
1.5 kb
1 kb
0.5 kb
Fig. 4 Agarose gel electrophoresis of DNA isolated from PICh extracts. Size dis-
tribution of purified mouse embryonic stem cell DNA after sonication. 2 μg of
DNA is analyzed on a 0.8 % agarose gel, post-stained with ethidium bromide
extensively wash the migration tank with MilliQ water, use com-
mercial pre-cast gels, premade buffers, loading dye etc., that
have limited chances to be contaminated by keratins or other
unwanted protein. Stain your gel in a brand new plastic dish
(diam 150 mm) and open the cover only to change buffers.
6. A 4–5 mL volume of cell pellet is sufficient for the sequence-
specific pull-down and the scramble pull-down.
7. Flash freeze the cross-linked material in liquid nitrogen and
store it at −80 °C to minimize cross-linking reversal of pro-
teins from the chromatin. It is not recommended to use mate-
rial stored longer than one month as we observed significant
de-cross-linking and much lower protein retrieval. The best
results are obtained when fresh material is used and the stor-
age steps are skipped.
8. It is extremely important that the same number of cells is
mixed for quantitative mass spectrometry analysis.
9. Resuspend pellets by vortexing until the sonication step 22.
To avoid foam vortex very gently and slowly increase to the
maximal speed. Using a pipette to resuspend results in losing
a significant amount of chromatin that sticks to the plastic. Pay
attention to minimize material loss at each step, as the sample
sticks to plastic ware.
10. This RNase A step is to optimize the subsequent LNA probe–
chromatin hybridization step and to avoid with the nonspe-
cific capture of RNA–protein complexes.
11. Do not vortex, as vortexing leads to excessive foaming of the
sample which inhibits sonication efficiency. Resuspend by
pipetting with 1 mL micropipette.
12. At this level of cross-linking, indirect ultrasonication in water
baths does not solubilize the chromatin. We have recently
developed an alternative solubilization method involving
restriction enzyme digestion and high pressure solubilization
with a French high pressure system [3].
13. Heating at 58 °C favors the unmasking of endogenously bio-
tinylated proteins which have to be cleared from the sample.
14. Pre-clearing is necessary to remove most of endogenously bio-
tinylated proteins that might compete with desthiobiotinyl-
ated probe during the streptavidin binding step. If the signal is
weak on silver-stained gel, reduce the pre-clear to 2 h at room
temperature instead of overnight.
15. This step is very important for the quality of the purification.
The ultra-fast gel filtration reduces the salt concentration of
the sample by about two-thirds leaving roughly 30 mM of
NaCl. Low salt concentrations make the hybridization step
more stringent. Also, it will prevent nonspecific precipitation
of chromatin to the dynabeads during the capture. The use of
32 Sophie L. Kan et al.
26. Put the 1.5 mL microcentrifuge tube with the hinge outwards.
The protein pellet will end up all along on this side of the wall
of the tube which explains why it is sometimes difficult to see.
27. If the PICh assay is very clean the scrambled extract should be
devoid of any proteins. Most frequent contaminants are his-
tones. The PICh extract should show different banding pat-
terns that input meaning the probe used specifically purified
proteins bound to the genomic region of interest.
28. Silver staining is about 50–100 times more sensitive than
Coomassie blue staining. Although some Silver staining meth-
ods are compatible with protein identification by mass spec-
trometry, we found that the analysis from Silver stained bands
strongly decreased the sensitivity of the analysis. Thus, while
the identification of proteins from such silver stained gels is
technically feasible, we do not recommend this.
29. The direct mass spectrometric analysis of liquid samples is
doable. However, it reproducibly resulted in a much
smaller and less complex proteome and therefore is not
recommended.
30. The DNA smear ranges from below 0.5 to more than 3 kb.
This is higher than the average fragment size distribution
obtained after the typical lower cross-linking conditions used
in ChIP (1 % HCHO for 10 min). Do not pre-stain, but post-
stain the gel for accurate size estimation.
Acknowledgments
We would like to thank Titia de Lange for the kind gift of the
TRF1 antibody.
References
1. Carey MF, Peterson CL, Smale ST (2009) 6. Landgraf R, Chen CH, Sigman DS (1995)
Chromatin immunoprecipitation (ChIP). Cold R-loop stability as a function of RNA structure
Spring Harb Protoc 9:pdb prot5279 and size. Nucleic Acids Res 23:3516–3523
2. Dejardin J, Kingston RE (2009) Purification of 7. Hirsch JD, Eslamizar L, Filanoski BJ,
proteins associated with specific genomic Loci. Malekzadeh N, Haugland RP, Beechem JM,
Cell 136:175–186 Haugland RP (2002) Easily reversible desthio-
3. Ide S, Dejardin J (2015) End-targeting pro- biotin binding to streptavidin, avidin, and other
teomics of isolated chromatin segments of a biotin-binding proteins: uses for protein label-
mammalian ribosomal RNA gene promoter. ing, detection, and isolation. Anal Biochem
Nat Commun 6:6674 308:343–357
4. Braasch DA, Corey DR (2001) Locked nucleic 8. Saksouk N, Barth TK, Ziegler-Birling C, Olava
acid (LNA): fine-tuning the recognition of N, Nowak A, Rey E, Mateos-Langerak J,
DNA and RNA. Chem Biol 8:1–7 Urbach S, Reik W, Torres-Pedilla ME, Imhof
5. Belotserkovskii BP, Reddy G, Zarling DA A, Dejardin J, Simboeck E (2014) Redundant
(1999) DNA hybrids stabilized by heterologies. mechanisms to form silent chromatin at peri-
Biochemistry 38:10785–10792 centromeric regions rely on BEND3 and DNA
methylation. Mol Cell 56:580–594
Chapter 4
Abstract
Proteins are very dynamic within the cell and their localization and trafficking between subcellular com-
partments are critical for their correct function. Indeed, the abnormal localization of a protein might lead
to the pathogenesis of several diseases. The association of cell fractionation methods and mass spectrom-
etry based proteomic methods allow both the localization and quantification of proteins in different sub-
compartments. Here we present a detailed protocol for enrichment, identification, and quantitation of the
nuclear proteome in cell lines combining nuclear subproteome enrichment by differential centrifugation
and high-throughput proteomics.
Key words Nuclear fractionation, Subcellular proteomics, Protein localization, Cell Line, Mass
spectrometry
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_4, © Springer Science+Business Media LLC 2017
35
36 Aline Poersch et al.
2 Materials
2.1 Cell Culture 1. Cell culture media: Mammary Epithelial Cell Growth
Medium (MEBM) supplemented with 100 ng/mL cholera
toxin (Sigma), MEGM SingleQuots kit (bovine pituitary
extract—BPE, human epidermal growth factor—hEGF,
Profiling Nuclear Sub-proteome 37
2.2 Cell Lysis 1. Lysis buffer: 50 mM HEPES pH 7.4, 10 mM NaCl, 5 mM
Components MgCl2, 0,1 mM EDTA, 1 mM Na3VO4 (fosfatase inhibitor),
1 mM NaF and 1 mM Na4P2O7.10dH2O, with protease inhib-
itor cocktail 5 % (v/v) (Sigma-Aldrich) (see Note 2).
2. Cell scraper.
3. Ice-cold phosphate buffered saline (PBS), pH 7.2.
4. 3 mL syringe with needle
5. Tissue homogenizer model D-130 (Biosystems) or equivalent
model.
6. 2 mL microcentrifuge tubes.
7. Refrigerated microcentrifuge.
2.3 Nuclear 1. Nuclear protein extraction buffer: 8 M urea and 2 % (v/v)
Fractionation CHAPS containing 5 % (v/v) protease inhibitor cocktail and
Components 1 mM Na3VO4 (see Note 3).
2. Vortex mixer.
3. Refrigerated microcentrifuge.
2.4.2 In-Gel Trypsin 1. Clean glass plate and sterile scalpel (see Note 7).
Digestion 2. Ammonium bicarbonate 100 mM pH 8 solution: Dissolve
1.58 g of ammonium bicarbonate in 200 mL of ultrapure
water (see Note 8).
3. Destain solution: 50 mL of ammonium bicarbonate 100 mM
pH 8 with 50 mL of 100 % acetonitrile.
4. Trypsin solution: Resuspend one vial containing 20 μg of
sequencing-grade modified trypsin (Promega) in 100 μL of
ammonium bicarbonate 100 mM pH 8 solution (see Note 9).
5. Thermomixer.
3 Methods
3.1 Cell Lysis 1. Cultivate MCF10A cell line in 75 cm2 flasks in described cul-
ture media. For the fractionation, plate 5 × 106–1 × 107 cells in
a 100 mm culture dish.
2. After 24 h and/or cellular confluence >80 %, remove the
media, wash the cells twice with pre-warmed PBS, substitute
with new culture media and proceed with the treatments of
interest.
Profiling Nuclear Sub-proteome 39
3.2 Preparation 1. After treatments of interest, remove the cell media and wash
of Cell Lysate the cells twice with ice-cold PBS. Remove any solution excess
that remains on the dish. Add 150 μL of lysis buffer contain-
ing protease inhibitors in each dish and scrape cells using a cell
scraper. Collect cell lysate and transfer to 2 mL microtubes.
Keep samples on ice.
2. Pass the cell lysate 20 times through a thin 25-gauge needle
using a 3 mL syringe and homogenize in a tissue homogenizer
(dounce homogenizer) for 1 min. Keep samples on ice.
3. Centrifuge samples at 500 × g for 20 min at 4 °C to pellet
nuclei (no brake applied during the deceleration of the centri-
fuge). Carefully transfer all of the supernatant to a 1.5 mL
microtube (see Note 10).
3.3 Nuclear 1. Add to the dry pellet obtained previously (see Subheading 3.2)
Fractionation 100 μL of nuclear extraction buffer. Nuclear proteins extrac-
tion are obtained by sonication of samples using an ultrasound
probe for 5 min followed by cycles of vortexing for 20 s and
ice bath for 5 min. Repeat this cycle three times.
2. Centrifuge samples at 20,000 × g for 30 min at 4 °C and trans-
fer the supernatant enriched in nuclear proteins to a 1.5 mL
microtube. Store samples at −80 °C until use.
3.4.2 In-Gel Trypsin 1. Place the gel on a clean glass plate and excise each lane con-
Digestion taining proteins with a clean scalpel. Cut each lane into
approximately 1 cm square pieces and transfer to a clean
1.5 mL microtube pre-rinsed with methanol.
40 Aline Poersch et al.
3.6 High-Throughput 1. Carry out the high throughput LC-MS/MS data collection
Mass Spectrometry for each individual fraction. Inject 10 μL of peptide extract
and analyze samples over a 90 min linear gradient from 5 to
35 % of organic solvent at 350 nL/min in the system described
in section 2.6 (see Note 14).
2. Process LC-MS/MS files through data bank search, protein
inference and quantitative analysis (see Note 15).
3. Match the lists of proteins identified in the nuclear enriched
fraction with proteins identified in the total cell extract, cyto-
plasmic or membrane enriched fractions (see Note 10). Select
for nuclear proteins based on the higher value of enrichment
obtained from the ratio of spectral counts observed in enriched
nuclear fractions/total cell extract cytoplasmic or membrane
enriched fraction profile (see Note 16). A Sample dataset
obtained for MCF10A cell line is presented in Table 1 and
Fig. 1 (see Note 17).
4 Notes
MCF10A
mitochondrion (GO:0005739)
2% 1%
2% 2% 2% 1% endosome (GO:0005768)
2%
4% Golgi apparatus (GO:0005794)
5%
6%
vacuole (GO:0005773)
53%
18% 28%
Fig. 1 Protein dataset for nuclear profile of MCF10A breast epithelial cell line. (a) Venn diagram of proteins
identified by LC-MS/MS in the nuclear and cytoplasmic enriched fractions of MCF10A cells. The fractionation
methodology allowed confident overall protein identification of 2220 proteins with less than 1 % FDR. The
overlap between nuclear enriched and cytoplasmic enriched fraction (55 %) as well as the number of proteins
identified in only one of those fractions (45 %) shows the complementarity of both subcellular proteomic pro-
files. (b) Gene Ontology classification of proteins enriched in the nuclear and cytoplasmic compartments also
supports the complementarity of the proteomic profiles
Acknowledgments
References
1. Hung MC, Link W (2011) Protein localiza- generative model. Bioinformatics 31(12):i365–
tion in disease and therapy. J Cell Sci 124(Pt i374. doi:10.1093/bioinformatics/btv264
20):3381–3392. doi:10.1242/jcs.089110 3. Butler GS, Overall CM (2009) Proteomic identifi-
2. Simha R, Briesemeister S, Kohlbacher O, cation of multitasking proteins in unexpected loca-
Shatkay H (2015) Protein (multi-)location tions complicates drug targeting. Nat Rev Drug
prediction: utilizing interdependencies via a Discov 8(12):935–948. doi:10.1038/nrd2945
46 Aline Poersch et al.
4. Tomas A, Futter CE, Eden ER (2014) EGF export. J Biol Chem 277(10):8517–8523.
receptor trafficking: consequences for signal- doi:10.1074/jbc.M108867200
ing and cancer. Trends Cell Biol 24(1):26–34. 17. Drissi R, Dubois ML, Boisvert FM (2013)
doi:10.1016/j.tcb.2013.11.002 Proteomics methods for subcellular pro-
5. Chahine MN, Pierce GN (2009) Therapeutic teome analysis. FEBS J 280(22):5626–5634.
targeting of nuclear protein import in patho- doi:10.1111/febs.12502
logical cell conditions. Pharmacol Rev 18. Lee YH, Tan HT, Chung MC (2010)
61(3):358–372. doi:10.1124/pr.108.000620 Subcellular fractionation methods and strate-
6. Turner JG, Sullivan DM (2008) CRM1- gies for proteomics. Proteomics 10(22):3935–
mediated nuclear export of proteins and drug 3956. doi:10.1002/pmic.201000289
resistance in cancer. Curr Med Chem 19. Ramsby ML, Makowski GS, Khairallah EA
15(26):2648–2655 (1994) Differential detergent fractionation of iso-
7. Takeda A, Yaseen NR (2014) Nucleoporins lated hepatocytes: biochemical, immunochemical
and nucleocytoplasmic transport in hemato- and two-dimensional gel electrophoresis charac-
logic malignancies. Semin Cancer Biol 27:3– terization of cytoskeletal and noncytoskeletal
10. doi:10.1016/j.semcancer.2014.02.009 compartments. Electrophoresis 15(2):265–277
8. Kau TR, Way JC, Silver PA (2004) Nuclear 20. Sawhney S, Stubbs R, Hood K (2009)
transport and cancer: from mechanism to Reproducibility, sensitivity and compatibility
intervention. Nat Rev Cancer 4(2):106–117. of the ProteoExtract subcellular fractionation
doi:10.1038/nrc1274 kit with saturation labeling of laser microdis-
9. Wang SC, Hung MC (2005) Cytoplasmic/ sected tissues. Proteomics 9(16):4087–4092.
nuclear shuttling and tumor progression. Ann doi:10.1002/pmic.200800949
N Y Acad Sci 1059:11–15. doi:10.1196/ 21. Walther TC, Mann M (2010) Mass
annals.1339.002 spectrometry-based proteomics in cell biology.
10. Fabbro M, Henderson BR (2003) Regulation J Cell Biol 190(4):491–500. doi:10.1083/
of tumor suppressors by nuclear-cytoplasmic jcb.201004052
shuttling. Exp Cell Res 282(2):59–69 22. Ong SE, Mann M (2006) A practical recipe for
11. Salmena L, Pandolfi PP (2007) Changing ven- stable isotope labeling by amino acids in cell
ues for tumour suppression: balancing destruc- culture (SILAC). Nat Protoc 1(6):2650–
tion and localization by monoubiquitylation. 2660. doi:10.1038/nprot.2006.427
Nat Rev Cancer 7(6):409–413. doi:10.1038/ 23. Rauch A, Bellew M, Eng J, Fitzgibbon M,
nrc2145 Holzman T, Hussey P, Igra M, Maclean B, Lin
12. Santiago A, Li D, Zhao LY, Godsey A, Liao D CW, Detter A, Fang R, Faca V, Gafken P, Zhang
(2013) p53 SUMOylation promotes its H, Whiteaker J, States D, Hanash S, Paulovich
nuclear export by facilitating its release from A, McIntosh MW (2006) Computational pro-
the nuclear export receptor CRM1. Mol Biol teomics analysis system (CPAS): an extensible,
Cell 24(17):2739–2752. doi:10.1091/mbc. open-source analytic system for evaluating and
E12-10-0771 publishing proteomic data and high through-
13. Ohtani N, Brennan P, Gaubatz S, Sanij E, put biological experiments. J Proteome Res
Hertzog P, Wolvetang E, Ghysdael J, Rowe M, 5(1):112–121. doi:10.1021/pr0503533
Hara E (2003) Epstein-Barr virus LMP1 blocks 24. Keller A, Nesvizhskii AI, Kolker E, Aebersold
p16INK4a-RB pathway by promoting nuclear R (2002) Empirical statistical model to esti-
export of E2F4/5. J Cell Biol 162(2):173– mate the accuracy of peptide identifications
183. doi:10.1083/jcb.200302085 made by MS/MS and database search. Anal
14. Yang H, Zhao R, Yang HY, Lee MH (2005) Chem 74(20):5383–5392
Constitutively active FOXO4 inhibits Akt 25. Nesvizhskii AI, Keller A, Kolker E, Aebersold
activity, regulates p27 Kip1 stability, and sup- R (2003) A statistical model for identifying
presses HER2-mediated tumorigenicity. proteins by tandem mass spectrometry. Anal
Oncogene 24(11):1924–1935. doi:10.1038/ Chem 75(17):4646–4658
sj.onc.1208352 26. Faca VM, Ventura AP, Fitzgibbon MP, Pereira-
15. Hu MC, Lee DF, Xia W, Golfman LS, Ou-Yang Faca SR, Pitteri SJ, Green AE, Ireton RC,
F, Yang JY, Zou Y, Bao S, Hanada N, Saso H, Zhang Q, Wang H, O’Briant KC, Drescher
Kobayashi R, Hung MC (2004) IkappaB kinase CW, Schummer M, McIntosh MW, Knudsen
promotes tumorigenesis through inhibition of BS, Hanash SM (2008) Proteomic analysis of
forkhead FOXO3a. Cell 117(2):225–237 ovarian cancer cells reveals dynamic processes
16. Alt JR, Gladden AB, Diehl JA (2002) of protein secretion and shedding of extra-
p21(Cip1) Promotes cyclin D1 nuclear accu- cellular domains. PLoS One 3(6):e2425.
mulation via direct inhibition of nuclear doi:10.1371/journal.pone.0002425
Chapter 5
Abstract
Phosphorylation is among the most important post-translational modifications of proteins and has numer-
ous regulatory functions across all domains of life. However, phosphorylation is often substoichiometric,
requiring selective and sensitive methods to enrich phosphorylated peptides from complex cellular digests.
Various methods have been devised for this purpose and we have recently described a Fe-IMAC HPLC
column chromatography setup which is capable of comprehensive, reproducible, and selective enrichment
of phosphopeptides out of complex peptide mixtures. In contrast to other formats such as StageTips or
batch incubations using TiO2 or Ti-IMAC beads, Fe-IMAC HPLC columns do not suffer from issues
regarding incomplete phosphopeptide binding or elution and enrichment efficiency scales linearly with the
amount of starting material. Here, we provide a step-by-step protocol for the entire phosphopeptide
enrichment procedure including sample preparation (lysis, digestion, desalting), Fe-IMAC column chro-
matography (column setup, operation, charging), measurement by LC-MS/MS (nHPLC gradient, MS
parameters) and data analysis (MaxQuant). To increase throughput, we have optimized several key steps
such as the gradient time of the Fe-IMAC separation (15 min per enrichment), the number of consecutive
enrichments possible between two chargings (>20) and the column recharging itself (<1 h). We show that
the application of this protocol enables the selective (>90 %) identification of more than 10,000 unique
phosphopeptides from 1 mg of HeLa digest within 2 h of measurement time (Q Exactive Plus).
Abbreviations
ACN Acetonitrile
AGC Acquisition gain control
CAA Chloroacetamide
DTT Dithiothreitol
FA Formic acid
FCS Fetal calf serum
HCD Higher energy collision induced dissociation
HCl Hydrochloride
HPLC High-performance liquid chromatography
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_5, © Springer Science+Business Media LLC 2017
47
48 Benjamin Ruprecht et al.
1 Introduction
2 Material
2.1 Preparation 1. Cell culture: RPMI 1640 medium supplemented with 10 %
of Proteome Digests fetal calve serum (FCS). Add 55 ml FCS to 500 ml RPMI
for Phosphopeptide 1640 medium. Sterile phosphate buffered saline (PBS) with-
Analysis out calcium and magnesium. 150 × 20 mm cell culture dishes.
Cell scraper. HeLa S3 cervix carcinoma cells (DMSZ,
Braunschweig, Germany).
2. Lysis buffer: Prepare a 40 mM Tris(hydroxymethyl)amino-
methane (Tris)–HCl, pH = 7.6 solution, containing 8 M urea,
protease and phosphatase inhibitors. Prepare a stock solution of
2 M Tris–HCl by dissolving 2.42 g Tris in 5 ml water, adjust the
pH to 7.6 using a 5 M HCl solution and fill up to 10 ml with
water. A 100 fold stock solution of phosphatase inhibitor cock-
tail 1, 2, and 3 is commercially available (Sigma Aldrich, Munich,
Germany). Add 4.8 g of urea to a 15 ml falcon tube, add 200 μl
Tris–HCl stock solution, add one protease inhibitor tablet com-
plete mini EDTA-free (Roche, Mannheim, Germany) and
100 μl of each phosphatase inhibitor stock solution and fill up
to 10 ml with water. Store the lysis buffer on ice.
3. Reducing agent: 1 M stock solution of dithiothreitol (DTT) in
water. Dissolve 1.54 g DTT in a falcon tube and fill up to 10 ml
50 Benjamin Ruprecht et al.
with water. Prepare 200 μl aliquots and store the reducing agent
at −20 °C.
4. Alkylating agent: 550 mM stock solution of CAA (CAA) in
water. Dissolve 514 mg CAA in a falcon tube and fill up to
10 ml with water. Prepare 200 μl aliquots and store the alkyl-
ating agent at −20 °C.
5. 40 mM Tris–HCl solution, pH 7.6: Prepare a stock solution
of 2 M Tris–HCl by dissolving 2.42 g Tris in 5 ml water.
Adjust the pH to 7.6 using a 5 M HCl solution and fill up to
10 ml with water. Take 200 μl of the 2 M stock solution and
fill up to 10 ml with water.
6. Trypsin stock solution: Prepare a stock solution of 1 μg/μl
trypsin (sequencing grade modified trypsin, Promega) in
50 mM acetic acid, store at −80 °C.
7. Sep-Pak C-18 peptide purification: 50 mg Sep-Pak cartridges
(Waters Corp., Eschborn, Germany). Solvent A: 0.07 % (v/v)
TFA in water. Dilute 70 μl of 100 % TFA in 99.93 ml water.
Solvent B: 50 % (v/v) ACN, 0.07 % (v/v) TFA in water.
Prepare 50 ml ACN and 70 μl of TFA and dilute in 49.93 ml
water. Store at 4 °C.
8. Vacuum manifold for Sep-Pak desalting.
2.4 LC-MS/MS 1. 50 mM citric acid and 1 % (v/v) FA in water. Dissolve 2.1 g of
and Data Analysis citric acid in 9.9 ml of water and add 100 μl of FA.
2. LC-MS/MS: nano-HPLC setup coupled to a high resolution
mass spectrometer. Here, we use an Eksigent NanoLC-Ultra
1D+ (Eksigent, Dublin, CA) coupled to a Q Exactive Plus
mass spectrometer (Thermo Scientific, Bremen). LC-trap col-
umn: 75 μm × 2 cm, packed with 5 μm Reprosil-Pur ODS-3
C-18 material (Dr. Maisch, Ammerbuch, Germany). Analytical
column: 75 μm × 42 cm, packed with 3 μm Reprosil-Gold
C-18 material (Dr. Maisch, Ammerbuch, Germany).
3. Nano-HPLC solvents: Loading solvent: 0.1 % (v/v) FA in
water. Solvent A: 0.1 % (v/v) FA and 5 % (v/v) DMSO [18] in
water. Solvent B: 0.1 % (v/v) FA and 5 % (v/v) DMSO in ACN.
4. Data analysis: Freely available MaxQuant [19] software pack-
age (e.g., version 1.5.2.8) with the integrated search engine
Andromeda [20]. Protein sequence database in FASTA format
(e.g., UniprotKB).
5. Spreadsheet editor or the freely available Perseus software
package.
3 Methods
3.1 Preparation A schematic overview of the experimental steps covered in this pro-
of Proteome Digests tocol is provided in Fig. 1.
for Phosphopeptide
1. Seed HeLa cells under sterile conditions in RPMI 1640
Analysis medium supplemented with 10 % FCS. Use 30 ml medium for
150 mm cell culture dishes. Grow the cells to 80 % confluency
under humidified atmosphere, 5 % CO2 at 37 °C. For lysis,
place the cell culture dishes on ice or work at 4 °C. Wash cells
two times with cold PBS. Use a pipette to aspirate residual
PBS from cell culture plates after the final washing step (see
Note 2).
52 Benjamin Ruprecht et al.
Fig. 1 Experimental workflow for comprehensive phosphopeptide enrichment depicting cell culture and lysis,
phosphopeptide enrichment using the Fe-IMAC column, desalting of enriched phosphopeptides, and LC-MS/
MS analysis
3.2 Phosphopeptide For first time use, the column can be directly charged with FeCl3
Enrichment solvent (it does not have to be stripped) (see Note 8). The column
by Fe-IMAC Column is usually operated below 1000 psi. Column stripping and charging
Chromatography should be repeated after 20 enrichments or in case the column has
not been used for more than one week (see Note 9).
1. Column stripping: Connect the IMAC column to your HPLC
system and rinse it with ultrapure water (1 ml/min, 10 ml). Inject
1 ml of IMAC stripping solvent into the sample loop and let it
run through the column (1 ml/min). After 1 min, inject another
1 ml IMAC stripping solvent. Repeat this step eight more times.
Make sure that the sample loop is flushed with water afterwards.
Rinse the column with ultrapure water (2 ml/min, 5 ml).
2. Column charging: Inject 1 ml of IMAC charging solvent into
the sample loop and let it run over the column (0.2 ml/min).
After 5 min, inject another 1 ml of IMAC charging solvent.
Repeat this step four more times. Wash the sample loop with
3 × 1 ml EDTA to scavenge remaining Fe3+ ions (from both
syringe and sample loop) and 10 × 1 ml water to get rid of
residual EDTA. Rinse the column with 50 ml FA solvent to
wash away unbound Fe3+ ions (2 ml/min).
3. Fe-IMAC enrichment: Connect the IMAC loading solvent and
IMAC elution solvent to the HPLC system. Flush the column
with 5 ml of 50 % IMAC elution solvent (3 ml/min). Re-
equilibrate the column with 20 ml IMAC loading solvent (3 ml/
min) (see Note 10). Perform a standard enrichment to ensure
proper charging (15 min gradient, see Table 1, see Note 10).
Table 1
Settings for a 15 min Fe-IMAC column enrichment, displaying the programmed
time, the flow (in ml/min), and percentages IMAC elution solvent used
3.3 Desalting Although most of the ammonia will evaporate during the vacuum
of the Fe-IMAC Eluate centrifugation/lyophilization step, residual ammonia salts might
remain. Hence, it is recommended to desalt Fe-IMAC eluates
using C-18 StageTips [21]. Pass all liquids through the tips by
centrifugation (~800 × g, room temperature; see Note 12).
1. Dissolve the dried sample in 250 μl of solvent A and keep the
sample on ice while the StageTips are prepared. Check the pH
of the dissolved peptide solution and, if required, adjust it to
pH 2 using FA.
2. Sequentially activate the tips using 250 μl of MeOH, 250 μl of
solvent B and equilibrate with 250 μl of solvent A. Empty the
eppendorf tube in between.
Phosphoproteome Enrichment by Fe-IMAC 55
Table 2
Group-specific and global parameters for data analysis using MaxQuant version 1.5.2.8
Group-specific parameters
Type Standard
Label No
Variable modifications Acetyl (Protein N-term), Oxidation (M), Phospho (STY)
Digestion mode Specific (Trypsin/P)
Max. missed cleavages 2
Main search peptide tolerance 5 parts per million (ppm)
Max. number of modifications per peptide 5
Global parameters
Database UniProtKB
Fixed modifications Carbamidomethyl
PSM FDR 0.01
Protein FDR 0.01
Site decoy fraction 0.01
Min. peptide length 7
Min. score for (un)modified peptides 0
Min. delta score for (un)modified peptides 0
MS/MS match tolerance 20 ppm
Second peptide search Enabled
Phosphoproteome Enrichment by Fe-IMAC 57
Table 3
Overview of results typically expected from a single Fe-IMAC enrichment,
measured on a 2 h LC-MS/MS gradient on a Q Exactive Plus. The Fe-IMAC
column eluate was reconstituted in 20 μl of 50 mM Citrate, 1 % FA and
5 μl were injected
Phosphopeptides (MaxQuant—evidence.txt)
Identified unique phosphopeptides 10089
Quantified unique phosphopeptides 9392
Mono phosphorylated 8392 (83 %)
Multiply phosphorylated 1697 (17 %)
Identification-based phosphopeptide selectivity 81 %
Intensity-based phosphopeptide selectivity 94 %
Phosphorylation sites (MaxQuant—Phospho(STY)sites.txt)
Identified phosphorylation sites 8973
Quantified phosphorylation sites 7451
Class I sites (Loc. prob. > 0.75) 6566
pS sites (class I) 5674 (86 %)
pT sites (class I) 727 (11 %)
pY sites (class I) 165 (3 %)
4 Notes
References
1. Lu Z, Jiang G, Blume-Jensen P et al (2001) 6. Andersson L, Porath J (1986) Isolation of
Epidermal growth factor-induced tumor cell phosphoproteins by immobilized metal (Fe3+)
invasion and metastasis initiated by dephos- affinity chromatography. Anal Biochem 154:
phorylation and downregulation of focal adhe- 250–254
sion kinase. Mol Cell Biol 21:4016–4031 7. Posewitz MC, Tempst P (1999) Immobilized
2. Ruprecht B, Lemeer S (2014) Proteomic anal- gallium(III) affinity chromatography of phos-
ysis of phosphorylation in cancer. Expert Rev phopeptides. Anal Chem 71:2883–2892
Proteomics 11:259–267 8. Zhou H, Xu S, Ye M et al (2006) Zirconium
3. Lemeer S, Heck AJ (2009) The phosphopro- phosphonate-modified porous silicon for
teomics data explosion. Curr Opin Chem Biol highly specific capture of phosphopeptides and
13:414–420 MALDI-TOF MS analysis. J Proteome Res
4. Pinkse MWH, Uitto PM, Hilhorst MJ et al 5:2431–2437
(2004) Selective Isolation at the femtomole 9. Zhou H, Low TY, Hennrich ML et al (2011)
level of phosphopeptides from proteolytic Enhancing the identification of phosphopep-
digests using 2D-nanoLC-ESI-MS/MS and tides from putative basophilic kinase substrates
titanium oxide precolumns. Anal Chem 76: using Ti (IV) based IMAC enrichment. Mol
3935–3943 Cell Proteomics 10:M110.006452
5. Kweon HK, Håkansson K (2006) Selective zir- 10. Bodenmiller B, Mueller LN, Mueller M et al
conium dioxide-based enrichment of phos- (2007) Reproducible isolation of distinct,
phorylated peptides for mass spectrometric overlapping segments of the phosphopro-
analysis. Anal Chem 78:1743–1749 teome. Nat Methods 4:231–237
60 Benjamin Ruprecht et al.
11. Tsai C-F, Hsu C-C, Hung J-N et al (2014) 17. Zhou H, Di Palma S, Preisinger C et al (2013)
Sequential phosphoproteomic enrichment Toward a comprehensive characterization of a
through complementary metal-directed immo- human cancer cell phosphoproteome.
bilized metal ion affinity chromatography. Anal J Proteome Res 12:260–271
Chem 86:685–693 18. Hahne H, Pachl F, Ruprecht B et al (2013)
12. Thingholm TE, Jensen ON, Robinson PJ et al DMSO enhances electrospray response, boost-
(2008) SIMAC (sequential elution from ing sensitivity of proteomic experiments. Nat
IMAC), a phosphoproteomics strategy for the Methods 10:989–991
rapid separation of monophosphorylated from 19. Cox J, Mann M (2008) MaxQuant enables
multiply phosphorylated peptides. Mol Cell high peptide identification rates, individual-
Proteomics 7:661–671 ized p.p.b.-range mass accuracies and
13. Ruprecht B, Koch H, Medard G et al (2015) proteome- wide protein quantification. Nat
Comprehensive and reproducible phospho- Biotechnol 26:1367–1372
peptide enrichment using iron immobilized 20. Cox J, Neuhauser N, Michalski A et al (2011)
metal ion affinity chromatography (Fe-IMAC) Andromeda: a peptide search engine inte-
columns. Mol Cell Proteomics 14:205–215 grated into the MaxQuant environment.
14. Larsen MR, Thingholm TE, Jensen ON et al J Proteome Res 10:1794–1805
(2005) Highly selective enrichment of phos- 21. Rappsilber J, Mann M, Ishihama Y (2007)
phorylated peptides from peptide mixtures Protocol for micro-purification, enrichment,
using titanium dioxide microcolumns. Mol pre-fractionation and storage of peptides for
Cell Proteomics 4:873–886 proteomics using StageTips. Nat Protoc
15. Kettenbach AN, Gerber SA (2011) Rapid and 2:1896–1906
reproducible single-stage phosphopeptide 22. Olsen JV, Blagoev B, Gnad F et al (2006)
enrichment of complex peptide mixtures: Global, in vivo, and site-specific phosphoryla-
application to general and phosphotyrosine- tion dynamics in signaling networks. Cell
specific phosphoproteomics experiments. Anal 127:635–648
Chem 83:7635–7644 23. Winter D, Seidler J, Ziv Y et al (2009) Citrate
16. Li Q, Ning Z, Tang J et al (2009) Effect of boosts the performance of phosphopeptide
peptide-to-TiO2 beads ratio on phosphopep- analysis by UPLC-ESI-MS/MS. J Proteome
tide enrichment selectivity. J Proteome Res Res 8:418–424
8:5375–5381
Chapter 6
Abstract
A true and accurate bottom-up global proteomic measurement will only be achieved when all proteins in
a sample can be digested efficiently and at least some peptides recovered on which to base an estimate of
abundance. Integral membrane proteins make up around one-third of the proteome and require special-
ized protocols if they are to be successfully solubilized for efficient digestion by the enzymes used in bot-
tom-up proteomics. The protocol described relies upon solubilization using the detergents sodium
deoxycholate and lauryl sarcosine with heating to 95 °C. A subset of peptides is purified by reverse-phase
solid-phase extraction and fractionated by strong-cation exchange prior to nano-liquid chromatography
with data-dependent tandem mass spectrometry. For quantitative proteomics experiments a protocol is
described for stable-isotope coding of peptides using dimethylation of primary amines allowing for three-
way sample multiplexing.
Key words Trypsin, Electrospray ionization, Proteome, StageTip, Dimethylation, Phase transfer
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_6, © Springer Science+Business Media LLC 2017
61
62 Joseph Capri and Julian P. Whitelegge
2 Materials
3 Methods
3.1 Cell Lysis 1. Adherent cells are washed twice directly on plate with ice-cold
and Proteolytic Digest PBS pH 7.6 (see Note 1).
2. 0.5 mL per 1 × 107 cells of cell lysis buffer is added directly to
plate, cells are scraped with a cell scraper, and lysates are tritu-
rated with P1000 pipettor.
3. Cell lysates are transferred to 1.5 mL Eppendorf lo-bind
microcentrifuge tubes, water bath sonicated at RT for 5 min,
and heated at 95 °C for 5 min.
4. Bicinchoninic acid protein assay (Pierce) is performed to
determine protein concentration.
5. Disulfide bridges are reduced with 5 mM tris(2-carboxyethyl)
phosphine (final concentration) at RT for 30 min and subse-
quently alkylated with 10 mM iodoacetamide (final concentra-
tion) at RT in the dark for 30 min.
6. Cell lysates are transferred to 15 mL Falcon tubes and diluted
1:5 (v:v) with 50 mM ammonium bicarbonate pH 8.5.
7. Proteins are digested with sequencing grade trypsin 1:100
(enzyme:protein by mass) for 4 h at 37 °C under gentle agita-
tion followed by a second aliquot of trypsin 1:100
(enzyme:protein) overnight at 37 °C under gentle agitation.
8. Samples are acidified with 0.5 % trifluoroacetic acid (final con-
centration), vortexed rapidly for 5 min, and centrifuged at
16,000 × g for 5 min at RT to pellet sodium deoxycholate.
9. Transfer supernatant to a new tube and proceed to peptide
desalting. If needing to store for up to a week, keep peptide
samples at 4 °C, otherwise freeze at −80 °C.
3.2 Peptide 1. 200 mg tC18 Sep-Pak cartridges (Waters) are wetted with
Desalting 2 mL of 100 % methanol, with solvent pulled through the car-
and Reductive tridge using a vacuum manifold. It is critical to stop the flow
Dimethylation before all solvent has passed through to prevent any air from
Full Membrane Protein Coverage 65
3.3 Strong Cation For each of the following steps, C18-SCX StageTips will be
Exchange denoted (S) and C18 StageTips will be denoted (C) if the follow-
Fractionation ing step is to be performed on that particular StageTip. Unless
otherwise noted, all solvent can be discarded properly (see Note 2).
1. Pierce Quantitative Colorimetric Peptide Assay is performed.
2. Peptide samples are reconstituted in StageTip loading buffer
at a concentration of 0.2 mg/mL.
3. Light, medium, and heavy labeled peptides are mixed 1:1:1.
4. (S,C) StageTips are wetted with 20 μL of 100 % methanol,
pushing solvent through by applying pressure with hand
syringe. It is critical to prevent air from entering frits, leaving
~1–2 μL above frits.
5. (S,C) 20 μL of StageTip elution buffer is passed through using
pressure from hand syringe.
6. (S,C) 20 μL of StageTip loading buffer is passed through
using pressure from hand syringe.
7. (C) 100 μL of 0.5 % acetic acid is deposited into C18 StageTip
and set aside for later use.
8. (S) 20 μL of SCX elution buffer 8 is passed through using
pressure from hand syringe.
9. (S) 20 μL of StageTip loading buffer is passed through using
pressure from hand syringe.
10. (S) 32 μg of differential-labeled peptides are loaded to C18-
SCX StageTip using pressure from hand syringe.
66 Joseph Capri and Julian P. Whitelegge
11. (S) 20 μL of StageTip loading buffer are passed through using
pressure from hand syringe.
12. (S) 20 μL of StageTip elution buffer is passed through using
pressure from hand syringe.
13. (S) 20 μL of 30 % acetonitrile with 0.5 % acetic acid is passed
through using pressure from hand syringe.
14. (S) 20 μL of SCX elution buffer 1 is passed through using
pressure from hand syringe and collected in 100 μL of pre-
deposited 0.5 % acetic acid above pre-conditioned C18
StageTip from step 7. This is repeated for SCX elution buffers
2–8 and collected into separate pre-conditioned C18 StageTips
from step 7.
15. (C) SCX fractions are pipetted up and down to mix with pre-
deposited 100 μL of 0.5 % acetic acid and then passed through
C18 using pressure from hand syringe.
16. (C) 20 μL of StageTip loading buffer is passed through using
pressure from hand syringe.
17. (C) 20 μL of StageTip elution buffer is passed through using
pressure from hand syringe and collected in 1.5 mL microcen-
trifuge tubes.
18. Peptide fractions are concentrated in vacuum centrifuge to
~2 μL, typically 4 min.
19. Concentrated peptide fractions are reconstituted with 10 uL
of 2 % acetonitrile with 0.1 % formic acid and transferred to
autosampler injection vials.
charge states for ions surpassing 6000 counts, target ion count
of 5000 or fill time of 50 ms, CID collision energy of 35, and
dynamic exclusion of 30 s.
5. Raw data is searched against respective species Uniprot fasta
database using MaxQuant 1.5.3.30 with standard preset
search parameters. The search parameters are as follows: 3-plex
dimethyl labeling to lysine and peptide N-terminus, trypsin
cleavage allowing up to two missed cleavages, fixed modifica-
tion of carbamidomethyl to cysteines, variable modifications
of acetylation to protein N-terminus and methionine oxida-
tion, 10 ppm and 0.5 Da mass errors for Full MS and MS/
MS, respectively, 1 % false-discovery rate on peptide and pro-
tein identifications, and peptide match between run feature
with 1.5 min time window.
4 Notes
References
1. Ryan CM, Souda P et al (2010) Post- 6. Erde J, Loo RR, Loo JA (2014) Enhanced
translational modifications of integral mem- FASP (eFASP) to increase proteome coverage
brane proteins resolved by top-down Fourier and sample recovery for quantitative proteomic
transform mass spectrometry with collisionally experiments. J Proteome Res 13:1885–1895
activated dissociation. Mol Cell Proteomics 7. Kulak NA, Pichler G et al (2014) Minimal,
9:791–803 encapsulated proteomic-sample processing
2. Whitelegge JP (2013) Integral membrane pro- applied to copy-number estimation in eukary-
teins and bilayer proteomics. Anal Chem otic cells. Nat Methods 11:319–324
85:2558–2568 8. León IR, Schwämmle V et al (2013)
3. Manza LL, Stamer SL et al (2005) Sample Quantitative assessment of in-solution diges-
preparation and digestion for proteomic tion efficiency identifies optimal protocols for
analyses using spin filters. Proteomics 5:
unbiased protein analysis. Mol Cell Proteomics
1742–1745 12:2992–3005
4. Wiśniewski JR, Zougman A, Mann M (2009) 9. Rappsilber J, Mann M, Ishihama Y (2007)
Combination of FASP and StageTip-based Protocol for micro-purification, enrichment,
fractionation allows in-depth analysis of the pre-fractionation and storage of peptides for
hippocampal membrane proteome. J Proteome proteomics using StageTips. Nat Protoc
Res 8:5674–5678 2:1896–1906
5. Masuda T, Tomita M, Ishihama Y (2008)
10. Wilson-Grady JT, Haas W, Gygi SP (2013)
Phase transfer surfactant-aided trypsin diges- Quantitative comparison of the fasted and re-fed
tion for membrane proteome analysis. mouse liver phosphoproteomes using lower pH
J Proteome Res 7:731–740 reductive dimethylation. Methods 61:277–286
Chapter 7
Abstract
The bottom-up proteomic analysis of cell line and tissue samples to a depth > 10,000 proteins still repre-
sents a considerable challenge because of the sheer number of peptides generated by proteolytic digestions
and the high dynamic range of protein expression. As a result, comprehensive protein coverage requires
multidimensional peptide separation. Recently, off-line hydrophilic strong cation exchange (hSAX) chro-
matography has proven its merits for high resolution separation of peptides due to its high degree of
orthogonality to reversed-phase liquid chromatography. Here we describe the use of hSAX for the deep
analysis of tissue proteomes. The protocol includes optimized sample preparation steps (lysis with the aid
of mechanical disruption, one-step disulfide bridge reduction and alkylation), setup and operation of hSAX
columns and gradients, desalting of hSAX fractions prior to LC-MS/MS analysis, and suggestions for the
choice of data acquisition parameters and data analysis using MaxQuant. Application of the protocol to the
fractionation of 300 μg human brain tissue digest led to the identification of more than 100,000 unique
peptide sequences representing over 10,195 proteins and 9,500 genes in 3 days of measurement time on
a Q Exactive Plus mass spectrometer.
