Multi-spectroscopic Metabolic Profiling
Comprehensive Metabolite Identification
by Simultaneous 1
H-NMR and GC-MS
Parag Acharya, Shahid Malik, David Chang, Jack Newton
Chenomx Inc., Edmonton, Alberta, Canada
Nuclear magnetic resonance (NMR) spectroscopy and gas chromatography
coupled mass spectrometry (GC-MS) are two important technology platforms
for metabolite profiling of biological samples. Due to the detection limit of
high-resolution 1
H NMR, certain portions of spectral regions in bio-fluid
samples remain unidentified. Thus, GC-MS has been employed to uncover the
hitherto non-identified metabolites and also to validate the NMR-identified
metabolites in cell extract and serum samples. Finally, the combined NMR and
GC-MS experimental approach has unambiguously provided comprehensive
coverage of metabolites, enabling identification of >90 metabolites in a single
sample.
Abstract
Metabolomics [1-3] has emerged as an important technique for understanding
biological systems. Metabolomics studies typically involve identifying and
quantifying metabolites, which essentially reflect the combined effects of
many influences on physiological functions and phenotype (drug,
environment, nutrition, genetic modification etc.). The basic research and
development in this field has been mainly focused on two areas: (a)
developing analytical technology platforms for metabolite identification and
quantification, as well as (b) software and algorithm development for data
analysis and chemometric analysis. The bio-analytical steps involved in
metabolite identification consists of following steps: Sample selection 
Sample Preparation  Chromatographic separation (mainly for MS
detection)  Detection and Characterization of metabolites.
Results
1. Anal. Chem. 78, 7954 (2006).
2. Trends in Biotech. 22, 245 (2004).
3. Proteomics 6, 4716 (2006).
4. Analyst 130, 606 (2005)
5. Anal. Chem. 78, 1272 (2006).
6. Brain 129, 877 (2006).
References
Experimental Methods
Metabolite Identification by NMR and GC-MS
Metabolite Coverage
Conclusion
Samples:
Human Serum
Human cell extract
¥
Not all detected metabolites indicated, only some representative ones are mentioned.
Comparison of NMR versus GC-MS
Discussions
Acknowledgements
Introduction
Challenges in Metabolite Analysis
The lower limit of detection for NMR means it is not possible to
unambiguously identify metabolite lower than ~2-3µM in bio-fluid samples. It
has been observed that ~5-30% of NMR spectral area (the actual percentage
varies depending on particular bio-fluid and sample quality) remains
unidentified when using deconvolution techniques such as targeted profiling
(because of spectral overlap, ambiguous assignments, etc.). Thus, the key
challenge of the present study is to recover an unequivocal and
comprehensive list of metabolites identified using combined NMR and GC-MS
techniques in cell extract and serum samples.
• Broad range in abundance level of
metabolites (from mM to nMfrom mM to nM)
• Diverse physicochemical nature of
metabolites (widely differing in structure,widely differing in structure,
functional groups, polarity etc.functional groups, polarity etc.)
• Experimental variability (in samplein sample
selection and preparationselection and preparation)
• Impossible for a single technique to cover
identification of all metabolites during
profiling (Figure 1).
Sensitivity
High-
throughput
Identification
NMR
?
LC-MS
GC-MS
DIMS
IR and
Raman
Figure 1. Trade-off between
different analytical techniques [4].
Objectives 1D-NMR GC-MS
Sample preparation Minimal (15 min.)†
Extensive (18.5 hrs)†
Sensitivity mM – µM mM – nM
Resolution Fair to good Very Good‡
Bias in detection Unbiased Relatively biased
Reproducibility Extremely High Fairly good
Quantification
Highly Quantifiable with
one internal standard
Required multiple isotope
labeled internal standards
Experiment Time 10 min†
~ 40 min†
Data Analysis Time ~ 25 min†
~ 90 min†
Throughput Very high Fairly high
†
Values given are as per our laboratory protocol ‡
Can be improved using special columns & mass
analyzer.
