Application of omics technologies in toxicology:

Proteomics and Metabolomicsroteomics and Metabolomics

Most Commonly Used Proteomics Techniques: Antibody arrays

Protein activity arrays

2-D gels

“Shotgun” proteomics

ICAT technology

SELDI

100% protein sequence coverage: a modern form of surrealism in proteomics. Meyer et al Amino Acids. 2010 Jul 13. 

• Screening protein-protein interactions • Studying protein posttranslational modifications• Examining protein expression patterns

Antibody Arrays

Antibody Arrays

The layout design of the BD Clontech™ Ab Microarray 380. The BD Clontech™ Ab Microarray 380 (#K1847-1) contains 378 monoclonal antibodies arrayed in a 32 x 24 grid. Each antibody is printed in duplicate. Dark gray dots at the corners represent Cy3/Cy5-labeled bovine serum albumin (BSA) spots, which serve as orientation markers. The open circles correspond to unlabeled BSA spots, which serve as negative controls. For complete descriptions of the proteins profiled by the Ab Microarray 380, visit bdbiosciences.com

Limitations, Challenges and Bottlenecks

• Protein production: ►cell-based expression systems for recombinant proteins► purification from natural sources► production in vitro by cell-free translation systems► synthetic methods for peptides

• Immobilization surfaces and array formats:► Common physical supports include glass slides, silicon,

microwells, nitrocellulose or PVDF membranes, microbeads

• Protein immobilization should be:► reproducible► applicable to proteins of different properties (size, charge,

…)► amenable to high throughput and automation, and

compatible with retention of fully functional protein activity► such that maintains correct protein orientation

• Array fabrication:► robotic contact printing► ink-jetting► piezoelectric spotting► photolithography

Panomics® Transcription Factor Arrays:

A set of biotin-labeled DNA binding oligonucleotides (TranSignal™ probe mix) is preincubated with any nuclear extract of interest to allow the formation of protein/DNA (or TF/DNA) complexes;

The protein/DNA complexes are separated from the free probes;

The probes in the complexes are then extracted and hybridized to the TranSignal™ Array. Signals can be detected using either x-ray film or chemi-luminescent imaging. All reagents for HRP-based chemiluminescent detection are included.

Protein Activity Arrays

Source: Panomics, Inc.

Protein Activity Arrays

Gel Shift Assay Protein Array

Source: Panomics, Inc.

2D Gel Electrophoresis + Mass Spectrometry

Meyer et al Amino Acids. 2010 Jul 13. 

2D Gel Electrophoresis Protein Resolution

Bandara & Kennedy (2002)

2D Gel Electrophoresis Image Analysis

Courtesy of Decodon

Courtesy of Alphainnotech

Inlet

Ionization

Mass Analyzer

Mass Sorting (filtering)

Ion Detector

Detection

Ion Source

• Solid• Liquid• Vapor

Detect ionsForm ions

(charged molecules)Sort Ions by Mass (m/z)

1330 1340 1350

100

75

50

25

0

Mass Spectrum

Acquiring a Mass SpectrumAcquiring a Mass Spectrum

All compounds must be ionized, but ionization efficiency is variable with different compounds

High voltage applied to metal sheath (~4 kV)

Sample Inlet Nozzle(Lower Voltage)

Charged droplets

++

++++

++

++++

++

++++ +++

+++ +++

+++ +

++

+

+

+

+

+++

+++

+++

MH+

MH3+

MH2+

Pressure = 1 atmInner tube diam. = 100 um

Sample in solution

N2

N2 gas

Partialvacuum

Electrospray ionization:

Ion Sources make ions from sample molecules

Typical MS SpectraTypical MS Spectra

2D Gel ElectrophoresisMass Spectrometry

Source: UNC Proteomics Core Facility

SEQUEST is a program that uses raw peptide MS/MS data (off TSQ-7000 or LCQ) to identify unknown proteins. It works by searching protein and nucleotide databases (in FASTA format) on the web for peptides that match the molecular weight of the unknown peptides produced by digestion of your protein(s) of interest. Theoretical MS/MS spectra are then generated and a score is given to each one. The top 500 scored theoretical peptides are retained and a cross correlation analysis is then performed between the un-interpreted MS/MS spectra (real MS/MS spectra) of unknown peptides with each of the retained theoretical MS/MS spectra. Highly correlated spectra result in identification of the peptide sequences and multiple peptide identification and thus determine the protein and organism of origin corresponding to the unknown protein sample.

