Standardizing

Gas Chromatography-Mass Spectrometry

Metabolomics

Maria I. KlapaMetabolic Engineering and Systems Biology Laboratory

Institute of Chemical Engineering and High Temperature Chemical Processes, Foundation for Research and Technology-Hellas (FORTH),

Patras, GREECE

Free Metabolite Pool ExtractionPolar

Metabolomic Profiling: a multi-step procedure

Data Acquisition

Non - Polar

NMRGC-MSLC-MSCE-MS

GC-MS

Data Analysis

Metabolite Identification

Pool Quantification

Multivariate Statistical Analysis

Schematic Diagram of Metabolomic Analysis

Biological Sample

BiologicalConclusions

DriedMetabolite

Mixture

Peak Area Profile

List of

Marker Ion Peak Areas

Schematic Diagram of Metabolomic Analysis

Biological Sample

BiologicalConclusions

DriedMetabolite

Mixture

Peak Area Profile

List of

Marker Ion Peak Areas

Original Metabolite j

Concentration

Measured Marker Ion i Peak Area

×=)( j

iMIjRF1

Internal Standard Normalizationonly biases that change RF to the same extent for all metabolites (Type A) might be presente.g. variation in the injected volumes, variation in drying, variation in replicate division, &Equipment’s operating conditions remain constantamong runs

Original Metabolite j

Concentration

Measured Marker Ion i Peak Area

×=)( j

iMIjRF1

Internal Standard (IS)

OriginalConcentration

Measured IS Marker Ion k Peak Area

×=)( IS

kMIISRF1 RPA j

ratio between 2 states (Metabolite’s j Concentration) = ratio (RPAj )

Schematic Diagram of Metabolomic Analysis

Biological Sample

BiologicalConclusions

DriedMetabolite

Mixture

Peak Area Profile

List of

Marker Ion Peak Areas

GC-MS

Mixture of Metabolite Derivatives

Derivative

Derivative

From Original Metabolite to Derivative Peak Area

concentration of the original metaboliteconcentration of a derivative of the original metaboliteMeasured peak area of the derivative’s marker ion(s)

Derivative’s I concentration

Measured Marker Ion h Peak Area

×=)( l

hMIlRF1

Internal Standard (IS)

OriginalConcentration

Measured IS Marker Ion k Peak Area

×=)( IS

kMIISRF1 RPAderiv. l of Mj

ratio between 2 states (Mj Concentration) ? ratio(RPAderiv. l of Mj )

Type B Biases

Incomplete derivatizationMultiple Derivatives for some MetabolitesPotential Change in Equipment’s Conditions between Runs

Need for a NEW Data Normalization, Correction and Validation Strategy

not jeopardizing the high-throughput nature of metabolomic analysis

Η. Κanani and M.I. Klapa #. 2007. Data Correction Strategy for Metabolomics Analysis using Gas Chromatography-Mass Spectrometry, Metabolic Engineering

Vol.9:39-51

TMS and MeOX Derivatization

R1C=OR2

+ Methoxyamine HCL

MSTFA

R1C=N-O-CH3

R2R1C=N-O-CH3

R2

syn

anti

R-COOHR-OH

R-NH2

R-COO-Si(CH3)3

R-O-Si(CH3)3

R-NH-Si(CH3)3

Metabolite Category 1kM + MSTFA MD

[ ] [ ] MDMD RPAwMDM *==

MDj

ISl

MD RFRFw =

SILYLATION

silylation time

conc

entr

atio

n

for t > tM

SILYLATIONsilylation time

conc

entr

atio

n

Metabolite Category 2

[ ] [ ] [ ]21 MDMDM +=

[ ][ ]

22

110

2

1

2

1

MDMD

MDMD

RPAwRPAw

kkk

MDMD

**

===

(+MSTFA)

k3

k1

k2

M + MD1

(+MSTFA)Methoxy

amine MD2

MD1

MD2ox

k3

ox

SILYLATIONsilylation time

conc

entr

atio

n

Metabolite Category 2

[ ] [ ] [ ]21 MDMDM +=

[ ][ ]

22

110

2

1

2

1

MDMD

MDMD

RPAwRPAw

kkk

MDMD

**

===Data Validation Criterion!

(+MSTFA)

k3

k1

k2

M + MD1

(+MSTFA)Methoxy

amine MD2

MD1

MD2ox

k3

ox

0

0.2

0.4

0.6

0.8

1

1 101 201 301 401 501 601 701 801

Injection Number

MD

1/ M

D2

Glucose Fructose

Published Metabolomic Analysis based on Metabolomic Data Acquired at Different Equipment conditions

SILYLATIONsilylation time

conc

entr

atio

n

Metabolite Category 3M + MSTFA M(TMS)x

M(TMS)x+1M(TMS)x+n

k

k1(+MSTFA)

[ ] [ ] ∑∑==

==n

iMDMD

n

ii ii

RPAwMDM11

*

Peak Area Variation with derivatization time among replicates of the same sample

15-100%

0.0001

0.0010

0.0100

0.1000

1.0000

5 6 7 8 9 10 11 12 13 14 15 16 17

Derivatization Time

Rel

ativ

e Pe

ak A

rea

Raw Data from Standard Amino Acid Mixture

New Normalization Algorithm

⎥⎥⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢⎢⎢

⎣

⎡

=

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

•

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

][][

.