Objective
Flowchart for Combined NMR and GC-MS
NMR:
Cell extract samples were diluted with D2O, DSS was added as
internal standard to the final volume of 600µl.
Blood sample was centrifuged and supernatant serum is collected.
After filtration, the aliquot was diluted with D2O and DSS as internal
standard added to the final volume of 600µl.
1D 1
H NMR spectra were obtained using Varian 800MHz
spectrometer at 300K [NOESYPR1D pulse sequence (relaxation delay
- 90º - t1 - 90º - tm - 90º - acquire) used with mixing time tm = 100
ms and delay = 2 sec]
NMR processing and profiling performed with Chenomx NMR suite
v4.62.
GC-MS:
250µl aliquot from NMR sample (for both cell extract and serum),
were completely dried for 5 hrs.
An aliquot of 60µl (for cell extract) and 30µl (for serum) of MAH
(40mg/ml) in pyridine were added to the dried samples and kept at
room temperature for 17 hrs. Then 80µl (for cell extract) and 40µl
(for serum) of MSTFA were added and heated to 40°C for 1.5 hrs.
Derivatized samples were diluted (1:5) with heptane‡
.
Samples were analyzed using Agilent 6890N GC-MS fitted with
HP5-MS GC column and single-Quad mass analyzer.
Injected volume for all experiments was 5µl in split-less mode.
GC parameters‡
: For cell extract - isothermal for 2 min at 70ºC,
followed by 5ºC per min ramp to 310ºC, holding for 5 min; Transfer
line temp.: 250ºC; Helium gas flow 1.2ml/min. For serum -
isothermal for 4.5 min at 70ºC, followed by 10ºC per min ramp to
300ºC, holding for 10 min; Transfer line temp.: 250ºC; Helium gas
flow 0.8ml/min.
MS parameters‡
: Ion source temp.: 250ºC; Recorded mass range
m/z 70 – 600 at 2 scans/sec; Solvent delay for detector: 2 min.
GC-MS analysis (including deconvolution) performed using AMDIS
software with NIST05 library. For library matching: match factor
(MF) threshold set at ≥60 and MF penalty included for retention
index mismatch. All detected peak for relatively low scored matching
(60≤ Match Factor <75) were manually examined (for corresponding
m/z patterns) and confirmed. For any ambiguity, low MF peaks have
been excluded from the list.
Sample
Preparation
NMR
Acquisition
Profiling of
NMR data
Metabolite
Identification
Drying
NM
R
Sam
ple
Derivatization
for GC
GC-MS
Acquisition
Analysis of
GC-MS data
Metabolite
Identification
Comparative
Analysis
Sample
‡
In an effort to derive an optimal GC-MS protocol in our laboratory, the
sample dilution factor and GC / MS parameters were standardized
(details not shown). Moreover, unlike many previous studies [5,6], GC-
MS has been performed on the same sample as of NMR, which helps
control for sample-to-sample variability.
Sample
Total
Peaks
Target
Total Identified
(Match Factor ≥60)
Identified with
Match Factor ≥75
Cell Ext. 189 140 62 43
Serum 180 149 67 52
NMR profiling has identified:
53 metabolites from cell extract and 51 metabolites from serum
• Alberta Ingenuity Fund for providing Alberta Ingenuity R&D Associateship
• National Institute for Nanotechnology (NINT) for providing GC-MS facility
• NANUC for providing high field NMR facility
Serum
GC-MSNMR
432427
GC-MSNMR
431934
Cell Extract
 More than 90 metabolites are identified in a single sample using
combined GC-MS and NMR
 ~20 metabolite overlap between NMR and GC-MS based detection
 GC-MS is useful for the identification of metabolites, but NMR has
a significant advantage in simultaneously identifying and
quantifying metabolites of interest; quantitative GC-MS is
significantly more complicated than quantitative NMR.
