Feny? D. (Ed.) Computational Biology

Подождите немного. Документ загружается.

202 Fenyö, Eriksson, and Beavis

24. J. Eriksson and D. Fenyo (2004) Probity, a

protein identiﬁcation algorithm with accurate

assignment of the statistical signiﬁcance of the

results, J Proteome Res, 3, 32–6.

25. R. Craig and R.C. Beavis (2003) A method

for reducing the time required to match pro-

tein sequences with tandem mass spectra,

Rapid Commun Mass Spectrom, 17, 2310–6.

26. R. Craig and R.C. Beavis (2004) TANDEM:

matching proteins with tandem mass spectra,

Bioinformatics, 20, 1466–7.

27. R. Craig, J.P. Cortens, and R.C. Beavis (2005)

The use of proteotypic peptide libraries for

protein identiﬁcation, Rapid Commun Mass

Spectrom, 19, 1844–50.

28. R. Craig, J.P. Cortens, and R.C. Beavis (2004)

Open source system for analyzing, validating,

and storing protein identiﬁcation data,

J Proteome Res, 3, 1234–42.

29. D. Fenyo, B.S. Phinney, and R.C. Beavis

(2007) Determining the overall merit of

protein identiﬁcation data sets: rho-diagrams

and rho-scores, J Proteome Res, 6,

1997–2004.

30. D.N. Perkins, D.J. Pappin, D.M. Creasy, and

J.S. Cottrell (1999) Probability-based protein

identiﬁcation by searching sequence databases

using mass spectrometry data, Electrophoresis,

20, 3551–67.

31. W. Zhang and B.T. Chait (2000) ProFound:

an expert system for protein identiﬁcation

using mass spectrometric peptide mapping

information, Anal Chem, 72, 2482–9.

32. J. Magnin, A. Masselot, C. Menzel, and

J. Colinge (2004) OLAV-PMF: a novel scor-

ing scheme for high-throughput peptide

mass ﬁngerprinting, J Proteome Res, 3,

55–60.

203

Chapter 12

Unbiased Detection of Posttranslational Modiﬁcations

Using Mass Spectrometry

Maria Fälth Savitski and Mikhail M. Savitski

Abstract

A major challenge in proteomics is to fully identify and characterize posttranslational modiﬁcation (PTM)

patterns present at any given time in cells, tissues, and organisms. Currently, the most frequently used

method for identifying PTMs is tandem mass spectrometry combined with searching a protein sequence

database. Although, database searching has been highly successful for the identiﬁcation of proteins, it has

a number of signiﬁcant drawbacks for identiﬁcation of modiﬁcations. The user needs to specify all

expected modiﬁcations, and the search engine needs to consider all possible combinations of these modi-

ﬁcations for all peptide sequences. If several potential modiﬁcations are considered, the search can take

much longer than the data acquisition, creating a bottleneck in high-throughput analysis. In addition,

the many possible assignments that need to be tested increase the noise and require better quality data

for conﬁdent identiﬁcation of modiﬁcations. Here, we describe a method for identifying both known and

unknown PTM using mass spectrometry that does not suffer from these problems. The method is based

on the observation that, in many samples, peptides are usually present both with and without modiﬁca-

tions. By identifying the unmodiﬁed peptide with conventional database searches, the modiﬁed species of

the peptide can be identiﬁed by searching for peptides with common and similar fragments as the unmod-

iﬁed peptide. After identifying both the modiﬁed and unmodiﬁed peptide, the elemental composition of

the modiﬁcation can be deduced if the mass accuracy of the precursor ion is sufﬁciently high.

Key words: Bioinformatics, Mass spectrometry, Posttranslational modiﬁcations, MS/MS, ModiﬁComb

Posttranslational modiﬁcations (PTMs) are covalent-processing

events that often are, as the name indicates, among the last steps

in the protein biosynthesis. PTMs can be produced by proteolytic

cleavage or by addition of a modifying group to one or more

amino acid residues. PTMs change the intrinsic properties of the

protein, and can determine its activity state, localization, turnover,

and interactions with other proteins. In signaling, for example,

1. Introduction

1.1. Posttranslational

Modiﬁcations

David Fenyö (ed.), Computational Biology, Methods in Molecular Biology, vol. 673,

DOI 10.1007/978-1-60761-842-3_12, © Springer Science+Business Media, LLC 2010

204 Savitski and Savitski

kinase cascades are turned on and off by the reversible addition

and removal of phosphate groups (1), and in the cell cycle,

ubiquitination marks cyclins for destruction at deﬁned time points

(2). PTMs are key regulators of protein function, localization,

and interactions taking place inside the cell (3, 4).

