Quantitative Proteomics: Applications and Strategies October 2013 Gustavo de Souza IMM, OUS

A little history… 1985 – First use: up to a 3 kDa peptide could be ionized 1987 – Method to ionize intact proteins (up to 34 kDa) described Instruments have no sequence capability 1989 – ESI is used for biomolecules (peptides) Sequence capability, but low sensitivity 1994 – Term «Proteome» is coined 1995 – LC-MS/MS is implemented «Gold standard» of proteomic analysis

2DE-based approach

“I see 1000 spots, but identify 50 only.”

Gradient elution:200 nl/min Column (75 mm)/spray tip (8 mm) Reverse-phase C18 beads, 3 mm Platin-wire 2.0 kV Sample Loading:500 nl/min No precolumn or split ESI 15 cm Fenn et al., Science 246:64-71, LC-MS

MS-based quantitation Inlet Ion Source Mass Analyzer Detector MALDI ES Time-of-Flight Quadrupole Ion Trap Quadrupole-TOF LC Peak intensities can vary up to 100x between duplicate runs. Quatitative analysis MUST be carried on a single run.

Ion Intensity = Ion abundance

MS measure m/z m/z Intensity Sample 2 Sample 1

Isotopic Labeling Unlabeled peptide: Labeled peptide: a)b)a)b)

Enzymatic Labeling

Metabolic Labeling

SILAC * m/z Passage cells to allow incorporation of labelled AA By 5 cell doublings cells have incorporated * m/z Grow SILAC labelled cells to desired number of cells for experiment * m/z Start SILAC labelling by growing cells in labelling media (labelled AA / dialized serum) m/z Media with Normal AA (  ) Media with Labelled AA (*) X 3 Cells in normal culture media Ong SE et al., 2002

Chemical Labeling Gygi SP et al., 1999

ICAT (Isotope-Coded Affinity Tag)

ICAT Thiol-specific group = binds to Cysteins

ICAT Thiol-specific group = binds to Cysteins

Quantitation at MS1 level m/z Intensity Double sample complexity, i.e. instrument have more “features” to identify, i.e. decrease in identification rate

iTRAQ (isobaric Tag for Relative and Absolute Quantitation) Sample prep Total mass of label = 145 Da ALWAYS Recognizes Arg or Lys

iTRAQ

Multiplexing

Metabolic VS Chemical Labeling

Summary Kolkman A et al., 2005

Label-free Mobile phase A A = 5% organic solvent in water B = 95% organic solvent in water B C18 column, 25cm long Time 20 s

Label-free Strassberger V et al., 2010

Summary

Take home message 1.Quantitation can be done gel-free 2.Labeling can be performed at protein or peptide level, during normal cell growth or in vitro 3.Quantitation can be achieved at MS1 or MS2 level 4.Method choice depends on experimental design, costs, expertise etc 5.In my PERSONAL OPINION, chemical label should be avoided at all costs unless heavy multiplexing is required

Applications State AState B Light IsotopeHeavy Isotope Mix 1:1 Optional Protein Fractionation Digest with Trypsin Protein Identification and Quantitation by LC-MS Upregulated protein - Peptide ratio >1 Control vs Tumor Cell? Control vs drug treated cell? Control vs knock-out cell?

Applications – Cell Biology Geiger T et al., 2012

Applications – Cell Biology

Applications – Immunology Meissner et al, Science 2013

Clinical Proteomics A. Amyloid tissue stained in Congo Red; B. After LMD. Wisniewski JR et al., 2012

Interactomics Schulze and Mann, 2004 Schulze WX et al., 2005

Signaling Pathways

Take home message 1.Anything is possible!

SILAC October 2013 Gustavo de Souza IMM, OUS

SILAC * m/z Passage cells to allow incorporation of labelled AA By 5 cell doublings cells have incorporated * m/z Grow SILAC labelled cells to desired number of cells for experiment * m/z Start SILAC labelling by growing cells in labelling media (labelled AA / dialized serum) m/z Media with Normal AA (  ) Media with Labelled AA (*) X 3 Cells in normal culture media Ong SE et al., 2002

Importance of Dialyzed Serum non-dialzed serum contains free (unlabeled) amino acids!

No alterations to cell phenotype C2C12 myoblast cell line Labeled cells behaved as expected under differentiation protocols

Why SILAC is convenient?

Convenient - no extra step introduced to experiment, just special medium Labeling is guaranteed close to 99%. All identified proteins in principle are quantifiable Quantitation of proteins affected by different stimuli, disruption of genes, etc. Quantitation of post-translational modifications (phosphorylation, etc.) Identification and quantitation of interaction partners

Catch 22 - SILAC  custom formulation media (without Lys and/or Arg) $$$$$$ - Labeled amino acids – Lys4, Lys6, Lys8, Arg6, Arg10. Use formulation accordingly to media formula (RPMI Lys, 40mg/L) ***** When doing Arg labeling, attention to Proline conversion! (50% of tryptic peptides in a random mixture predicted to contain 1 Pro)

Proline Conversion!