S
E
R
U
M
Lactate, Glycolate, Ala, Val,
Leu, Ile, Proline, Gly, Ser, Thr,
Met, Glu, Phe, Lys, Tyr, Urea,
Acetic acid, Citric acid,α-
hydroxy butyric acid,β-
hydroxy Glycerol, butyric
acid, Glucose, Mannose,
Ornithine,
His, Imidazole, TMAO, Taurine,
Acetone, Methanol,Succinate,
Isopropanol, Betaine, Pyruvate,
Creatine, Creatinine,2-hydroxy
isovalerate, 2-oxoglutarate, 2-
amino butyrate, Asn, 1,3-
dimethylurate, Oxalacetate,
Pyruvate, butyrate, Carnitine, τ-
Methylhistidine, 4-Pyridoxate
Exclusive GC-MS¥
Exclusive NMR¥
GC-MS & NMR
C
E
L
L
Ethylene glycol, Malonic
acid, Borate, Erythronic
acid, N-acyl Ser, N-formyl
Gly, Benzoic acid, Urea,
Putrescine, Urea, Citrate,
Stearic acid, Palmitic acid,
Mannitol, Fructose, Oxalic
acid, Uracil, α-hydroxy
butyraldehyde
Val, L-Ser, Leu, L-Proline, Gly,
L-Ala, β-Ala, 4-Hydroxy Proline,
Phe, Tyr, Pyroglutamate,
Glycerol, Succinate, Maleic acid,
Fumaric acid, Myo-Inositol,
Lactic acid, Hypoxanthine,
Imidazole
GTP, ADP, AMP, ATP, NAD+
Acetic acid, Ethanol, Formate
Pyruvate, Guanosine, Taurine,
Choline, O-Phospho Choline,
Ade, Arg, Asp, Asn, N-acetyl
Asp, Betaine, Creatine,
Creatinine, GSH, Glycolate, Hyp,
Ile, Lys, Met, NAA, Thr, Trp,
Taurine, Nicotinate
Ethylene glycol, Erythronic
acid, Stearic acid, Palmitic
acid, Fructose, 5,6-dihydro
Uracil, Pyroglutamate,
Indole-3-acetate, Glyceric
acid, picolinic acid, Myo-
Inositol, Aminomalonic
acid, Glu, δ-amino levulinic
acid
Total peaks = number of peaks, above the S/N threshold limit, detected after deconvolution.
Target = number of peaks matching with compound spectra in NIST library (these include
repetition due to same compound with different derivatization, peaks coming from derivatizing
agent, solvent etc.).
GC-MS has identified:
Irrespective of the sample matrix, most of the natural amino acids are
prevalently identified by both GC-MS and NMR. The long chain fatty acids
like stearic and palmitic acids can exclusively be detected by GC-MS,
whereas metabolites like nucleotides, creatine, creatinine, pyruvate, taurine,
betaine etc. are usually exclusively detected by NMR. Myo-Inositol can be
detected by both NMR and GC-MS in cell extract, whereas only GC-MC can
detect Myo-Inositol in serum. Similarly, Acetic acid is detected by both
techniques in serum, whereas only NMR can detect Acetic acid in cell extract.
Thus, multiple analytical techniques should bring more comprehensive
coverage of metabolites in bio-fluids/cells analysis, which is significant in
terms of bio-marker discovery, where one or many markers metabolites may
remain uncovered by mere profiling by single analytical technique. Our next
step is to compare the accuracy and effort involved in quantitative GC-MS
experiments as compared to NMR.
Though NMR and GC-MS are complimentary in nature, the application of
these two analytical techniques on identical samples has provided interesting
results. As similar sample preparation methodologies (e.g. same extraction
method etc.) were applied prior to NMR and GC-MS acquisition, the variations
in metabolite identification signify different identification capabilities of these
two analytical techniques. Thus, a large portion of the total identified
metabolites by each techniques (for cell extract: 34 out of 53 by NMR and 43
out of 62 by GC-MS; for serum: 27 out of 51 by NMR and 43 out of 67 by GC-
MS) remain exclusive with respect to the particular analytical platform being
used. Only 19 and 24 metabolites from cell extract and serum respectively
have been identified by both techniques, thus cross-validated.
