High Performance Gas Chromatography Mass Spectrometry in Addressing the Challenges of Metabolomic Studies – Separation in Time and

Mass

Jeffrey S. Patrick Director of Marketed Technology

LECO Separation Science

Outline of Presentation The Metabolomics Problem – Technologies and Challenges

The Zucker Rat Biological Problem and Other

Metabolomic Data

The Outcome of the Study – What is Expressionist?

GC-HRMS – Technology and Data

GCxGC-TOF-HRMS – Rats and Breath

Objectives • GCMS provides capabilities to define modulated analytes in populations

and phenotypes which complements LCMS • HRMS enables identification of unknowns and confident identification of

knowns – Accurate m/z for fragments – Isotopic Abundance for knowns and unknown – Mass accuracy and Isotopic abundance confirm formulae for m/z – Chemical Ionization with accurate m/z enables unknown ID – Provides linearity and sensitivity needed for metabolomics analysis

• Deconvolution enables the ability to: – detect and quantify metabolites – provide searchable spectra from difficult peak pairs – provide interpretable spectra from difficult peak pairs

• GCxGC TOF MS – – Separation of additional analytes – Differential Analysis and enhanced Sensitivity

• Genedata enables an HRMS-optimized tool for differential analysis of phenotypes and populations.

The Metabolomic Problem

Analytical Challenges in Metabolomics Accurate

Differential Analysis of

Biochemistry

Complex Samples and Analytes •Diverse samples (plants, phys

fluids, insects, tissue, etc) •Matrix Effects

•Spectral Dynamic Range •Isomer Differentiation

•Reproducibility

Range of Analyte Concentrations

Peak Capacity (1000s of

Analytes/hour)

Comprehensive Analysis (vs.

Targeted) > 60% Unknowns!

Data Interpretation Systems Biology and Contextual

Information

© R. Goodacre 13th Annual Fera/JIFSAN 2012

Can’t cover needed analytes with GC or LC Alone

Why GCMS? Fast and easy to

adopt

Sensitive, Reliable, Robust and Quantitative Linear Response

and Good Dynamic Range

Universal Application for M <

600 (w/ Deriv)

Over 50 yrs of Application and Largest Presence of ANY MS

System

HUGE Well-established

Database (>250k spectra)

Highest Peak Capacity

Chromatography

Variety of MS Opportunities

(Nominal, HRMS, MSMS)

System Under Study • Zucker Rats

– 3 Phenotypes/Strains w/ Animals bred to be • Lean (n=12), Fatty (n=12), Obese

(n=12) – 7-9 Weeks old (terminal bleed)

• Disodium EDTA as anti-coagulant • 0.1 µm filtered

• Objectives – Identify analytes which are up or down

regulated with phenotype using high performance MS

– Test the capabilities of HRTs

“Husky??”

“Fit”

Zucker Rat Study: Sample Preparation

1)Plasma (100 µL) 3) Vortex

4) Centrifuge & RemoveProtein Pellet

2) MeOH(400 µL)

5) Dry (2 hrs)6) Lyophilize(Overnight)

7) MeONH2

8) MSTFA

Pegasus GC-HRT9) FAMEs

Why High Resolution and Accurate Mass?

What are these values?

63%

48%

56%

51%

71%

68%

Percentage of Unidentified Analytes in a Metabolomic Study

Why?? Corrupt or Impure Spectra Uncertainty in m/z EI only with no M Limited MS/MS or accurate mass databases Limited Libraries to match Derivatives or Analytes Inadequate Chemical Analysis Tools

What is the real value of Accurate Mass?