Ima

ge

co

urt

esy

of

Un

ive

rsity

of

Ariz

on

a P

rote

om

ics

Co

re

Proteins are analyzed by standard shotgun proteomics, beginning with tryptic digest of a protein mixture, liquid chromatographic separation of the mixture (2D HPLC), analysis of peptide masses by mass spectrometry (MS) and fragmentation of peptides and subsequent analysis of the fragmentation spectra (MS/MS). Each step introduces bias into the peptides ultimately interpreted from the analysis, thereby affecting the probability pij of observing each peptide j from protein i. APEX involves training a classifier to estimate Oi, the prior estimate of the number of unique peptides expected from a given protein during such an experiment. By correcting for Oi, the number of peptides observed per protein thereby provides an estimate of the protein's abundance. HPLC, high-performance liquid chromatography. Nature Biotechnology 25, 117 - 124 (2007)

SHOTGUN PROTEOMICS

Limitations, Challenges and Bottlenecks

• Resolution: ► number of proteins that can be separated/distinguished (500,000?!?)► pI resolution► mass resolution (gels and mass spectrometry)

• Amount of the protein in the sample:► too little to be seen on a 2D gel?► too little to be extracted and digested?

• Protein solubility• Database searching and peptide identification

Bandara & Kennedy (2002)

Schneider LV, Hall MP. Drug Discov Today. 2005 10:353-63.

Two-dimensional electrophoretic analysis of rat liver total proteins. The proteins were separated on a pH 3–10 nonlinear IPG strip (left), or pH 4-7 IPG strip (right), followed by a 10% SDS–polyacrylamide gel. The gel was stained with Coomassie blue. The spots were analyzed by MALDI-MS. The proteins identified are designated with the accession numbers of the corresponding database.

From Fountoulakis & Suter (2002)

• In total, 273 different gene products were identified from all gels:65 gene products were only detected in the gels carrying total52 in the gels carrying cytosolicremaining proteins were found in both samples

• 45 proteins out of the 62 found in the gels carrying total protein samples were detected in the broad pH range 3–10 gel, 11 in the narrow pH range and nine in both types of gels

• 52 proteins only detected in the gels carrying the cytosolic fraction, except for 6 which were found in the broad pH range 3–10 gel, were found in one of the narrow pH range gels only (narrow pH range strips helped to detect 46 proteins not found in the broad range gels)

• Protein distribution was based on the protein identification by mass spectrometry and may not be complete due to:

spot loss during automatic excisionpeptide loss mainly from weak spotsspot overlappingsmall protein size

• About 5000 spots were excised from 13 2-D gels, 5 carrying total and 8 carrying cytosolic proteins. The analysis resulted in the identification of about 3000 proteins, which were the products of 273 different genes

Summary of the 2-D gel electrophoresis data

From Fountoulakis & Suter (2002)

From Fountoulakis & Suter (2002)

Summary of the 2-D gel electrophoresis data

Animals:

Male Wistar rats

(10–12 weeks, bw: 225±8 g)

Treatment:

Bromobenzene

(i.p., 5.0 mmol/kg bw)

dissolved in corn oil (40% v/v)

Duration of treatment:

24 hrs

The bromobenzene dose was hepatotoxic, and this was confirmed by the finding of a nearly complete glutathione depletion at 24 hr after bromobenzene administration. The low level of oxidised (GSSG) relative to reduced glutathione (GSH) indicates that the depletion is primarily due to conjugation and to a much lesser extent due to oxidation of glutathione. The bromobenzene administration resulted in on average 7% decrease in body weight after 24 hr.

