Global Internal Standard Technology GIST A Tool for Protein Expression Analysis

1
Global Internal Standard Technology (GIST) A
Tool for Protein Expression Analysis
Xiang Zhang Bindley Bioscience Center Purdue
University
2
Identification of changes in protein expression
and its modification is essential for
understanding biological processes

• Cellular response to stimuli is reflected by
  changes, i.e. protein expression,
  post-translation modification or processing
• Stimuli origin
• Chemical (drug, toxin etc.)
• Physical (cell interaction, changes in
  temperature, pressure etc.)
• Combination of both (disease)

The search for differences in protein expression
and modification is called Comparative Proteomics.
3
The Key Element of Comparative Proteomics is
Quantitation of Changes in Protein (Peptide)
Levels

• It is not easy to determine a change in a single
  protein level (such as Western Blot)
• In comparative proteomics, the challenge is to
  identify changes for as many proteins (peptides)
  as possible
• There are several approaches for quantitation
• Pattern recognition approach
• Isotopic labeling approach

4
Pattern Recognition
Alignment, normalization and peak
intensity comparison
Individual analysis of sample A and B by LC-MS

• Works well but has some potential issues
• Strongly depends on LC-MS system reproducibility
• Intensity of any peak is not only function of
  peptide concentration. It also depends on analyte
  composition
• It is difficult to obtain direct fold changes
  between samples
• Some of these potential issues could be overcome
  by using isotopic labeling strategies.

5
Isotopic Labeling

• Biosynthetic labeling
• In vivo incorporation of isotopic labeled species
  (growing cells in media enriched in 14N vs. 15N)
• Impossible to use it with human subjects
• Post-biosynthetic labeling.
• Labeling amino groups (GIST)
• Labeling cysteine residues (ICAT)
• 18O incorporation during proteolysis

6
Isotopic Labeling Basic principles (ICAT)
7
GIST Isotopic Labeling Technique
Concept was first introduced by Fred Regnier lab
at Purdue University in 2000

• Labeling reagents
• Succinimidyl propionate (12C vs 13C)
• Succinimidyl acetate (1H vs 2H)
• Target group
• Primary amine (N-terminus, Lysine residue)
• Sample is labeled following digestion

8
GIST Chemical structure of Labeling Reagents
9
Generating primary amines
Trypsin cleaves polypeptides C-terminal to lysine
and arginine
-NH-CH(R1)-CO-NH-CH(R2)-CO-
trypsin
-NH-CH(R1)-COOH
H2N-CH(R2)-CO-
Primary amine groups are present globally - every
peptide generated will be labeled by GIST reagents
10
Amino groups are easily alkylated
Note that Arg is not acetylated.
All primary amino groups are labeled.
11
GIST Experimental Design
12
Labeling samples from two different sources
13
Example of GIST labeled peptide analyzed by Mass
Spectrometry
542
545
peptide from experimental sample labeled with
heavy form
peptide from control sample labeled with light
form
m/z
14
How to Analyze GIST Data?
603.81
600.79

• Peak picking
• Identify peptide peak cluster
• Doublet identification
• Ratio calculation

630.28
627.25
15
Process Flow Chart of GISTool
16
Data Acquired on qTOF
Profile data
Centroid data
17
Chemical Noise Filter

• I. Peak Density Filter
• spectrum is segmented
• noise level of each segment is calculated based
  on local peak density
• noise level is smoothed across spectrum

II. Spike Filter
One third of user defined peak width is used as
minimum width of isotopic peak
18
Charge Deconvolution simple case

• Peptide can carry different charges in ESI
  experiment
• Peptide charge can be used for doublet
  recognition
• Some overlapped peptides can be resolved

19
Charge Deconvolution complicated case
white and red peptides both have 1 charge, but
they are shifted by 0.5 Da

• Identify peak group
• Find base peak
• Try different charges
• Assigned 2 to the group
• Check isotopic peak profile
• Flag white peaks
• Check M/Z space of the white peaks
• Assign white peaks as 1
• Search other white peaks in the group
• Try 1 on red peaks
• Search other reds in peak group

20
Deisotope

• Quantitatively resolve overlapped peptide peaks
• Simplify peak list

Simple case
Overlapped peptides
M2M0
M0
M0
M1
M3M1
M1
M2
M4M2
M3
21
How Do We Deisotope ?
Peptide isotopic peak profile can be calculated
if AA composition is known.
Peptide CmHnOoNpSq (12C13C)m(1H2H)n(16O17O1
8O)o(14N15N)p(32S33S34S36S)q

• peptide sequence is unknown during data
  processing
• in-silico prediction of isotopic peak profile
• comparing in-silico profile with experimental data

22
Peptide Isotopic Distribution Can Be Predicted
Large peptide
Small peptide
Relative intensity
Peptide B
Peptide A
Molecular weight (amu)
Molecular weight (amu)
Peptide A and B have the same molecular weight
but different AA composition

• Peptide isotopic peak profile varies
  significantly with MW
• To certain MW, the variation of isotopic peak
  profile is not significant