Rachel L. Sleighter and Patrick G. Hatcher (2011). Fourier Transform Mass Spectrometry for the Molecular Level Characterization of Natural Organic Matter: Instrument Capabilities, Applications, and Limitations, Fourier Transforms - Approach to Scientific Principles, Prof. Goran Nikolic (Ed.), ISBN: 978-953-307-231-9, InTech, DOI: 10.5772/15959. Available from: http://www.intechopen.com/books/fourier-transforms-approach-to-scientific-principles/fourier-transform-mass-spectrometry-for-the-molecular-level-characterization-of-natural-organic-matt

Zucker Rat Study: Workflow

Pegasus GC-HRT® (EI and CI-MS)

Separation &

Detection

ChromaTOF HRT Untargeted

and/or Targeted Peak Find

GeneData (Differential Analysis)

Alignment, Normalization & Statistics

Metabolite Identification

Sample Preparation

Zucker Rat Study: Instrument Parameters GC Agilent 7890 with 7693 Auto Sampler Column Restek Rxi-5Sil MS (30m x 0.25mm x 0.25mm) & 5m

Guard Carrier Gas, Flow He, 1.0 mL/min Constant Flow Injection/Volume Splitless, 1 µL (CI 2 µL) Temp. Program 70 °C (4 min) to 300 °C at 20 °C/min (6 min)

MS LECO Pegasus® GC-HRT Transfer Line Temp.

300 °C

Ion Source Temp. 250 °C (CI 200 °C) Ionization EI (70 eV); CI (140 eV) Mass Range 60 – 520 (CI 100 – 1000, Reagent Gas = 5% NH3 in CH4) Acquisition Rate 6 sps Mass Calibration PFTBA (Internal)

General Findings

500 600 700 800 900 1000

0.0e01.0e62.0e63.0e64.0e65.0e66.0e6

0.0e01.0e62.0e63.0e64.0e65.0e66.0e6

0.0e01.0e62.0e63.0e64.0e65.0e66.0e6

Time (s)

AIC Fatty 6

AIC Lean 6

AIC Obese 6

• Total Average Features Found (S/N > 100) – 662 (+/- 57) • Analytes having ID Match > 800 – 274 (+/- 34) • Analytes at > 600 and M , 2ppm – 266 (+/- 26) (N = 36)

ChromaTOF HRT: Deconvolution

500 600 700 800 900 1000 1100 12000.0e0

1.0e7

2.0e7

3.0e7

4.0e7

5.0e7

Time (s)

AIC

636 638 640 642 644 646 648

XIC(176.0921±0.0005) XIC(156.0836±0.0005) XIC(74.0360±0.0005)

A

B

Methionine (2TMS)

5-Oxo-Proline (2TMS)

Methyl Decanoate

Obese

100 150 200 250 300

0.0e01.0e52.0e53.0e54.0e55.0e5

02004006008001000

M/Z

73.04673 156.08361

258.

0973

8

230.

1024

7

156

73

147

75 230

258

Peak True - sample "Obese 9", L-Proline, 5-oxo-1-(trimethylsilyl)-, trimethylsilyl ester (CAS), at 639.86 s , Height (Counts)

Library Hit - Similarity: 835 - Library: Wiley9 - L-Proline, 5-oxo-1-(trimethylsilyl)-, trimethylsilyl ester (CAS), Abundance

EI-HRT: Accuracy & Spectral Similarity N

Si

O

O

O

Si-0.77 ppm

835/1000

80 100 120 140 160 180 200 220

0.0e01.0e52.0e53.0e54.0e55.0e5

02004006008001000

M/Z

74.03605 87.04401

69.0

6988

143.

1066

6

171.

1378

8

129.

0909

7

101.

0597

1

183.

1742

5

115.

0752

9

71.0

8552

84.0

5698

67.0

5429

140.

0316

6

185.

1535

5

74

87

143

69 129

171

101

83 97 183

115

71 157

185

67 85

Peak True - sample "Obese 9", Dodecanoic acid, methyl ester (CAS), at 641.033 s , Height (Counts)

Library Hit - Similarity: 750 - Library: Wiley9 - Dodecanoic acid, methyl ester (CAS), Abundance

-1.0 ppm

OO

750/1000

215

214.19251

215.

1991

8

218

0663

2

216.

