brochure: posterhall metabolomics, 09-2011 - bruker
TRANSCRIPT
Metabolomics
Poster Hall 2011
Innovation with IntegrityMass Spectrometry, NMR
Bruker Daltonik GmbH
Bremen · GermanyPhone +49 (0)421-2205-0 Fax +49 (0)421-2205-103 [email protected]
Bruker Daltonics Inc.
Billerica, MA · USAPhone +1 (978) 663-3660 Fax +1 (978) 667-5993 [email protected]
www.bruker.com/metabolomics
For research use only. Not for use in diagnostic procedures.
Bru
ker
Dal
toni
cs is
con
tinua
lly im
prov
ing
its p
rodu
cts
and
rese
rves
the
rig
ht
to c
hang
e sp
ecifi
catio
ns w
ithou
t no
tice.
© B
DA
L 01
-201
1, #
274
059
Table of content
The Need for Speed in Metabolomics:UHPLC with maXis UHR-Q-TOF-MS Analysis of Tea Extracts 4-5
Plant MetabolomicsDevelopment of an HPLC-TOF-MS screening-platform to assess ppHypSys-dependent metabolic changes in Nicotiana attenuata during insect herbivory 6
Challenges in the investigation of the metabolic changes in Nicotiana attenuata during insect herbivory using an improved HPLC-TOF-MS method 7
Bacterial MetabolomicsAnalysis of the Intraspecies Variation of Secondary Metabolite Production by Myxococcus xanthus using ESI-TOF-MS and PCA 8
Mining the Myxobacterial Secondary Metabolome using High-resolution Mass Spectrometry: Natural Product Discovery as an Analytical Challenge 9
Challenges in Metabolomics addressed by targeted and untargeted UHR-Q-TOF analysis 10
NutritionCharacterization of black and green tea using ESI-Q-TOF-MS anddata evaluation by principle component analysis 11
Analysis of Whisky by Electrospray FT-ICR Mass Spectrometry: Proof of Origin by Statistical Methods 12
Metabolic Profiling and Fingerprinting of Beverages using LC-MS and NMR 13
Clinical MetabolomicsIdentification of Nucleosides as Potential Biomarkers for Breast Cancer in Urine by ESI-TOF-MS 14
Combined analysis of NMR and LC/MS data: observations in urine samples from a lifestyle study 15
Searching biomarkers for the Complex Regional Pain Syndrome by metabolic profiling of urine using CE/MS 16
Metabolomics Technology: FT-MS, LC-MS-NMR, CE-MS, GC-APCI-TOF-MSElemental analysis of fulvic acids of Shilajit by ultra-high resolution FTMS 17
Characterization of dissolved organic matter in marine porewaters by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry 18
Hybrid LC/MS/NMR as a Unique Tool for Analysis of Highly Complex Samples and Integrated Chemical Identification 19
Integration and Robustness: a Touchstone in Metabolomics Studies 20
GC/APCI with ultra high resolution TOF-MS - analytical validation and applicability to metabolic profiling 21
Characterization of polyphenols of products derived from bees using CE-ESI-TOF 22
Metabolomics Data evaluation, Identification of unknownsFully Unsupervised Automatic Assignment and Annotation of Sum Formulae for Product Ion Peaks, Neutral Losses in MS and Product Ion Spectra 23
Sum Formula Calculation and Identification of a Bacterial Metabolite with m/z > 1100 24
Rapid and automated identification of phenolic compounds in food using LC/ESI-Qq-TOF-MS/MS and library search 25
Deconvolution and Automatic Formula Assignment of Alternating MS and Broad-Band CID Analyses 26
Small Molecule ImagingSmall Molecule Imaging Mass Spectrometry using solariX-MI 27
Powerful Solutions Tailored for Metabolomics Applications
Metabolomics has emerged as an exciting new area of biological research and gained substantial attention in recent years. Previously, significant efforts have been made to probe and understand the genomes, transcriptomes and proteomes of a variety of organisms. Despite massive investments to sequence and identify cellular DNA, RNA and proteins, our understanding of many key biological functions remains rudimentary and incomplete. Many researchers believe that perhaps more complete answers can be gleaned from studying small molecules that are utilized for a variety of metabolic and cellular functions. Because of this belief, in many laboratories, Metabolomics is catching up with traditional Proteomics and Transcriptomics efforts and has been christened “Biochemistry’s new look” [1].
Typically, during a Metabolomics study, a large number of cell or patient samples are analyzed and compared to identify potential biomarker compounds which correlate to a disease, drug toxicity, or genetic or environmental variation. Validated biomarkers can form the basis for new diagnostic assays, and lead the way to help establish truly personalized medicine, since changes in small molecule concentrations are closely related to the observed phenotype.
Answering the Challenges of Metabolomics
Metabolomics researchers are regularly confronted with the daunting tasks of acquiring large sets of data and then analyzing them in minute detail. Due to the chemical diversity of small molecule metabolites, it is impossible to study the entire metabolome using a single analytical technique or technology. Comprehensive analysis of the metabolome requires multiple solutions.
The Metabolome’s chemical space can be fully explored by utilizing Bruker´s high performance LC-MS, GC-MS, CE-MS and NMR systems in conjunction with dedicated software for data evaluation.
One of the most challenging tasks in the Metabolomics studies is the identification of unknown compounds. Exact mass and isotopic pattern information in MS and MS-MS spectra
-2- -3-
acquired by the solariX (FT-MS) or micrOTOF series and maXis instruments can be used to definitively identify unknown compounds. The intrinsic capabilities of these instruments in conjunction with the SmartFormula3D software enables accurate and reliable molecular formula generation fora wide variety of unknowns.
In order to enhance separation and increase throughput, many laboratories have turned to utilization of UHPLC for Metabolomics studies. Since the mass accuracy, isotopic fidelity and resolution of maXis and micrOTOF-series ESI-(Q)-TOF instruments are independent of the acquisition rate, they are perfectly suited for a coupling to UHPLC separations.
By utilizing a Bruker solution, their high-resolution and accurate mass LC-MS capabilities enable both targeted and untargeted Metabolomics workflows. In addition, the broad dynamic range characteristics of these systems allows both, high- and low-abundance compounds to be studied within the same dataset – even within the same mass spectrum.
Finally, Bruker offers the capability to integrate both LC-MS and NMR techniques into a single analytical system for maximum metabolite coverage and molecular identification. The MetabolicProfilerTM is a powerful combination of techniques and is exclusively designed to address all analytical needs for metabolic profiling and structure elucidation.
This poster hall is a “snapshot” of some of the current work utilizing a variety of Bruker solutions for Metabolomics studies. We would like to thank all of our partners and customers for their constant input and feedback. This information has served as an invaluable guide for us in our solution development efforts. We would like to extend our special thanks to everyone who has been involved in the generation of the outstanding posters compiled in this poster hall for your benefit.
Sincerely yours,
Aiko BarschMarket Manager Metabolomics
References / further reading
Development and validation of a liquid chromatography-electrospray ionization-time-of-flight mass spectrometry method for induced changes in Nicotiana attenuata leaves during simulated herbivory.E. Gaquerel, S. Heiling, M. Schoettner, G. Zurek, I.T. BaldwinJ Agric Food Chem. 2010 Sep 8;58(17):9418-27.
Discovering the Hidden Secondary Metabolome of Myxococcus xanthus: a Study of Intraspecific Diversity.D. Krug, G. Zurek, O. Revermann, M. Vos, G.J. Velicer, and R. MüllerAppl Environ Microbiol. 2008 May; 74(10): 3058–3068.
Efficient mining of myxobacterial metabolite profiles enabled by liquid chromatography-electrospray ionisation-time-of-flight mass spectrometry and compound-based principal component analysis.D. Krug, G. Zurek, B. Schneider, R. Garcia, R. MüllerAnal Chim Acta. 2008 Aug 22;624(1):97-106.
Gas chromatography/atmospheric pressure chemical ionization-time of flight mass spectrometry: analytical validation and applicability to metabolic profiling.A. Carrasco-Pancorbo , E. Nevedomskaya , T. Arthen-Engeland, T. Zey, G. Zurek, C. Baessmann, A.M. Deelder, O. MayborodaAnal Chem. 2009 Dec 15; 81(24):10071-9.
Evaluation of GC-APCI/MS and GC-FID As a Complementary Platform.T. Pacchiarotta, E. Nevedomskaya, A. Carrasco-Pancorbo, A.M. Deelder, and O.A. MayborodaJ Biomol Tech. 2010 December; 21(4): 205–213.
Combined Reversed Phase HPLC, Mass Spectrometry, and NMR Spectroscopy for a Fast Separation and Efficient Identification of Phosphatidylcholines.J. Willmann, H. Thiele, and D. LeibfritzJ Biomed Biotechnol. 2011; 2011: 385786.
Exploratory analysis of human urine by LC-ESI-TOF MS after high intake of olive oil: understanding the metabolism of polyphenols. R. García-Villalba, A. Carrasco-Pancorbo, E. Nevedomskaya, O. Mayboroda, A. Deelder, A. Segura-Carretero, and A. Fernández-Gutiérrez Analytical and bioanalytical chemistry. 2010 Sep 398(1):463-75
Amino acid profiling in urine by capillary zone electrophoresis - mass spectrometry. O.A. Mayboroda, C. Neusüss, M. Pelzing, G. Zurek, R. Derks, I. Meulenbelt, M. Kloppenburg, E. Slagboom, and A. DeelderJournal of chromatography A. 2007 Aug 1159(1-2):149-53
A strategy for the determination of the elemental composition by fourier transform ion cyclotron resonance mass spectrometry based on isotopic peak ratios.D. Miura, Y. Tsuji, K. Takahashi, H. Wariishi, K. SaitoAnal Chem. 2010 Jul 1; 82(13): 5887-91
Automated identification of phenolics in plant-derived foods by using library search approach. M. Gómez-Romero, G. Zurek, B. Schneider, C. Baessmann, A. Segura-Carretero, and A. Fernández-GutiérrezFood Chemistry. 2011 Jan 124(1): 379-386
[1] Blow N., Nature 455, 697-700 (2008).
The Need for Speed in Metabolomics:UHPLC with maXis MS Analysis of Tea Extracts
Untargeted metabolic profiling of different black and green tea extracts was performed using a UHPLC and maXis UHR-TOF analytical system. Processed data was submitted to principal component analysis (PCA) to identify compounds differentiating the tea varieties. The compounds’ sum formulae were queried in public databases, which enabled the identification of compounds characteristic for green and black tea.
Introduction
Fast separation of complex samples using high-resolution chromatography and state-of-the-art mass spectrometry are required to meet the simultaneous need for high sample throughput and high-quality data in metabolomics.Hyphenation of UHPLC and the maXis UHR-TOF (Ultra High Resolution Time-Of-Flight) mass spectrometer delivers speed without compromising on any performance factor, such as sensitivity, mass accuracy or resolution.
Black and green tea account for more than 95% of tea consumed worldwide. Health benefits have been hypothesized for both types of tea. A deeper understanding of potential health promoting effects – as well as an improvement in quality and taste – is a topic of interest in academia and the food industry.
Data generated from different tea extracts via UHR-TOF-MS was subjected to statistical analysis. Highly accurate mass data and isotopic pattern information in MS and MS-MS spectra enable reliable sum formula generation. Combining these formulae with database queries facilitates identification of unknown compounds – often described as the major bottleneck in metabolomics.
Experimental
In brief, 14 different tea extracts (6 black, 6 green and 2 black Darjeeling) were prepared by steeping tea samples in 100 mL hot water for 5 min. Analysis was performed with an UltiMate 3000 Rapid Separation system (Dionex) and maXis UHR-TOF (Bruker Daltonics) operating in ESI negative mode (acquisition rate 4 Hz). Data pre-processing and feature extraction were performed with CompassDataAnalysis; Principal Component Analysis (PCA) and sum formula generation were performed using SmartFormula in ProfileAnalysis (Bruker Daltonics). Complete experimental conditions are given in Reference [1].
Results and Discussion
LC–MS data obtained from the different tea extracts were processed using the “Find Molecular Features” (FMF) peak detection algorithm, which efficiently differentiates real signal from background noise. FMF compounds were used as input for PCA analysis.
The PCA scores plot (Figure 1; Explained Variance PC1 vs PC2: 71.0%) revealed a grouping of samples reflecting the oxidation grade of tea leaves: Green < Darjeeling <Black. Although Darjeeling tea is commonly sold as “black tea” it belongs to the group of “oolong teas” for which an incomplete oxidation is obtained during production [2]. This observation is reflected in the PCA scores
plot, where the Darjeeling samples are located between black and green teas. Interestingly, the extract of “Green Tea A” is clearly separated in the scores plot from all other green tea samples.
Several FMF compounds contributing to the differences between the tea types are highlighted in the loadings plot (Figure 1). Sum formulae were calculated usingSmartFormula, which provides a ranking according to the best fit of measured and theoretical isotopic pattern within a specified mass accuracy window. The quality of the isotopic pattern fit is expressed in the mSigma-Value (Table 1): the smaller the mSigma-Value, the better the isotopic matching. All elemental compositions given in Table 1 represent the first Sigma rank within a 1 ppm mass accuracy window. A query of the sum formulae in public databases using the CompoundCrawler enabled a tentative identification of these compounds. The identity of epicatechin, gallocatechin, epigallocatechin, epicatechin-3-gallate and epigallocatechingallate was confirmed by comparison with pure standards.
Loading 5 was among the compounds responsible for the separation of “Green Tea A” from all other samples (Figure 1). Sum formula generation based on MS data provided only three possible candidates within 1 ppm mass accuracy with C27H23O18 being ranked first by mSigma value (data not shown). An Auto MS–MS experiment was performed to provide fragmentation information for verification of the candidate sum formula. SmartFormula3D, which combines exact mass and isotopic pattern information from both MS and MS–MS spectra as well as neutral loss information [3] corroborated the candidate formula. A CompoundCrawler database query for this sum formula suggested trigalloylglucose as likely structure.This preliminary identification is in close agreement with the assigned neutral losses and fragment ion sum formulae within the MS–MS spectrum.
Figure 1: PCA scores and loadings plot of 14 tea extracts:black (O), green (X) and Darjeeling (Δ) tea extracts analysed in triplicate.
PCA Scores and loadings plot
Scores Loadings
black
Darjeelinggreen
12
3
4
56
7
8
9
1011
12
A
Assignments based on maXis MS–MS data can be made with a high level of confidence, as demonstrated by the sub-mDa mass deviations shown in the Figure 2. The structure of the candidate is consistent with the observed cleavages of ester and neighboring bonds of the gallic acid-sugar conjugate Trigalloyl-glucose.
Conclusion
Fast UHPLC separation coupled to maXis UHR-TOF-MS analysis enabled the comprehensive analysis of complex mixtures of small molecules, namely black and green tea samples. PCA analysis enabled differentiation of tea samplesaccording to their oxidation level.
The sub-ppm mass accuracy and the True Isotopic Pattern information delivered by the maXis instrument facilitated the straightforward sum formula generation of compounds characteristic for different types of tea. By combining accurate mass and isotopic pattern information in MS and MS–MS spectra, SmartFormula3D facilitates the faster interpretation of fragmentation data. Additionally, SmartFormula3D extends the capability for unique sum formula generation to compounds with higher molecular weight.Used in combination with CompoundCrawler database queries, the feasibility of the UHPLC and maXis UHR-TOF analytical approach for the identification ofunknown compounds in metabolomics studies has been demonstrated.
References
[1] A. Barsch et al., Application Note ET-19, Bruker Daltonik GmbH, www.bdal.com (2010). (Metabolic profiling of tea extracts by high-resolution LC in combination with maXis UHR-TOF-MS analysis)
[2] H.N. Graham, Prev Med., 21(3), 334–50 (1992).
[3] C. Stacey et al., Technical Note #TN-23, Bruker Daltonik GmbH, www.bdal.com (2008).
Authors
Aiko Barsch, Gabriela Zurek, Wiebke Lohmann, Bruker Daltonik GmbH, Bremen, Germany
Loadings (see Figure 1)
Retention time (min)
m/zmeasured
Error(ppm)
mSigma Sum formula[M-H]-
Possiblecompound
1 0.8 191.0560 0.51 6.7 C7H11O6 Quinic acid
2 6.7 337.0926 0.70 10.4 C16H17O8 Coumaroyl-quinic acid
3 2.4 169.0143 –0.59 6.6 C7H5O5 Gallic acid
4 2.8 343.0670 0.08 7.9 C14H15O10 Theogallin
5 6.9 635.0889 0.13 5.4 C27H23O18 Trigalloyl-glucose
6 5.3 483.0780 –0.04 6.8 C20H19O14 Digalloyl-glucose
7 4.9 633.0731 0.35 13.7 C27H21O18 Corilagin
8 7.0 441.0828 –0.22 8.4 C22H17O10 Epcatechin-3-gallate
9 6.0 457.0076 0.07 4.0 C22H17O11 Epigallocatechin gallate
10 6.4 289.0718 –0.24 6.1 C15H13O6 Epicatechin
11 5.0 305.0665 0.58 2.4 C15H13O7 Epigallocatechin
12 3.5 305.0666 0.16 5.5 C15H13O7 Gallocatechin
Towards the Identification of Trigalloyl-Glucose using SmartFormula 3D
C27H23O18Trigalloyl-glucose
C20H17O130.59 mDa
C27H23O180.33 mDa
C13H13O9-0.22 mDa 483.0780
C20H19O14-0.16 mDa
313.0567
465.0669
635.0887
C7H4O4 + 0. 47 mDa
C7H6O5 - 0.29 mDaC7H4O4 + 0.82 mDa
C7H6O5 + 0.06 mDa
-MS2(635.0887)
0
25
50
75
100
Intens.
[%]
300 350 400 450 500 550 600 m/z
C27H23O18Trigalloyl-glucose
C20H17O130.59 mDa
C27H23O180.33 mDa
C13H13O9-0.22 mDa 483.0780
C20H19O14-0.16 mDa
313.0567
465.0669
635.0887
C7H4O4 + 0. 47 mDa
C7H6O5 - 0.29 mDaC7H4O4 + 0.82 mDa
C7H6O5 + 0.06 mDa
-MS2(635.0887)
0
25
50
75
100
Intens.
[%]
300 350 400 450 500 550 600 m/z
Sum formula of precursorSum formulae of
corresponding fragments
Figure 2: SmartFormula 3D result and MS–MS spectrum of Compound 5.
Table 1: Selected loadings differentiating the tea samples.
-4- -5-
Metabolomics Posterbook 2011
Conclusions
MPI Chemical Ecology & Bruker Daltonics
Introduction
Methods
Summary
Development of an HPLC-TOF-MS screening-platform to assess ppHypSys-dependent metabolic changes in Nicotiana attenuata during insect herbivory
ESI-TOF-MS
• The proposed method combining an ESI-TOF-MS based metabolic screening with statistical analysis, kinetic profiles and subsequent identification was successfully applied to a large plant metabolomics data set investigating the influence of ppHypSys during insect herbivory.
• The nitrogen containing molecules are currently independently explored using WT plants grown in N15 media. In a next step experiments using accurate and high resolution MS/MS data and classical NMR analysis will be carried out.
Using an untargeted ESI-TOF-MS combined with PCA, we verified that large reconfigurations of N. attenuatametabolome occur during insect attack. By comparing profiles obtained from WT and genetically modified plants, we discovered the regulatory role played by ppHypSys on the biosynthesis of caffeoylputrescine, an alkaloid accumulating during insect feeding1.
1Lou and Baldwin (2004) Nitrogen Supply Influences Herbivore-Induced Direct and Indirect Defenses and Transcriptional Responses in Nicotiana attenuata. Plant Physiology
A reliable HPLC-TOF-MS-based metabolomics platform to screen for chemical changes in leaf extracts of Nicotiana attenuata after simulated insect attack W+OS (W = mechanical wounding, OS = application of Manduca sexta’s oral secretions) is presented.
The platform was used to investigate the role of the hydroxyproline-rich glycopeptide systemin precursor (ppHypSys) as a signaling molecule during insect herbivory. After either reducing endogenous ppHypSys transcript levels by RNAi (IRsys) or over-expressing the ppHypSys precursor under the control of a 35S-promotor (OVsys), we compared metabolomic changes of wild-type (WT), IRsys and OVsys in response to W+OS.
Matthias Schöttner1; Eva Rothe1; Emmanuel Gaquerel1; Beatrice Berger1; Aiko Barsch2, Birgit Schneider2; Robert Fezatte3; Gabriela Zurek2; Ian T. Baldwin1
1Max Planck Institute for Chemical Ecology 07745 Jena – Germany2Bruker Daltonik GmbH, 23359 Bremen - Germany3Bruker Daltonics, Billerica MA 01821 -USA
Fig. 1 (A) Nicotiana attenuata leaf attacked by Manduca sexta. (B) Insect herbivory was simulated by wounding leaves with a fabric pattern wheel and applying insect’s oral secretions.
A B
Fig. 2 Experimental design, sample preparation and HPLC-ESI-TOF-MS settings, data evaluation
EX
TR
AC
TIO
N
200 ng/ml Atropine as an internal standard 100 mg frozen tissue 1000 µl MeOH/buffer 40:60 (v/v)Buffer = ammonium acetate 50mM, pH 4.8
Fast-prep45 secCentrifugation (2x 13000rpm)
0 1 4 14 72 86 120 168 Time (h)
WTIRsys
OVsysDES
IGN
3 biological replicates per time point, treatment and genotype
flash-frozen after harvesting
DA
TA
EV
ALU
ATIO
N Total size of data set: 504 analyses.
Principal Component Analysis (PCA) in Bruker ProfileAnalysisTM software.Formula generation using SmartFormulaTM.Kinetic Profiles in Microsoft Office Excel.
m/z 251.1384
0
10000
20000
30000
40000
50000
60000
70000
1 4 14 72 86 120 168
ir
irc
ov
ovc
wt
wt c
Drop Page Fields Here
Average of Area7
Time
Line
WT
0
10000
20000
30000
40000
50000
60000
70000
1 4 14 72 86 120 168
ir
irc
ov
ovc
wt
wt c
Drop Page Fields Here
Average of Area7
Time
Line
OVsys
IRsysW+OS
Time (h)
Norm
aliz
ed a
rea
m/z 251.1384m/z 251.1384
0
10000
20000
30000
40000
50000
60000
70000
1 4 14 72 86 120 168
ir
irc
ov
ovc
wt
wt c
Drop Page Fields Here
Average of Area7
Time
Line
WT
0
10000
20000
30000
40000
50000
60000
70000
1 4 14 72 86 120 168
ir
irc
ov
ovc
wt
wt c
Drop Page Fields Here
Average of Area7
Time
Line
OVsys
IRsysW+OS
Time (h)
Norm
aliz
ed a
rea
m/z 251.1384
HP
LC
-ES
I-TO
F-M
S
Column: Macherey & Nagel Nucleodur C18 Pyramid 50x2mm, 5µm particles
A: Water 0.5 % formic acid (FA)B: Acetonitrile 0.5 % FABinary Gradient: 1min 8% B, in 5 min to 65% B, isocratic for 10min, 6 min flushing with 90% B, reconditioning 11 min (total runtime 32 min)
MS: Bruker MicrOTOFTM
ESI in positive and negative modeScan Range: 50-1400m/zCalibrant: Sodium formate clusterSpectra rate 1Hz
To filter out the changes influenced by ppHypSys the principal component analysis (PCA) was applied and the impact of specific ions on the grouping of samples was investigated on both a time- and genotype-scale. Prior to PCA, peak finding was performed using the FindAllCompounds peak finder delivering isotopic as well as chromatographic information.
Fig. 3 presents an example of the general workflow: All datasets are first screened using PCA. Influential loadings explaining the trends in the scores plot are investigated more closely. Here, m/z 251 at 60s, corresponding to caffeoylputrescine, is one of the highest loading on PC1 (A). Kinetic plots for this ion showed that plants silenced for ppHypSys expression failed to accumulate this compound (B) during herbivory.
The peak of interest was then localized in the original data (C) and the elemental composition was calculated using both the accurate mass and isotopic pattern information (D+E).
Results
Fig. 3 General work-flow for the filtering and identification of ions of interest using ProfileAnalysis : Example of caffeoylputrescine
NHNH2
O
OH
OHMolecular Formula = C13H18N2O3
Possible structure:
Localization of the exact mass
Elemental compositions for the exact mass 251.1384
Comparison with the calculated isotope pattern of C13H19N2O3 [M+H]+
Caffeoylputrescine
C
A
PCA Loadings PlotPCA Scores Plot
Compound bucket
m/z 251.1384
0
10000
20000
30000
40000
50000
60000
70000
1 4 14 72 86 120 168
ir
irc
ov
ovc
wt
wt c
Drop Page Fields Here
Average of Area7
Time
Line
WT
0
10000
20000
30000
40000
50000
60000
70000
1 4 14 72 86 120 168
ir
irc
ov
ovc
wt
wt c
Drop Page Fields Here
Average of Area7
Time
Line
OVsys
IRsysW+OS
Time (h)
Norm
aliz
ed a
rea
m/z 251.1384B
D
E
PCA of all wild type samples (nicotine excluded; explained variance 44%).
