Introduction into Selected Reaction Monitoring (SRM) Christina Ludwig

Transcription

1 Introduction into Selected Reaction Monitoring (SRM) Christina Ludwig EuPA Bioinformatics course

2 Overview A) What is selected reac/on monitoring, how does it work and why is it useful? B) How do I design a proteomic SRM measurement? C) How does SRM data look like and how do I analyze it? D) How can I quan/fy protein abundances with SRM? 1

3 Naming conven/on Selected Reac/on Monitoring (SRM) Mul/ple Reac/on Monitoring (MRM) 2

4 Overview A) What is selected reac/on monitoring, how does it work and why is it useful? B) How do I design a proteomic SRM measurement? C) How does SRM data look like and how do I analyze it? D) How can I quan/fy protein abundances with SRM? 1

5 Discovery versus targeted proteomics Shotgun Proteomics Selected Reac/on Monitoring Data dependent acquisi/on intensity MS1 m/z intensity m/z m/z m/z MS2 shortcomings à limited sensi/vity, biased towards more abundant proteins à reproducibility, missing data points à iden/fica/on of non- relevant proteins benefits à highly sensi/ve à reproducible à focus exclusively onto proteins of interest 3

6 Discovery versus targeted proteomics Shotgun Proteomics Selected Reac/on Monitoring Data dependent acquisi/on intensity m/z MS1 COMPLEMENTARY (not compe//ve) methods intensity m/z m/z m/z MS2 shortcomings à limited sensi/vity, biased towards more abundant proteins à reproducibility, missing data points à iden/fica/on of non- relevant proteins benefits à highly sensi/ve à reproducible à focus exclusively onto proteins of interest 3

7 The standard SRM workflow Protein extrac/on Protein diges/on Pep/de separa/on Proteins Pep/des organic solvent, % Ioniza/on Mass spectrometry HPLC time ESI triple quadrupole mass spectrometer (QQQ) 4

8 SRM is performed on a triple quadrupole mass spectrometer (QQQ) Q1/Q3 pair = Transi/on Q1 mass filter Q2 collision cell Q3 mass filter SRM trace 0.7 Da 0.7 Da intensity precursor ion selection fragmentation fragment ion selection time Advantages of SRM carried out on QQQ: efficient reduc/on of noise high transmission efficiency (>95%) à high sensitivity and selectivity Disadvantages of SRM carried out on QQQ: low resolu/on (achievable resolu/on 0.7 FWHM) You need to know what to look for in advance!!! 6

9 Principle of SRM data acquisi/on NH 2 b 1 y 11 y 10 y 9 y 8 y 7 y 6 y 5 y 4 y 3 y 2 y 1 F F G F T P E G V A E R b 2 b 3 b 4 b 5 b 6 b 7 b 8 b 9 b 10 b 11 COOH Transi/on 1 sequence Q1 [m/z] Q3 [m/z] frag- ment ion CE [ev] dwell /me [ms] FFGFTPEGVAER y FFGFTPEGVAER y FFGFTPEGVAER y FFGFTPEGVAER y Q1 Q2 Q3 Intensity record over 250 ms for transi/on

10 Principle of SRM data acquisi/on NH 2 b 1 y 11 y 10 y 9 y 8 y 7 y 6 y 5 y 4 y 3 y 2 y 1 F F G F T P E G V A E R b 2 b 3 b 4 b 5 b 6 b 7 b 8 b 9 b 10 b 11 COOH 30 sec sequence Q1 [m/z] Q3 [m/z] frag- ment CE [ev] dwell /me [ms] FFGFTPEGVAER y FFGFTPEGVAER y FFGFTPEGVAER y FFGFTPEGVAER y Reten/on /me, min sum of all dwell /mes is 30 data points over the the cycle /me = 1 second chromatographic peak à good peak extrapola/on à good quan/ta/ve accuracy 8

11 Quan/fica/on accuracy as a func/on of dwell and cycle /me Lange et al., 2008, Mol Syst Biol, 4, 222 5

12 SRM method design is a balancing act Goal: Achieve the best sensi/vity at a high quan/ta/ve accuracy with as many proteins as possible! Best sensi/vity Rule of thumb: Typical dwell /mes are between 10 ms and 100 ms dwell /me as long as possible High quan/ta/ve accuracy Rule of thumb: At least 8 data points over the peak are required (typically cycle /mes between 1.5 s and 3.5 s) cycle /me as short as possible No. of transi/ons/ pep/des/proteins as many as possible Maximal mul/plexing Rule of thumb: 3 to 6 transi/ons per pep/de, 3 to 5 pep/des per protein To date maximal number of transi/ons per run ~1000 (scheduled SRM) 9