2035

3

ChromaTOF HRT: Representative Compounds (0.9 ppm mass error)

Name Formula R.T. (s) Area LM (1000) Ion Observed Ion m/z Mass Accuracy (ppm)Alanine (3TMS) C9H23NO2Si2 448.6 127332975 829 [M-CH3]+ 218.10235 -1.62

Oxalic Acid (2TMS) C8H18O4Si2 464.5 27487671 889 [M-CH3]+ 219.05012 -1.00

Valine (2TMS) C11H27NO2Si2 508.2 50953068 820 [M-C3H7]+ 218.10250 -0.95

Serine (3TMS) C12H31NO3Si3 572.5 16161374 811 [M-CH3]+ 306.13720 0.16

Threonine (3TMS) C13H33NO3Si3 583.6 17103360 807 [M-CH3]+ 320.15264 -0.51

L-Proline, 5-oxo- (2TMS) C11H23NO3Si2 639.9 47799304 835 M+● 273.12089 -0.77

[M-CH3]+ 258.09738 -0.94

Citric Acid (4TMS) C18H40O7Si4 739.9 39155972 861 [M-CH3]+ 465.16066 -0.92

Galactose (MEOX, 5TMS) C22H55NO6Si5 795.467 38182656 792 [M-C10H27O2Si3]+ 364.1788657 -0.4088

Glucose (MEOX, 5TMS) C22H55NO6Si5 801.498 12382092 821 [M-C11H31O3Si3]+ 332.1529305 0.3934

Inositol (6TMS) C24H60O6Si5 824.445 11117824 850 [M-C6H20O2Si2]+ 432.1992864 -1.2284

Octadecanoic acid (TMS) C21H44O2Si 865.1 18735936 895 M+● 356.30982 -1.94

Arachidonic acid (TMS) C23H40O2Si 899.0 16205056 889 M+● 376.27853 -1.80

Cholestadiene C27H44 1061.05 384812 782 M+● 368.3441255 1.0118

Cholesterol TMS C30H54OSi 1164.7 24326300 736 M+● 458.39349 -0.76

Campesterol, TMS C31H56OSi 1224.6 1668712 888 M+● 472.40950 0.01Ave = 0.90 ppm

Genedata Expressionist MSX

Data Processing

MS Data Statistical Analysis

Integrated

Integrated MS data processing and analysis

Software truly optimized to handle GC-HRMS data

GeneData Results

Obese

Lean

Fatty Fold Change > 1.5

P values < 0.05

Pathways Implicated in GCMS

Statistics and Pathways are important but individual metabolites are important as well

BCAAs which Feed fatty acids and lipids

Serotonin – Obesity, Sleep and Appetite

0

50000

100000

150000

200000

250000

Serotonin

Fatty Lean Obese Total

5-Hydroxytryptamine

Implicated Pathways and Physiology Tryptophan Biosynthesis Neurotransmission Appetite control Depression Obesity

Branched Amino Acids

Figures of merit • Average m/z =

86.096368 • Target m/z

86.096430 • Avg. Mass Error =

0.72 ppm (Abs) (N = 36 over 2 days)

0

500000

1000000

1500000

2000000

2500000

Isoleucine

Fatty Lean Obese Total

Leucine and Valine show similar trends

Branched Chain AA, Fatty Acid and other Metabolism

Modulated Fatty Acid

0

500000

1000000

1500000

2000000

2500000

3000000

3500000

4000000

4500000

Hexadecanoic Acid

Figures of merit • Average m/z =

328.278977 • Target m/z

328.27889 • Avg. Mass Error =

0.97 ppm (Abs) (N = 36 over 2 days) Fatty Lean Obese Total

Other Modulated Metabolites

0

200000

400000

600000

800000

1000000

1200000

1400000

Myoinositol (6TMS)

m/z 217.1073

m/z 318.1483

Fatty Lean Obese Total

Monitored by each of 2 accurate m/z values one is more consistent/selective than the other Linked to selected lipid metabolism