From: Heijne et al. (2003)

• Liver samples, total RNA (50 g/array experiment)• cDNA microarrays (3000 genes)• Reference sample:

pooled RNA from liver (~50% w/w), kidneys, lungs, brain, thymus, testes, spleen, heart, and muscle of untreated Wistar rats

• Duplicated microarray/sample• 2-Fold cutoff (p<0.01) relative to the vehicle control:

32 genes were found to be significantly upregulated and 17 were

repressed following bromobenzene treatment • 1.5-Fold cutoff (p<0.01) relative to the vehicle control:

63 genes were found to be significantly upregulated and 35 genes

were repressed following bromobenzene treatment • Functional groups:

Drug metabolismGlutathione metabolismOxidative stressAcute phase responseProtein synthesisProtein degradationOthers

Gene Expression Profiling

From: Heijne et al. (2003)

Glutathione metabolism:

Oxidative stress:

From: Heijne et al. (2003)

From: Heijne et al. (2003)

• 3 two-dimensional gels were prepared from each sample

• A reference protein pattern contained 1124 protein spots

• 24 proteins were differentially expressed (BB or Corn oil)

Protein Expression Profiling

Liver is unique in its capability to regenerate after an injury. Liver regeneration after a 2/3 partial hepatectomy served as a classical model and is adopted frequently to study the mechanism of liver regeneration. In the present study, semi-quantitative analysis of protein expression in mouse liver regeneration following partial hepatectomy was performed using an iTRAQ technique. Proteins from pre-PHx control livers and livers regenerating for 24, 48 and 72 h were extracted and inspected using 4-plex isotope labeling, followed by liquid chromatography fractionation, mass spectrometry and statistical differential analysis. A total of 827 proteins were identified in this study. There were 270 proteins for which quantitative information was available at all the time points in both biologically duplicate experiments. Among the 270 proteins, Car3, Mif, Adh1, Lactb2, Fabp5, Es31, Acaa1b and LOC100044783 were consistently down-regulated, and Mat1a, Dnpep, Pabpc1, Apoa4, Oat, Hpx, Hp and Mt1 were up-regulated by a factor of at least 1.5 from that of the controls at one time point or more. The regulation of each differential protein was also demonstrated by monitoring its time-dependent expression changes during the regenerating process. We believe this is the first report to profile the protein changes in liver regeneration utilizing the iTRAQ proteomic technique.

Dettmer et al., MS Reviews, 26, 51, 2007

Metabolomics is the Most Closely Metabolomics is the Most Closely Related to PhenotypeRelated to Phenotype

Studying the Whole MetabolomeStudying the Whole Metabolome

CH2OP

CHOH

CH2O-

3-phosphoglyceric

acid dehydrogenase

CH2OP

CO

CH2O-

Focused analysis of a single metabolic pathwayFocused analysis of a single metabolic pathway

Unbiased analysis of the entire metabolomeUnbiased analysis of the entire metabolome

Some DefinitionsSome Definitions

Typical Size Range of MetabolitesTypical Size Range of Metabolites

Douglas B. Kell, Curr Opin Microbiol. 7, 296, 2004

NMR

LC/UV

GC/MS

LC/MS

M (10-6)

nM (10-9)

pM (10-12)

fM (10-15)

Range of Tools Required to Cover the Entire MetabolomeRange of Tools Required to Cover the Entire Metabolome

Adapted from Sumner, LW, et al., Phytochem, 62, 817,2003

Main Analytical Approaches to MetabolomicsMain Analytical Approaches to Metabolomics

MS

Chromatography

LC/MS

GC/MS

CE/MS

NMR

LC/NMR

Off-linehyphenation

Analytical Considerations NMR MS

Sensitivity

Reproducibility – w/in lab

Reproducibility – across labs

Quantitation

Sample Prep Requirements

Sample Analysis Automation

Versatility

Selectivity

Non-selectivity

Comparison of NMR vs MS for MetabonomicsComparison of NMR vs MS for Metabonomics

Taken from D.G. Robertson, Toxicological Sciences, 85, 809, 2005

Features of GC/MS MetabolomicsFeatures of GC/MS Metabolomics• Useful for volatiles or compounds that can be derivatized to volatile

compounds (derivatization often required)• Ideal for long chain compounds e.g. FFA, acyl carnitines, etc• More stable and reproducible than LC/MS• Most advanced metabolomics libraries• Standards are typically required for positive identification• Inexpensive technology