13 Scheduled SRM Increase the number of transi/ons per run without loosing sensi/vity and quan/ta/ve accuracy Intensity Seg1 Seg2 Seg3 Seg4 Pep1 Pep2 Pep3 Pep reten/on /me (min) pep/de Q1 Q3 start /me stop /me pep pep pep pep pep pep pep Scheduled SRM can help to increase the number of measurable transi/ons in one run (at a given dwell and cycle /me) by a factor of 5 to 20 fold BUT: accurate reten/on /me informa/on is required J. Stahl- Zeng et al., Mol Cell Proteomics, 2007, 6,

14 Pros and Cons of SRM Advantages of SRM: à Ideal technique for monitoring abundance of a set of pre- defined target proteins over many samples à A sensi/ve, selec/ve and quan/ta/vely accurate mass- spectrometric technique à Large dynamic quan/fica/on range spanning 4 to 5 orders of magnitude (in yeast proteins from 1E6 down to 50 copies/cell could be quan/fied) 1 à High degree of reproducibility and repeatability 2 Bojlenecks of SRM: à You need to have prior knowledge about the proteins of interest (targeted MS) à You need to develop, op/mize and validate SRM assays for each target pep/de, which is cost and labor intensive à Low resolu/on of the mass spectrometer à Limited mul/plexing capability (ca transi/ons per run) 1 Picok et al., 2009, Cell, 138, Addona et al., 2009, Nat Biotechnology, 27,

15 Take- home messages What is selected reac/on monitoring (SRM), how does it work and why is it useful? à Mass spectrometric approach falling into the category of targeted mass spectrometry à Measured on triple quadrupole instruments à Q1 and Q3 are used as mass filters à Q1 filters for the pep/de precursor ion à Q3 filters for the fragment ion à A Q1/Q3 pair is called a transi/on à For proteomic SRM several transi/ons per pep/de (3 to 6) and several pep/des per protein (3 to 5) are measured à For every SRM measurement a good balance between dwell /me, cycle /me and number or transi/ons/pep/des/proteins needs to be set à With scheduled SRM the number of targetable transi/ons/pep/des/proteins per run (at a given dwell and cycle /me) can be significantly increased 12

16 Overview A) What is selected reac/on monitoring, how does it work and why is it useful? B) How do I design a proteomic SRM measurement? C) How does SRM data set look like and how do I analyze it? D) How can I quan/fy protein abundances with SRM? 1

17 Design of an SRM experiment requires three types of informa/on 1. Selec/on of protein targets 2. Selec/on of representa/ve pep/des per protein 3. Selec/on of best transi/ons per pep/de Lange et al., 2008, Mol Syst Biol, 4,

18 1. Protein selec/on protein targets à depends on each study/project à can contain 10s to 100s of proteins etc. literature shotgun MS gene expression data protein- protein interac/on data 14

19 2. Pep/de selec/on pep/de targets à Each target protein typically yields 10s to 100s of (tryp/c) pep/des. à Only a few representa/ve pep/des per protein should be selected to infer protein abundance. WHICH ONES? 15

20 2. Pep/de selec/on pep/de targets 1. Not all peptides of a given protein have the same MS response factor. It is important to find the best-flying peptides to achieve maximal sensitivity. Examples of online repositories for shotgun proteomics: Pep@deAtlas PRIDE Human Proteinpedia The global proteome machine hgp://gpmdb.thegpm.org/ Perform your own proteome mapping experiments à Discovery proteomics: Shotgun or data-independent acquisition à Extensive sample fractionation à Overexpression of protein standards 15

21 2. Pep/de selec/on pep/de targets 1. Not all peptides of a given protein have the same MS response factor. It is important to find the best-flying peptides to achieve maximal sensitivity. 2. Only peptides which are unique for a given protein should be used to study the regulation of this particular protein. Uniqueness can be defined at different levels 1. Genome sequence of an organism 2. Splice isoforms 3. SNPs /sites/entrez Proteotypic peptides (PTPs): Good MS-response factor and unique for a given protein. 15

22 2. Pep/de selec/on pep/de targets How to deal with proteins that have not yet been observed by MS? à Predict the proteotypic peptides based on the protein sequence Pep/deSieve Prediction based on physico-chemical properties - Total amino acid composition of peptide - Peptide length - Amino acid charge - Amino acid secondary structure propensity - Amino acid hydrophobicity Mallick, P., et al. (2006). Computational prediction of proteotypic peptides for quantitative proteomics. Nature Biotechnology 25, 125.