Typical SILAC experiment workflow State AState B Light IsotopeHeavy Isotope Mix 1:1 Optional Protein Fractionation Digest with Trypsin Protein Identification and Quantitation by LC-MS Upregulated protein - Peptide ratio >1 Background protein - Peptide ratio 1:1

Additional validation criteria * Never use labelled Arg or Lys with same mass difference (Lys6/Arg6)

Triple SILAC Triple Encoding SILAC allows:  Monitoring of three cellular states simultaneously  Study of the dynamics of signal transduction cascades even in short time scales m/z Intensity 3 2 Blagoev B et al., 2004

Five time-point “multiplexing” profile Blagoev B et al., 2004

Quantitative phosphoproteomics in EGFR signaling Blagoev B et al., 2004 SILAC- HeLa cells 0’ EGF 1’ EGF 5’ EGF 10’ EGF 20’ EGF min. Cytoplasmic ext. Nuclear extract Lysis and Fractionation Anf digestion min. Cytoplasmic ext. Nuclear extract SCX / TiO2 Phospho- peptide enrichment 44 LC-MS runs 4x (10 SCX-frac- tions +FT) ID and quantitation 8x

MAP kinases activation 40 EGF (minutes) EGFr-pY1110 ShcA-pY427 ERK1-pY204 ERK2-pY187 EMS1-pS405 Relative ratios Signal progression

Spatial distribution of phosphorylation dynamics Cytosolic STAT5 translocates to the nucleus upon phosphorylation

Interactomics Schulze and Mann, 2004 Schulze WX et al., 2005

Limitations -Expensive -Quantitation at MS1 level  increased sample complexity -Cells has to grow in culture. Not a choice for primary cells, tissues or body fluids. -Cell lines have to be dyalized serum-friendly.

SILAC-labeled organism Sury MD et al., 2010

Super-SILAC Geiger T et al., 2010

Spike-In SILAC Geiger T et al., 2013

Take home message 1.Arguably the best labeling strategies: easy to handle, no chemical steps, >98% incorporation  low variability 2.Successfully used in the most diverse applications 3.Cells must be stable and growing in the media 4.There are decent alternative strategies for primary cells or organisms.

Label-free October 2013 Gustavo de Souza IMM, OUS

Label-free

Strassberger V et al., 2010 Time 10 s Time 500 fmol peptide 100 fmol peptide

Label-free Kiyonami R. et al, Thermo-Finnigan application note 500, 2010.

Label-free Replicates Measurements x x x x x x Ideal (low std) Replicates Measurements x x x x x x Reality (late 90’s)

Label-free Strassberger V et al., 2010

Label-free Neilson et al., Proteomics 2011

Spectral Count MS1 (or MS) MS2 (or MS/MS)

Spectral Count Time 20 s Time Depending on how complex the sample is at a specific retention time, the machine might be busy (i.e., doing many MS2) or idle (i.e., few or none MS2)

Limitation in Spectral Count Time MS scan MS2 scan 2 counts

Area Under Curve measurement Retention Time Ion Intensity AUC

Area Under Curve measurement MS2 scan Ion intensity in one MS1 Retention Time

Importance of Resolution for label-free RT m/z RT m/z 2+ 3+

Cox and Mann, Nature Biotechnol 26, Label-free became reliable (*) Importance of Resolution for label-free

1. Retention time 2. Peak intensity 3. Monoisotopic mass accuracy x Cox and Mann, Nature Biotechnol 26, Area Under Curve measurement

Regarding Label-free… - Calculate individual peptide “Intensity”. Protein Intensity = mean of peptides intensities - LFQ normalization

Data without Normalization proteins identified proteins quantified (95.72%)

How this was demonstrated? Yeast model Ghaemmagami S. et al., Nature 425, 2003 Huh WK. et al., Nature 425, 2003

How this was demonstrated? Ghaemmagami S. et al., Nature 425, 2003

MaxQuant and Yeast De Godoy LM. et al, Label-free became reliable AND showed good correlation with a well-established model

Label-free in primary cells

Higher CD4+ Higher CD8+ Pattern Recognition Receptors Pathway

Infection with Sendai virus (activate RIG-I PRR) RIG-I knockout Label-free in primary cells

Take home message 1.“Labe-free” represents a myriad of ANY method that does not use any labeling 2.Area Under Curve calculations are the most appropriate 3.Reliability is heavily dependent in good instrumentation and good bioinformatics (MaxQuant) 4.Currently, almost as good as SILAC (yet slightly less accurate)

SRM / MRM October 2013 Gustavo de Souza IMM, OUS

A little history…

So far, ID everything we can Mobile phase AB C18 column, 25cm long Time 20 s

Targeted analysis In some cases, the researcher don’t want the MS instrument to waste time trying to sequence as much as possible, but just to “search” and sequence pre-determined peptides. -Biomarker research -Tracking specific metabolic pathways -Tracking low abundant proteins in challenging sample (f.ex., in serum)

Plasma dynamic range Schiess R et al., 2009

Improving detection through tergeting Michalski A et al., 2011

Biomarker

Discovery phase Screening the sample gives you the following info: -For protein X  most intense peptides (not all peptides from same protein have the same intensity) - most common m/z format (+2, +3, PTM?) - their Retention times - their fragmentation profiles (does the +2 fragments well?)

Biomarker

Shorter gradient = More complex MS1 As you decrease separation resolution, you increase the chance that two or more peptides with different sequences BUT very close m/z elutes at the same time.

SRM (Selected Reaction Monitoring)

Different transitions from same peptide

Performance with synthetic peptides

Shorter gradient = More complex MS1 As you decrease separation resolution, you increase the chance that two or more peptides with different sequences BUT very close m/z elutes at the same time.

Number of biomarkers discovered so far by MS 0

Spiking sinthetic labeled peptide for absolute quantitation

Applying SRM to a proper model Bacterial genomic structure genes -No alternative splicing -Limited PTM presence

Discovery Phase

Validation on metabolic network

-It open possibilities to study molecular function implications at metabolic level. -Generate knockout, discovery phase to visualize pahways possibly altered by the KO, targeted the candidate pathways for in-depth quantitation.

Take home message - 1 st step is to make the regular analysis to collect acquisition features for as many peptides as possible. - Relevant in Biomarker research - Very challenging for complex samples, very powerful for simpler organisms and for pure biology projects. - Targeted analysis: ignore whole sample and focus in few protein.