Trend for late timepoints in treated samples
For research use only. Not for use in diagnostic procedures.
Conclusions
MPI Chemical Ecology & Bruker Daltonics
Challenges in the investigation of the metabolic changes in Nicotiana attenuata during insect herbivory using an improved HPLC-TOF-MS method
ESI-Qq-TOF-MS
The target of this study is the investigation of the metabolic changes in leaf extracts of Nicotiana attenuata after simulated insect herbivory (see Fig. 1) by mechanically wounding and applying oral secretions of Manduca sexta (W+OS) using HPLC-ESI-TOF-MS. Responses of wild-type (WT) plants are primarily compared before investigating the influence of a genetic modification in which the hydroxyproline-rich glycopeptide systemin precursor (ppHypSys) is either over-expressed (OV) or down-regulated (IR). The challenges in this type of study and strategies for quality control are discussed.
Preliminary results obtained from a first data set of the same sample origin (7 time-points with three biological replicates per time-point, treatment and genotype analyzed with a 17min gradient) were used to design the experiment presented here. The number of biological replicates was increased from three to five to access the biological variation. The chromatographic separation was optimized in order to minimize suppression effects due to coelutions. A pooled sample generated from all individual samples was spiked with internal standards (reserpine, atropine) served as quality control (QC) sample to monitor instrumental variation.
Matthias Schöttner1, G. Zurek2, E. Rothe1, E. Gaquerel1, B. Berger1, B. Schneider2, A. Barsch2, Mike McDonell3, Ian T. Baldwin1
1Max Planck Institute for Chemical Ecology 07745 Jena – Germany2Bruker Daltonik GmbH, 28359 Bremen –Germany3Bruker Biosciences, Delta, BC, Canada
Introduction
Fig. 1 (A) Nicotiana attenuata leaf attacked by Manduca sexta. (B) Simluated insect herbivory by wounding leaves with fabric pattern wheel & applying insect’s oral secretions.
A B
Materials & MethodsSampling & Sample PreparationTreated and untreated leaves (control) from WT, OV, IR (5 biological replicates) were harvested 1, 14, 86, and 120 h after treatment & flash frozen.Extraction: MeOH/acetic acid buffer 40:60 v/vInternal Standards (ISTD): Atropine, reserpine.QC Samples: QC1: Extraction buffer + ISTD; QC 2: pooled sample of all extracts + ISTD. N= 16 injections.Chromatographic MethodColumn: Phenomenex GeminiNX 150x2mmEluent A: water +0.1%AcN+ 0.05% HCOOHEluent B: AcN + 0.05% HCOOHGradient: 0min-5%B; 2min- 5%B; 2-30min-5-80%B; 35min-80%B; 40min- 5%B; 45min-stop Flow rate: 0.2mL/min Mass Spectrometry MethodMS: Bruker micrOTOF & micrOTOF-Q ESI-oa-Qq-TOF MS (Bruker Daltonik, Bremen)Ionization: ESI/ positive modeScan range: 50-1600 m/z; 1Hz acquisition rateCalibration: external with NaFormate clustersData EvaluationSoftware: ProfileAnalysisTM for PCA and t-test.Bucketing: time range: 0.5-30min mass range: 50-1600m/z Advanced bucketing; Normalization: None; Scaling: Pareto.Molecular features need to be present in at least 85% of all buckets for statistical calculation.
Quality ControlThe reproducibilty of the method was checked by means of the ISTDs reserpine & atropine. The RSD values of peak areas & heights were well below 10% for the entire sequence (Fig. 2A). A pooled sample QC2 representing a ‚chemical average‘ of all samples nicely clusters in the middle of a principle component analysis (PCA) scores plot of all measured samples. The QC data is used to verify the analytical performance and enables valid statitical interpretation in complex data sets.
Fig. 2: A- Reproducibility of reserpine & atropine in QC1 (buffer) and QC2 (pooled sample), N=16; B- PCA scores plot of all samples and QC2.
-0.2 0.0 0.2 0.4 PC 1
-0.2
0.0
0.2
0.4
PC 2
+QC
IR
xOV
WT
Group
+QC
IR
xOV
WT
Group
RS
D V
alue
s [%
]
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0AreaHeight
A B
• By means of internal standards and a pooled quality control sample, the reproducibility of the analytical method was monitored.• PCA and t-test were successfully applied to filter out metabolic changes. • The identification of regulated metabolites can be achieved based on molecular formulae derived from accurate mass spectra plus chemical knowledge.• This method will be applied to investigate the influence of the genetic modification.
Fig. 3: PCA and t-test results for all wildtype samples and molecular formulae.
A: PCA (pareto scaling) of all WT samples, explained variance PC1 vs PC2: 48.4%.
B: Excerpt of t-test result table and selected bucket statistics plots for WT 120h samples. Bucket Exact m/z Formula [M+H]+
11.4min-484.24m/z
484.2437 C26H34N3O6
14.2min-598.31m/z
598.3083 C26H45O14
5.1min-251.14m/z
251.1381 C13H19N2O3
9.6min-488.24m/z
488.2380 C25H34N3O7
11.7min-484.24m/z
484.2427 C26H34N3O6
2.1min-163.12m/z
163.123 C10H15N2
C: Example for SmartFormula result.
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 PC 1
-1.0
-0.5
0.0
0.5
1.0
PC 2
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 PC 1
-1.0
-0.5
0.0
0.5
1.0
PC 2
Late timepoints W+OS
Scores Loadings11.4min : 484.24m/z
0 10 20 30 Analysis
0
1
2
3
4
5
x104Int.
120 h
86 h
14 h
1 h
WS+OSCtrl
120 h
86 h
14 h
1 h
WS+OSCtrl
BucketStatistics
Fig. 3: PCA and t-test results for all wildtype samples and molecular formulae.
11.4min : 484.24m/z
2 4 6 8 Analysis
1
2
3
4
54x10
Int.
o/o: 0.17
14.2min : 598.31m/z
2 4 6 8 Analysis
0.2
0.4
0.6
0.84x10
Int.
5.1min : 251.14m/z
2 4 6 8 Analysis
0
2
4
6
4x10Int.
9.6min : 488.24m/z
2 4 6 8 Analysis
0
2000
4000
Int.
11.7min : 484.24m/z
2 4 6 8 Analysis
2
4
6
4x10Int.
2.1min : 163.12m/z
2 4 6 8 Analysis
2.5
3.0
3.5
4.0
5x10Int.
o/o: 0.22
o/o: 0.11
o/o: 0.21
o/o: 0.17
o/o: 0.64
o: WT Controlo: WT+OS
Results
Identification
Fig. 4: A- SmartFormula 3D results, B- MS/MS and simulated spectrum with annotated formulae.
150 200 250 300 350 400 450 m/z
0.0
0.5
1.0
1.5
2.0
X104Intens.
0.0
0.5
1.0
1.5
2.0
x104
129.1385145.0294
177.0547220.0963
291.1712
322.2112484.2416
C 10 H 8 O 3
C 9 H 6 O 3
NH
3
NH
3
C 3 H 7 N
C 3 H 7 N
C 4 H 12 N 2
C 4 H 12 N 2
+MS2(484.2428), 35-35eV, 11.43-11.84min #677-#701
C 7
H 1
7 N
2 ,
0.17
mD
a
C 9
H 5
O 2
, -0
.95
mD
a
C 9
H 7
O 3
, 0.
0053
mD
aC
8 H
18
N 3
O 1
, 1.
4 m
Da,
C
10
H 9
O 3
, -0
.11
mD
a
C 1
2 H
14
N 1
O 3
, 0.
49 m
Da
C 1
3 H
16
N 1
O 3
, 0.
52 m
Da
C 1
2 H
17
N 2
O 3
, 1.
1 m
Da,
C
14
H 1
8 N
1 O
3 ,
1.6
mD
a
C16
H 2
3 N
2 O
3 , 3
.2 m
Da
C17
H25
N2
O3
, 0.5
5 m
Da
C 2
6 H
34
N 3
O 6
, 2
mD
a
SmartFormula3D spectrum: C 26 H 34 N 3 O 6
163.0390
C17
H28
N3
O3
, 1.3
mD
a
C16
H26
N3
O3
, 1.2
mD
a
B
NH
NH
NH
O
O
OHOH
OHOMe
Structural Hypothesis for C26H33N3O6Caffeoyl-feruloyl-spermidine
A
MS peak list MS/MS peak list
Fragment formulae
All wildtype (WT) samples were investigated first to identify changes related to time and treatment only using PCA and a Student‘s t-test. The late timepoints of the treated samples are clearly separated in the PCA scores plot (Fig. 3A). The bucket statistics plots for the selected loading at 484.24m/[email protected] shows the same trend. A t-test of the treated and non-treated WT samples after 120h was applied to narrow down the number of compounds for further identification. The t-test result is returned for each bucket with the additional information about the regulation level, the maximum intensity in all analyses and the number of detected compounds per group (Fig. 3B). The first six bucket statistics plot are given as examples (Fig. 3B). In total 59 compounds with p-values < 0.05 and a minimum of 2-fold upregulation in the treated samples were found. This list of compounds was used as the basis for a scheduled precursor list for a targeted AutoMS/MS experiment. Molecular formulae of the target compounds have been generated based on the accurate MS and MS/MS data (Fig. 3C) using Smart FormulaTM.
After statistical analysis, the major challenge is the identification of the compounds involved in regulation. The accurate MS/MS data of the compound 484.24m/[email protected] has been processed with SmartFormula 3D (Fig. 4A), leading to an MS/MS spectrum assigned with fragment formulae and the corresponding neutral losses (Fig. 4B). With the molecular formula containing 3 nitrogen atoms, a spermidine derivative can be hypothesized. The fragment at 163m/z is typically observed for caffeic acid derivatives, 177m/z for ferulic acid derivatives. The combined information leads to a caffeoyl-feruloyl-spermidine [1] as potential metabolite being upregulated during herbivory. [1] Youhnovsky, N. et. al. Helvetica Chimica Acta 1998 81(9) 1654-1671.
D: Summary of molecular formulae for buckets from B.
For research use only. Not for use in diagnostic procedures.
-6- -7-
Plant Metabolomics Plant Metabolomics
Metabolomics Posterbook 2011
Conclusions
Universität des Saarlandes & Bruker Daltonics
Bacteria producing secondary metabolites with biological activities are important sources of potential new drugs [1]. An example for natural products discovered from myxobacteria are the Epothilones being promising anti-cancer agents. Chromatographic separation of bacterial culture extracts combined with ESI-TOF-MS analysis is a powerful tool for the comparison of production patterns and the identification of compounds via their molecular formula. Manual data examination is time consuming due to high complexity of samples. Statistical methods can facilitate the extraction of relevant information. We describe here a metabolomics based approach for extracts from myxobacteria using ESI-TOF-MS data, principal component analysis (PCA) and a sub-sequent identification via the molecular formula from accurate mass data.
[1] Gerth K, J. Biotechnol., 106(2-3), 233-253 (2003)
Bacteria were fermented in a 50 mL scale using complex media. Cultures contained XAD-16 resin as absorber. Cells and absorber resin were harvested by centrifugation, extracted with methanol and diluted in acetonitrile/water prior to injection. LC-MS measurements were performed using a micrOTOFTM orthogonal ESI-TOF mass spectrometer (Bruker Daltonik, Bremen) coupled to a UPLC system (Waters, Milford) equipped with photodiode array detector.
LC-MS data was prepared for PCA analysis in ProfileAnalysis using a bucketing approach (Fig. 2). The LC-MS was integrated from 0.2-7.5 min and 200-1200 m/z in time- and m/z-buckets of 0.2 min and 1 m/z. Each data set was normalized to the total intensity in an analysis. The variability of the instrumental analysis was regularly checked by two quality control (QC) samples. The first QC samples consisted of a standard mixture of compounds unrelated to the real samples in order to control retention time and mass calibration stability. The second QC sample was one of the bacterial extracts injected several times randomly within the sequence of measurements.In order to asses the variation in the fermentation experiment, five closely related strains producing the same predominant compounds were fermented with six biological replicates each. The replicate extracts of the five strains revealed considerable deviation within replicate fermentations, but still resulted in formation of groups indicative of subtle systematic differences between their metabolite profiles (Fig. 3).
The exploration of myxobacterial metabolite profiles by LC–MS screening for the presence of new natural products is described. Extracts from fermentations of Myxococcus strains are analysed by UPLC-coupled ESI-TOF mass spectrometry and the obtained data are processed using principal component analysis (PCA). The generation of molecular formulae from accurate mass measurements facilitates rapid compound identification. PCA is used to find trends in the different secondary metabolite production patterns from Myxococcus strains which may reveal correlations regarding not only the geographical distribution of the samples, but also biological properties of the strains.
Introduction
Methods
Summary
Results
Analysis of the Intraspecies Variation of Secondary Metabolite Production by Myxococcus xanthus using ESI-TOF-MS and PCA
ESI-TOF-MS
Daniel Krug1, Gabriela Zurek2,Birgit Schneider2, Rolf Müller1
1 Universität des Saarlandes, Saarbrücken, Germany2 Bruker Daltonik GmbH, Bremen, Germany
Fig. 1 UPLC chromatogram and high-resolutionESI-TOF-MS measurement from a Myxococcusxanthus extract.
t (sec)
Swarming M. xanthus colonyPhotography: G.J Velicer
DKxanthene-518
Fig. 2 PCA analysis of extracts from 6 M.xanthus strains plus one novel Myxococcusisolate facilitates comparative metaboliteprofiling and identification of compounds viatheir molecular formula.
-0.2 0.0 0.2 0.4 0.6 0.8 1.0 PC 1
-0.4
-0.2
0.0
0.2
0.4
0.6
PC 2
-0.2 0.0 0.2 0.4 0.6 0.8 1.0 PC 1
-0.4
-0.2
0.0
0.2
0.4
0.6
PC 2
1 2 3 4 5 6 Analysis0.00
0.01
0.02
0.03Int.
450 475 500 525 550 575 600 625 650 675 Time [s]0
1
2
3
4
5x10Intens.
DK1622A_63458.d: EIC 320-350 +All MS FullScan BDAL120_63429.d: EIC 314.305±0.002 +All MS FullScan
311.1852
317.1141324.2165327.2553
341.2284 346.2916 355.2557
358.3665
365.2294
+MS, 495.3s #981
0
2000
4000
6000
Intens.
310 320 330 340 350 360 m/z
261.1233
521.2394
+MS, 125.9-131.5s #(249-260)
158.9616 229.1536
287.1386 356.7092425.7259
483.2867540.3092
631.2763
712.4111 850.4457 965.5645
+MS, 307.8-310.9s #(610-616)0.0
0.5
1.0
1.5
2.0
2.5
5x10Intens.
0
1
2
3
4
5
4x10
200 400 600 800 1000 m/z
Illustration of rectangular bucketing approach for PCA of complex LC-MS datasets like myxobacterial extracts
PCA model output: Scores & Loadings Plot
6x Myxococcus xanthus
Myxococcus,novel isolate
m/z 314.31@ 485 sec
Bucket Statistics
EIC +(314.305)
GMF for m/z = 314.30523:C19H40N1O2
Fig. 3 PCA for extracts from 5 closely related M. xanthus strains reveals systematic differencesbetween their metabolite profiles. UPLC-ESI-TOF MS data is used for GMF and allows to evaluatealso quantitative variation among samples.
Scores & Loadings Plot
Dk1622
A B
C D
Photography: Michiel Vos
400 600 800 1000 1200 1400 Time [s]
0.00
0.25
0.50
0.75
1.00
1.25
5x10Intens.
A47A_63473.d: EIC 535.254±0.01 +All MS FullScan A47C_63452.d: EIC 535.254±0.01 +All MS FullScanA9F_63474.d: EIC 535.255±0.005 +All MS, Smoothed (0.51,1,GA) A9B_63437.d: EIC 535.254±0.005 +All MS FullScan, Smoothed (0.51,1,GA)A96B_63460.d: EIC 535.254±0.01 +All MS FullScan A96C_63457.d: EIC 535.254±0.01 +All MS FullScan
515 520 525 530 535 540 545 550 555 m/z
D
t (sec)
B
C
EIC 535.25
400 600 800 1000 1200 1400 1600 1800 2000 Time [s]0
1
2
3
5x10Intens.
A9B_63437.d: EIC 519.26±0.01 +All MS FullScan, Smoothed (0.51,1,GA) A9A_63466.d: EIC 519.26±0.01 +All MS FullScan, Smoothed (0.51,1,GA)A9F_63474.d: EIC 519.26±0.005 +All MS, Smoothed (0.51,1,GA) DK1622A_63458.d: EIC 519.26±0.01 +All MS FullScan, Smoothed (0.50,1,GA)DK1622B_63471.d: EIC 519.26±0.01 +All MS FullScan, Smoothed (0.51,1,GA) DK1622C_63477.d: EIC 519.26±0.01 +All MS FullScan
510 520 530 540 550 560 570 580 590 m/z
DK1622
B
t (sec)
EIC 519.26 Clustering of analyses fromreplicate fermentations of 5closely related Myxococcusxanthus strains in the PCAmodel.
Exploration of quantitative variation among samples, which accounts in part for the observed grouping pattern of strains, using EIC‘s from precise ESI-TOF MS measurements.
Calculation of putative molecular formulae for compounds of interest with „sub-ppm confidence“ utilizing GMF and SigmaFit®.
Myxococcus xanthus fruitingbodies emerging from a soil sample.
GMF results:m/z 519.26 C29H35N4O5 [M+H]+
= DKxanthene-518m/z 535.25 C29H35N4O6 [M+H]+
= DKxanthene-534
• We demonstrate that differences in secondary metabolite production patterns of myxobacterial strains can be extracted by PCA, even con-sidering significant biological variation within replicate fermentations.
• It is shown that accurate mass and isotopic pattern data from ESI-TOF–MS is an important key to enable the identification of compounds according to their molecular formula.
• We project an application to the discovery of novel compounds produced by myxobacteria, by enabling at the same time rapid de-replication and definition of putative target compounds for further studies. This is of special importance for research concerning natural products from myxobacterial strains, which are often talented multiproducers of secondary metabolites.
Compounds representing such differences were identified using the accurate mass data. Clustering of these analyses occured in part due to a high degree of quantitative variation regarding production of DKxanthenes, a recently characterized family of compounds which are important for the complex lifecylce of M. xanthus [4].In a second sequence, 110 bacterial isolates of the species M. xanthus covering largely variant spatial ranges (from centimeter scale to locations worldwide [5]) were cultured and analyzed under the above described conditions. Comparison of subsets of these extracts with samples from novel uncharacterized strains in a PCA model (Fig. 2) serves to quickly obtain an overview of their metabolite profiles and identify novel compounds.
[4] Meiser P, Proc. Natl. Acad. Sci. USA, 103 (50), 19128-19133 (2006)[5] Vos M, Velicer GJ, Appl Environ Microbiol., 72(5), 3615-25 (2006)
The separation was carried out with a binary acetonitrile-water gradient with 0.1 % formic acid and a reversed phase column (50x2 mm, 1.7 µm particles) (Fig. 1). Data was acquired in ESI positive and negative mode (Scan range: 150-1200 m/z). Sodium formate solution was injected as external calibration standard. Data was evaluated using ProfileAnalysisTM software for principal component analysis and Generate Molecular FormulaTM (GMF) with SigmaFit [2,3].
[2] Krug D, LC•GC Europe, March 2007 Applications Book, 41-42 (2007)[3] Ojanperä S, Rapid Commun. Mass Spectrom., 20 (7), 1161-1167 (2006)
AcknowledgementsWork in the laboratory of Rolf Müller was funded by „Deutsche Forschungsge-meinschaft“ (DFG) and „Bundesministerium für Bildung und Forschung“ (BMB+F).
For research use only. Not for use in diagnostic procedures.
Universität des Saarlandes & Bruker Daltonics
Conclusions
UHR-TOF-MS
Introduction
Mining the Myxobacterial Secondary Metabolome using High-resolution Mass Spectrometry: Natural Product Discovery as an Analytical Challenge
Daniel Krug1, Gabriela Zurek2, Niña S. Cortina1, Aiko Barsch2 and Rolf Müller1
1Pharmaceutical Biotechnology, Universität des Saarlandes, Saarbrücken, Germany 2Bruker Daltonik GmbH, Bremen, Germany
Myxobacteria and other secondary metabolite producers represent an important source of biologically active natural products with considerable promise for human therapy. While several studies have recently highlighted the enormous genetic potential of many myxobacterial species for secondary metabolite biosynthesis,[1,2] the number of compound classes reported from individual strains clearly falls behind genetic capabilities. Thus, the discovery of novel secondary metabolites from genetically proficient producers currently constitutes a substantial bottleneck. Improved analytical methods, based on the combined use of LC-coupled high-resolution mass spectrometry and statistical data evaluation, can significantly facilitate the process of uncovering these "hidden" bacterial secondary metabolomes.[3,4]
• Efficient detection of molecular features in high-resolution LC-MS datasets is a crucial pre-requisite for the susequent mining of complex myxobacterial samples for the presence of novel natural products, using statistical tools.
• Data acquired on the UPLC-ESI UHR-TOF platform are well suited for comprehensive profiling applications, enabling the in-depth comparison of myxobacterial secondary metabolomes with previously unmatched success.
• In the present work, M. xanthus DK1622 could be shown to produce 5 novel secondary metabolite classes, in addition to the 5 pre-viously known natural products. Future efforts using the analytical approach presented here aim to correlate compounds
•Myxococcus xanthus extracts were separated under UPLC conditions using a RP C18 column (50 x 2 mm, 1.7 µm particle size).
•ESI-MS measurements were performed using positive ionization on the maXis UHR-TOF (Bruker Daltonik; mass range: 100-2000 m/z, repetition rate: 3 Hz).
•Targeted screening was carried out using precise EICs and isotope pattern evaluation via SigmaFitTM with TargetAnalysisTM software (Bruker Daltonik).
•Statistical interpretation using PCA or Student‘s t-test including all data pre-processing was performed with ProfileAnalysisTM software (Bruker Daltonik).
Materials & Methods Genomics meets Metabolomics
Metabolomics-based experiments employing high-resolution LC-MS measurements are convenient because they provide the opportunity to apply both targeted queries and unbiased statistical treatment to the same dataset. Extracts derived from cultivations of Myxococcus xanthus DK1622 were subjected to targeted analysis, confirming the presence of 5 known secondary metabolite classes. Moreover, the secondary metabolomes of DK1622 mutants obtained by the inactivation of biosynthetic pathways were compared to wildtype metabolite profiles, in order to identify molecular features missing in the mutant extracts. Such experiments enable the discovery of previously unrecognized secon-dary metabolites and allow at the same time their assignment to a specific biosynthetic pathway encoded in the myxobacterial genome.
www.myxo.uni-saarland.de
I
0
1
2
3
x105
1 2 3 4 5 6 7 8 Ret. [min]
BPC(100-2000 m/z)
DKx
anth
ene
Myx
oche
lin
Myx
alam
ide
Myx
ochr
omid
eM
yxov
iresc
in
I
0
1
2
3
x105
1 2 3 4 5 6 7 8 Ret. [min]
I
0
1
2
3
x105
1 2 3 4 5 6 7 8 Ret. [min]
BPC(100-2000 m/z)
DKx
anth
ene
Myx
oche
lin
Myx
alam
ide
Myx
ochr
omid
eM
yxov
iresc
in
+MS, 3.05-3.10 min
400.2499404.1812
409.3392
0.0
0.5
1.0
1.5
399 401 403 405 407 409 411 m/z
I x105
Myxochelin AC20H26N3O6
[M+H]+ 404.1816 m/z
EICs
A
B
Fig. 1: High-resolution LC-MS analysis of an M. xanthus DK1622 extract. Targeted screening reveals known compounds with high confidence, even if they constitute only minor components in the complex mixture (A). Signals potentially representing novel metabolites span a high dynamic range (B).