The reversible phosphorylation of serine, threonine, and

tyrosine residues in proteins is considered to be one of the most

important PTMs (3, 5–7). It is estimated that in higher organ-

isms more than 25% of all proteins are phosphorylated and more

than 2% of the human genes code for protein kinases and the

respective phosphatases. Protein phosphorylation is believed to

play a major role in many important cellular processes such as the

cell cycle, cell differentiation, metabolism, cell motility, and sig-

naling (8). Other PTMs such as glycosylation (9), methylation

(10, 11), and ubiquitination (12) play important roles in cellular

processes. Other types of known and unknown PTMs are also

involved. Many types of PTMs have been discovered serendipi-

tously and many more remain yet to be discovered.

Mass spectrometry (MS) is an analytical technique that measures

the mass-to-charge (m/z) ratio of ions. A mass spectrometer con-

sists of an ion source where the analytes are ionized, a mass ana-

lyzer where the ionized analytes are separated according to their

m/z ratio, and a detector that measures the relative abundance of

the ions at each m/z value. Ionizing proteins and peptides for MS

analysis is most commonly done by electrospray ionization (ESI)

(13) or by matrix-assisted laser desorption/ionization (MALDI)

(14, 15). ESI and MALDI are called soft ionizations techniques

because they can ionize a large molecule (e.g., protein or peptide)

without fragmenting it, which makes it possible to determine the

mass of the intact peptide or protein.

To be able to decipher the amino acid sequence of a peptide,

it has to be broken into pieces. Ions of the peptide of interest are

gathered at its speciﬁc m/z and the peptide ions are fragmented

by, for example, collision with an inert gas such as helium. This is

called tandem mass spectrometry (MS/MS) (16). To identify the

amino acid sequence, the MS/MS spectra are searched against

databases using search engines such as Mascot (17), Sequest (18),

and X! Tandem (19).

There are two major techniques to fragment peptides, either

through vibrational excitation or through electron capture/

transfer. Collision-activated dissociation (CAD) (20, 21) is the

most common vibrational dissociation method, whereas electron-

capture dissociation (ECD) (22) and electron-transfer dissocia-

tion (ETD) (23) are examples of dissociation through electron

capture/transfers. The different types of fragmentation methods

yield different types of fragment ions. The two most common ion

types in CAD fragmentation are b and y ions, which are formed

1.2. Mass

Spectrometry

205

Unbiased Detection of Posttranslational Modiﬁcations Using Mass Spectrometry

from backbone cleavage of the peptide. b ions are formed when

the N-terminal fragment retains the charge and y ions are formed

when the C-terminal fragment retains the charge. If the ionized

peptide contains more than one charge, complementary ion pairs

can be produced, for example, fragmentation of a doubly charged

peptide can produce a b

ion pair, where the sum of n and m

equals the total number of residues in the peptide. Although

CAD fragmentation most often yield b and y ions, ECD/ETD

mainly produce c and z·ions. ECD and ETD provide more exten-

sive peptide fragmentation than CAD and are better for charac-

terizing PTMs, because they remain intact during the backbone

fragmentation (vide infra) (24).