.

.][][

.

.

.

...............

...

.

.

.

o

o

o

o

M

M

MDt

MDt

MDt

MDt

ISM

ISM

w

w

RPA

RPA

RPA

RPA

N

N

V

N

V

11

1

1

1

# of derivatives

# of

tim

epo i

nts

: relative (with respect to the peak area of the internal standard) peak area corresponding to the i-th derivative of M metabolite at derivatization time tj

1

1

MDtRPA

• Η. Κanani and M.I. Klapa #. 2007. Data Correction Strategy for Metabolomics Analysis using Gas Chromatography-Mass Spectrometry, Metabolic Engineering

Vol.9:39-51

• U.S. Patent Application No. 11/362,717• Best University of Maryland Invention of the Year 2005 in Information Sciences

Peak Area Variation with derivatization time among replicates of the same sample

dropped from 15-100% to 2-8%

0.0001

0.0010

0.0100

0.1000

1.0000

5 6 7 8 9 10 11 12 13 14 15 16 17

Derivatization Time

Rel

ativ

e Pe

ak A

rea

Normalized Data from Standard Amino Acid Mixture

0.0001

0.0010

0.0100

0.1000

1.0000

5 6 7 8 9 10 11 12 13 14 15 16 17

Derivatization TimeEf

fect

ive

Peak

Are

a

Category - 1 and 2 Metabolites

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

0 100 200 300 400 500

Time After Addition of MSTFA (minutes)

log2

(Pea

k A

rea

time

at s

ilyla

tion

time

t /

Peak

are

a at

30

min

of s

ilyla

tion)

citrate TMS sorbitol TMS iso-citrate TMS

ribitol 5 TMS threonate TMS fumarate TMSGlycerol 3TMS fructose meox2 TMS glucose MeOX1 5TMS

Kanani HH, Chrysanthopoulos P, Klapa MI. 2008. Standardizing GC-MS Metabolomics. J. Chromatogr. B Analyt Technol Biomed Life Sci. 871: 191-201

Matrix Effect-Derivatization time 14hr

0.10

1.00

10.00

100.00

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Plant Sample Number

Rat

io (P

A M

D1

/ PA

MD

2)

Fructose Glutamate Threonine Asparagine

Matrix Effects Limit the Accuracy of the Measurements even in the presence of an automated derivatization scheme

Kanani HH, Chrysanthopoulos P, Klapa MI. 2008. Standardizing GC-MS Metabolomics. J. Chromatogr. B Analyt Technol Biomed Life Sci. 871: 191-201