1
H NMR
of serum
Red lines
indicates
profiled
region
1
H NMR
of cell
extract
Blue lines
indicates
profiled
region
TIC of serum
TIC of
cell extract

Metabomeeting2008_rev230408-Jack-parag-final1

  • 1.
    Multi-spectroscopic Metabolic Profiling ComprehensiveMetabolite Identification by Simultaneous 1 H-NMR and GC-MS Parag Acharya, Shahid Malik, David Chang, Jack Newton Chenomx Inc., Edmonton, Alberta, Canada Nuclear magnetic resonance (NMR) spectroscopy and gas chromatography coupled mass spectrometry (GC-MS) are two important technology platforms for metabolite profiling of biological samples. Due to the detection limit of high-resolution 1 H NMR, certain portions of spectral regions in bio-fluid samples remain unidentified. Thus, GC-MS has been employed to uncover the hitherto non-identified metabolites and also to validate the NMR-identified metabolites in cell extract and serum samples. Finally, the combined NMR and GC-MS experimental approach has unambiguously provided comprehensive coverage of metabolites, enabling identification of >90 metabolites in a single sample. Abstract Metabolomics [1-3] has emerged as an important technique for understanding biological systems. Metabolomics studies typically involve identifying and quantifying metabolites, which essentially reflect the combined effects of many influences on physiological functions and phenotype (drug, environment, nutrition, genetic modification etc.). The basic research and development in this field has been mainly focused on two areas: (a) developing analytical technology platforms for metabolite identification and quantification, as well as (b) software and algorithm development for data analysis and chemometric analysis. The bio-analytical steps involved in metabolite identification consists of following steps: Sample selection  Sample Preparation  Chromatographic separation (mainly for MS detection)  Detection and Characterization of metabolites. Results 1. Anal. Chem. 78, 7954 (2006). 2. Trends in Biotech. 22, 245 (2004). 3. Proteomics 6, 4716 (2006). 4. Analyst 130, 606 (2005) 5. Anal. Chem. 78, 1272 (2006). 6. Brain 129, 877 (2006). References Experimental Methods Metabolite Identification by NMR and GC-MS Metabolite Coverage Conclusion Samples: Human Serum Human cell extract ¥ Not all detected metabolites indicated, only some representative ones are mentioned. Comparison of NMR versus GC-MS Discussions Acknowledgements Introduction Challenges in Metabolite Analysis The lower limit of detection for NMR means it is not possible to unambiguously identify metabolite lower than ~2-3µM in bio-fluid samples. It has been observed that ~5-30% of NMR spectral area (the actual percentage varies depending on particular bio-fluid and sample quality) remains unidentified when using deconvolution techniques such as targeted profiling (because of spectral overlap, ambiguous assignments, etc.). Thus, the key challenge of the present study is to recover an unequivocal and comprehensive list of metabolites identified using combined NMR and GC-MS techniques in cell extract and serum samples. • Broad range in abundance level of metabolites (from mM to nMfrom mM to nM) • Diverse physicochemical nature of metabolites (widely differing in structure,widely differing in structure, functional groups, polarity etc.functional groups, polarity etc.) • Experimental variability (in samplein sample selection and preparationselection and preparation) • Impossible for a single technique to cover identification of all metabolites during profiling (Figure 1). Sensitivity High- throughput Identification NMR ? LC-MS GC-MS DIMS IR and Raman Figure 1. Trade-off between different analytical techniques [4]. Objectives 1D-NMR GC-MS Sample preparation Minimal (15 min.)