From a mass spectrometric point of view, PTMs can be roughly

classiﬁed into two distinct classes: labile and stable modiﬁcations

(Fig. 1). The loss of a labile modiﬁcation is a preferred fragmentation

1.3. Identiﬁcation

of PTMs Using MS

Fig. 1. (a) Fragmentation of an unmodiﬁed doubly charged peptide (only y

ions are

shown for simplicity). (b) Fragmentation of the same peptide with a stable modiﬁcation on

serine. y

, y

, and y

are shifted by the mass of the modiﬁcation X compared with the

unmodiﬁed peptide. (c) Fragmentation of the same peptide with a labile modiﬁcation on

serine. The most dominant peak corresponds to a loss of the neutral modiﬁcation mass Y

from the modiﬁed peptide. All y ions are low abundant and do not contain information on

the position of Y, because they arise through secondary fragmentation of (M − Y )

206 Savitski and Savitski

channel in CAD. A tandem mass spectrum of a peptide with a

labile modiﬁcation frequently, only exhibits one very abundant

fragment ion corresponding to the loss of the modiﬁcation and

few to none backbone fragment ions, making it very difﬁcult to

derive the identity of the peptide and the location of the modiﬁ-

cation (Fig. 1c). A stable modiﬁcation, however, has no such

effect. Fragmentation of a peptide with a stable modiﬁcation gives

rise to a spectrum dominated by backbone fragmentation

(Fig. 1b). Many modiﬁcations fall in the gap between the two

classes (4). Phosphorylation is such an example: serine/threonine

phosphorylation is a labile type of modiﬁcation, whereas the less

frequent tyrosine phosphorylation is a stable modiﬁcation.

One of the most striking features of the ECD/ETD approach

is that it always produces backbone fragmentation regardless of

what type of modiﬁcation is attached to the peptide, even such

labile ones as phosphorylation (25) and glycosylation (26).

Basically, ECD/ETD MS/MS spectra of modiﬁed peptides always

give rise to the type of fragmentation pattern shown in Fig. 1b

(with c and z ions instead of y and b). This makes the combined

use of ECD/ETD and CAD very effective for the study of phos-

pho-peptides. The strong neutral loss of phosphoric acid in CAD

serves as a ﬂag for identifying the spectra of phospho-peptides,

and the accompanying ECD spectrum can reveal the identity of

the peptide by providing sequence information and possibly also

the location of the phosphorylation site (27).

In order to detect stable modiﬁcations using database

searching, the following strategy is employed. Assume that one

is looking for a modiﬁcation of mass ∆M located on the amino

acid residue X (X is any of the 20 common amino acids). Then,

the search engine will expand the search space by adding for

every peptide containing X an additional modiﬁed peptide with

(X + ∆ M) instead of X (e.g., a peptide containing three X will

give rise to seven additional modified peptides, three with one

modification, three with two modiﬁcations, and one with three

modiﬁcations). The experimental spectra will be compared in

addition to the “usual” theoretical spectra, also to the spectra

stemming from these added modiﬁed peptides.

The addition of a large number of new peptides to the data-

base leads to a longer overall search time and to a decreased sen-

sitivity. It is very computationally demanding to perform searches

assuming the presence of more than ten types of modiﬁcations.

Also no new types of modiﬁcations can be discovered by using

this standard approach.

Craig and Beavis (28) have suggested a strategy for reduc-

ing the time required to search MS/MS data with variable modi-

ﬁcations. The strategy works under the assumption that an

unmodiﬁed, tryptic (without missed cleavages) peptide will be

selected for MS/MS for each identiﬁable protein in the sample.

207

Unbiased Detection of Posttranslational Modiﬁcations Using Mass Spectrometry

In the ﬁrst stage of identiﬁcation, only search criteria for

unmodiﬁed tryptic peptides (if desired one or more potential

modiﬁcations can be added) will be used. In subsequent stages,

searches allowing for modiﬁcations, non-speciﬁc hydrolysis

and missed cleavages will be conducted against proteins that

were identiﬁed in the ﬁrst stage by tryptic peptides. This strat-

egy allows for a more sensitive and speciﬁc identiﬁcation of

known modiﬁcations.

Here, we are going to describe a method for how to identify

both known and unknown PTMs utilizing MS. These types of

approaches have already uncovered various new and hitherto

unreported types of modiﬁcations.

The type of data required for the analysis described in this chapter

would preferentially come from a high-resolution mass spectrom-

eter such as the Fourier Transform mass spectrometer [29] or the

Orbitrap [30]. The data analysis would require access to a regular

PC and to one of the established search engines, e.g., X! Tandem

or Mascot.