Targeted screening & MS performance evaluation
Detection of molecular features and statistical interpretation
PCA Scores & Loadings Plot (PC1 versus PC2, 67% explained variance)
-0.4 -0.2 0.0 0.2 0.4 PC 1
-0.4
-0.2
0.0
0.2
0.4
PC 2
-0.4 -0.2 0.0 0.2 0.4 PC 1
-0.2
0.0
0.2
0.4
PC 2
wildtype
mutant
506.271 m/z@ 2.20 min
Results of t-test model (wildtype versus mutant, 4 replicates each)
No. compound P-value Ratio WT/Mut
x-fold change
Max. Intensity
WT count(of 4)
Mut.Count (of 4)
1 159.9s : 506.27m/z - - - 52493 4 013 151.5s : 515.25m/z - - - 25005 4 017 138.8s : 523.25m/z - - - 4374 4 0…597 215.9s : 601.32m/z 0.67039 0.76 -1.32 21203 4 2880 574.6s : 545.35m/z 0.78097 0.90 -1.11 1777 2 41290 433.2s : 416.32m/z 0.00244 1.41 1.41 94026 4 41397 258.0s : 535.25m/z 0.12887 0.65 -1.53 182171 4 42067 133.9s : 259.11m/z 0.38485 0.66 -1.51 9210 4 4
A B
Compoundname
PseudomolecularIon
Calculated [M+H]+ (Da)
Avg. error(ppm)
Avg. error(mDa)
Avg.mSigma
Myxalamid A C26H41NO3 416,3159 0,91 0,37 9,1
Myxalamid B C25H39NO3 402,3003 0,52 0,21 16,7
Myxochelin B C20H25N3O6 404,1816 0,96 0,4 7,2
DKxanthen-534 C29H34N4O6 535,2551 0,72 0,4 10,1
Myxovirescin A C35H61NO8 624,4470 1,28 0,8 9,1
Myxochromid A3 C45H63N7O9 846,4760 0,48 0,4 11,0
A B Avg. error(ppm)
Avg. error(mDa)
Avg. errormSigma
N=10 samples
Averages derivedfrom spectra acrossthe respectivemolecular features
0.95 0.39 19.3
0.74 0.3 23.5
0.23 0.1 12.2
0.91 0.51 9.3
1.08 0.68 16.7
0.51 0.4 19.1
Fig. 2: A, PCA scores & loadings plot, revealing the grouping pattern of wildtype and mutant M. xanthusDK1622 extracts (4 biological replicates each). The significance of a compound 506.271 m/z is highlighted. B, t-test model based on 2067 molecular features found in the LC-MS datasets. In comparison to the PCA model, additional low-abundance metabolites are revealed. These are putative derivatives of the compound 506.271 m/z.
Fig. 3: Target screening results for 5 known secondary metabolite classes from M. xanthus DK1622, using measurements from the UHR-TOF (A) and nanomate-LTQ-Orbitrap (B) platforms.
M. xanthus DK1622wildtype
M. xanthus DK1622mutant
A B C
BPC 100-2000 m/zEIC 506.2713 m/z
DK1622 wildtype
BPC 100-2000 m/zEIC 506.2713 m/z
Ret. (min)
DK1622 mutant
M. xanthus DK1622Genome ~ 9 Mbp
wildtype
?
M. xanthus DK1622Genome ~ 9 Mbp
mutant
Fig. 4: Identification of a novel secondary metabolite (506.271 m/z @ 2.20 min) from M. xanthus DK1622 by comprehensive analysis of metabolite profiles from the wildtype strain and a genetically engineered "knockout" mutant (A). Such novel compounds may be represented by subtle signals relative to the abundance of known metabolites and matrix substances (B,C; orange arrow). m/z range in (B) 200-1200 m/z; Ret. time 1.40 – 4.40 min.
1
2
3
5x10Intens.
0.02 4 Ret. (min)
1
2
3
5x10Intens.
0.02 4 Ret. (min)
1
2
3
02 4
5x10Intens.
1
2
3
02 4
5x10Intens.
1. Wenzel SC, Müller R (2009). Curr. Opin. Drug Disc. Dev. 12 (2), 220-230.2. Garcia RO, Krug D, Müller R (2009). Methods in Enzymology 458, 59-913. Krug D, Zurek G, Revermann O, Vos M, Velicer GJ, Müller R (2008). Appl.Environ.Microbiol. 74, 3058-684. Krug D, Zurek G, Schneider B, Garcia RO, Müller R (2008). Anal. Chim. Acta 624, 97-106
References
Funding from DFG, BMB+F and DAAD is gratefully acknowledged. For research use only. Not for use in diagnostic procedures.
-8- -9-
Bacterial Metabolomics Bacterial Metabolomics
Metabolomics Posterbook 2011
Conclusions
Bruker Daltonics & HIPS
ESI-UHR-QTOF-MS
Challenges in Metabolomics addressed by targeted and untargeted UHR-Q-TOF analysis
• The challenge of simultaneously achieving selectivity (hrEIC-traces) and speed (up to 20Hz) can be readily met with ESI-UHR-Q-TOF technology.
• The high resolution and mass accuracy LC-MS data sets allow for both targeted and untargeted analysis.
• Identification can be facilitated with SmartFormula 3D and accurate MS/MS data.
For research use only. Not for use in diagnostic procedures.
Name Formula[M+H]+
m/z theo. m/z meas. Error [ppm]
mSigma N
Cittilin A C34H39N4O8 631.2762 631.2762 0.34 6.07 19
DK-Xanthen 534 C29H35N4O6 535.2551 535.2547 0.79 5.01 19
Myxalamid A C26H42NO3 416.3159 416.3159 0.26 4.46 19
Myxochelin B C20H36N3O6 404.1816 404.1814 0.47 3.49 19
Myxovirsecin A C35H62NO8 624.447 624.4469 0.29 6.19 18
Gabriela Zurek1, Aiko Barsch1, Wiebke Lohmann1, Daniel Krug2, Rolf Müller2
1 Bruker Daltonik GmbH, 28359 Bremen, Germany2 Helmholtz-Institute for Pharmaceutical Research Saarland (HIPS) & Universität des Saarlandes, 66123 Saarbrücken, Germany
IntroductionMyxobacteria are promising producers of natural products exhibiting potent biological activities, and several myxobacterial metabolites are currently under investigation as potential leads for novel drugs. However, the myxobacteria are also a striking example for the divergence between the genetic capacity for the production of secondary metabolites and the number of compounds that could be characterised to date. The number of identified metabolites is usually significantly lower than expected from genome sequence information.
Here, we present an ESI-UHR-Q-TOF based analysis of myxobacterial secondary metabolites, which permits to solve several challenges frequently encountered in metabolite profiling studies. The challenges comprise the simultaneous need for fast, robust, and sensitive analysis with high resolution, accuracy and excellent reproducibility. Analytical solutions for targeted and untargeted metabolomics experiments by means of ESI-UHR-Q-TOF-MS are discussed.
Methods
• Myxococcus xanthus extracts were separated using a RP C18 column (50 x 2 mm, 1.7 µm particle size) on an UltiMate 3000TM rapid separation liquid chromatography system RSLC (Dionex).
• ESI-MS measurements were performed using positive ionization on the maXis UHR-Q-TOF-MS (Bruker Daltonik) m/z range: 100-1200 m/z, repetition rate: 3, 5, 10, 20Hz.
• Targeted screening was carried out using precise EICs and isotope pattern evaluation via SigmaFitTM with TargetAnalysisTM (Bruker Daltonik).
• Statistical interpretation using PCA including all data pre-processing was performed with ProfileAnalysisTM 2.0 (Bruker Daltonik) using FindMolecularFeatures (FMF) compounds and advanced bucketing (Bucket filter 75%; Pareto scaling).
0 2 4 6 8 10 Time [min]0.00
0.25
0.50
0.75
1.00
1.25
Inte
nsity
x 1
06
Myx
oche
lin
Myx
alam
id
DK
Xan
then
e
Myx
ochr
omid
eM
yxov
iresc
in
Citt
ilin
416.3160
417.3188
418.32160.00
0.25
0.50
0.75
1.00
1.25
Inte
nsity
x 1
05
416.0 417.0 418.0 419.0 m/z
Myxalamid A[M+H]+ C16H42NO3Error -0.2ppmmSigma 5.2Res. 39282
Sample Set
• The sample set consisted of five different Myxococcus xanthus strains with different genetic knockouts (1380, 2004, 2152, 4125) cultivated as four biological replicates each. DK 1622 served as reference strain. For performance evaluation and quality control (QC) purposes a pooled sample of all extracts was created (5µL of each extract). Samples were diluted 10x with methanol prior to injection of 2 technical replicates for each sample.
Fig. 1: Basepeak chromatogram of Myxococcus xanthus extract with EIC traces of known metabolites and mass spectrum of Myxalamid A. Fig. 3: A: Result of target screening as presented in
TargetAnalysisTM; B: summary of averaged mass accuracy and mSigma values for different secondary metabolites.
Sensitivity & Dynamic Range
• The extracts from myxobacteria represent a complex mixture of matrix components from the medium and bacterial secondary metabolites. Figure 1 depicts the base peak chromatogram (BPC) of a quality control sample composed of all extracts used in this study. The colored traces represent extracted ion chromatogram (EIC) traces of known metabolite classes from Myxococcus Xanthus. A very sensitive method with a high dynamic range is required to detect the low concentrated metabolites often covered by matrix components.
Selectivity & Speed
• When screening complex mixtures, it is important that selectivity and speed do not compromise each other. The selectivity in accurate mass and high resolution LC-MS measurements can e.g. be visualized with high resolution EIC traces (hrEIC). Figure 2A shows hrEIC traces of 1mDa width for five metabolites acquired at acquisition rates up to 20Hz. The achieved accuracy is thus independent from the acquisition speed. The mass spectra of Myxochelin B are depicted in Figure 2B. The mass accuracy, mSigma fit values as measure for the goodness of fit of the isotopic pattern and the resolution are again not influenced by the change in acquisition rate.
Targeted Analysis
• The targeted analysis is focused on known metabolites using retention time, accurate mass and isotopic pattern information. The selectivity for screening is achieved by hrEIC traces. Screening results with compound name, molecular formula, retention time (RT), mass accuracy and mSigma fit values are given as well as peak area, intensity values and RT deviation (see Figure 3A). The screening was carried out with 2mDa traces. The pooled QC sample was injected randomly distributed between samples and as block of 3 replicates before & after the samples, resulting in a total of 19 technical replicates.
Fig. 2: High resolution extracted ion chromatogram traces of 1mDa width of various metabolites and mass spectra of Myxochelin B at different acquisition speeds.
404.1814
405.1841406.1869
+MS, 2.63-2.73min #(402-417)
404.1814
405.1843406.1868
+MS, 2.64-2.72min #(716-739)
404.1815
405.1842406.1868
+MS, 2.65-2.72min #(1347-1386)
404.1813
405.1840406.1866
+MS, 2.64-2.71min #(2429-2499)
404.1816
405.1849406.1883
simulated: C 20 H 26 N 3 O 6 ,404.18
0246
x104
Inte
nsity 0
2
4x104
0
1
2x104
0.0
0.5
1.0x104
0.0
0.5
1.0x104
404.0 404.5 405.0 405.5 406.0 406.5 407.0m/z
3Hz
5Hz
10Hz
20Hz
0.5 ppm6.0 mSigmaRes. 38647
0.4 ppm0.6 mSigmaRes. 38036
0.2 ppm0.8 mSigmaRes. 38383
0.8 ppm8.8 mSigmaRes. 39034
MyxoQCMix_BA3_01_366.d
MyxoQCMix_BA3_05_321.d
MyxoQCMix_BA3_06_328.d
MyxoQCMix_BA3_06_335.d
Inte
nsity
0.0
0.5
1.0
x105
0.00
0.25
0.500.75
1.00x105
0
2
4
x104
0
1
2
x104
1 2 3 4 5 6 7 8Time [min]
3Hz
5Hz
10Hz
20Hz
Myxochelin B
Myxalamid A
DKXanthene 534
Myxovirescin A
Cittilin A
Simulated isotope pattern C20H26N3O6
A B
A
B
[1] Wenzel SC, Müller R (2009). Curr. Opin. Drug Disc. Dev. 12 (2), 220-230[2] Garcia RO, Krug D, Müller R (2009). Methods in Enzymology 458, 59-91[3] Krug D, Zurek G, Revermann O, Vos M, Velicer GJ, Müller R (2008). Appl. Environ. Microbiol. 74, 3058-3068[4] Krug D, Zurek G, Schneider B, Garcia R, Müller R (2008). Anal. Chim. Acta 624, 97-106
References
Funding from BMB+F, DFG and DAAD is gratefully acknowledged.
Untargeted analysis using PCA
• The data set of the different strains was subjected to Principle Component Analysis (PCA) to investigate different production patterns and metabolites missing potentially due to the genetic modification. One important step prior to the statistical analysis is the extraction of compounds using FindMolecularFeatures (FMF) for data reduction and comprehensive detection of all compounds in the LC-MS run. The PCA result for all strains is presented in Figure 4A: the different strains cluster in the PCA scores plot with the technical replicates being close to each other. From the loadings plot, the compounds related to the clustering are derived. Figure 4B gives three examples of loadings with different production patterns in metabolites. The m/z value 1011.53 represents a novel metabolite deficient in strain 2004. The two others are known metabolites Myxalamid A and DKXanthen-548.
Identification
• The identification of unknown compounds is one of the most challenging tasks in the metabolomics workflow. Accurate MS/MS data from metabolites can help to reduce the number of possible sum formulae. The identification with SmartFormula 3D is exemplified for Myxalamid A: based on sum formulae calculated for precursor and product ions it is possible to limit the number of candidates and achieve an assignment of formulae to fragments.
Fig. 4: A: PCA analysis of M. xanthus strains with scores and loadings plots PC1 vs PC2 (expl. variance 48%); B: bucket statistics plot of selected loadings visualizing the relative production of two known and one unknown metabolite.
O 1380O 2004O 2152O 4125O DK1622
Scores Plot Loadings PlotA
0 5 10 15 20 25 30 35 400.0
0.2
0.4
0.6
0.8
1.0 Bucket: 2.6min : 1011.53m/z
Analysis
Inte
nsity
x 1
05
0 5 10 15 20 25 30 35 40
Bucket: 7.5min : 416.32m/z
Analysis
1.0
1.5
2.0
2.5
3.0
Inte
nsity
x 1
05
0 5 10 15 20 25 30 35 40
Bucket: 4.8min : 549.27m/z
Analysis
0
1
2
3
4
Inte
nsity
x 1
04
-1.5 -1.0 -0.5 0.0 0.5 1.0 PC 1
-0.5
0.0
0.5
1.0
PC 2
-1.5 -1.0 -0.5 0.0 0.5 1.0 PC 1
-0.5
0.0
0.5
1.0
PC 2
Novel Metabolite DKXanthen-548Myxalamid A
B
Technical replicates
127.
1120
187.
1481
215.
1431
267.
1741
290.
2111
323.
2366
398.
3047
0.00
0.25
0.50
0.75
1.00
1.25
x104
100 150 200 250 300 350 400 450 500m/z
208.
1333
416.
3155 fragment sum formulae of
the selected suggestion
MS/MS-peaklist
sum formulae of precursor ion combining fullscan and MS/MS information
MS peaklist(precursor region only)
Fig. 5: SmartFormula3D result for Myxalamid A with MS/MS spectrum as insert.
Conclusions
Bruker Daltonics
Characterization of black and green tea using ESI-Q-TOF-MS and data evaluation by principle component analysisAiko Barsch, Gabriela Zurek, Wiebke LohmannBruker Daltonik GmbH, Fahrenheitstraße 4, 28359 Bremen, Germany
IntroductionBlack and green tea account for more than 95% of the total tea consumption around the world. For both types of teas possible health benefits have been hypothesised. Possible anti-carcinogenic, anti-inflammatory, anti-thrombic, antiviral and anti-bacterial effects have been described, although the protective actions remain to be thoroughly proven. In order gain a deeper understanding for these possible health promoting effects, as well as to improve the quality and taste, characterisation of tea is a topic of interest in academia and a large market in industry. In our study, we used high resolution electrospray time-of-flight mass spectrometry to study different teas using statistical methods.
Methods
HPLC: UPLC (Waters) Column: 50 x 2.0 mm Acquity BEH C18 1.7 µmMobile phase: (A) H2O + 0.1% HCOOH (B)ACN + 0.1% HCOOH Gradient: 0 min 2% B; 1 min 2% B; 5 min 30% B; 5.5 min 95% B; 6 min 95% B; 6.5 min 2% B; 8 min 2% B Flow rate: 0.35 mL/minInjection: 1 µL (in triplicate)Sample: 6 different commercial tea samples (1 Green, 2 Darjeeling, 2 “Black English”, 1 “Black English” without caffeine) were extracted with 100 mL hot water for 5 min and subsequently centrifuged for 10min @ 13000 rpm. Samples were injected without dilutionMS: micrOTOF-Q II (ESI-Q-TOF MS, Bruker Daltonik GmbH). Ionization: ESI(+) Scan range: m/z 75-1000. Scan mode: Fullscan, autoMSMSData Evaluation: ProfileAnalysis 2.0TM; All relevant information was extracted from the raw data by applying the “Find Molecular Features” algorithm. This algorithm efficiently differentiates real signal from background noise.
• LC-ESI-Qq-TOF-MS analysis in combination with PCA was applied to differentiate green, Darjeeling and black tea samples
• SmartFormula3D (combining exact mass and isotopic pattern information in MS and MS/MS spectra) allowed for sum formula generation of relevant compounds differentiating samples
• Removing caffeine for PCA calculation revealed a separation of the samples according to oxidation grade obtained during tea processing
• Compounds with higher abundance in black tea could be shown to belong to the class of flavonoids
•Green tea samples contained higher amounts of catechins compared to black teas
Results• All relevant “features” were extracted from the LC-MS data of the different tea samples by applying the “Find Molecular Features” peak detection algorithm. The PCA scores plot based on the extracted information revealed a clear separation of the “decaffeinated” black tea from all other tea samples (Fig. 1 A). Visualizing the peak intensities for the loading which is mainly contributing to this difference (bucket statistics plot) highlights the absence of a peak with m/z 195 in decaffeinated tea. Sum formula generation for this compound (Fig. 1 C and D) based on exact mass and isotopic pattern information followed by a database query could prove that the expected absence of caffeine in decaffeinated tea constitutes the main difference to the other tea samples.
Fig. 1: PCA of 6 different tea samples. The “decaffeinated” black tea sample is clearly separated in the scores plot. The loading which is mainly responsible for this difference could be identified as caffeine.
For research use only. Not for use in diagnostic procedures.
ESI-Qq-TOF-MS
-0.2 0.0 0.2 0.4 0.6 PC 1
-0.15
-0.10
-0.05
0.00
0.05
0.10
PC 2
-0.2 0.0 0.2 0.4 0.6 PC 1
-0.15
-0.10
-0.05
0.00
0.05
0.10
PC 2
A: PCA Scores & Loadings Plot of all tea samples (explained variance PC1 vs PC2: 86.6% + 8.3%)
Scores Loadings
3.0min;195.09m/z
Bucket: 3.0min : 195.09m/z
5 10 Analysis0
1
2
3
4
55x10
Int.
B: Bucket Statistics –intensity of selected loading displayed for all samples
C: SmartFormula result for selected loading
D: Query in Public databases via Compound Crawler:Caffeine!
-1.0 -0.5 0.0 0.5 PC 1
-0.50
-0.25
0.00
0.25
0.50
PC 2
-1.0 -0.5 0.0 0.5 PC 1
-0.50
-0.25
0.00
0.25
0.50
PC 2Scores Loadings
Green
Darjeeling
“Black English”
4.0min; 611.16m/z
4.4min; 595.16m/z
4.3min; 449.11m/z
decaffeinated black
139.0391
289.0703
C 7 H 6 O 5 + 0.00058
C 7 H 4 O 4 + 0.00145
C 15 H 12 O 8 + 0.00107
C 14 H 12 O 8 + 0.00061
7. +MS2(459.0923), 11.5929-17.3893eV, 3.73-3.83min #650-#667
0
2000
4000
6000
Intens.
150 200 250 300 350 400 450 m/z
MS peak list
MS/MS peak list
Fragment formulae
only one sum formula suggestion
remains for the precursor ion
Bucket Statistics SmartFormula Results Putative Structure
Fig. 3: A: Scores and loadings plot highlighting loadings differentiating green and black tea. B: Bucket statistics, Smart Formula and putative structure for 4 selected loadings. C:SmartFormula3D exemplified on putative epigallocatechin gallate; combining exact mass and isotopic pattern information of MS and MS/MS spectra only one sum formula suggestion remains for the compound. D: MS/MS spectrum of putative epigallocatechin gallate with annotated sum formulae for neutral losses.
B:
D:
C:
decaffeinated black
• Removing caffeine as well as theobromine and theanine peaks for PCA calculation revealed a clear separation of green, Darjeeling and black teas along PC1 (Figure 2). This clustering is in accordance to the different oxidation grades obtained during tea fermentation: green tea < Darjeeling tea < black tea
Fig. 2: PCA Scores & Loadings Plot of all tea samples after removing caffeine, theobromine and theanine for model calculation (explained variance PC1 vs PC2, 52.7% + 27.8%). Selected loadings contributing to the difference between green tea and black tea are highlighted by RT; m/z
•Investigating loadings mainly contributing to the difference between black and green tea revealed several compounds to be higher abundant in black tea compared to green or Darjeeling tea (highlighted in Fig. 2). A compound with 611m/zhad a very low intensity in Darjeeling tea “2” and was much higher abundant in the decaffeinated black tea (Fig. 3 A). The sum formula generated for this compound (C27H31O16) could possibly correspond to rutin (Fig. 3 B+C - verification based on standard in progress). For several further compounds which had a higher intensity particularly in decaffeinated black tea samples the corresponding sum formulae were calculated (Fig 3. D). A query in public databases indicated that these compounds most likely belong to the class of flavonoids.
-1.0 -0.5 0.0 0.5 PC 1
-0.50
-0.25
0.00
0.25
0.50
PC 2
-1.0 -0.5 0.0 0.5 PC 1
-0.50
-0.25
0.00
0.25
0.50
PC 2Scores Loadings
Green
Darjeeling
“Black English”
2.8min; 307.08m/z
3.5min; 459.09m/z
3.4min; 291.09m/z
4.1min; 443.10m/z
4.1min; 465.10m/z
A:
B: C:
D:Bucket Exact m/z
measuredFormula [M+H]+
4.2min-595.16m/z 595.1646 C27H31O15
4.1min- 465.10m/z 465.1021 C21H21O12
4.3min- 449.11m/z 449.1075 C21H21O11
Fig. 3: A: Bucket statistics plot for 4.1 min; 611.16 m/z. B:SmartFormula result. C: CompoundCrawler database query for best sum formula hit. D: Summary of molecular formulae highlighted in Fig.2
A:
Bucket: 4.0min : 611.16m/z
2.5 5.0 7.5 10.0 12.5 15.0 Analysis
1
2
3
4
4x10Int. 4.0min; 611.16m/z
•Several loadings had a higher abundance in green tea (and Darjeeling) in comparison to the black teas (Fig. 4 A+B). Calculating the sum formulae for these compounds in combination with a query in public databases (like ChemSpider) via the CompoundCrawler suggested several catechins as likely structural candidates for these compounds. Catechin derivatives are known to be higher abundant in green tea compared to black tea.
•By combining exact mass and isotopic pattern information in MS and MS/MS spectra SmartFormula3D could reduce the list of possible sum formulae from 5 to 1 for compound 3.5 min; 459.09 m/z (Fig. 4 C). The possible structure of epigallocatechin-gallate for the calculated sum formula (C22H18O11) could be fortified by evaluating the sum formulae corresponding to neutral losses in the MS/MS spectrum (Fig. 4 D).
Bucket: 3.5min : 459.09m/z
2 4 6 8 10 12 14 16 Analysis
1
2
3
4
4x10Int.
3.5min; 459.09m/z O
O
OH
OH
OH
OH
O
OH
OH
OH
OH
(-)-epigallocatechin gallate
Bucket: 4.1min : 443.10m/z
2 4 6 8 10 12 14 16 Analysis
1
2
3
4
4x10Int.
4.1min; 442.10m/z
(-)-epicatechin gallate
O
O
OH
OH
OH
OH
O
OH
OH
OH
Bucket: 3.4min : 291.09m/z
2 4 6 8 10 12 14 16 Analysis0
2
4
4x10Int.
3.4min; 291.09m/z
(-)-epicatechin
O
OH
OH
OH
OH
OH
Bucket: 2.8min : 307.08m/z
2 4 6 8 10 12 14 16 Analysis
0
2
4
4x10Int.
2.8min; 307.08m/z
(-)-epigallocatechin
O
OH
OH
OH
OH
OH
OH
-10- -11-
Bacterial Metabolomics Food & Nutrition
Metabolomics Posterbook 2011
Conclusions
Bruker Daltonics
ESI FT-ICR MS
Analysis of Whisky by Electrospray FT-ICR Mass Spectrometry: Proof of Origin by Statistical MethodsMatthias Witt, Rainer Paape,Jens Fuchser and Jochen Friedrich
Bruker Daltonik GmbH, Bremen, Germany
IntroductionWhisky is a high-class consumed alcoholic beverage witha several billion dollar market. Due to the high value ofthis liquor counterfeiting and manipulation have beenobserved. Therefore, the proof of the origin of this luxuryalcoholic drink is of major interest of distillers andbeverage importers. Whisky consists beside water andalcohol of a variety of volatile and non-volatile chemicalcomponents, e.g. organic acids and esters, aldehydes,phenols, polyphenols and lactones. [1] Several massspectrometry techniques have been applied to analyzevolatile compounds in whisky [2,3]. Recently the proof oforigin and authenticity of whisky has been studied byelectrospray mass spectrometry [4].
[1] Fujieda, M., Tanaka, T., Suwa, Y., Koshimizu, S., Kuono, I., J. Agric. Food Chem. 2008, 56, 7305.[2] Kinton, V. R., Adam, L., Abstr. Pap. Am. Chem. Soc. 2002, 224, U90.[3] Fitzgerald, G., James, K. J., MayNamara, K, Stack, M. A., J. Chromatogr. A 2000, 896, 351. [4] Møller, J. K. S., Catharino, R. R., Eberlin, M. N., Analyst 2005, 130, 890.