23 3. Transi/on selec/on transi/on targets NH 2 b 1 y 11 y 10 y 9 y 8 y 7 y 6 y 5 y 4 y 3 y 2 y 1 F F G F T P E G V A E R b 2 b 3 b 4 b 5 b 6 b 7 b 8 b 9 b 10 b 11 COOH à Each pep/de can fragment into a series of fragment ions (y-, b-, a-, c-, x-, z-, water- loss, amine loss, etc.) à Only a few representa/ve fragments per pep/de should be selected WHICH ONES?

24 Where do you get SRM- Assay parameter from?!? transi/on targets We want to select the most intense transi/ons à gain best sensi/vity We want to select transi/ons uniquely describing the target pep/de à gain best selec/vity Empirical data (MS2 spectra) Predic/on intensity m/z 1. Shotgun data / public data repositories 2. Crude synthe/c pep/de libraries 3. In- vitro synthesized proteins Bioinfoma/c tool predict best transi/ons per pep/de based on the pep/de sequence: S. Li, R. J. Arnold, H. Tang, P. Radivojac, Anal Chem, 2011, 83, 790. A. Bertsch et al., J Proteome Res, 2010, 9, 2696.

25 SRM assay parameters strictly required op/onal target protein proteotypic peptide LC retention time Precursor m/z Fragments m/z Selected reaction monitoring (SRM) Q1 Q2 Q3 0.7 Da 0.7 Da Protein diges/on Pep/de separa/on precursor fragmentation fragment Collision energy Relative intensities of transitions Once developed, SRM assays can be used universally and stored in public repositories (

26 Take- home messages How do I design a proteomic SRM measurement? à Three important types of informa/on 1. protein targets 2. proteotypic pep/des per protein 3. SRM assays for each proteotypic pep/de à An SRM assay for any given proteotypic pep/de includes parameters for 1. Precursor m/z value (Q1) 2. Fragment m/z values (Q3s) 3. collision energy 4. pep/de reten/on /me 5. rela/ve intensi/es for all fragments à SRM assays can be generated from 1. empirical data (shotgun, crude synthe/c pep/des, in- vitro synthesized proteins, etc.) 2. predic/on à Once an SRM- assay is op/mized and validated it can be used uniformly and stored in pubic databases 18

27 Overview A) What is selected reac/on monitoring, how does it work and why is it useful? B) How do I design a proteomic SRM measurement? C) How does SRM data look like and how do I analyze it? D) How can I quan/fy protein abundances with SRM? 1

28 Triple Quadrupoles and SRM data analysis somware QQQs Q- Trap (ABSciex) TSQ (Thermo) QQQ (Agilent) Xevo (Waters) somware EVOQ (Bruker) PACERTM Mul/quant Pinpoint MassHunter MassLynx 19

29 Criteria for reliable SRM peak iden/fica/on and quan/fica/on 1. Co- elu/on 2. Peak shape 3. Signal intensity intensity intensity intensity /me (min) /me (min) /me (min) can be always applied - no prior knowledge required 20

30 Criteria for reliable SRM peak iden/fica/on and quan/fica/on 1. Co- elu/on 2. Peak shape 3. Signal intensity /me (min) 5. Correla/on reten/on /me (empiric or predicted*) intensity intensity intensity intensity /me (min) /me (min) 4. Correla/on peak intensi/es to library spectrum 6. Correla/on with heavy- labeled standard light heavy intensity y6 y7 y9 y11 /me (min) intensity light heavy m/z /me (min) *SSRCalc: Krokhin et al., 2004, Mol Cell Proteomics, 3,

31 Real- life SRM... target pep/de: IGQEVDGEAVGDEFK protein: YBR181C (Rps6a)...only some/mes looks like this! 21

32 Real- life SRM... target pep/de: QAINVEEAFYTLAR target protein: YNL098C (Ras2) library MS2 spectrum dotp: 0.82 dotp:

33 Real- life SRM... target pep/de: EAAAAAYGPDTDIPR target protein: YHR205W (Sch9) dotp 0.93? library MS2 spectrum dotp 0.93? dotp 0.99? dotp 0.99?? dotp 0.83 dotp 0.99? dotp 0.89? 23

34 Real- life SRM... Use of 15 N labeled total yeast extract as an internal standard light ( 14 N labeled) light ( 14 N) heavy ( 15 N) overlay light and heavy heavy ( 15 N labeled) Confident iden/fica/on possible!!! 24

35 Real- life SRM... light ( 14 N labeled) target pep/de: GYGEAEFVFSFIK target protein: YKR039W (Gap1) heavy ( 15 N labeled) NOT DETECTABLE!!! 25

36 Take- home messages How does SRM data look like and how do I analyze it? à For SRM peak iden/fica/on the following criteria can be applied: 1. co- elu/on 2. peak shape 3. peak intensity 4. correla/on with spectral library 5. reten/on /me correla/on 6. correla/on with heavy internal standard à In complex samples and with low abundant target proteins the use of heavy- labeled internal standards is highly recommendable 26