† Extensive (18.5 hrs)† Sensitivity mM – µM mM – nM Resolution Fair to good Very Good‡ Bias in detection Unbiased Relatively biased Reproducibility Extremely High Fairly good Quantification Highly Quantifiable with one internal standard Required multiple isotope labeled internal standards Experiment Time 10 min† ~ 40 min† Data Analysis Time ~ 25 min† ~ 90 min† Throughput Very high Fairly high † Values given are as per our laboratory protocol ‡ Can be improved using special columns & mass analyzer. Objective Flowchart for Combined NMR and GC-MS NMR: Cell extract samples were diluted with D2O, DSS was added as internal standard to the final volume of 600µl. Blood sample was centrifuged and supernatant serum is collected. After filtration, the aliquot was diluted with D2O and DSS as internal standard added to the final volume of 600µl. 1D 1 H NMR spectra were obtained using Varian 800MHz spectrometer at 300K [NOESYPR1D pulse sequence (relaxation delay - 90º - t1 - 90º - tm - 90º - acquire) used with mixing time tm = 100 ms and delay = 2 sec] NMR processing and profiling performed with Chenomx NMR suite v4.62. GC-MS: 250µl aliquot from NMR sample (for both cell extract and serum), were completely dried for 5 hrs. An aliquot of 60µl (for cell extract) and 30µl (for serum) of MAH (40mg/ml) in pyridine were added to the dried samples and kept at room temperature for 17 hrs. Then 80µl (for cell extract) and 40µl (for serum) of MSTFA were added and heated to 40°C for 1.5 hrs. Derivatized samples were diluted (1:5) with heptane‡ . Samples were analyzed using Agilent 6890N GC-MS fitted with HP5-MS GC column and single-Quad mass analyzer. Injected volume for all experiments was 5µl in split-less mode. GC parameters‡ : For cell extract - isothermal for 2 min at 70ºC, followed by 5ºC per min ramp to 310ºC, holding for 5 min; Transfer line temp.: 250ºC; Helium gas flow 1.2ml/min. For serum - isothermal for 4.5 min at 70ºC, followed by 10ºC per min ramp to 300ºC, holding for 10 min; Transfer line temp.: 250ºC; Helium gas flow 0.8ml/min. MS parameters‡ : Ion source temp.: 250ºC; Recorded mass range m/z 70 – 600 at 2 scans/sec; Solvent delay for detector: 2 min. GC-MS analysis (including deconvolution) performed using AMDIS software with NIST05 library. For library matching: match factor (MF) threshold set at ≥60 and MF penalty included for retention index mismatch. All detected peak for relatively low scored matching (60≤ Match Factor <75) were manually examined (for corresponding m/z patterns) and confirmed. For any ambiguity, low MF peaks have been excluded from the list. Sample Preparation NMR Acquisition Profiling of NMR data Metabolite Identification Drying NM R Sam ple Derivatization for GC GC-MS Acquisition Analysis of GC-MS data Metabolite Identification Comparative Analysis Sample ‡ In an effort to derive an optimal GC-MS protocol in our laboratory, the sample dilution factor and GC / MS parameters were standardized (details not shown). Moreover, unlike many previous studies [5,6], GC- MS has been performed on the same sample as of NMR, which helps control for sample-to-sample variability. Sample Total Peaks Target Total Identified (Match Factor ≥60) Identified with Match Factor ≥75 Cell Ext. 189 140 62 43 Serum 180 149 67 52 NMR profiling has identified: 53 metabolites from cell extract and 51 metabolites from serum • Alberta Ingenuity Fund for providing Alberta Ingenuity R&D Associateship • National Institute for Nanotechnology (NINT) for providing GC-MS facility • NANUC for providing high field NMR facility Serum GC-MSNMR 432427 GC-MSNMR 431934 Cell Extract  More than 90 metabolites are identified in a single sample using combined GC-MS and NMR  ~20 metabolite overlap between NMR and GC-MS based detection  GC-MS is useful for the identification of metabolites, but NMR has a significant advantage in simultaneously