MethodsIn our study we used ultra-high resolved massspectrometry to study whiskies from different originsusing statistical methods and fingerprinting. The whiskieshave been diluted 1:20 in 50% MeOH for direct infusionmeasurements using electrospay FT-ICR in negative ionmode. The spectra have been acquired with the newBruker solariX FTMS platform equipped with a 12Tmagnet using a resolving power of 300.000 at m/z 400.Three repetitive measurements have been performed ofeach whisky to proof the concept of the principalcomponent analysis (PCA) for this kind of samples.
ESI-FT-ICR-mass spectrometry has beenproven as a powerful tool for thecharacterization of complex mixtures suchas whisky.
Figure 2. PCA score plots of whisky mass spectra: Analysis of a) full mass spectrum with 1 Da buckets, b) full mass spectrum with 2mDa buckets using fine structure, c) mass range m/z 310-330 with 1 Da buckets and d) mass range m/z 310-330 with 2 mDa bucketsusing fine structure (origin market with different colours; red: Whisky from Spey, green: japanese Whisky, blue: Whisky from Highland;repetitive measurements have the same symbol)
Figure 1. Electrospray mass spectrum of a Highland whisky innegative ion mode: a) full mass spectrum, b) part of spectrum toshow the fine structure, mass resolution and peak annotationincluding molecular formula and mass errors.
Summary The mass spectra of whisky are very complex even if
only polar compounds are detected in electrospraynegative ion mode.
Using negative ion mode mainly CxHyOz compoundshave been detected.
Specific compounds can be allocated in whisky likeellagic acid or gluconic acid with accurate massmeasurement.
Using the fine structure of a mass spectrum areamainly consisting of humic compounds, the origin ofunknown whiskys can be allocated using bothstatistical methods, PCA or cluster analysis.
ResultsSeveral Scottish whiskies from two different origins(Highland and Spey) as well as several whiskies from thejapanese distillery Suntory have been analyzed. Usingelectrospray ionization the most polar components ofwhisky are detected. Ellagic acid, a polyphenol present infruits and vegetables with antiproliferative and antioxidantproperties, and gluconic acid have been found as themost dominant species (Figure 1a). Beside majorcomponents, the mass spectra of whisky show a complexpattern with several peaks at one nominal mass resultingin several thousand peaks in a mass spectrum (Figure1b). The molecular formulae of more than thousandcompounds have been identified.
315.01464
315.07216
315.10855
315.18134
315.25410
314.95 315.00 315.05 315.10 315.15 315.20 315.25 m/z0.0
0.5
1.0
1.5
2.0
8x10Intens.
ESI_neg_Sample_28_000001.d: -isCID MS
Whisky Highland 28
C14H3O9-
0.0 ppm
C15H7O8-
0.0 ppm
C13H15O9-
0.1 ppm
C14H19O8-
0.1 ppm
C16H11O7-
0.1 ppm
C17H15O6-
0.1 ppm
C16H27O6-
0.1 ppm
C18H35O4-
0.1 ppm
191.05611
300.99899
439.12460
200 300 400 500 600 700 m/z0
1
2
3
49x10
Intens.
ESI_neg_Sample_28_000001.d: -isCID MS
Whisky Highland 28
C14H5O8-
0.0 ppm
Ellagic acida)
b)
-0.6 -0.4 -0.2 0.0 0.2 0.4 PC 1
-0.2
0.0
0.2
PC 2
-0.6 -0.4 -0.2 0.0 0.2 0.4 PC 1
-0.2
0.0
0.2
PC 2
-0.20 -0.10 0.00 0.10 0.20 PC 1
-0.10
-0.05
0.00
0.05
0.10
PC 2
-0.20 -0.10 0.00 0.10 0.20 PC 1
-0.10
-0.05
0.00
0.05
0.10
PC 2
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 PC 1
-0.3
-0.2
-0.1
0.0
0.1
0.2
PC 2
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 PC 1
-0.3
-0.2
-0.1
0.0
0.1
0.2
PC 2
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 PC 1-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
PC 2
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 PC 1-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
PC 2
a)
d)c)
b)
Figure 3. Cluster analysis: a) whiskies of the same origin havebeen correctly clustered using 2 mDa buckets in the mass rangem/z 310-330. b) the three different origins of whiskies have beenclustered using cluster analysis with the full mass spectrum andfine structure (2 mDa buckets)
PCA (Figure 2) as well as cluster analysis (Figure 3)have been carried out of the full and of a part of themass spectra (m/z 310-330) with and without using thefine structure to validate the origin of the studiedwhiskies and to probe the relevance of the highlyresolved mass spectra for the characterization of whisky.Whiskies of the same origin are in the same area of thePCA score plot and grouped in the cluster analysisusing a mass spectrum area mainly of humiccompounds. The fine structure improves the correctclustering based on the origin of the samples. Repetitivemeasurements in the PCA plot (identical colour andsymbol) show the reproducibility of the measurements.
Hig
hlan
d_M
31_0
1.d
Hig
hlan
d_M
32_0
1.d
Hig
hlan
d_M
28_0
1.d
Hig
hlan
d_M
34_0
1.d
Hig
hlan
d_M
27_0
1.d
Hig
hlan
d_M
29_0
1.d
Hig
hlan
d_M
30_0
1.d
Japa
n_M
21_0
1.d
Japa
n_M
20_0
1.d
Japa
n_M
82_0
1.d
Spe
y_M
39_0
1.d
Spe
y_M
38_0
1.d
Spe
y_M
40_0
1.d
0.00
0.05
0.10
0.15
0.20
0.25
Hei
ght
92859375
9495
62 97
7890
84
81556814
1761
48 57
3566
62
a)
Hig
hlan
d_M
28
Hig
hlan
d_M
34
Hig
hlan
d_M
30
Hig
hlan
d_M
32
Hig
hlan
d_M
31
Spe
y_M
38
Spe
y_M
39
Hig
hlan
d_M
27
Spe
y_M
40
Hig
hlan
d_M
29
Japa
n_M
82
Japa
n_M
20
Japa
n_M
21
0.02
0.06
0.10
0.14
Hei
ght
89
7493
87
90 65
67 92
99
72
76
85
7431
33
53 59
25 27
25
19
12b)
Mass range m/z 150-700 low resolution: 1 Da buckets
Mass range m/z 150-700 high resolution: 2 mDa buckets
Mass range m/z 310-330 low resolution: 1 Da buckets
Mass range m/z 310-330 high resolution: 2 mDa buckets
For research use only. Not for use in diagnostic procedures.
Conclusions
Bruker BioSpin & Bruker Daltonics
The fruit juice industry purchases its raw materials in a world-wide market. Therefore, the control of authenticity and legal conformity is of outmost importance. In this context the relatively large natural variability makes it difficult to differentiate from adulteration, e.g. the addition of sugar, water, flavor components or cheaper fruit types than those declared on the label. Consequently, there is a significant interest in analytical high throughput techniques for quality and authenticity control. Especially the combination of spectroscopic techniques like NMR and LC-MS with multivariate statistical evaluation of the data shows great potential to accomplish this task. The Metabolic ProfilerTM
combines automatic acquisition of NMR and LC-MS data of a large number of samples in a fully automated system (Fig. 1). Here, it is demonstrated that apple juice samples from different origin can be easily distinguished and that analytes responsible for the discrimination can be identified using multivariate statistical methods.
M. Godejohann1; E. Humpfer1; H. Schaefer1; P. Rinke2; G. Zurek3; Carsten Baessmann3; M. Spraul11Bruker BioSpin GmbH, Rheinstetten, Germany; 2SGF International, Nieder-Olm, Germany ; 3Bruker Daltonik GmbH, Bremen, Germany
LC-MS and NMR provide a complementary set of data: While LC-ESI-MS shows best results for medium polar, medium weight molecules at lowest concentrations, the NMR is able to see extremely polar, small and medium size molecules even in complex mixtures which would elute at the void volume of a regular RP chromatography. There, this results in suppression of electrospray ionization caused by coelution with inorganic salts and high concentrated polar components.
Reducing variables for statistical analysisFor both, LC-MS and NMR all entries in the bucket table can be screened on base of t-tests and by a PCA-DA approach to look for the most discriminating m/z and ppm variables for the two group problem of Chinese and Polish apple juices.
Most discriminating variables for LC-MSFor LC-MS, 55 most discriminating (m/z –intensity – retention time) variables were extracted and used to run a principle component analysis as shown in Figure 2. Molecular formulae can be readily calculated from these variables as shown for the most discriminating loading with m/z 435.1312. Another representation of this result is shown on bottom of Figure 2. Here, the intensity for each discriminating mass peak for each sample is colour coded, such that red indicates the highest intensity and blue the lowest. This clearly shows that Chinese apple juices are richer in early eluting low molecular size components while Polish juices are distinguished by higher concentrations of late eluting analytes in the higher mass range. The only German juice compares more to the Polish juices. In the PCA scores plot, this juice labeled with number 61 clearly separates from all others.
Combined statistical approachA combined LC-MS-NMR correlation matrix evaluating only m/z values, retention times and NMR-ppm-shifts with a correlation factor larger than 0.75 was created. An m/z of 435.1312 can be extracted as most discriminating vector between juice from China and Poland (Figure 4). After calculation of the combined covariance matrix, variables for the NMR buckets correlating to this mass can be extracted. In comparison to the reference data base and with the molecular formula generated from the exact mass data, this component can be identified as phloridizin. This study nicely demonstrates the strength of a combined LC-MS and NMR platform in food screening as both quality control issues and identification can be addressed simultaneously.
Juices from different geographical origin were analysed by a combined LC-MS and NMR approach. The statistical analysis directly revealed discriminating analytes typical for each origin. Furthermore, a combined statistical approach resulted in identification of Phloridzin as being one of the main discriminators.
Fig. 1: The Metabolic ProfilerTM combines NMR and LC-MS data acquisition in a self-contained, fully automated system
• The combined acquisition of LC-MS and NMR data for a large set of similar samples together with a multivariate statistical analysis allows the easy differentiation of juices from different origin. Moreover, it can also accomplish the direct identification of discriminating analytes without laborious isolation and structural elucidation based on statistical results.
Introduction
Summary
Results
Metabolic Profiling and Fingerprinting of Beverages using LC-MS and NMR
LC-MS and NMR
MethodsApple juices: 60 original re-diluted juice concentrates from Poland and China and one German Market juice sample were used for combined LC-MS and NMR analysisSample preparation: re-diluted juices were automatically prepared from one source into two separate 96-well plates.NMR prep.: addition of 10% D2O buffer, pH 3.LC-MS prep.: 2-fold dilution with H2O. Data acquisition: Data were acquired on a Metabolic ProfilerTM:LC-MS on ESI-TOF micrOTOF mass spectrometer (Bruker Daltonik, Bremen, Germany) in negative ionisation mode coupled to an HPLC system from Agilent (Waldbronn, Germany). LC separation was carried out on a short reversed phase column with a 10 min solvent gradient. Sodium formate solution was injected as calibrant at the start of each chromatographic run.NMR on Bruker AVANCE 600 NMR Spectrometer equipped with a 5mm TCI cryo-probe converted to a flow probe using a 120 µl cryofit employing flow injection technology (BEST-NMR, Bruker BioSpin, Rheinstetten, Germany. T=300K).Statistical evaluation: LC-MS: advanced bucketing method in AMIX; Mass range: m/z = 50.5-800.5; m=0.01; Time range from 0.5-10min; NMR bucketing in AMIX: rectangular bucketing form 12 to 0.1 ppm; bucket width 0.04 ppm; exclusion range 4.9-4.5 ppm. Statistical plots were created from Matlab 7.0.
1
2
3
Instrument status
Instrument status
Instrument status
1
2
3
Instrument status
Instrument status
Instrument status
Fig. 2: Result for the PCA calculation (top) and intensity distribution for LC-TOF-MS data (bottom) for most discriminating mass buckets sorted according to their importance with regard to the discriminating vector. The sum formula calculation for the discriminating m/z 435.1312 results in C21H23O10- mass deviation -3.5 ppm.
Most discriminating variables from NMRThe same statistical analysis approach was performed for the NMR data (Figure 3).The intensity distribution clearly indicates that Chinese juices show higher concentrations in the sugar region, while Polish juices are richer in components showing aromatic signals and especially malic acid showing a signal at 4.2 ppm.
Fig. 4: A: Slice taken from combined covariance matrix for m/z 435.1312; B: The aromatic region from the original NMR spectra of a Chinese and Polish juice together with the library spectrum of Phloridzin shows good agreement to the correlations on top.
Fig. 3: Intensity distribution of NMR buckets which are most discriminating between Chinese and Polish samples. The German market sample exhibits features common in both juices.
For research use only. Not for use in diagnostic procedures.
-12- -13-
Food & Nutrition Food & Nutrition
Bru
ker
Dalt
on
ics
Iden
tifi
cati
on
of
Nu
cleo
sid
es
as
Po
ten
tial
Bio
mark
ers
fo
r B
reast
Can
cer
in U
rin
e b
y E
SI-
TO
F-M
SThom
as Z
ey1, M
atth
ias
Pelz
ing
1,
Din
o B
ulli
nger
2,
Antj
e Fr
icke
nsc
hm
idt2
, G
abriel
a Zure
k1,
S. La
ufe
r2,
Ber
nd K
amm
erer
2
1 B
ruke
r D
alto
nik
Gm
bH
, Fa
hre
nhei
tstr
. 4,
28359 B
rem
en,
Ger
man
y; 2
Univ
ersi
ty o
f Tueb
ingen
, In
stitute
of Ph
arm
acy,
Auf der
Morg
enst
elle
8,
72076 T
ueb
ingen
, G
erm
any
1 A
ffin
ity C
hro
mato
gra
ph
y
The
affin
ity c
hrom
atog
raph
y is
bas
ed o
n th
e se
lect
ive
and
reve
rsib
le b
indi
ng o
f cis
-dio
ls to
phe
nylb
oron
ic a
cid
imm
obili
zed
on
a po
lyac
ryla
mid
e st
atio
nary
pha
se. E
lutio
n is
car
ried
out u
sing
0,
1M fo
rmic
aci
d in
MeO
H/H
2O 1
:1 (v
/v).
Intr
od
uct
ion
RN
A, p
artic
ular
y t-R
NA
, con
tain
s a
num
ber o
f mod
ified
nuc
leos
ides
in
add
ition
to th
e re
gula
r rib
onuc
leos
ides
ade
nosi
ne, g
uano
sine
, ur
idin
e an
d cy
tidin
e. M
odifi
catio
ns li
ke m
ethy
latio
n, h
ydro
xyla
tion,
S
/O-s
ubst
itutio
n an
d fo
rmat
ion
of c
ompl
ex s
ide
chai
ns a
re ta
king
pl
ace
post
-tran
scrip
tiona
lly w
ithin
the
poly
nucl
eotid
e m
olec
ule
by
vario
us m
odify
ing
enzy
me
syst
ems
such
as
met
hyltr
anfe
rase
s an
d lig
ases
. Dur
ing
RN
A tu
rnov
er, f
ree
norm
al a
nd m
odifi
ed
nucl
eosi
des
are
form
ed b
y hy
drol
ytic
act
ion
of ri
bonu
clea
ses
and
phos
phat
e-el
imin
atio
n by
pho
spha
tase
s.
In c
ontra
st to
unm
odifi
ed n
ucle
osid
es, m
any
mod
ified
one
s ar
e bi
oche
mic
al e
nd p
rodu
cts
and
cann
ot b
e re
utili
zed
for R
NA
sal
vage
pa
thw
ay. T
oget
her w
ith s
mal
l am
ount
s of
regu
lar r
ibon
ucle
osid
es,
the
mod
ified
spe
cies
are
circ
ulat
ing
in th
e bl
ood
stre
am p
rior t
o ur
inar
y ex
cret
ion.
Esp
ecia
lly m
odifi
ed n
ucle
osid
es a
re e
leva
ted
in c
ance
r pat
ient
s`
urin
e du
e to
the
up-re
gula
ted
met
abol
ism
and
cel
l gro
wth
in tu
mor
tis
sue
and
thus
incr
ease
d R
NA
turn
over
. The
se o
bser
vatio
ns h
ave
form
ed th
e ba
sis
for f
urth
er in
vest
igat
ions
of n
ucle
osid
es a
s po
tent
ial t
umor
mar
kers
in c
ance
r dia
gnos
is.
The
mic
rOTO
F E
SI-T
OF
has
prov
en a
n in
disp
ensa
ble
tool
for t
he
iden
tific
atio
n of
unk
now
ns b
ased
on
accu
rate
mas
s da
ta a
nd th
e tru
e is
otop
ic p
atte
rn. N
ovel
TO
F-te
chno
logy
has
lead
to th
e co
nser
vatio
n of
isot
ope
ratio
s an
d ov
erco
me
the
past
dyn
amic
ra
nge
limit
of th
is te
chno
logy
. Thi
s is
esp
ecia
lly im
porta
nt fo
r the
id
entif
icat
ion
from
bio
logi
cal m
atric
es a
s sh
own
in th
e ex
ampl
e fo
r nu
cleo
side
s fro
m u
rine.
Exp
eri
men
tal
A s
ensi
tive
met
hod
for t
he a
naly
sis
of u
rinar
y rib
onuc
leos
ides
in
clud
ing
affin
ity c
hrom
atog
raph
y pu
rific
atio
n (A
ffiG
el60
1, B
ioR
ad,
Mun
ich)
prio
r to
LC-M
S a
naly
sis
usin
g a
mic
rOTO
FTME
SI-o
a-TO
F-M
S s
yste
m (B
ruke
r Dal
toni
k, B
rem
en) h
as b
een
deve
lope
d.
Rev
erse
d ph
ase
chro
mat
ogra
phic
sep
arat
ion
of th
e pr
epur
ified
nu
cleo
side
s w
as p
erfo
rmed
usi
ng a
n A
gile
nt 1
100
bina
ry g
radi
ent
pum
p H
PLC
sys
tem
with
pho
todi
ode
arra
y de
tect
ion
(Agi
lent
Te
chno
logi
es, W
aldb
ronn
) and
a X
Terra
C18
col
umn
(100
x2m
m,
3µm
par
ticle
s; W
ater
s).
Dat
a ac
quis
ition
was
con
trolle
d by
Com
pass
1.0
for m
icrO
TOF.
Th
e m
ass
spec
trom
eter
was
ope
rate
d in
ES
I pos
itive
mod
e in
a
scan
rang
e fro
m 5
0-11
00 m
/z ;
as c
alib
ratio
n st
anda
rd s
odiu
m
form
ate
solu
tion
was
inje
cted
with
in th
e vo
id v
olum
e of
eac
h ch
rom
atog
raph
ic ru
n.
Dat
a ev
alua
tion
was
per
form
ed u
sing
the
‘Gen
erat
e M
olec
ular
Fo
rmul
a (G
MF)
TM’ s
oftw
are
suite
with
in D
ataA
naly
sis
3.2
eval
uatin
g bo
th a
ccur
ate
mas
s po
sitio
n an
d tru
e is
otop
ic p
atte
rns
for t
he c
alcu
latio
n of
mol
ecul
ar fo
rmul
ae.
Bas
ed o
n m
olec
ular
form
ula
sugg
estio
n an
d ch
emic
al in
form
atio
n,a
data
base
sea
rch
(Sci
Find
er) w
as p
erfo
rmed
resu
lting
in d
iffer
ent
stru
ctur
e pr
opos
als.
The
sea
rch
was
rest
ricte
d to
ribo
syl-c
onta
inin
g st
ruct
ures
con
side
ring
othe
r kno
wn
fragm
enta
tion
moi
etie
s su
ch a
s ca
rbox
ylic
aci
d. T
he m
ass
spec
trom
etric
resu
lts o
btai
ned
by E
SI-
oa-T
OF-
MS
mea
sure
men
ts w
ere
com
pare
d w
ith o
ur re
sults
(c
hara
cter
istic
frag
men
tatio
n pa
ttern
s) fr
om L
C-E
SI I
on T
rap
MS
, ta
ndem
mas
s sp
ectro
met
ry a
nd P
SD
-MA
LDI-T
OF-
MS
. Fi
ve y
et u
nkno
wn
nucl
eosi
des
in u
rine
coul
d be
ver
ified
(Tab
le 1
, N
o. 1
-5).
Furth
erm
ore,
urin
ary
com
poun
ds o
f diff
eren
t orig
in, e
.g.
hist
amin
e-m
etab
olite
s (N
o. 6
), co
uld
easi
ly b
e di
stin
guis
hed
from
R
NA
-met
abol
ites.
Con
fiden
ce in
to th
e pr
esen
ted
met
hod
was
pr
oven
by
iden
tifyi
ng m
odifi
ed u
rinar
y nu
cleo
side
s (N
o. 7
-12)
kn
own
to b
e pr
esen
t in
urin
e.Th
is p
roof
-of-c
once
pt s
tudy
sho
ws
that
it is
pos
sibl
e to
gen
erat
e st
ruct
ural
hyp
othe
sis
for u
rinar
y nu
cleo
side
s ba
sed
on a
ccur
ate
mas
s an
d tru
e is
otop
ic p
atte
rn in
com
bina
tion
with
dat
abas
e se
arch
.
Ref
eren
ces
[1] D
udle
y, E
., N
ewto
n, R
.P.,
Rap
id C
omm
. in
Mas
s S
pect
rom
etry
, 14
(200
0).
2] B
ergm
ark,
J.,
Gra
neru
s, G
., S
cand
. J. c
lin. L
ab. I
nves
t., 3
4 (1
974)
[3] K
amm
erer
, B.,
Fric
kens
chm
idt,
A.,
Mal
di-T
OF-
MS
ana
lysi
s of
urin
ary
nucl
eosi
des,
JA
SM
S, a
ccep
ted
(200
5)
3 C
hara
cteri
stic
Neu
tral Lo
ss
The
char
acte
ristic
dec
ay o
f the
nuc
leos
ides
into
a n
eutra
l sug
ar
moi
ety
(m/z
= 1
32) a
nd th
e co
rresp
ondi
ng n
ucle
ic b
ase
fragm
ent i
s ob
serv
ed u
nder
col
lisio
n in
duce
d di
ssoc
iatio
n (C
ID) c
ondi
tions
. Fi
gure
2 d
epic
ts th
e m
ass
spec
trum
of g
uano
sine
incl
udin
g th
e ne
utra
l los
s. T
he n
eutra
l los
s of
132
m/z
repr
esen
ts a
sec
ond
inde
pend
ent i
dent
ifica
tion
crite
rion
for n
ucle
osid
es b
esid
es a
ffini
ty
chro
mat
ogra
phy.
Furth
erm
ore,
132
-neu
tral l
oss
of th
e se
lect
ed c
ompo
unds
was
co
nfirm
ed b
y M
Sn -
expe
rimen
ts (E
SI-H
CT-
MS
, tan
dem
mas
s sp
ectro
met
ry).
4 G
en
era
te M
ole
cula
r Fo
rmu
la
Ele
men
tal c
ompo
sitio
ns o
f nuc
leos
ides
wer
e su
gges
ted
usin
g th
e G
ener
ateM
olec
ular
Form
ula(
GM
F)-e
dito
r whi
ch e
valu
ates
the
form
ula
sugg
estio
n us
ing
both
acc
urat
e m
olec
ular
mas
s an
d is
otop
ic p
atte
rn m
atch
. To
impr
ove
conf
iden
ce in
the
resu
lts, t
he
mea
sure
d is
otop
ic p
atte
rn is
com
pare
d w
ith th
e th
eore
tical
one
, ge
nera
ted
from
the
sugg
este
d el
emen
tal c
ompo
sitio
n (s
igm
a va
lue)
. The
sm
alle
r the
sig
ma
valu
e th
e be
tter t
he fi
t. Fi
gure
3
show
s th
e m
olec
ular
form
ula
calc
ulat
ed fr
om th
e m
ass
spec
trum
of
1-m
ethy
lgua
nosi
ne. T
able
1 c
onta
ins
a se
lect
ion
of id
entif
ied
nucl
eosi
des
from
urin
e in
clud
ing
the
mas
s er
ror a
nd s
igm
a va
lue.
Figu
re 2
: in-
sour
ce fr
agm
enta
tion,
neu
tral l
oss
of
m/z
= 1
32 fo
r gu
anos
ine.
No.
Mea
sure
dM
ass
(M+H
)+
Gen
erat
ed
Mol
ecul
ar F
orm
ula
(M+H
)+
ppm
er
ror
Sigm
aVa
lue
SciF
inde
r Sea
rch
Res
ult
134
6,12
59C
13H
20N
3O8
-3,9
300,
0038
3-(3
-am
ino-
3-ca
rbox
y-pr
opyl
)-ur
idin
e
229
8.11
54C
11H
16N
5O5
-2.6
850.
0556
7-m
ethy
lgua
nosi
ne
332
8.10
67C
12H
18N
5O4S
2.00
90.
0475
2-m
ethy
lthio
-N6-
met
hyla
deno
sine
438
4.11
71C
14H
18N
5O8
-5.3
740.