This strategy for identifying PTMs is based on the assumption

that a modiﬁed peptide (dependent peptide) will be present

along with its unmodiﬁed counterpart (base peptide) in the sample

(most PTMs are substoichiometric) (Fig. 2). The computational

approach can identify PTMs without making any assumptions

about which PTMs are present in a given sample. It requires that

the MS data are in high resolution, and preferably the MS/MS

data should be in high resolution as well. The mass resolution is

the dimensionless ratio of the mass of the peak divided by its

width. Usually, the peak width is taken as the full width at half

maximum intensity. The resolution required for a reliable unbi-

ased PTM identiﬁcation should be above 50,000. If the MS data

are of low resolution, it will be very difﬁcult to reliably assign the

elemental composition to a potential modiﬁcation mass. If the

MS/MS data are of low resolution, the procedure can still work,

but only for high-quality spectra and will require more matching

fragments.

Step 1. Perform a regular database search to identify base pep-

tides in the dataset.

2. Materials

3. Methods

208 Savitski and Savitski

Step 2. Make a new dataset of the unidentiﬁed MS/MS spectra,

these are potential-dependent peptides.

Step 3. For every identiﬁed base peptide, search the dataset of

unidentiﬁed MS/MS spectra for spectra with similar fragment

ions (same mass or shifted by the mass of the potential modiﬁca-

tion, ∆M = M

modiﬁed

− M

unmodiﬁed

) (see Note 1). For an MS/MS spec-

trum to be considered as a dependent peptide to the base peptide,

there should be at least six common fragments (see Note 2). The

validity of the six fragment approach was tested on high-resolution

data using a special decoy database approach described in (31)

and was shown to yield less than 2% false-positive hits.

Step 4. The last step is to identify the elemental composition of

the modiﬁcation. This can be done, unambiguously in most cases,

when the mass of the peptide is measured within 5 ppm (see Notes

3 and 4).

1. Retention time difference can also be used to identify the pairs

of base and dependent peptides, because the dependent peptide

may be expected to elute within a limited time window before or

after the base peptide. For instance, in a typical 2 h liquid chro-

matography gradient, the phosphorylated version of a peptide

will rarely elute 10 min later or earlier than its unmodiﬁed

version. By using the retention time, the number of MS/MS

spectra of possible dependent peptides can be limited. Although

the retention time resolution is relatively low compared with

the m/z resolution (a standard deviation of around 5 min can

4. Notes

Fig. 2. Flowchart for the unbiased PTM detection. The text explaining the individual steps

can be found in the methods part.

209

Unbiased Detection of Posttranslational Modiﬁcations Using Mass Spectrometry

be expected in replicate experiments using a 90 min gradient),

it is still a highly useful parameter. For instance, retention time

alone can predict whether formylation (+CO) occurs on a ser-

ine/threonine side-chain or the N-terminus.

2. If there are MS/MS spectra from both CAD and ECD fragmen-

tation available, these can be added together to ﬁlter out real peptide

peaks by using golden complementary pairs (32, 33). Golden com-

plementary pairs are fragment ions pairs that can be reliably

identiﬁed from both CAD and ECD spectra by utilizing the fact

that different bonds are cleaved (in CAD the peptide bond and

in ECD the N–Ca bond). The fragment ions identiﬁed in such

a way have a very high, >95%, probability of being true fragment

ions. The golden pair matching also gives information about

whether the ions are N-terminal or C-terminal fragments. When

using golden pair, a criterion of four common fragments between

the base and the dependant peptides could be used.

3. The mass accuracy of the precursor ion is important when it

comes to identifying the elemental composition of the modi-

ﬁcation. The unique elemental composition of modiﬁcations

can be deduced in the majority of the cases when working

with MS data of <5 ppm accuracy.

4. An important part of the characterization of PTMs is to eval-

uate their biological signiﬁcance –do the modiﬁcations have a

function in vivo, or are they in vitro modiﬁcations due to the

sample preparation.

References

1. Cohen P. (2000) The regulation of protein

function by multisite phosphorylation – a 25

year update. Trends in Biochemical Sciences

25, 596–601.

2. Tyers M, Jorgensen P. (2000) Proteolysis and

the cell cycle: with this RING I do thee

destroy. Current Opinion in Genetics &

Development 10, 54–64.