Experiment

Library match

• Chromatography can be tailored to specific chemical classes• Various MS analyzers can be coupled e.g. triple quad, TOF, ion trap each with it’s own

advantages in speed, resolution and sensitivity.• Very high mass accuracy available with TOF instruments (< 2ppm)• Variable ionization efficiencies and matrix suppression leads to poor quantitation w/out

standards• Excellent for targetted metabolomics, more challenging for global “unbiased” profiling• Q-TOF can acquire high res data + MS/MS for fragmentation analyses• Libraries are available but suffer from inconsistent retention times in the LC front end.

Features of LC/MS MetabolomicsFeatures of LC/MS Metabolomics

The NMR PhenomenonThe NMR Phenomenon(Hydrogen nuclei act like little magnets)(Hydrogen nuclei act like little magnets)

Hydrogen nuclei out and about Hydrogen nuclei in a magnetic field

RF

pulse

Aligned with the big magnetic field

Precessionbased on magnetic

environment& detection

Excited statetransverse to the field

The NMR ExperimentThe NMR Experiment

detector

The Chemical ShiftThe Chemical Shift

Different hydrogen atoms (gray) are in unique Different hydrogen atoms (gray) are in unique chemicalchemical and and magneticmagnetic environments environments

This results in different precession frequencies and This results in different precession frequencies and distinct spectral features.distinct spectral features.

Features of NMRFeatures of NMR• High structural information content

• Very high intra/inter-lab reproducibility

• Inherently quantitative (no need for authentic standards)

• Minimal sample processing required

• Non-destructive

• Expensive instrumentation• Relatively low sensitivity (typically >M

concentrations required)• Spectral crowding can hinder

interpretation• Long chain aliphatics are challenging

(e.g. fatty acids)

PROS

CONS

Day 1

Day 2

Day 3

Day 4

Day 5

Normal Metabolic Profiles (rat urine)

Adapted from D. Robertson, Pfizer Global Research and Development

Functional NMR Spectrum of Rat Urine

Nature Reviews: Drug Discovery Nicholson et al. (2002)

“Biomarker Windows”

Quantitative Fitting with NMR DatabaseQuantitative Fitting with NMR Database

Nature Reviews: Drug Discovery Nicholson et al. (2002)

Data Analysis in Metabolomics

NMR Spectra Primary Data Processing

Unsupervised mapping of data in 3D space

Supervised classification and calculation of

confidence intervals

Adapted from D. Robertson, Pfizer Global Research and Development

-25

-20

-15

-10

-5

0

5

10

15

20

25

-30 -20 -10 0 10

PC1

PC2

PAP

ANIT

Control

-15

-10

-5

0

5

10

15

-40 -30 -20 -10 0 10

PC1

PC2

ANIT

Control

-naphthylisothiocyanite (ANIT)

p-Aminophenol (PAP)

Control

Adapted from D. Robertson, Pfizer Global Research and Development

Tools to Identify BiomarkersTools to Identify Biomarkers

Set of 1D1H spectra

2D spectra1H & 13C

NMRDatabase

1H & 13CPrediction

KEGG Analysis

MetaboliteID

The Human Metabolome DatabaseThe Human Metabolome Database

http://www.metabolomics.ca/

Mapping to Pathway DatabasesMapping to Pathway Databases

Proteome Res., 5 (7), 1586 -1601, 2006

Systems Toxicology: Integrated Genomic, Proteomic and Metabonomic Analysis of Methapyrilene Induced Hepatotoxicity in the Rat

