Transcript Label-free

Quantitative Proteomics:
Applications and Strategies
Gustavo de Souza
IMM, OUS
October 2013
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
2DE-based approach
“I see 1000 spots, but identify 50 only.”
LC-MS
Column (75 mm)/spray tip (8 mm)
Reverse-phase C18 beads, 3 mm
No precolumn or split
15 cm
Sample Loading:500 nl/min
Gradient elution:200 nl/min
ESI
Platin-wire
2.0 kV
Fenn et al., Science 246:64-71, 1989.
MS-based quantitation
Inlet
LC
Ion
Source
Mass
Analyzer
MALDI
ES
Time-of-Flight
Quadrupole
Ion Trap
Quadrupole-TOF
Peak intensities can vary up to
100x between duplicate runs.
Detector
Quatitative analysis MUST be
carried on a single run.
Ion Intensity = Ion abundance
MS measure m/z
Sample 2
Intensity
Sample 1
m/z
Isotopic Labeling
Unlabeled
peptide:
a)
b)
Element
1
H
Labeled
peptide:
a)
b)
Stable Isotope
2
H
12C
13C
14
15
N
16O
N
18O
Enzymatic Labeling
Metabolic Labeling
SILAC
Cells in normal culture media
Media with
Normal AA ()
Media with
Labelled AA (*)
Start SILAC labelling by growing
cells in labelling media
m/z
m/z
(labelled AA / dialized serum)
*
m/z
Passage cells to allow incorporation
of labelled AA
m/z
*
m/z
By 5 cell doublings cells have
incorporated
m/z
*
X3
m/z
Grow SILAC labelled cells to desired
number of cells for experiment
X3
m/z
Ong SE et al., 2002
Chemical Labeling
ICAT Reagents: Heavy reagent: d8-ICAT (X=deuterium)
Light reagent: d0-ICAT (X=hydrogen)
O
N
N
S
Biotin
tag
O
XX
N
O
XX
O
XX
O
XX
Linker
(heavy or light)
O
N
I
Thiol specific
reactive group
Gygi SP et al., 1999
ICAT (Isotope-Coded Affinity Tag)
ICAT
Cell State 1
(All cysteines labeled with
light ICAT)
ICAT)
Combine
Optional fractionation
Pr
ot
ei
n
A
Pr
ot
ei
n
B
Pr
ot
ei
n
C
Pr
ot
ei
n
D
Pr
ot
ei
n
E
Pr
ot
ei
n
F
Relative Abundance
100
....
0
Time
Quantitate relative protein levels by measuring peak ratios
Proteolyze
Identify proteins by sequence information (MS/MS scan)
Cell State 2
(All cysteines labeled
with heavy ICAT)
Analyze by LCLC-MS/MS
Relative Abundance
100
Affinity separation
=Protein A
NH2-EACDPLREACDPLR-COOH
0
200
400
600
m/z
Thiol-specific group = binds to Cysteins
800
ICAT
Thiol-specific group = binds to Cysteins
Intensity
Quantitation at MS1 level
m/z
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)
Recognizes
Arg or Lys
Total mass of label
= 145 Da ALWAYS
Sample prep
iTRAQ
iTRAQ
Multiplexing
Metabolic VS Chemical Labeling
• Metabolic labeling
- 15N labeling
- SILAC
• Chemical methods
- many… but ICAT is
prototype
Living cells
Efficient labeling
Simple!
Isolated protein sample
Depends on chemistry
Multi-step protocols
Require optimization
Summary
Kolkman A et al., 2005
Label-free
Mobile phase
C18 column, 25cm long
A
B
20 s
Time
A = 5% organic solvent in water
B = 95% organic solvent in water
Label-free
Strassberger V et al., 2010
Summary
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 A
State B
Upregulated protein - Peptide ratio >1
Arg13C
6
Light Isotope
Heavy Isotope
Mix 1:1
Arg12C
6
Optional Protein
Fractionation
m/z
Digest with Trypsin
Control vs Tumor Cell?