3. Jensen ON. (2004) Modiﬁcation-speciﬁc pro-

teomics: characterization of post-translational

modiﬁcations by mass spectrometry. Current

Opinion in Chemical Biology 8, 33–41.

4. Mann M, Jensen ON. (2003) Proteomic anal-

ysis of post-translational modiﬁcations. Nature

Biotechnology 21, 255–61.

5. Blagoev B, Ong SE, Kratchmarova I, Mann M.

(2004) Temporal analysis of phosphotyrosine-

dependent signaling networks by quantitative

proteomics. Nature Biotechnology

22, 1139–45.

6. Blume-Jensen P, Hunter T. (2001) Oncogenic

kinase signalling. Nature 411, 355–65.

7. Steen H, Mann M. (2002) A new derivatization

strategy for the analysis of phosphopeptides by

precursor ion scanning in positive ion mode.

Journal of the American Society for Mass

Spectrometry 13, 996–1003.

8. Johnson SA, Hunter T. (2005) Kinomics:

methods for deciphering the kinome. Nature

Methods 2, 17–25.

9. Wells L, Vosseller K, Hart GW. (2001)

Glycosylation of nucleocytoplasmic proteins:

signal transduction and O-GlcNAc. Science

291, 2376–78.

10. Huang Y, Fang J, Bedford MT, Zhang Y, Xu

RM. (2006) Recognition of histone H3 lysin

e-4 methylation by the double tudor domain

of JMJD2A. Science 312, 748–51.

11. Philips MR, Pillinger MH, Staud R, et al.

(1993) Carboxyl methylation of ras-related

proteins during signal transduction in neutro-

phils. Science 259, 977–80.

12. Cenciarelli C, Hou D, Hsu KC, et al.

(1992) Activation-induced ubiquitination

of the T-cell antigen receptor. Science 257,

795–7.

210 Savitski and Savitski

13. Yamashita M, Fenn JB. (1984) Electrospray ion-

source – another variation on the free-jet theme.

Journal of Physical Chemistry88, 4451–9.

14. Karas M, Bachmann D, Hillenkamp F. (1985)

Inﬂuence of the wavelength in high-irradiance

ultraviolet laser desorption mass spectrometry

of organic molecules. Analytical Chemistry

57, 2935–9.

15. Tanaka K, Waki H, Ido Y, et al. (1988) Protein

and polymer analyses up to m/z 100 000 by

laser ionization time-of-ﬂight mass spectrometry.

Rapid Communications in Mass Spectrometry

2, 151–3.

16. McLafferty FW. (1981) Tandem mass spec-

trometry. Science (New York, NY) 214, 280–7.

17. Perkins DN, Pappin DJ, Creasy DM, Cottrell

JS. (1999) Probability-based protein identiﬁca-

tion by searching sequence databases using mass

spectrometry data. Electrophoresis 20, 3551–67.

18. Eng JK, McCormack AL, Yates JR. (1994) An

approach to correlate tandem mass-spectral

data of peptides with amino-acid-sequences in

a protein database. Journal of the American

Society for Mass Spectrometry 5, 976–89.

19. Craig R, Beavis RC. (2004) TANDEM: match-

ing proteins with tandem mass spectra.

Bioinformatics (Oxford, England)20, 1466–67.

20. Hunt DF, Yates JR, 3rd, Shabanowitz J,

Winston S, Hauer CR. (1986) Protein

sequencing by tandem mass spectrometry.

Proceedings of the National Academy of Sciences

of the United States of America 83, 6233–7.

21. Paizs B, Suhai S. (2005) Fragmentation path-

ways of protonated peptides. Mass Spectrometry

Reviews 24, 508–48.

22. Zubarev RA, Kelleher NL, McLafferty FW.

(1998) Electron capture dissociation of multi-

ply charged protein cations. A nonergodic

process. Journal of the American Chemical

Society 120, 3265–6.

23. Syka JE, Coon JJ, Schroeder MJ, Shabanowitz J,

Hunt DF. (2004) Peptide and protein

sequence analysis by electron transfer dissocia-

tion mass spectrometry. Proceedings of the

National Academy of Sciences of the United

States of America 101, 9528–33.