0039
N6-
succ
inyl
aden
osin
e
529
6.13
58C
12H
18N
5O4
-1.4
730.
0044
1,N
6-di
met
hyla
deno
sine
625
9.09
25C
10H
15N
2O6
-0.3
290.
0072
1--
D-r
ibof
uran
osyl
-1H
-imid
azol
e-4-
acet
ic
acid
7
286.
1036
C11
H16
N3O
6-1
.003
0.00
42N
4-ac
etyl
cytid
ine
824
7.09
26C
9H15
N2O
6-0
.651
0.
0026
dihy
drou
ridin
e
941
3.14
37C
15H
21N
6O8
-5.1
730.
0064
N6-
carb
amoy
lthre
onyl
aden
osin
e
1028
2.11
90
C11
H16
N5O
4-0
.752
0.00
511-
met
hyla
deno
sine
1131
2.13
09C
12H
18N
5O5
-2.1
900.
0086
N2,
N2-
dim
ethy
lgua
nosi
ne
1228
3.10
41C
11H
15N
4O5
-1.5
370.
0068
1-m
ethy
linos
ine
2 C
hro
mato
gra
ph
ic S
ep
ara
tio
n
A m
ixtu
re o
f 20
know
n nu
cleo
side
s (c
once
ntra
tion
from
0.8
to 6
40
nmol
/ml >
bas
ed o
n ur
ine
sam
ple
valu
es) a
s w
ell a
s pu
rifie
d re
al
urin
e sa
mpl
es w
ere
sepa
rate
d w
ith re
vers
ed p
hase
ch
rom
atog
raph
y. T
he s
epar
atio
n sh
own
in F
ig. 1
was
ach
ieve
d w
ith a
bin
ary
acet
onitr
ile-w
ater
gra
dien
t with
0.1
% H
CO
OH
as
mod
ifier
. UV
det
ectio
n at
260
nm c
an b
e ap
plie
d as
add
ition
al
dete
ctio
n sy
stem
in c
ase
of a
n ar
omat
ic b
ase
in th
e nu
cleo
side
.It
is c
lear
ly o
bser
ved
that
the
urin
e sa
mpl
es c
onta
ins
mor
e nu
cleo
side
s th
an th
e st
anda
rd m
ixtu
re.
Anal
ysis
Pat
hway
Urin
e Sa
mpl
e
Affin
ity C
hrom
atog
raph
y
LC-E
SI-o
a-TO
F-M
SN
ucle
osid
e ?
Neu
tral
Los
s
m/z
= 1
32
Gen
erat
e M
olec
ular
For
mul
aco
mbi
ning
Accu
rate
Mas
s an
d Tr
ue Is
otop
ic P
atte
rn
Dat
aBas
eSea
rch
-Sci
Find
er
Prop
osed
Str
uctu
re fo
r Mod
ified
Nuc
leos
ide
132
B
O O
1
2
3
4
5
5 G
MF a
nd
Data
base
Searc
h R
esu
lts
Tabl
e 1:
GM
F an
d S
ciFi
nder
Sea
rch
Res
ults
-lig
ht b
lue:
new
nuc
leos
ides
in u
rine;
dar
k bl
ue: c
onfir
mat
ion
of k
now
n nu
cleo
side
Figu
re1:
UV
/BP
-Chr
omat
ogra
mof
nucl
eosi
dest
anda
rdm
ixtu
re(A
)an
dpu
rifie
dre
alur
ine
sam
ple
(B).
UV
@26
0nm
BP
C E
SI p
ositi
ve m
ode
Aca
libra
tion
stan
dard
UV
@26
0nm
BP
C E
SI p
ositi
ve m
ode
Bca
libra
tion
stan
dard
284.
0994
152.
0631
m
/z =
132
N
NH
N
N
O
OH
OH
HH
OH
O
NH
3+
N H
NH
N
N
O
NH
3+
1 23
456
7
8
9-
ribos
e
Figu
re 3
: GM
F-di
alog
with
mol
ecul
ar fo
rmul
a pr
opos
als
for 1
-m
ethy
lgua
nosi
ne
298.
1152
299.
1182
NN
NN
O
O
NH
2
OH
OH
OH
CH
3
ES
I-TO
F-M
SFo
r re
sear
ch u
se o
nly
. N
ot
for
use
in d
iagnost
ic p
roce
dure
s.
-14-
Clinical Metabolomics
Metabolomics Posterbook 2011
Bru
ker
Dalt
on
ics
& B
ruker
Bio
spin
Co
mb
ined
an
aly
sis
of
NM
R a
nd
LC
/M
S d
ata
: o
bse
rvati
on
s in
uri
ne s
am
ple
s fr
om
a lif
est
yle
stu
dy
Mar
kus
God
ejoh
ann1 ;
Har
tmut
Sch
äfer
1 ; C
hris
tian
Fisc
her1 ;
Pete
r Nei
dig1 ;
Man
fred
Spr
aul1 ;
Duk
-You
ng H
an2 ;
Eun
il Le
e3 ; G
abrie
la Z
urek
4 ; C
arst
en B
aess
man
n41 B
ruke
r Bio
Spin
Gm
bH, R
hein
stet
ten,
Ger
man
y; 2 K
BSI
Seo
ul, S
eoul
, Sou
th-K
orea
; 3 Kor
ea U
nive
rsity
, Seo
ul, S
outh
-Kor
ea; 4 B
ruke
r Dal
toni
k G
mbH
, Bre
men
, Ger
man
y
Co
ncl
usi
on
s
Info
rmat
ion
from
NM
R a
nd L
CM
S c
an b
e m
erge
d in
to o
ne c
ovar
ianc
e m
atrix
, i.e
. int
erpr
etin
g th
e ov
eral
l exp
erim
enta
l des
ign
as s
tatis
tical
he
tero
spec
trosc
opy
[1].
Met
abol
ites
and
met
abol
ic n
etw
orks
dom
inat
ing
the
stat
istic
s in
bo
th L
CM
S a
nd N
MR
spe
ctra
, can
in p
rinci
ple
be
iden
tifie
d by
the
cros
s-pe
aks
in th
e th
ree
dim
ensi
onal
che
mic
al s
hift
-mas
s -r
eten
tion
time
dom
ain.
The
com
bine
d co
varia
nce
mat
rix o
f N
MR
and
LC
-MS
dat
a fro
m th
e ur
ine
sam
ples
is
depi
cted
in F
ig. 4
.
Com
bine
d C
ovar
ianc
e M
atrix
Fig.
4:A
Sub
sect
ion
of th
e co
mbi
ned
cova
rianc
e m
atrix
of N
MR
and
LC
-MS
dat
a sh
owin
g co
rrela
tion
betw
een
the
arom
atic
pr
oton
s of
the
para
ceta
mol
glu
curo
nide
and
the
mas
s re
spon
se a
t 328
m/z
; Bin
tens
ity
dist
ribut
ion
for 3
28 m
/z a
nd 7
.34p
pm o
f all
sam
ples
; CM
S a
nd N
MR
spe
ctra
from
sam
ple
539.
para
ceta
mol
glu
curo
nide
para
ceta
mol
sul
fate
7.34
ppm
# 53
9
MS
sam
ple
539
NM
R s
ampl
e 53
932
8.10
44 m
/z
232.
0282
m/z#
539
NM
R
MS
m/z
ppm
Inte
nsity
of 3
28 m
/z in
all
sam
ples
Inte
nsity
of 7
.34
ppm
in a
ll sa
mpl
es
328 m/z
ppm
m/z
A B C
In m
etab
onom
ic s
tudi
es, t
he a
pplic
atio
n of
mor
e th
an o
ne a
naly
tical
pla
tform
is a
pre
requ
isite
to
succ
ess.
In p
artic
ular
, the
com
bina
tion
of L
C-E
SI-
TOF-
MS
and
flow
inje
ctio
n N
MR
add
ress
es th
e an
alyt
ical
nee
d to
cov
er a
larg
e dy
nam
ic ra
nge
of
chem
ical
ly d
iver
se c
onst
ituen
ts in
bod
y flu
ids.
In a
re
cent
med
ium
sca
le s
tudy
on
hum
an m
etab
olis
m
base
d on
urin
e sa
mpl
es, t
he in
fluen
ces
of li
fest
yle
-in
parti
cula
r sm
okin
g -o
n th
e m
etab
olic
pro
file
have
bee
n ex
amin
ed. N
MR
and
LC
-MS
m
easu
rem
ents
wer
e pe
rform
ed u
sing
the
Met
abol
icP
rofil
erTM
(see
Fig
. 1).
Cur
rent
resu
lts a
re
base
d on
dat
a fro
m 1
026
sam
ples
(466
sm
oker
, 24
2 no
n sm
oker
, 225
sto
p sm
oker
, and
93
unkn
own
stat
us).
Fig.
1: M
etab
olic
Pro
filer
TM: S
ynch
roni
zed
NM
R
and
MS
& c
ompl
ete
data
eva
luat
ion.
Intr
oduc
tion
PCA
on L
C-M
S an
d N
MR
dat
aFi
gure
2: P
CA
on
a bu
cket
tabl
e de
rived
from
LC
MS
dat
a re
veal
s ef
fect
s of
life
styl
e an
d m
edic
atio
n (F
ig. 2
A: P
C9/
PC
10 s
core
s pl
ot).
Rel
ated
met
abol
ites
can
be id
entif
ied
from
lo
adin
gs p
lot d
ue to
acc
urat
e m
ass
info
rmat
ion
(see
Fig
. 2C
) and
true
isot
opic
pat
tern
. Exa
mpl
es
for i
llust
ratio
n ar
e E
phed
rine,
a c
onst
ituen
t of
tradi
tiona
l Chi
nese
med
icin
e (m
/z16
6.12
33 =
C
10H
16N
O a
nd c
hara
cter
istic
frag
men
t at m
/z
148.
1129
= C
10H
14N
) and
Pen
icill
in ty
pe
antib
iotic
, (m
/z35
0.11
82 =
C16
H20
N3O
4S).
LC-M
S
data
of t
wo
affe
cted
indi
vidu
als
(hig
hlig
hted
by
the
red
arro
w in
Fig
. 2A
) are
dis
play
ed a
s co
ntou
r pl
ots
(Fig
. 2B
) for
spe
ctra
l ver
ifica
tion.
S
ampl
e co
lour
: Sm
oker
; N
on-s
mok
er ●
; sto
p-sm
oker
●, u
nkno
wn
stat
us ●
.
Ref
eren
ces
[1] D
. J. C
rock
ford
et.
al.,
Anal
. Che
m. 2
006
7836
3-37
1.[2
] http
://w
ww
.gen
ome.
jp/k
egg/
kegg
2.ht
ml
Expe
rimen
ts /
Met
hods
Urin
e Sa
mpl
es:S
pot u
rine
sam
ples
(Kor
ea U
nive
rsity
, Seo
ul, S
outh
Kor
ea)
wer
e fro
zen
imm
edia
tely
afte
r sam
plin
g an
d on
ly th
awed
for p
repa
ratio
n.Sa
mpl
e pr
epar
atio
n:S
ampl
es w
ere
auto
mat
ical
ly p
repa
red
from
one
sou
rce
into
two
sepa
rate
96-
wel
l pla
tes.
NM
R p
rep.
:sam
ples
wer
e bu
ffere
d by
add
ing
60µl
of b
uffe
r (pH
7 in
D2O
co
ntai
ning
0.1
% T
SP
and
0.0
13%
NaN
3) to
540
µl o
f urin
e.LC
-MS
prep
.:2-
fold
dilu
tion
with
H2O
. N
MR
on
Bru
ker A
vanc
e 60
0 N
MR
Spe
ctro
met
er u
sing
flow
inje
ctio
n te
chno
logy
(B
ES
T-N
MR
, Bru
ker B
ioS
pin,
Rhe
inst
ette
n, G
erm
any.
T=3
00K
).LC
-MS
on E
SI-T
OF
mic
rOTO
FTMm
ass
spec
trom
eter
(Bru
ker D
alto
nik,
Bre
men
, G
erm
any)
cou
pled
to a
n H
PLC
sys
tem
from
Agi
lent
(Wal
dbro
nn, G
erm
any)
. LC
se
para
tion
was
car
ried
out o
n a
reve
rsed
pha
se c
olum
n (W
ater
s A
tlant
is
100*
2.1m
m, 3
µm p
artic
les)
with
a 1
0min
-gra
dien
t fro
m 9
8% H
2O /
2%
acet
onitr
ile (A
cN) +
0.2
% H
CO
OH
to 2
0% H
2O w
ithin
8 m
in fo
llow
ed b
y a
2min
flu
shin
g w
ith 1
00%
AcN
. The
flow
rate
was
0.5
ml/m
in. 5
% o
f the
flow
wer
e sp
lit
into
the
MS
. Sod
ium
form
iate
sol
utio
n w
as in
ject
ed a
s ca
li-br
ant a
t the
sta
rt of
ea
ch c
hrom
atog
raph
ic ru
n.D
ata
Acq
uisi
tion
Softw
are:
Tops
pin
1.3
(Bru
ker B
iosp
in, R
hein
stet
ten,
G
erm
any)
, Com
pass
1.0
(Bru
ker D
alto
nik,
Bre
men
, Ger
man
y).
Dat
a In
terp
reta
tion
& S
tatis
tics
Softw
are:
AM
IX 3
.6a
(Bru
ker B
iosp
in) f
or
com
bine
d st
atis
tical
LC
-MS
and
NM
R in
terp
reta
tion,
Dat
aAna
lysi
s 3.
4 an
d P
rofil
e A
naly
sis
1.0
(Bru
ker D
alto
nik)
for M
S in
terp
reta
tion,
Mat
Lab
(Mat
hwor
ks,
MA
, US
A) f
or a
dditi
onal
inte
rpre
tatio
n &
vis
ualiz
atio
n.B
ucke
ting
para
met
er:(
1) d
ynam
ic b
ucke
ting
met
hod
in A
MIX
; Mas
s ra
nge:
m
/z=
50.5
-800
.5
m=0
.01;
Tim
e ra
nge
from
0.5
-10m
in;
t = 0
.8; 5
966
buck
ets
in to
tal.
(2) m
ass
rang
e fro
m m
/z=
150.
5-20
0.5,
m
= 0.
01 a
nd ti
me
rang
e fro
m
0.5-
3min
, t =
0.8
usi
ng p
eak
inte
nsiti
es fr
om 1
000
to 1
0000
cou
nts
(exc
ludi
ng
inte
nse
mas
s pe
aks)
resu
lting
in 5
00 b
ucke
ts.
Whe
n do
ing
dist
ribut
ion
com
paris
on o
f all
indi
vdua
l ion
s re
gist
ered
in th
e bu
cket
tabl
e,
abou
t 150
MS
pea
ks a
re fo
und
to b
e si
gnifi
cant
ly
diffe
rent
(sig
nific
ance
of s
imila
rity
p <
0.05
) ad
ditio
nally
to th
e ob
viou
s m
etab
olite
s as
ni
cotin
e an
d pr
oduc
ts, m
entio
ned
abov
e. F
or
exam
ple,
one
MS
pea
k w
ith a
hig
h si
gnifc
ance
(p
= 1)
was
m/z
153.
065
at 2
.26m
in. T
he fo
rmul
a of
this
pea
k w
as id
entif
ied
usin
g ac
cura
te m
ass
and
isot
opic
pat
tern
info
rmat
ion
as C
7H8N
2O2
(neu
tral).
Whe
n se
arch
ing
this
mol
ecul
ar fo
rmul
a in
KE
GG
[2],
two
poss
ible
and
like
ly m
etab
olite
s ar
e id
entif
ied:
N1-
met
hyl-2
-pyr
idon
e-5-
carb
ox-
amid
e an
d N
1-m
ethy
l-4-p
yrid
one-
5-ca
rbox
amid
e bo
th d
eriv
ed fr
om n
icot
inam
ide.
NM
R a
nd L
CM
S p
rovi
de c
ompl
emen
tary
pe
rspe
ctiv
es, w
hich
are
ext
rem
ely
pow
erfu
l w
hen
com
bine
d w
ith s
tatis
tical
app
roac
hes
such
as
PC
A, P
CA
-DA
, and
com
bine
d co
varia
nce
anal
ysis
.
Whi
le m
ass
spec
trosc
opy
is g
ivin
g be
st re
sults
fo
r pol
ar a
naly
tes
elut
ing
with
a re
ason
able
re
tent
ion
from
the
HP
LC c
olum
n, th
e N
MR
giv
es
best
resu
lts fo
r sm
all e
ndog
enou
s m
etab
olite
s (F
ig. 3
). W
hile
the
MS
resp
onse
of a
n an
alyt
e us
ually
incl
udes
the
pseu
dom
olec
ular
ion
and
som
e fra
gmen
ts, t
he N
MR
resp
onse
can
spr
ead
out o
ver s
ever
al re
sona
nces
(see
Fig
. 3B
).
„glu
cose
ve
ctor
“
gluc
ose
refe
renc
e sp
ectru
m
cont
rol s
pect
rum
take
n fro
m m
iddl
e of
sco
res
plot
spec
trum
take
n fro
m e
nd o
f blu
e ve
ctor
„TM
AO
“
NM
RSc
ores
Load
ings
Plo
tA BG
ener
ate
Mol
ecul
ar F
orm
ula
(GM
F) R
esul
ts
166.
1233
350.
1182
148.
1129 m
= 1
8.01
03H
2O =
18.
0105
C Dspec
trum
on
left
spec
trum
on
right
mas
s re
spon
sibl
e fo
r di
ffere
nce
Scor
es P
lot
Load
ings
Plo
t
LC-M
S co
ntou
r plo
tBA
Fig.
2: P
CA
on L
CM
S D
ata,
see
text
for d
etai
ls
Fig.
3: P
CA
on N
MR
Dat
a, s
ee te
xt fo
r det
ails
Effe
cts
of S
mok
ing
A n
arro
w s
ettin
g of
the
buck
etin
g pa
ram
eter
s fo
r th
e LC
-MS
dat
a al
low
s to
iden
tify
trend
s fo
r the
di
scrim
inat
ion
betw
een
smok
ing
and
non-
smok
ing
indi
vidu
al a
ccor
ding
to n
icot
ine
and
its m
etab
olite
s (F
ig. 5
). It
has
to b
e no
ted
that
als
o fo
r non
-and
st
op-s
mok
ers
sign
ifica
nt le
vels
of t
hese
mar
kers
ca
n be
foun
d w
hich
mig
ht b
e ex
plai
ned
by th
e ef
fect
of p
assi
ve s
mok
ing.
Whe
n ap
plyi
ng P
CA
-DA
to
LC
MS
and
NM
R b
ucke
t tab
les
excl
udin
g ni
cotin
e an
d re
late
d m
etab
olite
s, it
can
be
obse
rved
that
the
scor
es d
istri
butio
n m
axim
a ar
e se
para
ted
for s
mok
ers
and
non-
smok
ers.
This
is
givi
ng s
ome
indi
catio
n th
at s
mok
ing
mig
ht in
duce
ch
ange
s in
met
abol
ism
. Fur
ther
det
aile
d an
alys
is
is n
eede
d fo
r ver
ifica
tion,
how
ever
.
Fig.
5:T
op:P
CA
ana
lysi
s of
LC
-MS
dat
a fro
m
1026
urin
e sa
mpl
es P
C1
vs. P
C2;
1ni
cotin
e (m
/z16
3.12
3), 2
cotin
ine
(m/z
177.
102)
, 3-
hydo
xy-c
otin
ine
(m/z
193.
097)
and
4hy
drox
y-ni
cotin
e (m
/z17
9.11
8). C
ente
r:
cotin
ine
inte
nsity
(m/z
177.
102
) as
func
tion
of s
ampl
e nu
mbe
r.B
otto
m: P
CA
-DA
sco
res
dist
ribut
ions
(LC
MS
: lef
t, N
MR
: rig
ht).
non-
smok
ersm
oker
non-
smok
ersm
oker
nico
tine
1co
tinin
e 2
OH
-nic
otin
e 4
OH
-cot
inin
e 3
ASc
ores
Plo
tLo
adin
gs P
lot
LC
-MS
/ N
MR
For
rese
arch
use
only
. N
ot
for
use
in d
iagnost
ic p
roce
dure
s.
-15-
Clinical Metabolomics
Co
ncl
usi
on
s
LU
MC
& U
trech
t U
niv
ers
ity &
Bru
ker
Dalt
on
ics
Intr
od
uct
ion
Meth
od
s
Su
mm
ary
Resu
lts –
Sta
tist
ical
Evalu
ati
on
Searc
hin
g b
iom
ark
ers
fo
r th
e C
om
ple
x R
eg
ion
al P
ain
Syn
dro
me b
y m
eta
bo
lic
pro
fili
ng
of
uri
ne u
sin
g C
E/
MS
Raw
i R
am
au
tar1
, A
nn
e A
. v.d
. P
las3
, K
ars
ten
Mic
helm
an
n4,
Ric
o J
.E.
Derk
s2,
Gab
riela
Zu
rek
4,
Go
vert
W.
So
mse
n1,
Gerh
ard
us
J. d
e J
on
g1
, J.
J. v
an
Hilte
n3,
An
dré
M.
Deeld
er2
, O
leg
A.
Mayb
oro
da
2
1 D
ept
of Bio
med
ical
Anal
ysis
, U
trec
ht
Univ
ersi
ty,
Utr
echt,
NL
2 B
iom
ole
cula
r M
ass
Spec
trom
etry
Unit,
Dep
t of
Para
sito
logy,
LU
MC,
Leid
en,
NL
3 D
ept
of N
euro
logy,
LU
MC, Le
iden
, N
L4 B
ruke
r D
alto
nik
Gm
bH
, Bre
men
, G
erm
any
Com
ple
x Reg
ional
Pai
n S
yndro
me
(CRPS
) is
a c
hro
nic
pai
n c
onditio
n c
har
acte
rize
d b
y a
variet
y of au
tonom
ic,
senso
ry, m
otor
and t
rophic
chan
ges
. In
mos
t ca
ses
CRPS
dev
elops
afte
r an
inju
ry o
r su
rger
y. H
owev
er,
the
pat
hogen
esis
of th
is s
yndro
me
rem
ains
uncl
ear. A
s a
conse
quen
ce, dia
gnosi
s of CRPS is
difficu
lt m
ainly
due
to
the
lack
of la
bor
atory
tes
ts. The
dev
elopm
ent
of r
elia
ble
dia
gnost
ic t
ests
req
uires
cle
ar u
nder
stan
din
g o
f bio
chem
ical
mec
han
ism
s of th
e pat
holo
gy.
Met
abolit
e pro
filin
g c
an p
rovi
de
insi
ght
into
mec
han
ism
s of dis
ease
an
d n
ovel
pro
gnost
ic o
r dia
gnost
ic m
arke
rs.
One
of th
e im
port
ant
clin
ical
man
ifes
tations
of CRPS is
aber
rant
inflam
mat
ion a
nd u
rine
could
be
an a
ppro
priat
e body
fluid
to a
cces
s th
e in
flam
mat
ory
com
pon
ent
of dis
order
.
The
des
crib
ed w
ork
is
par
t of th
e TREN
D (
Trau
ma
Rel
ated
Neu
ronal
Dys
funct
ion)
pro
ject
, fo
r m
ore
info
rmat
ion p
leas
e re
fer
to w
ww
.tre
ndco
nso
rtiu
m.n
l
•CE:
Bec
kman
Coulter
PA 8
00 (
Fulle
rton,
CA).
•N
onco
vale
ntly
coat
ed fuse
d s
ilica
cap
illar
y usi
ng
poly
bre
ne
and p
oly
(vin
ylsu
lfon
ate)
: 130 c
m x
50
µm
ID
.•
Bac
kgro
und e
lect
roly
te (
BG
E):
1M
form
ic a
cid p
H 1
.8.
•Sam
ple
inje
ctio
n:
hyd
rodyn
amic
inje
ctio
n w
ith p
H-
med
iate
d s
tack
ing u
sing 1
2.5
% N
H4O
H.
•ESI-
TOF-
MS:
Mic
rOTO
F (B
ruke
r D
alto
nik
, Bre
men
, G
erm
any)
.•
CE-M
S c
ouplin
g:
shea
th-f
low
inte
rfac
e an
d g
rounded
ESI
spra
yer.
•Shea
th liq
uid
: 50%
met
han
ol an
d 0
.1%
form
ic a
cid.
•Ele
ctro
spra
y: p
osi
tive
ioniz
atio
n m
ode,
-4.5
kV
ioniz
atio
n v
oltag
e.•
Dat
a ev
aluat
ion:
Dat
aAnal
ysis
soft
war
e (B
ruke
r D
alto
nik
Gm
bH
, G
erm
any)
for
mol
ecula
r fe
ature
ex
trac
tion, XCM
S (
Scr
ipps
Cen
ter
for
Mas
s Spec
trom
etry
, La
Jolla
, CA)
and S
IMCA-P
+ s
oft
war
e (U
met
rics
, U
mea
, Sw
eden
).•
Rea
l sa
mple
s: 2
9 u
rine
sam
ple
s fr
om
pat
ients
with
chro
nic
CRPS a
nd 1
8 u
rine
sam
ple
s fr
om
hea
lthy
contr
ols
.•
Sam
ple
pre
par
atio
n:
urine
1x
dilu
ted w
ith B
GE a
nd
centr
ifugat
ed for
5 m
in a
t 13200 g
.