identifying and quantifying metabolites of interest; quantitative GC-MS is significantly more complicated than quantitative NMR. S E R U M Lactate, Glycolate, Ala, Val, Leu, Ile, Proline, Gly, Ser, Thr, Met, Glu, Phe, Lys, Tyr, Urea, Acetic acid, Citric acid,α- hydroxy butyric acid,β- hydroxy Glycerol, butyric acid, Glucose, Mannose, Ornithine, His, Imidazole, TMAO, Taurine, Acetone, Methanol,Succinate, Isopropanol, Betaine, Pyruvate, Creatine, Creatinine,2-hydroxy isovalerate, 2-oxoglutarate, 2- amino butyrate, Asn, 1,3- dimethylurate, Oxalacetate, Pyruvate, butyrate, Carnitine, τ- Methylhistidine, 4-Pyridoxate Exclusive GC-MS¥ Exclusive NMR¥ GC-MS & NMR C E L L Ethylene glycol, Malonic acid, Borate, Erythronic acid, N-acyl Ser, N-formyl Gly, Benzoic acid, Urea, Putrescine, Urea, Citrate, Stearic acid, Palmitic acid, Mannitol, Fructose, Oxalic acid, Uracil, α-hydroxy butyraldehyde Val, L-Ser, Leu, L-Proline, Gly, L-Ala, β-Ala, 4-Hydroxy Proline, Phe, Tyr, Pyroglutamate, Glycerol, Succinate, Maleic acid, Fumaric acid, Myo-Inositol, Lactic acid, Hypoxanthine, Imidazole GTP, ADP, AMP, ATP, NAD+ Acetic acid, Ethanol, Formate Pyruvate, Guanosine, Taurine, Choline, O-Phospho Choline, Ade, Arg, Asp, Asn, N-acetyl Asp, Betaine, Creatine, Creatinine, GSH, Glycolate, Hyp, Ile, Lys, Met, NAA, Thr, Trp, Taurine, Nicotinate Ethylene glycol, Erythronic acid, Stearic acid, Palmitic acid, Fructose, 5,6-dihydro Uracil, Pyroglutamate, Indole-3-acetate, Glyceric acid, picolinic acid, Myo- Inositol, Aminomalonic acid, Glu, δ-amino levulinic acid Total peaks = number of peaks, above the S/N threshold limit, detected after deconvolution. Target = number of peaks matching with compound spectra in NIST library (these include repetition due to same compound with different derivatization, peaks coming from derivatizing agent, solvent etc.). GC-MS has identified: Irrespective of the sample matrix, most of the natural amino acids are prevalently identified by both GC-MS and NMR. The long chain fatty acids like stearic and palmitic acids can exclusively be detected by GC-MS, whereas metabolites like nucleotides, creatine, creatinine, pyruvate, taurine, betaine etc. are usually exclusively detected by NMR. Myo-Inositol can be detected by both NMR and GC-MS in cell extract, whereas only GC-MC can detect Myo-Inositol in serum. Similarly, Acetic acid is detected by both techniques in serum, whereas only NMR can detect Acetic acid in cell extract. Thus, multiple analytical techniques should bring more comprehensive coverage of metabolites in bio-fluids/cells analysis, which is significant in terms of bio-marker discovery, where one or many markers metabolites may remain uncovered by mere profiling by single analytical technique. Our next step is to compare the accuracy and effort involved in quantitative GC-MS experiments as compared to NMR. Though NMR and GC-MS are complimentary in nature, the application of these two analytical techniques on identical samples has provided interesting results. As similar sample preparation methodologies (e.g. same extraction method etc.) were applied prior to NMR and GC-MS acquisition, the variations in metabolite identification signify different identification capabilities of these two analytical techniques. Thus, a large portion of the total identified metabolites by each techniques (for cell extract: 34 out of 53 by NMR and 43 out of 62 by GC-MS; for serum: 27 out of 51 by NMR and 43 out of 67 by GC- MS) remain exclusive with respect to the particular analytical platform being used. Only 19 and 24 metabolites from cell extract and serum respectively have been identified by both techniques, thus cross-validated. 1 H NMR of serum Red lines indicates profiled region 1 H NMR of cell extract Blue lines indicates profiled region TIC of serum TIC of cell extract