Andrew Craig, James Sidaway, Elaine Holmes, Terry Orton, David Jackson, Rachel Rowlinson, Janice Nickson, Robert Tonge, Ian Wilson, and Jeremy Nicholson

Abstract:

Administration of high doses of the histamine antagonist methapyrilene to rats causes periportal liver necrosis. The mechanism of toxicity is ill-defined and here we have utilized an integrated systems approach to understanding the toxic mechanisms by combining proteomics, metabonomics by 1H NMR spectroscopy and genomics by microarray gene expression profiling. Male rats were dosed with methapyrilene for 3 days at 150 mg/kg/day, which was sufficient to induce liver necrosis, or a subtoxic dose of 50 mg/kg/day. Urine was collected over 24 h each day, while blood and liver tissues were obtained at 2 h after the final dose. The resulting data further define the changes that occur in signal transduction and metabolic pathways during methapyrilene hepatotoxicity, revealing modification of expression levels of genes and proteins associated with oxidative stress and a change in energy usage that is reflected in both gene/protein expression patterns and metabolites. The difficulties of combining and interpreting multi-omic data are considered.

Vehicle 10 mg/kg, 7 days

100 mg/kg, 7 days100 mg/kg, 7 days

Methapyrilene-induced liver injury in the rat

Hamadeh et al 2002 Tox Path

Proteins altered and identified between control and methapyrilene dosed groups. Proteins are numbered

Ex where elevated and Rx where reduced.

Average standard 1H NMR spectra of liver from each treatment group. This figure shows clearly dose related elevationsand composition changes in fatty acid species…

“Systems Toxicology: Integrated Genomic, Proteomic and Metabonomic Analysis of Methapyrilene Induced Hepatotoxicity in the Rat”

“Systems Toxicology: Integrated Genomic, Proteomic and Metabonomic Analysis of Methapyrilene Induced Hepatotoxicity in the Rat”

“Systems Toxicology: Integrated Genomic, Proteomic and Metabonomic Analysis of Methapyrilene Induced Hepatotoxicity in the Rat”

•Our aim was to determine the impact of drug toxicity on hepatic metabolic pathways and also ascertain whether a multiomic systems biology approach would result in improved understanding of the mechanism of hepatotoxicity of the drug

•The combination of information from gene, protein and metabolite levels provides an integrated picture of the response to methapyrilene-induced hepatotoxicity with mutually supporting and mutually validating evidence arising from each biomolecular level. As expected there were several instances where genes and proteins, either encoded by the same gene or by other genes within the same pathway, were both co regulated by methapyrilene toxicity, and sometimes this was in concert with an associated metabolic product

However:

Strategy of parallel omic data sets: It should be noted that alterations in expression of genes or enzyme levels and modification of protein forms, while suggesting a potential target of toxic effects, do not imply that function or activity must be altered… Alterations to metabolic profiles reflect function and so may serve to aid interpretation of corresponding gene expression and proteomic analyses… Furthermore, as metabolites unlike genes do not suffer the problem of orthology, observed metabolic effects are likely to be highly conserved between species and integrated systems approaches applied to two species may be one framework within which to reconcile and understand the similarities and differences in genetic wiring of common biological processes between different species.

Issue of experimental design: …looking at time points where toxicity is already well developed mitigates against obtaining a clear understanding of the temporal dynamics of the mechanism, especially as changes at the gene, protein and metabolite level may proceed at different rates and on different time scales. As such we might expect highly non linear relationships between the concentrations of various species at the different levels of biomolecular organization…

Issue of molecular resolution: …we detected 100s of gene expression changes compared to the relatively small number of changes detected by the other two technologies. It may thus be likely that insufficient detail was obtained at each biomolecular level to elaborate fully on mechanism of methapyrilene toxicity…

Statistical difficulties: Since each data type usually requires tailored preprocessing (normalization, transformation, scaling, etc.) combining multiple data sets presents a significant analytical challenge. Here, we have performed a separate analysis at the gene, protein, and metabolite level and integrated the knowledge gained from each data set to uncover pathways which responded to the methapyrilene-induced toxicity.