Targeted(MS(and(its(Applicaon( in(Biology(and(Medicine( · PDF fileTargeted proteomics:...

Targeted MS and its Applica2on in Biology and Medicine

Hasmik Keshishian Proteomics Group The Broad Ins2tute of MIT and Harvard

Perturbations: • Disease • Development • Drug • KO/KI

Discovery defines a reduced set of “sentinel” marks that need to be repeatedly measured in a range perturbations

Precisely measure selected analytes in all experiments – no

missing data!

Desired assay properties:

• Highly specific • Sensitive • Highly precise • Multiplexed •  Interference-free

Past: Westerns; Immunoassays

Analyte Valley of Death

Not all proteins and PTMs of interest observed in all experiments

Conventional protein measurement methods have major limitations

Properties of the Measurements Western blotting Immunoassay

Specificity of method Highly variable Good Standards (proteins) available for analytical validation No Few

Internal standards used for high inter-laboratory reproducibility No No

Validated assays for high % of proteome No No Quantitative No to semi Yes Moderate-to-high throughput No Yes Can be highly (e.g., >20) multiplexed No No Interferences detectable and avoidable No No New assays easy to generate (time, reagents, cost, expertise)? No No

Targeted proteomics: hypothesis-driven technology for biology and medicine

Targeted MS Methods

•  Fewer arrows shot •  Fewer pep2des hit •  What you hit is defined by user •  All selected pep2des hit all the 2me •  Quan2fica2on is highly precise and can

be accurate using internal standards. •  50-‐1000x more sensi2ve than Discovery

Classic “Discovery” MS

•  Lots of “arrows” shot •  Many, many pep2des hit •  What you hit is not up to you •  Pep2des not repeatedly detected

across experiments •  Good rela2ve quan2fica2on if

labeling is used

How targeted MS (MRM-MS) differs from conventional MS/MS

Triple quadrupole mass spectrometer Q1 Q2 Q3

Ionize all peptides

Mass-select peptide ion

Fragment peptide ion

MRM-MS

MRM spectrum

10 ms

Monitor 3 fragment ions

Triple quadrupole mass spectrometer Q1 Q2 Q3

Detect all fragment ions

Ionize all peptides

Mass-select peptide ion

Fragment peptide ion

200 400 600 800 1000

Mass spectrum

200 ms

MS/MS Operating Mode

Terminology:

SRM – Selected Reac2on Monitoring MRM – Mul2ple Reac2on Monitoring PRM – Parallel Reac2on Monitoring Transi;on – Precursor coupled to a specific product ion

(Q1/Q3 pair) Peak Area Ra;o (PAR) –

Refer to the same targeted MS method on a triple quadrupole Mass spectrometer

Peak Area (light) Peak Area (heavy)

Refers to a targeted MS method on high resolu2on MS instrument

Proteins of interest

Define “signature peptides” for the protein

1. Existing data from in-house unbiased experiments

2. Proteomics data in public domain (GPM, PeptideAtlas)

3. Predictive algorithms (ESP, PeptideSieve)

Configure MRM assay

Synthesize heavy internal standard peptides

Apply MRM assay to samples

MRM assay construction

Ratio “light” peptide to “heavy” peptide

m/z A

bund

ance

*

*

*

Q1 Q2 Q3

Endogenous “light” peptides

Spike “heavy” Internal standard

peptides

568.8/716.4 568.8/829.5 568.8/916.5

y6

y8

y7

571.8/722.4 571.8/835.5 571.8/922.5

y6

y8

y7

G Y S I F S Y A T [12C6]K y8 y7 y6

MH+ = 1136.6 [M+2H]2+ = 568.8

G Y S I F S Y A T [13C6]K y8 y7 y6

MH+ = 1142.6 [M+2H]2+ = 571.8

PAR = [12C6] Peak Area [13C6] Peak Area

Monitoring a pair of 12C/13C peptides by MRM-MS

Key Steps in Developing Targeted Quant. Assays

•  Selection of Targets –  Empirical MS data or known targets of interest? –  Synthesis of heavy and light peptide targets

•  Characterization of Targets –  Selection of transitions and assay optimization

•  Determination of Assay Figures of Merit –  LOD/LOQ –  Assay Precision

•  Testing on typical samples

–  Demonstrates robustness of assay –  Surveys assay for utility in samples of interest

•  Application to real samples

Light Heavy 0 0.5

0.05 0.5

0.1 0.5

0.5 0.5

1 0.5

Q1 Q2 Q3

Response Curves and Figures of Merit

LOD Reproducibility Calibration Curve

Light peptide at varying conc.