37 Overview A) What is selected reac/on monitoring, how does it work and why is it useful? B) How do I design a proteomic SRM measurement? C) How does SRM data look like and how do I analyze it? D) How can I quan/fy protein abundances with SRM? 1

38 Chromatographic SRM signal intensity scales linear with pep/de abundance Final goal of any SRM experiment: Quan/fica/on of predefined target proteins across mul/ple biological samples SRM pep/de intensity (cps) GSMADVPK(heavy) 1E+07 R² = E+06 1E+05 1E+04 1E pep/de concentra/on (fmol) 100 Linearity over up to 5 orders of magnitude have been reported in the literature à SRM is the MS method with the highest linear dynamic range P. Picotti, B. Bodenmiller, L. N. Mueller, B. Domon, R. Aebersold, Cell, 2009, 138, 795. J. Stahl-Zeng et al., Mol Cell Proteomics, 2007, 6,

39 Relative versus absolute protein quantification Rela/ve quan/fica/on Absolute quan/fica/on samples 2 samples 1 samples healthy disease quan/ta/ve readout specific protein intensity value specific protein intensity value 600 output protein ra/o 1 : 2 2- fold upregulated protein concentra/on copies/cell fmol/µg extract ng/ml body fluid rela/ve comparison of the same protein between samples comparison of the same protein between samples + comparison of different proteins within the same sample 28

40 Label- free versus stable isotope- labeled quan/fica/on LABEL- FREE QUANTIFICATION based on SRM signal intensity changes between samples PROs: cheap applicable to any kind of samples CONs: prone to possible sample prepara/on and MS measurement variabili/es à less precise and accurate quan/fica/on STABLE ISOTOPE- LABELED QUANTIFICATION based on SRM signal intensity changes rela/ve to an isotopically- labeled standard CONs cost intensive not always applicable PROs: can account for biases in sample prepara/on and for MS run to run varia/ons à more precise and accurate quan/fica/on Accepted gold standard for protein quan/fica/on using mass spectrometry is to perform stable isotope- labeled quan/fica/on in combina/on with SRM 29

41 Stable- isotope labeled protein quan/fica/on Facts Introduc/on of heavy standard proteins or pep/des into the sample of interest which carry isotope- labeled atoms (typically 13 C, 15 N, 18 O). Heavy standards are chemically iden/cal to the endogenous targets and hence show the same behavior in terms of à chromatography à ioniza/on à fragmenta/on à ion suppression Heavy standards and endogenous targets are dis/nguishable by the induced mass shiw 30

42 Stable- isotope labeled standards AQUA/SIS pep/des concatenated pep/des fully labeled proteins metabolic labeling chemical labeling NH 2 COOH plasmid plasmid synthesis of pep/des with 13 C- 15 N- labeled C- terminal amino acid cloning of concatenated proteotypic pep/des (from different proteins), overexpression in labeled medium, overexpression of the target protein in labeled medium cell growth in labeled medium (SILAC, 15N labeling) chemical labeling of amino acid side chains or termini at the pep/de or protein level (ICAT, dimethyl) rela/ve quan/fica/on addi/on of a heavy- labeled standard in an UNKOWN concentra/on à output format: ra/o light/heavy 31

43 Stable- isotope labeled standards AQUA/SIS pep/des concatenated pep/des fully labeled proteins metabolic labeling chemical labeling NH 2 COOH plasmid plasmid synthesis of pep/des with 13 C- 15 N- labeled C- terminal amino acid cloning of concatenated proteotypic pep/des (from different proteins), overexpression in labeled medium overexpression of the target protein in labeled medium cell growth in labeled medium (SILAC, 15N labeling) chemical labeling of amino acid side chains or termini at the pep/de or protein level (ICAT, dimethyl) absolute quan/fica/on addi/on of a heavy- labeled standard in a precisely KNOWN concentra/on output format: copies/cell, ng/ml 31

44 Take- home messages How can I quan/fy protein abundances with SRM? à Protein quan/fica/on with SRM is possible due to a linear correla/on between the chromatographic SRM signal intensity and protein/pep/de abundance à Protein quan/fica/on results can be classified into rela/ve and absolute quan/fica/on à Rela/ve quan/fica/on leads to ra/o changes à Absolute quan/fica/on leads to protein concentra/ons (like ng/ml or copies/cell) à Protein quan/fica/on can be carried out label- free or based on stable- isotope labeled standards à Stable- isotope labeled quan/fica/on in combina/on with SRM is the accepted gold standard technique for protein quan/fica/on à The ideal quan/fica/on approach needs to be chosen for each project accordingly 32

45 Ques/ons?