Key words Proteomics, Deep fractionation, Chromatography, Strong anion exchange, Tissue
proteomics
Abbreviations
ACN Acetonitrile
AGC Acquisition gain control
CAA Chloroacetamide
DTT Dithiothreitol
FA Formic acid
FDR False discovery rate
HCl Hydrochloric acid
HPLC High-performance liquid chromatography
hSAX Hydrophilic strong anion exchange
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_7, © Springer Science+Business Media LLC 2017
69
70 Benjamin Ruprecht et al.
1 Introduction
Despite the fact that tremendous progress has been made recently
in mapping out the human proteome [1, 2] and breathtaking
advances at all levels of proteomic sample preparation and mass
spectrometric instrumentation have been realized [3, 4], identifi-
cation and quantification of a single proteome to a depth > 10,000
proteins is still a considerable challenge. This has initially been
accomplished by two independent research groups in 2011 using
the cell lines HeLa [5] and U2OS [6], respectively. At the time,
this effort comprised the use of multiple enzymes and the analysis
of 72 fractions in 288 h of LC-MS/MS measurement time [5].
Such in-depth proteomic analysis was subsequently extended to a
greater number of cell lines [7, 8] but also tissue proteomes [9,
10], the latter of which represents an even greater challenge
because protein expression in tissues tends to span a broader
dynamic range than cell lines and the analysis is often complicated
by the presence of blood, fat or connective tissue. High-resolution
two-dimensional peptide separation/fractionation is an efficient
means to boost proteome coverage, sequence coverage and quan-
tification performance in bottom-up proteomics experiments.
Given that the stationary phase used as the second dimension sepa-
ration in nano-LC-MS/MS setups is almost exclusively comprised
of reversed-phase (RP) material, the first peptide separation dimen-
sion should ideally be orthogonal to RP and offer high chromato-
graphic resolution. Many different techniques have been employed
for this purpose such as HILIC [11], ZIC-HILIC [12], ERLIC
[13], WAX [14], high-pH reversed-phase [15], SAX [16], or SCX
[17]. Recently Ritorto et al. have demonstrated the merits of hSAX
chromatography [18], which separates peptides primarily based on
the number of acidic residues and the stationary phase is character-
ized by ultralow hydrophobicity. This combination enables
Hydrophilic Strong Anion Exchange (hSAX) Chromatography Enables Deep… 71
2 Materials
2.1 Preparation 1.
Tissue preparation: Precellys 24 Homogenizer (Bertin
of Tissue Technologies, France), Precellys ceramic kit (1.4 mm “small”,
and Proteome Digest 0.5 ml tubes).
2. 550 mM CAA stock solution in water. Dissolve 514 mg CAA
in a falcon tube and fill up to 10 ml with water. Prepare 1 ml
aliquots and store at −20 °C.
3. 1 M TCEP-HCl stock solution in water: Dissolve 2.87 g of
TCEP-HCl in 10 ml of water. Prepare 25 μl aliquots and store
at −20 °C.
4. Tissue lysis solution (see Notes 1 and 2): 50 mM Tris–HCl,
pH = 7.6, containing 8 M urea, 10 mM TCEP-HCl, 40 mM
CAA, protease and phosphatase inhibitors. A 100 fold stock
solution of phosphatase inhibitor cocktail 1, 2, and 3 is com-
mercially available (Sigma Aldrich, Munich, Germany). Prepare
a stock solution of 2 M Tris–HCl by dissolving 2.42 g Tris in
5 ml water. Adjust the pH to 7.6 using HCl and fill up to
10 ml with water. Transfer 4.8 g of urea to a 15 ml falcon tube.
Add 250 μl Tris–HCl stock solutions, one protease inhibitor
72 Benjamin Ruprecht et al.
2.2 hSAX 1. hSAX solvent A: 5 mM Tris, pH 8.5. Fill 900 ml water in a
Chromatography graduated 1 l cylinder. Use a magnetic stirrer to dissolve
0.606 g of Tris(hydroxymethyl)aminomethane (Tris) and
adjust the pH to 8.5 with 1 M HCl. Fill up to 1 l with water.
2. hSAX solvent B: 5 mM Tris, 1 M NaCl, pH 8.5. Fill 400 ml
water in a graduated 1 l cylinder. Use a magnetic stirrer to dis-
solve 0.303 g of Tris and 29.221 of sodium chloride and adjust
the pH to 8.5 with 1 M HCl. Fill up to 500 ml with water.
3. hSAX analytical column: Dionex IonPac AS24, hydroxide-
selective anion-exchange analytical column (2 × 250 mm, Thermo
Fisher Scientific, Waltham, USA, Product No. 064153).
4. hSAX guard column: Dionex IonPac AG24, hydroxide-selective
anion-exchange guard column (2 × 50 mm, Thermo Fisher
Scientific, Waltham, USA, Product No. 064151) (see Note 3).
5. HPLC system with the following requirements: flow rates
ranging from 0.1 ml/min to 1 ml/min; 0.1–1 ml sample loop;
UV detector set to read fixed wavelengths of 214 nm and
280 nm (here we used a Dionex Ultimate 3000 system with a
flow rate of 0.25 ml/min and a 100 μl sample loop).
2.3 StageTip 1. StageTip construction: Small, round punch to cut out C-18
Desalting of hSAX disks. 200 μl plastic pipette tip, 1.5 ml reaction vessel, 5 ml
Fractions Eppendorf CombiTip.
2. Empore Octadecyl C18 47 mm Solid Phase Extraction Disks
#2215 (3 M Purification, Eagan, MN, USA).
Hydrophilic Strong Anion Exchange (hSAX) Chromatography Enables Deep… 73
3 Methods
3.1 Preparation 1. Add 250 μl of precooled tissue lysis solution to 5–20 mg of wet
of Tissue Proteome tissue and transfer into Precellys tubes containing ceramic
Digest beads. Mount Precellys tubes in the Precellys 24 bead-milling
device and perform tissue lysis and homogenization (5500 rpm,
1 × 25 s, 5 s pause).
2. Use a Bradford assay (or a similar photometric assay) to deter-
mine the protein concentration. Store lysates at −80 °C or
continue directly. Continue with a lysate volume correspond-
ing to 200–300 μg of total protein (see Notes 4 and 5).
●● Dilute sample with four volumes of 50 mM Tris–HCl,
pH 7.6 to decrease urea concentration to 1.6 M. Add tryp-
sin in a protease-to-protein ratio of 1:100 (w/w) and pre-
digest 4 h in a thermoshaker at 37 °C and 700 rpm. Add
another 1:100 (w/w) trypsin and incubate the digestion
mixture over night in a thermo-shaker at 37 °C and 700 rpm.
●● Cool samples down to room temperature and acidify the
sample to a pH of ~2 by addition of 1 % (v/v) FA (check the
pH afterwards). Centrifuge acidified peptides at 14,000 × g
to p
recipitate insoluble matter. Use 50 mg Sep-Pak columns
and place them into a vacuum manifold (see Note 6). Prime
Sep-Pak columns by adding 1 ml of solvent B. Equilibrate
column by adding 2 × 1 ml of solvent A. Transfer the acidi-
74 Benjamin Ruprecht et al.
3.2 hSAX Connect the analytical hSAX column and the hSAX guard column
Chromatography to you HPLC system. Upon first time use the column has to be
properly flushed with hSAX solvent A until the pressure is stable (see
Note 8). Setup a gradient following the specifications in Table 1.
Monitor the UV absorption at fixed wavelengths of 280 and 214 nm
and use a flow rate of 0.25 ml/min throughout the gradient.
1. Run a standard digest: A Standard ensures column integrity
and proper column equilibration (see Note 9). Inject ~100 μg
in 100 μl hSAX solvent A.
2. Run a blank: Inject 100 μl hSAX solvent A to clean and equili-
brate the column and to avoid carry-over from the previous
sample (see Note 10).
3. Sample fractionation (see Note 11): Dissolve the desalted
digest in your 1.5 ml reaction vessel in 105 μl solvent A (see
Note 12, for sonication). Centrifuge the sample at 20,000 × g
for 10 min to pellet insoluble debris which might lead to
column clogging. Inject 100 μl of the dissolved sample. Use
a 96-well plate to collect the eluting fractions in 1 min inter-
vals (0.25 ml/fraction; see Note 13) starting 2 min into the
gradient. Collect a total of 38 fractions (see Fig. 1). Freeze
the fractions at −80 °C and dry them down using a vacuum
centrifuge.
Table 1
Settings for a 50 min hSAX column gradient, including the programmed time, the solvent flow
(in ml/min) and percentage of hSAX solvents used
Retention time
[min] hSAX solvent A [%] hSAX solvent B [%] Flow [ml/min]
0 100 0 0.25
3 100 0 0.25
27 75 25 0.25
40 0 100 0.25
44 0 100 0.25
45 100 0 0.25
50 100 0 0.25
Hydrophilic Strong Anion Exchange (hSAX) Chromatography Enables Deep… 75
Fig. 1 Typical UV (216 nm) chromatogram of a 50 min hSAX separation using 300 μg of tissue digest and the
specified gradient composition (in % of hSAX solvent B). The inset below the chromatogram illustrates the
suggested fractionation and pooling scheme applied for this example
3.3 StageTip Given the high salt concentration, it is necessary to desalt the hSAX
Desalting of hSAX fractions using C-18 StageTips [24]. Pass all liquids through the
Fractions tips by centrifugation (~800 × g, room temperature, see Note 14).
1. Resuspend all dried fractions, except for fraction 6 and 38 in
250 μl of solvent A (see Note 15). Pool fractions according to
the scheme depicted in Fig. 1 by transferring the dissolved
fraction 5 into the well containing fraction 6 and the dissolved
fraction 37 into the well containing fraction 38. This results in
a total of 36 fractions.
2. Preparation of C-18 StageTips [24]: Use the small, round
punch to cut out five C18 extraction disks from Empore mate-
rial. Construct a micro-column by packing the disks into a
200 μl pipette tip. Use a sharp scalpel to cut the lid of a 1.5 ml
reaction vessel. The reaction vessel will serve as container for
the flow-through. Push the micro-column through the cut lid
into the 1.5 ml reaction tube. Prepare one tip containing five
C-18 disks (Empore Octadecyl C-18 47 mm Solid Phase
Extraction Disks #2215, 3 M Purification, Eagan, MN, USA)
for each of the 36 hSAX fractions (see Note 16).
76 Benjamin Ruprecht et al.
3.4 LC-MS/MS 1. Reconstitute the desalted hSAX fractions in 50 μl of 0.1 % FA.
and Data Analysis 2. Inject 5 μl per fraction (see Note 18) and wash peptides bound
to the trap column for 10 min using loading solvent (0.1 % FA
in water) at a flow rate of 5 μl/min. Then, transfer peptides to
the analytical column and separate them at a flow rate of
300 nl/min using the following gradient: elute peptides using
a linear gradient from 4 to 32 % solvent B for the first 100 min
followed by a 10 min wash out and re-equilibration phase
(increase to 80 % solvent B within 1 min, hold at 80 % solvent
B for 4 min, decrease to 2 % solvent B within 1 min, hold at 2 %
solvent B for 4 min).
3. During peptide elution, directly inject peptides into the mass
spectrometer via electrospray ionization in positive ionization
mode. Suggested parameters for data dependent acquisition
on a Q Exactive Plus are specified in Table 2 (see Note 19).
4. Analyze the data using a proteomics software capable of label-
free quantification. All results shown in this chapter are based
on peptide identifications by search of raw data against the
UniProtKB human database, version July 2013 (88,354
sequences) using the freely available MaxQuant version 1.5.1.0
and its built-in Andromeda search engine. Parameters applied
are specified in Table 3.
5. To obtain the number of protein groups and unique genes
open the proteinGroups.txt and exclude reverse and contami-
nant hits. Count the unique entries in the ”Protein IDs” col-
umn and the “Gene names” column. Average and report the
sequence coverage in percent for the unique proteins. Use the
peptides.txt, remove reverse and contaminant hits and subse-
quently remove the duplicates from the sequence column.
Count and report the number of unique sequences. The num-
ber of acquired PSMs can be extracted from the “MS/MS
identified” column in the summary.txt (see Table 4).
Hydrophilic Strong Anion Exchange (hSAX) Chromatography Enables Deep… 77
Table 2
Suggested parameters for the measurement of hSAX fractions on a Q
Exactive Plus
Full MS
Resolution 70,000
AGC target 3e6
Maximum IT 100 ms
Scan range 360–1300 m/z
MS2
Resolution 17,500
AGC target 1e5
Maximum IT 50 ms
TopN 20
Isolation window 1.7 m/z
Fixed first mass NCE 25
Additional settings
Underfill ratio 1.0 %
Charge exclusion unassigned, 1, 7, 8, >8
Peptide match Preferred
Exclude isotopes On
Dynamic exclusion 35.0 s
Table 3
Search parameters used for data analysis with MaxQuant version 1.5.1.0.
In case nothing is specified, default parameters were used
Group-specific parameters
Type Standard
Label No
Variable modifications Acetyl (Protein N-term),
Oxidation (M)
Digestion mode Specific (Trypsin/P)
Max. missed cleavages 2
Main search peptide tolerance 4.5 ppm
Max. number of modifications per peptide 5
Global parameters
Database UniProtKB
Fixed modifications Carbamidomethyl
PSM FDR 0.01
Protein FDR 0.05
Site decoy fraction 0.01
Min. peptide length 7
Min. score for unmodified peptides 0
Min. score for modified peptides 40
Min. delta score for unmodified peptides 0
Min. delta score for modified peptides 6
MS/MS match tolerance 20 ppm
Second peptide search Enabled
Table 4
Overview of results typically expected from a 36 fraction hSAX separation
of human brain tissue digests, where each fraction was measured using a
2 h LC-MS/MS gradient on a Q Exactive Plus
PSMs, proteins and peptides (summary.txt, proteinGroups.txt and peptides.txt)
PSMs 473,882
Unique peptides 111,840
Proteins 10,195
Genes 9,516
Average sequence coverage [%] 30.6
Hydrophilic Strong Anion Exchange (hSAX) Chromatography Enables Deep… 79
Fig. 2 Two-dimensional peptide separation characteristics. (a) Unique peptide sequences per hSAX fraction
across the 110 min LC-MS gradient (10 min LC-MS retention time bins). The size of the dots scales with the
number of identified peptides. This clearly shows that hSAX separation is highly orthogonal to RP chromatog-
raphy. (b) Separation efficiency of the hSAX fractionation shown as the percentage of peptides found in one or
more fractions (the numbers above the bars indicate percentages). (c) Number of peptide sequences identified
per hSAX fraction
4 Notes
13. The fraction volume and number is roughly adjusted to the sepa-
ration power of the hSAX column. In our experience collecting
narrower fractionations does not considerably improve protein
identification. However, depending on the sample complexity,
the available MS machine time and the MS performance, it is also
possible to pool the 38 collected fractions into 24 or 12 fractions
prior to measurement. Although the protein identification is not
severely decreased, the achieved sequence coverage and the num-
ber of identified peptides might be considerably lower.
14. Make sure that each tip equilibration step, the washing step,
each sample loading step (sample and flow-through application)
and each elution step takes approximately 5 min. Since parame-
ters such as C18 material packing density might vary in between
experiments, we suggest to separately adjust the centrifugation
speed to fit the specified time scale for each experiment.
15. Using volumes of 250 μl prevents columns from running dry
even upon prolonged centrifugation. This is especially benefi-
cial for parallelized desalting because not all columns run at the
same speed.
16. Although only 36 StageTips are necessary for desalting of
hSAX fractions, we recommend preparing and equilibrating 40
StageTips. This way you can save yourself the trouble of start-
ing from the beginning should one or more tips be of insuffi-
cient quality.
17. To ensure that the C18 material is not running dry, make sure
that there is no air trapped between the applied liquid and the
packed C18 material.
18. Although we inject 5 μl per fraction (which is a good starting
point) we advise to initially test different injection volumes in
order to choose the right amount of loading. This will depend
on the hSAX input amount, the capacity of the LC-MS trap
and analytical columns and ultimately also on the sensitivity of
your mass spectrometer.
19. For label-free experiments, a Top N method should be chosen
such that an adequate sampling of the chromatographic peak is
assured (~10 MS1 scans per peak). Given the high input
amount and dynamic range of full proteome digests, some
peptides are highly abundant. Thus the cycle time of the instru-
ment has to be adjusted to the chromatographic peak width
which should be determined beforehand. Likewise, the MS
dynamic exclusion time should be adjusted to the chromato-
graphic peak width. Since every LC-MS system has somewhat
different separation and dead volume characteristics, we sug-
gest adjusting this value to the median peak width at base.
82 Benjamin Ruprecht et al.
References
1. Wilhelm M, Schlegl J, Hahne H et al (2014) 14. Hennrich ML, Groenewold V, Kops GJPL et al
Mass-spectrometry-based draft of the human (2011) Improving depth in phosphopro-
proteome. Nature 509:582–587 teomics by using a strong cation exchange-
2. Kim M-S, Pinto SM, Getnet D et al (2014) A weak anion exchange-reversed phase
draft map of the human proteome. Nature multidimensional separation approach. Anal
509:575–581 Chem 83:7137–7143
3. Richards AL, Merrill AE, Coon JJ (2015) 15. Gilar M, Olivova P, Daly AE et al (2005) Two-
Proteome sequencing goes deep. Curr Opin dimensional separation of peptides using
Chem Biol 24:11–17 RP-RP-HPLC system with different pH in first
4. Mann M, Kulak NA, Nagaraj N et al (2013) and second separation dimensions. J Sep Sci
The coming age of complete accurate, and 28:1694–1703
ubiquitous proteomes. Mol Cell 49:583–590 16. Zhou F, Sikorski TW, Ficarro SB et al (2011)
5. Nagaraj N, Wisniewski JR, Geiger T et al Online nanoflow reversed phase-strong anion
(2011) Deep proteome and transcriptome exchange-reversed phase liquid
mapping of a human cancer cell line. Mol Syst chromatography- tandem mass spectrometry
Biol 7:548 platform for efficient and in-depth proteome
sequence analysis of complex organisms. Anal
6. Beck M, Schmidt A, Malmstroem J et al (2011) Chem 83:6996–7005
The quantitative proteome of a human cell
line. Mol Syst Biol 7:549 17. Wolters DA, Washburn MP, Yates JR (2001)
An automated multidimensional protein iden-
7. Geiger T, Wehner A, Schaab C et al (2012) tification technology for shotgun proteomics.
Comparative proteomic analysis of eleven com- Anal Chem 73:5683–5690
mon cell lines reveals ubiquitous but varying
expression of most proteins. Mol Cell 18. Ritorto MS, Cook K, Tyagi K et al (2013)
Proteomics 11:M111.014050 Hydrophilic strong anion exchange (hSAX)
chromatography for highly orthogonal peptide
8. Azimifar SB, Nagaraj N, Cox J et al (2014) separation of complex proteomes. J Proteome
Cell-type-resolved quantitative proteomics of Res 12:2449–2457
murine liver. Cell Metab 20:1076–1087
19. Ruprecht B, Koch H, Medard G et al (2015)
9. Deshmukh AS, Murgia M, Nagaraj N et al Comprehensive and reproducible phosphopep-
(2015) Deep proteomics of mouse skeletal tide enrichment using iron immobilized metal
muscle enables quantitation of protein iso- ion affinity chromatography (Fe-IMAC) col-
forms, metabolic pathways, and transcription umns. Mol Cell Proteomics 14:205–215
factors. Mol Cell Proteomics 14:841–853
20. Cox J, Mann M (2008) MaxQuant enables
10. Wiśniewski R, Dus-Szachniewicz K, high peptide identification rates, individualized
Ostasiewicz P et al (2015) Absolute proteome p.p.b.-range mass accuracies and proteome-
analysis of colorectal mucosa, adenoma and wide protein quantification. Nat Biotechnol
cancer reveals drastic changes in fatty acid 26:1367–1372
metabolism and plasma membrane transport-
ers. J Proteome Res 14:4005–4018 21. Kulak NA, Pichler G, Paron I et al (2014)
Minimal, encapsulated proteomic-sample pro-
11. Alpert AJ (1990) Hydrophilic-interaction cessing applied to copy-number estimation in
chromatography for the separation of peptides, eukaryotic cells. Nat Methods 11:319–324
nucleic acids and other polar compounds.
J Chromatogr 499:177–196 22. Hahne H, Pachl F, Ruprecht B et al (2013)
DMSO enhances electrospray response, boost-
12. Boersema PJ, Divecha N, Heck AJR et al ing sensitivity of proteomic experiments. Nat
(2007) Evaluation and optimization of ZIC- Methods 10:989–991
HILIC-RP as an alternative MudPIT strategy.
J Proteome Res 6:937–946 23. Cox J, Neuhauser N, Michalski A et al (2011)
Andromeda: a peptide search engine integrated
13. Hao P, Guo T, Li X et al (2010) Novel applica- into the MaxQuant environment. J Proteome
tion of electrostatic repulsion-hydrophilic Res 10:1794–1805
interaction chromatography (ERLIC) in shot-
gun proteomics: comprehensive profiling of rat 24. Rappsilber J, Mann M, Ishihama Y (2007)
kidney proteome. J Proteome Res Protocol for micro-purification, enrichment,
9:3520–3526 pre-fractionation and storage of peptides for
proteomics using StageTips. Nat Protoc
2:1896–1906
Chapter 8
Abstract
Despite recent advances in mass spectrometric sequencing speed and improved sensitivity, the in-depth
analysis of proteomes still widely relies on off-line peptide separation and fractionation to deal with the
enormous molecular complexity of shotgun digested proteomes. While a multitude of methods has been
established for off-line peptide separation using HPLC columns, their use can be limited particularly when
sample quantities are scarce. In this protocol, we describe an approach which combines high pH reversed-
phase peptide separation into few fractions in StageTip micro-columns. This miniaturized sample prepara-
tion method enhances peptide recovery and hence improves sensitivity. This is particularly useful when
working with limited sample amounts obtained from e.g., phosphopeptide enrichments or tissue biopsies.
Essentially the same approach can also be applied for multiplexed analysis using tandem mass tags (TMT)
and can be parallelized in order to deliver the required throughput. Here, we provide a step-by-step pro-
tocol for TMT6plex labeling of peptides, the construction of StageTips, sample fractionation and pooling
schemes adjusted to different types of analytes, mass spectrometric sample measurement, and downstream
data processing using MaxQuant. To illustrate the expected results using this protocol, we provide results
from an unlabeled and a TMT6plex labeled phosphopeptide sample leading to the identification of
>17,000 phosphopeptides in 8 h (Q Exactive HF) and >23,000 TMT6plex labeled phosphopeptides (Q
Exactive Plus) in 12 h of measurement time. Importantly, this protocol is equally applicable to the frac-
tionation of full proteome digests.
Abbreviations
CAN Acetonitrile
AGC Acquisition gain control
FA Formic acid
HCD Higher energy collision induced dissociation
HPLC High-performance liquid chromatography
hSAX Hydrophilic strong anion exchange
IMAC Immobilized metal ion affinity chromatography
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_8, © Springer Science+Business Media LLC 2017
83
84 Benjamin Ruprecht et al.
IT Injection time
MS Mass spectrometer
MS/MS Tandem mass spectrometry
NH4FA Ammonium formate
RP Reversed-phase
SAX Strong anion exchange
SCX Strong cation exchange
StageTip Stop and go extraction tip
TEAB Triethylammonium bicarbonate
TFA Trifluoroacetic acid
TiO2 Titanium dioxide
TMT Tandem mass tag
Ppm Parts per million
PSM Peptide spectrum match
pY/pS/pT Phosphotyrosine, -serine, -threonine
ZIC-HILIC Zwitterionic hydrophilic interaction liquid chromatography
1 Introduction
2 Materials
2.3 High pH 1. Empore Octadecyl C18 47 mm Solid Phase Extraction Disks
Reversed-Phase #2215 (3 M Purification, Eagan, MN, USA).
Micro-Column 2. Small, round punch to cut out C-18 disks (Fig. 1b).
Fractionation
3. 200 μl plastic pipette tip.
4. 1.5 ml reaction vessel.
5. High pH reversed-phase stock solution: 50 mM ammonium
formate (NH4FA, Sigma-Aldrich, St. Louis, MO, Product
number 156264, reagent grade 97 %), pH 10. Weigh in
315.3 mg of NH4FA and transfer it into a glass beaker or cyl-
inder. Add water to a volume of 90 ml. Mix with a magnetic
stirrer and adjust the pH to 10 using ammonium hydroxide.
Fill up to 100 ml with water.
Fig. 1 Construction of a micro-column for high pH reversed-phase fractionation. (a) The lid of a 1.5 ml reaction
tube is cut crosswise with a small scalpel. (b) A punching device, assembled from a syringe and a piece of wire
is used to punch out 5 disks of Empore C18 material. (c) The punching device is used to push the disks into
the 200 μl pipette tip. (d) The constructed column is placed in the reaction tube by pushing the tip through the
cut lid
High ph Micro-Columns for Simple Efficient Proteome Fractionation 87
2.4 LC-MS/MS 1. 50 mM citric acid and 1 % (v/v) FA in water. Dissolve 105 mg
and Data Analysis of citric acid monohydrate (VWR, Product No. 20278.298) in
9.90 ml of water and add 100 μl of 100 % FA.
2. LC-MS/MS: nano-HPLC setup coupled to a high resolution
mass spectrometer. Here, we used an Eksigent NanoLC-Ultra
1D+ (Eksigent, Dublin, USA; to measure fractions of
TMT6plex labeled phosphopeptides) or an Thermo Ultimate
Table 1
Mixing scheme for high pH reversed-phase elution solvents
3 Methods
3.1 Sample For methods related to lysis, digestion, and peptide desalting using
Preparation Sep-Pak columns, we refer to Chapter 5. The Fe-IMAC protocol
also describes the purification of phosphopeptides using Fe-IMAC
columns. The procedure can be readily applied to combined,
TMT6plex labeled digests. We recommend using at least 1.5 mg of
protein digest for phosphopeptide enrichment (Fe-IMAC input
amounts ranging between 1.5 and 3 mg are required to obtain
optimal results). For a TMT6plex experiment this translates into
250 μg of protein per channel (see Note 6).
3.2 TMT6plex 1. Desalting of samples prior to TMT6plex labeling (see Note 7):
Labeling of Whole Acidify digested samples to ~ pH 2 by addition of 100 % FA to
Proteome Digests Prior a final concentration of 1 % (v/v) FA (check the pH). Centrifuge
to Phosphopeptide peptides at 5000 × g to precipitate insoluble matter. Place six
Enrichment 50 mg Sep-Pak columns into a vacuum manifold (see Note 8).
Prime Sep-Pak columns by adding 1 ml of solvent B. Equilibrate
column by adding 2 × 1 ml of solvent A. Load the supernatant
of the acidified sample slowly onto the column, reapply the
flow-through and discard the second flow-through afterwards.
Wash the column with 2 × 1 ml of solvent A. Elute peptides of
each column with 2 × 150 μl of solvent B into a 1.5 ml reaction
vessel. Freeze the samples at −80 °C and make sure that they
are still frozen when they are placed in the vacuum centrifuge
to dry them down.
2. Labeling reaction (see Note 9): Reconstitute the six desalted sam-
ples in 200 μl of 50 mM TEAB. To start the labeling reaction, add
High ph Micro-Columns for Simple Efficient Proteome Fractionation 89
3.3 High pH The following steps for fractionation and desalting are performed
Reversed-Phase by centrifugation of the micro-column at ~800 × g (see Note 12).
Micro-Column Pre-cool and store all solvents on ice. Unless stated otherwise,
Fractionation avoid letting the columns run dry.
1. High pH reversed-phase micro-column construction: Use the
small, round punch to cut out five C18 extraction disks from
Empore material (see Note 13 and Fig. 1). Construct a micro-
column by packing the disks into a 200 μl pipette tip. Use a
sharp scalpel to cut the lid of a 1.5 ml reaction vessel. The
reaction vessel will serve as container for the flow-through.
Push the micro-column through the cut lid into the 1.5 ml
reaction tube.
2. Dissolve your dried sample in 250 μl solvent A (see Note 14).
Vortex, spin down and store on ice while the column is equili-
brated. Check if the pH of the dissolved sample is ~10 using
pH indicator strips.
90 Benjamin Ruprecht et al.
Fig. 2 Total ion current chromatograms displaying peptide elution patterns across the different high pH
reversed-phase fractions for unlabeled phosphopeptides (blue, Fig. 2a) and TMT6plex labeled phosphopep-
tides (red, Fig. 2b). The number in the corner of each plot indicates the absolute ion current intensity and the
respective fraction number
92 Benjamin Ruprecht et al.
4 Notes
Table 2
Suggested parameters for the measurement of high pH reversed-phase phosphopeptide fractions on
a Q Exactive HF (unlabeled phosphopeptides) and a Q Exactive Plus (TMT6plex labeled
phosphopeptides) mass spectrometer
Table 3
Search parameters used for data analysis with MaxQuant version 1.5.2.8. In case default parameters
were used, nothing is specified
Fig. 3 Orthogonality and peptide separation characteristics of high pH reversed-phase micro-column fraction-
ation. (a) Unique phosphopeptide sequences per micro-column fraction across the 110 min LC-MS gradient
(blue: unlabeled phosphopeptides; red: TMT6plex labeled phosphopeptides; grouped into 10 min LC-MS reten-
tion time bins). The size of the dots scales with the number of identified phosphopeptides. (b) Separation
efficiency of the high pH reversed-phase micro-column fractionation shown as the percentage of peptides
found in one or more fractions (blue: unlabeled phosphopeptides; red: TMT6plex labeled phosphopeptides; the
numbers above the bars indicate percentages)
Table 4
Summary of expected results for fractionation of TMT6plex and unlabeled phosphopeptides prepared
from 1.5 mg and 2 mg of digest, respectively
References
1. Wilhelm M, Schlegl J, Hahne H et al (2014) 3. Gauci S, Helbig AO, Slijper M et al (2009)
Mass-spectrometry-based draft of the human Lys-N and trypsin cover complementary parts
proteome. Nature 509:582–587 of the phosphoproteome in a refined SCX-
2. Villén J, Gygi SP (2008) The SCX/IMAC based approach. Anal Chem 81:4493–4501
enrichment approach for global phosphoryla- 4. McNulty DE, Annan RS (2008) Hydrophilic
tion analysis by mass spectrometry. Nat Protoc interaction chromatography reduces the com-
3:1630–1638 plexity of the phosphoproteome and improves
98 Benjamin Ruprecht et al.
global phosphopeptide isolation and detection. 12. Wiśniewski JR, Zougman A, Mann M (2009)
Mol Cell Proteomics 7:971–980 Combination of FASP and StageTip-based
5. Ruprecht B, Koch H, Medard G et al (2015) fractionation allows in-depth analysis of the
Comprehensive and reproducible phosphopep- hippocampal membrane proteome. J Proteome
tide enrichment using iron immobilized metal Res 8:5674–5678
ion affinity chromatography (Fe-IMAC) col- 13. Thompson A, Schäfer J, Kuhn K et al (2003)
umns. Mol Cell Proteomics 14:205–215 Tandem mass tags: a novel quantification strat-
6. Batth TS, Francavilla C, Olsen JV (2014) Off- egy for comparative analysis of complex protein
line high-pH reversed-phase fractionation for mixtures by MS/MS. Anal Chem
in-depth phosphoproteomics. J Proteome Res 75:1895–1904
13:6176–6186 14. Cox J, Mann M (2008) MaxQuant enables
7. Kettenbach AN, Gerber SA (2011) Rapid and high peptide identification rates, individualized
reproducible single-stage phosphopeptide p.p.b.-range mass accuracies and proteome-
enrichment of complex peptide mixtures: wide protein quantification. Nat Biotechnol
application to general and phosphotyrosine- 26:1367–1372
specific phosphoproteomics experiments. Anal 15. Cox J, Neuhauser N, Michalski A et al (2011)
Chem 83:7635–7644 Andromeda: a peptide search engine integrated
8. Ishihama Y, Rappsilber J, Mann M (2006) into the MaxQuant environment. J Proteome
Modular stop and go extraction tips with Res 10:1794–1805
stacked disks for parallel and multidimensional 16. Olsen JV, Blagoev B, Gnad F et al (2006)
peptide fractionation in proteomics. Global, in vivo, and site-specific phosphoryla-
J Proteome Res 5:988–994 tion dynamics in signaling networks. Cell
9. Rappsilber J, Mann M, Ishihama Y (2007) Protocol 127:635–648
for micro-purification, enrichment, pre-fraction- 17. Winter D, Seidler J, Ziv Y et al (2009) Citrate
ation and storage of peptides for proteomics using boosts the performance of phosphopeptide
StageTips. Nat Protoc 2:1896–1906 analysis by UPLC-ESI-MS/MS. J Proteome
10. Lawrence RT, Perez EM, Hernández D et al Res 8:418–424
(2015) The proteomic landscape of triple- 18. Ow SY, Salim M, Noirel J et al (2009) iTRAQ
negative breast cancer. Cell Rep 11:630–644 underestimation in simple and complex mix-
11. Kitata RB, Dimayacyac-Esleta BRT, Choong tures: “the good, the bad and the ugly”.
W-K et al (2015) Mining missing membrane pro- J Proteome Res 8:5347–5355
teins by high-pH reverse-phase StageTip frac- 19. Werner T, Sweetman G, Savitski MF et al (2014)
tionation and multiple reaction monitoring mass Ion coalescence of neutron encoded TMT 10-plex
spectrometry. J Proteome Res 14:3658–3669 reporter ions. Anal Chem 86:3594–3601
Chapter 9
Abstract
Protein glycosylation is considered to be one of the most abundant post-translational modifications and is rec-
ognized for playing key roles in cellular functions. Aberrant N-linked glycosylation has been associated with
several human diseases and has prompted the development and constant improvement of analytical tools to
separate, characterize, and quantify glycoproteins in complex mixtures extracted from various biological
samples (such as blood and tissue). Lectins, or carbohydrate-binding proteins, have been used as valuable
tools for enriching for glycoproteins and selecting for specific types of glycosylation. Herein a method using
multidimensional intact protein fractionation and LC-MS/MS analysis is described. Immunodepletion is
used to remove highly abundant proteins from human plasma, followed by glycoform separation using
multi-lectin affinity chromatography, in which specific lectins are chosen to capture and elute specific types
of glycosylation. Reversed-phase chromatography prior to digestion is used for further fractionation, allow-
ing for an increased number of protein identifications of moderate- to low-abundant proteins detectable in
plasma . This method also incorporates isotopic labeling during alkylation for relative quantitation between
two samples (such as a case and control). A bottom-up, tandem mass spectrometry-based proteomics
approach is used for protein identification and quantitation, and allows for screening glycoform-specific
changes across hundreds of plasma proteins.
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_9, © Springer Science+Business Media LLC 2017
99
100 Sarah M. Totten et al.
200 µL 200 µL
Sample 1 (Case) Sample 2 (Control)
Immunodepletion Immunodepletion
Take Flow-through Take Flow-through
Reduction/Alkylation Reduction/Alkylation
Light Acrylamide Labeling Heavy Acrylamide Labeling
Multi-Lectin Affinity
Chromatography
Unbound AAL PHA-L/E
Reversed-Phase
Fractionation (x3)
Tryptic Digestion
LC-MS/MS of RP Fractions
Protein Mixture
AAL
PHA-L/E
2390 A. Immunodepleon
1990 Flow-through
1590
1190
790
390
-10
-1 5 11 17 23 29 35 41 47 53 59 65
150
B. M-LAC PHA
130
110 Unbound AAL
Absorbance (mAU)
90
70
50
30
10
-10 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
35
30 C. Reversed-Phase
25
20
15
10
5
0
-5
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Retenon Time (minutes)
Fig. 3 A series of chromatographic separations on intact plasma proteins. In our
multidimensional separation approach, immunodepletion (a) is used to deplete
plasma of highly abundant proteins. The flow-through from that separation con-
tains moderate- to low-abundant plasma proteins, which are then separated by
glycoforms using M-LAC (b). Boxes are inset to show suggested fraction collection.
Three separate fractions are collected from M-LAC and are further fractionated at
the intact protein level by reversed-phase chromatography (c). Inset vertical lines
represent the fractions collected in 1 min intervals
104 Sarah M. Totten et al.
2 Materials
2.2 Reduction 1. Bradford protein assay kit (including protein standards and
and Alkylation Coomassie dye) and spectrophotometer for measuring protein
of Depleted Plasma concentration.
2. Protein denaturation buffer: 8 M urea, 50 mM Tris–HCl,
0.05 % octyl β-d-glucopyranoside, pH 7.5, prepared in
100 mM ammonium bicarbonate.
3. Dithiothreitol (DTT) for protein reduction.
4. Acrylamide for protein alkylation. For isotopic labeling, both
light (12C) and heavy (13C) acrylamide is required. 1,2,3-13C3
(heavy) acrylamide can be purchased from Cambridge Isotope
Laboratories. If not labeling, alternative alkylating agents can
be used, such as iodoacetamide or iodoacetic acid (not covered
in this protocol, see Note 2).
5. 3000 NMWL centrifugal filters, 4 mL (Amicon® Ultra—4,
Millipore) for buffer exchange.
6. Phosphate-buffered saline, 1× solution
2.6 LC-MS/MS 1. Conical mass spectrometry vials (250 μL, with snap septum
Analysis of Tryptic lids, or other autosampler vials).
Peptides 2. Buffer A: 0.1 % formic acid (FA) in water.
and Glycopeptides
3. Buffer B: 0.1 % FA in acetonitrile.
4. C18 analytical column (Picofrit 75 μm ID, New Objective,
packed in-house with MagicC18 AQ 5 μm, 100 A resin, packed
to 25 cm, or similar).
5. C18 trap column (Thermo Scientific Acclaim PepMap 100,
5 μm particle size, 5 mm length, 300 μm ID).
3 Procedures
3.2 Reduction The amount of depleted protein from plasma is determined using
and Alkylation a standard Bradford protein assay. Typically between 0.75 and
of Depleted Plasma 1.0 mg of protein is recovered from the depletion of 200 μL of
human plasma. Before separation by multi-lectin affinity chroma-
tography, proteins must be reduced and alkylated.
MLAC for Protein Glycoforms 107
3.4 Reversed-Phase Each of the three serial-eluted lectin fractions collected from M-LAC
Chromatography are then further separated by reversed-phase fractionation at the
of Intact Proteins intact protein level. By separating at the protein level, the mixture of
from M-LAC Fractions proteins is further de-complexed, allowing for deeper analysis and
more protein identifications. Furthermore protein glycoforms are
confined to more distinct reversed-phase fractions prior to digestion,
allowing separation of protein isoforms. Reversed-phase fraction-
ation is performed on a 100 mm × 2.1 mm ID stainless steel column
pre-packed with POROS®R2 (Applied Biosystems), 2000 Å particle
size poly styrene-divinylbenzene immobile phase. Maximum pres-
sure should not exceed 170 bar.
1. Load the concentrated sample onto a 500 μL sample loop at
0.8 mL/min for 5 min with reversed-phase Solvent A (0.1 %
trifluoroacetic acid in water).
2. Separate the proteins using a gradient of increasing organic
content in the mobile phase. Between 5 and 38 min, bring
reversed-phase Buffer B (0.1 % trifluoroacetic acid in aceto-
nitrile) to 90 %. Hold the gradient at 90 % Buffer B for 2 min,
then re-equilibrate the column with 95 % Buffer A for
10 min.
3. Collect reversed-phase fractions in an automatic analytical scale
fraction collector (Agilent 1260) into 1 mL racked microtubes.
MLAC for Protein Glycoforms 109
In total 24, 0.8 mL fractions are collected (one per minute from
8 to 31 min, as shown in Fig. 3c). Fractions can be collected
manually if automated fraction collection is unavailable.
4. Repeat steps 1–3 for remaining M-LAC fractions. Overall,
three RP separations are performed (one per M-LAC frac-
tion—unbound, AAL, and PHA-L/E). Per sample, there are 3
sets of 24 RP fractions.
5. Freeze RP fractions at −80 °C and lyophilize.
3.6 LC-MS/MS RP fractions are then analyzed by nano LC-MS/MS. In total, each
Analysis of Tryptic sample will have 72 RP fractions in total (24 unbound, 24 AAL,
Peptides and 24 PHA-L/E—or less if some were combined, which is rec-
and Glycopeptides ommended to improve signal in the mass spectrometer).
1. Transfer the 50 μL of each RP fraction to a mass spectrometry
vial.
2. Load 15 μL (approximately between 5 and 10 μg of peptides)
onto a C18 trap column at 5 μL/min, briefly de-salt and con-
centrate, then subsequently separate on a 25 cm C18 analytical
column (Picofrit 75 μm ID, New Objective, packed in-house
with MagicC18 AQ resin). Tryptic peptides are separated using
a multistep gradient at a flow rate of 0.6 μL/min in which Buffer
B (0.1 % FA in acetonitrile) is increased from 0 % (100 % Buffer
A, 0.1 % FA in water) to 85 % over 120 min. Re-equilibrate the
analytical column for 20 min at 98 % Buffer A.
3. Ionize nano-LC eluent by electrospray ionization at 2.25 kV
with the capillary temperature set to 200 °C.
4. In each MS/MS experiment, perform an initial MS1 scan over
an m/z range of 400–1800, followed by 10 data-dependent
collision-induced dissociation fragmentation events on the 10
most intense +2 or +3 ions from the MS1 spectrum over an
acquisition time of 140 min.
5. Acquired data can be processed by the Computational Proteomics
Analysis System (CPAS) [35] pipeline using X! Tandem search
algorithm [36, 37] for peptide identifications and the Q3 quan-
titation algorithm to calculate heavy-to-light ratios for cysteine-
containing acrylamide labeled peptides [38]. PeptideProphet
[39, 40] and ProteinProphet can be used to validate peptide and
protein identifications respectively.