Heavy peptide at constant conc.

Relevant Background

Output: Spectral Library

Input: DDA Search results

(mzXML, pepXML, etc)

Output: Vendor Specific MRM

Instrument Method

Input: Targeted Peptide List

Output: Peak identification and

automatic integration

Input: Vendor specific acquisition files

Output: Comprehensive

results report

Input: Integrated peaks, report parameters

Selection of target

peptides

MRM data acquisition

User data

analysis

skyline.gs.washington.edu

11

QuaSAR try 2012−05−15

Overall Reproducibility

Theoretical Concentration (fmol/ul)

Coeff

icien

t of V

ariat

ion (%

)

50

100

150●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●●●

●

●

●

●●●

0 0.01 0.05 0.135 0.375 1 4 20 100

try 2012−05−15Distribution of LOD values

(Outliers not shown)

Conc

entra

tion (

fmol/

ul)

0.00

0.05

0.10

0.15

0.20

LOD

Calibration curves for Peptide: AAYLQETGKPLDETLK

Theoretical Conc (fmol/ul)

Meas

ured

Con

c (fm

ol/ul)

0 50 100 150

050

100

150

Title : try (Concentration Plots)Agilent6490_SurrogateStudy_Fwd_MinMatrix_011912_jmr−calcurves 2012−05−15



LOD IS.Area 3.y10.1 0.39 1.0E+03 3.y13.2 0.06 8.1E+03 3.y15.2 0.04 6.4E+03 Spike level : 10

Data points for Peptide: AAYLQETGKPLDETLK


Meas

ured

Con

c (fm

ol/ul)

0.01 0.1 1 10 100

0.01

0.11

1010

0Title : try (Concentration Plots)

Agilent6490_SurrogateStudy_Fwd_MinMatrix_011912_jmr−calcurves 2012−05−15






Calibration curves for Peptide: AEVNGLAAQGK


Meas

ured

Con

c (fm

ol/ul)

0 50 100 150

050

100

150





Data points for Peptide: AEVNGLAAQGK


Meas

ured

Con

c (fm

ol/ul)

0.01 0.1 1 10 100

0.01

0.11

1010








Calibration curves for Peptide: ALQASALNAWR


Meas

ured

Con

c (fm

ol/ul)

0 50 100 150

050

100

150





Data points for Peptide: ALQASALNAWR


Meas

ured

Con

c (fm

ol/ul)

0.01 0.1 1 10 100

0.01

0.11

1010








Overall Reproducibility

Skyline: an essential free tool for targeted assays QuaSAR: fully integrated into Skyline

Numerous, well documented Studies •  Addona (2009) Nature Biotech •  Whiteaker (2011) Mol Cell Proteomics •  Addona (2011) Nature Biotech •  Kuhn (2011) Mol Cell Proteomics •  HüNenhain (2012) Sci Transl Med •  Kennedy (2013) Nature Methods •  Keshishian (2014) Mol Cell Proteomics 800 MRM assays; 2400 transi;ons

High Mul;plex and Informa;on Content

MRM-MS is precise, reproducible, robust, and can be highly multiplexed

Robust Precise and Reproducible

Lab 1

Lab 6

Lab 2 Lab 3

Lab 4 Lab 5

Automated Sample Processing

Gillette and Carr Nature Methods, 2013

Enrichment methods increase limits of detection

Peptides in digested sample Add 13C-labeled

signature peptide

Capture on Ab-coated magnetic beads

Native (12C) peptide Heavy Stnd (13C) peptide

MRM-MS with Ab-capture of peptides decreases assay complexity and increases robustness (iMRM or SISCAPA1)

1. Anderson et al. (2004) J. Prot. Res.

Ratio 13C-peptide to 12C-peptide by MRM-MS

12C-peptide 13C6-peptide

Time, minutes

Abu

ndan

ce

* * * * MRM-MS

Wash; elute peptides

Advantages of SISCAPA •  Simpler sample handling prior

to LC-MS/MS to reach ng/mL •  Only requires 1 Ab •  Easy to obtain useful anti-

peptide Abs (>75% success)

1000’s - >10,000 fold enrichment

When Would You Use Targeted Quantification?