23
Correlating in-silico Results with Experimental
Results
Detect the significantly intense isotopic peaks
by comparing experimentally measured isotopic
peak profile with with in-silico predicted peak
profile - No, there is no other peptide - Yes, a
peak from other peptide contributes to the
current peak
Relative Intensity
M/Z
MW
24
Charge Deconvolution and Deisotoping Process
25
Doublet Recognition

• Mass difference
• Retention time

GIST Acetate 3, 6, ICAT 8, 16,
Shifting according to labeling reagent.
26
Calculation of Peak Ratio
I. Ratio is calculated in each scan good for
12C/13C pairs
II. Ratio is calculated after smoothing peptide
peaks at chromatographic level good for H/D or
16O/18O pairs

• SG smooth each peak
• Peak detection
• Doublet recognition

Peptide Regulation
27
Ranking Doublets
RANK 1, 2, 3, 4,
5
Good doublet Doublet complex
Singlet
28
Decharging to Rescue Mixed Doublets
PeptideA 2 light heavy
PeptideB 2 light heavy
500
505
6 possible combinations
29
Overview of MS Proteomics Platform
Proteins
GIST label
Mass Spectrometere.g. Ion Trap, Q-Tof
Separation
Ionization
Digest
RPLC/CapLC
ESI
Purified Separated
MS/MS (fragmentation pattern)
Survey Scan MS
MS-only Ion Chromatogram (stop here for
quantitative profiling)
select ion
Time
m/z
m/z
Protein Database
Theoretical MS/MS Spectrum
SEQUEST(CorrelationAnalysis)
ProteinIdentification
30
Typical GIST Doublet LightHeavy

• shows 3 m/z unit separation
• (6 Da difference) since 2 ion

State 1 (normal) 1H or 12C
m/z
31
Are Singlets Ever Observed?
PQVSTPTLVEVSR
singlet

• Pro has no primary amine,
• also no Lys present

doublet

• There is usually a scientific explanation for
  singlets

32
Effectiveness of GIST Technology

• Peak ratio distribution is rather tight
• Labeling efficiency is gt 95 no mixed
  populations of
• labeled and unlabeled peptides are observed
• 13C labeled peptides show greater deviation from
  the
• mean than 2H
• Mean ratio values below 1.0 and 3.0 are most
  likely
• due to experimental error

33
GIST Results for 8 Protein Mixture
Experimental design Create an unknown mixture
of 8 standard proteins
in different concentrations and
different ratios to test
the entire GIST process (labeling,
MS, GISTool, etc.)
Protein Actual Ratio Exp.
Ratio peptides IDd BSA 51
4.81 17 Transferrin 31
3.31 10 Glucose oxidase 13
12.1 13 Lysozyme 11 1.11
4 Carbonic Anhydrase 71
5.11 4 Lactoglobulin 101
only Light detected 7 Myoglobin
11 1.31 1 Ovalbumin
15 N/A unidentified

• Proteins identified by MS/MS sequences

• Ratios determined by GISTool from MS only data

• Ratios were successfully determined for most
  proteins in mix

34
A Real Sample Human Serum
Experimental design
Samples 1 normal vs. 1 colon cancer
Protein Level Fractionation

• No peptide level fractionation other than
  RP-LC/MS
• A single protein fraction (175mM, 40 ACN) from
  1 normal and 1 diseased
• sample was digested separately with trypsin,
  GIST labeled, and mixed

35
Serum Fold Changes Using Deuterium and 13C GIST
Reagents
GIST Acetate Normal(1H) vs. Disease(2H)
GIST Propionate Normal(12C) vs. Disease(13C)
10
10
1
1
log (Ratio)
log (Ratio)
350
600
850
1100
1350
1600
350
600
850
1100
1350
1600
0.1
0.1
m/z
m/z

• Only well-resolved doublets ranked 1-3 by
  GISTool are shown
• Only fold changes 21 were considered
  significant

36
Largest Fold Change Identified in Human Serum
Heavy
colon cancer
LC/MS
normal
Light

• Up-regulation of deoxyhemoglobin indicates
  inadequate oxygen levels in blood
• Diseases such as cancer usually result in
  deficient oxygen supplies to tissues

LC/MS/MS
deoxyhemoglobin
MS/MS data
37
GIST Fold-Change Results from Normal vs. Diseased
Serum

• Identifications from a single data-dependent
  LC/MS/MS experiment

38
Conclusions

• GIST is an efficient global peptide isotopic
  labeling technique that can be used for protein
  expression analysis and can be used with many
  selections of chromatography
• GISTool is a very successful software algorithm
  for analyzing any isotope-labeled data generated
  by high-resolution MS
• (ICAT, etc.-not just GIST)
• Since GIST is capable of easily obtaining direct
  fold changes,
• PTM and/or post-translational processing
  experiments can be designed
• Peptide fractionation will be necessary for
  complex samples since
• isotopic-labeling strategies double the
  complexity of samples