We
dem
onst
rate
d t
hat
CE-E
SI-
TOF-
MS is
an a
nal
ytic
al
pla
tform
fully
am
enab
le for
clin
ical
met
abolo
mic
s st
udie
s. T
he
bio
logic
al a
nd c
linic
al inte
rpre
tation o
f th
is
dat
a se
t is
the
prim
ary
goal
of fu
rther
eva
luat
ion.
•CE-T
OF-
MS u
sing n
on-c
oval
ently
coat
ed c
apill
arie
s ca
n b
e use
d f
or f
ast
and r
epro
duci
ble
met
abolic
pro
filin
g o
f urine
with m
inim
al s
ample
pre
-tr
eatm
ent.
•
Met
abolic
pro
filin
g o
f urine
from
pat
ients
with C
RPS b
y PB-P
VS C
E-T
OF-
MS o
ffer
s pote
ntial
to a
dva
nce
the
under
stan
din
g o
f th
e bio
chem
ical
bas
is
of
CRPS.
•Fu
rther
stu
die
s ar
e re
quired
to
valid
ate
CRPS
spec
ific
ity
of
com
pounds
resp
onsi
ble
for
cla
ssific
atio
n.
The
optim
ized
CE-T
OF-
MS m
ethod w
as t
hen
applie
d t
o a
se
t of cl
inic
al u
rine
sam
ple
s of 29 C
RPS
pat
ients
and 1
8
contr
ols
. All
dat
a file
s w
ere
reca
libra
ted a
nd e
xport
ed a
s net
CD
F file
s fo
r fu
rther
pro
cess
ing.
Mig
ration t
ime
alig
nm
ent
was
per
form
ed u
sing X
CM
S s
oft
war
e. T
he
resu
ltin
g file
(TSV d
ata)
was
exp
ort
ed t
o S
IMCA-P
+ 1
1.2
fo
r fu
rther
anal
ysis
. In
itia
l an
alys
is w
as p
erfo
rmed
usi
ng
princi
ple
com
ponen
t an
alys
is (
PCA)
yiel
din
g a
co
nsi
der
able
gro
upin
g first
PCs,
but
with v
ery
limited
pre
dic
tive
abili
ty (
dat
a not
show
n).
Fig.
1 S
chem
atic
of Anal
ytic
al W
ork
flow
Fig.
3 S
tatist
ical
Anal
ysis
usi
ng O
PLS-D
A.
CE-E
SI-
TOF-
MS
A.T
wo
cla
ss P
LS-D
A m
odel
al
low
s di
scrim
inat
ion
of p
atie
nt a
nd c
ontr
ol
grou
p0
1000
00
2000
00
3000
00
4000
00
5000
00
06.
2512
.525
NH
4OH
conc
entr
atio
n(%
)
Plate number
Valin
ePh
enyl
alan
ine
0
1000
000
2000
000
3000
000
4000
000
5000
000
Lysi
neA
rgin
ine
Valin
ePh
enyl
alan
ine
Tyro
sine
Peak area
BG
E
Spik
ed b
efor
e pr
etre
atm
ent
Spik
ed a
fter p
retr
eatm
ent
Inte
nsity
2 x1
05Ar
bitr
ary
units
1014
12
Tim
e [m
in]
Uri
ne
dilu
ted
wit
h B
GE
(1:1
)Fa
st s
epar
atio
n o
n
bila
yer
coat
ed c
apill
ary
Det
ecti
on u
sin
g M
S w
ith
h
igh
mas
s ac
cura
cyC
E-To
F-M
S pr
ofile
s
m/z
50
-4
50
XC
MS
PC
AP
LS-D
A
Pre
-pro
cess
ing
of
raw
CE-
ToF-
MS
data
Mig
rati
on t
ime
alig
nm
ent
and
nor
mal
izat
ion
Fig.
2 O
ptim
izat
ion o
f an
alyt
ical
met
hod in t
erm
s of se
par
atio
n e
ffic
iency
and s
ample
pre
par
atio
n.
A. In
fluen
ce o
f th
e N
H4O
H c
once
ntr
atio
n
in p
re-i
nje
ctio
n p
lug o
n p
late
num
ber
.AB
C
B.
Ext
ract
ed ion e
lect
ropher
ogra
ms
(EIE
) of
an a
min
o a
cid (
AA)
test
mix
ture
(5
0 μ
M e
ach).
C.
Peak
are
as o
f se
lect
ed a
min
o a
cids
afte
r th
eir
spik
ing (
50
M e
ach)
into
BG
E a
nd
into
uri
ne
bef
ore
& a
fter
pre
trea
tmen
t.
1,
Lysi
ne
2,
Arg
inin
e3,
Ala
nin
e4,
Val
ine
5 &
6,
Isole
uci
ne
&Le
uci
ne
7,
Met
hio
nin
e8,
Glu
tam
ic a
cid
9,
Phen
ylal
anin
e10,
Tyro
sine.
In o
rder
to o
bta
in g
ood
focu
sing, th
e urine
sam
ple
has
to
be
acid
ifie
d p
rior
to inje
ctio
n w
hic
h is
achie
ved b
y dilu
tion
with B
GE (
1:1
; v/
v).
The
effe
ct o
f th
e urine
mat
rix
on t
he
signal
inte
nsi
ty o
f se
lect
ed A
As
wer
e in
vest
igat
ed b
y sp
ikin
g e
xper
imen
ts.
Figure
2.C
dep
icts
th
e pea
k ar
eas
of se
lect
ed A
As
in B
GE a
nd s
pik
ed into
urine
prior
and a
fter
pre
trea
tmen
t (1
x dilu
tion in B
GE +
ce
ntr
ifugat
ion).
A s
ignific
ant
dec
reas
e of si
gnal
inte
nsi
ty
is o
bse
rved
for
AAs
in t
he
urine
sam
ple
s due
to
ioniz
atio
n s
uppre
ssio
n. The
mag
nitude
of th
is e
ffec
t w
as
consi
sten
t fr
om
one
urine
sam
ple
to a
noth
er.
Resu
lts –
An
aly
tica
l M
eth
od
Bas
ed o
n p
revi
ous
work
on a
min
o a
cid (
AA)
pro
filin
g
usi
ng C
E-T
OF-
MS [
May
boro
da
et.
al.
J Chro
m A
1159
(2007)
149-5
3],
the
stab
ility
of m
igra
tion t
imes
has
bee
n im
pro
ved b
y usi
ng c
apill
arie
s dyn
amic
ally
coat
ed
with a
bila
yer
of poly
bre
ne
(PB)
and p
oly
(vin
yl)s
ulfonat
e (P
VS).
In-c
apill
ary
pre
conce
ntr
atio
n u
sing p
H-m
edia
ted
stac
king a
llow
ed f
or
limits
of det
ection in t
he
nan
om
ola
r co
nce
ntr
atio
n r
ange.
The
effe
ct o
f th
e N
H4O
H
conce
ntr
atio
n in
the
pre
-inje
ctio
n p
lug o
n t
he
separ
atio
n
effici
ency
of va
line
and p
hen
ylal
anin
e is
giv
en in F
ig 2
.A.
The
separ
atio
n o
f an
AA s
tandar
d m
ixtu
re is
pre
sente
d in
figure
2.B
usi
ng a
n o
ptim
ized
inje
ctio
n v
olu
me
of ca
. 100
nL
(with p
rein
ject
ion o
f 12.5
% N
H4O
H s
olu
tion
).
Part
ial le
ast
squar
e dis
crim
inan
t an
alys
is w
ith o
rthogonal
si
gnal
corr
ection (
OPL
S-D
A)
apply
ing t
he
know
n c
lass
in
form
atio
n w
as s
ubse
quen
tly
applie
d t
o t
he
dat
a. T
he
OPL
S-D
A s
core
s plo
t giv
en in F
ig.
3A s
how
s th
e ex
pec
ted
gro
upin
g w
ith a
n inital
iden
tifica
tion o
f co
rres
pondin
g
load
ings
in F
ig.
3B.
R2X(1
) an
d R
2X(2
) ar
e th
e Sum
of
Squar
es o
f al
l th
e X’s
exp
lain
ed b
y th
e ex
trac
ted
com
ponen
ts. The
iden
tifica
tion o
f in
fluen
tial
load
ings
is
still
ongoin
g a
nd w
ill f
orm
the
bas
is for
the
bio
logic
al
inte
rpre
tation.In
ord
er t
o judge
on t
he
pre
dic
tive
abili
ty
of th
e m
odel
, va
lidat
ion b
y a
random
per
muta
tion
tes
t has
bee
n p
erfo
rmed
(Fi
g.
3C).
-2.0
007
-1.0
007
0.00
00
1.00
07
2.00
07
-2.0
007
-1.0
007
0.00
001.
0007
2.00
07
t[2]
t[1]
step
_3_b
w_30
_20_
OSC
.M16
(PLS
-DA)
t[Com
p. 1
]/t[C
omp.
2]
Colo
red
acco
rdin
g to
cla
sses
in M
16
R2X[
1] =
0.1
9776
R2X
[2]
= 0.
1108
32
Ell
ipse
: Ho
tell
ing
T2 (
0.95
)
Clas
s 1
Clas
s 2
SIM
CA-
P+ 1
1.5
- 5/2
/200
8 5:
31:2
8 PM
B.O
PLS-
DA
Load
ings
Plo
tId
entif
icat
ion
of m
olec
ular
dis
crim
inat
ors
base
d on
load
ings
is c
urre
ntly
ong
oing
C.A
ran
dom
per
mut
atio
n te
st t
o as
sess
ro
bust
ness
of
the
OPL
S-D
A m
odel
(Fi
g 3.
A)
-0.3
-0.2
-0.1
-0.00.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
step
_3_b
w_30
_20_
OSC
.M16
(PLS
-DA)
: Val
idat
e M
odel
$M16
.DA1
Inte
rcep
ts: R
2=(0
.0, 0
.315
), Q
2=(0
.0, -
0.27
4)
50 p
ermu
tati
ons
2 co
mpon
ents
R2 Q2
SIM
CA-
P+ 1
1.5
- 5/2
/200
8 5:
37:2
5 PM
Inte
rcep
ts:
R2=
(0.0
, 0.
315)
, Q
2=(0
.0,
-0.2
74)
-0.4
-0.3
-0.2
-0.1
-0.00.1
0.2
0.3
0.4
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
0.1
w*c[2]
w*c
[1]
step
_3_b
w_30
_20_
OSC
.M16
(PLS
-DA)
w*c[
Com
p. 1
]/w*c
[Com
p. 2
]Co
lore
d ac
cord
ing
to V
ar ID
(Prim
ary)
R2X[
1] =
0.1
9776
R2
X[2]
= 0
.110
832 S
IMC
A-P+
11.
5 - 5
/2/2
008
5:36
:21
PM
Hip
puric
Hip
puric
acid
acid
Hip
puric
Hip
puric
acid
acid
Met
hylh
istid
ine
Met
hylh
istid
ine
Met
hylh
istid
ine
Met
hylh
istid
ine
22 --Am
inoo
ctan
oic
acid
Amin
ooct
anoi
c ac
id22 --
Amin
ooct
anoi
c ac
idAm
inoo
ctan
oic
acid
Prol
ine
Prol
ine
beta
ine
beta
ine
Prol
ine
Prol
ine
beta
ine
beta
ine
AB
C
R2X
[1]=
0.19
776
R2X
[2]=
0.11
083
R2X
[1]=
0.19
776
R2X
[2]=
0.11
083
A.T
wo
cla
ss O
PLS-
DA
mod
el
allo
ws
disc
rimin
atio
n of
pat
ient
and
co
ntro
l gro
up
For
rese
arch
use
only
. N
ot
for
use
in d
iagnost
ic p
roce
dure
s.
-16-
Clinical Metabolomics
Metabolomics Posterbook 2011
Conclusions
Bruker Daltonics
Matthias Witt1, Khalid Anwer2, Boris Koch3 and Jens Fuchser1
1Bruker Daltonik GmbH, Bremen, Germany2Jamia Hamdard, Hamdrad University, New Delhi, India3Alfred-Wegener Institute for Polar and Marine Research Bremerhaven, Germany
Fig. 1 Broadband Electrospray spectrum of Fulvic acids from Shilajit (top) and a zoom view (bottom)
Introduction
Methods
Results
Elemental analysis of fulvic acids of Shilajit by ultra-high resolution FTMS
ESI-FTICRMS
• From a single measurement more than 5000 elemental compositions could be identified in the fulvic acid extract of shilajit.
• For complex mixtures, which are not separated sufficiently by chromatographic methods, high resolution (≥ 300.000) FTICR MS is mandatory.
• Similar distributions of elemental compositions have been found for compounds bearing no, one or two nitrogen atoms.
• The measured average H/C ratio of 1.27 is relatively high compared to the one of Suwannee river fulvic acids standard from the IHSS (1.02).
389.01510
389.03619
389.05145
389.06703
389.08779
389.09906
389.10888
389.12417
389.14528
389.16058 389.17365
389.18173
389.19143 389.21005
389.025 389.050 389.075 389.100 389.125 389.150 389.175 389.200 389.225 m/z0.0
0.5
1.0
1.5
8x10Intens.
389.025 389.050 389.075 389.100 389.125 389.150 389.200389.175m/z
389.225
389.00 389.05 389.10 389.15 389.20
0.0
5.0x104
1.0x105
1.5x105
2.0x105
Inte
nsity
(a.i.
)
m/z
Shilajit: Spectrum simulation around m/z 389Resolving power: 100000
Analysis Type ResultAverage
molecular mass (N=0)401 Da
Average O/C (N=0) 0.55Average H/C (N=0) 1.27Average DBE (N=0) 7.7
Identified elemental compositions (N=0) 2540
Identified elemental compositions (N=1) 1566
Identified elemental compositions (N=2) 1071
Elemental analysisby FTMS
C (%): 54.40H (%): 5.67N (%): 1.14O (%): 38.79
Van Krevelen analysis of fulvic acids of Shilajit
O/C(Center of mass)
H/C(Center of mass)
Van Krevelen analysisN=0
0.55 1.27
Van Krevelen analysisN=1
0.51 1.22
Van Krevelen analysisN=2
0.57 1.19
Shilajit is a pale-brown exudation of variable consistency, oozing out from layer of rocks in many mountain ranges of the world. It originates from bio-organic material having been exposed to high pressures and high temperatures. Shilajit has been used as a rejuvenator and an adaptogen for thousands of years. Its major physiological action has been attributed to the presence of bioactive dibenzo-alpha-pyrons along with humic and fulvic acids which act as carrier molecules for the active ingredients. Fulvic acids are known as a tremendously complex mixture of organic compounds. However, due to the extremely high resolving power and mass accuracy of Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry the elemental composition of these compounds can be determined directly from this complex mixture.
Mass spectrometer: APEX Qe 9.4 T equipped with an Apollo II source (Bruker Daltonics, Billerica, USA) using negative electrospray ionization. Analyte: Fulvic acids of Shilajit after extraction with macrophorous ion-exchange resin and desalting with styrene divinyl benzene polymer adsorber. The used solution was approx. 0.5 mg/mL in water/methanol 50/50 sprayed @ 0.12 mL/h.
Figure 1 displays the complexity of the sample. Automated assignment of molecular formulae was carried out using TargetAnalysis. For the analyses different number of nitrogen atoms were allowed.Figure 2 illustrates the need for high resolution mass spectrometry: A resolution of 100.000 is insufficient to completely resolve all present compounds.
Table 2 Summery of characteristics of fulvic acids from Shilajit
Fig. 3 Kendrick Intensity plot (top)VanKrevelen Molecular mass plot (middle)VanKrevelen Intensity plot (bottom)
200 300 400 500 600 7000.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
Kend
rick
Mas
s D
efec
t
Kendrick Nominal Mass
Kendrick plot - Shilajit - ESI negative ion mode
Center of mass values:KM: 395, KMD: 0.317
0.0 0.2 0.4 0.6 0.8 1.00.0
0.5
1.0
1.5
2.0
0.0
0.5
1.0
1.5
2.0
Shilajit - Van Krevelen Molecular mass plot ESI negative ion mode, N=0
H/C
O/C
200
300
700
400
600
500
Mol. mass
Average molecular mass (weighted by peak intensity): 401 dalton
Nitrogen atoms = 0
0
100
%
Intensity
0.0 0.2 0.4 0.6 0.8 1.00.0
0.5
1.0
1.5
2.0
0.0
0.5
1.0
1.5
2.0
Shilajit - Van Krevelen plotESI negative ion mode; N=0
H/C
O/C
Nitrogen atoms = 0
0
100
%
Intensity
Fig. 2 Comparison of simulated spectrum (top) and measured high-resolution spectrum (bottom)
1
2
Instrument status
Instrument status
257.10310
286.06487
355.03057
385.18675
403.06697
431.19223
449.10897
493.13515
519.17191549.16135
611.16174
250 300 350 400 450 500 550 600 650 700 m/z0.0
0.5
1.0
1.5
2.0
2.5
8x10Intens.
-MS
389.01495
389.03612
389.05138
389.06698
389.08784
389.09896
389.10892
389.12420
389.14532
389.16059
389.17374
389.19710 389.21014
389.025 389.050 389.075 389.100 389.125 389.150 389.175 389.200 389.225 m/z0.0
0.5
1.0
1.5
2.0
2.5
7x10Intens.
C19H33O80.3 ppm
C19H34O6P0.7 ppm
C22H29O60.3 ppm
C29H25O0.8 ppm
C18H29O90.0 ppm
C18H39O7P0.7 ppm
C21H25O70.0 ppm
C17H25O100.0 ppm
C17H26O8P0.8 ppm
C19H21N2O70.8 ppm
C20H21O80.0 ppm
C15H21N2O100.1 ppm
C16H21O110.0 ppm
C18H17N2O80.2 ppm
C19H17O90.1 ppm
C14H17N2O110.2 ppm
C15H17O120.2 ppm
C15H21N2O4S30.2 ppm
C18H13O100.1 ppm
C14H13O130.1 ppm
C17H9O110.2 ppm
Resolving power: 300.000
389.025 389.050 389.075 389.100 389.125 389.150 389.200389.175 m/z389.225
250 300 350 400 450 500 600550 650 700 m/z
C17H13N2O90.0 ppm
Table 1 Detailed Van Krevelen Analysis
Resolution = 700.000
Resolution = 100.000
Simulated spectrum
Measured spectrum
m/z
For research use only. Not for use in diagnostic procedures.
-17-
Metabolomics Technology: FT-MS
Conclusions
Bruker Daltonics
Introduction
Methods
Summary
Characterization of dissolved organic matter in marine pore waters by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry
FT-ICR-MS
The study demonstrate that FT-ICR-MS is able to resolve the molecular composition of DOM, extracted from sediment pore water even in a volume as small as 50 ml. The observed variations can be linked to environmental processes and are an important contribution to the basic understanding of the organic matter cycle at continental shelves.
The molecular compositions of DOM in marine sediment pore water and of river DOM was determined by FT-ICR-MS.
Molecular variations in DOM could either be linked to a change in source (O/C, H/C) or to transformation processes (m/z, DBE) within the sediment.
While the analyzed DOM represents a rather labile pool, POM is more stabile and therefore preserved in the sediment; degradation of POM becomes more important with longer transport and residence times.
Recently, the application of ultrahigh resolution Fourier transform ion cyclotron mass spectrometry (FT-ICR-MS) has been extended to the characterization of DOM in natural environments. We used FT-ICR-MS to elucidate the fate of organic matter shortly after deposition at a continental shelf. Although OM accumulated in shelf sediments play an essential role in the global carbon cycle, the source variations and transformation processes within the sediment are so far poorly understood and require further investigations. In this study, we compare pore water DOM to the lipid biomarker composition in particulate organic matter (POM) of the associated sediment.
Frauke Schmidt1; Boris Koch2; Marcus Elvert1; Matthias Witt3, Gabriela Zurek3, Kai-Uwe Hinrichs11DFG-Research Center Ocean Margins, Bremen University, Germany; 2Alfred-Wegener-Institute for Marine Research, Bremerhaven, Germany; 3Bruker Daltonik GmbH, Bremen, Germany
Pore water DOM:•Higher abundance of N-containing molecules and compounds with lower m/z (Fig. 1) source signal•Progressive loss of fulvic acids and first appearance of compounds with high H/C ratios (Fig. 3a) along the transect source signal
•Weighted average m/z & DBE compared to the marine and river water column (Fig. 3b) transformation of DOM induced by a higher microbial activity in the sedimentWithin the pore water DOM the transformation strength seemed to be linked to sediment properties (e.g. grain size, TOC (data not shown)) and therefore potential substrate for microbes.
Contrariwise POM shows a decrease of terrestrial markers (lignin) and a higher degradation along the transect (Fig.4) POM represents a more stabile organic matter pool and degradation is a factor of the residence time
E
W
a) b)Minho
11002
Figure 1. a) FT-ICR-MS spectrum of the Minho River and the mudbelt sample (11002) b) FT-ICR-MS spectrum of the transect at 449 m/z. marine pore waters show a decrease in pure CHO-compounds and the appearance of N-containing formulas in the analytical window.
50 ml of pore water was obtained from the surface sediment by rhizon sampling. DOM was extracted from the pore water and the river water by solid phase extraction (Varian, PPL). Samples were ionized by negative electrospray ionization and detected with a 9.4 T FT-ICR mass spectrometer. Unequivocal molecular formulas were determined on the base of exact masses. Lipid biomarkers were extracted with DCM/MeOH (2:1) by microwave and analyzed by GC-MS/FID for the identification and quantification of the compounds.
Results
Major river input
bottom currents
Figure 2. Sampling locations at the Galician shelf. The mid-shelf mudbelt (gray area) consists of fine grained, TOC-rich material.
1 0.1 8
449.57
0.52
1 .1 3
Minho
8.408.908.629.1 2DBE
1 .261 .241 .291 .26H/C
41 1 .99426.58422.50461 .44m/z
0.50
Atlantic
0.500.500.49O/C
1 10021 10061 1033
1 0.1 8
449.57
0.52
1 .1 3
Minho
8.408.908.629.1 2DBE
1 .261 .241 .291 .26H/C
41 1 .99426.58422.50461 .44m/z
0.50
Atlantic
0.500.500.49O/C
1 10021 10061 1033
fulvicacids
fermented sugarsfatty acids
fulvicacids
fermented sugarsfatty acids
HOHO
O
OH
OH
OHO
O
HO
O
OH
HOHOHO
O
OH
OH
OHO
O
HO
O
OH
HO
O
OH
OH
OHO
O
HO
O
OH
a) b)
1 0.1 8
449.57
0.52
1 .1 3
Minho
8.408.908.629.1 2DBE
1 .261 .241 .291 .26H/C
41 1 .99426.58422.50461 .44m/z
0.50
Atlantic
0.500.500.49O/C
1 10021 10061 1033
1 0.1 8
449.57
0.52
1 .1 3
Minho
8.408.908.629.1 2DBE
1 .261 .241 .291 .26H/C
41 1 .99426.58422.50461 .44m/z
0.50
Atlantic
0.500.500.49O/C
1 10021 10061 1033
fulvicacids
fermented sugarsfatty acids
fulvicacids
fermented sugarsfatty acids
HOHO
O
OH
OH
OHO
O
HO
O
OH
HOHOHO
O
OH
OH
OHO
O
HO
O
OH
HO
O
OH
OH
OHO
O
HO
O
OH
a) b)
Figure 3. a) Van Krevelen diagram of all identified CHO-compounds (200<m/z<600, ppm<0.5). The loss of terrestrial fulvic acids and the appearance of fatty acids and fermented sugars along the transect result in a shift to higher H/C and lower O/C ratios. b) Weighted average molecular element ratios, m/z and double bond equivalents (DBE = sum of rings and double bonds in a molecule) along the transect.
We identified distinct molecular changes from the river to the pore water DOM and linked them to either source variations or transformation processes. River DOM:•High O/C and low H/C values (Fig. 3b) source signal: higher complexity and oxygenation of terrestrial OM
Figure 4. Relative contributions of algae markers and lignin to the POM in the marine sediment. During degradation vanillin is transformed into vanillic acid.
W E
OH
OCH3
C O
CH3OH
OCH3
CO H
degradation
For research use only. Not for use in diagnostic procedures.
-18-
Metabolomics Technology: FT-MS
Metabolomics Posterbook 2011
Applications & References
Bruker BioSpin & Bruker Daltonics
Hybrid LC/MS/NMR as a Unique Tool for Analysis of Highly Complex Samples and Integrated Chemical Identification
MetabolicProfilerTM LC-MS/NMR
Introduction
One of the most challenging tasks in analytical chemistry today is the characterization of complex biological mixtures, i.e. for biomarker discovery, or the discrimination of disease states, origin etc. LC/MS is an established tool for high throughput analysis with good sensitivity at reasonable costs. NMR is the most prominent analytical technique for structure elucidation and in screening for e.g., metabolomics or routine applications like food and beverages analysis. The combination of LC/MS and Flow-NMR, however, can effectively yield qualitative and quantitative information by applying statistical tools. Data will be shown such as newborn metabolomics and the geographical discrimination of beverages by (bio)marker detection.