110 Sarah M. Totten et al.
4 Notes
References
1. Wang H, Hanash S (2011) Intact-protein 10. Zaia J (2010) Mass spectrometry and glycomics.
analysis system for discovery of serum-based OMICS 14(4):401–418
disease biomarkers. Methods Mol Biol 11. Mechref Y, Hu Y, Desantos-Garcia JL, Hussein
728:69–85 A, Tang H (2013) Quantitative glycomics strat-
2. Apweiler R, Hermjakob H, Sharon N (1999) egies. Mol Cell Proteomics 12(4):874–884
On the frequency of protein glycosylation, as 12. An HJ, Kronewitter SR, de Leoz MLA, Lebrilla
deduced from analysis of the SWISS-PROT CB (2009) Glycomics and disease markers.
database. Biochim Biophys Acta 1473(1):4–8 Curr Opin Chem Biol 13(5-6):601–607
3. Varki A, Cummings RD, Esko JD, et al., edi- 13. Fujitani N, J-i F, Araki K, Fujioka T, Takegawa
tors. Cold Spring Harbor (NY): Cold Spring Y, Piao J, Nishioka T, Tamura T, Nikaido T,
Harbor Laboratory Press; 2009. Bookshelf ID: Ito M, Nakamura Y, Shinohara Y (2013) Total
NBK1963 PMID: 20301279 cellular glycomics allows characterizing cells
4. Mkhikian H, Grigorian A, Li CF, Chen H-L, and streamlining the discovery process for cel-
Newton B, Zhou RW, Beeton C, Torossian S, lular biomarkers. Proc Natl Acad Sci U S A
Tatarian GG, Lee S-U, Lau K, Walker E, 110(6):2105–2110
Siminovitch KA, Chandy KG, Yu Z, Dennis JW, 14. Hu Y, Zhou S, Yu C-Y, Tang H, Mechref Y
Demetriou M (2011) Genetics and the environ- (2015) Automated annotation and quantita-
ment converge to dysregulate N-glycosylation tion of Glycan by LC-ESI-MS analysis using
in multiple sclerosis. Nat Commun 2:334 MultiGlycan-ESI computational tool. Rapid
5. Palmigiano A, Barone R, Sturiale L, Sanfilippo Commun Mass Spectrom 29(1):135–142
C, Bua RO, Romeo DA, Messina A, Capuana 15. Ruhaak LR, Miyamoto S, Lebrilla CB (2013)
ML, Maci T, Le Pira F, Zappia M, Garozzo D Developments in the identification of glycan
(2016) CSF N-glycoproteomics for early diag- biomarkers for the detection of cancer. Mol
nosis in Alzheimer’s disease. J Proteomics Cell Proteomics 12(4):846–855
131:29–37 16. Zhou S, Hu Y, DeSantos-Garcia JL, Mechref Y
6. Theodoratou E, Campbell H, Ventham NT, (2015) Quantitation of permethylated
Kolarich D, Pucic-Bakovic M, Zoldos V, N-glycans through multiple-reaction monitor-
Fernandes D, Pemberton IK, Rudan I, ing (MRM) LC-MS/MS. J Am Soc Mass
Kennedy NA, Wuhrer M, Nimmo E, Annese V, Spectrom 26(4):596–603
McGovern DPB, Satsangi J, Lauc G (2014) 17. Wu S-W, Pu T-H, Viner R, Khoo K-H (2014)
The role of glycosylation in IBD. Nat Rev Novel LC-MS2 product dependent parallel data
Gastroenterol Hepatol 10:588–600, advance acquisition function and data analysis workflow
online publication for sequencing and identification of intact glyco-
7. Liang H-C, Russell C, Mitra V, Chung R, Hye A, peptides. Anal Chem 86(11):5478–5486
Bazenet C, Lovestone S, Pike I, Ward M (2015) 18. Saba J, Dutta S, Hemenway E, Viner R (2012)
Glycosylation of human plasma clusterin yields a Increasing the productivity of glycopeptides
novel candidate biomarker of Alzheimer’s dis- analysis by using higher-energy collision
ease. J Proteome Res 14(12):5063–5076 dissociation-accurate mass-product-dependent
8. Goulabchand R, Vincent T, Batteux F, J-f E, electron transfer dissociation. Int J Proteomics
Guilpain P (2014) Impact of autoantibody gly- 2012:560391
cosylation in autoimmune diseases. Autoimmun 19. Hong Q, Ruhaak LR, Stroble C, Parker E,
Rev 13(7):742–750 Huang J, Maverakis E, Lebrilla CB (2015) A
9. Lauc G, Huffman JE, Pučić M, Zgaga L, method for comprehensive glycosite-mapping
Adamczyk B, Mužinić A, Novokmet M, and direct quantitation of serum glycoproteins.
Polašek O, Gornik O, Krištić J, Keser T, Vitart J Proteome Res 14(12):5179–5192
V, Scheijen B, Uh H-W, Molokhia M, Patrick 20. Mayampurath A, Yu C-Y, Song E, Balan J,
AL, McKeigue P, Kolčić I, Lukić IK, Swann O, Mechref Y, Tang H (2014) Computational
van Leeuwen FN, Ruhaak LR, Houwing- framework for identification of intact glycopep-
Duistermaat JJ, Slagboom PE, Beekman M, de tides in complex samples. Anal Chem
Craen AJM, Deelder AM, Zeng Q, Wang W, 86(1):453–463
Hastie ND, Gyllensten U, Wilson JF, Wuhrer
M, Wright AF, Rudd PM, Hayward C, 21. Hu H, Khatri K, Zaia J (2016) Algorithms and
Aulchenko Y, Campbell H, Rudan I (2013) design strategies towards automated glycopro-
Loci associated with N-glycosylation of human teomics analysis. Mass Spectrom Rev n/a-n/a
immunoglobulin G show pleiotropy with auto- 22. Mayampurath A, Song E, Mathur A, Yu C-y,
immune diseases and haematological cancers. Hammoud Z, Mechref Y, Tang H (2014)
PLoS Genet 9(1):e1003225 Label-free glycopeptide quantification for
MLAC for Protein Glycoforms 113
Abstract
Complete understanding of cellular function requires knowledge of the composition and dynamics of
protein interaction networks, the importance of which spans all molecular cell biology fields. Mass
spectrometry-based proteomics approaches are instrumental in this process, with affinity purification coupled
to mass spectrometry (AP-MS) now widely used for defining interaction landscapes. Traditional AP-MS
methods are well suited to providing information regarding the temporal aspects of soluble protein–
protein interactions, but the requirement to maintain protein–protein interactions during cell lysis and AP
means that both weak-affinity interactions and spatial information is lost. A more recently developed
method called BioID employs the expression of bait proteins fused to a nonspecific biotin ligase, BirA*,
that induces in vivo biotinylation of proximal proteins. Coupling this method to biotin affinity enrichment
and mass spectrometry negates many of the solubility and interaction strength issues inherent in traditional
AP-MS methods, and provides unparalleled spatial context for protein interactions. Here we describe the
parallel implementation of both BioID and FLAG AP-MS allowing simultaneous exploration of both spatial
and temporal aspects of protein interaction networks.
Key words Mass spectrometry, BioID, Biotin, Streptavidin, FLAG tag, Proximity labeling, Affinity
purification, Proteomics, Protein interactions, Protein network, Protein identification
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_10, © Springer Science+Business Media LLC 2017
115
116 Geoffrey G. Hesketh et al.
A.
Gateway
entry clone
Tet repressor Flp-In T-REx
attL1 attL2
ORF
Gateway inhibits expression expression cell line
expression clone
attB1 attB2
co-transfect into HygR PCMV
BirA* ORF
/2X TetO 2
KANR/SpecR BirA* ORF Flp-In T-REx host
FLAG
LR clonase cell line
2
FLAG
tO
Hy
/2X TeV
M
g
R
PC
+ tetracycline
AmpR pOG44
Gateway (encodes Flp
pDEST-pcDNA5/FRT/TO-BirA*-FLAG-ORF recombinase)
destination vector HygR PCMV
/2X TetO 2
BirA* ORF
FLAG
Hy
tO
/2X TeV
g
M
R
PC
AmpR
pDEST-pcDNA5/FRT/TO-BirA*-FLAG
B. Expand Hygromycin
resistant cell population
Freeze 15 cm plate
BioID FLAG-AP
BirA*
PREY BAIT
PREY
anti-FLAG antibody
streptavidin biotinylated prey
biotin
2 Materials
2.1 Expression 1. Gateway entry vector for the protein(s) of interest (see
Constructs Subheading 3). Either “open” (no stop codon) or “closed”
(with stop codon) ORFs should be used depending on whether
C-terminal or N-terminal fusions, respectively, are desired.
2. Gateway destination vector (pDEST-pcDNA5-BirA*-FLAG_
Nterm for N-terminal fusions or pDEST-pcDNA5- BirA*FLAG_
Cterm for C-terminal fusions). These can be requested through
the Gingras lab. Non-Gateway versions of these vectors can be
requested through the Raught lab.
3. Gateway LR Clonase II (Invitrogen, Cat #11791-100).
4. Competent bacteria (e.g., DH5α).
5. SOC and LB media.
6. LB-ampicillin (100 μg/mL) agar plates.
7. Ampicillin stock (100 mg/mL).
8. DNA miniprep kit.
9. BsrGI restriction enzyme and standard DNA agarose gel system
(for clone validation).
2.2 Generation 1. Flp-In T-REx HEK293 cells (Invitrogen, for cell culture).
of Pooled Stable Cell 2. DMEM (high glucose, with pyruvate and l-glutamine) supple-
Lines mented with 10 % FBS, 100 U/mL penicillin/streptomycin
(Flp-In and T-REx cassettes can be selected for by the addition
of Zeocin (100 μg/mL) and Blasticidin (3 μg/mL), respec-
tively, to the media).
3. pOG44 vector (encodes Flp recombinase) (1 μg/mL), puri-
fied expression clone (100 ng/μL), transfection reagent.
4. Standard growth media supplemented with 200 μg/mL
Hygromycin B.
Fig. 1 (continued) constitutively repressed by the Tet repressor until tetracycline is supplemented. For complete
details, please see the Gateway and Flp-In™ T-Rex™ manual available on the Invitrogen website. (b)
Schematic representation of the parallel BioID and FLAG AP-MS workflow. Hygromycin resistant clones are
pooled and expanded to five 15 cm plates; one for generating frozen stock, and two biological replicates for
BioID, the two biological replicates for FLAG-AP. For BioID, pooled clones expressing BirA*-FLAG-ORF construct
and parallel control cell lines are grown to approximately 80 % confluence, and treated with tetracycline and
biotin (4–24 h) to induce bait expression and biotinylation of proximal proteins. Cells are lysed under strong
solubilizing conditions (protein–protein interactions are lost) and AP is performed with streptavidin-sepharose
beads (buffer conditions are outlined in Fig. 2). For FLAG-AP, cells are grown to approximately 80 % confluence,
and treated with tetracycline (24 h) to induce bait expression. Cells are lysed under gentle solubilizing condi-
tions (protein–protein interactions are maintained) and AP is performed with anti-FLAG magnetic beads. In
both cases, proteins are digested on-bead for subsequent mass-spectrometric analysis
120 Geoffrey G. Hesketh et al.
2.3 Biotin– Use “clean” (non-autoclaved) pipette tips and tubes throughout
Streptavidin Affinity procedure (see Note 1). Use ultrapure/HPLC grade water for all
Purification (BioID) buffers and reagents.
1. BioID Lysis Buffer Stock, for 50 mL: 50 mM Tris pH 7.5
(2.5 mL of 1 M stock), 150 mM NaCl (1.5 mL of 5 M stock),
0.4 % SDS (2 mL of 10 % stock), 1 % IGEPAL CA-630 (or other
Nonidet P-40 substitute) (5 mL of 10 % stock), 1.5 mM MgCl2
(150 μL of 0.5 M stock), 1 mM EGTA (200 μL of 0.25 M
stock), fill up to 50 mL with water.
2. BioID Lysis Buffer Complete (see Note 2): Per 1 mL Lysis
Buffer add (freshly) 2 μL of 500× stock of Protease Inhibitors
(PI) (Cat #P8340, Sigma-Aldrich), and 1 μL Benzonase
(250 U/μL) (Cat #71205, EMD-Millipore) (alternatively
Turbo Nuclease (Cat #9207, BioVision) may be used).
3. BioID Wash Buffer (for 10 mL): 2 % SDS (2 mL of 10 % stock),
50 mM Tris pH 7.5 (500 μL of 1 M stock), fill up to 10 mL
with water.
4. Biotin stock solution (for 20 mL of 20 mM which is a 400×
stock): 100 mg Biotin, add 2 mL of 30 % NH4OH and place
on ice. Slowly add 5 mL of 1 N HCl, wait 5 min, and repeat
for a total of 18 mL added. Store at 4 °C protected from light
(see Note 3).
5. Tetracycline stock solution (10 mg/mL which is a 10,000×
stock): 100 mg in 10 mL 50 % EtOH (anhydrous) in water.
Store at −20 °C protected from light (see Note 4).
6. Streptavidin-sepharose beads “high performance” (Cat #17-
5113-01, GE Healthcare).
7. Ammonium bicarbonate (ABC) (50 mM, pH 8.0): 200 mg in
water (see Note 5).
8. Trypsin from porcine pancreas (proteomics grade) (see Note 6):
for the solution to digest on the Streptavidin-sepharose beads,
resuspend one 20 μg tube (Cat #T6567, Sigma-Aldrich) in a
final volume of 2 mL ABC (concentration 10 ng/μL). For
spike-in trypsin addition, resuspend a 20 μg tube of trypsin in
100 μL ABC (concentration 0.2 μg/μL). When using trypsin
“singles” (Cat #T7575, Sigma-Aldrich) instead, resuspend
each in 10 μL ABC and dilute appropriately to achieve desired
final concentration.
9. Formic acid (mass spectrometry grade): 5 % and 10 % stocks
freshly prepared in HPLC grade water and stored in clean glass
vials (Cat #14-955-319, Fisher Scientific).
10. Autosampler vials (Cat #160134, Dionex) and caps (Cat #03-
391-43, Fisher Scientific).
2.4 FLAG Affinity 1. FLAG Lysis Buffer: 50 mM Hepes–KOH pH 8.0, 100 mM
Purification (FLAG-AP) KCl, 2 mM EDTA, 0.1 % NP40, and 10 % glycerol, supple-
mented with 1 mM PMSF, 1 mM DTT, and 1× protease
Parallel BioID and FLAG Affinity-Purification Coupled to Mass Spectrometry 121
3 Methods
3.1 Preparation A Gateway “entry” clone is defined as the open reading frame
of Expression (ORF) for the protein of interest cloned into a Gateway donor vec-
Constructs Using tor such as pDONR233 (Invitrogen). For N-terminal fusions, we
Gateway Cloning recommend using an entry clone harboring a stop codon to pre-
vent cloning scars [35]; C-terminal fusions should be generated
with “open” entry vectors. The protocol assumes that an entry
clone for Gateway cloning is available for the protein of interest
(and that it has been sequence-verified); a miniprep provides a
sufficient amount. New “Entry” clones can be generated through
PCR amplification of inserts with flanking attB sites and recombi-
nation into a donor vector with attP sites by Gateway “BP” reac-
tion. Entry clones may also be generated via a BP reaction with an
expression clone and a donor vector. Alternatively, BirA*-FLAG
vectors compatible with standard restriction enzyme/ligation cloning
can be used.
1. For the Gateway LR reaction, combine the following: (a)
Entry clone (for LR reaction) (ideally 150 ng, although less
can be used); (b) Destination vector (1 μL of a 150 ng/μL
stock); (c) Water or TE pH 8.0 up to 4 μL.
2. Thaw LR Clonase on ice and briefly mix/vortex, then add
1 μL of LR Clonase to reaction (total volume of 5 μL), mix
and spin down. Incubate reaction at RT for 1 hour up to over-
night (see Note 7).
3. Add 1 μL of Proteinase K solution to reaction and incubate for
10 min at 37 ° C.
4. To transform, add 1–2 μL of LR reaction to 20–50 μL chemi-
cally competent E. coli (e.g., DH5α) and incubate on ice for
122 Geoffrey G. Hesketh et al.
3.2 Generation Our groups have been primarily using the Flp-In T-REx HEK293
of Pools of Stable Cell cell system from Invitrogen for interaction proteomics (other Flp-In
Lines T-Rex cells can alternatively be used with the same constructs). The
system enables tetracycline-inducible expression of the transgenes
expressed at a single copy (from the same Flp Recombination Target
(FRT) containing locus) through a Flp-mediated recombination
event. This system enables robust establishment of cell lines and
pools of cell lines, even for proteins whose expression is toxic for
cell growth (see Note 8).
1. On DAY 0, plate approximately 6 × 105 cells in wells of 6-well
plates such that they are approximately 60–80 % confluent at
the time of transfection (use media that does not contain anti-
biotics for optimal cell viability). In order to transfect with
pcDNA5-FRT-TO based vectors, ensure that media does not
contain Zeocin, to which resistance is lost upon successful
recombination.
2. On DAY 1, transfect cells with 1 μg pOG44 plasmid (expresses
the Flp recombinase) and 100 ng pDEST-pcDNA5-BirA*-
FLAG-ORF (aka, the expression clone) construct using mam-
malian cell transfection reagent of choice according to
manufacturer’s directions (see Note 9). Remember to transfect
your selected controls in parallel (see Note 10).
3. On DAY 2, split each well into a 10 cm dish in complete media
(see Note 11).
4. On DAY 3, aspirate media and replace with complete media
supplemented with 200 μg/mL Hygromycin B to begin selection
of integrated cells. Be careful to avoid cross-contamination
during the selection process.
Parallel BioID and FLAG Affinity-Purification Coupled to Mass Spectrometry 123
3.3 Preparing Cell The protocol for the preparation of the cell pellets and the BioID
Pellets and FLAG-AP purification assumes that only identification and
quantification of the proteins is desired. Note that cell pellets are
typically harvested at room temperature for BioID, whereas pel-
lets should be harvested on ice with chilled PBS for FLAG-AP.
For projects that involve the characterization of specific posttransla-
tional modifications, extra care should be taken (e.g., to inactivate
phosphatases in the case of phosphorylation). This can be done
through the addition of inhibitors, handling all steps with ice-cold
reagents, and shortening some of the incubation times. Also note
that if the analysis of specific peptides is needed, it is recommended
to include alkylation and reduction steps prior to mass spectrometric
identification.
1. Grow 1 × 15 cm plate of cells stably expressing Flp-In BirA*-
FLAG tagged construct (or suitable control) to approximately
80 % confluence in complete media (see Note 13).
2. Add 1 μg/μL of tetracycline (2 μL of a 10 mg/mL stock per
20 mL media) for both BioID and FLAG-AP analysis and
50 μM biotin (50 μL of a 20 mM stock per 20 mL media) for
BioID only. Incubate the cells for 24 h (see Note 14). For
BioID, note that you can pre-build protein expression by incu-
bating with tetracycline for 12–24 h prior to adding biotin for
4–12 h. Keep in mind, however, that you should always time
the incubation of your controls and samples carefully to enable
comparative studies.
124 Geoffrey G. Hesketh et al.
3.4 Parallel BioID The BioID protocol has been optimized for the capture of proxim-
and FLAG-AP ity partners for membrane-associated proteins and other non-
Purification soluble cellular structures (such as RNA granules and bodies), and
differs slightly from those we previously reported for signaling
molecules, identification of E3 ligase substrates, chromatin associ-
ated proteins or the centrosome/cilia, notably in the concentra-
tion of detergents used ([27, 29, 31, 32]; see Fig. 2, and Note 17).
The protocol starts from 1 × 15 cm plate per replicate (normally
corresponds to a dry cell pellet weighing approximately
75–150 mg), which is sufficient for up to six injections on AB Sciex
TripleTOF mass spectrometers (e.g., enabling acquisition of tech-
nical replicates, or parallel Data Dependent and Data Independent
Acquisition runs [13]; see Note 18).
3.4.1 Biotin–Streptavidin Steps 1–8 should be performed at 4 °C and/or on ice (ensure that
Affinity Purification (BioID) all solutions are chilled prior to use), with the remaining steps
and on-Bead Trypsin being carried out at RT.
Digest
Parallel BioID and FLAG Affinity-Purification Coupled to Mass Spectrometry 125
A. B.
1. RIPA 2. BioID Lysis Buffer 1. mild wash 2. harsh wash
Tris-HCl (pH 7.5) 50 mM 50 mM 1x RIPA buffer BioID Wash Buffer
NaCl 150 mM 150 mM 2x FLAG Lysis Buffer BioID Lysis Buffer
SDS 0.1% 0.4%
3x ABC Buffer ABC Buffer
IGEPAL CA-630 - 1%
Triton X-100 1% -
sodium deoxycholate 0.5% - 1 2
EDTA 1 mM - C.
EGTA 1 mM 1 mM
MgCl2 - 1.5 mM 10 20 31
Fig. 2 (a) Detailed comparison of the composition of buffers previously reported: (1) RIPA lysis buffer used for
BioID in our previous study [29], and (2) the modified BioID Lysis Buffer described in this protocol. (b).
Comparison of wash steps after streptavidin affinity-purification (AP): (1) mild wash steps as previously
reported [9], and (2) the harsh wash steps described in this protocol. (c) Number of high confidence preys
identified using BirA-FLAG-DCP1A as bait using the two different methods outlined in A. and B.; (1) RIPA
lysis + mild wash post-AP and (2) BioID Lysis Buffer + harsh wash post-AP
15 μL to the 20 μL packed beads). Beads are best pipetted using
a tip cut at an angle with a clean razor blade; when preparing
beads for multiple samples, prepare a bit extra to account for
losses due to pipetting.
8. Add 20 μL dry bead equivalent as bead slurry (35 μL) to each
sample. Ensure that bead slurry is homogeneously mixed prior
to pipetting beads for each sample. This is effectively achieved
by gently vortexing the tube (at lowest vortex intensity required
to effectively suspend beads) for a few seconds immediately
prior to pipetting. Rotate samples “end-over-end” overnight
at 4 °C (see Note 21).
9. Centrifuge beads at 500 × g for 2 min (see Note 22).
10. Remove supernatant with a vacuum line equipped with a fine
pipette tip (e.g., P20 or “gel loading” tip), being careful to not
disturb or lose beads.
11. Transfer beads in 500 μL of BioID Lysis Buffer into a new
1.5 mL tube (see Note 23). Centrifuge at 500 × g for 30 s and
remove the supernatant.
12. For all washes, add 500 μL of the indicated wash solution, and
suspend beads with brief, gentle vortexing/mixing. Centrifuge
beads at 500 × g for 30 s and remove supernatants as above.
Perform the following washes: (a) Wash once with BioID Wash
Buffer; (b) Wash twice with BioID Lysis Buffer; (c) Wash three
times with 50 mM ammonium bicarbonate (ABC) solution
(see Note 24).
13. Add 100 μL trypsin solution (i.e., 1 μg trypsin in 100 μL ABC)
to beads and incubate 4 h with rotation at 37 °C (see Notes 25
and 26).
14. Add a fresh 1 μg of trypsin (5 μL if stock is prepared as described
in Materials) to each sample and continue incubation over-
night with rotation at 37 °C.
15. Centrifuge beads at 500 × g for 2 min and retrieve 100 μL of
the supernatant (containing the digested peptides) into a new
1.5 mL tube.
16. Add 100 μL of HPLC grade water to the beads and gently
vortex/mix. Centrifuge and collect 100 μL of supernatant as
above, and pool with peptide solution collected in previous
step.
17. Add formic acid to digested peptides to a final concentration of
2 % (use 50 μL of 10 % stock yielding 250 μL total solution)
and vortex briefly.
18. Centrifuge at 16,100 × g for 5 min to eliminate debris and any
residual beads that may have carried over, then transfer exactly
230 μL (being careful not to disrupt the debris “pellet” in the
remaining 20 μL) to new 1.5 mL tubes.
Parallel BioID and FLAG Affinity-Purification Coupled to Mass Spectrometry 127
3.4.2 FLAG Affinity This protocol was developed to perform affinity purification from
Purification (FLAG-AP) 1 × 15 cm plate prepared in Subheading 3.3. In this procedure cells are
and on-Bead Trypsin lysed by passive lysis assisted by freeze–thaw and affinity purification is
Digest performed on a magnetic bead support (see ref. [34] for alternative
protocol using an agarose bead support and for protocol optimiza-
tion). Ensure that all solutions are chilled on ice prior to use.
1. Weigh tubes and subtract tube weight to determine cell pellet
weight.
2. Thaw cell pellets on ice and add FLAG Lysis Buffer at a 4:1
ratio vol (μL): wt (mg) (e.g., 400 μL buffer to a 100 mg
pellet).
3. Break up pellet with pipette (by repeated aspirations) and then
freeze–thaw once by freezing samples on dry ice, then imme-
diately moving tube to a 37 °C water bath with agitation, just
until the visible ice is about to thaw.
4. Centrifuge at 16,100 × g for 20 min at 4 °C and transfer the
supernatants to fresh tubes (remove or avoid the lipid layer on
top of the lysate if present). Save a 20 μL aliquot to monitor
bait expression by Western blot.
5. Prepare the anti-FLAG M2 magnetic beads. Use 25 μL per
sample (supplied by vendor as a 50 % slurry). Wash the required
volume of beads 3–4× with 1 mL of FLAG Lysis Buffer (using
a magnetic tube rack). Resuspend the beads in FLAG Lysis
Buffer to make a 50 % slurry and distribute 25 μL of this into
each sample.
128 Geoffrey G. Hesketh et al.
3.4.3 Mass Spectrometry Mass spectrometric analysis uses a standard LC-MS/MS set-up,
and Data Analysis which we have also described elsewhere [13, 27, 29]. Briefly, using
an autosampler (Eksigent), samples are loaded onto fused silica
capillary columns (0.75 μm ID, 350 μm OD) pre-loaded with
10 cm of C18 reversed phase material (3.5 μm diameter). Ionized
peptides are emitted by nanoelectrospray ion source in-line with a
nano-HPLC system and analyzed using either Data Dependent
(DDA) or Data Independent (DIA) Acquisition methods.
Important considerations include identifying and mitigating
potential carryover issues, and being able to identify which of the
identified proteins are likely to represent bona fide interactors
(or proximity partners in the case of BioID) and which are likely to
be contaminants.
We attempt to address these considerations by: (1) the inclusion
of negative controls in our experimental design (see Note 10); (2) the
processing of at least two biological replicates per bait analyzed,
Parallel BioID and FLAG Affinity-Purification Coupled to Mass Spectrometry 129
4 Notes
Acknowledgment
References
1. Snider J, Kotlyar M, Saraon P, Yao Z, Jurisica I, 7. Breitkreutz A, Choi H, Sharom JR, Boucher L,
Stagljar I (2015) Fundamentals of protein Neduva V, Larsen B, Lin ZY, Breitkreutz BJ,
interaction network mapping. Mol Syst Biol Stark C, Liu G, Ahn J, Dewar-Darch D, Reguly
11(12):848. doi:10.15252/msb.20156351 T, Tang X, Almeida R, Qin ZS, Pawson T,
2. Gingras AC, Gstaiger M, Raught B, Aebersold Gingras AC, Nesvizhskii AI, Tyers M (2010) A
R (2007) Analysis of protein complexes using global protein kinase and phosphatase interac-
mass spectrometry. Nat Rev Mol Cell Biol tion network in yeast. Science 328(5981):1043–
8(8):645–654. doi:10.1038/nrm2208 1046. doi:10.1126/science.1176495
3. Gavin AC, Bosche M, Krause R, Grandi P, 8. Sowa ME, Bennett EJ, Gygi SP, Harper JW
Marzioch M, Bauer A, Schultz J, Rick JM, (2009) Defining the human deubiquitinating
Michon AM, Cruciat CM, Remor M, Hofert enzyme interaction landscape. Cell 138(2):389–
C, Schelder M, Brajenovic M, Ruffner H, 403. doi:10.1016/j.cell.2009.04.042
Merino A, Klein K, Hudak M, Dickson D, 9. Choi H, Larsen B, Lin ZY, Breitkreutz A,
Rudi T, Gnau V, Bauch A, Bastuck S, Huhse B, Mellacheruvu D, Fermin D, Qin ZS, Tyers M,
Leutwein C, Heurtier MA, Copley RR, Gingras AC, Nesvizhskii AI (2011) SAINT:
Edelmann A, Querfurth E, Rybin V, Drewes probabilistic scoring of affinity purification-mass
G, Raida M, Bouwmeester T, Bork P, Seraphin spectrometry data. Nat Methods 8(1):70–73.
B, Kuster B, Neubauer G, Superti-Furga G doi:10.1038/nmeth.1541
(2002) Functional organization of the yeast 10. Dunham WH, Mullin M, Gingras AC (2012)
proteome by systematic analysis of protein Affinity-purification coupled to mass spectrom-
complexes. Nature 415(6868):141–147. etry: basic principles and strategies. Proteomics
doi:10.1038/415141a 12(10):1576–1590. doi:10.1002/pmic.
4. Ho Y, Gruhler A, Heilbut A, Bader GD, Moore 201100523
L, Adams SL, Millar A, Taylor P, Bennett K, 11. Bisson N, James DA, Ivosev G, Tate SA,
Boutilier K, Yang L, Wolting C, Donaldson I, Bonner R, Taylor L, Pawson T (2011) Selected
Schandorff S, Shewnarane J, Vo M, Taggart J, reaction monitoring mass spectrometry reveals
Goudreault M, Muskat B, Alfarano C, Dewar the dynamics of signaling through the GRB2
D, Lin Z, Michalickova K, Willems AR, Sassi adaptor. Nat Biotechnol 29(7):653–658.
H, Nielsen PA, Rasmussen KJ, Andersen JR, doi:10.1038/nbt.1905
Johansen LE, Hansen LH, Jespersen H, 12. Zheng Y, Zhang C, Croucher DR, Soliman
Podtelejnikov A, Nielsen E, Crawford J, MA, St-Denis N, Pasculescu A, Taylor L, Tate
Poulsen V, Sorensen BD, Matthiesen J, SA, Hardy WR, Colwill K, Dai AY, Bagshaw
Hendrickson RC, Gleeson F, Pawson T, Moran R, Dennis JW, Gingras AC, Daly RJ, Pawson
MF, Durocher D, Mann M, Hogue CW, Figeys T (2013) Temporal regulation of EGF signal-
D, Tyers M (2002) Systematic identification of ling networks by the scaffold protein Shc1.
protein complexes in Saccharomyces cerevisiae Nature 499(7457):166–171. doi:10.1038/
by mass spectrometry. Nature 415(6868):180– nature12308
183. doi:10.1038/415180a
13. Lambert JP, Ivosev G, Couzens AL, Larsen B,
5. Hein MY, Hubner NC, Poser I, Cox J, Nagaraj Taipale M, Lin ZY, Zhong Q, Lindquist S,
N, Toyoda Y, Gak IA, Weisswange I, Mansfeld Vidal M, Aebersold R, Pawson T, Bonner R,
J, Buchholz F, Hyman AA, Mann M (2015) A Tate S, Gingras AC (2013) Mapping differen-
human interactome in three quantitative tial interactomes by affinity purification cou-
dimensions organized by stoichiometries and pled with data-independent mass spectrometry
abundances. Cell 163(3):712–723. acquisition. Nat Methods 10(12):1239–1245.
doi:10.1016/j.cell.2015.09.053 doi:10.1038/nmeth.2702
6. Huttlin EL, Ting L, Bruckner RJ, Gebreab F, 14. Collins BC, Gillet LC, Rosenberger G, Rost
Gygi MP, Szpyt J, Tam S, Zarraga G, Colby HL, Vichalkovski A, Gstaiger M, Aebersold R
G, Baltier K, Dong R, Guarani V, Vaites LP, (2013) Quantifying protein interaction dynam-
Ordureau A, Rad R, Erickson BK, Wuhr M, ics by SWATH mass spectrometry: application
Chick J, Zhai B, Kolippakkam D, Mintseris J, to the 14-3-3 system. Nat Methods 10(12):
Obar RA, Harris T, Artavanis-Tsakonas S, 1246–1253. doi:10.1038/nmeth.2703
Sowa ME, De Camilli P, Paulo JA, Harper JW,
Gygi SP (2015) The bioplex network: a sys- 15. Hubner NC, Bird AW, Cox J, Splettstoesser
tematic exploration of the human interac- B, Bandilla P, Poser I, Hyman A, Mann M
tome. Cell 162(2):425–440. d oi:10.1016/j. (2010) Quantitative proteomics combined
cell.2015.06.043 with BAC TransgeneOmics reveals in vivo
Parallel BioID and FLAG Affinity-Purification Coupled to Mass Spectrometry 135
Abstract
Protein–protein interactions (PPIs) play an essential role in all biological processes. In vivo, PPIs occur
dynamically and depend on extracellular cues. To discover novel protein–protein interactions in mamma-
lian cells, we developed a high-throughput automated technology called LUMIER (LUminescence-based
Mammalian IntERactome). In this approach, we co-express a Luciferase (LUC)-tagged fusion protein
along with a Flag-tagged protein in an efficiently transfectable cell line such as HEK-293T cells. The inter-
action between the two proteins is determined by co-immunoprecipitation using an anti-Flag antibody,
and the presence of the LUC-tagged interactor in the complex is subsequently detected via its luciferase
activity. LUMIER can easily detect transmembrane protein partners, interactions that are signaling- or
splice isoform-dependent, as well as those that may occur only in the presence of posttranslational modifi-
cations. Using various collections of Flag-tagged proteins, we have generated protein interaction networks
for several TGF-β family receptors, Wnt pathway members, and have systematically analyzed the effect of
neural-specific alternative splicing on protein interaction networks. The results have provided important
insights into the physiological and functional relevance of some of the novel interactions found. LUMIER
is highly scalable and can be used for both low- and high-throughput strategies. LUMIER is thus a valu-
able tool for the identification and characterization of dynamically regulated PPIs in mammalian systems.
Here, we describe a manual version of LUMIER in a 96-well format that can be easily implemented in any
laboratory.
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_11, © Springer Science+Business Media LLC 2017
137
138 Miriam Barrios-Rodiles et al.
Fig. 1 (a) LUMIER strategy for the detection of protein–protein interactions. A protein of interest (Bait) tagged
with Renilla luciferase (RL or RLUC) is co-expressed in mammalian cells along with a 3Flag-tagged putative
partner (Prey). If the two proteins interact (in the presence or absence of a stimulus), the complex can be
immunoprecipitated using the anti-Flag antibody. The presence of the Bait is detected by light emission upon
Renilla substrate addition. (b) LUMIER strategy for domain mapping. TBRI-HA-RL wild type was expressed
alone or together with wild type (WT) or mutant Occludin (OCLN) harboring the indicated domain deletions (see
schematic), and interactions were measured by LUMIER (top panel) or by immunoblotting with antibody against
HA. Expression of proteins was confirmed by immunoblotting total cell lysates, as indicated. From Barrios-
Rodiles M. et al., 2005 Science (307):1621–1625. Reprinted with Permission from AAAS. (c) LUMIER strategy
for the detection of a ternary complex. The RLUC-tagged Bait is co-expressed with a 3Flag-tagged Prey1 and
a second 3HA-tagged Prey2. A first immunoprecipitation is performed using the anti-Flag antibody and an
aliquot is taken to perform a luciferase assay and western blot to confirm the presence of the Bait and the
Flag-tagged Prey1. Following elution using Flag-peptide to release the complex from the beads, a second
immunoprecipitation is conducted using the anti-HA antibody. A second luciferase assay and western blot is
perfomed to determine the presence of the Bait and the HA-tagged Prey2 (see ref. 19)
2 Materials
Fig. 2 Plasmids used in LUMIER. The vectors are pCMV5-based for high protein expression when transiently
transfected into HEK-239T cells. Only unique restriction enzyme sites at the multiple cloning site (MCS) are
shown. (a and b) NhRLUC and N3Flag are used to tag the Bait and Prey at the N-terminus. There is no stop codon
at the 3′ end sequence of Renilla and N3Flag. (c and d) ChRLUC and C3Flag are used to tag the Bait and Prey at
the C-terminus. The MCS of ChRLUC is a composite of pCMV5 and pCDNA3 polylinker regions. There is a stop
codon at the 3′ end of Renilla and 3Flag. The GenBank accession number for the humanized Renilla luciferase
(hRLUC) sequence is AF362545. The amino acid sequence for the 3Flag is: MDYKDHDGDYKDHDIDYKDDDDK.
The cytomegalovirus (CMV) promoter drives high levels of gene expression for the Bait or Prey. The AmpR gene
encodes beta-lactamase for ampicillin resistance selection in bacteria
2.3 Equipment 1. HandEvac aspirator 8-channel with ejector for disposable tips
(Argos EV514 and EV520).
2. Multichannel pipets (for 10 μl, 20–300 μl, and 1 ml).
142 Miriam Barrios-Rodiles et al.
2.4 Buffers 1. PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, and
1.8 mM KH2PO4. Adjust pH to 7.4 with HCl.
2. Blocking Buffer: 3 % bovine serum albumin, 5 % sucrose, 0.5 %
Tween 20 in PBS.
3. Lysis Buffer: 50 mM Tris–HCl pH 7.4, 150 mM NaCl, 1 mM
tetrasodium EDTA, 0.5 % Triton X-100. A 10× stock can be
stored at 4 °C and diluted to 1× just before lysis when adding
phosphatase and protease inhibitors.
4. Phosphatase Inhibitors (final concentrations): 10 mM sodium
pyrophosphate tetrabasic decahydrate, 1 mM sodium orthovan-
adate, 25 mM NaF. Stock solutions at 10×, 100× and 20×
respectively, can be prepared and stored at 4 °C to be added
just before lysis.
5. Protease Inhibitors: 1 mM phenylmethanesulfonylfluoride
(PMSF; stock 100 mM in ethanol), 10 μg/ml pepstatin A
(stock 1 mg/ml in DMSO), 100 μg/ml soy trypsin inhibitor
(stock 10 mg/ml in Tris–EDTA buffer). 10 μg/ml leupeptin,
10 μg/ml antipain, 50 μg/ml aprotinin, 100 μg/ml benzami-
dine hydrochloride (stocks 1 mg/ml, 1 mg/ml, 5 mg/ml, and
10 mg/ml, respectively in Tris–EDTA buffer). Stocks for pro-
tease inhibitors are stored at −20 °C.
6. Tris–EDTA Buffer for protease inhibitors: 10 mM Tris–HCl
pH 7.4, 1 mM tetrasodium EDTA.
7. Wash buffer: 50 mM Tris–HCl pH 7.4, 150 mM NaCl, 1 mM
tetrasodium EDTA, and 0.1 % Triton X-100.
8. Working Renilla-Glo substrate: Dilute the 100× substrate to
1× using the provided buffer in the kit.
9. PBS-T: 0.05 % Tween 20 in PBS.
10. Buffer to dilute Flag-HRP antibody (Flag-HRP buffer): 5 %
Tween 20 and 1 % fetal bovine serum in PBS.
11. Working Super signal ELISA pico substrate: Reagent A and B
are mixed 1:1 ratio and then diluted 1:5 with distilled water.
3 Methods
Fig. 3 (a) A LUMIER workflow diagram. Forty-eight hours post-transfection in 96-well plates, the cells are
lysed. Lysate is split into Immunoprecipitation (IP) plate and Totals plate. After 1 h incubation the IP plate is
washed and Renilla substrate added to IP and Totals plates to assess the interaction and to measure the
expression levels of the Bait. Subsequently, an anti-Flag-HRP-conjugated antibody is added to the IP plate to
measure the expression levels of the 3Flag-tagged Prey. After lysis, LUMIER results can be obtained within 4 h.
(b) LUMIER data shows that exon 7 of Bridging Integrator 1 (Bin1) promotes an interaction with Dynamin 2
(Dnm2). HEK-293T cells were transfected as indicated. Dnm2-3Flag was used as prey while Bin1-RL (without
exon 7) or Bin1 + ex7-RL (with exon 7) were used a baits. Averages for Renilla luciferase signals from IP plate,
Totals plate and Flag-HRP measurements are shown on the left-side graphs (error bars show ± standard
deviations from four technical replicates). The LIR IP, NLIR, and NtFLIR values calculated as described in Data
Analysis are shown on the right-side graphs indicating that exon 7 promotes the interaction between Bin1 and
Dnm2
3.1 Day 1: Plate 1. Manually plate 15,000 HEK293T cells per well in 70 μl/well,
HEK-293T Cells in poly-l-lysine coated 96-well plates (see Note 3) in DMEM
and Coat Plates with with 10 % fetal bovine serum and antibiotics (penicillin/strep-
Anti-Flag Antibody tomycin). Incubate overnight at 37 °C.
144 Miriam Barrios-Rodiles et al.
3.3 Day 4: Lyse Cells 1. Forty-eight hours post-transfection aspirate media from cells
and Perform and dispense 150 μl/well of Lysis Buffer that includes phos-
Immunoprecipitation phatase and protease inhibitors (see Note 7). Shake plate at low
(IP) and Flag-Tagged speed for 40 min at 4 °C.
Prey Measurement 2. Transfer 90 μl of lysate to anti-Flag-coated plate (Lumitrac; IP
plate) and 20 μl to a round bottom-white plate (white round-
bottom; Totals plate). Incubate both plates at 4 °C for 1 h,
without shaking.
3. Aspirate liquid from IP plate using the 8-channel aspirator and
dispense 250 μl of Wash buffer for each wash. Repeat wash step
5–6 times. Aspirate liquid from last wash and flick off remain-
ing liquid. Dispense 100 μl/well of Working Renilla-Glo sub-
strate to IP plates, incubate for 10 min at room temperature,
and read in ENVISION luminometer 100 ms/well to detect
interactions (see Note 8).
4. Measure expression levels of the Renilla-tagged proteins.
Dispense 20 μl/well of Working Renilla-Glo substrate to
Totals plate containing 20 μl of lysate, incubate for 10 min at
room temperature, and read in ENVISION luminometer
100 ms/well.
5. Aspirate the liquid from IP plate and flick to remove all liquid.
Wash IP plate eight times with 250 μl/well of PBS-T each
time. Dispense 100 μl/well of anti-Flag-HRP antibody at
1:20,000 dilution (see Note 9) in Flag-HRP Buffer. Incubate
for 30 min at 4 °C without shaking.
LUMIER and Mammalian Protein Interaction Networks 145
6. Aspirate all liquid and wash eight times with 250 μl/well PBS-
T. Aspirate liquid from last wash and flick off remaining liquid.
Dispense 100 μl/well of Working Super signal ELISA pico
substrate. Incubate at room temperature for 7 min and read in
ENVISION for 20 ms/well.
3.4 Data Analysis The Normalized LUMIER Intensity Ratio (NLIR) value for each
interaction is calculated as follows: The LUMIER Intensity Ratio
for each Bait/Prey combination is first determined from the lucif-
erase activity in the immunoprecipitates (LIR IP). This LIR IP is
calculated as the luciferase intensity in IP plate when each Bait
(RLUC-fusions) and Prey (3Flag-tagged cDNAs) are co-
transfected, over the luciferase intensity from immunoprecipitates
when the Bait (RLUC-fusion) is co-transfected with empty vector
(pCMV5).
LIR IP = signal IP (Bait + Prey)/signal IP (Bait + pCMV5).
The LIR TOT is caculated in a similar manner from the lucif-
erase activity measured in the total cell lysates (Totals plate) of each
Bait/Prey combination over the luciferase activity from total cell
lysates of each Bait/empty vector.
LIR TOT = signal Totals (Bait + Prey)/signal Totals (Bait + pCMV5).
The Normalized LIR (NLIR) is the ratio of LIR IP/LIR TOT
for each Bait/Prey combination. In general, an NLIR value equal
or higher than 3 is considered as a positive interaction.
To give a more quantitative interaction score a Flag Intensity
Ratio (FIR) is calculated to account for changes in the immuno-
precipitation of Flag-tagged Prey protein with different RLUC-
tagged Baits. This FIR is calculated with the Flag-HRP signal value
from immunoprecipitates when Bait and Prey are co-transfected,
over the Flag-HRP signal value from immunoprecipitates when
the Prey is co-transfected with empty vector.
FIR = signal Flag-HRP (Bait + Prey)/signal Flag-HRP
(Bait + pCMV5).
A final Normalized to Flag LIR value (NtFLIR) is the ratio of
NLIR/FIR for each interaction tested. See Fig. 3b for an example
on data analysis.