•  You have a candidate list (10’s - >100) of proteins and/or phosphopeptides you want to repeatedly measure under varying conditions, cell types, etc. –  Precise, relative quantification across samples –  Control vs stimulated cell lines –  Normal vs disease-state tissue or plasma

•  Wherever change of abundance is needed –  Exact peptide sequence is monitored and quantified –  Small changes in abundance can be determined (<<2-fold)

•  Case Studies

Discovery and Verification proteomics in a human model of myocardial injury: planned myocardial infarction (PMI)

Discovery Goal: Generate list of candidate protein biomarkers

and specific pep2des to target Technique: LC-‐MS/MS with extensive frac2ona2on Sample: Proximal fluid (coronary sinus)

Femoral Vein Samples

Time (min): Baseline, 10, 60,

240, 1440

Coronary Sinus Samples

Time (min): Baseline, 10, 60

Addona et al. Nature Biotechnology (2011) 29: 635

QuanCtaCve VerificaCon Goal: Quan2fy protein abundance in cases, controls Technique: frac2on-‐MRM-‐MS and immunoassay Sample: Peripheral plasma of cases (PMI, SMI, ischemia) and controls (rou2ne catheteriza2on)

Fuzzy C-means clustering identified 333 regulated proteins in 5 clusters: pick peptides from proteins of interest from each cluster

Keshishian et al. Mol Cell Proteomics, 2015

23 Antibodies for 23 Peptides/ 13 Protein targets

Protein Peptide Ug Ab in the

mix Troponin I NITEIADLTQK 2

IL 33 TDPGVFIGVK 2 VLLSYYESQHPSNESGDGVDGK 2

ACLP Aortic carboxypeptidase-like protein 1

ILNPGEYR 2 DTPVLSELPEPVVAR 2

FHL1 four and a half LIM domains 1 isoform 5

AIVAGDQNVEYK 2 NPITGFGK 2

MYL3 Myosin light chain 3 AAPAPAPPPEPERPK 2 ALGQNPTQAEVLR 2 HVLATLGER 2

TPM1 Isoform 4 of Tropomyosin alpha-1 chain

LVIIESDLER 2 SIDDLEDELYAQK 1 HIAEDADR 2

ITGB1 Isoform Beta-1C of Integrin beta-1

GEVFNELVGK 1

TAGLN2 Transgelin-2 ENFQNWLK 2 TAGLN1 Transgelin-1 AAEDYGVIK 2

FGL2 Fibroleukin ELESEVNK 1 EEINVLHGR 2

SCUBE2 Signal peptide GSVACECRPGFELAK 2

FSTL1 Follistatin-related protein 1 IQVDYDGHCK 2 LDSSEFLK 2

SPON1 Spondin-1 VEGDPDFYKPGTSYR 2 AQWPAWQPLNVR 2

Patient samples for immunoMRM (iMRM)

Blood Draws

Baseline (pre-heparin)

10 min 1 hrs 4 hrs Baseline (post-heparin)

2 hrs 24 hrs

Planned MI patient samples (12)

MI patient samples (22)

Blood Draws

Baseline 4 hrs 24 hrs

Cath. Control patient samples (8)

Blood Draws

Baseline (pre-heparin)

10 min 1 hrs

ABI 4000QTRAP

Immunoaffinity enrichment

LC-MRM

Overview of Broad’s Automated iMRM Workflow

Desalt

Desalt using

vacuum manifold

KingFisher

Overnight capture using

Labquake

Bravo

Digest

Capture Target Pep;des •  23 an2body Mastermix •  2ug beads per ug Ab •  Crosslinked Ab to bead