24. Zubarev RA. (2004) Electron-capture disso-

ciation tandem mass spectrometry. Current

Opinion in Biotechnology 15, 12–6.

25. Stensballe A, Jensen ON, Olsen JV, Haselmann

KF, Zubarev RA. (2000) Electron capture

dissociation of singly and multiply phosphory-

lated peptides. Rapid Communications in

Mass Spectrometry 14, 1793–800.

26. Mirgorodskaya E, Roepstorff P, Zubarev RA.

(1999) Localization of O-glycosylation sites

in peptides by electron capture dissociation in

a Fourier transform mass spectrometer.

Analytical Chemistry 71, 4431–6.

27. Kocher T, Savitski MM, Nielsen ML, Zubarev

RA. (2006) PhosTShunter: a fast and reliable

tool to detect phosphorylated peptides in liq-

uid chromatography Fourier transform tan-

dem mass spectrometry data sets. Journal of

Proteome Research 5, 659–68.

28. Craig R, Beavis RC. (2003) A method for

reducing the time required to match protein

sequences with tandem mass spectra. Rapid

Communications in Mass Spectrometry 17,

2310–6.

29. Marshall AG, Hendrickson CL. (2002)

Fourier transform ion cyclotron resonance

detection: principles and experimental con-

ﬁgurations. International Journal of Mass

Spectrometry 215, 59–75.

30. Olsen JV, de Godoy LMF, Li GQ, et al.

(2005) Parts per million mass accuracy on an

orbitrap mass spectrometer via lock mass

injection into a C-trap. Molecular & Cellular

Proteomics 4, 2010–21.

31. Savitski MM, Nielsen ML, Zubarev RA.

(2006) ModiﬁComb, a new proteomic tool

for mapping substoichiometric post-transla-

tional modiﬁcations, ﬁnding novel types of

modiﬁcations, and ﬁngerprinting complex

protein mixtures. Molecular & Cellular

Proteomics 5, 935–48.

32. Horn DM, Zubarev RA, McLafferty FW.

(2000) Automated de novo sequencing of

proteins by tandem high-resolution mass

spectrometry. Proceedings of the National

Academy of Sciences of the United States of

America 97, 10313–7.

33. Nielsen ML, Savitski MM, Zubarev RA.

(2005) Improving protein identiﬁcation

using complementary fragmentation tech-

niques in fourier transform mass spectrome-

try. Molecular & Cellular Proteomics 4,

835–45.

211

Chapter 13

Protein Quantitation Using Mass Spectrometry

Guoan Zhang, Beatrix M. Ueberheide, Soﬁa Waldemarson,

Sunnie Myung, Kelly Molloy, Jan Eriksson, Brian T. Chait,

Thomas A. Neubert, and David Fenyö

Abstract

Mass spectrometry is a method of choice for quantifying low-abundance proteins and peptides in many

biological studies. Here, we describe a range of computational aspects of protein and peptide quantita-

tion, including methods for ﬁnding and integrating mass spectrometric peptide peaks, and detecting

interference to obtain a robust measure of the amount of proteins present in samples.

Key words: Proteomics, Quantitation, Proteins, Peptides, Mass spectrometry

Mass spectrometry (MS)-based quantitative proteomics has been

applied to solve a wide variety of biological problems, and several

MS-based workﬂows have been developed for protein and pep-

tide quantitation (Fig. 1). In mass spectrometric quantitation

methods it is usually assumed that the measured signal has a lin-

ear dependence on the amount of material in the sample for the

entire range of amounts being studied. A prerequisite for accu-

rate quantitation is that unwanted experimental variations in

sample extraction, preparation, and analysis be minimized, and it

is therefore critical that each step in the workﬂow is optimized

for reproducibility.

One way of optimizing the reproducibility is to label the

samples with stable isotopes, mix them together and perform the

subsequent sample-handling steps on the mixed sample. The earlier in

the workﬂow that the stable isotope label is introduced and the sam-

ples mixed, the smaller is the effect of variations in sample handling.

1. Introduction

David Fenyö (ed.), Computational Biology, Methods in Molecular Biology, vol. 673,