4 Notes
1. The plasmids to tag the Bait with Renilla luciferase and the
Prey with 3Flag are pCMV5-based. There is also a Gateway®
version for each.
2. The Baits and Preys can be tagged at either N- or C-terminus.
This is an important consideration when designing a LUMIER
experiment. If the Bait of interest has an important domain at
146 Miriam Barrios-Rodiles et al.
Acknowledgments
References
1. Fields S, Song O (1989) A novel genetic sys- 7. Pelletier JN, Campbell-Valois FX, Michnick
tem to detect protein-protein interactions. SW (1998) Oligomerization domain-directed
Nature 340(6230):245–246 reassembly of active dihydrofolate reductase
2. Lievens S, Lemmens I, Tavernier J (2009) from rationally designed fragments. Proc Natl
Mammalian two-hybrids come of age. Trends Acad Sci U S A 95(21):12141–12146
Biochem Sci 34(11):579–588 8. Wilson CG, Magliery TJ, Regan L (2004)
3. Braun P, Tasan M, Dreze M, Barrios-Rodiles Detecting protein-protein interactions with
M, Lemmens I, Yu H, Sahalie JM, Murray RR, GFP-fragment reassembly. Nat Methods
Roncari L, de Smet AS, Venkatesan K, Rual JF, 1(3):255–262
Vandenhaute J, Cusick ME, Pawson T, Hill 9. Remy I, Michnick SW (2006) A highly sensi-
DE, Tavernier J, Wrana JL, Roth FP, Vidal M tive protein-protein interaction assay based on
(2009) An experimentally derived confidence Gaussia luciferase. Nat Methods 3(12):
score for binary protein-protein interactions. 977–979
Nat Methods 6(1):91–97 10. Boute N, Jockers R, Issad T (2002) The use of
4. Eyckerman S, Verhee A, der Heyden JV, resonance energy transfer in high-throughput
Lemmens I, Ostade XV, Vandekerckhove J, screening: BRET versus FRET. Trends
Tavernier J (2001) Design and application of a Pharmacol Sci 23(8):351–354
cytokine-receptor-based interaction trap. Nat 11. Ramachandran N, Hainsworth E, Bhullar B,
Cell Biol 3(12):1114–1119 Eisenstein S, Rosen B, Lau AY, Walter JC,
5. Stagljar I, Korostensky C, Johnsson N, te LaBaer J (2004) Self-assembling protein
Heesen S (1998) A genetic system based on microarrays. Science 305(5680):86–90
split-ubiquitin for the analysis of interactions 12. Barrios-Rodiles M, Brown KR, Ozdamar B,
between membrane proteins in vivo. Proc Natl Bose R, Liu Z, Donovan RS, Shinjo F, Liu Y,
Acad Sci U S A 95(9):5187–5192 Dembowy J, Taylor IW, Luga V, Przulj N,
6. Petschnigg J, Groisman B, Kotlyar M, Taipale Robinson M, Suzuki H, Hayashizaki Y, Jurisica
M, Zheng Y, Kurat CF, Sayad A, Sierra JR, I, Wrana JL (2005) High-throughput mapping
Mattiazzi Usaj M, Snider J, Nachman A, of a dynamic signaling network in mammalian
Krykbaeva I, Tsao MS, Moffat J, Pawson T, cells. Science 307(5715):1621–1625
Lindquist S, Jurisica I, Stagljar I (2014) The 13. Ozdamar B, Bose R, Barrios-Rodiles M, Wang
mammalian-membrane two-hybrid assay HR, Zhang Y, Wrana JL (2005) Regulation of
(MaMTH) for probing membrane-protein the polarity protein Par6 by TGFbeta receptors
interactions in human cells. Nat Methods controls epithelial cell plasticity. Science
11(5):585–592 307(5715):1603–1609
148 Miriam Barrios-Rodiles et al.
14. Miller BW, Lau G, Grouios C, Mollica E, embryonic stem cells. Cell Rep 5(6):
Barrios-Rodiles M, Liu Y, Datti A, Morris Q, 1611–1624
Wrana JL, Attisano L (2009) Application of an 20. Ellis JD, Barrios-Rodiles M, Colak R, Irimia
integrated physical and functional screening M, Kim T, Calarco JA, Wang X, Pan Q,
approach to identify inhibitors of the Wnt O’Hanlon D, Kim PM, Wrana JL, Blencowe
pathway. Mol Syst Biol 5:315 BJ (2012) Tissue-specific alternative splicing
15. Hall MP, Unch J, Binkowski BF, Valley MP, remodels protein-protein interaction networks.
Butler BL, Wood MG, Otto P, Zimmerman K, Mol Cell 46(6):884–892
Vidugiris G, Machleidt T, Robers MB, Benink 21. Taipale M, Krykbaeva I, Koeva M, Kayatekin
HA, Eggers CT, Slater MR, Meisenheimer PL, C, Westover KD, Karras GI, Lindquist S
Klaubert DH, Fan F, Encell LP, Wood KV (2012) Quantitative analysis of HSP90-client
(2012) Engineered luciferase reporter from a interactions reveals principles of substrate rec-
deep sea shrimp utilizing a novel imidazopyr- ognition. Cell 150(5):987–1001
azinone substrate. ACS Chem Biol 7(11): 22. Ryzhakov G, Teixeira A, Saliba D, Blazek K,
1848–1857 Muta T, Ragoussis J, Udalova IA (2013)
16. Xu G, Barrios-Rodiles M, Jerkic M, Turinsky Cross-species analysis reveals evolving and con-
AL, Nadon R, Vera S, Voulgaraki D, Wrana JL, served features of the nuclear factor kappaB
Toporsian M, Letarte M (2014) Novel protein (NF-kappaB) proteins. J Biol Chem 288(16):
interactions with endoglin and activin receptor- 11546–11554
like kinase 1: potential role in vascular net- 23. Blasche S, Mortl M, Steuber H, Siszler G, Nisa
works. Mol Cell Proteomics 13(2):489–502 S, Schwarz F, Lavrik I, Gronewold TM, Maskos
17. Kittanakom S, Barrios-Rodiles M, Petschnigg K, Donnenberg MS, Ullmann D, Uetz P, Kogl
J, Arnoldo A, Wong V, Kotlyar M, Heisler LE, M (2013) The E. coli effector protein NleF is a
Jurisica I, Wrana JL, Nislow C, Stagljar I caspase inhibitor. PLoS One 8(3):e58937
(2014) CHIP-MYTH: a novel interactive pro- 24. Deng Q, Wang D, Li F (2014) Detection of
teomics method for the assessment of agonist- viral protein-protein interaction by microplate-
dependent interactions of the human format luminescence-based mammalian inter-
beta(2)-adrenergic receptor. Biochem Biophys actome mapping (LUMIER). Virol Sin 29(3):
Res Commun 445(4):746–756 189–192
18. Varelas X, Miller BW, Sopko R, Song S, Gregorieff 25. Tahoun A, Siszler G, Spears K, McAteer S, Tree
A, Fellouse FA, Sakuma R, Pawson T, Hunziker J, Paxton E, Gillespie TL, Martinez-Argudo I,
W, McNeill H, Wrana JL, Attisano L (2010) The Jepson MA, Shaw DJ, Koegl M, Haas J, Gally
Hippo pathway regulates Wnt/beta-catenin sig- DL, Mahajan A (2011) Comparative analysis of
naling. Dev Cell 18(4):579–591 EspF variants in inhibition of Escherichia coli
19. Beyer TA, Weiss A, Khomchuk Y, Huang K, phagocytosis by macrophages and inhibition of
Ogunjimi AA, Varelas X, Wrana JL (2013) E. coli translocation through human- and
Switch enhancers interpret TGF-beta and bovine-derived M cells. Infect Immun 79(11):
Hippo signaling to control cell fate in human 4716–4729
Chapter 12
Abstract
Generating molecular information in a clinically relevant time frame is the first hurdle to truly integrating precision
medicine in health care. Reverse phase protein microarrays are being utilized in clinical trials for quantifying
posttranslationally modified signal transduction proteins and cellular signaling pathways, allowing direct com-
parison of the activation state of proteins from multiple specimens, or individual patient specimens, within the
same array. This technology provides diagnostic and therapeutic information critical to precision medicine. To
enhance accessibility of this technology, two hurdles must be overcome: data normalization and data acquisi-
tion. Herein we describe an unamplified, dual-color signal detection methodology for reverse phase protein
microarrays that allows multiplex, within spot data normalization, reduces data acquisition time, simplifies
automated spot detection, and provides a stable signal output. This method utilizes Quantum Nanocrystal fluo-
rophore labels (Qdot) substituted for organic fluorophores coupled with an imager (ArrayCAM) that captures
images of the microarray rather than sequentially scanning the array. Streamlining and standardizing the data
analysis steps with ArrayCAM high-resolution, dual mode chromogenic/fluorescent array imaging overcomes
the data acquisition hurdle. The spot location and analysis algorithm provides certain parameter settings
that can be tailored to the particular microarray type (fluorescent vs. colorimetric), resulting in greater than 99 %
spot location sensitivity. The described method demonstrates equivalent sensitivity for a non-amplified Qdot
immunoassay when using automated vs. manual immunostaining procedures.
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_12, © Springer Science+Business Media LLC 2017
149
150 Solomon Yeon et al.
Table 1
Examples of US laboratories performing reverse phase protein microarray technology
(continued)
Dual-Color Protein Microarray Analysis 151
Table 1
(continued)
Fig. 2 Equivalent resolution in the detection of relevant biologically pathways previously observed using colorimet-
ric detection can be achieved on the ArrayCAM platform. Strong concordance in spot signal intensities between
a β-Actin stained array imaged on a flatbed scanner (x-axis, Image Quant ver 5.1 software spot analysis) and
the ArrayCAM imager (y-axis, ArrayCAM software spot analysis). β-Actin was detected with a catalyzed signal
amplification method using colorimeteric detection [18]
154 Solomon Yeon et al.
2 Materials
2.1 Manual Dual- 1. Reverse phase protein microarrays printed with whole cell
Color Qdot lysates, serum, vitreous, laser capture microdissected cell
Immunostaining lysates, peripheral blood mononuclear cells, or other protein
containing body fluids.
2. ProPlate slide chambers and spring clips (Grace Bio-Labs):
1-well format.
3. Orbital shaker.
4. Pipettes (1 μL–50 mL delivery volume).
5. Plastic trays/petri dishes with lids.
6. Type 1 reagent grade water (dH2O).
7. Antigen retrieval solution, 10 % sodium hydroxide in water,
10×. Store at 4 °C. Dilute to 1× prior to use: Prepare 100 mL
of 1× Antigen Retrieval by adding 10 mL of 10× Antigen
Retrieval to 90 mL dH2O. Mix thoroughly. Store at 4 °C.
8. Q-Block blocking buffer, proprietary composition (Grace Bio-
Labs). Store at 4 °C.
9. Primary antibodies against analytes of interest (anti-rabbit
and/or anti-mouse, unconjugated) (see Note 1).
10. Antibody diluent proprietary composition (Grace Bio-Labs).
Store at 4 °C.
11. 10× Wash Buffer Q, proprietary composition (Grace Bio-
Labs). Store at 4 °C. Prior to use prepare 400 mL of 1× Wash
Buffer Q by adding 40 mL of 10× Wash Buffer Q to 360 mL
Dual-Color Protein Microarray Analysis 155
2.2 Automated 1. Reverse phase protein microarrays printed with whole cell
Dual-Color Qdot lysates, serum, vitreous, laser capture microdissected cell
Immunostaining lysates, peripheral blood mononuclear cells, or other protein
containing body fluids.
2. Automated slide stainer with open-source reagents (e.g., Dako
Autostainer).
3. Orbital shaker.
4. Plastic trays/petri dishes with lids.
5. Pipettes (1 μL–50 mL delivery volume).
6. Type 1 reagent grade water (dH2O).
7. Antigen retrieval solution 10×, 10 % sodium hydroxide in
water. Store at 4 °C. Dilute to 1× prior to use as described in
item 7 of Subheading 2.1.
8. Q-Block blocking buffer, proprietary composition (Grace Bio-
Labs). Store at 4 °C.
9. Primary antibodies (anti-rabbit and/or anti-mouse, unconju-
gated) (see Note 1).
10. Antibody diluent, proprietary composition (Grace Bio-Labs).
Store at 4 °C.
11. 10× Wash Buffer Q, proprietary composition (Grace Bio-
Labs). Store at 4 °C. Dilute to 1× prior to use as described in
item 11 of Subheading 2.1.
12. 10× Rinse buffer, proprietary composition (Grace Bio-Labs).
Store at 4 °C. Dilute to 1× prior to use as described in item 12
of Subheading 2.1.
13. Qdot nanocrystals conjugated to secondary antibodies: 800 nm
goat anti-rabbit and 655 nm goat anti-mouse. Store at 4 °C.
Do not freeze.
14. Detection reagent diluent, proprietary composition (Grace
Bio-Labs). Store at 4 °C.
156 Solomon Yeon et al.
3 Methods
3.1 Manual Dual- 1. Retrieve RPPA slides from storage (−20 °C). Allow the slides
Color QDot RPPA to reach room temperature. The number of slides to stain
Immunostaining equals the number of primary antibodies plus 1 for a secondary
antibody only control [12, 18].
2. Wash RPPA slides 4 × 15 min in fresh dH2O. Place slides with
the nitrocellulose film facing up, in a plastic dish. Add adequate
amounts of water to completely immerse the slides. Place the
dish on an orbital rocker with gentle rocking for 15 min.
Remove and discard the water. Repeat the wash steps three
times. Do not allow the slides to dry out at any point during
the staining.
Dual-Color Protein Microarray Analysis 157
18. Remove and discard the Rinse Buffer. Remove the ProPlate
chamber from each array slide.
19. Rinse the slides in copious volumes of fresh dH2O.
20. Allow the slides to dry at room temperature, protected from
light.
Table 2
Autostainer program for multiplex, two-color Qdot RPPA staining
Fig. 4 ArrayCAM spot finding features. The ArrayCAM spot finding software depicts spots that meet the spot
search parameters as green circles and those spots that fail the search parameters are depicted as red circles
with lines through them. The 3-dimensional features of the spot can be readily visualized with a 3-D spot
inspection tool
3.3.2 Image Acquisition 1. Load a slide by pushing the ArrayCAM lid backwards to expose
the slide stage.
2. Orient the slide face down with the label on the left (see Note 2).
3. Secure the slide on the stage by pushing the two clasps backwards
and close the lid.
4. Click on “File Info” tab (Fig. 1).
162 Solomon Yeon et al.
3.3.3 Image Analysis 1. Click the “File Info” tab, and load a previously created naviga-
tional .gal file by clicking on the “Load .gal File” button.
2. Select the “1 × 1 Sub-array” and click ok, then choose the
appropriate .gal file.
3. Click on the “Image Control” tab. Click on the “Invert Color”
button if the image represents a colorimetric assay.
4. Under the “Auto tab”, set values as (see Note 8):
5. Click the “Params” button and set the values as (see Note 9):
4 Notes
negative
buffer
blank
control
content
blocker
Each type designator can be of all lower case letters and the
first letter can be capitalized. No other case options are permit-
ted. Each must be spelled exactly as shown.
The user may construct any navigational .gal file as long as
it obeys the general rules for .gal file construction and includes
the additional information as described above.
Save the navigational .gal file to the directory C:\GBL\Gal\
Controls
ATF 1
8 6
Type = GenePix ArrayList V1.0
Supplier = Grace Bio-Labs
ArrayerSoftwareName = None
ArrayerSoftwareVersion = None
BlockCount = 1
BlockType = 0
SlideType = 1
“Block1 = 4400,1400,200,13,433,13,433”
Block Column Row ID Name Type
1 1 1 ID FIDUCIAL fiducial
1 2 1 ID BLANK blank
1 3 1 ID BLANK blank
1 4 1 ID FIDUCIAL fiducial
1 5 1 ID BLANK blank
1 6 1 ID BLANK blank
1 7 1 ID BLANK blank
1 8 1 ID FIDUCIAL fiducial
1 9 1 ID BLANK blank
1 10 1 ID BLANK blank
1 11 1 ID BLANK blank
1 12 1 ID BLANK blank
1 13 1 ID FIDUCIAL fiducial
1 1 2 ID Content analyte
(continued)
168 Solomon Yeon et al.
1 2 2 ID Content analyte
1 3 2 ID Content analyte
1 4 2 ID Content analyte
1 5 2 ID Content analyte
1 6 2 ID Content analyte
1 7 2 ID Content analyte
1 8 2 ID Content analyte
1 9 2 ID Content analyte
1 10 2 ID Content analyte
1 11 2 ID Content analyte
1 12 2 ID Content analyte
1 13 2 ID Content analyte
Acknowledgments
References
1. Holmes FA, Espina V, Liotta LA, Nagarwala study utilizing multi-omic molecular profiling
YM, Danso M, McIntyre KJ (2013) Pathologic to find potential targets and select individual-
complete response after preoperative anti- ized treatments for patients with previously
HER2 therapy correlates with alterations in treated metastatic breast cancer. Breast Cancer
PTEN, FOXO, phosphorylated Stat5, and Res Treat 147:579–588
autophagy protein signaling. BMC Res Notes 14. Pierobon M, Silvestri A, Spira A, Reeder A, Pin
6:507 E, Banks S (2014) Pilot phase I/II personal-
2. Korf U, Derdak S, Tresch A, Henjes F, ized therapy trial for metastatic colorectal can-
Schumacher S, Schmidt C (2008) Quantitative cer: evaluating the feasibility of protein pathway
protein microarrays for time-resolved measure- activation mapping for stratifying patients to
ments of protein phosphorylation. Proteomics therapy with imatinib and panitumumab.
8:4603–4612 J Proteome Res 13:2846–2855
3. Liotta LA, Espina V, Mehta AI, Calvert V, 15. Robertson FM, Petricoin EF III, Van Laere SJ,
Rosenblatt K, Geho D (2003) Protein micro- Bertucci F, Chu K, Fernandez SV (2013)
arrays: meeting analytical challenges for clinical Presence of anaplastic lymphoma kinase in
applications. Cancer Cell 3:317–325 inflammatory breast cancer. Springerplus 2:497
4. MacBeath G, Schreiber SL (2000) Printing pro- 16. Wulfkuhle JD, Speer R, Pierobon M, Laird J,
teins as microarrays for high-throughput function Espina V, Deng J (2008) Multiplexed cell sig-
determination. Science 289:1760–1763 naling analysis of human breast cancer applica-
5. Paweletz CP, Charboneau L, Bichsel VE, tions for personalized therapy. J Proteome Res
Simone NL, Chen T, Gillespie JW (2001) 7:1508–1517
Reverse phase protein microarrays which cap- 17. Xia W, Petricoin EF III, Zhao S, Liu L, Osada
ture disease progression show activation of T, Cheng Q (2013) An heregulin-EGFR-
pro-survival pathways at the cancer invasion HER3 autocrine signaling axis can mediate
front. Oncogene 20:1981–1989 acquired lapatinib resistance in HER2+ breast
6. Petricoin EF III, Espina V, Araujo RP, Midura cancer models. Breast Cancer Res 15:R85
B, Yeung C, Wan X (2007) Phosphoprotein 18. Gallagher RI, Silvestri A, Petricoin EF III,
pathway mapping: Akt/mammalian target of Liotta LA, Espina V (2011) Reverse phase
rapamycin activation is negatively associated protein microarrays: fluorometric and colori-
with childhood rhabdomyosarcoma survival. metric detection. Methods Mol Biol
Cancer Res 67:3431–3440 723:275–301
7. Gallagher RI, Espina V (2014) Reverse phase 19. Tibes R, Qiu Y, Lu Y, Hennessy B, Andreeff
protein arrays: mapping the path towards per- M, Mills GB (2006) Reverse phase protein
sonalized medicine. Mol Diagn Ther array: validation of a novel proteomic technol-
18:619–630 ogy and utility for analysis of primary leukemia
8. Huels C, Muellner S, Meyer HE, Cahill DJ specimens and hematopoietic stem cells. Mol
(2002) The impact of protein biochips and Cancer Ther 5:2512–2521
microarrays on the drug development process. 20. VanMeter AJ, Rodriguez AS, Bowman ED, Jen
Drug Discov Today 7:S119–S124 J, Harris CC, Deng J (2008) Laser capture
9. Jameson JL, Longo DL (2015) Precision med- microdissection and protein microarray analysis
icine—personalized, problematic, and promis- of human non-small cell lung cancer: differen-
ing. N Engl J Med 372:2229–2234 tial epidermal growth factor receptor (EGPR)
10. Mueller C, Liotta LA, Espina V (2010) Reverse phosphorylation events associated with
phase protein microarrays advance to use in mutated EGFR compared with wild type. Mol
clinical trials. Mol Oncol 4:461–481 Cell Proteomics 7:1902–1924
11. Cremona M, Espina V, Caccia D, Veneroni S, 21. Chiechi A, Mueller C, Boehm KM, Romano A,
Colecchia M, Pierobon M (2014) Stratification Benassi MS, Picci P (2012) Improved data nor-
of clear cell renal cell carcinoma by signaling malization methods for reverse phase protein
pathway analysis. Expert Rev Proteomics microarray analysis of complex biological
11:237–249 samples. Biotechniques 1–7
12. Espina V, Liotta LA, Petricoin EF III (2009) 22. Mannsperger HA, Gade S, Henjes F, Beissbarth
Reverse-phase protein microarrays for ther- T, Korf U (2010) RPPanalyzer: analysis of
anostics and patient tailored therapy. Methods reverse-phase protein array data. Bioinformatics
Mol Biol 520:89–105 26:2202–2203
13. Jameson GS, Petricoin EF, Sachdev J, Liotta 23. Stanislaus R, Carey M, Deus HF, Coombes K,
LA, Loesch DM, Anthony SP (2014) A pilot Hennessy BT, Mills GB (2008) RPPAML/
170 Solomon Yeon et al.
Abstract
The ability to enumerate all of the proteins in a cell is quickly becoming a reality. Quantitative proteomics
adds an extra dimension to proteome-wide discovery experiments by enabling differential measurements
of protein concentrations, characterization of protein turnover, increased stringency of co-
immunoprecipitation reactions, as well as many other intriguing applications. One of the most widely used
techniques that enable relative protein quantitation is stable isotope labeling by amino acids in cell culture
(SILAC) (Ong et al., Mol Cell Proteomics 1(5):376–386, 2002). Over the past decade, SILAC has
become the preferred approach for proteome-wide quantitation by mass spectrometry. This approach
relies on the metabolic incorporation of isotopically enriched amino acids into the proteome of cells—the
proteome of “light” (1H, 12C, 14N) cells can then be compared to “heavy” (2H, 13C, 15N) cells as the iso-
topically labeled proteins and peptides are easily distinguished in a mass spectrometer. Since cellular uptake
and response to isotopically different amino acid(s) is naïve, it is without impact on cell physiology. We
provide a detailed step-by-step procedure for performing SILAC-based experiment for proteome-wide
quantitation in this chapter.
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_13, © Springer Science+Business Media LLC 2017
171
172 Kian Kani
Quantitation
>5
Number of samples Isobaric mass tags
<2
ICAT/Acrylamide/di-
methyl
Fig. 1 Quantitative proteomics flowchart for either metabolic labeling or chemical addition of isobaric mass tags
one amino acid and introduces various artifacts that affect quanti-
fication by increasing the number of unwanted peptide ion peaks.
Of particular concern is the biosynthesis of proline from excess
arginine [7]. This impacts all SILAC experiments that utilize argi-
nine. The best workaround for this complication is to supplement
SILAC media with standard l-proline, thereby suppressing the
endogenous conversion enzymes [8]. We suggest that all SILAC
media contain 100 mg/L of l-proline in order to prevent the con-
version of isotope-coded arginine to proline.
The application of the SILAC approach is based on the research
aims of the investigators. Some of the early works focused on dif-
ferential expression of proteins in various model systems (yeast, E.
coli, C. elegans, Arabidopsis, etc.) [9–11]. The combination of
SILAC based quantitation with enrichment strategies that enable
measurement of protein posttranslational modifications has been a
powerful application of this technique. For example, Bose et al.
discovered novel downstream phosphorylation targets of the onco-
gene HER2 [12]. A number of other reports have applied SILAC
based approaches to measure changes in glycosylation [13], ubiq-
uitinylation [14], methylation [15], and histone modification [16].
Global changes in various host systems may also be monitored by
the SILAC approach. For example, Kruger et al. metabolically
labeled a mouse with heavy lysine and demonstrated the applica-
tion of quantitative proteomics with knockout mice [17]. A num-
ber of other experimental strategies involving SILAC have been
developed. For instance, Schwanhausser et al. utilized pulsed
incorporation of amino acids to measure kinetics of protein turn-
over (pulsed-SILAC) [18]. SILAC may also be used to enhance
the reliability of absolute protein quantitation [19] and facilitate
improvement in reducing the false positive identification of con-
taminant proteins in CO-IP LC-MS/MS experiments [20, 21]. In
the following section, we describe a step-by-step procedure for a
typical two-plex based SILAC experiment (Fig. 2).
2 Materials
All cell culture material should be sterile and cell lines should be
tested for mycoplasma contamination prior to amino acid incorpo-
ration with a suitable kit (Mycoprobe Mycoplasma detection kit,
R&D Systems, Minneapolis, MN). All buffers and solvents should
be prepared with Milli-Q water (Millipore, Billerica, MA).
1. Appropriate media for your cell line deficient in arginine,
lysine, and glutamine (see Note 1). For example, RPMI 1640
Medium deficient in arginine, lysine, and glutamine (Life
Technologies, Carlsbad, CA).
A. Ctrl Treatment
13 12
Heavy ( C Lysine) Light( C Lysine)
6 6
Mix lysates
(equal mass)
Reduction/Alkylation
Digest (trypsin/lys-c)
Analysis
MS
MS/MS
6 Da
Intensity
M/Z
Fig. 2 Schematic of a typical two-plex SILAC experiment. Metabolic labeling of cells occurs in rapidly dividing
cultures. Cell are passaged six times in SILAC media. Cells are lysed in an appropriate buffer and subject to
protein concentration determination by Bradford or BCA. Equal masses of proteins are mixed and subject to
reduction and alkylation. Lysates are treated with trypsin or lys-c and analyzed by LC-MS/MS. The area of the
“light” and “heavy” peptides are used for quantitation
Table 1
Recommended amino acids for SILAC experiments
amino acids 13
C6, 15N4
arginine
3 plex 1
H/12C/14N light D4 lysine and C6 15N2 Lysine and
13
amino acids 13
C6 arginine 13
C6, 15N4 arginine
4 plex 1
H/12C/14N light D4 lysine and 13
C6 15N2 Lysine and 13
C6 15N2 D9
amino acids 13
C6 arginine 13
C6, 15N4 arginine Lysine and D7
arginine
5 plex 12
C6, 14N4 12
C6, 15N4 13
C6, 14N4 arginine 13
C6, 15N4 arginine D7 13C6, 15N4
arginine arginine arginine
3 Methods
3.1 Cell Culture 1. Seed one 10 cm plate per condition in normal media at 25–50 %
and Labeling confluence the day before starting the SILAC experiment (Day
0, see Note 4 for comments on experiments requiring a larger
quantity of cells for final analysis).
2. Obtain one (1 L) Sterile Vacuum Filter for each growth condi-
tion (e.g., “Heavy” and “Light”) inside a sterile laminar bio-
logic safety bench.
3. Prepare SILAC media by adding the following items to the top
chamber of the vacuum filter:
(a) 860 mL of media.
(b) 1 mL of 1000× proline (to suppress arginine conversion).
(c) 1 mL of heavy or light 1000× amino acids as appropriate
(see Note 5).
(d) 10 mL 100× glutamine.
(e) 18 mL 1 M HEPES.
(f) 10 mL 100× of Penicillin/streptomycin.
(g) 100 mL dialyzed FBS.
4. Allow components to filter through the bottom chamber using
vacuum, discard the upper section. Label the media appropri-
ately (e.g., “Heavy” or “Light”).
5. On Day 1, aspirate the media and wash cells with PBS thrice.
Quantitative Proteomics Using SILAC 177
3.2 Cell Lysis Each lysis condition is application and protocol dependent. We will
describe a simple urea lysis procedure that is applicable to many
subsequent analysis pipelines.
1. Seed the final passage of cells such that they reach 85 % conflu-
ence 48 h after passaging.
2. For Adherent cells, place 10 cm plates on ice and remove by
aspiration.
3. Wash cells thrice with 10 mL of ice cold PBS. Aspirating PBS
between washes (Note: Do not wash cells off the plate).
4. After the last wash, tilt plate at 45° angle for 1 min thereby
allowing the remaining PBS to drain to bottom of plate.
5. Completely aspirate remaining PBS.
6. Add 1 mL of urea lysis buffer.
7. Scrape cells of plate with a Nunc Cell scraper (Fisher Scientific,
Pittsburg, PA).
8.
Remove cells and pipette into a prechilled 1.5 mL
microcentrifuge.
9. Freeze cells at −20 °C for at least 2 h (Note: cells can be kept
frozen for later analysis).
10. Thaw cells on ice.
11. Prepare cell lysate by passing cell material through a 27 gauge
needle at least ten times (BD Sciences, Thermo Scientific,
Pittsburg, PA).
178 Kian Kani
3.3 Practical The following section will outline some of the practical consider-
Example ations that are involved in a typical two-plex SILAC based experi-
ment. This example can be modified to fit various experimental
conditions, cell lines, and controls. One of the key challenges in
clinical oncology is to determine which patient is going to benefit
from a particular therapeutic regime. Biomarkers that indicate
therapy response are useful clinical tools for patient stratification.
Quantitative Proteomics Using SILAC 179
We will utilize the cellular dependence of the A431 cell line (ATCC,
Manassas, VA) to EGFR signaling as a surrogate for lung cancer
patients that harbor EGFR activating mutations in their tumors.
These patients have dramatic response rates to tyrosine kinase
inhibitors (TKIs) that target EGFR. The goal of the following
experiment is to identify proteins that are either upregulated or
downregulated in the EGFR dependent A431 cell line upon treat-
ment with an EGFR TKI (Iressa).
1. Two-plex SILAC experiments can be done with various com-
binations of “light” and “heavy” amino acids. In this example,
utilize SILAC with 13C6 lysine which results in a mass shift of
6 Da.
2. Maintain two cultures of A431 cells in either heavy or light
SILAC media and passage them for at least six generations (see
Note 7).
3. Perform technical replicates by dosing “heavy” labeled cells
with drug (in this case Iressa) in order to comparing the pro-
teome with “light” labeled cells treated with vehicle. The
reciprocal experiment is “heavy” labeled cells treated with
vehicle and compared to “light” labeled cells treated with
drug.
4. Two 15 cm plates of “heavy” labeled A431 cells and two 15 cm
plates of “light” labeled A431 cells are seeded at 50 % conflu-
ence (Note: the confluence and general health of cells is an impor-
tant component to SILAC based in vitro experiments. Proper
care should be placed on cell health).
5. The following day, dose cells “heavy” labeled cells with 100 nM
Iressa and “light” labeled cells with vehicle control.
6. Repeat the dosing on the reciprocal labeled cells.
7. After 16 h, proceed to cell lysis step as detailed in Subheading 3.2.
8. Perform LC-MS/MS analysis; in general, peptides identifica-
tions are based on the MS/MS spectra and the m/z of the
monoisotopic peak.
9. Peptide quantitation is obtained from the extracted ion chro-
matogram (XIC) of two differently labeled versions of the pep-
tide. The XIC is the contribution of a given m/z to the total
ion chromatogram (TIC). The main components of the XIC
are the m/z value, the elution time profile, and the intensity of
the peak. A number of different software packages are available
including Proteowizard [22], MSQuant [23], MaxQuant [24,
25], and Census [26, 27].
10. In order to assess the robustness of each quantitative peptide,
manual inspection of the XIC may be performed (Fig. 3).
Peptides with poor XIC overap and or low intensity should be
excluded.
180 Kian Kani
A. B.
Good quantitation Poor quantitation
500,000 500,000
450,000 450,000
Light 400,000 Light 400,000
350,000
Intensity
350,000
Intensity
Scans: 1172 - 1215 Scans: 1172 - 1215
300,000 300,000
Mass: 1122.63 Mass: 1122.63
250,000 250,000
Area: 1,300,000 Area: 75,000
200,000 200,000
150,000 150,000
100,000 100,000
50,000 50,000
1110 1130 1150 1170 1190 1210 1230 1250 1290 1310 1110 1130 1150 1170 1190 1210 1230 1250 1290 1310
Scan Scan
500,000 500,000
450,000 450,000
Heavy 400,000 Heavy 400,000
Intensity
Intensity
Scans: 1172 - 1215 350,000
Mass: 1128.65 300,000 300,000
Mass: 1128.65
Area: 1,360,000 250,000 250,000
Area: 890,000
200,000 200,000
150,000 150,000
100,000 100,000
50,000 50,000
1110 1130 1150 1170 1190 1210 1230 1250 1290 1310 1110 1130 1150 1170 1190 1210 1230 1250 1290 1310
Scan Scan
500,000 500,000
450,000 450,000
Combined 400,000 400,000
Combined
Intensity
350,000
Intensity
H to L ratio: 1.05:1 350,000
300,000 H to L ratio: 11.9:1 300,000
L to H ratio: 0.95:1 250,000 L to H ratio: 0.08:1 250,000
200,000 200,000
150,000 150,000
100,000 100,000
50,000 50,000
1110 1130 1150 1170 1190 1210 1230 1250 1290 1310 1110 1130 1150 1170 1190 1210 1230 1250 1290 1310
Scan Scan
Fig. 3 (a) Accurate quantitation is obtained when the elution profile of the light and heavy peptide overlap in
time but differ in m/z. (b) In this example, the elution profile of the heavy and light peptide show traces of a
secondary peak which results in poor overlap of the XIC and unreliable quantitation
11. The final ratio for the protein expression reflects the expression
change as a function of Iressa dosing. The SILAC ratio of any
given protein can be determined by averaging the area of the
“light” and “heavy” peptides. The statistical significance of the
SILAC ratio must include the following parameters:
(a) The number of quantitative peptides (at least two).
(b) The intensity of the quantitative peptides should be greater
than 1000.
(c) The mean ratio for all peptides from this protein.
(d) The standard deviation (lower the better).
(e) Error determined by the student’s t-test (less than 0.05).
12. Lists of proteins may be exported to a suitable spreadsheet for
further analysis.
13. The initial step is to convert the “heavy” and “light” fold
changes into a log2 scale (Fig. 4a).
14. In order to compare results from different proteomic experi-
ments (reciprocal mixing) and eliminate to systematic errors in
the mixing step the data must be normalized using the median
of the log2 ratio. This may be achieved by dividing the ratio of
each protein by the median value.
Quantitative Proteomics Using SILAC 181
A. 8.0
4.0
0.0
2.0
4.0
8.0
Proteins
B.
# Pep. Avg H/L p value # Pep. Avg H/L p value
C.
1000 nM
100 nM
500 nM
CTRL
p-EGFR-Y1068
CBFB
Actin
16 hours (Iressa)
Fig. 4 (a) Distribution of protein fold change for the heavy treated sample. (b)
Example of several experimental parameters used to determine robustness of
data. (c) Western blot validation of protein fold change with loading control
4 Notes
References
1. Ong SE, Blagoev B, Kratchmarova I, ments with embryonic stem cells. Mol Cell
Kristensen DB, Steen H, Pandey A, Mann M Proteomics 7(9):1587–1597. doi:10.1074/
(2002) Stable isotope labeling by amino acids mcp.M800113-MCP200, M800113-
in cell culture, SILAC, as a simple and accurate MCP200 [pii]
approach to expression proteomics. Mol Cell 9. de Godoy LM, Olsen JV, de Souza GA, Li G,
Proteomics 1(5):376–386 Mortensen P, Mann M (2006) Status of com-
2. Beck S, Michalski A, Raether O, Lubeck M, plete proteome analysis by mass spectrom-
Kaspar S, Goedecke N, Baessmann C, Hornburg etry: SILAC labeled yeast as a model system.
D, Meier F, Paron I, Kulak NA, Cox J, Mann Genome Biol 7(6):R50. doi:10.1186/gb-
M (2015) The impact II, a very high-resolution 2006-7-6-r50, gb-2006-7-6-r50 [pii]
quadrupole time-of-flight instrument (QTOF) 10. Gruhler A, Schulze WX, Matthiesen R, Mann
for deep shotgun proteomics. Mol Cell M, Jensen ON (2005) Stable isotope label-
Proteomics 14(7):2014–2029. doi:10.1074/ ing of Arabidopsis thaliana cells and quantita-
mcp.M114.047407, M114.047407 [pii] tive proteomics by mass spectrometry. Mol Cell
3. Zhang G, Fenyo D, Neubert TA (2009) Proteomics 4(11):1697–1709. d oi:10.1074/mcp.
Evaluation of the variation in sample prepa- M500190-MCP200, M500190-MCP200 [pii]
ration for comparative proteomics using 11. Gruhler A, Olsen JV, Mohammed S, Mortensen
stable isotope labeling by amino acids in cell P, Faergeman NJ, Mann M, Jensen ON (2005)
culture. J Proteome Res 8(3):1285–1292. Quantitative phosphoproteomics applied to the
doi:10.1021/pr8006107 yeast pheromone signaling pathway. Mol Cell
4. Bantscheff M, Schirle M, Sweetman G, Rick J, Proteomics 4(3):310–327. doi:10.1074/mcp.
Kuster B (2007) Quantitative mass spectrome- M400219-MCP200, M400219-MCP200 [pii]
try in proteomics: a critical review. Anal Bioanal 12. Bose R, Molina H, Patterson AS, Bitok JK,
Chem 389(4):1017–1031. doi:10.1007/ Periaswamy B, Bader JS, Pandey A, Cole PA
s00216-007-1486-6 (2006) Phosphoproteomic analysis of Her2/
5. Tzouros M, Golling S, Avila D, Lamerz J, neu signaling and inhibition. Proc Natl Acad
Berrera M, Ebeling M, Langen H, Augustin Sci U S A 103(26):9773–9778. doi:10.1073/
A (2013) Development of a 5-plex SILAC pnas.0603948103, 0603948103 [pii]
method tuned for the quantitation of tyro- 13. Boersema PJ, Geiger T, Wisniewski JR, Mann
sine phosphorylation dynamics. Mol Cell M (2013) Quantification of the N-glycosylated
Proteomics 12(11):3339–3349. doi:10.1074/ secretome by super-SILAC during breast cancer
mcp.O113.027342, O113.027342 [pii] progression and in human blood samples. Mol
6. Hilger M, Mann M (2012) Triple SILAC to Cell Proteomics 12(1):158–171. doi:10.1074/
determine stimulus specific interactions in the mcp.M112.023614, M112.023614 [pii]
Wnt pathway. J Proteome Res 11(2):982–994. 14. Dhungana S, Merrick BA, Tomer KB, Fessler
doi:10.1021/pr200740a MB (2009) Quantitative proteomics analy-
7. Van Hoof D, Pinkse MW, Oostwaard DW, sis of macrophage rafts reveals compartmen-
Mummery CL, Heck AJ, Krijgsveld J (2007) talized activation of the proteasome and of
An experimental correction for arginine- proteasome- mediated ERK activation in
to-
proline conversion artifacts in SILAC- response to lipopolysaccharide. Mol Cell
based quantitative proteomics. Nat Methods Proteomics 8(1):201–213. doi:10.1074/mcp.
4(9):677–678. doi:10.1038/nmeth0907-677, M800286-MCP200, M800286-MCP200 [pii]
nmeth0907-677 [pii] 15. Ong SE, Mittler G, Mann M (2004) Identifying
8. Bendall SC, Hughes C, Stewart MH, Doble and quantifying in vivo methylation sites by
B, Bhatia M, Lajoie GA (2008) Prevention heavy methyl SILAC. Nat Methods 1(2):119–
of amino acid conversion in SILAC experi- 126. doi:10.1038/nmeth715, nmeth715 [pii]
184 Kian Kani
Abstract
Measuring protein changes over time or following stimuli is one of the important tasks of proteomics. In
the past decade, several strategies have been developed for the relative quantification of proteins using mass
spectrometry (MS). Isobaric labeling strategies for relative quantitative proteomics allow for parallel mul-
tiplexing of quantitative experiments. With this technique, multiple peptide samples are chemically labeled
with isobaric chemical tag variants and each variant has the same molecular structure and mass. Each vari-
ant, however, is designed to produce a unique “reporter ion” when fragmented inside a mass spectrometer.
Once peptide samples are labeled, combined, and analyzed using MS, differentially labeled peptides are
indistinguishable in a first, MS spectrum of intact peptides. However, since each tag variant contains a
labile component with different mass, “reporter ions” can be generated and recorded in a subsequent MS2
spectrum. Intensities from each variant are recorded to represent the relative abundances of the peptide in
each sample. Isobaric tags for relative and absolute quantitation (iTRAQ) and tandem mass tags (TMT)
are commercially available reagents for performing this technique. Here, we describe the general workflow
of relative quantification of proteins using TMT by MS2, or an additional MS3 spectrum.
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_14, © Springer Science+Business Media LLC 2017
185
186 Lichao Zhang and Joshua E. Elias
Fig. 1 Structure of TMT reagents. (a) TMT reagent contains mass reporter, mass
normalizer and amine reactive group. A HCD cleavable linker is between the
mass reporter and mass normalizer. (b) 10plex TMT. Heavy labeled C12 and N13
are indicated by asterisks. Figures are adapted from Thermo Scientific instruc-
tions for TMT10plex mass tag labeling kits and reagents
Fig. 2 Workflow for relative quantification using 10plex TMT. Ten different protein samples are denatured,
reduced, alkylated and digested into peptide samples. These peptide samples are then labeled with 10plex
TMT reagent kit and then combined into one sample. The combined sample is fractionated using HPRP liquid
chromatography and the fractions are concatenated into 12 samples. After cleaning up, each sample is sub-
jected to HPLC-MS analysis and the data are then analyzed to obtain the identity and relative quantification of
proteins
2 Materials
2.3 Sample 1. Sep-Pak C18 1 cc vac cartridge, 50 mg sorbent per cartridge
Purification (Waters, Milford, MA).
and Fractionation 2. Vacuum manifold (Varian, Palo Alto, CA).
3. Vortex mixer.
TMT Quantitation Mass Spectrometry 189
Fig. 3 MS analysis of 10plex TMT samples for peptide sequencing and quantification. Top figure shows a typi-
cal MS2 HCD spectrum. Bottom left figure shows the zoomed spectrum of the reporter ions region. Theoretical
HCD monoisotopic reporter ion masses for 10plex TMT reagent are listed at bottom right
4. Centrifuge.
5. Methanol.
6. Acetonitrile.
7. Acetic acid, glacial.
8. Ammonium formate for HPLC ≥ 99.0.
9. Ammonium hydroxide (Fisher Scientific, Fair Lawn, NJ).
10. Centrivap concentrator (Labconco, Kansas City, MO).
190 Lichao Zhang and Joshua E. Elias
11. Empore™ SPE Disks, C18, 47 mm (Sigma-Aldrich, St. Louis, MO).
12. ZORBAX Extend-C18 column (Agilent technologies, Santa
Clara, CA).
13. HPLC binary pump.
3 Methods
with 1 mL of 0.5 % acetic acid (three times) and elute the pep-
tides with 1 mL of 40 % ACN in 0.5 % acetic acid. The eluted
samples are dried in a centrivap concentrator and stored in
−20 °C (see Note 6).