Identification of Unknown Peaks

Derivatives formed from chemical transformations

Derivatives not present inmajor public databases

n/d: Not detected consistentlyin all the samples

n/dLeucine N N O21.000Lecine N On/dLeucine OLeucine11

n/dTryptophan N N O1.0Tryptophan N On/dTryptophan O2 (putative)Tryptophan17

0.26Tyrosine N O O0.94Tyrosine O O1.18Tyrosine O2 (putative)Tyrosine18

7.87Serine NNOO20.299Serine N O O2.97Serine O OSerine15

0.48Phenylalanine N O1.30Phenylalanine OPhenylalanine14

2.124Lysine NNNNO21.005Lysine N N N On/dLysine N N OLysine*12

n/diso-Leucine N N O20.92iso-Leucine N O2.55iso-Leucine Oiso-Leucine*10

0.369Methinonine N N O21.42Methionine N OMethionine*13

1.00Histidine N N On/dHistidine N On/dHistidine O2 (putative)Histidine9

0.37Cysteine N N O12.67Cysteine N S On/dCysteine N O2Cysteine5

1.595Asparagine N N N N O2,3

(putative)1.904Asparagine N N N O0.726Asparagine N N OAsparagine3

n/dOrnithine N N N N O0.48Ornithine N N N O21.10Ornithine N N N OArginine2

0.73Dopamine N N O O4.16Dopamine N O ODopamine23

2.67Homoserine N N O O0.231Homoserine N O O6.51Homoserine O OHomoserine24

n/dOrnithine N N N N O0.48Ornithine N N N O21.10Ornithine N N N OOrnithine25

0.774Alanine N N O1.025Alanine N OAlanine*1

0.224Aspartate N O O3.824Aspartate O O2,3Aspartate*4

Derivative 3Derivative 2Derivative 1Amino acid

33.5Threonine NNOO20.321Threonine N O O3.30Threonine O OThreonine16

0.988Pyroglutamate N O11.014Glutamate N O OGlutamate6

9.000Pyroglutamine NNO 1,2,3

(putative)10.3Glutamine N N N O0.667Glutamine N N OGlutamine7

b–Alanine N N O

Allantoin N N N N N

Valine N N O2,3

MD3

Gaba N N OB- Alanine N O

Allantoin N N N N

Valine N O

Glycine N N O

MD2

Gaba N OB-Alanine O

Allantoin N N N

Valine O

Glycine N O

MD1

1.0n/dGaba22

2.120.53025.3Allantoin20

n/d0.8421.638Valine*19

0.80n/d8.88Beta-Alanine21

0.7739.397Glycine*8

w3w2w1(M)

n/dLeucine N N O21.000Lecine N On/dLeucine OLeucine11

n/dTryptophan N N O1.0Tryptophan N On/dTryptophan O2 (putative)Tryptophan17

0.26Tyrosine N O O0.94Tyrosine O O1.18Tyrosine O2 (putative)Tyrosine18

7.87Serine NNOO20.299Serine N O O2.97Serine O OSerine15

0.48Phenylalanine N O1.30Phenylalanine OPhenylalanine14

2.124Lysine NNNNO21.005Lysine N N N On/dLysine N N OLysine*12

n/diso-Leucine N N O20.92iso-Leucine N O2.55iso-Leucine Oiso-Leucine*10

0.369Methinonine N N O21.42Methionine N OMethionine*13

1.00Histidine N N On/dHistidine N On/dHistidine O2 (putative)Histidine9

0.37Cysteine N N O12.67Cysteine N S On/dCysteine N O2Cysteine5

1.595Asparagine N N N N O2,3

(putative)1.904Asparagine N N N O0.726Asparagine N N OAsparagine3

n/dOrnithine N N N N O0.48Ornithine N N N O21.10Ornithine N N N OArginine2

0.73Dopamine N N O O4.16Dopamine N O ODopamine23

2.67Homoserine N N O O0.231Homoserine N O O6.51Homoserine O OHomoserine24

n/dOrnithine N N N N O0.48Ornithine N N N O21.10Ornithine N N N OOrnithine25

0.774Alanine N N O1.025Alanine N OAlanine*1

0.224Aspartate N O O3.824Aspartate O O2,3Aspartate*4

Derivative 3Derivative 2Derivative 1Amino acid

33.5Threonine NNOO20.321Threonine N O O3.30Threonine O OThreonine16

0.988Pyroglutamate N O11.014Glutamate N O OGlutamate6

9.000Pyroglutamine NNO 1,2,3

(putative)10.3Glutamine N N N O0.667Glutamine N N OGlutamine7

b–Alanine N N O

Allantoin N N N N N

Valine N N O2,3

MD3

Gaba N N OB- Alanine N O

Allantoin N N N N

Valine N O

Glycine N N O

MD2

Gaba N OB-Alanine O

Allantoin N N N

Valine O

Glycine N O

MD1

1.0n/dGaba22

2.120.53025.3Allantoin20

n/d0.8421.638Valine*19

0.80n/d8.88Beta-Alanine21

0.7739.397Glycine*8

w3w2w1(M)

Derivatives treated as unknowns in public

databases

Conclusions

We developed a GC-MS metabolomic data validation, normalization and correction strategy that does NOT jeopardize the high-throughput nature of the analysis

The method is easy to implement and increases the accuracy of measurements by an order of magnitude for some metabolites (NH2 containing compounds)

In light of the importance of metabolomics research, this method is expected to provide a valuable tool for

the acquisition of accurate metabolomic data

Objective

To analyze stress-induced molecular interaction networks

in the context of plant primary metabolism

during the first (30) hours of the stress treatment

under a variety of individual or combined perturbations

using integrated time-series transcriptomic & metabolomicanalyses

Model System: Αrabidopsis thaliana Whole Plant Liquid Cultures Well-controlled growth environment

Experimental Design & Setup

X 4

9h6h 12h 18h 24h1 h

3h

X 2

HarvestingDay 12

LIGHT

Humidity

(80-100 µmole/cm2/s)

60%TEMP (23 °C)

A. thaliana whole-plant liquid culturesDay 0

Exp 3,4,5

Elevated CO2

Exp 2

ControlSet

Combined StressesExp 6,7,8

Gamborg Media + Sucrose (58 mM)Air 0.04% CO2

30h

NaCl (50mM) or Trehalose (12mM) or ACC

Air 1% CO2

NaCl (50mM) or Trehalose (12mM) or ACCAir 1% CO2

Dutta et. al. 2008. Time-series integrated “omic” analyses to elucidate short-term stress-induced responses

in plant liquid cultures. Biotech. Bioeng. (In Press; E-print Available)