Metabolite ID in Tobacco Leaf

Cholesterol in Green Tobacco Leaf

1100 1200 1300 1400 1500 1600 1700 18000.0e0

2.0e6

4.0e6

6.0e6

8.0e6

1.0e7

1.2e7

1.4e7

1.6e7

Time (s)TIC

1645 1647.5 1650 1652.5 1655 1657.5 1660 1662.5 16650.0e0

5.0e4

1.0e5

1.5e5

2.0e5

2.5e5

3.0e5

3.5e5

4.0e5

4.5e5

Time (s)XIC(145.101145±3ppm) XIC(165.090939±3ppm)

1645 1647.5 1650 1652.5 1655 1657.5 1660 1662.5 1665

2.0e6

3.0e6

4.0e6

5.0e6

6.0e6

7.0e6

Time (s)TIC

Tocopherol

Cholesterol

1650 1660

dl-à

-Toc

ophe

rol

Chol

este

rol (

CAS)

XIC(165.0909)

Green Leaf : Co-eluting Compounds

50 100 150 200 250 300 350 400 450 5000

200

400

600

800

1000

Area

(Ab

unda

nce)

M/Z

145.1011581.06991 105.06991

95.0

8561

43.0

5449

55.0

5439

159.

1167

9

121.

1011

1

213.

1636

3

133.

1011

1

368.

3433

3

67.0

5431

255.

2105

0

247.

2417

2

353.

3199

7

163.

1480

6

71.0

8556

129.

0698

1

173.

1324

3

301.

2886

7

39.0

2315

199.

1480

3

275.

2729

2Peak True - sample"GL 9aT Splitless", Cholesterol (CAS), at 1652.88 s

50 100 150 200 250 300 350 400 450 5000.0e0

2.0e2

4.0e2

6.0e2

8.0e2

1.0e3

1.2e3

Area

(Ab

unda

nce)

M/Z

43 81

95 145

368

105

9155

247

159

121

133

67 213

255

18 353

386

275

173

36

Library Hit - Library: Wiley9 - Cholesterol (CAS)

LM = 880/1000

Peak True

*

HO

50 100 150 200 250 300 350 400 450 5000.0e0

2.0e2

4.0e2

6.0e2

8.0e2

1.0e3

1.2e3

Area

(Ab

unda

nce)

M/Z

165.09094

430.

3803

3

205.

1222

6

Peak True - sample"GL 9aT Splitless", dl-à-Tocopherol, at 1650.87 s

50 100 150 200 250 300 350 400 450 5000.0e0

2.0e2

4.0e2

6.0e2

8.0e2

1.0e3

1.2e3

Area

(Ab

unda

nce)

M/Z

165

430

43 205

57

Library Hit - Library: mainlib - dl-à-Tocopherol LM = 833/100

Peak True

OHO

50 100 150 200 250 300 350 400 450 5000

100

200

300

400

500

600

700

800

900

1000

M/Z

145.10115

81.0

6991

105.

0699

0

91.0

5432

95.0

8561

43.0

5449

55.0

5439

27.9

9492

159.

1167

9

121.

1011

0

213.

1636

3

133.

1011

1

368.

3433

3

67.0

5431

255.

2105

024

7.24

171

353.

3199

7

173.

1324

3

301.

2886

7

199.

1480

0

275.

2729

1

Area (Abundance), Peak True - sample"GL 9aT Splitless", Cholesterol, at 1652.88 s

Interpretative Power of Accurate Mass – Cholesterol (in Tobacco Leaf Extract NOT TMS)

CH3

CH3

CH3

CH3+ CH31.1 ppm

CH2+

CH3CH3CH3

CH3

CH3

0.9 ppm

Match = 852

M●+ 0.5 ppm

Structures and fragmentation with ACD/MS Workbook Suite

GCxGC TOF MS

Sample Preparation

Centrifuge

SpeedVac BSTFA

GCxGC TOF - MS

Pool by Condition

Control Ethanol

Freeze; Pulverize; Extract with MeOH

DISEASED ETHANOLIC

1942.5 1945 1947.5 1950 1952.5 1955 1957.5 1960 Time (s)