3.3 Sample 1. Combine 1 % of each labeled sample and dry down in a cen-
Preparation for Ratio trivap concentrator.
and Labeling 2. Dry down the rest of the labeled samples and store in −20 °C.
Efficiency Check
3. Purify the combined sample using STAGE-Tip [7]. Cut out
(See Note 10) four cookies (2 mm o.d.) from the Empore C18 Disks and
pack into a P200 pipette tip. The STAGE-Tip column is con-
ditioned and equilibrated with 80 μL of methanol, 40 μL of
40 % ACN in 5 % formic acid, 40 μL of 5 % formic acid (two
times). Resuspend the dried sample in 40 μL of 0.1 % formic
acid. Add the sample, wash with 40 μL of 5 % formic acid (two
times) and elute the peptides with 40 μL of 40 % ACN in 5 %
formic acid. The eluted samples are dried in a centrivap con-
centrator and stored in −20 °C.
4. Perform HPLC-MS analysis described in Subheading 3.5 and
data analysis in Subheading 3.6.
5. Verify if the reporter ion distributions follow desired overall
ratio (e.g., 1:1:1:1:1:1:1:1:1:1). Calculate sample ratio adjust-
ments, if necessary (see Note 10).
3.4 Sample 1. Resuspend the samples in 50 μL of 0.1 % formic acid and combine
Purification the samples with the adjusted ratio (Subheading 3.3, step 5).
and Fractionation 2. Purify the combined sample using Sep-Pak.
See Note 11)
192 Lichao Zhang and Joshua E. Elias
3.5 HPLC-MS 1. Prepare the following buffers: buffer A is 0.2 % formic acid in
Analysis water; buffer B is 0.2 % formic acid in ACN.
2. Peptides are separated on a 100 μm inner diameter microcapil-
lary column. The tip of the column is pulled in-house with
laser puller and the column is packed with Reprosil C18 resin
(0.5 cm of 5 μm resin close to the tip followed by 18 cm of
3 μm resin). There are also commercially available nano-flow
C18 columns that can be used.
3. Run the HPLC gradient as follows: 98 % A to 60 % A in
130 min, to 40 % A in 20 min, and then to 2 % A in 18 min.
The total gradient time is 180 min.
4. The TMT reporter ions can be analyzed at either MS2 or MS3
level on a LTQ-Orbitrap instrument (Velos or Elite), and the
“synchronous precursor selection” MS3 method is available with
Obitrap Fusion Tribrid Mass Spectrometer (see Note 12) [5, 6].
(a) MS2 method. The MS1 scans are performed in the Orbitrap
in the mass range of 300–1500 m/z and the resolution is
set to 60,000. Ten most intense ions (intensity above 500
counts) are selected for HCD fragmentation. The precur-
sor isolation width is set to 2 m/z. The AGC settings are
1E6 and 5E4 for FTMS1 and FT MSn scans, respectively.
Maximum ion times for FTMS1 and FTMSn are both
250 ms. The normalized collision energy for HCD is 45 %
at 30 ms activation time. Monoisotopic precursor selection
and charge state rejection are enabled. Singly charged ion
species and ions with no unassigned charge states are
excluded from MS2 analysis. Ions within ±10 ppm m/z
window around ions selected for MS2 are excluded from
further selection for fragmentation for 90 s.
TMT Quantitation Mass Spectrometry 193
4 Notes
Acknowledgments
The authors would like to thank Dr. Lihua Jiang from Stanford
University and Dr. Xiaoyue Jiang from Thermo Scientific for help-
ful discussions about TMT methods on Obitrap Fusion Tribrid
Mass Spectrometer and data processing using Proteome Discoverer.
198 Lichao Zhang and Joshua E. Elias
References
1. Oda Y, Huang K, Cross F, Cowburn D, Chait plexed quantitative proteomics. Nat Methods
B (1999) Accurate quantitation of protein 8:937–940
expression and site-specific phosphorylation. 6. McAlister GC, Nusinow DP, Jedrychowski MP,
Proc Natl Acad Sci U S A 96:6591–6596 Wuehr M, Huttlin EL, Erickson BK, Rad R,
2. Thompson A, Schafer J, Kuhn K, Kienle S, Haas W, Gygi SP (2014) MultiNotch MS3
Schwarz J, Schmidt G, Neumann T, Hamon C enables accurate, sensitive, and multiplexed
(2003) Tandem mass tags: a novel quantifica- detection of differential expression across cancer
tion strategy for comparative analysis of com- cell line proteomes. Anal Chem 86:7150–7158
plex protein mixtures by MS/MS. Anal Chem 7. Rappsilber J, Mann M, Ishihama Y (2007)
75:1895–1904 Protocol for micro-purification, enrichment,
3. Gygi S, Rist B, Gerber S, Turecek F, Gelb M, pre-fractionation and storage of peptides for
Aebersold R (1999) Quantitative analysis of proteomics using StageTips. Nat Protoc
complex protein mixtures using isotope-coded 2:1896–1906
affinity tags. Nat Biotechnol 17:994–999 8. Wang Y, Yang F, Gritsenko MA, Wang Y, Clauss
4. Ross P, Huang Y, Marchese J, Williamson B, T, Liu T, Shen Y, Monroe ME, Lopez-Ferrer
Parker K, Hattan S, Khainovski N, Pillai S, Dey D, Reno T, Moore RJ, Klemke RL, Camp DG
S, Daniels S, Purkayastha S, Juhasz P, Martin S, II, Smith RD (2011) Reversed-phase chroma-
Bartlet-Jones M, He F, Jacobson A, Pappin D tography with multiple fraction concatenation
(2004) Multiplexed protein quantitation in strategy for proteome profiling of human
Saccharomyces cerevisiae using amine-reactive MCF10A cells. Proteomics 11:2019–2026
isobaric tagging reagents. Mol Cell Proteomics 9. Stark G, Stein W, Moore S (1960) Reactions of
3:1154–1169 cyanate present in aqueous urea with amino
5. Ting L, Rad R, Gygi SP, Haas W (2011) MS3 acids and proteins. J Biol Chem
eliminates ratio distortion in isobaric multi- 235:3177–3181
Chapter 15
Abstract
Recent advancements in mass spectrometry (MS) and data analysis software have enabled new strategies
for biological discovery using proteomics. Proteomics has evolved from routine discovery and identifica-
tion of proteins to integrated multi-omics projects relating specific proteins to their genes and metabolites.
Using additional information, such as that contained in biological pathways, has enabled the use of tar-
geted protein quantitation for monitoring fold changes in expression as well as biomarker discovery. Here
we discuss a full proteomic workflow from discovery proteomics on a quadrupole Time-of-Flight (Q-TOF)
MS to targeted proteomics using a triple quadrupole (QQQ) MS. A discovery proteomics workflow
encompassing acquisition of data-dependent proteomics data on a Q-TOF and protein database searching
will be described which uses the protein abundances from identified proteins for subsequent statistical
analysis and pathway visualization. From the active pathways, a protein target list is created for use in a
peptide-based QQQ assay. These peptides are used as surrogates for target protein quantitation. Peptide-
based QQQ assays provide sensitivity and selectivity allowing rapid and robust analysis of large batches of
samples. These quantitative results are then statistically compared and visualized on the original biological
pathways with a more complete coverage of proteins in the studied pathways.
Key words Mass spectrometry, Proteomics, QTOF, QQQ, Time-of-flight, Quadrupole, Proteins,
Informatics, Data-dependent acquisition, Protein quantitation
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_15, © Springer Science+Business Media LLC 2017
199
200 Caroline S. Chu et al.
2 Materials
2.1 Protein Digest 1. Protein Lysate: As a quality assessment sample, we use either a
Reagents Pyrococcus furiosus (Pfu, Agilent p/n 400510) [1, 2] or E. coli
protein lysate (BioRad p/n 163-2110).
2. Ammonium bicarbonate (100 mM): add 100 mL water to
0.7906 g ammonium bicarbonate.
3. Dithiothreitol (DTT, 200 mM) >99 + %: add 1 mL water to
0.031 g DTT in a 1.5 mL Eppendorf tube.
4. Iodoacetamide (IAM, 200 mM) 97 %: add 1 mL water to
0.037 g IAM in a 1.5 mL Eppendorf tube.
5. Trypsin Stock Solution: Sequencing-grade trypsin and resus-
pension buffer (Agilent p/n 204310): Prepare a 1 μg/μL solu-
tion in a) 50 mM acetic acid if not all the solution will be used
(store remainder in freezer) or b) 50 mM ammonium bicar-
bonate if it will all be used immediately.
2.3 Discovery 1. Agilent 1260 Infinity HPLC-Chip/MS for use with all Agilent
and Targeted 6000 series Mass Spectrometers.
Proteomics Nanoflow 2. Q-TOF Internal Reference Mass (IRM) Low Mass Solution
HPLC-Chip Mass (1000 μg/mL methyl stearate in acetonitrile): Put 10 mg
Spectrometry methyl stearate into a 15 mL Falcon conical tube, add 10 mL
of acetonitrile, vortex to get methyl stearate into solution.
202 Caroline S. Chu et al.
Table 1
60 min Gradient with the Agilent 1290 UPLC
LC conditions
Column AdvanceBio Peptide Mapping, 2.1 × 250 mm, 2.7 μm (Agilent p/n
651750-902)
Column temperature 50 °C
Injection volume 20 μL
Autosampler temp 4 °C
Needle wash 10 s in wash port (50:50 water:methanol with 0.1 % formic acid)
Mobile phase A = 0.1 % formic acid in water
B = 0.1 % formic acid in 90 % acetonitrile in water
Flow rate 0.40 mL/min
Gradient program Time, mins %B
0.0 3
52.0 35
55.0 70
57.0 70
58.0 3
Stop time 60.0 min
Post time 5.0 min
Table 2
90 min Gradient with the Agilent 1290 UPLC
LC conditions
Column AdvanceBio Peptide Mapping, 2.1 × 250 mm, 2.7 μm (Agilent p/n
651750-902)
Column temperature 50 °C
Injection volume 20 μL
Autosampler temp 4 °C
Needle wash 10 s in wash port (50:50 water:methanol with 0.1 % formic acid)
Mobile phase A = 0.1 % formic acid in water
B = 0.1 % formic acid in 90 % acetonitrile in water
Flow rate 0.40 mL/min
Gradient program Time, mins %B
0.0 3
82.0 35
85.0 70
87.0 70
88.0 3
Stop time 90.0 min
Post time 5.0 min
Pathway-Informed Discovery and Targeted Proteomic Workflows… 203
Table 3
120 min Gradient with the Agilent 1290 UPLC
LC conditions
Column AdvanceBio Peptide Mapping, 2.1 × 250 mm, 2.7 μm (Agilent p/n
651750-902)
Column temperature 50 °C
Injection volume 20 μL
Autosampler temp 4 °C
Needle wash 10 s in wash port (50:50 water:methanol with 0.1 % formic acid)
Mobile phase A = 0.1 % formic acid in water
B = 0.1 % formic acid in 90 % acetonitrile in water
Flow rate 0.40 mL/min
Gradient program Time, min %B
0.0 3
110.0 40
115.0 70
117.5 70
118.0 3
Stop time 120.0 min
Post time 5.0 min
Table 4
150 min GRADIENT with the Agilent 1290 UPLC
LC conditions
Column AdvanceBio Peptide Mapping, 2.1 × 250 mm, 2.7 μm (Agilent p/n
651750-902)
Column temperature 50 °C
Injection volume 20 μL
Autosampler temp 4 °C
Needle wash 10 s in wash port (50:50 water:methanol with 0.1 % formic acid)
Mobile phase A = 0.1 % formic acid in water
B = 0.1 % formic acid in 90 % acetonitrile in water
Flow rate 0.40 mL/min
Gradient program Time, mins %B
0.0 3
140.0 40
145.0 70
147.5 70
148.0 3
Stop time 150.0 min
Post time 5.0 min
204 Caroline S. Chu et al.
2.5 Mass 1.
ESI-L Low Concentration Tuning Mix (Agilent p/n
Spectrometry: 6500 G1969-8500) (for Dual ESI or ESI with Jet Stream Technology).
Series Q-TOF and 6400 2. Glass Calibrant Delivery System (CDS) bottle (Agilent p/n
Series QQQ 9300-2576) and cap (Agilent p/n 9300-2575).
3. Tuning and calibrating with the ESI with Jet Stream Technology
on the 6500 series iFunnel Q-TOFs in positive ion mode, add
10 mL ESI-L to a clean CDS bottle, add 88.5 mL acetonitrile,
1.5 mL water, and 5 μL of 0.1 mM HP-0.321 (included in the
Biopolymer Reference Mass Kit (Agilent p/n G1969-85003).
Mix thoroughly and place in position Bottle B in the CDS.
4. Tuning and calibrating with the dual ESI on the 6500 series
iFunnel Q-TOFs in positive ion mode, add 25 mL ESI-L to a
clean CDS bottle, add 71.25 mL acetonitrile, and 3.75 mL
water. Mix thoroughly and place in position Bottle B in the
CDS.
Pathway-Informed Discovery and Targeted Proteomic Workflows… 205
5. Internal Reference Mass Solution for the ESI with Jet Stream
Technology on the 6500 Series Q-TOFs; ES-TOF Reference
Mass Solution Kit (Agilent p/n G1969-85001) containing
two ampoules (2.2 mL/ampoule) of the following reference
ions: 100 mM ammonium trifluoracetate (TFA-NH4) in 90:10
acetonitrile:water, 5 mM purine in 90:10 acetonitrile:water,
and 2.5 mM hexakis(1H, 1H, 3H-tetrafluoropropoxy)phosp-
hazine (HP-0921) in 90:10 acetonitrile:water (see Note 1). To
a 1 L Nalgene bottle, add 950 mL acetonitrile, 50 mL water,
0.4 mL purine, and 1.0 mL HP-0921. Cap and invert the bot-
tle several times to mix the reference solution. Pour 100 mL
into a CDS bottle and place onto Bottle A in the
CDS. Alternatively an isocratic pump can be used with the 1 L
stock solution with a 1:100 splitter (Agilent p/n G1607-
60000) connected to the reference nebulizer.
6. Internal Reference Mass Solution for the ESI on the 6500
Series Q-TOFs; ES-TOF Reference Mass Solution Kit (Agilent
p/n G1969-85001) containing two ampoules (2.2 mL/
ampoule) of the following reference ions: 100 mM TFA-NH4
in 90:10 acetonitrile:water, 5 mM purine in 90:10
acetonitrile:water, and 2.5 mM HP-0921 in 90:10
acetonitrile:water. To a 1 L Nalgene bottle, add 950 mL ace-
tonitrile, 50 mL water, 0.5 mL TFA-NH4, 1.0 mL purine, and
0.45 mL HP-0921 (see Note 2) Tuning and calibrating with
the ESI with Jet Stream Technology and ESI on the 6400
series QQQs in positive ion mode, add 100 mL ESI-L to a
clean CDS bottle and place in position Bottle B in the CDS.
3 Methods
3.1 Protein Digestion 1. Dissolve sample in 50 % TFE in 50 mM ammonium bicarbon-
ate buffer to yield a 1.35 mg/mL solution (see Note 5) For
larger amounts of sample (such as the E. coli lysate), aliquot
100 μL per tube into 1.5 mL Eppendorf tubes (E. coli sample
yields 18 tubes; 135 μg per tube).
2. Add 2.5 μL DTT stock solution (200 mM) to each tube and
vortex to mix. Heat at 60 °C for 45 min.
3. Add 10 μL IAM stock solution (200 mM). Vortex briefly.
Allow to stand at room temperature for 45 min in the dark
(foil covered rack).
4. Add 2.5 μL DTT stock solution (200 mM) to remove excess
IAM. Allow to stand at room temperature for 30 min in the
dark.
5. Add 600 μL water and 200 μL ammonium bicarbonate to each
vial (Note: pre-mix this to reduce pipetting).
6. Add 6 μL trypsin stock solution at 1:20 or 1:50 enzyme:substrate.
Vortex briefly. Incubate overnight at 37 °C.
7. Add 4 μL neat formic acid or TFA to stop trypsin activity.
Vortex briefly. Digest is ready to analyze by LC/MS. Final
concentration should be ~150 ng/μL. For larger amounts of
sample, such as the quality assessment samples, it is convenient
208 Caroline S. Chu et al.
to mix all the digest vials together (15 mL Falcon tube), then
aliquot to Eppendorf tubes (100 μL per vial) and store at
−80 °C (or −20 °C if you don’t have a −80 °C freezer).
3.2 Discovery 1. With the ESI with Jet Stream Technology source [3] on the
Proteomics: Data- Q-TOF, in MassHunter Acquisition, go to the Tune page.
Dependent Acquisition Tune and calibrate the Q-TOF in the Extended Mass Range,
(DDA) on an Agilent 2 GHz at mass range m/z 3200.
6500 Series Q-TOF 2. Select Quadrupole Tune and perform a Quad Tune.
3. Once the tune is completed, change the mass range to m/z
1700 and select “Apply”.
4. Once the 20 min equilibration is complete, select m/z 118 to
m/z 1622, and select calibrate the TOF.
5. Save the Tune File before switching back to acquisition
window.
6. In the acquisition window, generate the following acquisition
method for standard flow with the ESI with Jet Stream
Technology source:
Liquid Chromatography Gradient: Table 2.
Source: Table 5.
MS Acquisition: Table 6.
7. Save method as “DDA_90min_AJS.m”.
8. In the acquisition window, generate the following acquisition
method for nanoflow with the HPLC-Chip source [4, 5]:
Liquid Chromatography Gradient: Table 7.
Table 5
Source conditions for the 6500 series QTOF with the ESI with Agilent Jet
Stream source
Table 6
Discovery proteomics acquisition method for the 6500 series QTOF
Parameter Setting
Acquisition mode Extended dynamic range (2 GHz), high sensitivity, low mass range
m/z 1700
Mass range m/z 300–1700
Acquisition rate/time 8 spectra/s
Auto MSMS range m/z 50-1700
MSMS acquisition rate/time 3 spectra/s (max)
Isolation width Narrow (~1.3 Hz)
Precursors/cycle Top 20
Collision energy 3.6*(m/z)/100–4.8
Threshold for MSMS 1000 counts and 0.001
Dynamic exclusion On; 1 repeat then exclude for 0.2 min
Precursor abundance-based scan Yes
speed
Target 25,000
Use MS/MS accumulation time Yes
limit
Purity 100 % stringency, 30 % cutoff
Isotope model Peptides
Sort precursors By abundance only; +2, +3, > + 3
Source: Table 8.
MS Acquisition: Table 9.
9. Save method as “DDA_130min_Chip.m.
3.3 Discovery 1. Move your DDA data from the QTOF to a new folder under
Proteomics: Data the “smdata” folder on your Spectrum Mill server. Typically
Analysis Using data is organized in subfolders under smdata to organize proj-
Spectrum Mill (Fig. 3) ects (see Note 6).
2. Open Internet Explorer and load Spectrum Mill, navigate to
the Data Extractor page.
3. In the Data Directories section, click the Select… button to
select the folder or folders that contain your files.
4. For Agilent Q-TOF data and most extractions, adjust the
parameters as outlined below:
●● Click the Choose… button to select the
“Carbamidomethylation (C) “modification”.
210 Caroline S. Chu et al.
Table 7
Source and LC gradient conditions for the 6500 series QTOF with the HPLC Chip
Table 8
Source conditions for the 6500 series QTOF with HPLC Chipcube
Table 9
Discovery proteomics acquisition method for the 6500 series QTOF with HPLC-Chip
Parameter Setting
Acquisition mode Extended Dynamic Range (2GHz), High Sensitivity, Low Mass Range m/z
1700
Mass range m/z 275–1700 (MS) and 50–1700 (MS/MS)
Acquisition rate/time 8 spectra/s
Auto MSMS range m/z 50–1700
MSMS acquisition rate/ 3 spectra/s (max)
time
Isolation width Narrow (~1.3 Hz)
Precursors/cycle Top 20 precursors per cycle using precursor abundance-based acquisition
rate with accumulation time limit enabled; active exclusion after one
spectrum for 0.5 min
Collision energy 3.6*(m/z)/100–4.8
Threshold for MSMS 1000 counts and 0.001
Dynamic Exclusion On; 1 repeat then exclude for 0.2 min
Precursor abundance- Yes
based scan speed
Target 25,000
Use MS/MS Yes
accumulation time
limit
Purity 100 % stringency, 30 % cutoff
Isotope model Peptides
Sort precursors By abundance only; +2, +3, > + 3
3.4 Discovery Data Analysis Using Mass Profiler Professional (MPP) to perform
and Targeted statistical and correlation analysis between samples groups. Using
Proteomics the optional Pathway Architect module allows the results to be
mapped to publically available biological pathways databases along
with metabolomics and gene data, if available.
1. Open MPP and create a new project.
2. Create a New Experiment within MPP.
Pathway-Informed Discovery and Targeted Proteomic Workflows… 215
Binary pump:
Gradient: Time %B
0 5
1 25
1.5 70
1.6 5
Table 10
HSA in BSA dilution table for 6400 Series QQQ checkout
Sample Final concentration Volume of HSA standard Volume of BSA dilution solution (μL)
A 100 fmol/μL 10 μL of HSA Stock 90
B 10 fmol/μL 10 μL of A 90
C 1 fmol/μL 10 μL of B 90
D 100 amol/μL 10 μL of C 90
E 10 amol/μL 10 μL of D 90
F 5 amol/μL 20 μL of E 20
G 2 amol/μL 10 μL of E 40
H 1 amol/μL 10 μL of E 90
Capillary pump:
HPLC-Chip MS Interface:
QQQ MRM:
ume. For the sample type, use “Blank” for the BSA blank and
“Calibration” for the HSA spiked into BSA. For the Level for
each sample enter 1 for Sample H and go up in increments of
1 till you get to Sample A with at Level 8 (Table 10).
3.7 Targeted 1. For initial performance checkout of the QQQ for daily or
Proteomics: monthly assessment of the LC/MS platform, use the
Commercial Kits PeptiQuant™ LC/MS Platform Performance kit. Each
for Quality Control Platform Performance kit includes seven tryptic-digested
Using Standard Flow plasma samples spiked with stable-isotope labeled standards
on the 6400 Series (SIS). These samples are rehydrated and ready for use with the
QQQ [6, 7, 8] LC/MRM/MS. Protocols are included within each kit.
2. For quality control of the bottom-up workflow (from denatur-
ation through to detection), use the PeptiQuant™ MRM-MS
Workflow Performance Kit. Each Workflow Performance kit
contains: human plasma, trypsin, and a SIS peptide mixture.
The protocol is included with the kit providing the detailed
procedure for the sample preparation. The protocol describes
the sequential reduction, alkylation, and quenching of the 10×
diluted plasma prior to digestion overnight with trypsin.
Digestion is stopped by the addition of the chilled SIS peptide
mixture from 250 to 0.025 fmol/μL for standard samples G to
A (A being the lowest concentration) and chilled formic acid
solution. The samples are allocated into separate Eppendorf
tubes, centrifuged, and the peptide supernatant is removed for
desalting. The desalted supernatant is lyophilized and rehy-
drated in 0.1 % formic acid for a final concentration of 1 μg/μL
for LC-MRM/MS analysis.
4 Notes
1. Before you break open each ampoule, invert the ampoule sev-
eral times to mix. Inspect the ampoule’s contents to ensure
that all the solution is contained in the lower cylindrical base.
Shake the ampoule, if needed, to dislodge any air pocket that
may prevent solution from settling in the lower portion of the
ampoule.
2. Cap and invert the bottle several times to mix the reference
solution. Pour 100 mL into a CDS bottle and place onto
Bottle A in the CDS. Alternatively an isocratic pump can be
used with the 1 L stock solution with a 1:100 splitter (Agilent
p/n G1607-60000) connected to the reference nebulizer.
3. System cleanliness is the biggest challenge as this peptide will
exhibit some carryover.
Pathway-Informed Discovery and Targeted Proteomic Workflows… 221
References
1. Wong CC, Cociorva D, Miller CA et al (2013) proteomics results. Agilent Technical Overview
Proteomics of Pyrococcus Furiosus (Pfu): identifi- 5991-0735EN http://www.agilent.com/cs/
cation of extracted proteins by three indepen- librar y/technicalover views/Public/5991-
dent methods. J Proteome Res 12(2):763–770 0735EN.pdf Accessed 31 August 2015
2. Vaudel M, Burkhart JM, Breiter D et al (2012) 6. Percy AJ, Chambers AG, Borchers CH, (2014)
A complex standard for protein identification, Application kits for standardizing MRM-based
designed by evolution. J Proteome Res quantitative plasma proteomic workflows on
11(10):5065–5071 the Agilent 6490 LC/MS system. Agilent
3. Yang Y, Bhat V, Miller CA. (2015) Jet Stream Application Note 5991-3601EN https://
proteomics for sensitive and robust standard w w w. a g i l e n t . c o m / c s / l i b r a r y /
flow LC/MS. Agilent Technical Overview applications/5991-3601EN.pdf Accessed 31
5991-5687EN http://www.agilent.com/cs/ August 2015
librar y/technicalover views/public/5991- 7. Percy AJ, Chambers AG, Yang J et al (2012)
5687EN.pdf Accessed 31 August 2015 Comparison of standard-and nano-flow liquid
4. Miller CA, Jenkins S, Sana TR, et al. (2013) chromatography platforms for MRM-based
Proteomics in multi-omics workflows using quantitation of putative plasma biomarker
yeast as a model system. Agilent Application proteins. Anal Bioanal Chem
Note 5991-2484EN https://www.agilent. 404(4):1089–1101
com/cs/library/applications/5991-2484EN. 8. Percy AJ, Mohammed Y, Yang J, Borchers CH
pdf 31 August 2015 (2015) A standardized kit for automated
5. Buckenmaier S, Mora J, van de GoorT, et al. (2012) quantitative assessment of candidate protein
Enhanced chromatography with the Agilent biomarkers in human plasma. Bioanalysis,
Polaris-HR-Chip-3C18 improved LC/MS/MS 7(23):2991–3004
Chapter 16
Abstract
Data-independent acquisition is a powerful mass spectrometry technique that enables comprehensive MS
and MS/MS analysis of all detectable species, providing an information rich data file that can be mined
deeply. Here, we describe how to acquire high-quality SWATH® Acquisition data to be used for large
quantitative proteomic studies. We specifically focus on using variable sized Q1 windows for acquisition of
MS/MS data for generating higher specificity quantitative data.
Key words Mass spectrometry, SWATH acquisition, Data-independent acquisitions, Variable win-
dows, Quantitation, Proteomics
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_16, © Springer Science+Business Media LLC 2017
223
224 Birgit Schilling et al.
2 Materials
2.1 Samples 1. Retention Time Calibration Peptides: The iRT peptides (kit
for Analysis P/N Ki-3003, Biognosys, Switzerland) are spiked into com-
plex samples, typically at dilutions of 1/10 (see Note 1).
2. LC-MS quality control (QC) sample: Beta-Galactosidase (BGal)
Stock Solution. Add 625 μL of buffer (10 % acetonitrile/0.1 %
formic acid) to the BGal vial (kit P/N 4465867, SCIEX) to cre-
ate a 1 pmol/μL stock solution. Aliquot the stock solution into
10–50 μL volumes and freeze at −20 °C for future use.
3. SWATH system suitability sample: pre-digested cell lysate simi-
lar to the study sample described under step 4 to assess SWATH
performance.
4. Pre-digested Cell Lysate samples: a typical study sample used
here consists of a tryptic digestion of a yeast cell lysate.
Proteomic samples were previously reduced, alkylated,
digested, and desalted, providing a solution of ~1 μg/μL
digested protein lysate for analysis.
2.4 Data Processing 1. SWATH® Acquisition data is processed with the SWATH
Acquisition MicroApp 2.0 in PeakView® Software [9], using a
spectral ion library [4] generated from prior data-dependent
acquisitions (see Note 4).
3 Methods
3.1 Set 1. Using the nanoLC 425 LC system with a cHiPLC System
Up LC-MS System (SCIEX), set up in standard ‘trap elute’ mode, to desalt first
then separate the peptides from the complex tryptic digestions.
Set up the autosampler (AS-3) valve with a 10 μL sample loop,
and connect to the loading pump, and the cHiPLC system.
226 Birgit Schilling et al.
3.2 Create a Basic 1. Click Build Acquisition Method to start a new method. Set
LC-MS experiment 1 to TOF MS and set the m/z range to 400 – 1500
Acquisition Method and the Accumulation Time to 250 msec. Next, open the Edit
Parameters and set the source conditions that were optimized
for the system, these should typically be in the following ranges.
On the Compound tab, ensure the Declustering Potential (DP) is
set to between 80 and 100 V. On the Source/Gas tab, typical
conditions will be 2300–2600 V for the IonSpray Voltage (IVSF),
3-6 for the Ion Source Gas 1 (GS1), 0 for the Ion Source Gas 2
(GS2), 20–25 for the Curtain Gas (CUR), and 100–150 for the
Interface Heater Temperature (IHT).
2. Create a trap loading method for desalting and add this to the
LC-MS acquisition method; for example using a loading pump
flow rate of 0.5 μL/min of 100 % Buffer A for 30 min (see Note 6).
3. Create a μL-pickup method for the autosampler, injecting 2 μL
of sample using a 10 μL loop and add this to the LC-MS acqui-
sition method.
4. Create the analytical gradient method and add this to the
LC-MS method. An example gradient for a complex sample is
to use a flow rate of 300 nL/min and the following gradient:
5–30 % (v/v) solvent B with solvent A making up the differ-
ence (from 0 to 120 min), 30–80 % solvent B (from 120 to
130 min), and at 80 % solvent B (from 130 to 136 min) then
back to 5 % solvent B from 136 to 138 min for mobile phase
equilibration, with a total run-time of 155 min (see Note 9).
5. Set the MS acquisition Duration to 150 min. Save the LC-MS
acquisition method.
3.3 Convert 1. Click on the Create SWATH Exp button in the top right hand
to a SWATH side of the LC-MS acquisition method window. The Create
Acquisition Method SWATH Experiment UI appears (Fig. 1). Click on the ‘Manual’
Tab of the UI on the right to set up a custom SWATH
experiment.
SWATH Acquisitions Using TripleTOF Mass Spectrometers 227
Fig. 1 Build a Variable Window SWATH Method. Any window strategy can be constructed in text file format and
loaded into the SWATH Acquisition method editor for automatic method building. (a) A text file describing the
Q1 isolation window strategy can be constructed. (b) Load into SWATH Acquisition method editor and adjust
the MS and MS/MS settings as shown. (c) Click OK to automatically build a SWATH Acquisition method
Prepare the SWATH Variable Window text file that defines the
Q1 isolation window strategy as shown in Fig. 1a (see Note 10).
Generate the Variable Window Table manually in Excel or
compute from previously acquired MS data on the specific
228 Birgit Schilling et al.
3.4 Creating 1. In order to obtain consistent and reliable data from study sam-
a Practical SWATH ples, regular LC-MS tests need to be performed before and
Study during the entire SWATH study. Initially, use the pre-digested
Acquisition Batch BGal standard (see Note 5) and perform QC acquisitions, or
SWATH Acquisitions Using TripleTOF Mass Spectrometers 229
3.5 Data Processing Once the SWATH data is acquired, several data processing pipe-
lines can be used to process the data (see Note 4). As an example,
a SWATH data set was acquired using the above methods and a
number of different Q1 window strategies, then processed using a
spectral library approach as previously described for the SWATH
2.0 algorithm by Lambert et al. [9], also referred to as targeted
extraction. Figure 2 highlights that higher numbers of peptides are
robustly identified and quantified when more, smaller (variable)
Q1 windows are used for SWATH acquisitions [6].
4 Notes
25000
15000
100 VW
10000
80 VW
60 VW
5000
0
0 10 20 30 40 50
% CV
Fig. 3 Variable Q1 Window Widths for SWATH Acquisition. To achieve better spec-
ificity in complex matrices, smaller Q1 windows are desirable especially in the
m/z dense regions where many peptide precursors are measured. The m/z den-
sity histograms constructed from the TOF MS data for the proteome of interest
(blue line) can be used to construct variable sized windows (red line), where the
density of precursors in each of the isolation windows is equalized across the
m/z range
Acknowledgments
References
1. Collins BC, Gillet LC, Rosenberger G et al tor for creating and analyzing targeted pro-
(2013) Quantifying protein interaction dynam- teomics experiments. Bioinformatics
ics by SWATH mass spectrometry: application 26:966–968
to the 14-3-3 system. Nat Methods 9. Lambert JP, Ivosev G, Couzens AL et al (2013)
10:1246–1253 Mapping differential interactomes by affinity
2. Gillet LC, Navarro P, Tate S et al (2012) purification coupled with data-independent
Targeted data extraction of the MS/MS spec- mass spectrometry acquisition. Nat Methods
tra generated by data-independent acquisition: 10:1239–1245
a new concept for consistent and accurate pro- 10. Bruderer R, Bernhardt OM, Gandhi T et al
teome analysis. Mol Cell Proteomics (2015) Extending the limits of quantitative
11(O111):016717 proteome profiling with data-independent
3. Liu Y, Buil A, Collins BC et al (2015) acquisition and application to acetaminophen-
Quantitative variability of 342 plasma proteins treated three-dimensional liver microtissues.
in a human twin population. Mol Syst Biol Mol Cell Proteomics 14:1400–1410
11:786 11. Wang J, Tucholska M, Knight JD et al (2015)
4. Selevsek N, Chang CY, Gillet LC et al (2015) MSPLIT-DIA: sensitive peptide identification
Reproducible and consistent quantification of for data-independent acquisition. Nat Methods
the Saccharomyces cerevisiae proteome by 12(12):1106–1108, Online
SWATH-mass spectrometry. Mol Cell 12. Tsou CC, Avtonomov D, Larsen B et al (2015)
Proteomics 14:739–749 DIA-Umpire: comprehensive computational
5. Egertson JD, MacLean B, Johnson R et al framework for data-independent acquisition
(2015) Multiplexed peptide analysis using proteomics. Nat Methods 12:258–264, 7 p
data-independent acquisition and Skyline. Nat following 264
Protoc 10:887–903 13. Li Y, Zhong CQ, Xu X et al (2015) Group-
6. Hunter CL, Collins B, Gillet L, et al. (2014) DIA: analyzing multiple data-independent
Increasing depth of coverage in data indepen- acquisition mass spectrometry data files. Nat
dent acquisition with acquisition improve- Methods 12(12):1105–1106, Online
ments and higher sample loads. Proccedings of 14. Rardin MJ, Schilling B, Cheng LY et al (2015)
the 61st Annual ASMS Conference on Mass MS1 peptide ion intensity chromatograms in
Spectrometry & Allied Topics, Baltimore, MD, MS2 (SWATH) data independent acquisitions.
June 15–19, 2014 improving post acquisition analysis of pro-
7. Rost HL, Rosenberger G, Navarro P et al teomic experiments. Mol Cell Proteomics
(2014) OpenSWATH enables automated, tar- 14:2405–2419
geted analysis of data-independent acquisition 15. Zhang Y, Bilbao A, Bruderer T et al (2015)
MS data. Nat Biotechnol 32:219–223 The use of variable Q1 isolation windows
8. MacLean B, Tomazela DM, Shulman N et al improves selectivity in LC-SWATH-MS acqui-
(2010) Skyline: an open source document edi- sition. J Proteome Res 14:4359–4371
Chapter 17
Abstract
In this review we describe a protocol to annotate the effects of missense mutations on proteins, their func-
tions, stability, and binding. For this purpose we present a collection of the most comprehensive databases
which store different types of sequencing data on missense mutations, we discuss their relationships, pos-
sible intersections, and unique features. Next, we suggest an annotation workflow using the state-of-the
art methods and highlight their usability, advantages, and limitations for different cases. Finally, we address
a particularly difficult problem of deciphering the molecular mechanisms of mutations on proteins and
protein complexes to understand the origins and mechanisms of diseases.
1 Introduction
*
Author contributed equally with all other contributors.
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_17, © Springer Science+Business Media LLC 2017
235
236 Minghui Li et al.
2 Materials
3 Methods
Table 1
A summary of databases integrating the data on human genetic variations, mutations, and their
clinical relevance
Table 2
Online resources for exploring the structural, cellular, and genomic context of mutated proteins
Table 3
A summary of online resources for predicting the phenotypic effects of mutations
Table 3
(continued)
Type of output,
method, and energy
Resource Input function URL Reference
FoldX Structure ΔΔG using empirical http://foldxsuite.crg.eu/ [78]
force fields
PoPMuSiC-2.0 Structure ΔΔG using a http://dezyme.com/ [79]
combination of
statistical potentials
and neural
networks
ERIS Structure ΔΔG using physical http://dokhlab.unc.edu/tools/eris/ [80]
force fields with
atomic modeling
CUPSAT Structure ΔΔG using mean http://cupsat.tu-bs.de/ [81]
force atom pair and
torsion angle
potentials
Hunter Structure ΔΔG using http://bioinfo41.weizmann.ac.il/hunter/ [82]
knowledge-based
potentials
MultiMutate Structure ΔΔG using four-body http://www.math.wsu.edu/math/faculty/bkrishna/DT/Mutate/ [83]
scoring functions
AUTO-MUTE Structure ΔΔG using http://proteins.gmu.edu/automute [84]
knowledge-based
Effects of Mutations on Protein-Protein Interactions
potentials
(continued)
241
Table 4
242
(continued)
Type of output,
method, and energy
Resource Input function URL Reference
NeEMO Structure ΔΔG using residue http://protein.bio.unipd.it/neemo/ [85]
Minghui Li et al.
interaction
networks
DUET Structure ΔΔG using an http://bleoberis.bioc.cam.ac.uk/duet/stability [86]
integrated
computational
approach of mCSM
and SDM
MAESTRO Structure ΔΔG using multi http://biwww.che.sbg.ac.at/MAESTRO [87]
agent stability
prediction
I-Mutant3.0 Structure/Sequence ΔΔG using SVMs http://gpcr2.biocomp.unibo.it/cgi/ [88]
predictors/I-Mutant3.0/I-Mutant3.0.cgi
MUPro Structure/ Predicts qualitative http://mupro.proteomics.ics.uci.edu/ [89]
Sequence decrease/increase
of stability using
SVM
iStable Structure/Sequence ΔΔG using SVM http://predictor.nchu.edu.tw/iStable [90]
MuStab Sequence Predicts qualitative http://bioinfo.ggc.org/mustab/ [91]
decrease/increase
of stability using
SVM
iPTREE-STAB Sequence ΔΔG using decision http://bioinformatics.myweb.hinet.net/iptree.htm [92]
tree methods
Effects of Mutations on Protein-Protein Interactions 243
Table 5
A summary of online and software resources for predicting the effects of mutations on protein–
protein binding affinity. Here “ΔΔG” refers to ΔΔGbind. All resources need structure as an input
3.1 Collecting First, we will describe a protocol for extracting clinically relevant
and Integrating mutations from ClinVar, COSMIC, and TCGA databases to fur-
the Data on Human ther analyze them with respect to their clinical and functional
Polymorphisms impacts. Here we use human monomeric Casitas B-lineage lym-
and Mutations phoma c-CBL protein (CBL) as an example [20]. The links and
references to web resources used in this protocol are listed in
Table 1.
1. NCBI variation viewer (http://www.ncbi.nlm.nih.gov/varia-
tion/view/) [21] can be searched by gene name (CBL), Refseq
accession (NP_005179), or Uniprot ID (P22681.2). A snap-
shot of variation viewer shows the genetic variation from
ClinVar in the locus of CBL gene (Fig. 2a). Different variants
in this viewer are depicted as separate tracks below the CBL
gene locus. The ClinVar track displays multiple variants as
boxes with the number of variants listed within each box.
Variants with benign effect are shown using green color,
whereas the purple boxes show pathogenic variants.
2. Additionally, the user can download all mutations from the
COSMIC ftp server in VCF format (CosmicCodingMuts.vcf.gz)
and display these mutations in a separate track in the variation
viewer (choose menu “your data” and select “add track” option).
Note that it is necessary to select the same genome assembly in
both variation viewer and in COSMIC (e.g., GRCh37).
3. Each variant with a valid ClinVar annotation is linked to a cor-
responding dbVar or dbSNP record. Here we will focus on a
single nucleotide variant of CBL gene with the dbSNP acces-
sion rs267606706. As shown in Fig. 2b, it is a missense muta-
tion where nucleotide substitution T->C results in p.Y371H
amino acid substitution. There is clinical evidence that this
variant can cause a Noonan syndrome-like disorder and/or
juvenile myelomonocytic leukemia. The GTR studies cited in
ClinVar show [22] that p.Y371H is a heterozygous germline
substitution.
4. The dbSNP page for accession rs267606706 contains a link to
“3D structure mapping” (found under the “NCBI resources”
section) which points to the SNP3D page where several syn-
onymous and missense variants are mapped onto the CBL pro-
tein structure. By default, this variant is selected, but it is
possible to select other variants for display by clicking on the
“Cn3D selected” button. Additionally, the SNP3D page con-
tains a link “CD” that shows conserved domains from the
CDD database mapped onto this protein [23]. Figure 2c
depicts the structure of CBL using Cn3D with mutated Tyr
residue side chain colored in yellow.
5. As was shown previously using ClinVar, the germline p.Y371H
mutation may be linked to leukemia, however, many cancer
246 Minghui Li et al.
Fig. 2 Identification of clinically relevant mutations in ClinVar, COSMIC, and TCGA. (a) NCBI Variation Viewer
showing the CBL gene locus on chromosome 11; ClinVar and dbSNP data shown as tracks below the gene,
pathogenic mutations are presented as purple squares and closely-located mutations are grouped together in
ClinVar track; (b) one of the pathogenic mutations p.Y371H shown in dbSNP with the corresponding disease
annotation in GTR; (c) a representative structure of CBL protein visualized in SNP3D with Cn3D software,
mutated p.Y371H is shown with the yellow side chain; (d) cBioPortal view of CBL protein with missense muta-
tions mapped onto the corresponding domains; (e) a representative structure visualized with JMol directly in
cBioPortal with all mapped missense mutations shown in green
3.2 Finding Protein Now we consider the domain context of CBL mutations described
Domains in the previous section (p.Y371H and p.C384R) and annotate
and Functional Sites their impacts on protein interactions and signaling pathways. The
Affected by Mutations web resources used in this protocol are listed in Table 2.
1. Evolutionarily conserved sites in a multiple sequence align-
ment usually correspond to functionally important sites and
mutations in these sites can be harmful to protein function. If
a protein of interest has a known PDB structure, conservation
profiles can be downloaded from the PDBsum resource other-
wise ConSurf server [24] can be used. In addition, the CDD
server can offer functional annotations of sites in conserved
protein domains, whereas IBIS server provides locations of
binding sites for different types of binding partners (protein,
small molecule, nucleic acid, ion, and peptides), and facilitates
the mapping of a comprehensive biomolecular interaction net-
work for a given protein query (with or without structure)
[25–27]. Similar binding sites in IBIS are clustered together
based on their sequence and structure conservation.
2. Open IBIS web page and search for 1FBV structure, chain
A. Go to “protein–protein” tab and click on the balloon with
the annotation “RING” domain to display interactions of the
CBL RING domain with other domains/proteins (Fig. 3a).