Gabriela Zurek1, Aiko Barsch1, Carsten Baessmann1, HartmutSchaefer2, Manfred Spraul21Bruker Daltonik GmbH, Fahrenheitstr. 428359 Bremen - Germany2Bruker BioSpin GmbH, Silberstreifen76287 Rheinstetten - Germany
Fig. 1 Metabolic Profiler consisting of liquid handler for sample preparation, Bruker 600MHz NMR spectrometer with flow injection and micrOTOFTM
LC/ESI-TOF-MS system.
The Metabolic Profiler™ introduced in 2005 combines a liquid handler, a 600 MHz NMR spectrometer with flow probe and a LC-ESI-(Q)TOF MS all under control of an oracle based order management system SampleTrackTM. The complete process with preparation and acquisition is fully automated. No manual data processing is needed. Statistical evaluation is done on the basis of PCA or other algorithms to extract all relevant information from a series of complex samples. The data is summarized in a single comprehensive reporting system.
The MetabolicProfilerTM
• ESI-TOF-MS of modified tobacco plants, M. Schöttner & I. Baldwin, MPI Chemical Ecology Jena, Germany**• Screening of myxobacterial extracts using UPLC & ESI-TOF-MS, D. Krug & R. Müller, Uni Saarbrücken, Germany [ACA, 624 (2008), p. 97–106].• GC/APCI-TOF-MS of cerebrospinal fluid, O. Mayboroda & A. Carrasco-Pancorbo, LUMC Leiden, Netherlands**• Personalized Metabolic Profiling using NMR, CEMET Florence, Italy• Newborn screening by high resolution flow-NMR, C. Fusch & M. Nauck, University Hospital Greifswald, Germany• Fruit Juice ScreenerTM SGF & Bruker BioSpin for Quality Assurance, P. Rinke, SGF Germany
Analytical Challenges in Metabolomics• Experimental design & Documentation & Reporting• Sample handling & throughput• Dynamic range & sensitivity• Reproducibility and long term stability• Chemical diversity of metabolites e.g. fatty acids, sugars • Diversity of matrices e.g. urine, cell extracts, plants• Data preprocessing & statistical evaluation• De-novo Identification & Structure Elucidation• Quantitation
MS Highlights & News
Fig. 2 The maXis UHR-QqTOF instrument.
The novel ultra high resolution (UHR)-Qq-TOF maXisTM provides the maximum of mass spectral information (resolution > 40000) on the UHPLC speed scale (max. scan rate 20Hz). Accurate MS and MS/MS data with mass accuracies better than 1ppm in combination with trueisotopic pattern are unique tools for analysis of complex samples and subsequent structural identification. Novel IonCooler™ tech-nology is delivering full width MS/MS sensitivity.
Spectral Database NMR & MSThe identification of putative markers is the major bottleneck in metabolomics nowadays. High quality spectral libraries of reference compounds are the key for fast dereplication of already known compounds. This enables the scientist to faster pinpoint really novel biomarker candidates. The NMR spectral database currently comprises 514 compounds (> 14599 spectra) measured from pH 3-8 in 0.5 pH unit steps. 2D-NMR spectra were acquired at pH= 3,5,7. The accurate MS/MS spectral library is composed of 170 compounds (> 1000 spectra) in ESI positive & negative mode. An important feature of the database is the combination of spectral, structural, chemical and physiological information.
Fig. 4 MS/MS and NMR spectral libraries providing spectral, structural, and physiological information.
NMR Highlights & NewsThe new AvanceIII console provides the latest developments achieved for digital RF generators and 2nd generation digital receivers. The patented digital RF generation guarantees most precise and stable RF pulses. In addition to the highest available phase and frequency resolution, RF parameters can be switched in only 25 ns, including simultaneous phase, frequency and amplitude switching. The 2nd generation digital receiver significantly increases dynamic range and digital resolution based on an ADC architecture, leading to over 22 usable bits. The AvanceII console guarantees stability and reproducibility combined with excellent spectral quality as demanded in metabolomics experiments.
Fig. 3 New AvanceIII console.
pH dependance in human urineAlanine Lactate 3-OH-Isoval. Ethanol
pH=3.0
pH=3.5
pH=4.0
pH=4.2
pH=4.4
pH=4.6
pH=4.8
pH=5.0
pH=5.2
pH=5.5
pH=5.8
pH=6.1
pH=6.2
pH=6.5
pH=7.0
pH=7.6
pH=8.0
pH dependance in human urineAlanine Lactate 3-OH-Isoval. Ethanol
pH=3.0
pH=3.5
pH=4.0
pH=4.2
pH=4.4
pH=4.6
pH=4.8
pH=5.0
pH=5.2
pH=5.5
pH=5.8
pH=6.1
pH=6.2
pH=6.5
pH=7.0
pH=7.6
pH=8.0
pH=3.0
pH=3.5
pH=4.0
pH=4.2
pH=4.4
pH=4.6
pH=4.8
pH=5.0
pH=5.2
pH=5.5
pH=5.8
pH=6.1
pH=6.2
pH=6.5
pH=7.0
pH=7.6
pH=8.0
MS/MS
meta-information
** Posters on display at this meeting.For research use only. Not for use in diagnostic procedures.
-19-
Metabolomics Technology: LC-MS-NMR
Conclusions
Bruker BioSpin & Bruker Daltonics
Markus Godejohann1, Silke Keller1, Hartmut Schäfer1, Christian Fischer1, Peter Neidig1, Gabriela Zurek2, Aiko Barsch2, Carsten Baessmann2, Manfred Spraul11Bruker BioSpin GmbH, Rheinstetten, Germany2Bruker Daltonik GmbH, Bremen, Germany
Fig. 1: Fully integrated LC-MS and NMR system including sample preparation.
Introduction
Methods
Integration and Robustness: a Touchstone in Metabolomics Studies
LC-MS/NMR
When integrating multiple analytical techniques like NMR & LC-MS, it is a key requirement to assess the robustness and reproducibility of the system to guarantee the generation of valid data for statistical analysis. A systematic study with two different human urine samples followed by series of blanks has been undertaken in a fully integrated LC-MS-NMR system including sample preparation.
The integrated platform consisted of:• oracle based sample administration and order management system SampleTrack (Bruker BioSpin, Germany) • Liquid Handler for Sample Preparation: All samples were automatically prepared in tubes or 96-well plates for NMR (buffered to pH=7) and diluted with water for LC-MS. Details on the off-line and on-line operation modes are given in Box 1.• LC-MS on ESI-TOF micrOTOF mass spectrometer (Bruker Daltonik, Bremen, Germany) in postive ionisation mode coupled to an HPLC system from Agilent (Waldbronn, Germany).• flow injection NMR on Bruker AVANCE 600 NMR Spectrometer (BEST-NMR, Bruker BioSpin, Rheinstetten, Germany)• automated data evaluation in AMIX software (Bruker BioSpin).
In NMR automation on the integrated platform, baseline correction is not necessary and phasing quality is excellent such that no manual processing steps are needed prior to further data analysis. Further it is important, that automated preparation excludes introduction of extra positional variability due to residual variation in buffer concentration.
The reproducibility of the overall system was evaluated under real sample conditions using two different human urine samples. The LC-MS and NMR data was evaluated using principle component analysis (PCA). A clear separation of the urines and the blanks is observed both in NMR and LC-MS as expected. Within-groupvariance contributions are small reflecting high reproducibility of the automation procedure.
Instrument status
Reproducibility
LC-MS Preparation:• 100µL sample + 100µL H2O• 8 Injections: 20µL loop overfill• Overlay of BPCs
NMR Preparation:• 360µL sample + 40µL buffer pH 7 (0.1% TSP)• 80 BEST-NMR experiments in automation
Time [min]5 6 7 8 9 10 110
1
2
3 RSD values of peak intensities
7.2%7.9%
7.4% 9.1%
LC-MSIntegralPrecision typically : RSD < 5%
TSP-Signal
NMR
Samples: Urine 1 (16 Aliquots); Urine 2 (32 Aliquots); Blank (8 Aliquots)Protocol for preparation &measurement:8 x Urine 1 - 8 x Urine 2 - 8 x Urine 1 - 8 x Urine 2 - 8 x Blank, Wait for approx. 200 min - 8 x Urine 2 Wait for approx. 100 min - 8 x Urine 2
Creatinine
Urine 1
Urine 2
Blank
Hippuric acid
LC-MS: on-line
Creatinine
Urine 1
Urine 2
Blank
NMR: on-line
Comparison of PCA results
• In summary, accuracy of repeated measurements in NMR and LC-MS is below 10% RSD including preparation.• Data from this integrated system is fully amenable to subsequent combined statistical analysis.
Off-line OperationLiquid Handler prepares samples• into sample tubes (NMR)• into 96-well plates (LC-MS)Manual transfer of• NMR tubes into sample changer• wellplate into LC-MS autosampler
On-line OperationLiquid Handler prepares samples• into cooled 96-well plates (BEST-NMR andLC-MS)Liquid handler • injects into NMR flow probe• acts as autosampler for LC-MS
The reproducibility of the entire preparation, injection and analysis process was investigated for LC-MS using a standard solution of four p-hydroxybenzoic acid esters (5 ng/uL), for NMR evaluating the integral of the internal standard D4-(Trimethylsilyl)propionic acid (TSP). The RSD values of peak intensities in the LC-MS chromatograms were < 3% without preparation and < 10% including the preparation. The RSD values of the retention time were < 0.5%. In NMR, the integral precision of the TSP signal was < 5%.
Box 1
NMR data quality and reprocibility under full automation, including sample preparation, flow injection, measurement and processing.Note, each spectrum corresponds to a newlyprepared and injected replicate sample from the same initial urine.
For research use only. Not for use in diagnostic procedures.
-20-
Metabolomics Technology: LC-MS-NMR
Metabolomics Posterbook 2011
Conclusions
LUMC & Bruker Daltonics
GC/APCI UHR-TOF-MSFor research use only. Not for use in diagnostic procedures.
Introduction
GC/APCI with ultra high resolution TOF-MS - analytical validation and applicability to metabolic profiling
Fig. 2 Standard Mixture for analytical method development covering different compound classes and polarity.
Tiziana Pacchiarotta1, Alegría Carrasco-Pancorbo1, Ekaterina Nevedomskaya1, Thomas Arthen-Engeland2, Gabriela Zurek2, Jacobus J. van Hilten3, Mike McDonell4, Andre M. Deelder1, Oleg A. Mayboroda1.
1LUMC, Biomolecular Mass Spectrometry Unit, Dept. of Parasitology, Leiden, NL2Bruker Daltonik GmbH, Bremen, D3LUMC, Dept. of Neurology, Leiden, NL4Bruker Biosciences, Delta, BC, Canada
Most of the commercial GC/MS systems use ionization under vacuum conditions: electron ionization (EI) and chemical ionization (CI). Atmospheric pressure ionization sources (API), which are probably the key of the “overnight success” of MS-detectors in analytical sciences due to coupling with liquid chromatography, are rarely used with GC instruments. A recently introduced multipurpose AP source [1] created the opportunity to reconsider the importance of AP ionization for GC. Here, we present an analytical evaluation of GC/APCI-MS showing the benefits of soft atmospheric pressure chemical ionization for GC in combination with an orthogonal-acceleration ultra high-resolution TOF mass spectrometer (UHR-TOF-MS) maXis. First data of clinical cerebrospinal fluid (CSF) samples from CRPS patients are presented to demonstrate the huge potential of GC/APCI-UHR-TOF-MS in metabolic profiling.
GC/APCI UHR-TOF-MS
Fig. 1 GC/APCI UHR-TOF MS instrument: Transfer line enlarged & source schematics.
[1] Schiewek et al (2008) Analytical Bioanalytical Chemistry, v.392, p.87
Strategy for Analytical Method Development
The first step towards a GC method applicable to metabolic profiling is the composition of a standard mixture covering simultaneously both a variety of compound classes found in biological samples as well as different polarity ranges & molecule sizes. The compound classes covered in the standard mixture are listed in Fig. 2. A chromatogram of the standard mixture is presented in Fig. 3. Excellent repeatability was obtained, with relative standard deviations (RSDs) of peak areas between 0.7% and 2.1% in the intra-day study, and between 3.8% and 6.4% in the inter-day study. The analytical method optimization as well as analytical parameters are described in detail in [2].
[2] Carrasco-Pancorbo et al (2009); Analytical Chemistry 81(24):10071-9
GC+
Standard Mix
Amines
Compound with Indoles group
Xanthines and related coumpounds
Alcohols
Organic Acids
Amino acids
Compounds with Imidazol groups
Compounds with hydroxyl and amine groups
Derivatization: methoxyamination + silylation with MSTFA
CSF Analysis & Profiling
Future directions• To continue the study of the analyses obtained for CSF samples to identify more compounds in the profile.
• To improve derivatization robustness by applying automated in-line derivatization.
• To create a GC/APCI-MS database for automatic compound assignment.
• The derivatization reaction is a crucial step in GC-MS analysis. It is mandatory to optimize its conditions in depth as it can be the major source of variation in complex matrices.
• During GC/APCI ionization of TMS derivatives the predominant ion observed is [M+H]+.
• The comparison of APCI and EI spectra reveals similarities and differences concerning presence/absence of the (pseudo) molecular ion and key fragments.
• The combination of GC/APCI with UHR-TOF MS directly enables the assignment of molecular formulae to the TMS derivatives and its fragments. This significantly facilitates spectra interpretation.
After optimization of the method using the standard mixture, the analysis of CSF real samples from patients suffering from Complex Regional Pain Syndrome (CRPS) was performed (see Fig. 5.) CRPS is a chronic pain condition characterized by a variety of sensory and autonomic disturbances. CRPS usually develops after an injury. However, the pathogenesis remains unclear and consequently no diagnostic laboratory tests do exist nowadays. Metabolic profiling, as performed here, can provide insight into mechanisms of disease and novel prognostic or diagnostic markers.In total, a set of 32 CSF samples was measured: 12 controls, 12 CRPS patients and one pooled CSF sample as quality control (QC, 8 technical replicates). The principle component analysis (PCA) of CSF samples from control and CRPS patients is presented in Fig. 6. The QC sample represents one identical pooled CSF sample. The observations are labeled by the order of measurement. Clearly, the variation is associated to the derivatization order in the batch.Therefore, another effort to optimize derivatization conditions and waiting times prior to analysis needs to be made. This is necessary to improve reproducibility and robustness of the method to a level acceptable for statistical analysis.
Fig. 5 Chromatogram of derivatized CSF sample.
Fig. 6 PCA analysis of CSF samples from control & CRPS patients, and CSF quality control sample.
-20
-15
-10
-5
0
5
10
15
20
-20 -15 -10 -5 0 5 10 15 20
Sec
ond
prin
cipa
l com
pone
nt
First principal component
controlCRPSQC
23
4
20
28
12
2
7
183115
26
10
24
11
6
19
8
22
3
3027
3216
14
1
5
9
13
172125
2933
SIMCA-P+ 12 - 2009-07-22 14:25:38 (UTC+1)
PC1: 21.7% explained variance
PC
2: 1
9.5%
exp
lain
ed v
aria
nce
PCA calculationbased on32 analyses340 features (XCMS)
-20
-15
-10
-5
0
5
10
15
20
-20 -15 -10 -5 0 5 10 15 20
Sec
ond
prin
cipa
l com
pone
nt
First principal component
controlCRPSQC
23
4
20
28
12
2
7
183115
26
10
24
11
6
19
8
22
3
3027
3216
14
1
5
9
13
172125
2933
SIMCA-P+ 12 - 2009-07-22 14:25:38 (UTC+1)
Legend
Material and MethodsCerebrospinal Fluid (CSF) Sample PreparationCSF extracts were prepared by adding cold methanol precipitation (total time 2 hours). Afterwards they were centrifuged and the supernatant was evaporated under N2.Derivatization reactionDerivatization reaction was based on a two step procedure: methoxyamination (60 min at 40ºC) and silylation with MSTFA+1%TMCS (30 min at 40ºC).Chromatographic methodThe samples (1µL) were injected in a HP-5-MS column (30 m, 0.25 mm ID, 0.25 m film) and analyzed by a temperature gradient of 5ºC/min over 57 min (oven initial T= 70ºC kept over 5 min).Mass spectrometry methodThe maXis was operated in APCI positive mode at 2.0bar nebulizer pressure, 200°C dry gas temperature (N2) and a dry gas flow rate of 4L/min. The scan range was set to 100-1000m/z, number of summations to 5000x.
Fig. 3 Separation of the derivatized standard mixture (100 M) (A) & assignment of trimethylsilyl-(TMS)-derivatives (B).
Fig. 4 Comparison of APCI and EI spectra of silylated A. valine and B. 5-hydroxyindole-3-acetic acid.
Valine + 2TMS(M.W. + 2TMS = 261 Dalton)
APCI[M+H]+
262.1659C11H28NO2Si2
172.1157C8H18NOSi
144.1207C7H18NSi
14473
218M+
?
EI
NH2
CHCH
CH3
COOH
CH3
A
5-Hydroxindole-3-acetic acid+3TMS(M.W. + 3TMS = 407 Dalton)
APCI[M+H]+
C19H34NO3Si3408.1843
290.1394C15H24NOSi2
264.1751C11H32OSi3
117.0707C9H9
EI290
407M+
73
NH
OH
COOHNH
OH
NH
OH
COOH
B
Results Standard MixtureA
B
9.20.29C19H22NNortriptyline +H327.7- 0.08C19H34NO3Si35-Hydroxyindole-3-acetic+3TMS+H315.6- 0.11C17H31N2O3Si2Phenyl-Gly+2TMS+H309.1- 0.18C17H37N4O3Si4Uric acid+4TMS+H294.0- 0.39C20H44NO2Si4Dopamine+4TMS+H284.20.61C18H38NO2Si34-Methyldopamine +3Si+H275.5- 0.12C18H36NO3Si3Tyrosine+3TMS+H262.51.00C18H47N2O2Si4Lysine+4TMS+H252.9- 0.10C10H17N4O2SiTheophylline+1TMS+H243.20.10C10H15O2SiCaffeine+H231.80.30C12H8NO3SiHippuric acid+1TMS+H223.3- 0.29C11H15N2O3Phenyl-Gly+H215.80.00C15H28NO2Si2Phenylalanine+2TMS+H205.50.40C14H34NO4Si3Glutamic acid+3TMS+H191.60.32C13H32NO4Si3Aspartic acid+3TMS+H181.3- 0.20C11H28NO2SSi2Methionine+2TMS+H172.90.71C13H34NO3Si3Threonine+3TMS+H165.10.29C12H32NO3Si3Serine+3TMS+H144.70.09C11H30NO2Si3Glycine+3TMS+H146.40.80C12H30NO2Si2Leucine+2TMS+H136.0- 0.30C12H30NO2Si2Isoleucine+2TMS+H121.6- 0.10C9H24NO3Si2Serine+2TMS+H111.80.70C8H11N4O2Benzoic acid+1TMS+H106.1- 0.29C11H28NO2Si2Valine+2TMS+H93.40.31C14H29N4O3Si3Uric acid+3TMS+H81.80.49C9H22NO2SiLeucine+1TMS+H72.2- 0.70C8H18NO2SiProline+1TMS+H61.80.00C9H22NO2SiIsoleucine+1TMS+H54.40.21C9H24NO2Si2Sarcosine+2TMS+H44.60.31C8H22NO2Si2Glycine+2TMS+H34.50.20C9H24NO2Si2Alanine+2TMS+H23.00.21C8H20NO2SiValine+1TMS+H1
mSigmaValue
Error (mDa)
Formula (peak found)
Compound Name + Silylation Rate + Ionization
Peak ID
9.20.29C19H22NNortriptyline +H327.7- 0.08C19H34NO3Si35-Hydroxyindole-3-acetic+3TMS+H315.6- 0.11C17H31N2O3Si2Phenyl-Gly+2TMS+H309.1- 0.18C17H37N4O3Si4Uric acid+4TMS+H294.0- 0.39C20H44NO2Si4Dopamine+4TMS+H284.20.61C18H38NO2Si34-Methyldopamine +3Si+H275.5- 0.12C18H36NO3Si3Tyrosine+3TMS+H262.51.00C18H47N2O2Si4Lysine+4TMS+H252.9- 0.10C10H17N4O2SiTheophylline+1TMS+H243.20.10C10H15O2SiCaffeine+H231.80.30C12H8NO3SiHippuric acid+1TMS+H223.3- 0.29C11H15N2O3Phenyl-Gly+H215.80.00C15H28NO2Si2Phenylalanine+2TMS+H205.50.40C14H34NO4Si3Glutamic acid+3TMS+H191.60.32C13H32NO4Si3Aspartic acid+3TMS+H181.3- 0.20C11H28NO2SSi2Methionine+2TMS+H172.90.71C13H34NO3Si3Threonine+3TMS+H165.10.29C12H32NO3Si3Serine+3TMS+H144.70.09C11H30NO2Si3Glycine+3TMS+H146.40.80C12H30NO2Si2Leucine+2TMS+H136.0- 0.30C12H30NO2Si2Isoleucine+2TMS+H121.6- 0.10C9H24NO3Si2Serine+2TMS+H111.80.70C8H11N4O2Benzoic acid+1TMS+H106.1- 0.29C11H28NO2Si2Valine+2TMS+H93.40.31C14H29N4O3Si3Uric acid+3TMS+H81.80.49C9H22NO2SiLeucine+1TMS+H72.2- 0.70C8H18NO2SiProline+1TMS+H61.80.00C9H22NO2SiIsoleucine+1TMS+H54.40.21C9H24NO2Si2Sarcosine+2TMS+H44.60.31C8H22NO2Si2Glycine+2TMS+H34.50.20C9H24NO2Si2Alanine+2TMS+H23.00.21C8H20NO2SiValine+1TMS+H1
mSigmaValue
Error (mDa)
Formula (peak found)
Compound Name + Silylation Rate + Ionization
Peak ID
The separation of the standard mixture is shown in Fig. 3, while Fig. 4 is giving more details on the different ionization behaviour of TMS derivatives at EI and APCI conditions. In both examples, similar key fragments (e.g. 144m/z for valine) derived from in-source CID can be observed for both ionization techniques. In the APCI spectra, the EI fragment of 73m/z is not observed. The big advantage of APCI ionization in combination with an accurate mass analyzer is the simple assignment of the molecular formulae to both the pseudomolecular ion [M+H]+ and the fragments as demonstrated in Fig. 4.
-21-
Metabolomics Technology: GC-APCI-TOF-MS
Co
ncl
usi
on
s
Bru
ker
Dalt
on
ics
Intr
od
uct
ion
Meth
od
s
Su
mm
ary
Resu
lts
In t
he
pas
t se
vera
l ye
ars,
ther
e has
bee
n
incr
easi
ng e
viden
ce o
f th
e an
tiox
idan
t ca
pac
ity
of
pro
duct
s der
ived
fro
m b
ees.
Polle
n is
a hiv
e der
ived
pro
duct
of
gre
at c
om
mer
cial
inte
rest
ow
ing t
o its
hig
h n
utr
itio
nal
qual
ity
and c
an b
e co
nsi
der
ed a
pote
ntial
sou
rce
of
antiox
idan
ts.
In t
his
work
, a
new
, ea
sy a
nd r
apid
cap
illar
y el
ectr
ophore
sis
elec
trosp
ray
tim
e-of-
flig
ht
mas
s sp
ectr
omet
ry (
CE-E
SI-
TOF-
MS)
pro
cedure
for
anal
yzin
g p
hen
olic
com
pounds
in
polle
n h
as b
een d
evel
oped
. The
flora
l polle
n
has
bee
n o
bta
ined
fro
m s
elec
ted h
oney
bee
-co
llect
ed p
olle
n p
elle
ts m
ade
up b
y Ran
uncu
lus
pet
iola
ris
HKB (
Ran
uncu
lace
ae)
polle
n g
rain
s.
It is
dem
onst
rate
d t
hat
a f
ine
optim
izat
ion o
f se
vera
l se
par
atio
n p
aram
eter
s is
ess
ential
in
ord
er t
o a
chie
ve s
uitab
le C
E-E
SI-
TOF-
MS
anal
ysis
of
poly
phen
ols
in n
atura
l pro
duct
s.
Twen
ty g
ram
s of
polle
n p
elle
ts w
ere
extr
acte
d
five
tim
es w
ith e
than
ol:
wat
er (
1:1
) w
ith 6
0
min
mac
erat
ion.
The
extr
acts
wer
e se
par
ated
by
centr
ifugat
ion (
15269 g
) fo
r 10 m
in.