Oct

adec

anoi

c ac

id, t

rim

ethy

lsily

l est

er

132 POOLED ETOH 132 POOLED CTRL REF

Oct

adec

anoi

c ac

id, t

rim

ethy

lsily

l est

er

100 200 300 400

100 200 300 400

500

1000 129

341 201 257

Peak True - sample "POOLED ETOH ", peak 866, at 1951.76 s

100 200 300 400

500

1000 117 73 341

201 297

Library Hit - similarity 811, "Octadecanoic acid, trimethylsilyl ester"

100 200 300 400

100 200 300 400

500

1000 117 73

341 201

Peak True - sample "POOLED CTRL REF", peak 989, at 1951.74 s

100 200 300 400

500

1000 73 117

341 201

Library Hit - similarity 869, "Octadecanoic acid, trimethylsilyl ester (CAS)"

Octadecanoic Acid in Mouse Liver extracts – arrow indicates up regulation

NORMAL CONTROL

Similarity > 800

Match Present in reference and sample within user-specified concentration tolerance

Out of Tolerance

Present in reference and sample, but not within user-specified concentration tolerance

Not found

Present in reference, but not in sample

Unknown Present in sample, but not in reference

Normal Control Pooled (Reference)

Disease Ethanolic Pooled (Sample)

How can we identify differences between similar samples?

ChromaTOF’s Reference Feature computes the relative concentration of peaks in a sample with respect to a user specified reference. Type tags are assigned to each analyte:

Representative Examples

Match: Proline, di-TMS Out of Tolerance: Linoleic Acid, TMS

Not Found: Ala-Gly, N-TMS-, TMS ester Unknown: Tryptophan, tri-TMS

Type Concentration Name Similarity R.T. (s) Quant S/N Area Height Quant Masses Match 105.68 Proline, di-TMS 888 1044 , 1.025 5419.4 665035 28371 142 Out of Tolerance 199 Linoleic Acid, TMS 901 2232 , 1.175 12237 1858318 141038 67

Not Found Ala-Gly, N-TMS-, TMS ester 850 1548 , 1.325 116

Unknown Tryptophan, tri-TMS 844 2229 , 1.360 664.14 48291 3073.3 202

Control Disease Control Disease

Control Disease Control Disease

Library ID: Undecane

Breath Analysis

Breath tests using 1D GC MS

Demonstrated proof of principle in: • Lung cancer • Breast cancer • Heart transplant rejection • Pulmonary tuberculosis • Environmental toxicology • Influenza

All of these studies were performed with 1D GC MS…

…and then 2D GC MS came along

Human breath VOCs Two-dimensional gas chromatography and time-of-flight mass

spectrometry

Breath

Breath minus air

The limitations of 1D GC MS Poor selectivity ~ 200 VOCs in breath Co-elution +++ Biomarker ID not consistent

The advantages of GCxGC TOF MS Excellent selectivity ~ 2,000 VOCs in breath Co-elution – minimal Biomarker ID highly consistent

Conclusions • Metabolomics Studies using GCMS provide capabilities to define modulated

analytes in populations and phenotypes which complements LCMS • HRMS provides an ability to identify unknowns and to have confident

identification of knowns – Accurate m/z for fragments – Isotopic Abundance for knowns and unknown – Mass accuracy and isotopic Abundance confirm formulae for m/z – CI enables molecular ion m/z – Provides linearity and sensitivity needed for metabolomics analysis

• Deconvolution enables the ability to: – detect and quantify metabolites – provide searchable spectra from difficult peak pairs – provide interpretable spectra from difficult peak pairs

• GCxGC TOF MS – – Separation of additional analytes – Differential Analysis and enhanced Sensitivity

• Genedata enables an HRMS-optimized tool for differential analysis of phenotypes and populations.

People Doing the Work

For More Information

Contact LECO at:

World Headquarters/United States In United States: 800-292-6141 or 269-985-5496

Outside U.S.A.: 269-983-5531

Email: info@leco.com www.leco.com