3

2

1 4

1% CO2

7

8

AmbientCO2

Sucrose (58 mM)

Glucose (58 mM)

NaCl (50 mM)

910

ACC (0.01 mM)

6

6

Media Pert. effectElevated CO2 Effect

1. Control2. 1% CO23. 50 mM NaCl4. 1% CO2 + 50 mM NaCl Trehalose (12 mM)

cDNA MicroarrayTranscriptomic analysis

(Bhaskar Dutta, UMD)

GC-MS Metabolomic analysis(Harin Kanani, UMD)

10 Exp * 20 samples * 2 Injections * 550 Peaks = 220,000 Total Measurements (8 Exp * 20 samples) Trizol extractions 160 mRNA amplifications 640 cDNA syntheses

320 Dye Injections 320 Micro-array hybridizations (flip-dye)

GCGC--MS MS MetabolomicMetabolomic Data Correction MethodologyData Correction Methodology

Without Data CorrectionPaired-SAM analysis (TIGR MeV v.3.1)

delta = 1.2, FDR (median)= 0%

Cat-3Cat-1 Cat-2 Η. Κanani and M.I. Klapa #. 2007. Data Correction Strategy for Metabolomics Analysis

using Gas Chromatography-Mass Spectrometry, Metabolic Engineering 9:39-51Kanani HH, Chrysanthopoulos P, Klapa MI. 2008. Standardizing GC-MS Metabolomics.

J. Chromatogr. B Analyt Technol Biomed Life Sci. 871: 191-201

Unknown 022

3,4-

Glyoxylate

Glycerol 3 P

4-Hydroxybutanoate

Unknown 083Unknown 024Unknown 116Unknown 040Unknown 044Unknown 048Unknown 074Unknown 345

Unknown 161

Unknown 133Unknown 088Unknown 285Unknown 010

Unknown 059Unknown 097Unknown 089Unknown 039Unknown 136

Unknown 013Unknown 073

Unknown 195

Dihydroxybutyrate

Nicotinate

Glycerate

With Data Correction

Relative Peak Areas

27 + 1 significant

Unknown 022

3,4-Dihydroxybutyrate

NicotinateGlycerol 3 P

Unknown 083

Unknown 024Unknown 116

Unknown 040Unknown 044Unknown 048

Unknown 074

Unknown 345

Unknown 161

Unknown 133

Unknown 088

Unknown 285

Unknown 010

Unknown 059

Unknown 097Unknown 089

Unknown 039

Unknown 136

4-Hydroxybutanoate

Unknown 381

Unknown 371Unknown 391

Unknown 368

Unknown 387

Unknown 376

LysineUnknown 412

Unknown 390

Unknown 415

Uracil

Glyoxylate

26+1+11 significant

Unknown 013Unknown 073

Unknown 195

Relative Peak Areas

PCA- Metabolomic Data: Individual Stress Response

PC1: 48%

PC2: 14%

PC3: 9%

Total: 71%

1% CO2

Control

PC 3

PC 1

PC 2

MiMicroarray TimeTime-series SSignificance Analysis

Consists of 4 modules

Comparison of significant genes’

GO analysis results between time-points

Identification of Significant Genes at each time-point

Correlation Analysis between timepoints with respect to their common

significant genes

US Patent Application, 2006

Dutta B, Snyder R, Klapa MI. 2007. Significance Analysis of Time-Series Transcriptomic Data: A methodology that enables the identification

and further exploration of the differentially expressed genesat each time-point. Biotech. Bioeng. 98: 668-678.

Analysis of Gene Variability in

Significance Level Among Time Points

Elevated CO2 stress : Time-profile of No of Significant Genes

Metabolomic Analysis

20

40

60

80

1 h 3 h 6 h 9 h 12 h 18 h 24 h 30 h PairedSAMfra

ctio

n of

tota

l num

ber (

295)

of

met

abol

ites

(%)

% FDR 2.6 2.1 2.4 1.7 1.9 1.6 4.8 1.8 0

PositivelySignificant NegativelySignificant

Total Significant

% FDR 0 0 0.011 0.014 0 0.007 0 0 0

Metabolomic Analysis

Dutta et. al. 2008. Time-series integrated “omic” analyses to elucidate short-term stress-induced responses in plant liquid cultures. Biotech. Bioeng. (In Press; E-print Available)

0

10

20

30

40

1h 3h 6h 9h 12h 18h 24h 30h PairedSAMfra

ctio

n of

tota

l num

ber (

1123

1)

of g

enes

(%)

Transcriptomic Analysis

Acknowledgements

FundingUS NSF Grant: QSB-0331312, UMD Minta Martin Foundation,UMD Department of Chemical & Biomolecular Engineering,FORTH/ICE-HTBayer HealthCare LLC