Binding sites are shown on CBL sequence as triangles and
highly conserved binding sites are shown in red color. In the
list of interaction partners below, the first conserved binding
site cluster is formed between RING domain of CBL and
ubiquitin conjugating enzyme from UBCc family. By clicking
on the plus sign next to “UBCc”, one can see the correspond-
ing binding site, the alignment of similar binding sites found in
different CBL-UBCc complexes. By opening the link to Cn3D
viewer, one can explore the interfaces and binding sites in these
protein complexes (Fig. 3c).
248 Minghui Li et al.
Fig. 3 Analysis of conserved functional and binding sites in mutated proteins using IBIS method. (a) conserved
protein binding site in the RING domain of CBL; (b) interaction graph of CBL protein (represented by 1FBV PDB
structure); (c) visualization of protein interface between CBL and UBCc in Cn3D software. Position 384 in 1FBV
corresponds to position 338 in the full-length PDB sequence
3.3 Assessing If Many methods have been proposed to predict the effects of mis-
Mutations Have sense mutations on proteins, classifying them as damaging or
Damaging or Benign benign. These methods differ in terms of the properties of muta-
Effects on Proteins tions or proteins used during the training procedure as well as in
terms of the algorithms applied for decision-making. For example,
machine learning algorithms train their models to distinguish
known disease-associated from neutral mutations. Other methods
do not explicitly train their models but almost all methods described
in this section exploit the evolutionary conservation assuming that
changes at conserved positions tend to be more deleterious. Besides
sequence conservation various other sequence and structural fea-
tures are used, which may include: changes in physicochemical
properties between wild-type and substituted amino acid, struc-
tural features (mostly solvent accessibility), site mutability in DNA,
and sequence context of the site.
An unbiased testing and comparison of machine learning meth-
ods is obviously an issue since they are usually trained on all avail-
able datasets of mutations and it is difficult to obtain a test set which
would not overlap with training set. There are several experimental
studies on variants in P53, LacI, and ABCA1 proteins which can be
regarded as unbiased test cases. Comparisons of different methods
on these experimental sets reported up to 70 % TPR (True positive
rate) at 10 % FPR (False positive rate) [30]. Models trained to dis-
tinguish Mendelian variants with pronounced deleterious effects
250 Minghui Li et al.
mut WT
DDG fold = DG fold - DG fold (1)
DGbind = G AB - G A - G B (3)
There are very few methods that estimate actual ΔΔGbind values
and these methods require all-atom or at least protein backbone
atom coordinates of a wild-type and/or mutated protein. Some of
the methods use coarse-grained predictors based on statistical or
empirical potentials, others apply molecular mechanics force fields
with different solvation models. For example, the molecular
mechanics Poisson–Boltzmann surface area (MM-PBSA) method
and its derivatives have been shown to yield very good agreement
between predicted and experimental values with correlation coef-
ficients up to 0.69 [41]. For all methods, the right choices of mini-
mization protocols, energy functions, and solvation models are
Effects of Mutations on Protein-Protein Interactions 253
3.6 Assessing Proteins may adopt different conformations along the pathway of
the Changes in Protein a biochemical reaction and their intrinsic flexibility and ability to
Conformations sample alternative conformations are crucial for protein function.
and Hydrogen Bond Mutations might shift the equilibrium between different confor-
Networks Induced mations (Fig. 4) and as a result, the most populated conformation
by Mutations of a mutated protein can be different in structure, stability, and
functional activity from the wild-type conformation. It is extremely
difficult to model structural changes in a protein backbone pro-
duced by mutations and large conformational shifts can be pre-
dicted correctly only in a few cases. In fact, most algorithms
discussed in the previous sections do not account for the backbone
flexibility. If several conformations of the same protein are available
in the structural databank, all of them ideally should be used to
provide a complete picture of dynamical and energetic effects of
mutations [20].
Mutations can either change the global conformation of an
entire molecule or have a localized effect in a small region. In a
recent study of the NFAT5 transcription factor [47], different
mutations from the same DNA-binding loop were analyzed and it
was shown that effects of these mutations on protein dynamics and
DNA binding were drastically different although they were located
very close to each other in sequence and structure. Protein dynam-
ics can be studied through performing molecular dynamics (MD)
simulations using NAMD [2], CHARMM [3] and Amber [48]
MD packages. NAMD, for example, is fast and easy to use; it can
Effects of Mutations on Protein-Protein Interactions 255
Acknowledgements
References
1. Abecasis GR, Altshuler D, Auton A, Brooks variation. Nucleic Acids Res 41(Database
LD, Durbin RM, Gibbs RA, Hurles ME, issue):D936–D941
McVean GA, Genomes Project C (2010) A 11. Kibbe WA, Arze C, Felix V, Mitraka E, Bolton
map of human genome variation from E, Fu G, Mungall CJ, Binder JX, Malone J,
population- scale sequencing. Nature Vasant D, Parkinson H, Schriml LM (2015)
467(7319):1061–1073 Disease Ontology 2015 update: an expanded
2. Phillips JC, Braun R, Wang W, Gumbart J, and updated database of human diseases for
Tajkhorshid E, Villa E, Chipot C, Skeel RD, linking biomedical knowledge through disease
Kale L, Schulten K (2005) Scalable molecular data. Nucleic Acids Res 43(Database
dynamics with NAMD. J Comput Chem issue):D1071–D1078
26(16):1781–1802 12. Ramos EM, Hoffman D, Junkins HA, Maglott
3. Brooks BR, Bruccoleri RE, Olafson BD, States D, Phan L, Sherry ST, Feolo M, Hindorff LA
DJ, Swaminathan S, Karplus M (1983) (2014) Phenotype-Genotype Integrator
Charmm – a program for macromolecular (PheGenI): synthesizing genome-wide associ-
energy, minimization, and dynamics calcula- ation study (GWAS) data with existing
tions. J Comput Chem 4(2):187–217 genomic resources. Eur J Hum Genet
4. Humphrey W, Dalke A, Schulten K (1996) 22(1):144–147
VMD: visual molecular dynamics. J Mol Graph 13. Stenson PD, Ball EV, Mort M, Phillips AD,
14(1):33–38, 27-38 Shiel JA, Thomas NS, Abeysinghe S, Krawczak
5. Pettersen EF, Goddard TD, Huang CC, M, Cooper DN (2003) Human Gene Mutation
Couch GS, Greenblatt DM, Meng EC, Ferrin Database (HGMD): 2003 update. Hum Mutat
TE (2004) UCSF Chimera—a visualization 21(6):577–581
system for exploratory research and analysis. 14. Amberger JS, Bocchini CA, Schiettecatte F,
J Comput Chem 25(13):1605–1612 Scott AF, Hamosh A (2015) OMIM.org:
6. Wang Y, Geer LY, Chappey C, Kans JA, Bryant Online Mendelian Inheritance in Man
SH (2000) Cn3D: sequence and structure (OMIM(R)), an online catalog of human genes
views for Entrez. Trends Biochem Sci and genetic disorders. Nucleic Acids Res
25(6):300–302 43(Database issue):D789–D798
7. Mailman MD, Feolo M, Jin Y, Kimura M, 15. Landrum MJ, Lee JM, Riley GR, Jang W,
Tryka K, Bagoutdinov R, Hao L, Kiang A, Rubinstein WS, Church DM, Maglott DR
Paschall J, Phan L, Popova N, Pretel S, Ziyabari (2014) ClinVar: public archive of relationships
L, Lee M, Shao Y, Wang ZY, Sirotkin K, Ward among sequence variation and human pheno-
M, Kholodov M, Zbicz K, Beck J, Kimelman type. Nucleic Acids Res 42(Database
M, Shevelev S, Preuss D, Yaschenko E, Graeff issue):D980–D985
A, Ostell J, Sherry ST (2007) The NCBI 16. Forbes SA, Bindal N, Bamford S, Cole C, Kok
dbGaP database of genotypes and phenotypes. CY, Beare D, Jia M, Shepherd R, Leung K,
Nat Genet 39(10):1181–1186 Menzies A, Teague JW, Campbell PJ, Stratton
8. Rubinstein WS, Maglott DR, Lee JM, Kattman MR, Futreal PA (2011) COSMIC: mining
BL, Malheiro AJ, Ovetsky M, Hem V, complete cancer genomes in the Catalogue of
Gorelenkov V, Song G, Wallin C, Husain N, Somatic Mutations in Cancer. Nucleic Acids
Chitipiralla S, Katz KS, Hoffman D, Jang W, Res 39(suppl 1):D945–D950
Johnson M, Karmanov F, Ukrainchik A, 17. Weinstein JN, Collisson EA, Mills GB, Shaw
Denisenko M, Fomous C, Hudson K, Ostell KR, Ozenberger BA, Ellrott K, Shmulevich I,
JM (2013) The NIH genetic testing registry: a Sander C, Stuart JM, Cancer Genome Atlas
new, centralized database of genetic tests to Research N (2013) The Cancer Genome Atlas
enable access to comprehensive information Pan-Cancer analysis project. Nat Genet
and improve transparency. Nucleic Acids Res 45(10):1113–1120
41(Database issue):D925–D935 18. Singh A, Olowoyeye A, Baenziger PH, Dantzer
9. Sherry ST, Ward M, Sirotkin K (1999) dbSNP- J, Kann MG, Radivojac P, Heiland R, Mooney
database for single nucleotide polymorphisms SD (2008) MutDB: update on development of
and other classes of minor genetic variation. tools for the biochemical analysis of genetic
Genome Res 9(8):677–679 variation. Nucleic Acids Res 36(Database
10. Lappalainen I, Lopez J, Skipper L, Hefferon T, issue):D815–D819
Spalding JD, Garner J, Chen C, Maguire M, 19. Mottaz A, David FP, Veuthey AL, Yip YL
Corbett M, Zhou G, Paschall J, Ananiev V, (2010) Easy retrieval of single amino-acid
Flicek P, Church DM (2013) DbVar and polymorphisms and phenotype information
DGVa: public archives for genomic structural using SwissVar. Bioinformatics 26(6):851–852
Effects of Mutations on Protein-Protein Interactions 257
20. Li M, Kales SC, Ma K, Shoemaker BA, Crespo- 29. Hernansaiz-Ballesteros RD, Salavert F,
Barreto J, Cangelosi AL, Lipkowitz S, Panchenko Sebastian-Leon P, Aleman A, Medina I,
AR (2015) Balancing protein stability and activ- Dopazo J (2015) Assessing the impact of
ity in cancer: a new approach for identifying mutations found in next generation sequenc-
driver mutations affecting CBL ubiquitin ligase ing data over human signaling pathways.
activation. Cancer Res 76(3):561–571 Nucleic Acids Res 43(W1):W270–W275
21. NR Coordinators (2014) Database resources 30. Choi Y, Sims GE, Murphy S, Miller JR, Chan
of the National Center for Biotechnology AP (2012) Predicting the functional effect of
Information. Nucleic Acids Res 42(Database amino acid substitutions and indels. PLoS One
issue):D7–D17 7(10):e46688
22. Perez B, Mechinaud F, Galambrun C, Ben 31. Adzhubei IA, Schmidt S, Peshkin L, Ramensky
Romdhane N, Isidor B, Philip N, Derain-Court VE, Gerasimova A, Bork P, Kondrashov AS,
J, Cassinat B, Lachenaud J, Kaltenbach S, Sunyaev SR (2010) A method and server for
Salmon A, Desiree C, Pereira S, Menot ML, predicting damaging missense mutations. Nat
Royer N, Fenneteau O, Baruchel A, Chomienne Methods 7(4):248–249
C, Verloes A, Cave H (2010) Germline muta- 32. Thusberg J, Olatubosun A, Vihinen M (2011)
tions of the CBL gene define a new genetic Performance of mutation pathogenicity predic-
syndrome with predisposition to juvenile tion methods on missense variants. Hum Mutat
myelomonocytic leukaemia. J Med Genet 32(4):358–368
47(10):686–691 33. Hashimoto K, Rogozin IB, Panchenko AR
23. Marchler-Bauer A, Derbyshire MK, Gonzales (2012) Oncogenic potential is related to acti-
NR, Lu S, Chitsaz F, Geer LY, Geer RC, He J, vating effect of cancer single and double
Gwadz M, Hurwitz DI, Lanczycki CJ, Lu F, somatic mutations in receptor tyrosine kinases.
Marchler GH, Song JS, Thanki N, Wang Z, Hum Mutat 33(11):1566–1575
Yamashita RA, Zhang D, Zheng C, Bryant SH 34. Schlebach JP, Narayan M, Alford C, Mittendorf
(2015) CDD: NCBI’s conserved domain data- KF, Carter BD, Li J, Sanders CR (2015)
base. Nucleic Acids Res 43(D1):D222–D226 Conformational stability and pathogenic mis-
24. Ashkenazy H, Erez E, Martz E, Pupko T, Ben- folding of the integral membrane protein
Tal N (2010) ConSurf 2010: calculating evolu- PMP22. J Am Chem Soc 137(27):8758–8768
tionary conservation in sequence and structure 35. Potapov V, Cohen M, Schreiber G (2009)
of proteins and nucleic acids. Nucleic Acids Res Assessing computational methods for predict-
38(suppl 2):W529–W533 ing protein stability upon mutation: good on
25. Shoemaker BA, Zhang D, Thangudu RR, average but not in the details. Protein Eng Des
Tyagi M, Fong JH, Marchler-Bauer A, Bryant Sel 22(9):553–560
SH, Madej T, Panchenko AR (2010) Inferred 36. Khan S, Vihinen M (2010) Performance of
Biomolecular Interaction Server – a web server protein stability predictors. Hum Mutat
to analyze and predict protein interacting part- 31(6):675–684
ners and binding sites. Nucleic Acids Res
38(Database issue):D518–D524 37. Zhang Z, Wang L, Gao Y, Zhang J,
Zhenirovskyy M, Alexov E (2012) Predicting
26. Goncearenco A, Shaytan AK, Shoemaker BA, folding free energy changes upon single point
Panchenko AR (2015) Structural perspectives mutations. Bioinformatics 28(5):664–671
on the evolutionary expansion of unique
protein-protein binding sites. Biophys 38. Nishi H, Tyagi M, Teng S, Shoemaker BA,
J 109(6):1295–1306 Hashimoto K, Alexov E, Wuchty S, Panchenko
AR (2013) Cancer missense mutations alter
27. Shoemaker BA, Zhang D, Tyagi M, Thangudu binding properties of proteins and their inter-
RR, Fong JH, Marchler-Bauer A, Bryant SH, action networks. PLoS One 8(6):e66273
Madej T, Panchenko AR (2012) IBIS (Inferred
Biomolecular Interaction Server) reports, pre- 39. Teng S, Madej T, Panchenko A, Alexov E
dicts and integrates multiple types of conserved (2009) Modeling effects of human single
interactions for proteins. Nucleic Acids Res nucleotide polymorphisms on protein-protein
40(Database issue):D834–D840 interactions. Biophys J 96(6):2178–2188
28. Dou H, Buetow L, Hock A, Sibbet GJ, 40. Ghersi D, Singh M (2014) Interaction-based
Vousden KH, Huang DT (2012) Structural discovery of functionally important genes in
basis for autoinhibition and phosphorylation- cancers. Nucleic Acids Res 42(3):e18
dependent activation of c-Cbl. Nat Struct Mol 41. Li M, Petukh M, Alexov E, Panchenko AR
Biol 19(2):184–192 (2014) Predicting the impact of missense
258 Minghui Li et al.
mutations on protein-protein binding affinity. cBio cancer genomics portal: an open platform
J Chem Theory Comput 10(4):1770–1780 for exploring multidimensional cancer genom-
42. Porollo A, Meller J (2007) Prediction-based ics data. Cancer Discov 2(5):401–404
fingerprints of protein–protein interactions. 55. Peri S, Navarro JD, Amanchy R, Kristiansen
Proteins 66(3):630–645 TZ, Jonnalagadda CK, Surendranath V,
43. Qin S, Zhou H-X (2007) meta-PPISP: a meta Niranjan V, Muthusamy B, Gandhi TK,
web server for protein-protein interaction site Gronborg M, Ibarrola N, Deshpande N,
prediction. Bioinformatics 23(24):3386–3387 Shanker K, Shivashankar HN, Rashmi BP,
44. Zhou H-X, Qin S (2007) Interaction-site pre- Ramya MA, Zhao Z, Chandrika KN, Padma N,
diction for protein complexes: a critical assess- Harsha HC, Yatish AJ, Kavitha MP, Menezes
ment. Bioinformatics 23(17):2203–2209 M, Choudhury DR, Suresh S, Ghosh N,
Saravana R, Chandran S, Krishna S, Joy M,
45. Porollo A, Meller J (2012) Computational Anand SK, Madavan V, Joseph A, Wong GW,
methods for prediction of protein-protein Schiemann WP, Constantinescu SN, Huang L,
interaction sites. Protein-Protein Interactions – Khosravi-Far R, Steen H, Tewari M, Ghaffari
Computational and Experimental Tools S, Blobe GC, Dang CV, Garcia JG, Pevsner J,
472:3–26 Jensen ON, Roepstorff P, Deshpande KS,
46. Li M, Simonetti FL, Goncearenco A, Chinnaiyan AM, Hamosh A, Chakravarti A,
Panchenko AR (2016) MutaBind estimates Pandey A (2003) Development of human pro-
and interprets the effects of sequence variants tein reference database as an initial platform for
on protein-protein interactions. Nucleic Acids approaching systems biology in humans.
Res. Jul 8;44(W1):W494-501. Genome Res 13(10):2363–2371
47. Li M, Shoemaker BA, Thangudu RR, Ferraris 56. Kerrien S, Aranda B, Breuza L, Bridge A,
JD, Burg MB, Panchenko AR (2013) Broackes-Carter F, Chen C, Duesbury M,
Mutations in DNA-binding loop of NFAT5 Dumousseau M, Feuermann M, Hinz U,
transcription factor produce unique outcomes Jandrasits C, Jimenez RC, Khadake J,
on protein-DNA binding and dynamics. J Phys Mahadevan U, Masson P, Pedruzzi I,
Chem B 117(42):13226–13234 Pfeiffenberger E, Porras P, Raghunath A,
48. Case DA, Cheatham TE, Darden T, Gohlke H, Roechert B, Orchard S, Hermjakob H (2012)
Luo R, Merz KM, Onufriev A, Simmerling C, The IntAct molecular interaction database in
Wang B, Woods RJ (2005) The Amber biomo- 2012. Nucleic Acids Res 40(Database
lecular simulation programs. J Comput Chem issue):D841–D846
26(16):1668–1688 57. Joshi-Tope G, Gillespie M, Vastrik I,
49. Stefl S, Nishi H, Petukh M, Panchenko AR, D'Eustachio P, Schmidt E, de Bono B, Jassal B,
Alexov E (2013) Molecular mechanisms of Gopinath GR, Wu GR, Matthews L, Lewis S,
disease-causing missense mutations. J Mol Biol Birney E, Stein L (2005) Reactome: a knowl-
425(21):3919–3936 edgebase of biological pathways. Nucleic Acids
50. Mizuguchi K, Deane CM, Blundell TL, Res 33(Database issue):D428–D432
Johnson MS, Overington JP (1998) JOY: pro- 58. Kanehisa M, Goto S, Hattori M, Aoki-
tein sequence-structure representation and Kinoshita KF, Itoh M, Kawashima S, Katayama
analysis. Bioinformatics 14(7):617–623 T, Araki M, Hirakawa M (2006) From genom-
51. Tina KG, Bhadra R, Srinivasan N (2007) PIC: ics to chemical genomics: new developments in
protein interactions calculator. Nucleic Acids KEGG. Nucleic Acids Res 34(Database
Res 35(suppl 2):W473–W476 issue):D354–D357
52. Stenson P, Mort M, Ball E, Shaw K, Phillips A, 59. Niknafs N, Kim D, Kim R, Diekhans M, Ryan
Cooper D (2014) The Human Gene Mutation M, Stenson PD, Cooper DN, Karchin R
Database: building a comprehensive mutation (2013) MuPIT interactive: webserver for map-
repository for clinical and molecular genetics, ping variant positions to annotated, interactive
diagnostic testing and personalized genomic 3D structures. Hum Genet
medicine. Hum Genet 133(1):1–9 132(11):1235–1243
53. Thorn CF, Klein TE, Altman RB (2010) 60. Peterson TA, Adadey A, Santana-Cruz I, Sun
Pharmacogenomics and bioinformatics: Y, Winder A, Kann MG (2010) DMDM:
PharmGKB. Pharmacogenomics domain mapping of disease mutations.
11(4):501–505 Bioinformatics 26(19):2458–2459
54. Cerami E, Gao J, Dogrusoz U, Gross BE, 61. Jegga AG, Gowrisankar S, Chen J, Aronow BJ
Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, (2007) PolyDoms: a whole genome database
Heuer ML, Larsson E, Antipin Y, Reva B, for the identification of non-synonymous cod-
Goldberg AP, Sander C, Schultz N (2012) The ing SNPs with the potential to impact disease.
Effects of Mutations on Protein-Protein Interactions 259
computational approach. Nucleic Acids Res based method for predicting protein stability
42(Web Server Issue):W314–319 changes upon mutations. Bioinformatics
87. Laimer J, Hofer H, Fritz M, Wegenkittl S, 23(10):1292–1293
Lackner P (2015) MAESTRO – multi agent 93.
Dehouck Y, Kwasigroch JM, Rooman
stability prediction upon point mutations. M, Gilis D (2013) BeAtMuSiC: predic-
BMC Bioinformatics 16(1):116 tion of changes in protein–protein bind-
88. Capriotti E, Fariselli P, Rossi I, Casadio R ing affinity on mutations. Nucleic Acids Res
(2008) A three-state prediction of single point 41(W1):W333–W339
mutations on protein stability changes. BMC 94. Berliner N, Teyra J, Çolak R, Garcia Lopez S,
Bioinformatics 9(Suppl 2):S6 Kim PM (2014) Combining structural model-
89. Cheng J, Randall A, Baldi P (2006) Prediction ing with ensemble machine learning to accu-
of protein stability changes for single-site muta- rately predict protein fold stability and binding
tions using support vector machines. Proteins affinity effects upon mutation. PLoS One
62(4):1125–1132 9(9):e107353
90. Chen CW, Lin J, Chu YW (2013) iStable: off- 95. Kruger DM, Gohlke H (2010) DrugScorePPI
the-
shelf predictor integration for predicting webserver: fast and accurate in silico alanine
protein stability changes. BMC Bioinformatics scanning for scoring protein-protein interac-
14(Suppl 2):S5 tions. Nucleic Acids Res 38(Web Server
91. Teng S, Srivastava A, Wang L (2010) Sequence Issue):W480–486
feature-based prediction of protein stability 96. Zhao N, Han JG, Shyu CR, Korkin D (2014)
changes upon amino acid substitutions. BMC Determining effects of non-synonymous SNPs
Genomics 11(Suppl 2):1–8 on protein-protein interactions using super-
92. Huang L-T, Gromiha MM, Ho S-Y (2007) vised and semi-supervised learning. PLoS
iPTREE-STAB: interpretable decision tree Comput Biol 10(5):e1003592
Chapter 18
Abstract
Characterization, especially quantification, of protein interactions in live cells is usually not an easy
endeavor. Here, we describe a straightforward method to identify and quantify the interaction of a mem-
brane protein (“bait”) and a fluorescently labeled interaction partner (“prey”) (membrane-bound or cyto-
solic) in live cells using Total Internal Reflection Fluorescence microscopy. The bait protein is immobilized
within patterns in the plasma membrane (e.g., via an antibody); the bait–prey interaction strength can be
quantified by determining the prey bulk fluorescence intensity with respect to the bait patterns. This
method is particularly suitable also for the analysis of weak, transient interactions that are not easily acces-
sible with other methods.
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_18, © Springer Science+Business Media LLC 2017
261
262 Gerhard J. Schütz et al.
specific patterns within the plasma membrane of living cells [7, 8].
We have extended this approach to use it as a tool for characteriza-
tion and quantification of protein interactions: One interaction
partner (bait) is restricted to specific regions (typically regular
micropatterns) in the live cell plasma membrane and the lateral dis-
tribution of a fluorescently labeled interaction partner (prey) is
monitored. In case of an interaction, prey molecules will follow the
bait pattern; homogeneous distribution of prey protein in the
plasma membrane indicates the absence of an interaction (Fig. 1).
Quantification can be achieved by comparing the prey signal inten-
sity within and outside the bait regions: the signal contrast between
these regions provides a measure of the interaction strength.
While patterned surfaces can be generated by different meth-
ods (e.g., photolithography [9] or dip-pen nanolithography [10]),
soft lithography [11] is probably the most convenient: it is fast,
simple, and lends itself to high throughput routines. In this proto-
col, the patterned cell substrate is produced by printing streptavi-
din patterns on a glass coverslip, to which a bait-specific biotinylated
antibody is then attached. We have first used this approach to
2 Materials
Prepare all work solutions fresh each time. Store epoxy-coated cov-
erslips in the desiccator after opening. This protocol is optimized
for PDMS stamps; if a different material is used, conditions may
need to be adjusted for optimal printing results.
1. Polydimethylsiloxane (PDMS) stamps (see Note 1).
2. Epoxy-coated coverslips: NEXTERION® slide E (Schott,
Germany).
3. Streptavidin stock solution: dissolve 0.5 mg/ml streptavidin
(Sigma, USA) in phosphate buffered saline (PBS) pH 7.4.
Store aliquots at −20 °C. Do not freeze and thaw.
4. Streptavidin work solution: dilute streptavidin stock solution
to 50 μg/ml in PBS pH 7.4.
5. Secure SealTM Hybridization chambers (Grace Biolabs, USA).
6. BSA-Cy5 stock solution (see Note 2): dissolve Cy5-labeled
bovine serum albumin (BSA-Cy5; Nanocs, USA) to 1 mg/ml in
PBS pH 7.4. Store aliquots at −20 °C. Do not freeze and thaw.
7. BSA-Cy5 work solution: dilute BSA-Cy5 stock solution to
100 μg/ml in PBS pH 7.4.
8. Antibody work solution: dilute biotinylated antibody to
10 μg/ml in PBS pH 7.4 containing 1 % BSA.
9. Imaging buffer: Hank’s Balanced Salt Solution (HBSS) with
Ca2+ and Mg2+ and 2 % fetal calf serum (FCS) (see Note 3).
10. Cells expressing bait proteins and fluorescent prey proteins (see
Note 4), Accutase (Sigma, USA) (see Note 5).
3 Methods
3.1 Soft Lithography The workflow of “3.1 Soft lithography and functionalization” is
and Functionalization sketched in Fig. 2.
1. Wash PDMS stamp by rinsing with ethanol (p.a.) and ultra-
pure water. Dry the PDMS stamp under a stream of a dry inert
gas such as nitrogen or argon.
2. Place ~50 μL of streptavidin work solution (50 μg/ml) on the
PDMS stamp (the whole pattern area should be covered). Let
protein adsorb to stamp for 15 min at room temperature (see
Note 6).
3. Wash the PDMS stamp by rinsing carefully with water and dry
under a stream of nitrogen or argon.
4. Place the PDMS stamp face-down under its own weight onto
an epoxy-coated coverslip and incubate for 30 min at room
temperature or overnight at 4 °C in a humidified atmosphere
(e.g., a petri dish with a wet tissue) (see Note 7).
5. Mark the position of the patterned area on the back of the
coverslip with a water-resistant marker and separate the stamp
from the slide using tweezers (see Note 8).
Protein Micropatterning Assay 265
3.2 Seeding Cells 1. Grow adherent cells expressing bait and prey proteins of inter-
est to 70 % confluency in a 10 cm tissue culture dish.
2. Detach cells with Accutase® solution and centrifuge 4 min at
300 × g. This protocol has been tested for T24, HeLa and
CHO cells (see Note 11).
3. Pellet cells by spinning for 5 min at ~300 × g.
4. Discard the supernatant and resuspend the cell pellet in 1 ml of
the appropriate growth medium. Then, dilute this ~1:10 in
growth medium (see Note 12).
5. Remove the PBS from the hybridization chamber on the
micropatterned coverslip and seed cell suspension.
6. Check cell density on a light microscope. Cells should be sin-
gle but not too sparse.
7. Put coverslips in a petri dish humidity chamber to prevent the
sample from running dry and incubate for 1.5–2 h at 37 °C in
a 5 % CO2 atm.
8. Before analyzing the cells on the microscope, replace the
medium with imaging buffer.
3.3 Total Internal 1. Place the coverslip on a TIRF microscope in a suitable mount
Reflection (see Note 13).
Fluorescence (TIRF) 2. The BSA-Cy5 grid needed for quantitative analysis is recorded
Microscopy at 647 nm.
3. Distribution of fluorescent prey protein (tagged with e.g.,
GFP) is recorded (at e.g., 488 nm (see Note 14)).
3.4 Contrast 1. Export microscopy images as 8-bit TIF image. For contrast
Quantitation quantitation it is necessary to export images of the fluorescent
prey/bait protein (Fig. 3a) and the respective image with the
BSA-Cy5 grid (Fig. 3b). Figure 3c shows the overlaid images.
266 Gerhard J. Schütz et al.
Fig. 3 Quantitation of protein interactions using “Spotty”. Image recorded of the fluorescently labeled prey
protein (a) and the corresponding BSA-Cy5 grid (b). (c) Overlay. (d) An automatic gridding algorithm automati-
cally optimizes the grid parameters and produces a grid that correctly fits the micropatterns. Yellow lines
denote the cell areas to be used for analysis. (e) The grid subdivides the total image into adjacent squares,
each of which is quantified according to the average signal within a central circle comprising the micropattern
spot (F+) and the signal outside this circle (F−). (f) Statistical analysis of multiple cells is shown in a 2D histo-
gram of the fluorescence brightness and contrast. The color scale corresponds to the number of events (i.e.,
individual analyzed spots)
Protein Micropatterning Assay 267
Fig. 4 Examples of generated 2D histograms (a) T24 cell transiently expressing CD4 and Lck-YFP grown on CD4
antibody patterns. Lck-YFP interacts strongly with the patterned CD4, which is also reflected in the high contrast
values shown in the 2D histogram on the right. The low contrast values at lower fluorescence intensities are
probably a result of low CD4 (and Lck-YFP) expression levels of a cell subpopulation. For calculating the mean
contrast <C>, we only consider data points above a certain intensity threshold (indicated by the yellow line). (b)
T24 cell transiently expressing CD4 and cytosolic YFP grown on CD4 antibody patterns. No copatterning of YFP
with CD4 can be observed, the contrast values fluctuate around zero. Scale bars are 10 μm. The color scale
corresponds to the number of events (i.e., individual analyzed spots). Figure modified from [12]
4 Notes
14. When using this assay for the first time, we recommend using
a fluorescently labeled bait protein as described in Note 4. If
the fluorescence signals of bait, prey and analysis grid are suf-
ficiently spectrally separated, labeled bait protein can be used
for all measurements.
15. “Spotty” can be obtained from www.protein-interaction-lab.at
upon request.
16. Evolutionary computation strategies are used for optimized
grid identification and detection of micropatterns in biological
samples.
17. A relevant factor for the success of contrast evaluation is the
size of the F+ region. It has to be adjusted to fit the actual size
of the printed patterns (as shown in Fig. 3e).
18. Taking into account the fluorescence brightness is especially
useful when dealing with a heterogeneous cell population with
very different expression levels of bait and prey protein (see also
Fig. 4). It may be advantageous to analyze cell subpopulations
of different expression levels separately, or to apply an intensity
threshold as shown in Fig. 4.
Acknowledgments
This work was funded by the Austrian Science Fund (FWF projects
P 26337 and P 25730), the Austrian Research Promotion Agency
(FFG project 842379), the program ‘Regionale
Wettbewerbsfähigkeit OÖ 2007–2013’ with the financial support
of the European Fund for Regional Development, as well as the
Federal State of Upper Austria.
References
Abstract
Because proteomics experiments are so complex they can readily fail, and do so without clear cause. Using
standard experimental design techniques and incorporating quality control can greatly increase the chances
of success. This chapter introduces the relevant concepts and provides examples specific to proteomic
workflows. Applying these notions to design successful proteomics experiments is straightforward. It can
help identify failure causes and greatly increase the likelihood of inter-laboratory reproducibility.
1 Introduction
1.1 Why Proteomics Proteomics aims to quantify thousands of proteins in complex bio-
Experiments Benefit logical backgrounds, such as tissue and plasma. It is typically done
from Design using sensitive instruments like mass spectrometers following mul-
tiple preparation steps, including protein extraction, enrichment,
fractionation, and digestion. Because these complex laboratory
processes can lack stability across a full study, they can negatively
impact the outcome unless the experiment is designed to take this
into account.
In their 2005 paper [5], Hu et al. describe three proteomics
studies that failed due to poor experimental design. In one study,
cancer sample data showed strong dissimilarity between run dates.
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_19, © Springer Science+Business Media LLC 2017
271
272 Daniel Ruderman
Because one cancer subtype’s samples were all run on a single day,
any biological signal was confounded with any technical changes
that day. In another study, samples were found to group in similarity
not by cancer subtype, as expected, but instead by the date the col-
lection protocol was changed. Their third example was the victim of
erroneous mass spectrometer calibration and sample degradation.
These studies would likely have succeeded with proper experimental
design and inclusion of quality controls. Poor experimental design is
cited as a major impediment to translational proteomics [6].
Science is ultimately held to the standard of reproducibility. So
it is surprising that attempts to reproduce the results of high-profile
biomedical studies mostly failed [7, 8]. Poor experimental design
has been highlighted as a cause reproducibility failure [9]. The US
National Institutes of Health has recently emphasized reproduc-
ibility, stating that grant applications will be reviewed specifically
for “rigorous experimental design for robust and unbiased results”
[10]. Thus experimental design is not only important in executing
research, but in securing the funds to do so in the first place.
2 Materials
3.1 Define A clear research goal determines the experiment you’ll want to
the Research Goal run. It embodies questions like: “What effect will you measure?”
and Relevant and “Which sample groups will you compare?” These in turn iden-
Response Variables tify the response variables you will measure and the experimental
factors you will change to determine their impact.
Response variables are the experimental outputs (numbers) used
to answer the research question. In quantitative proteomics they
may be the abundances of peptides or proteins. In proteome surveys
they may instead be the number of distinct proteins found across
samples. Core to designing an experiment is choosing the experi-
mental methodology. Other chapters in this volume can aid in the
decision, for example, between MRM (Multiple Reaction
Monitoring), DDA (Data-Dependent Acquisition), DIA (Data-
Independent Acquisition), or other methods [17]. The best choice
will optimize the ability to accurately measure the response variables,
possibly at the expense of reduced performance by other metrics.
For example, in setting a mass isolation window there is a trade-off
between analyte selectivity and quantitation sensitivity [18].
It is important to have an estimate of the effect size, as it will
help determine how many samples will need to be run (see below).
For early stage biomarker screening any significant difference may be
of interest, no matter how small. In other cases, such as toxicology
studies, only physiologically meaningful differences are of value.
Fig. 1 Four major activities and their component steps to designing an unbiased and efficient experiment
274 Daniel Ruderman
3.2 List Relevant Factors are those things that potentially affect response variables.
Factors Ideally the response would only be impacted by the physical prop-
erty the experimental system is designed to measure. But this is
almost never the case. For example, the measured peptide peak
area depends not just on that peptide’s abundance but also on elec-
trospray stability, chromatography column carryover, and the effi-
ciency of proteolytic digestion. Digestion efficiency may in turn
depend on the trypsin reagent lot number and which lab tech per-
forms the digestion.
Those factors we change as part of the experimental investiga-
tion are called experimental factors. For example, when comparing
the proteome between subcellular fractions, the proteomic fraction
(e.g., cytoplasm or nucleus) is such a factor. If we are instead test-
ing the effect of chromatography on peak height, then flow rate
might be an experimental factor. Once the experimental factors
have been chosen to answer the research question, all those remain-
ing in the list are called nuisance factors. These are conditions that
can affect response variables but do not remain fixed for all samples
or runs. Examples include reagent batches, intra-day LC-MS run
order, and personnel. The goal of identifying nuisance factors is to
minimize their effect on response variables while allowing experi-
mental factors to systematically alter those responses.
3.3 Develop Control Two kinds of control samples are important for verifying that the
Samples and Quality experimental platform actually works. Experimental controls ensure
Control Procedures accuracy: Quantitation should reflect what is in the sample; actual
differences should be detected; lack of change between sample groups
should not. Quality control samples (QCs) ensure that processes and
equipment perform within defined specifications [19, 20].
Experimental controls for accuracy are used to ensure that
quantities like retention time, m/z, and protein/peptide abun-
dance are well measured. They are typically created by spiking
compounds into the experimental samples. Controls can also be
run through the mass spectrometer at different times than the
experimental samples (e.g., lock masses during alternate MS scans,
separate QC samples). Krokhin and Spicer [21] describe a set of
spike-in peptides for normalizing reverse phase HPLC retention
times. The peptides enable both retention time normalization of
data between runs and hydrophobicity index prediction of sample
peptides [22] for improved identification. For protein abundance
normalization between runs of complex proteomic samples, the
“super-SILAC” approach of Geiger et al. can be used [23]. Here a
proteomic standard sample is derived from species-specific cells
cultured in SILAC media, providing mass shifts to tryptic peptides
so they do not overlap sample peptides. This approach has two
important benefits. First the proteins are matched to the sample, so
most peptides will have a corresponding quantitation standard.
Second, because the standard can be spiked into the sample prior
Proteomic Experimental Design 275
3.4 Optimize Process Protocols such as found in this volume and in vendor application
notes give detailed roadmaps for key parts of the experiment. But
there remain many parameters to adjust in making the experimen-
tal system work well as a whole. For example, Agilent’s Jet Stream
electrospray ionization source has six adjustments, such as drying
gas temperature, nebulizer voltage, and capillary voltage [32].
Additionally, MS data analysis packages have their own settings.
The OMSSA search engine [33], for example, has 9 of them, not
including PTM selection. An experiment’s success may hinge on
getting them right.
Fortunately, there are systematic methods for optimizing these
settings. The statistics sub-field of Response Surface Methodology
(RSM) [34] is used to design and analyze an experiment to esti-
mate the settings that maximize a performance metric. Although
process optimization is not strictly part of experimental design (it
need not be done for each individual experiment), it is equally
important and often overlooked. The basic idea is to perform a
designed experiment across a set of parameter adjustments. RSM
then fits the performance metric to a (usually) quadratic surface in
the space of parameters, and estimates the parameter settings that
optimize performance. To optimize data analysis parameters no
additional experiments need be performed. Instead, the analysis
software can be run on a single representative data set under differ-
ent parameter settings. A number of examples demonstrating
proteomics performance gains through process optimization have
been published [35–42].
It is not always obvious how to choose the performance met-
ric. For example, when trying to discover biomarkers, is it best to
maximize the total number of spectral “features” detected or
instead to maximize the number of features with high signal-to-
noise ratio? Is the number of identified proteins also important? A
standard approach in such situations is to mathematically combine
a number of relevant metrics using Derringer’s desirability metric
[34] (see Note 1). It tends to compromise well between competing
criteria. One such tradeoff is found during label-free MS1 quanti-
tation in tandem mass spectrometry, where higher accuracy can be
Proteomic Experimental Design 277
holds, where N is the minimum number of samples needed per
group, zq is the qth quantile of the normal distribution, α is the
desired significance level (typically 0.05), 1-β is the power, σ is the
estimated total noise standard deviation (biological and technical),
and δ is the expected effect size. At a significance level of 0.05 and
power of 80 %, z1-a / 2 = 1.96 and z1- b = 0.84 , so N = 7.84 (s / d ) .
2
3.6.1 Blocking Many factors change discretely over the course of an experiment.
Some examples are reagent batches, personnel, date, and run order.
As one of these factors changes it can alter the response of the
Proteomic Experimental Design 279
3.6.2 Randomization There may be nuisance factors with many levels where it is not
obvious how to perform blocking, for example the assignment of
samples to wells in a 96-well plate. Although it is known that plate-
based experiments can have systematic changes toward plate edges,
it is very effective simply to randomize sample placement. Similarly,
the order in which a day’s samples are processed can be random-
ized in case of changing equipment or personnel attentiveness.
Temporal randomization (both within and across days) in pro-
teomics is particularly important since there are many potential
sources of drift, including LC column degradation, electrospray
instability, transfer tube build-up, and tuning loss.
Randomization has the added advantage of counteracting bias
due to unrecognized influences. For example if a contiguous sub-
set of vials gets contaminated, and they were randomized relative
to sample groups, then the experimental result will show increased
variability but without a biased effect on the conclusion (although
with increased confidence intervals). Randomization is particularly
280 Daniel Ruderman
Fig. 2 A randomized block design for a proteomics LC-MS workflow. Two sample groups (A and B) are being
measured for proteomic differences. The experimental day is a blocking factor, and morning (AM) versus after-
noon (PM) is a blocking factor nested within the day. Each such block contains six samples run (balanced
between three samples of each group). The 6-sample run order is randomized to avoid confounding temporal
drift with the A/B proteomic difference, with laboratory personnel blinded to the ordering. At the middle of each
LC-MS run a QC sample is analyzed to monitor process. QC and sample data should be analyzed to ensure no
systematic AM versus PM differences or large day-to-day variation impact the signals of interest
3.6.3 Replication Unlike blocking and randomization, which seek to remove system-
atic bias that could be mistaken for or mask a real effect, the goal
of replication is to reduce the impact of random measurement vari-
ation. This leads to more precise estimates of the effects that answer
Proteomic Experimental Design 281
Fig. 3 (a) Blocked and balanced design for a 4-plex iTRAQ experiment contrasting two sample groups (A and
B) with eight samples per group (A1 through A8, B1 through B8) in four runs/pools. (b) Frequency table for
labels across sample groups showing balanced design. (c) Frequency table for runs across sample groups
showing balanced design
3.6.5 Control Samples An experiment’s design should include control samples to ensure
process. This is most crucial in complex experiments involving
many samples measured across multiple days or instruments.
However, even a simple experiment on just a few samples will ben-
efit, particularly when a negative result is buoyed by knowledge
that the measurements are reliable. The ~15 % additional cost in
time and resources is well worth the payoff.
Quality control samples measure whether technology is func-
tioning consistently and within specifications. Since most pro-
teomic workflows involve multiple steps (e.g., capture, fractionation,
LC-MS) it is best to insert the QC samples as early in the workflow
as possible to test them all. A daily series of sample preparations can
be “bookended” by QC samples at either end to measure perfor-
mance changes. If intra-day changes are unlikely then a single QC
sample could instead be placed in the middle of a day’s experimen-
tal samples. In either case, if a QC sample demonstrates the equip-
ment has failed, those runs can be rejected and the samples possibly
rerun. Additionally, the response measures on the QC samples can
be used to normalize experimental data for possible drift.