Plate of 30ul Digested Plasma (not depleted) •  Add heavy pep2de

mix over ager diges2on, prior desal2ng

Wash and Elute Target Pep;des •  Wash 2x in 1x PBS, 0.03% CHAPS •  Wash 1x in 0.1x PBS, 0.03% CHAPS •  Elute in 3%ACN, 5% HOAc

Analyze •  LC-‐MRM-‐MS •  Skyline •  QuaSAR

Troponin I

FHL1

PMI MI Control

Cell Cycle Control: G2/M DNA Damage Checkpoint Cell Cycle Control: G1/S Checkpoint

Regula2on of Apoptosis Erb B / HER Signaling JAK / STAT Signaling

TGFB Signaling

Case Study 2: Quantification of Phosphopeptides

Challenges with Phosphopeptide Isoforms

LSSLRASTSK

LSSLRASTSK LSSLRASTSK

LSSLRASTSK +

y8

v  Co-‐elu2on AND same molecular mass: Limits number of transi2ons for quan2fica2on

v  Even if pep2des

don’t co-‐elute, iden2fying which is which is challenging without standards

Lys-‐C; Trypsin

Tissue homogenate, cell lysate

Add labeled phosphopep2des

IMAC

Develop targeted-MS peptide and phosphopeptide marker panels

p

LC-MRM-MS

Enriched Phosphopep2des

MRM spectrum(light and heavy)

10 ms

Targeted MS/MS(MRM-MS)

Full scan (MS/MS)spectrum

100 ms

Data dependent MS/MSa

b

Flu

ores

cenc

e

Retention time (min)

Flu

ores

cenc

e

Retention time (min)

Q2 Q3

Ionize allpeptides

Mass selectpeptide ion

Fragmentpeptide ion

Detect allfragment ions

Q1

Q2 Q3

Ionize allpeptides

Mass selectpeptide ion

Fragmentpeptide ion

Monitor ≥3fragment ions

Q1

Endogenous analyte peptideSynthetic isotope-labeled peptide(a.k.a. “internal standard”)

Collisiongas

Electrosprayemitter

Selec2vely detect and quan2fy proteins and phosphopep2des, including ischemia-‐associated markers iden2fied

YAP1 QASTDAGTAGALTPQHVR

MAPK1 VADPDHDHTGFLTEYVATR

Enrichment is necessary for phosphopeptide measurements

•  There is no significant endogenous signal for phosphorylated targets in unenriched samples

•  Enrichment increases detection of some of the endogenous peptides in cell lysate

•  After enrichment, percent of SM hits that are phosphorylated is 98-99%

Unenriched Enriched (IMAC)

GSK3B (S9)

TTSFAESCKPVQQPSAFGSMK

Measuring phosphosites to discover similari;es in cell signaling among phenotypes elicited by drug treatment (“P100 assay”)

Abelin, Jaffe et al. Proteomics PlaMorm

Example application of P100 Assay: dose response to kinase inhibitor

Abelin et al. submiNed

Case Study 3: Histones and their post-translational modifications

Source: PDB 1AOI, Luger et al. Nature (1997)

ph

me

me me me

Associated with transcrip2onal regulatory states of genomic loci

Histones are not amenable to traditional proteomics approaches

Histone H3:

•  Trypsin diges2on would create many short pep2des and would do so inconsistently due to presence of numerous inhibi2ng modifica2ons

10 20 30 40 50 60 ARTKQTARKSTGGKAPRKQLATKAARKSAPATGGVKKPHRYRPGTVALREIRRYQKSTEL 70 80 90 100 110 120 LIRKLPFQRLVREIAQDFKTDLRFQSSAVMALQEACEAYLVGLFEDTNLCAIHAKRVTIM 130 PKDIQLARRIRGERA Red: Trypsin cleavage sites

Histones are not amenable to traditional proteomics approaches

Histone H3:

•  Propionylate lysine -  Consistent pep2des -  Helps with reten2on since histone pep2des are

very hydrophilic

10 20 30 40 50 60 ARTKQTARKSTGGKAPRKQLATKAARKSAPATGGVKKPHRYRPGTVALREIRRYQKSTEL 70 80 90 100 110 120 LIRKLPFQRLVREIAQDFKTDLRFQSSAVMALQEACEAYLVGLFEDTNLCAIHAKRVTIM 130 PKDIQLARRIRGERA

Source: Garcia et al. Nature Protoc. (2007)

0 5 10 15 20 25 30 35 40 45 50 55 60 Time (min)

50

100

Rela2ve Ab

undance

33.50 551.9957

42.41 575.8434

37.90 549.8186 26.10

541.3210 43.66 582.8521

31.88 547.3229 21.58

526.3099

59.23 309.1305 49.74

694.3874 9.96

399.7397 58.04 343.2953 21.06

490.7913 46.75 743.9099 57.48

320.2039 0.91 371.1012

0

H3K4me3 H3K4me2

H3K18ac0K23ac1

H3K27me2K36me1

H3K18ac0K23ac0 H3K27ac1K36me3

H3K9me3K14ac1

H3-‐41-‐49

H3K9me2K14ac0

Highly mul;plexed MRM-‐MS is an equally valuable tool in biology: assays for modified pep;des

modifica;ons

Cell lines

Monomethyl = 14.0157 Dimethyl = 28.0313 Trimethyl = 42.0470 Acetyl = 42.0106 Phosphoryl = 79.9663 Ubiquityl stub = 114.0429

Benefits of Targeted MS for peptide/protein assays

•  High molecular specificity •  Quantitative

•  Works in any matrix (cells, tissue, plasma, urine, etc.)

•  Works equally well for posttranslationally modified peptides •  ‘Sensitivity and assay performance already sufficient to assay many

proteins of interest (mid-pg/mL to high ug/mL)

•  Interferences can detected and avoided (unlike Westerns, IHC, immunoassays, aptamers)

•  Does not require immunoassay-grade antibodies (2/protein)

•  Assays can be highly multiplexed (>100 plex now routine) •  Not possible with antibodies; can be done with aptamers

•  Assays are transferable with high interlab reproducibility •  Multiple MS platforms now capable of targeted MS experiments (not just

triple quads any longer)

References

1.  Paola Picor & Ruedi Aebersold. Selected reac2on monitoring–based proteomics: workflows, poten2al, pitalls and future direc2ons. Nature Methods. 9, 555 – 566 (2012)

2.  Gilleue MA and Carr SA. Quan2ta2ve analysis of pep2des and proteins in biomedicine by targeted mass spectrometry. Nat Methods (2013) 10(1): 28-‐34.

3.  Carr SA et al. Targeted Pep2de Measurements in Biology and Medicine: Best Prac2ces for Mass Spectrometry-‐based Assay Development Using a Fit-‐for-‐Purpose Approach. Mol Cell Proteomics (2014) 13(3): 907-‐917.

4.  Addona T. et al., “A pipeline that integrates discovery and verifica2on of plasma protein biomarkers reveals candidate markers for cardiovascular disease”, Nature Biotechnology (2011) 29: 635

5.  Creech AL et al. Building the Connec2vity Map of epigene2cs: Chroma2n profiling by quan2ta2ve targeted mass spectrometry. Methods (2014) Nov 6.

6.  Abba2ello SE et al. Large-‐scale inter-‐laboratory study to develop, analy2cally validate and apply highly mul2plexed, quan2ta2ve pep2de assays to measure cancer-‐relevant proteins in plasma. Mol Cell Proteomics (2015) Feb 18. pii: mcp.M114.047050. in press

7.  Keshishian H Mul2plexed, Quan2ta2ve Workflow for Sensi2ve Biomarker Discovery in Plasma Yields Novel Candidates for Early Myocardial Injury. Mol Cell Proteomics (2015) Feb 27. pii: mcp.M114.046813. in press

Targeted(MS(and(its(Applicaon( in(Biology(and(Medicine( · PDF fileTargeted proteomics:...

Documents

Transcript of Targeted(MS(and(its(Applicaon( in(Biology(and(Medicine( · PDF fileTargeted proteomics:...