Protein Identification and
Quantitation by LC-MS
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
Gustavo de Souza
IMM, OUS
October 2013
SILAC
Cells in normal culture media
Media with
Normal AA ()
Media with
Labelled AA (*)
Start SILAC labelling by growing
cells in labelling media
m/z
m/z
(labelled AA / dialized serum)
*
m/z
Passage cells to allow incorporation
of labelled AA
m/z
*
m/z
By 5 cell doublings cells have
incorporated
m/z
*
X3
m/z
Grow SILAC labelled cells to desired
number of cells for experiment
X3
m/z
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?
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 A
State B
Upregulated protein - Peptide ratio >1
Arg13C
6
Light Isotope
Heavy Isotope
Mix 1:1
Arg12C
6
Optional Protein
Fractionation
m/z
Background protein - Peptide ratio 1:1
Digest with Trypsin
Arg12C
6
Arg13C
6
Protein Identification and
Quantitation by LC-MS
m/z
Additional validation criteria
* Never use labelled Arg or Lys with same
mass difference (Lys6/Arg6)
Triple SILAC
Triple Encoding SILAC allows:
3
Intensity
2
Monitoring of three cellular
states simultaneously
Study of the dynamics of
signal transduction cascades
even in short time scales
m/z
Blagoev B et al., 2004
Five time-point “multiplexing” profile
Blagoev B et al., 2004
Quantitative phosphoproteomics in EGFR signaling
8x
0’ EGF
8x
1’ EGF
0-5-10 min.
8x
5’ EGF
Cytoplasmic ext.
Nuclear extract
8x
5’ EGF
8x
10’ EGF
8x
20’ EGF
SILACHeLa cells
1-5-20 min.
Cytoplasmic ext.
Nuclear extract
Lysis and
Fractionation
Anf digestion
SCX / TiO2
SCX / TiO2
SCX / TiO2
4x (10 SCX-fractions +FT)
44 LC-MS runs
SCX / TiO2
Phosphopeptide
enrichment
ID and
quantitation
Blagoev B et al., 2004
MAP kinases activation
Signal progression
40
EGFr-pY1110
ShcA-pY427
ERK1-pY204
ERK2-pY187
EMS1-pS405
10
2
1
5
10
15
EGF (minutes)
20
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
Gustavo de Souza
IMM, OUS
October 2013
Label-free
Label-free
10 s
500 fmol peptide
Time
100 fmol peptide
Time
Strassberger V et al., 2010
Label-free
Kiyonami R. et al, Thermo-Finnigan application note 500, 2010.
Measurements
Label-free
Ideal (low std)
x
x
x
x
x
x
Measurements
Replicates
Reality (late 90’s)
x
x
x
x
x
x
Replicates
Label-free
Strassberger V et al., 2010
Label-free
Neilson et al., Proteomics 2011
Spectral Count
MS1 (or MS)
899.013
899.013
MS2 (or MS/MS)
899.013
Spectral Count
20 s
Time
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
MS scan
MS2 scan
2 counts
Time
2 counts
Time
Area Under Curve measurement
Ion Intensity
AUC
Retention Time
Area Under Curve measurement
Ion intensity
in one MS1
MS2 scan
Retention Time
Importance of Resolution for label-free
2+
2+
3+
3+
RT
RT
m/z
m/z
Importance of Resolution for label-free
-Label-free became reliable (*)
Cox and Mann, Nature Biotechnol 26, 2008.
Area Under Curve measurement
AV: 1 NL: 3.43E6
798.32
798.83
2
1
799.33
3
799.83
82
794
x
795.17
795
802.13
797
1. Retention time
2. Peak intensity
3. Monoisotopic mass accuracy
803.40
797.73
796.31
796
800.32
802.72
798
799
m/z
800
801
802
803
805.84
804
805
Cox and Mann, Nature Biotechnol 26, 2008.
Regarding Label-free…
- Calculate individual peptide “Intensity”.
Protein Intensity = mean of peptides intensities
- LFQ normalization
Data without Normalization
-7422 proteins identified
- 7105 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, 2008.
-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
Label-free in primary cells
Infection with Sendai virus
(activate RIG-I PRR)
RIG-I knockout
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
Gustavo de Souza
IMM, OUS
October 2013
A little history…
So far, ID everything we can
Mobile phase
C18 column, 25cm long
A
20 s
Time
B
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
- 700-6000 genes
- No alternative splicing
- Limited PTM presence
Discovery Phase
Validation on metabolic network
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
- Targeted analysis: ignore whole sample and focus in few
protein.
- 1st 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.