Figure 4 shows a design exemplifying blocking, randomiza-
tion, and QCs. This design was motivated by a set of blood samples
collected from eight subjects in pairs, one prior to and one after a
seizure. The research goal was to determine whether there are
detectable changes in the plasma proteome following seizure. The
design specifies how samples are run on LC-MS on a system with
two LC columns (LC-1 and LC-2) alternately driving a single
Proteomic Experimental Design 283
Fig. 4 Example of blocking by LC system and inclusion of QCs in a paired sample (pre- [no asterisk] and post-
seizure [asterisk]) workflow. See Subheading 3.6.5 for details
3.7 Choose In addition to vendor-specific software, there are now many freely
the Statistical Analysis available analysis platforms for proteomics data [55–58]. Analysis
steps include protein identification, quantification, and statistical
testing. The most common methods are t-tests and ANOVA. Since
these tests make certain assumptions about the data statistics (e.g.,
Gaussian noise, homogeneity of variance, independence across
samples) one should be sure these preconditions are met (particu-
larly for small data sets) [3]; if not, other methods such as nonpara-
metric tests (e.g., Mann–Whitney U) can be used. Quantitation
through spectral counting, for example, can often have non-
Gaussian noise statistics. With blocked designs one must ensure
blocking is accounted for in the analysis (so that the possibly large
variation between blocks is ignored when estimating the noise
level). This can typically be accomplished using a blocking factor in
ANOVA [59] or mixed effects models [60]. As a general rule the
284 Daniel Ruderman
statistical analysis must match both the research question and the
experimental design. It must not be altered after the collection of
data (“HARKing” [61]). I strongly recommend engaging a
statistician early in the experiment design process to address
these questions. After data are collected is often too late.
Control samples also must be analyzed. QC samples should be
checked for instrumentation and process failure. These QCs can
also be used to detect performance variation between blocks (e.g.,
across days) to characterize changes in equipment performance.
Analysis of positive and negative control samples should validate
that the entire experimental workflow functions properly.
4 Conclusions
5 Notes
Acknowledgments
References
16. Cairns DA (2014) Statistical issues in the Zimmerman LJ, Carr SA, Fisher SJ, Gibson
design and planning of proteomic profiling BW, Paulovich AG, Regnier FE, Rodriguez
experiments. Clin Proteomics 18:223–236 H, Spiegelman C, Tempst P, Liebler DC,
17. Leitner A, Aebersold R (2013) SnapShot: mass Stein SE (2010) Performance metrics for liq-
spectrometry for protein and proteome analy- uid chromatography- tandem mass spectrom-
ses. Cell 154(1):252–252e251. doi:10.1016/j. etry systems in proteomics analyses. Mol Cell
cell.2013.06.025 Proteomics 9(2):225–241
18. Gallien S, Bourmaud A, Kim SY, Domon B 27. Ma Z-Q, Polzin KO, Dasari S, Chambers
(2014) Technical considerations for large- MC, Schilling B, Gibson BW, Tran BQ, Vega-
scale parallel reaction monitoring analysis. Montoto L, Liebler DC, Tabb DL (2012)
J Proteomics 100:147–159. doi:10.1016/j. QuaMeter: multivendor performance metrics
jprot.2013.10.029 for LC–MS/MS proteomics instrumentation.
19. Montgomery DC (2013) Introduction to Anal Chem 84(14):5845–5850
statistical quality control, 7th edn. Wiley,
2 8. Wang X, Chambers MC, Vega-Montoto LJ,
Hoboken, NJ Bunk DM, Stein SE, Tabb DL (2014) QC
20. Bramwell D (2013) An introduction to statis- metrics from CPTAC raw LC-MS/MS data
tical process control in research proteomics. interpreted through multivariate statistics. Anal
J Proteomics 95(C):3–21, doi: papers3://pub- Chem 86(5):2497–2509
lication/doi/10.1016/j.jprot.2013.06.010
2 9. Bereman MS, Johnson R, Bollinger J, Boss
21. Krokhin OV, Spicer V (2009) Peptide reten- Y, Shulman N, MacLean B, Hoofnagle
tion standards and hydrophobicity indexes AN, MacCoss MJ (2014) Implementation
in reversed-phase high-performance liq- of statistical process control for proteomic
uid chromatography of peptides. Anal experiments via LC MS/MS. J Am Soc Mass
Chem 81(22):9522–9530, doi: papers3:// Spectrom 25(4):581–587. doi:10.1007/
publication/doi/10.1021/ac9016693 s13361-013-0824-5
22. Krokhin OV (2006) Sequence-specific retention
3 0. Burkhart JM, Vaudel M, Zahedi RP, Martens
calculator. algorithm for peptide retention predic- L, Sickmann A (2011) iTRAQ protein quan-
tion in ion-pair RP-HPLC: application to 300- tification: a quality-controlled workflow.
and 100-Å pore size C18 sorbents. Anal Chem Proteomics 11(6):1125–1134, doi: papers3://
78(22):7785–7795. doi:10.1021/ac060777w publication/doi/10.1002/pmic.201000711
23. Geiger T, Wisniewski JR, Cox J, Zanivan S,
3 1. Herbrich SM, Cole RN, West KP Jr, Schulze
Kruger M, Ishihama Y, Mann M (2011) Use K, Yager JD, Groopman JD, Christian P, Wu
of stable isotope labeling by amino acids in L, O'Meally RN, May DH, McIntosh MW,
cell culture as a spike-in standard in quantita- Ruczinski I (2013) Statistical inference from
tive proteomics. Nat Protoc 6(2):147–157, multiple iTRAQ experiments without using
doi: papers3://publication/doi/10.1038/ common reference standards. J Proteome Res
nprot.2010.192 12(2):594–604. doi:10.1021/pr300624g
24. Levin Y, Hradetzky E, Bahn S (2011)
3 2. Greco G, Boltner A, Letzel T (2014)
Quantification of proteins using data- Optimization of jet stream ESI parameters
independent analysis (MSE) in simple and- when coupling agilent 1260 infinity analytical
complex samples: a systematic evaluation. SFC system with agilent 6230 TOF LC/MS.,
Proteomics 11(16):3273–3287 Agilent Technologies, https://www.agilent.
com/cs/library/applications/5991-4510EN.
25. Ivanov AR, Colangelo CM, Dufresne CP, pdf
Friedman DB, Lilley KS, Mechtler K, Phinney
BS, Rose KL, Rudnick PA, Searle BC, Shaffer
3 3. Geer LY, Markey SP, Kowalak JA, Wagner L,
SA, Weintraub ST (2013) Interlaboratory Xu M, Maynard DM, Yang X, Shi W, Bryant
studies and initiatives developing standards SH (2004) Open mass spectrometry search
for proteomics. Proteomics 13(6):904–909, algorithm. J Proteome Res 3(5):958–964.
doi: papers3://publication/doi/10.1002/ doi:10.1021/pr0499491
pmic.201200532
3 4. Bezerra MA, Santelli RE, Oliveira EP, Villar
26. Rudnick PA, Clauser KR, Kilpatrick LE, LS, Escaleira LA (2008) Response surface
Tchekhovskoi DV, Neta P, Blonder N, methodology (RSM) as a tool for optimization
Billheimer DD, Blackman RK, Bunk DM, in analytical chemistry. Talanta 76(5):965–977
Cardasis HL, Ham A-JL, Jaffe JD, Kinsinger 35. Coscollà C, Navarro-Olivares S, Martí P, Yusà V
CR, Mesri M, Neubert TA, Schilling B, Tabb (2014) Application of the experimental design
DL, Tegeler TJ, Vega-Montoto L, Variyath of experiments (DoE) for the determination
AM, Wang M, Wang P, Whiteaker JR, of organotin compounds in water samples
Proteomic Experimental Design 287
using HS-SPME and GC–MS/MS. Talanta RR, Ioannidis JPA, Jankowski J, Julian BA,
119:544–552 Klein JB, Kolch W, Luider T, Massy Z, Mattes
36. Eliasson M, Rännar S, Madsen R, Donten MA, WB, Molina F, Monsarrat B, Novak J, Peter
Marsden-Edwards E, Moritz T, Shockcor JP, K, Rossing P, Sánchez-Carbayo M, Schanstra
Johansson E, Trygg J (2012) Strategy for opti- JP, Semmes OJ, Spasovski G, Theodorescu D,
mizing LC-MS data processing in metabolo- Thongboonkerd V, Vanholder R, Veenstra TD,
mics: a design of experiments approach. Anal Weissinger E, Yamamoto T, Vlahou A (2010)
Chem 84(15):6869–76 Recommendations for biomarker identification
37. Székely G, Henriques B, Gil M, Ramos A, and qualification in clinical proteomics. Science
Alvarez C (2012) Design of experiments as a Transl Med 2 (46): 46ps42
tool for LC–MS/MS method development for 44. Anderle M, Roy S, Lin H, Becker C, Joho K
the trace analysis of the potentially genotoxic (2004) Quantifying reproducibility for differen-
4-dimethylaminopyridine impurity in glucocor- tial proteomics: noise analysis for protein liquid
ticoids. J Pharmaceut Biomed Anal 70:251–8 chromatography-mass spectrometry of human
38. Maes K, Van Liefferinge J, Viaene J, Van serum. Bioinformatics 20(18):3575–3582
Schoors J, Van Wanseele Y, Béchade G, 45. Liu P, Hwang JTG (2007) Quick calculation
Chambers EE, Morren H, Michotte Y, for sample size while controlling false discov-
Vander Heyden Y, Claereboudt J, Smolders ery rate with application to microarray analysis.
I, Van Eeckhaut A (2014) Improved sensitiv- Bioinformatics 23(6):739–746
ity of the nano ultra-high performance liquid 46. Fisher RA, Others (1926) The arrangement of field
chromatography- tandem mass spectrometric experiments. doi: citeulike-article-id:10709753
analysis of low-concentrated neuropeptides 47. Westbrook JA, Noirel J, Brown JE, Wright
by reducing aspecific adsorption and optimiz- PC, Evans CA (2015) Quantitation with
ing the injection solvent. J Chromatogr A chemical tagging reagents in biomarker stud-
1360:217–228 ies. Proteomics Clin Appl 9(3-4):295–300.
39. Passeport E, Guenne A, Culhaoglu T, Moreau doi:10.1002/prca.201400120
S, Bouyé J-M, Tournebize J (2010) Design of 48. Oberg AL, Mahoney DW, Eckel-Passow JE,
experiments and detailed uncertainty analysis Malone CJ, Wolfinger RD, Hill EG, Cooper
to develop and validate a solid-phase micro- LT, Onuma OK, Spiro C, Therneau TM,
extraction/gas chromatography–mass spec- Bergen Iii HR (2008) Statistical analysis of rela-
trometry method for the simultaneous analysis tive labeled mass spectrometry data from com-
of 16 pesticides in water. J Chromatograph A plex samples using ANOVA. J Proteome Res
1217(33):5317–5327 7(1):225–233, doi: papers3://publication/
40. Raji MA, Schug KA (2009) Chemometric study doi/10.1021/pr700734f
of the influence of instrumental parameters on 49. Niepel M, Spencer SL, Sorger PK (2009) Non-
ESI-MS analyte response using full factorial genetic cell-to-cell variability and the conse-
design. Int J Mass Spectrom 279:100–106 quences for pharmacology. Curr Opin Chem
41. Zhou Y, Song J-Z, Choi FF-K, Wu H-F, Qiao Biol 13(5–6):556–561, doi: http://dx.doi.
C-F, Ding L-S, Gesang S-L, Xu H-X (2009) An org/10.1016/j.cbpa.2009.09.015
experimental design approach using response 50. Liu H, Sadygov RG, Yates JR 3rd (2004)
surface techniques to obtain optimal liquid A model for random sampling and estima-
chromatography and mass spectrometry condi- tion of relative protein abundance in shotgun
tions to determine the alkaloids in Meconopsi proteomics. Anal Chem 76(14):4193–4201.
species. J Chromatogr A 1216(42):7013–7023 doi:10.1021/ac0498563
42. Switzar L, Giera M, Lingeman H, Irth H, 51. Mercier C, Truntzer C, Pecqueur D, Gimeno
Niessen WMA (2011) Protein digestion opti- J-P, Belz G, Roy P (2009) Mixed-model
mization for characterization of drug–protein of ANOVA for measurement reproducibil-
adducts using response surface modeling. ity in proteomics. J Proteomics 72(6):974–
J Chromatogr A 1218(13):1715–1723, doi: 981, doi: http://dx.doi.org/10.1016/j.
papers3://publication/doi/10.1016/j. jprot.2009.05.005
chroma.2010.12.043 52. Day SJ, Altman DG (2000) Blinding in clinical
43. Mischak H, Allmaier G, Apweiler R, Attwood T, trials and other studies. BMJ 321(7259):504
Baumann M, Benigni A, Bennett SE, Bischoff 53. Begley CG (2013) Six red flags for suspect
R, Bongcam-Rudloff E, Capasso G, Coon JJ, work. Nature 497(7450):433–4
D'Haese P, Dominiczak AF, Dakna M,
Dihazi H, Ehrich JH, Fernandez-Llama P, Fliser 54. MacCoun R, Perlmutter S (2015) Blind analy-
D, Frokiaer J, Garin J, Girolami M, Hancock sis: hide results to seek the truth. Nature
WS, Haubitz M, Hochstrasser D, Holman 526(7572):187–189
288 Daniel Ruderman
55. Gatto L, Christoforou A (2014) Using R and 59. Kleinbaum DG, Kleinbaum DG (1998)
Bioconductor for proteomics data analysis. Applied regression analysis and other multivari-
Biochim Biophys Acta 1844(1 Pt A):42–51 able methods, 3rd edn. Duxbury, Pacific Grove
56. Colangelo CM, Chung L, Bruce C, Cheung 60. Choi M, Chang CY, Clough T, Broudy D,
K-H (2013) Review of software tools for Killeen T, MacLean B, Vitek O (2014)
design and analysis of large scale MRM pro- MSstats: an R package for statistical analysis of
teomic datasets. Methods 61(3):287–298 quantitative mass spectrometry-based pro-
57. Schwacke JH, Hill EG, Krug EL, Comte- teomic experiments. Bioinformatics
Walters S, Schey KL (2009) iQuantitator: a 30(17):2524–2526
tool for protein expression inference using 61. Kerr NL (1998) HARKing: hypothesizing
iTRAQ. BMC Bioinformatics 10:342. after the results are known. Pers Soc Psychol
doi:10.1186/1471-2105-10-342 Rev 2(3):196–217
58. Gonzalez-Galarza FF, Lawless C, Hubbard SJ, 62. Kuhn M. Desirability: desirability function
Fan J, Bessant C, Hermjakob H, Jones AR (2012) optimization and ranking. https://github.
A critical appraisal of techniques, software pack- com/topepo/desirability
ages, and standards for quantitative proteomic 63. Derringer G, Suich R (1980) Simultaneous
analysis. OMICS 16(9):431–442, doi: papers3:// optimization of several response variables.
publication/doi/10.1089/omi.2012.0022 J Qual Technol 12:214–219
Chapter 20
Abstract
Targeted mass spectrometry comprises a set of methods able to quantify protein analytes in complex mixtures
with high accuracy and sensitivity. These methods, e.g., Selected Reaction Monitoring (SRM) and SWATH
MS, use specific mass spectrometric coordinates (assays) for reproducible detection and quantification of
proteins. In this protocol, we describe how to analyze, in a targeted manner, data from a SWATH MS experi-
ment aimed at monitoring thousands of proteins reproducibly over many samples. We present a standard
SWATH MS analysis workflow, including manual data analysis for quality control (based on Skyline) as well
as automated data analysis with appropriate control of error rates (based on the OpenSWATH workflow). We
also discuss considerations to ensure maximal coverage, reproducibility, and quantitative accuracy.
Key words Targeted proteomics, SWATH, SWATH MS, SWATH acquisition, Data-independent
acquisition, DIA, OpenSWATH, pyProphet, TRIC aligner, Skyline
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_20, © Springer Science+Business Media LLC 2017
289
290 Hannes L. Röst et al.
A
1200
}
SWATH
isolation
1175 window
500
475 MS2
S2
m/z
MS2
S2
MS2
S2
MS2
MS2
450
425
400
{
Time
Cycle
time
B
MS2
MS2
MS2
MS2
Intensity
MS2
Intensity
m/z
Tim
e
Time
Fig. 1 SWATH MS. (a) While peptides elute from the LC, the mass spectrometer cycles through a large m/z
range (here from 400 to 1200 m/z) and co-fragments all peptide ions in relatively large isolation windows (or
“swathes”) of several m/z (here 25 m/z) [6]. For each isolation window, the resulting fragment ions are
recorded in a high-resolution composite mass spectrum (MS2). For every isolation window, the instrument
accumulates fragment ions for 100 ms, for 32 isolation windows covering 400–1200 m/z this results in a cycle
time of 3.2 s. At the beginning of each cycle a survey scan (MS1) is recorded (not shown). (b) In SWATH MS,
intensities of specific fragment ions are extracted from the highly multiplexed fragment ion spectra in a tar-
geted manner to produce extracted ion chromatogram peak groups
MSConvert
or qtofpeakpicker
(ProteoWizard)
OpenSWATH
(OpenMS)
Peak group identification
SWATH assay
Identified
peak groups
SWATH assay library
for iRT peptides
pyProphet
Scored
peak groups
TRIC Alignment
Cross-run
alignment
(msproteomicstools)
Aligned output
2 Materials
Any files described in this chapter, including example files to test the
protocol, can be downloaded from the following website: http://
www.peptideatlas.org/PASS/PASS00779. Most of the example files
are related to a recent study in Mycobacterium tuberculosis that used
SWATH MS [13]. The installation of software tools is described for a
64-bit Microsoft Windows 7 system. However, many of the tools can
be installed on other platforms, such as Mac OS X and Linux, as well
(OpenMS, Python, pyProphet, TRIC aligner). Specific software ver-
sions used to develop this protocol are noted in the SoftwareVersions.
txt file available at the above mentioned website. It is important to
have enough free disk space available as the data files and the interme-
diary analysis files tend to be very large (dozens of GB).
1. SWATH data files. A detailed protocol of how to obtain high-
quality SWATH MS data on a TripleTOF instrument is pro-
vided in Chapter 16 by Hunter et al., but other instrument
types can be used as well (see Note 1). See Note 2 for a discus-
sion of parameters that are important during acquisition, such
as LC gradient, SWATH window size, and acquisition time.
Importantly, the samples have to contain retention time calibra-
tion (iRT) peptides at the time they are measured (see Note 3).
In this protocol, we assume that the 11 synthetic iRT peptides
provided in the iRT-kit by Biognosys have been spiked into
each sample [14]. Please note that the AB SCIEX TripleTOF
instrument produces two data files per SWATH run, a .wiff and
a .wiff.scan file, which should always be stored together (see
three .wiff and corresponding .wiff.scan example files).
2. File describing SWATH window setup. The selection of appro-
priate SWATH windows is an important part of a SWATH MS
experiment (Chapter 16 by Hunter et al.). For the current pro-
tocol, we assume that a fixed window setup (32 windows of
26 m/z each, 1 m/z overlap: 400–425 m/z, 424–450 m/z, 449–
475 m/z, … 1149–1175 m/z, 1174–1200 m/z) has been used
for data acquisition (see example file SWATHwindows_acquisi-
tion.tsv). In contrast to the SWATH windows used for data
acquisition, the SWATH windows used for data analysis should
not be overlapping. Therefore, the 1 m/z overlap that was used
for acquisition among the neighboring SWATH windows is
split, such that each window is only 25 m/z (first and last win-
dows are slightly different): 400–424.5 m/z, 424.5–449.5 m/z,
449.5–474.5 m/z, … 1149.5–1174.5 m/z, 1174.5–1200 m/z
(see example file SWATHwindows_analysis.tsv).
294 Hannes L. Röst et al.
3 Methods
3.1 Visualization Before proceeding with automated SWATH data analysis, it is cru-
of Selected Peptides cial to ensure adequate quality of the raw data. Skyline offers great
for Quality Control visualization options to assess SWATH data quality and we recom-
mend loading all data files to be subjected to automated analysis by
OpenSWATH first into Skyline to manually inspect a few quality
control peptides, such as the spiked-in iRT peptides. Optimally,
296 Hannes L. Röst et al.
every SWATH data file is loaded into Skyline right after acquisition
to monitor LC stability as well as performance of the mass
spectrometer.
1. Open a blank Skyline document and go to Settings → Peptide
Settings → “Filter” tab. Uncheck the option to “Auto-select all
matching peptides” and click “OK”.
2. Go to Settings → Transition Settings → “Filter” tab. Uncheck the
option to “Auto-select all matching transitions” and click “OK”.
3. Go to Settings → Transition Settings → “Full-Scan” tab. Set the
following parameters: Acquisition method: DIA; Product mass
analyzer: TOF; Isolation scheme: From the drop-down menu,
select “Add…” and fill the “Edit Isolation Scheme” window as
shown in Fig. 3a (for this protocol, the window boundaries can
be copy-pasted from the file SWATHwindows_analysis.tsv).
Resolving power: 15,000 (depends on the state of the instru-
ment during SWATH data acquisition); Retention time filter-
ing: Include all matching scans. Click “OK”.
4. Go to Edit → Insert → Transition list. In the opening window,
paste assays of peptides you would like to monitor. For this
protocol, the assays for the iRT peptides can be copy-pasted
from the file iRTassays_Skyline.tsv. Click “Insert”.
5. Save the Skyline file.
6. Go to File → Import → Results and select “Add single-injection
replicates in files” and click “OK”.
7. Select the raw SWATH data .wiff files (the .wiff.scan files are not
showing up here, but need to be located in the same folder) and
click “Open”. When asked about removing the common prefix
of the file name, click “Remove”. Now it will take a while to
import all the data depending on file size and number of files.
8. After the data is loaded, arrange the Skyline file such that it is
convenient to inspect the runs (Fig. 3b). First, get the peak
area and retention time overview plots: View → Retention
Times → Click “Replicate Comparison” and View → Peak
Areas → Click “Replicate Comparison”. Next, arrange the
panels by drag-and-drop to the desired location. Then go to
“Settings” and click “Integrate all”. Finally, right-click on a
chromatogram plot → Auto-zoom X-axis → Best Peak.
9. Inspect the SWATH peak groups to judge performance of LC
(peak shape, retention time stability) and mass spectrometer
(signal intensity and signal-to-noise ratio for low abundant
peaks) over time (Fig. 3b). Note that Skyline uses a scoring
system to determine the correct peak group but it can still hap-
pen that it does not pick the correct one. In these cases, peak
boundaries can be changed manually by click-and-drag in the
retention time axis of the chromatogram plot. A confidently
Fig. 3 SWATH data visualization in Skyline. (a) Screen shot showing how SWATH windows are defined in Skyline.
This is essential for the software to know from which SWATH window the specified fragment ions should be
extracted. (b) Example of how to organize the Skyline window for fast and efficient monitoring of large numbers
of SWATH runs. The example here shows just three runs, but Skyline can easily handle dozens of runs
298 Hannes L. Röst et al.
3.2 Raw Data MSConvert, provided through the ProteoWizard software suite
Conversion Into mzML (Chapter 23 by Mallick et al.), enables conversion of proprietary
raw data files (.wiff for AB SCIEX, .raw files for Thermo Fisher
instruments) into an open format, such as mzML or mzXML.
OpenSWATH can handle both file types, but when running on a
Windows PC, the input files need to be in mzML format.
1. Start the software by opening the Start Menu (Windows icon
in lower left corner), type “MSConvert” in the search field and
click on “MSConvert”.
2. In the MSConvert window, select “List of Files” and use the
“Browse” button to locate the raw files to be converted (the
.wiff and .wiff.scan files need to be located in the same folder).
Click the “Add” button. Choose an output directory using the
second “Browse” button. Then set the following parameters:
Output format: mzML; Extension: mzML; Binary encoding
precision: 64-bit; Write index: yes; Use zlib compression: yes.
All other checkboxes should be unchecked and no filters
should be set (see Fig. 4). While OpenSWATH produces best
results on profile data, it is possible to run it on centroided data
to reduce disk space and execution time (see Note 5).
3.3 Automated OpenSWATH can be run as a single command, which executes all
Analysis steps of the OpenSWATH data analysis pipeline automatically. The
with OpenSWATH individual steps can be controlled through flags passed on the com-
mand line. Use the --helphelp command to see all options and
to learn about particular speed and memory optimizations avail-
able. Table 1 summarizes the most important options.
1. Start the software by opening the Start Menu (Windows icon
in lower left corner), type “TOPP” in the search field and click
on “TOPP command line” which will start a command prompt
preloaded with the necessary paths to execute OpenSWATH.
2. OpenSWATH requires four input files: (1) The actual data in
mzXML or mzML format (on Windows PC only mzML for-
mat is accepted). (2) An assay library containing assays for all
target peptides. OpenSWATH can take assay libraries in both
tab-separated table (tsv) and TraML format. For more details
see Notes 4 and 6. (3) An assay library containing assays for all
iRT peptides (see Note 3) in TraML format. 4) A file with a
SWATH Data Analysis 299
Fig. 4 SWATH data conversion with MSConvert. Conversion of raw data to mzML
format using the MSConvert GUI, showing the individual options that should be
selected during the conversion step for profile-mode conversion
Table 1
OpenSWATH parameters
Parameter Explanation
--helphelp To see all options
-out_chrom Additionally also output the generated XICs. This option is
required if the OpenSWATH results are to be inspected
manually with TAPIR [23] or for other downstream
processing not discussed here
-swath_windows_file Text file specifying the SWATH windows (see example file and
main text for more details)
-rt_extraction_window The size of the retention time extraction window in seconds
(centered around the theoretical retention time calculated
using the iRT alignment), usually 300–600 s are reasonable,
but may be adjusted according to the chromatography and the
accuracy of expected retention times in the assay library
-mz_extraction_window Size of the extraction window in Dalton. The default value of
0.05 may need to be adjusted depending on the instrument
resolution
-ppm Indicates that -mz_extraction_window is given in ppm
instead of Dalton
-Scoring:TransitionGroup Minimal peak width in seconds preventing spurious peaks from
Picker:min_peak_width appearing in the results. Defaults to 14 s
-TransitionGroupPicker: Indicates whether background subtraction should be performed
background_subtraction during quantification. This option may increase the accuracy
of the quantification
-Scoring:TransitionGroup Smoothing parameter for Savitzky-Golay smoothing (expressed
Picker:PeakPicker MRM: in number of scans). Increasing this parameter leads to
sgolay_frame_length stronger smoothing. Alternatively, Gaussian smoothing may be
used by -Scoring:TransitionGroupPicker:Peak
PickerMRM:use_gauss
-Scoring:stop_report_ Determines the number of peak groups per assay that are
after_feature reported in the result. A smaller number produces less output;
generally 5–10 is appropriate
-batchSize Determines the size of the batch of chromatograms that are
loaded into memory and analyzed at once. The smaller this
number, the less memory is required
-sort_swath_maps This parameter ensures that the SWATH maps in the input are
sorted by m/z (this assumes that the swath window file given
in -swath_windows_file also contains windows sorted by
m/z)
-readOptions Setting this option to cache instructs the software to not load
all data into memory at once but to create a local cache of the
data. Make sure to specify a directory for the temporary files
using the -tempDirectory parameter
-tempDirectory Defines the directory to store temporary files (required if
-readOptions cache is used)
SWATH Data Analysis 301
3.4 Error Rate After running OpenSWATH, q-values corresponding to the FDR
Estimation of peak identification can be estimated with the pyProphet soft-
with pyProphet ware tool. The only input for pyProphet is the OpenSWATH result
table file (osw_output.tsv).
1. Use the same command line window as above and enter (on a
single line):
C:\Anaconda2\Scripts\pyprophet.exe
--ignore.invalid_score_columns
--d_score.cutoff=0.5
osw_output.tsv
This command will output a set of files, of which osw_out-
put_with_dscore_filtered.csv will be used in the next step.
2. pyProphet will also output a table containing the number of
true and false positives for different q-value cutoffs (_sum-
mary_stat.csv). A q-value cutoff (or FDR) between 1 and 10 %
is appropriate for most studies. For further information, the
pdf report file should be consulted, which contains plots of the
fitted distributions as well as ROC curves. This step provides
an opportunity to identify common errors, such as using an
assay library without decoys or using an assay library unsuited
for the measured sample (e.g., from another organism). In a
successful run, target and decoy distributions should be clearly
separated as shown in Fig. 5.
Fig. 5 Peak group scoring with pyProphet. To ensure that the FDR estimation step using pyProphet worked
properly, the output should be manually inspected to identify potential problems. The plotted distributions
should be bimodal and the decoy distribution should be close in shape to the false positive distribution (left part
of the bimodal distribution)
302 Hannes L. Röst et al.
Table 2
TRIC alignment parameters
Parameter Explanation
--help To see all options
--method Alignment method, either reference-based alignment (best_overall or
global_best_overall) or reference-free, tree-based alignment
(LocalMST). The method best_overall is most conservative as it
does not change peak groups with a good q-value, thereby avoiding errors
introduced during alignment. Alternatively, global_best_overall
may be chosen, which, however, requires that the maximal shift in
retention time is low, otherwise even good peak groups may be removed
from the output if their retention time is too far off. The method
LocalMST is recommended for large-scale SWATH MS analysis
--realign_method Describes how the alignment between individual runs is performed and
may be either linear (linear or diRT) or nonlinear (lowess or
CVSpline). For small datasets, linear alignment methods typically
perform sufficiently well and are faster than the nonlinear methods
--max_rt_diff Maximal shift in retention time that is tolerated after alignment
(in seconds). The software will automatically estimate a sensible value if
it is set to auto_3medianstdev. Adjust this parameter to your
chromatography. Nonlinear alignment methods typically allow for lower
values than linear alignment methods
--target_fdr Desired q-value (or FDR) on assay level (estimated using decoys)
--max_fdr_quality q-value (or FDR) cutoff to still consider a feature for quantitation.
Typically, a value of 0.01 (1 %) is appropriate. Higher values (e.g., 0.05)
may be appropriate for small to medium size datasets as they increase
completeness of the data matrix, but may also introduce lower quality
peaks groups
SWATH Data Analysis 303
1. Use the same command line window as above and enter (on a
single line):
C:\Anaconda2\python.exe
C:\Anaconda2\Scripts\feature_alignment.py
--in file1_with_dscore.csv file2_with_dscore.
csv file3_with_dscore.csv
--out aligned.tsv
--method best_overall
--realign_method diRT
--max_rt_diff 90
--target_fdr 0.01
--max_fdr_quality 0.01
The command exemplified here will run alignment on three files
using linear iRT alignment and pick an appropriate peak group in
each run within the aligned window using a reference-based align-
ment. Only peptides that pass the identification threshold esti-
mated by the TRIC algorithm will be reported in the final output
(set to 1 % FDR using the --target_fdr parameter in the
example above). Note that recent results suggest substantial
improvements when using reference-free, nonlinear retention time
alignment, which can be enabled using the “LocalMST” tree-
based alignment option and the “lowess” RT correction option.
4 Notes
the required disk space and execution time of the whole work-
flow substantially. As centroiding reduces the number of peaks
in the data and sometimes may remove low-intensity peaks, the
XICs generated by OpenSWATH become less smooth and peak
detection becomes less sensitive. If centroiding is performed, it
is crucial to compare the results of OpenSWATH to those
obtained on profile data on the same dataset to obtain an accu-
rate estimation of how centroiding affects the identification
rate. Multiple centroiding algorithms are freely available,
including the internal msconvert centroiding (enabled during
conversion by the “Filter” settings) as well as the separate exe-
cutable qtofpeakpicker.exe (found in the same folder as
msconvert.exe after installation of ProteoWizard). The
qtofpeakpicker has been specifically designed for data derived
from QTOF instruments and we recommend to use it to cen-
troid TripleTOF data [10]. The qtofpeakpicker is run directly
on the .wiff files and we recommend using the same parame-
ters as Schubert et al. to generate centroided mzML files
[10]: --resolution=20000 --area=1 --thresh-
old=1 --smoothwidth=1.1.
6. A high-quality SWATH assay library is an essential feature for any
targeted analysis of DIA or SWATH MS data. The OpenSWATH
software uses this information to extract XICs for the targeted
peptide from the appropriate fragment ion spectra and to score
the peaks according to how well they match the information in
the assay library (expected retention time, fragment ion m/z,
etc.). Alongside the assays for the proteins of interest (target
assays), the library needs to contain a set of “decoy assays” which
represent peptides not present in the sample (usually shuffled tar-
get peptide sequences). These assays are scored in the same fash-
ion as the targeted assays and used by pyProphet to estimate the
null distribution of different OpenSWATH scores and compute
q-values (for FDR control). Having decoys in the assay library is
essential for this step to work. Repositories such as http://www.
SWATHAtlas.org provide assay libraries that already contain
decoys. Please refer to Röst et al. and Schubert et al. for more
information on the scoring algorithm and the importance of
decoys and FDR control [9, 10]. OpenSWATH accepts assay
libraries in both tab-separated table (tsv) and TraML format. For
large assay libraries, the tsv format is more memory efficient than
the TraML format. OpenMS comes with the tools
ConvertTraMLToTSV and ConvertTSVToTraML, which
enable conversion between the two formats (regardless of the
names of these conversion tools, the file ending for the table for-
mat has to be.csv, while the actual data format is tab-separated).
306 Hannes L. Röst et al.
References
1. Domon B (2012) Considerations on selected for creating and analyzing targeted proteomics
reaction monitoring experiments: implications experiments. Bioinformatics 26:966–968.
for the selectivity and accuracy of measure- doi:10.1093/bioinformatics/btq054
ments. Proteomics Clin Appl 6:609–614. 13. Schubert OT, Ludwig C, Kogadeeva M et al
doi:10.1002/prca.201200111 (2015) Absolute proteome composition and
2. Picotti P, Aebersold R (2012) Selected reaction dynamics during dormancy and resuscitation of
monitoring-based proteomics: workflows, Mycobacterium tuberculosis. Cell Host
potential, pitfalls and future directions. Nat Microbe. doi:10.1016/j.chom.2015.06.001
Methods 9:555–566. doi:10.1038/ 14. Escher C, Reiter L, MacLean B et al (2012)
nmeth.2015 Using iRT, a normalized retention time for
3. Picotti P, Bodenmiller B, Mueller LN et al more targeted measurement of peptides.
(2009) Full dynamic range proteome analysis of Proteomics 12:1111–1121. doi:10.1002/
S. cerevisiae by targeted proteomics. Cell pmic.201100463
138:795–806. doi:10.1016/j.cell.2009.05.051 15. Kohlbacher O, Reinert K, Gröpl C et al (2007)
4. Venable JD, Dong M-Q, Wohlschlegel J et al TOPP--the OpenMS proteomics pipeline.
(2004) Automated approach for quantitative Bioinformatics 23:e191–7. doi:10.1093/bio-
analysis of complex peptide mixtures from tan- informatics/btl299
dem mass spectra. Nat Methods 1:39–45. 16. Sturm M, Bertsch A, Gröpl C et al (2008)
doi:10.1038/nmeth705 OpenMS – an open-source software framework
5. Chapman JD, Goodlett DR, Masselon CD for mass spectrometry. BMC Bioinformatics
(2013) Multiplexed and data-independent tan- 9:163. doi:10.1186/1471-2105-9-163
dem mass spectrometry for global proteome 17. Röst HL, Schmitt U, Aebersold R, Malmström
profiling. Mass Spectrom Rev. doi:10.1002/ L (2014) pyOpenMS: a Python-based interface
mas.21400 to the OpenMS mass-spectrometry algorithm
6. Gillet LC, Navarro P, Tate S et al (2012) Targeted library. Proteomics 14:74–77. doi:10.1002/
data extraction of the MS/MS spectra generated pmic.201300246
by data-independent acquisition: a new concept 18. Röst HL, Schmitt U, Aebersold R, Malmström
for consistent and accurate proteome analysis. L (2015) Fast and efficient XML data access
Mol Cell Proteomics 11:O111.016717. for next-generation mass spectrometry. PLoS
doi:10.1074/mcp.O111.016717 One 10:e0125108. doi:10.1371/journal.
7. Gallien S, Duriez E, Crone C et al (2012) pone.0125108
Targeted proteomic quantification on 19. Junker J, Bielow C, Bertsch A et al (2012)
quadrupole- orbitrap mass spectrometer. Mol TOPPAS: a graphical workflow editor for the
Cell Proteomics 11:1709–1723. doi:10.1074/ analysis of high-throughput proteomics data.
mcp.O112.019802 J Proteome Res 11:3914–3920. doi:10.1021/
8. Peterson AC, Russell JD, Bailey DJ et al (2012) pr300187f
Parallel reaction monitoring for high resolu- 20. Aiche S, Sachsenberg T, Kenar E et al (2015)
tion and high mass accuracy quantitative, Workflows for automated downstream data
targeted proteomics. Mol Cell Proteomics.
analysis and visualization in large-scale computa-
doi:10.1074/mcp.O112.020131 tional mass spectrometry. Proteomics 15:1443–
9. Röst HL, Rosenberger G, Navarro P et al 1447. doi:10.1002/pmic.201400391
(2014) OpenSWATH enables automated, tar- 21. Teleman J, Röst HL, Rosenberger G et al
geted analysis of data-independent acquisition (2014) DIANA-algorithmic improvements for
MS data. Nat Biotechnol 32:219–223. analysis of data-independent acquisition MS
doi:10.1038/nbt.2841 data. Bioinformatics. Oxford, England.
10. Schubert OT, Gillet LC, Collins BC et al (2015) doi:10.1093/bioinformatics/btu686
Building high-quality assay libraries for targeted 22. Reiter L, Rinner O, Picotti P et al (2011)
analysis of SWATH MS data. Nat Protoc mProphet: automated data processing and sta-
10:426–441. doi:10.1038/nprot.2015.015 tistical validation for large-scale SRM experi-
11. Röst HL, Liu Y, D’Agostino G et al (2016) ments. Nat Methods 8:430–435. doi:10.1038/
TRIC: an automated alignment strategy for nmeth.1584
reproducible protein quantification in targeted 23. Röst HL, Rosenberger G, Aebersold R,
proteomics. Nat Methods 13:777–783. Malmström L (2015) Efficient visualization of
doi:10.1038/nmeth.3954 high-throughput targeted proteomics experi-
12. MacLean B, Tomazela DM, Shulman N et al ments: TAPIR. Bioinformatics. Oxford,
(2010) Skyline: an open source document editor England. doi:10.1093/bioinformatics/btv152
SWATH Data Analysis 307
24. Malmström L, Bakochi A, Svensson G et al 27. Parker SJ, Röst HL, Rosenberger G et al
(2015) Quantitative proteogenomics of human (2015) Identification of a set of conserved
pathogens using DIA-MS. Proteomics 129:98– eukaryotic internal retention time standards for
107. doi:10.1016/j.jprot.2015.09.012 data-independent acquisition mass spectrome-
25. Bruderer R, Bernhardt OM, Gandhi T et al try. Mol Cell Proteomics 14:2800–2813.
(2015) Extending the limits of quantitative doi:10.1074/mcp.O114.042267
proteome profiling with data-independent 28. Rosenberger G, Koh CC, Guo T et al (2014) A
acquisition and application to acetamino- repository of assays to quantify 10,000 human
phen treated 3D liver microtissues. Mol proteins by SWATH-MS. Sci Data 1:140031.
Cell Proteomics. doi:10.1074/mcp.M114. doi:10.1038/sdata.2014.31
044305 29. Selevsek N, Chang C-Y, Gillet LC et al (2015)
26. Egertson JD, Kuehn A, Merrihew GE et al Reproducible and consistent quantification of
(2013) Multiplexed MS/MS for improved the Saccharomyces cerevisiae proteome by
data-independent acquisition. Nat Methods SWATH-MS. Mol Cell Proteomics 14:739–
10:744–746. doi:10.1038/nmeth.2528 749. doi:10.1074/mcp.M113.035550
Chapter 21
Abstract
Laboratories tend to be amenable environments for long-term reliable operation of scientific measurement
equipment. Indeed, it is not uncommon to find equipment 5, 10, or even 20+ years old still being rou-
tinely used in labs. Unfortunately, the Achilles heel for many of these devices is the control/data acquisi-
tion computer. Often these computers run older operating systems (e.g., Windows XP) and, while they
might only use standard network, USB or serial ports, they require proprietary software to be installed.
Even if the original installation disks can be found, it is a burdensome process to reinstall and is fraught
with “gotchas” that can derail the process—lost license keys, incompatible hardware, forgotten configura-
tion settings, etc. If you have running legacy instrumentation, the computer is the ticking time bomb
waiting to put a halt to your operation.
In this chapter, I describe how to virtualize your currently running control computer. This virtualized
computer “image” is easy to maintain, easy to back up and easy to redeploy. I have used this multiple times
in my own lab to greatly improve the robustness of my legacy devices.
After completing the steps in this chapter, you will have your original control computer as well as a
virtual instance of that computer with all the software installed ready to control your hardware should your
original computer ever be decommissioned.
Key words Legacy hardware, Virtual computers, System reliability, Mass Spectrometry, VirtualBox,
Cloning, Systems management
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_21, © Springer Science+Business Media LLC 2017
309
310 Jonathan E. Katz
2 Materials
3 Methods
Fig. 1 Booting from secondary media. Panel (a) shows a representative example of a splash screen when a
computer is powered on. Pressing F12, as prompted by this screen brings up the menu shown in panel (b).
From this menu, the system can be booted from prepared USB flash drives, removable CD drives, or other
media
Fig. 2 Booting into OSFClone. OSFClone is built on an operating system with very rudimentary drivers. As such,
it is entirely keyboard controlled. In this screen shot, the user was prompted to press enter at the “boot:”
prompt and then asked to press the space bar to boot without further defining the video card
prompted for and press enter). Image will copy the entire
hard drive into a file (in contrast to “clone” which would
duplicate one device onto another device). You may be
presented with options of which “format you wish to
use” and you should select “dd” to perform this imaging
operation (i.e., type 1 and press the enter key).
(j) You will now need to select a source and destination. See
Fig. 3 as an example.
●● Type 1 and press enter to select the source; most
likely it will be “/dev/hda"—confirm this on your
system; you will be able to confirm this based on the
316 Jonathan E. Katz
reported size (it will match the size of your “c:” drive).
Return to the option menu.
●● Type 2 and press enter to select the destination; most
likely it will be “/dev/sda1"—confirm this on your
system; you will be able to confim this based on the
reported size (it will most likely be the largest size
available). Note that for the destination of an image
operation, the partition will be mounted and a file will
be created in that partition.
Fig. 3 Setting Source and Destination in OSFClone. OSFClone uses a command line interface to set all the
parameters. Initially, to set the source, type “1” and press enter (as shown in panel (a)). You will be presented
with a screen similar to panel (b). Most likely you will select the option corresponding to /dev/hda—confirm
this based on the device size. Then you will be back at panel (a) and you can select option 2 for the destination.
Presented with a screen similar to Panel (c), you will probably select the option for /dev/sda1 or /dev/sdb1;
confirm this by matching the size of your external hard drive to the option you wish to select
Legacy Instrumentation Virtualization 317
Fig. 4 Converting the dd image file into a VirtualBox image file. Panel (a) demonstrates how to open a com-
mand line to type commands. Press the “Start” button in the lower left, type “cmd” and press enter. Panel (b)
demonstrates how to copy the complete path for the command line from within Windows Explorer. Panel (c)
shows a sample of how the command window looks when executing the VBoxManage command under
Windows 7
Fig. 5 Installing Guest Additions in Your Virtual Machine. Within the started virtual machine, select Devices-
>Insert Guest Additions CD. Then install the guest additions as prompted
4 Notes
Acknowledgments
References
Abstract
Data quality assessment is important for reproducibility of proteomics experiments and reusability of pro-
teomics data. We describe a set of statistical tools to routinely visualize and examine the quality control
(QC) metrics obtained for raw LC-MS/MS data on different instrument types and mass spectrometers.
The QC metrics used here are the identification free QuaMeter metrics. Statistical assessments introduced
include (a) principal component analysis, (b) dissimilarity measures, (c) T2-chart for quality control, and
(d) change point analysis. We demonstrate the workflow by a step-by-step assessment of a subset of Study
5 for the Clinical Proteomics Technology Assessment for Cancer (CPTAC) using our R functions.
Key words Abnormal experiment, Change point, Dissimilarity, Euclidean distance, MS/MS pro-
teomics, Principal component analysis, Quality control
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_22, © Springer Science+Business Media LLC 2017
325
326 Xia Wang
2 Materials
Table 1
46 metrics produced in the “IDFree” mode of QuaMeter
Filename What is the name of the file from which the metrics were computed?
StartTimeStamp At what time did acquisition begin for this experiment?
XIC-WideFrac What fraction of precursor ions account for the top half of all peak width?
XIC-FWHM-Q1 What is the 25%ile of peak widths for the wide XICs?
XIC-FWHM-Q2 What is the 50%ile of peak widths for the wide XICs?
XIC-FWHM-Q3 What is the 75%ile of peak widths for the wide XICs?
XIC-Height-Q2 The log ratio for 50%ile of wide XIC heights over 25%ile of heights.
XIC-Height-Q3 The log ratio for 75%ile of wide XIC heights over 50%ile of heights.
XIC-Height-Q4 The log ratio for maximum of wide XIC heights over 75%ile of heights.
RT-Duration What is the highest scan time observed minus the lowest scan time
observed?
RT-TIC-Q1 The interval when the first 25 % of TIC accumulates divided by
RT-Duration.
RT-TIC-Q2 The interval when the second 25 % of TIC accumulates divided by
RT-Duration.
RT-TIC-Q3 The interval when the third 25 % of TIC accumulates divided by
RT-Duration.
RT-TIC-Q4 The interval when the fourth 25 % of TIC accumulates divided by
RT-Duration.
RT-MS-Q1 The interval for the first 25 % of all MS events divided by RT-Duration.
RT-MS-Q2 The interval for the second 25 % of all MS events divided by RT-Duration.
RT-MS-Q3 The interval for the third 25 % of all MS events divided by RT-Duration.
RT-MS-Q4 The interval for the fourth 25 % of all MS events divided by RT-Duration.
RT-MSMS-Q1 The interval for the first 25 % of all MS/MS events divided by
RT-Duration.
RT-MSMS-Q2 The interval for the second 25 % of all MS/MS events divided by
RT-Duration.
(continued)
328 Xia Wang
Table 1
(continued)
RT-MSMS-Q3 The interval for the third 25 % of all MS/MS events divided by
RT-Duration.
RT-MSMS-Q4 The interval for the fourth 25 % of all MS/MS events divided by
RT-Duration.
MS1-TIC-Change-Q2 The log ratio for 50%ile of TIC changes over 25%ile of TIC changes.
MS1-TIC-Change-Q3 The log ratio for 75%ile of TIC changes over 50%ile of TIC changes.
MS1-TIC-Change-Q4 The log ratio for largest TIC change over 75%ile of TIC changes.
MS1-TIC-Q2 The log ratio for 50%ile of TIC over 25%ile of TIC.
MS1-TIC-Q3 The log ratio for 75%ile of TIC over 50%ile of TIC.
MS1-TIC-Q4 The log ratio for largest TIC over 75%ile TIC.
MS1-Count How many MS scans were collected?
MS1-Freq-Max What was the fastest frequency for MS collection in any minute? (Hz)
MS1-Density-Q1 What was the 25%ile of MS scan peak counts?
MS1-Density-Q2 What was the 50%ile of MS scan peak counts?
MS1-Density-Q3 What was the 75%ile of MS scan peak counts?
MS2-Count How many MS/MS scans were collected?
MS2-Freq-Max What was the fastest frequency for MS/MS collection in any minute? (Hz)
MS2-Density-Q1 What was the 25%ile of MS/MS scan peak counts?
MS2-Density-Q2 What was the 50%ile of MS/MS scan peak counts?
MS2-Density-Q3 What was the 75%ile of MS/MS scan peak counts?
MS2-PrecZ-1 What fraction of MS/MS precursors is singly charged?
MS2-PrecZ-2 What fraction of MS/MS precursors is doubly charged?
MS2-PrecZ-3 What fraction of MS/MS precursors is triply charged?
MS2-PrecZ-4 What fraction of MS/MS precursors is quadruply charged?
MS2-PrecZ-5 What fraction of MS/MS precursors is quintuply charged?
MS2-PrecZ-more What fraction of MS/MS precursors is charged higher than +5?
MS2-PrecZ-likely-1 What fraction of MS/MS precursors lack known charge but look like +1 s?
MS2-PrecZ-likely-multi What fraction of MS/MS precursors lack known charge but look like >+1 s?
3 Methods
3.1 Initial Data A timeline plot of the data files, grouped by participating laborato-
Examination ries, individual instruments, sample types, or other features, helps
Statistical Assessment of QC Metrics 329
visualize the data files in the order that they were produced in the
study. The metric “StartTimeStamp” is used to extract starting
time of the experiment for a raw data file.
Many of the QuaMeter metrics are based on quartiles of a
quantity. For example, XIC-FWHM-Q1, XIC-FWHM-Q2, and
XIC-FWHM-Q3 are the 25%ile, 50%ile, and 75%ile of peak widths
for the wide XICs. This helps to provide a rough approximation of
the distribution of the peak widths for the wide XICs. However, it
also brings in potential high correlation among these measures. In
practice, we usually remove any metric that shows correlation
greater than 0.98 or less than −0.98 with the other metric(s). The
analysis loses little information from this removal because one can
predict the values for these removed variables almost perfectly
from the retained information. Also, metrics that are related to
directly configurable options, such as MS2-PrecZ-* and
RT-Duration, are also omitted in the statistical analysis (see Note 3
initial data examination).
3.2 Principal Principal component analysis (PCA) helps to visualize the multidi-
Component Analysis mensional quality control (QC) metrics in lower dimension. It
makes the interpretation easier by transform correlated metrics to
uncorrelated linear combinations of the metrics. To reduce impact
by any extreme data files, we use robust PCA method [7]: the vari-
ance–covariance matrix of the QC metrics as well as the center of
each metrics are estimated by cov.rob function in R and are used in
the PCA analysis (cov.rob function in R). The PCA analysis is car-
ried out by princomp R function using the QC metrics normalized
by the individual metric’s variance. The results contain PC scores
matrix (linear combinations of original metrics), whose columns
corresponding to the dimension of the QC metrics that are ana-
lyzed and whose rows are the newly transformed QC metrics for
each data file (see Note 3 principal component analysis for the
example). The data structure can then be visually examined by sys-
tematically exploring two- or three-dimensional plots of different
combinations of PC scores [8]. The one used most frequently is
the plot based on the first two PCs, which account for the largest
proportion of variance in the data. A more complete examination
is needed for a comprehensive view of the data structure.
3.3 Dissimilarity This step is to exploratorily examine how similar two experiments
Measures are in terms of the QC metrics. If a spectrometer has been operat-
ing stably, it is expected that QC metrics obtained from all the
experiments on this spectrometer should be randomly distributed
around a “true” stable status, and thus the distance between each
other is approximately equal. If an experiment is “far away” from
the rest of the experiments based on the QC metrics, it reflects the
overall difference in the QC metrics and thus implies this experi-
ment may have been operated under quite different conditions
330 Xia Wang
+ ( x12 - x 22 ) + ( x1 p - x 2 p ) .
2
( x11 - x 21 )
2 2
3.4 T2-Chart For experiment i with the p-dimensional PC scores xi, a T2 statistic
for Quality Control is computed as
Ti 2 = ( xi 1 - x1 ) / l1 + ( xi 2 - x 2 ) / l2 + + ( xip - x p ) / lp ,
2 2 2
where the p-dimensional vector x is the median of the p QC met-
rics and λj is variance of the jth PC component. The statistic Ti2 is
a normalized Euclidean distance. Since xi’s are PC scores and the
variance–covariance matrix is a diagonal matrix with λj, j = 1, …, p ,
as the diagonal element, it is assumed that Ti2 approximately fol-
lows a χ2-distribution with p degrees of freedom [9]. The upper
and lower control limits can be obtained at a given significance
level α as cutoff between in-control and out-of-control experi-
ments. While the comparison based on the dissimilarity measure is
more exploratory, this statistic provides a more rigorous evaluation
of the abnormal experiments with a cutoff significance level. It may
be impacted, however, by data files from extreme experiments.
3.5 Change Point Change point analysis is a tool to study the potential batch effects
Analysis in experiments. The multidimensional QC metrics is summarized
in the normalized p-dimensional Euclidean distance (Ti2). The
batch effects may be caused by the changes in the mean of Ti2, or
in the variability of Ti2, or both. With a smaller number of experi-
mental runs on each spectrometer, we only assume changes in the
mean. The R function cpt.mean in the R library changepoint is
used in studying the batch effects in the mean level [10]. When
Statistical Assessment of QC Metrics 331
4 Notes
ChromatogramMzUpperOffset = "1.5mz"
Instrument = "LTQ"
QqTOF configuration:
ChromatogramMzLowerOffset = "0.1mz"
ChromatogramMzUpperOffset = "0.1mz"
Instrument = "orbi"
#Here, “Orbi” means
“can resolve isotopes.”
3. Demonstration using CPTAC Study 5.
Read in the QuaMeter data. There were a total six .tsv files for
QuaMeter metrics with the 46 metrics. The sample type, the
spectrometer instrument type, and a label for the spectrometer
number were added. Thus the “QuaMeter.csv” file had a total
of 50 columns with the first column as the index number. In
this demonstration, we only examined 30 runs from Sample
1B, 5 runs for each spectrometer.
Initial data examination. Figure 1 shows the timeline plot for
all the data files in CPTAC Study 5. We first removed
RT-Duration and MS2.PrecZ.* as they are operating condi-
tions set by the operator. We then used cor R function to detect
and removed metrics with high correlations with other met-
rics. At the end, 27 numeric metrics, excluding Filename and
StartTimeStamp, were used in the following statistical analysis.
These 27 numeric metrics were:
XIC (5): XIC.WideFrac, XIC.FWHM.Q2, XIC.Height.Q2,
XIC.Height.Q3, XIC.Height.Q4
RT (11): RT.TIC.Q1, RT.TIC.Q2, RT.TIC.Q3, RT.TIC.Q4,
RT.MS.Q1, RT.MS.Q2, RT.MS.Q3, RT.MS.Q4, RT.
MSMS.Q1, RT.MSMS.Q2, RT.MSMS.Q3
MS1 (9): MS1.TIC.Change.Q2, MS1.TIC.Change.Q3, MS1.
TIC.Change.Q4, MS1.TIC.Q2, MS1.TIC.Q3, MS1.TIC.
Q4, MS1.Count, MS1.Freq.Max, MS1.Density.Q1, MS1.
Density.Q2, MS1.Density.Q3
MS2 (2): MS2.Freq.Max, MS2.Density.Q1
Principal component analysis. The R object metrics.pca con-
tained the 27 metrics for the 30 experiment runs. The follow-
ing R codes carried out robust PCA analysis.
metrics.pca<-metrics.1[,5:(ncol(metrics.1)-3)]
robust.cov<-cov.rob(metrics.pca)
robust.cor<-cov2cor(robust.cov$cov)
robust.cor.1<-robust.cov
robust.cor.1$cov<-robust.cor
metrics.pca.1<-metrics.pca
for ( i in 1:ncol(metrics.pca)){
Statistical Assessment of QC Metrics 333
1B
3A
OrbiW56
3B
OrbiP65
Orbi86
LTQc65
LTQ295
LTQ73
StartTimeStamp
Fig. 1 Timeline for experiments in CPTAC Study 5. It illustrates the run times for each LC-MS/MS experiment
in the course of Study 5 across six spectrometers. The run order was prescribed by the SOPs under which
Study 5 was conducted. Sample 1B: digested NCI-20 mixture; Samples 3A: yeast; 3B: yeast + BSA
robust.cor.1$center[i]<-robust.cov$center[i]/sqrt(robust.
cov$cov[i,i])
metrics.pca.1[,i]<-metrics.pca[,i]/sqrt(robust.cov$cov[i,i])
}
print(summary(pc.cr <- princomp(metrics.pca.1, covmat =
robust.cor.1, scores=T)))
Figure 2 shows the PC plot for the first two principal compo-
nents. The experimental runs from the same spectrometer
tended to cluster together, and LTQ and Orbi instruments
tended to cluster together, along with some potential abnormal
experimental runs, such as the one in LTQ73.
Dissimilarity measures. For pairwise comparison, we used run1
to represent the p-dimensional PC coordinate for one experi-
ment and run2 to represent that of the other experiment. The
dissimilarity measures were then calculated in R as distance
<-sqrt(sum((run1-run2)^2)). Figure 3 shows all the values of
dissimilarity measure grouped by spectrometers. Clearly, there
was a cluster of large distances on LTQ73, which were all
334 Xia Wang
0
LTQ295
LTQc65
Orbi86
OrbiP65
−2
OrbiW56
−4
−6
−6 −4 −2 0 2 4
The First Principal Component
Fig. 2 PCA plots based on the first two principal components of quality metrics
for experiments on Sample 1B on the six spectrometers
6
Euclidean distance
1 2 3 4 5
LTQc65 OrbiW56
40
0
LTQ73 OrbiP65
40
*
30
30
20
T2
20
10
10
0
LTQ295 Orbi86
40
30
30
20
20
10
10
0
1 2 3 4 5
date
Fig. 4 T2 statistic and change point analysis results for the six spectrometers. The dots represent the values of
T2 statistic with blue dots for “in-control” experiments, pink dots for “out-of-control” experiments, and “*” for
those with large dissimilarity measures. If a pink dot is circled by a black circle, then it is not included in the
change point study (as the one for OrbiW56). The red squares and green diamonds distinguish the two batches
of the experiments runs, if there is a change point
References
1. Bell AW, Deutsch EW, Au CE et al (2009) A 4. Chambers MC, Maclean B, Burke R et al (2012)
HUPO test sample study reveals common A cross-platform toolkit for mass spectrometry
problems in mass spectrometry-based pro- and proteomics. Nat Biotechnol 30:918–920
teomics. Nat Methods 6:423–430 5. Wang X, Chambers MC, Vega-Montoto LJ
2. Mann M (2009) Comparative analysis to guide et al (2014) QC metrics from CPTA raw
quality improvements in proteomics. Nat LC-MS/MS data interpreted through multi-
Methods 6:717–719 variate statistics. Anal Chem 86(5):2497–2509
3. Ma ZQ, Polzin KO, Dasari S (2012) QuaMeter: 6. R Core Team (2015) R: a language and envi-
multivendor performance metrics for LC–MS/ ronment for statistical computing. R
MS proteomics instrumentation. Anal Chem Foundation for Statistical Computing, Vienna,
84(14):5845–5850 Austria. http://www.R-project.org/
Statistical Assessment of QC Metrics 337
7. Rousseeuw PJ, van Zomeren BC (1990) 10. Killick R, Eckley IA (2014) changepoint: an R
Unmasking multivariate outliers and leverage package for changepoint analysis. J Stat Softw
points. J Am Stat Assoc 85:633–639 58(3):1–19
8. Ringnér M (2008) What is principal compo- 11. Tabb DL, Vega-Montoto L, Rudnick PA et al
nent analysis? Nat Biotechnol 26(3):303–304 (2010) Repeatability and reproducibility in
9. Johnson RA, Wichern DW (2007) Applied proteomic identifications by liquid
multivariate statistical analysis. Pearson Prentice chromatography- tandem mass spectrometry.
Hall, Upper Saddle River J Proteome Res 9:761–776
Chapter 23
Abstract
Recent advances in proteome informatics have led to an explosion in tools to analyze mass spectrometry
data. These tools operate across the analysis pipeline doing everything from assessing quality control to
matching peptides to spectra to quantification. Unfortunately, the vast majority of these tools are not able
to operate directly on the proprietary formats generated by the diverse mass spectrometers. Consequently,
the first step in many protocols is the conversion of data from vendor-specific binary files to open-format
files. This protocol details the use of ProteoWizard’s msConvert and msConvertGUI software for this
conversion, taking format features, coding options, and vendor particularities into account. We specifically
describe the various options available when doing conversions and the implications of each option.
Key words Proteomics, Proteome informatics, Data conversion, Open formats, mzML, mzXML,
ProteoWizard
1 Introduction
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6_23, © Springer Science+Business Media LLC 2017
339
340 Ravali Adusumilli and Parag Mallick
2 Materials
2.1 Installation 1. Make sure you have installed Microsoft .NET Framework 3.5
of ProteoWizard SP1 and 4.0.
(Includes Both (a) MS .NET Framework 3.5 SP1 can be found at: http://
msConvert, www.microsoft.com/en-us/download/details.aspx?id=22
msConvertGUI and (b) MS .NET Framework 4 can be found at: http://www.
msPicture): microsoft.com/en-us/download/details.aspx?id=17851
2. Download the latest version of proteoWizard from: http://
proteowizard.sourceforge.net/
(a) We typically recommend the Windows 64-bit installer for
most users.
(b) Upon download, double-click the .msi file to install.
(c) proteoWizard will typically be installed into a folder ‘C:\
Program Files\ProteoWizard\ProteoWizard 3.0.DDDD’
PWIZ Conversion 341
3 Methods
3.1 Using 1. Navigate to the proteowizard directory (Fig. 1) (For ex: C:\
msConvertGUI Program Files\ProteoWizard\ProteoWizard 3.0.8708)
to Convert the File 2. Start the msConvertGUI application
Format of Data Files Double-click on “msConvertGUI” application to open. This
should open the msconvert GUI application. Visually,
msconvertGUI is divided into four boxes (Fig. 2), which are
described in greater detail below.
(a)
Select input files and output files directory location
(Top left).
Fig. 2 MSConvertGUI
Select the output file encoding type. The output file can be
encoded as 32-bit (default—smaller file size) or 64-bit.
(d) Write index:
Type: Checkbox
Default: Checked
Check this box to write index to the output file (see Note 1).
(e) Use zlib compression:
Type: Checkbox
Default: Checked
Check this box to “zlib” compress the output file (see Note 2).
(f) TPP compatibility:
Type: Checkbox
PWIZ Conversion 347
Default: Checked
Check this box to make the output file “TPP compatible.”
(g) Package in gzip:
Type: Checkbox
Default: NOT Checked
Check this box to “gzip” the output file (see Note 3).
6. Select Filter Options (Figs. 5, 6 and 7) (see Note 4)
Filters:
Type: Dropdown list
Default: MS Level
Select the filter to apply to the output file. The following filters
can be selected:
(a) MS Level (Default)
(b) Peak Picking (see Note 5)
( c) Zero Samples
(d) ETD Peak Filter
(e) Threshold Peak Filter
( f) Charge State Predictor
(g) Activation
(h) Subset
Click “Add” to add the settings selected from step a and/
or b. Alternately, remove the settings added by clicking
“Remove”. View the selected filters and parameters in the box.
PWIZ Conversion 349
Fig. 10 msconvert commandline using -f (or --filelist) to convert multiple files together
4 Notes
1. A file index allows for more rapid scan level access of the file.
The index itself makes files slightly larger. However, instead of
having to read a file from beginning to end, it can now be read
from random positions.
2. zlib compression is a seemingly invisible change to the file.
Essentially, the ms spectra themselves are zlib compressed prior
352 Ravali Adusumilli and Parag Mallick
to writing the file. This greatly compresses the file, but also can
slow down its reading. Furthermore, some readers are unable
to handle zlib compressed files.
3. Unlike zlib compression, gzip actually fully gzips the entire
file. This shrink files by about ~20–30 %. Once gzipped, the
files are no longer directly human readable without being ung-
zipped first. ProteWizard and some other tools are able to read
files, even when gzipped. However, they are no longer able to
provide random access. The vast majority of open source tools
are not able to read gzipped files directly.
4. Filters are operations that are performed on spectra to alter
them prior to writing out a converted file. While some filters
are simple (e.g., only writing out MS/MS data), others can be
quite complex and substantially alter your data. The com-
mandLine msConvert has access to many more filters than are
available in the GUI. Please note, that as filters can fundamen-
tally change your data, they may substantially impact down-
stream operations, such as quantification.
5. Peak Picking, is the msConvert nomenclature for “centroid-
ing” which is a very common procedure for shrinking file sizes.
However, this is an irreversible operation that discards a sub-
stantial amount of data. It should be used with caution.
6. msConvert has a large number of available options (Fig. 8):
All the available options are listed on running the command
“msconvert” from the command prompt.
--outfile
●● The --outfile argument lets the user set the name of the
output file. If missing, the output filename will be based on
the given input filename.
-e or --ext
●● This argument can be used to select the extension of the
output filename.
●● Options available: mzML, mzXML, mgf, txt, mz5
●● If none selected, the extension will be based on the output
file conversion format (For example: using argument
--mzXML with msconvert will result in the <output_file-
name> .mzXML).
●● If no output file conversion is selected, then the default as
mentioned earlier will be “mzML”
--mzML
●● The output converted file will be “.mzML” format.
●● This is the default if no other format is selected
--mzXML
●● Very similar to the above “--mzML” format
●● The output file will be created as <output_filename> .
mzXML
--mz5
●● Similar to --mzML and --mzXML options
●● Writes the output file to “mz5” format
--64
●● The “--64” option is used to set the binary encoding to
64-bit precision for the converted file (same as 64bit preci-
sion option on the msconvertGUI)
●● Default if no other encoding flags are selected
--32
●● Like above, this is used to set the binary encoding.
●● As the flag suggests, this option lets us set the output file
conversion to 32-bit encoding precision
●● The m/z and intensity values can also be encoded inde-
pendently using the --mz23 and --inten64 flags.
--mz64
●● The “mz” prefix to the flag “--mz64” refers to m/z values
This flag along with the following “--mz32” flag below
356 Ravali Adusumilli and Parag Mallick
-g or --gzip
●● Use “-g” or “--gzip” option to gzip the entire output file.
●● Selecting this option reduces the file size and adds “.gz” to
the output filename
--filter
●● IMPORTANT: Adding this argument provides the option
to add/set filters before writing the data to output file.
●● Several options are available for filtering the data. All the
available filter types/options are explained in greater detail
under the next section C (FILTER OPTIONS).
●● Each type of filter mentioned in section C should be pre-
ceded by the “--filter” flag.
●● For Example:
–– msconvert C:\Users\Administrator\Desktop\pwizard_
pub\inputFile01.raw
–– --filter “peakPicking true 1-” --filter “threshold bpi-
relative .5 most-intense”
--merge
●● IMPORTANT: This argument/flag provides us the option
to merge the output of multiple input files into a single
output files. This option can be very useful if needed to
merge fractions with each fraction having an individual *.
RAW file.
--simAsSpectra
●● “sim” refers to selected ion-monitoring. This is a useful
option available in case the selected ion-monitoring is
required as spectra and NOT as chromatograms. This is
often used to represent MRM data in mzXML files, which
do not natively support chromatogram data.
--srmAsSpectra
●● “srm” refers to selected reaction monitoring. Like the
“--simAsSpectra” option, the “--srmAsSpectra” flag can be
used to write the selected reaction monitoring as sepctra
and NOT chromatograms. This is often used to represent
MRM data in mzXML files, which do not natively support
chromatogram data.
--combineIonMobilitySpectra
●● Usually, ion mobility spectra are written individually for
each scan. This option can be used to write all drift bins/
scans in a frame/block as one spectrum instead of indi-
vidual spectra.
358 Ravali Adusumilli and Parag Mallick
--acceptZeroLengthSpectra
●● This option lets you accept zero length spectra in case the
file contains empty spectra.
●● Some (but not all) vendor readers have an efficient way of
filtering out empty spectra.
●● NOTE: Takes more time to open the file.
--ignoreUnknownInstrumentError
●● The instrument information should be present in the ven-
dor file usually but in case it is missing, this option can be
useful.
●● If true, if an instrument cannot be determined from a ven-
dor file, it will not be an error
--help
●● Use this option to display all filtering options in detailed
7. The msconvert commandline has additional filtering options
when compared to the msConvertGUI.
For example, msLevel is present in both commandline and
msConvertGUI (as MS Levels on GUI). This filter selects only
spectra with indicated mslevels (ms level 1). Similarly, the ana-
lyzer option in the commandline allows the user to retain the
spectra with the specified mass analyzer (“FTMS” or “ITMS”)
type only.
Note: Filters are applied sequentially in the order that you
list them, and the sequence order can make a large difference
in your output. In particular, the peakPicking filter must be
first in line if you wish to use the vendor-supplied centroiding
algorithms since these use the vendor DLLs, which only oper-
ate on raw untransformed data.
(continued)
●● msLevel
This filter selects only spectra with the indicated <mslevels>.
–– The value set to this argument needs to be an integer.
–– For Example:
# extract MS1 scans only
msconvert C:\Users\Administrator\Desktop\pwizard_pub\
inputFile01.raw
--filter “msLevel 1”
# Alternatively, inside the configuration file from the
msconvert.config file (input as in section 3.2, step 2),
add the filter for “msLevel 1” as:
mzXML=true
zlib=true
filter=“index [3,7]”
filter=“msLevel 1” #For ms1 scans only
●● chargeState
–– This filter keeps spectra that match the listed charge state(s).
–– The value set to this argument needs to be an integer.
–– Both known/single and possible/multiple charge states
are tested. Use 0 to include spectra with no charge state
at all.
●● precursorRecalculation
–– This filter recalculates the precursor m/z and charge for
MS2 spectra.
–– Looks at the prior MS1 scan to better infer the parent mass.
–– Works only on orbitrap and FT data and does not use any
third party (vendor DLL) code.
–– ONLY need to use as a flag. For example:
# Add this line if using config file
filter=“precursorRecalculation”
# If using as a flag to the command line parameter, add the
following to the command:
--filter “precursorRecalculation”
●● mzRefiner
–– This filter recalculates the m/z and charges, adjusting precur-
sors for MS2 spectra and spectra masses for MS1 spectra.
–– It uses an ident file with a threshold field and value to calcu-
late the error and will then choose a shifting mechanism to
correct masses throughout the file. It only works on Orbitrap,
FT, and TOF data. It is designed to work on mzML files.
PWIZ Conversion 361
●● precursorRefine
–– Similar to precursorRecalculation
–– This filter refines the precursor m/z and charge for MS2
spectra. It looks at the prior MS1 scan to better infer the
parent mass. It only works on orbitrap, FT, and TOF data.
It does not use any third party (vendor DLL) code.
●● peakPicking
–– This filter performs centroiding on spectrawith the selected
ms_level argument
peakPicking [<PickerType>[snr=<minimum signal-to-
noise ratio>] [peakSpace=<minimum peak spacing>]
[msLevel=<ms_levels>]]
●● scanNumber
–– This filter takes input as an integer and lets the user select
spectra by scan number
–– Scan numbers are 1-based and not contiguous
●● scanEvent
–– scanEvent filters spectra by scan event and takes input in
the form of a set.
–– This filter offers the flexibility of excluding the selected
scan events as well.
–– For Example:
# All scan event EXCEPT 5 can be included using-
--filter “scanEvent 1-4 6- ”
●● scanTime
–– Spectra can be selected using a specific time range using
this filter.
●● sortByScanTime
–– As the name suggests, the spectra can be a sorted in ascend-
ing order by scan start time.
●● stripIT
–– “IT” in the filter name refers to “ion trap”.
–– This filter removes out the ion trap data spectra with MS
level 1.
●● metadataFixer
–– As the name suggests, this filter can be used to “fix” metadata—
meaning, it adds/replaces a spectra’s TIC/BPI metadata.
–– This filter traverses the m/z intensity arrays to find the sum
and max.
362 Ravali Adusumilli and Parag Mallick
Threshold Threshold
type format Explanation
count Integer (n) Retains intensity values for the selected number of “n” data points. Data
points where intensity is equal to nth intense data point are removed
count-after- Integer (n) Same as the count but retains data points where intensity = nth data
ties point
absolute Flag (no Keep data whose absolute intensity matches threshold
argument)
bpi-relative Flag (no Data is retained if the intensity matches the percentage of base-peak
argument) intensity. 0.75 = 75 %
tic-relative Flag (no Similar to above but for retains data for individual intensities greater
argument) than or less than the percentage of total ion current for the scan.
1 = 100 %
tic-cutoff Flag (no As the name suggests, this is a cutoff. Data is retained UPTO the
argument) percentage of the total ion current for the scan
ms_levels Integer (n) OPTIONAL.
If selected, only scans with MS level = n will be selected
PWIZ Conversion 363
●● mzWindow
–– As the name suggest, this filter is used to provide a range
to select the m/z within the selected window.
–– The m/z values falling ONLY within the selected range
will be retained.
–– Input type: range [mzLow, mzHigh]
–– For Example:
--filter “mzWindow [100.1,307.5]”
●● mzPrecursors
–– This filter allows us the flexibility of selecting spectra within
a given list of precursor m/z values.
–– The input type is a list of m/z precursor value
–– For Example:
--filter “mzPrecursors [123.4,567.8]”
●● defaultArrayLength
–– For this filter, the default array length refers to the range of
peak counts.
–– The name is derived from mzML format file where
“defaultArrayLength” refers to peak list.
–– This can be specified as range of integers
–– For Example:
# Retain only peakcounts >=100:
--filter “defaultArrayLength 100-”
●● zeroSamples
–– Usage: zeroSamples <mode> [<MS_levels>]
–– This is a useful filter to deal with zero values in the spec-
trum. This can be used one of two ways—either remove
spectra with zero values or add the sero value incase the
spectra is missing.
–– Mode options-
i. removeExtra: consecutive zero intensity peaks are
removed from spectra
“100.1,1000 100.2,0 100.3,0 100.4,0 100.5,0 100.6,1030”
would become
“100.1,1000 100.2,0 100.5,0 100.6,1030”
and a peak list
“100.1,0 100.2,0 100.3,0 100.4,0 100.5,0 100.6,1030
100.7,0 100.8,1020 100.9,0 101.0,0”
would become
“100.5,0 100.6,1030 100.7,0 100.8,1020 100.9,0”
364 Ravali Adusumilli and Parag Mallick
●● MS2Denoise
–– Usage: MS2Denoise [<peaks_in_window> [<window_
width_Da> [multicharge_fragment_relaxation]]]
–– Use to denoise, i.e., remove peaks with noise for spectra
with precursor ions.
i. <peaks_in_window>: the number peaks to select in
moving window, default is 6.
ii. <window_width_Da>: the width of the window in
Da, default is 30.
iii. <multicharge_fragment_relaxation>—if “true” (the
default), allows more data below multiply charged
precursors.
●● MS2Deisotope
–– MS2Deisotope [hi_res [mzTol=<mzTol>]] [Poisson
[minCharge=<minCharge>] [maxCharge=<maxCharge>]]
–– Uses the Markey method or a Poisson model to deisotope
the ms2 spectra.
i. hi_res: set to “false” (default) or “true”
●● mzTol: Input format is a decimal value. Sets the
m/z tolerances .Regular default is 0.5, high resolu-
tion default is 0.01
ii. Poisson: Used to define the search range for the
charge. Default is 1
1. minCharge: Default is 1.
2. maxCharge: Default is 3.
–– For Example:
--filter “MS2Deisotope true mzTol=0.4 1 minCharge=1
maxCharge=3”
●● ETDFilter
–– Usage: ETDFilter [<removePrecursor> [<rem-
oveChargeReduced> [<removeNeutralLoss> [<blanket-
Removal> [<matchingTolerance>]]]]]
i. <removePrecursor>—specify “true” to remove unre-
acted precursor (default is “false”)
ii. <removeChargeReduced>—specify “true” to remove
charge reduced precursor (default is “false”)
iii. <removeNeutralLoss>—specify “true” to remove neu-
tral loss species from charge reduced precursor (default
is “false”)
iv. <matchingTolerance>—specify matching tolerance in
MZ or PPM (examples: “3.1 MZ” (the default) or
“2.2 PPM”)
366 Ravali Adusumilli and Parag Mallick
●● chargeStatePredictor
–– Usage: chargeStatePredictor [overrideExistingCharge=
<true|false (false)>] [maxMultipleCharge=<int (3)>]
[minMultipleCharge=<int (2)>] [singleChargeFractionTIC=
<real (0.9)>] [maxKnownCharge=<int (0)>] [makeMS2=
<true|false (false)>]
i. <overrideExistingCharge> : always override existing
charge information (default:“false”)
ii. <maxMultipleCharge> (default 3) and <minMulti-
pleCharge> (default 2): range of values to add to the
spectrum’s existing “MS_possible_charge_state” val-
ues.If these are the same values, the spectrum’s MS_
possible_charge_state values are removed and replaced
with this single value.
iii. <singleChargeFractionTIC>: is a percentage expressed
as a value between 0 and 1 (the default is 0.9, or 90 %).
This is the value used as the previously mentioned ratio
of intensity above and below the precursor m/z.
iv. <maxKnownCharge> (default is 0, meaning no maxi-
mum): the maximum charge allowed for “known”
charges even if override existing charge is false. This
allows overriding junk charge calls like +15 peptides.
v. <algorithmMakeMS2>: default is “false”, when set to
“true” the “makeMS2” algorithm is used instead of
the one described above.
●● turbocharger
–– turbocharger [minCharge=<minCharge>] [maxCharge=
<maxCharge>] [precursorsBefore=<before>] [precursors
After=<after>] [halfIsoWidth=<half-width of isolation
window>] [defaultMinCharge=<defaultMinCharge>]
[defaultMaxCharge=<defaultMaxCharge>] [useVendorPe
aks=<useVendorPeaks>]
–– <maxCharge> (default: 8) and <minCharge> (default 1):
defines range of possible precursor charge states.
–– <before> (default: 2) and <after> (default 0): number of
survey (MS1) scans to check for precursor isotopes, before
and after a MS/MS in retention time.
–– <half-width of isolation window> (default: 1.25): half-
width of the isolation window from which precursor is
derived. Window is centered at target m/z with a total size
of ± the value entered.
–– <defaultMinCharge> (default: 0) and <defaultMaxCharge>
(default: 0): in the event that no isotope is found in the iso-
PWIZ Conversion 367
–– Available Options:
i. “negative”
ii. “positive”
iii. “+”
iv. “−”
References
1. Chambers MC, Maclean B, Burke R, Amodei D, MacCoss M, Tabb DL, Mallick P (2012) A
Ruderman DL, Neumann S, Gatto L, Fischer B, cross-platform toolkit for mass spectrometry and
Pratt B, Egertson J, Hoff K, Kessner D, Tasman proteomics. Nat Biotechnol 30(10):918–920.
N, Shulman N, Frewen B, Baker TA, Brusniak doi:10.1038/nbt.2377
MY, Paulse C, Creasy D, Flashner L, Kani K, 2. Kessner D, Chambers M, Burke R, Agus D,
Moulding C, Seymour SL, Nuwaysir LM, Mallick P (2008) ProteoWizard: open source
Lefebvre B, Kuhlmann F, Roark J, Rainer P, software for rapid proteomics tools develop-
Detlev S, Hemenway T, Huhmer A, Langridge ment. Bioinformatics 24(21):2534–2536.
J, Connolly B, Chadick T, Holly K, Eckels J, doi:10.1093/bioinformatics/btn323
Deutsch EW, Moritz RL, Katz JE, Agus DB,
Printed on acid-free paper
Index
Lucio Comai et al. (eds.), Proteomics: Methods and Protocols, Methods in Molecular Biology, vol. 1550,
DOI 10.1007/978-1-4939-6747-6, © Springer Science+Business Media LLC 2017
369
Proteomics: Methods and Protocols
370 Index
M Proteins����������������� 1, 11, 20, 35, 48, 61, 70, 84, 99, 115, 137,
149, 171, 185, 199, 223, 236, 261, 271, 289, 331, 339
Magnesium����������������������������������������������������������������� 49, 140 Protein extraction���������������������� 1–9, 11, 37, 41, 71, 102, 271
Mammalian cells������������������������������������������������������ 122, 139 Protein glycosylation������������������������������������������������ 100, 101
Mass spectrometry������������������������������������3, 7, 17, 20, 27–31, Protein identification���������������������������� 33, 35, 36, 43, 44, 67,
33, 36, 38, 41, 48, 51, 61–67, 70, 71, 76, 84, 85, 87, 91, 81, 102, 108, 109, 193, 277, 282, 283
93, 94, 100, 105, 109, 115–129, 131, 133, 171, Protein interactions������������ 28, 116, 138, 235–243, 245–247,
185–197, 199–221, 223–229, 231, 232, 272, 274, 276, 249–255, 261–269
284, 289–291, 294, 296, 309, 319, 326, 339 Protein localization�������������������������������������������������������35, 36
Membrane proteins�������������������������12, 61–66, 116, 132, 261 Protein microarray����������������������������������������������������149–168
Microarray����������������������������������������������������������������149–168 Protein network��������������������������������������������������������137–147
Microarray analysis���������������������������������������������������146–168 Protein phosphorylation�����������������������������������������������������48
Micropatterning��������������������������������������� 261–263, 266, 268 Protein-protein interactions (PPIs)����������������� 116–119, 132,
Microscopy�����������������������������������������������������������������������265 138–140, 235–243, 245–247, 249–255, 261–269
Multicolor detection���������������������������������������������������������156 Protein purification������������������������������������������������������������11
Multi-lectin affinity chromatography Protein quantitation���������������������������99–102, 104–111, 173,
(M-LAC)�������������������������������������������������������99–111 214, 266, 274, 275
Mutations������������������������������������������������� 235, 236, 243–254 Proteome�������������������������������� 2, 36, 59, 61, 70, 84, 151, 171,
MzML���������������������������������������������294, 298, 299, 305, 339, 193, 232, 273, 289, 339
340, 352, 355, 360, 363 Proteome informatics�������������������������������������������������������339
MzXML�������������������������������������������294, 298, 339, 340, 345, Proteomics�������������������������������� 1, 19, 36, 56, 61, 70, 84, 109,
350, 352, 354, 355, 357, 360 120, 150, 171, 199, 223, 271, 289, 325, 339
ProteoWizard�����������������������������������179, 294, 298, 305, 326,
N
340–341, 345, 349, 352
Nanocrystals��������������������������������������������� 152, 153, 155, 156 Proximity labeling�������������������������������������������������������������115
Nuclear fractionation����������������������������������������������������37, 39 pyProphet�������������������������������������������������������� 292, 293, 295,
301–302, 305
O
Q
Open formats������������������������������������������� 294, 298, 339, 340
OpenSWATH������������������������������������������224, 229, 292, 294, Qdot nanocrystal�������������������������������������� 152, 153, 155–160
295, 298–305 Q-TOF����������������������������������������������208–211, 223, 305, 332
Quadrupole�����������������������������������������97, 193, 199, 208, 224
P Quality control (QC)�������������������������195, 215, 220, 225, 229,
Peptide���������������������������������1, 12, 30, 38, 50, 61, 70, 84, 109, 272, 274–276, 278–280, 282–284, 293, 295–298, 325,
123, 139, 172, 185, 201, 223, 247, 272, 290, 366 326, 328–333, 335, 336
Pericentromeric repeats������������������������������������������������������20 Quantification���������������������������� 6, 8, 29, 35, 37, 41, 45, 56,
Phase-transfer��������������������������������������������������� 12, 15, 17, 62 70, 71, 76, 84, 92, 97, 123, 140, 151, 152, 154, 173,
Phenotype�������������������������������������������������������� 236–238, 240 185–197, 261, 283, 289, 290, 292, 300, 352
Phosphate-buffered saline (PBS)���������������������������22–24, 30, 37, Quantitation������������������������������ 5, 99–102, 104–111, 171, 173,
39, 49, 51, 58, 64, 79, 104, 111, 123, 124, 142, 144–146, 174, 176, 179, 180, 205–206, 214, 231, 232, 265–267,
175–177, 182, 263, 265 273–276, 283, 285, 302
Phosphocapture������������������������������������������������������������������47 Quantitative analysis������������������������������������������� 31, 41, 186,
Phosphorylation�����������������������������������48, 58, 117, 123, 139, 261–269
151, 173, 183, 249 Quantitative proteomics���������������������������������� 11–17, 61–66,
Plasma proteomics���������������������������������������������������� 275, 282 171–173, 175–182, 223–229, 231, 232, 273
Polylysine��������������������������������������������������������������������������268
R
Post-translational modification������������������������35, 48, 61, 99,
100, 123, 132, 238, 239 Randomization�������������������������������������������������������� 129, 229,
Precision medicine����������������������������������������������������149–168 278–280, 282, 284
Principal component analysis (PCA)��������������� 116, 215, 326, Receptor tyrosine kinase���������������������������������������������������179
329, 330, 332, 334
Proteomics: Methods and Protocols
371
Index
Reverse phase protein microarray (RPPA)�������������� 149–152, T
154–163
RNase A����������������������������������������������������������������� 24, 29, 31 Tandem mass tags (TMT)������������������������������ 84, 85, 87–90,
92–97, 185–197, 276, 280
S Targeted proteomics��������������������������������� 199–221, 289, 290
Telomeres����������������������������������������������������20, 22, 28, 29, 32
Sample preparation��������������������������� 2, 37–40, 49, 62, 70, 71,
Ternary complex������������������������������������������������������� 139, 140
85, 88, 150, 154, 191, 197, 220, 282, 325
Time-of-flight (TOF)������������������������������128, 186, 199, 223,
Serum����������������������������������� 29, 37, 100, 110, 130, 133, 140,
296, 303, 360, 367
142, 144, 150, 154, 155, 174, 205, 263, 331
TIRF microscopy�������������������������������������������������������������265
Signal transduction����������������������������������������������� 36, 48, 149
Transfection����������������������� 119, 122, 130, 140, 143, 144, 146
Signaling pathways����������������������������������� 137, 138, 247, 249
Transmembrane proteins������������������������������������������ 139, 263
Single nucleotide polymorphism
TRIC aligner�������������������������������������292, 293, 295, 302–303
(SNP)��������������������������������������������������� 235, 237, 238
Triple quadrupole (QQQ)���������������� 199, 204–207, 216–221
Skyline����������������������������������������������201, 206, 216, 221, 224,
Trypsin����������������������������� 3, 6, 7, 9, 12, 13, 15, 16, 38–40, 44,
229, 293–298, 303
50, 52, 62–64, 67, 72, 73, 78, 94, 101, 105, 109, 121,
Smad2 and Smad4�����������������������������������������������������������138
124–128, 130, 133, 142, 174–178, 188, 190, 193, 195,
Soft lithography���������������������������������������� 262, 264–265, 267
201, 207, 212, 220, 268, 274
Sonication������������������������������������������� 6, 8, 14, 16, 20, 21, 24,
30, 31, 39, 74, 80, 176 V
STAGE-Tip����������������������������������������17, 63, 65, 75, 81, 191
Streptavidin������������������������������������������������20, 22, 25, 26, 31, Variable windows���������������������������������������������� 227, 228, 231
118–120, 124–127, 132, 262–264, 268 Variance��������������������� 194, 277, 282, 283, 327, 329–331, 335
Subcellular proteomic���������������������������������������������������������43 Virtual computers������������������������������310, 311, 317, 319, 323
SWATH����������������������������������������������������������� 116, 223, 290 VirtualBox���������������������������������311–313, 317–320, 322–324
SWATH acquisition������������������223–229, 231, 232, 290, 303
W
SWATH MS��������������������� 224, 232, 290–294, 302, 303, 305
System reliability��������������������������������������������������������������309 Western-blot������������������� 20, 27, 28, 125, 139, 164, 181, 182
Systems management�������������������������������������������������������122 Wnt signaling�������������������������������������������������������������������138