All
super
nat
ants
wer
e co
mbin
ed a
nd e
vapora
ted
to d
rynes
s at
low
pre
ssure
. Anal
ysis
was
car
ried
out
with a
CE s
yste
m
(Agile
nt,
San
ta C
lara
) co
uple
d t
o a
mic
rOTO
F ESI-
TOF
syst
em (
Bru
ker
Dal
tonik
, G
erm
any)
under
the
follo
win
g o
ptim
ized
conditio
ns:
•
Unco
ated
fuse
d-s
ilica
cap
illar
y co
lum
n100
cm x
50 µ
m I
.D,
375 µ
m O
.D.
•Runnin
g b
uffer
80 m
M a
mm
oniu
m
acet
ate/
amm
onia
, pH
10.5
.•
Voltag
e 30 k
V.•
Inje
ctio
n t
ime
15 s
.•
MS a
nal
yses
ESI
neg
ativ
e pola
rity
.
•Shea
th liq
uid
2-p
ropan
ol/
wat
er 6
0:4
0 w
ith
0.1
% (
v/v)
TEA;
flow
rat
e 6
L/m
in.
•ESI:
dry
gas
4 L
/min
at
130ºC
; neb
uliz
er g
as
pre
ssure
of
0.3
bar
. •
Ext
ernal
cal
ibra
tion w
ith s
odiu
m f
orm
ate.
•Soft
war
e: G
ener
ateM
ole
cula
rForm
ula
TM
within
D
ataA
nal
ysis
TM.
A n
ovel
CE-E
SI-
TOF-
MS m
ethodolo
gy
for
the
char
acte
riza
tion o
f m
any
wel
l-kn
ow
n p
hen
olic
co
mpounds
pre
sent
in p
olle
n s
ample
s in
les
s th
an 1
7 m
in h
as b
een d
evel
oped
. Fo
r th
e id
entifica
tion o
f th
e phen
olic
com
pounds
in
polle
n e
xtra
cts,
acc
ura
te m
ass
dat
a of
the
pse
udom
ole
cula
r an
ions
from
ESI-
TOF-
MS w
as
eval
uat
ed.
•The
separ
atio
n b
y CE-E
SI-
TOF-
MS is
an a
ttra
ctiv
e m
ethod f
or
the
char
act
eriz
atio
n o
f phen
olic
co
mpounds
pre
sent
in c
om
ple
x m
atrice
s su
ch a
s polle
n.
•ESI-
TOF-
MS p
rovi
din
g e
xcel
lent
mas
s ac
cura
cy a
s w
ell as
iso
topic
pat
tern
info
rmat
ion h
as b
een
succ
essf
ully
applie
d t
o iden
tify
se
vera
l phen
olic
com
pounds
in
polle
n.
The
phen
olic
com
pounds
wer
e id
entified
ac
cord
ing t
o t
hei
r m
ole
cula
r fo
rmula
usi
ng t
he
exac
t m
ass
spec
tra
and iso
topic
pat
tern
in
form
atio
n (
Tab.
1).
Mas
s dev
iations
wer
e sm
alle
r th
an 3
ppm
with a
good iso
topic
pat
tern
fit a
s in
dic
ated
by
the
sigm
a va
lues
. A
mas
s sp
ectr
um
of
Quer
cetin-3
-Rutinosi
de
is
dep
icte
d in F
ig.
3 a
s w
ell as
the
sim
ula
ted
spec
trum
for
the
pse
udo-m
ole
cula
r an
ion.
Char
acte
rist
ic fra
gm
ents
fro
m t
he
conju
gat
ed
sugar
moie
ties
are
obse
rved
.
Fig.
1:B
PE o
f polle
n e
xtra
ct b
y CE-E
SI-
TOF-
MS
at t
he
optim
ized
conditio
ns.
Fig.
2:
EIE
s (±
5m
Da)
of
the
wel
l-kn
own
phen
olic
com
pounds
det
ecte
din
polle
n.
Table
1:
Phen
olic
com
pounds
det
erm
ined
by
CE-E
SI-
TOF-
MS in a
n e
xtra
ct o
f polle
n.
CE-E
SI-
TOF-
MS
an
d U
niv
ers
ity o
f G
ran
ad
a
Ch
ara
cteri
zati
on
of
po
lyp
hen
ols
of
pro
du
cts
deri
ved
fro
m b
ees
usi
ng
CE
-ES
I-TO
FD
avid
Arr
aez-
Rom
an
1
Gab
riela
Zu
rek
2
Cars
ten
Baess
man
n2
An
ton
io S
eg
ura
-Carr
ete
ro1
Alb
ert
o F
ern
án
dez-
Gu
tiérr
ez1
1U
niv
ersi
ty o
f G
ranad
a, G
ranad
a, S
pai
n.
2Bru
ker
Dal
tonik
Gm
bH
, Bre
men
, G
erm
any.
0.02.5
5.07.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Time [m
in]0.00.51.01.5
6x10
Inten
s.
BPE 5
0-800
–All M
s80
mMNH
4OAc
pH=1
0.5
Calib
rant
0.02.5
5.07.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Time [m
in]0.00.51.01.5
6x10
Inten
s.
0.02.5
5.07.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Time [m
in]0.00.51.01.5
6x10
Inten
s.
BPE 5
0-800
–All M
s80
mMNH
4OAc
pH=1
0.5
Calib
rant
Cal
ibra
nt
#C
om
pou
nd
Form
ula
Ion
m/
z exp
eri
men
tal
m/
z ca
lcu
late
d
Mass
Devia
tio
nS
igm
a
Valu
eC
E m
igra
tio
n
tim
e (
min
)(p
pm
)(m
Da)
1A
ceti
n-G
luco
sid
eC
22H
19O
10
[M-H
]-443.0
989
443.0
983
1.2
-0.5
50.0
311
9.6
2G
all
oyl-
Glu
cose
C13H
15O
10
[M-H
]-331.0
661
331.0
670
2.8
0.9
30.0
862
13.6
3A
pig
en
in-6
,8-d
i-C
-Gly
cosi
de
C27H
29O
15
[M-H
]-593.1
519
593.1
511
1.3
-0.7
50.0
187
15.0
4Q
uerc
eti
n-3
-Ru
tin
osi
de
C27H
29O
16
[M-H
]-609.1
468
609.1
461
1.2
-0.7
30.0
523
15.5
5G
en
iste
in 7
-O-β
-D-G
luco
sid
eC
21H
19O
10
[M-H
]-431.0
975
431.0
983
1.8
0.7
80.0
290
16.4
6Lu
teo
lin
-7-O
-Glu
cosi
de
C21H
19O
11
[M-H
]-447.0
924
447.0
932
1.9
0.8
30.0
493
16.7
7A
pig
en
in-7
-O-G
luco
sid
e
C21H
19O
10
[M-H
]-431.0
974
431.0
983
2.1
0.8
90.0
347
16.8
82
',4
',6
'-Tri
hyd
roxy-3
'-fo
rmyld
ihyd
roch
alc
on
eC
16H
13O
5[M
-H]-
285.0
765
285.0
768
0.9
0.2
60.0
293
17.0
Initia
lly,
the
CE-M
S c
onditio
ns
wer
e optim
ized
ac
cord
ing t
o t
he
follo
win
g c
rite
ria:
res
olu
tion,
sensi
tivi
ty,
anal
ysis
tim
e an
d p
eak
shap
e.U
nder
thes
e co
nditio
ns,
CE-E
SI-
TOF-
MS
reco
rds
such
as
the
one
giv
en in F
ig.
1 a
nd 2
w
ere
obta
ined
for
the
polle
n e
xtra
ct.
Fig.
1 is
show
ing t
he
bas
e pea
k el
ectr
opher
ogra
m (
BPE
) of
the
polle
n e
xtra
ct in E
SI
neg
ativ
e m
ode.
The
calib
rant
pea
k is
mar
ked a
t th
e en
d o
f th
e ru
n.
In F
ig.
2,
extr
acte
d ion e
lect
ropher
ogra
ms
(EIE
) ar
e sh
ow
n f
or t
he
iden
tified
poly
phen
ols
.
Fig.
3:
Mea
sure
d a
nd s
imula
ted m
ass
spec
trum
of
Quer
cetin-3
-Rutinosi
de.
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
0.0
0.2
0.4
0.6
0.8
1.04
x10
Inte
ns.
EIE 3
31
.06
6 ±
0.0
05
C1
3H
15
O1
0
EIE 4
31
.09
7 ±
0.0
05
C2
1H
19
O1
0
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0T
ime
[min
]0
2000
4000
6000
Inte
ns.
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
0
2000
4000
6000
8000
Inte
ns.
EIE 2
85
.07
6 ±
0.0
05
C1
6H
13
O5
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
0.0
0.5
1.0
1.5
2.0
2.54
x10
Inte
ns.
EIE 6
09
.14
6 ±
0.0
05
C2
7H
29
O1
6
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0T
ime
[min
]0.
00
0.2
5
0.5
0
0.7
5
1.0
0
1.2
54x1
0In
ten
s.
EIE 4
43
.09
8 ±
0.0
05
C2
2H
19
O1
0
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
01234x1
0In
tens.
EIE 5
93
.15
1 ±
0.0
05
C2
7H
29
O1
5
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0T
ime
[min
]0
100
0
200
0
300
0
400
0
500
0
Inte
ns.
EIE 4
47
.09
2 ±
0.0
05
C2
1H
19
O1
1
12
34
57
6
8
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
0.0
0.2
0.4
0.6
0.8
1.04
x10
Inte
ns. 0.
02.
55.
07.
510
.012
.515
.017
.520
.022
.525
.0T
ime
[min
]0.
0
0.2
0.4
0.6
0.8
1.04
x10
Inte
ns.
EIE 3
31
.06
6 ±
0.0
05
C1
3H
15
O1
0
EIE 4
31
.09
7 ±
0.0
05
C2
1H
19
O1
0
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0T
ime
[min
]0
2000
4000
6000
Inte
ns. 0
.02.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0T
ime
[min
]0
2000
4000
6000
Inte
ns.
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
0
2000
4000
6000
8000
Inte
ns. 0
.02.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
0
2000
4000
6000
8000
Inte
ns.
EIE 2
85
.07
6 ±
0.0
05
C1
6H
13
O5
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
0.0
0.5
1.0
1.5
2.0
2.54
x10
Inte
ns. 0.
02.
55.
07.
510
.012
.515
.017
.520
.022
.525
.0T
ime
[min
]0.
0
0.5
1.0
1.5
2.0
2.54
x10
Inte
ns.
EIE 6
09
.14
6 ±
0.0
05
C2
7H
29
O1
6
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0T
ime
[min
]0.
00
0.2
5
0.5
0
0.7
5
1.0
0
1.2
54x1
0In
ten
s. 0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0T
ime
[min
]0.
00
0.2
5
0.5
0
0.7
5
1.0
0
1.2
54x1
0In
ten
s.
EIE 4
43
.09
8 ±
0.0
05
C2
2H
19
O1
0
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
01234x1
0In
tens.
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
Tim
e [m
in]
01234x1
0In
tens.
EIE 5
93
.15
1 ±
0.0
05
C2
7H
29
O1
5
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0T
ime
[min
]0
100
0
200
0
300
0
400
0
500
0
Inte
ns. 0.
02.
55.
07.
510
.012
.515
.017
.520
.022
.525
.0Tim
e [m
in]
0
100
0
200
0
300
0
400
0
500
0
Inte
ns.
EIE 4
47
.09
2 ±
0.0
05
C2
1H
19
O1
1
12
34
57
6
8
119.
0350
89
151.
0461
14195.
0510
38
257.
0495
80
391.
1082
52
609.
1468
34
-MS
, 15.
5min
#93
0
024685x1
0In
tens
.
100
200
300
400
500
600
700
800
m/z
60
9.1
46
10
8 61
0.1
49
51
3 61
1.1
51
82
1
C2
7H
29
O1
6,
M ,
60
9.1
5
0
50
0
10
00
15
00
20
00
25
00
Inte
ns.
10
02
00
30
04
00
50
06
00
70
08
00
m/z
The
auth
or
DAR g
rate
fully
ack
now
ledges
the
finan
cial
support
of
a post
doct
ora
l co
ntr
act
from
the
Univ
ersi
ty o
f G
ranad
a (P
erfe
ccio
nam
iento
de
Doc
tore
s) in B
ruke
r D
alto
nik
Gm
bH
.
For
rese
arch
use
only
. N
ot
for
use
in d
iagnost
ic p
roce
dure
s.
-22-
Metabolomics Technology: CE-TOF-MS
Metabolomics Posterbook 2011
Conclusions
Bruker Daltonics
Sandy Yates1; Aiko Barsch2; Gabriela Zurek2; Ilmari Krebs2
1Bruker Daltonics, Fremont, CA2Bruker Daltonik GmbH, Bremen, Germany;
• ESI-Qq-TOF mass spectrometers providing accurate MS, MS/MS and IS-CID-MS3 data are important tools in structure elucidation.
• The isotopic pattern plays an important role in the generation of formulae for MS to MSn spectra.
• The automated generation and con-solidation of proposed formulae for observed ions and neutral losses increases the confidence in resulting formulae for precursor and product ions
• This significantly accelerates the interpretation of MS and MS/MS spectra.
Introduction
Fully Unsupervised Automatic Assignment and Annotation of Sum Formulae for Product Ion Peaks, Neutral Losses in MS and Product Ion Spectra
ESI-Qq-TOF
An essential component on the process of characterizing chemical unknowns, is the inspection of collision-induced dis-sociation (CID) mass spectra from quadrupole/time-of-flight (Q-TOF) in-struments. The high resolution and superior mass accuracy is key to establish the elemental composition of molecules. Unfortunately with increasing m/z values the number of possible sum formulae grows exponentially.3,5 Several techniques have been developed which use the information from MS/MS spectra as additional criterion for reducing the number of possible formulae for the precursor ion, by summing up the potential formulae for product ion and neutral loss to establish the identity of the precursor ion.1-6 Herein we present a novel software module which implements a different route, which is exemplified with the medium sized compound Paclitaxel (I).
Experimental• Infusion (3 µL/min) of standards (I) (~1 ng/µL)• Mass spectrometer: micrOTOF-QTM ESI-Qq-TOF• Ionization: ESI positive mode• Scan range 50-1000 m/z• Mass spectra: MS fullscan, MS/MS, IS-CID-MS3
• MS/MS collision energies: 0-50 eV• External recalibration of all spectra types on MS full scan spectrum
O OH
O
O
O
O
N
H
O
OOO
H
OH
O
O
H
OH
M-Series
S-Series T-Series
Scheme 1: Fragmentation pathway of I. Blue, bold ions has been subjected to fragmentation (MS/MS and isCID-MS3)
Fig. 1: Paxlitaxel (I) with indication of origin of fragments.6
854
836
369
551
758
309
716
776
794
569
449
431
431
491
327
387
509
655
-C16H15NO4
-C2H4O2
-C2H4O2
-C2H4O2
-C2H4O2
-C7H6O2 -H2O
-H2O
-H2O
-H2O
-H2O
-C16H
15N
O4
-C7H6O2
-C2H4O2
-C16H
15N
O4
-C2H4O2
-C2H4O2
-C2H4O2
-C7H6O2
-C9H6O2
-C7H7NO
-C2H2O
-C16H15NO4
T-SeriesM-Series
268
286
118 105
240
-H2O
-CO
-C7H
6O
2
-C8H
9N
O
-C31H36O10
S-Series
-C7H6O2
-C7H6O2
1)Kumar K., et al., Rapid Commun. Mass Spectrom. 1992, 6, 585-591
2)Tenhosaari A., Anal. Chim. Acta 1991, 248, 71-753)Grange A., et al., Rapid Commun. Mass Spectrom.
2006, 20, 89-1024)McDonald L., et al., Anal. Chem. 2003, 75, 2730-
27395)Kaufmann A., Rapid Commun. Mass Spectrom.
2007, 21, 2003-20136)Konishi Y., et al., Anal. Chem. 2007, 79, 1187-1197
References
Precursor m/z meas m/z calc err [mDa] mSigma
854 C 47 H 52 N 1 O 14 854.3390 854.3382 -0.8 4.1
Fragment m/z meas m/z calc err [mDa] mSigma
836 C 47 H 50 N 1 O 13 836.3280 836.3277 -0.3 15.7
794 C 45 H 48 N 1 O 12 794.3174 794.3171 0.7 6.9
758 C 45 H 44 N 1 O 10 758.2949 758.2960 1.0 25.2
776 C 45 H 46 N 1 O 11 776.3067 776.3065 -0.1 15.6
716 C 43 H 42 N 1 O 9 716.2835 716.2854 1.9 18.3
655 C 38 H 39 O 10 655.2575 655.2538 -1.6 16.3
569 C 31 H 37 O 10 569.2383 569.2381 -0.3 1.5
551 C 31 H 35 O 9 551.2279 551.2276 0.1 0.5
509 C 29 H 33 O 8 509.2170 509.2170 0.0 3.2
491 C 29 H 31 O 7 491.2067 491.2064 2.0 11.7
449 C 27 H 29 O 6 449.1963 449.1954 -0.4 7.0
431 C 20 H 27 O 5 431.1865 431.1853 -1.2 31.6
387 C 22 H 27 O 6 387.1801 387.1802 0.1 11.9
327 C 20 H 23 O 4 327.1590 327.1591 0.1 19.8
309 C 20 H 21 O 3 309.1483 309.1485 1.1 19.8
286 C 16 H 16 N 1 O 4 286.1074 286.1074 0.0 8.1
268 C 16 H 14 N 1 O 3 268.0972 268.0968 1.2 42.9
240 C 15 H 14 N 1 O 2 240.1033 240.1019 -1.4 11.4
118 C 8 H 8 N 1 118.0645 118.0651 0.6 3.7
105 C 7 H 5 O 1 105.0351 105.0335 -1.6 5.5
Table 1: Sum formulae for fragments shown in scheme 1. Blue, bold ions have been subjec-ted to MS/MS and isCID-MS3 fragmentation.
ASMS 2008, MPG152
Algorithm
The elucidation multiple precursor formulae and corresponding product ion formulae is pursued in two steps:
1.Generate all possible formulae for the precusor ion, as well as for all product ions, applying restrictive filters mass accuracy, formula boundaries, rings plus double bond equivalents and SigmaFitTM
thresholds.2.Filter out all product ion formulae which
are not a subset of the precursor ion formulae. Crosscheck every potential pair with neutral loss candidates. Eliminate all formulae not belonging to a triple consisting of formulae for precusor ion, product ion, and neutral loss.
If higher MSn data are available, they can be used for reducing recursivley the sets of possible candidate formulae.
Fig. 2: MS/MS spectrum of I with annotation of losses derived from isCID-MS3 spectra as indicated on the left hand side. The peaks in the spectrum can be associated mainly to fragments resulting from losses of side chains from the four membered core structure. S- and T series are initiated by breakage of the ester functionality connecting the left hand side chain with the core, whereas the M series starts by losses of water or acetic acid. The T series and the M series eventually join their paths at the common fragment 509 and finally reach fragment 309 which represents the four membered cylic core structure after loss of nearly all side chains.
286 50
9
55
15
69
77
6
85
4
+MS2(854.3)
0 100 200 300 400 500 600 700 800 m/z
C8H6O3H3N
C8H6O3C8H6O
H2OH3N
C9H11NO3
H2OC7H6O2
C2H4O2
C2H4O2C7H6O2 H2OC7H6O2C2H4O2H2O
C9H6O2
C7H6O2
C2H4O2C2H4O2
C31H34O9C16H15NO4 C7H6O2
C2H4O2
C29H32O8C16H15NO4
C7H6O2
C29H30O7
C2H4O2C16H15NO4 C7H7NO
C2H2OC16H15NO4H2O
C2H4O2 C2H4O2
C2H4O2C7H6O2H2OH2O
C2H4O2C7H6O2
C7H6O2C2H4O2
C22H26O6
C7H6O2
C10H9NO3C8H9NO
C7H6O2 CO
HCN
C31H36O10
C16H15NO4H2O
CO
79
4
83
6
65
5
240
286
491
509
551
569
655
776
794
836
854
For research use only. Not for use in diagnostic procedures.
-23-
Metabolomics Data evaluation, Identification of unknowns
Conclusions
Bruker Daltonics
Sum Formula Calculation and Identification of a Bacterial Metabolite with m/z > 1100Wiebke Lohmann, Petra Decker, Aiko Barsch, Gabriela ZurekBruker Daltonik GmbH, Fahrenheitstraße 4, 28359 Bremen, Germany
For research use only. Not for use in diagnostic procedures.
ESI-Qq-TOF-MS
0
10
20
30
40
50
60
70
80
90
100
50 100 150 200 250 300 350 400 450 500
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
50 100 150 200 250 300 350 400 450 m/z
# o
f su
m f
orm
ula
e
5 ppm 2 ppm 1 ppm 0.5 ppm 0.1 ppm 0.05 ppm
Fig. 1: Theoretical number of calculated sum formulae in dependence on compound mass and mass accuracy (search range: C200 H400 N50 O50 S5 P5 F5 Cl5 Br5 I5, even electron configuration, no further constraints).
b)
1153.6838
1170.7112
1175.6665
1191.6397
+NH4 - 1.0069
+Na - 1.0071
+K - 1.0077
1. +MS, 3.31-3.64min #(194-214)
0.0
0.5
1.0
1.5
2.0
2.5
5x10
Intens.
1155 1160 1165 1170 1175 1180 1185 1190 m/z
172.1320
357.2379
499.2999
556.3567
698.4184
741.4618
883.5223
940.5803
1125.6856
1175.6648
1. +MS2(1171.0000), 54-66eV, 3.51-3.83min #(205-224)
0.00
0.25
0.50
0.75
1.00
1.25
4x10Intens.
200 400 600 800 1000 m/z
a)
[M+H]+
Fig. 3: Sum formula calculations for m/z 1170.7112 using a calculation window of 5 ppm and chemical constraints: The number of sum formula suggestions depends extremely on the predefined element range.
Fig. 5: Principle of SmartFormula3DTM and result formicrOTOF-Q data (5 ppm calculation window): only sixsum formula candidates left.
Fig. 5: Online search result for the sum formula suggestions from SF3D: only for one compounds and structures can be assigned.
result no. 1
Fig. 6: Principle of SmartFormula3DTM and result for maXis data (1 ppm calculation window): only two sum formula candidates left.
a)
1170.7112
1171.7142
1172.71661173.7186
1. +MS, 3.31-3.64min #(194-214)
1170.7119
1171.7152
1172.71861173.7220
1. C 57 H 100 N 7 O 18 ,1170.71
1170.7079
1171.7112
1172.71451173.7179
1. C 52 H 100 N 9 O 20 ,1170.71
1170.7087
1171.7121
1172.71541173.7188
1. C 68 H 100 N O 15 ,1170.71
1170.70821171.7116
1172.7149
1173.71831174.7216
1. C 81 H 92 N 3 O 4 ,1170.71
0.0
0.5
1.0
1.5
5x10
Intens.
0.0
0.5
1.0
1.5
2.05x10
0.0
0.5
1.0
1.5
2.05x10
0.0
0.5
1.0
1.5
2.05x10
0.0
0.5
1.0
1.5
2.05x10
1170 1171 1172 1173 1174 1175 m/z
experimental
simulatedpatterns
b)a)
IntroductionIn the course of identification and structural characterization of compounds, the unambiguous sum formula calculation is an essential step. With increasing molecular mass, the number of possible sum formulae is increasing exponentially, so that already for m/z values around 500 it is not possible anymore to get just one sum formula suggestion from an accurate mass measurement only, even if a mass accuracy of 0.1 ppm is assumed (Fig. 1).
Calculation constraints like ranges for meaningful H/C ratios, number of rings/double bonds and careful selection of elements allowed for sum formula calculation will reduce the number of suggestions significantly and help to exclude chemically unlikely molecules from the result list. However, for m/z 1000 at 0.1 ppm mass accuracy more than 2200 sum formulae would still be calculated (>10.000 for mass accuracy 1 ppm).
On the other hand, the combination of accurate mass information with additional information like isotopic pattern derived both from full scan and MS/MS spectra can significantly extend the m/zrange for reliable sum formula suggestion to values higher than 1000.
MethodsHPLC: Ultimate 3000 Rapid Separation LC ("RSLC", Dionex). Column: Zorbax XDB-C8 2.1x50 mm, 1.8 µm (Agilent).Column temp. 30 °C. Mobile phase: A = H2O, B = MeOH (each 0.1% HCOOH). Gradient: linear gradient 85 – 99.5% in 3 min, hold 5 min. Flow rate: 0.2 mL/min. Injection volume: 1 µL. Sample: bacterial metabolite, provided by CVUA Stuttgart. MS: micrOTOF-Q II, maXis (ESI-Qq-TOF MS, UHR-TOF MS, Bruker Daltonik GmbH). Ionization: ESI(+). Scan range: m/z 50-1300. Scan mode: Fullscan, MSMS 1171.Calibration: external: sodium formate clusters injected at the beginning of each chromatographic run. For maXis: additional internal calibration, m/z1221.990637 (Hexakis(1H,1H,4H-hexafluoro-butyloxy)phosphazine, C24H19O6N3P3F36).
A solution containing a bacterial metabolite that gave an MS signal at about m/z 1171 was analyzed using a reversed phase gradient separation on an UHPLC system interfaced to an ESI-TOF-MS or an ultrahigh-resolution ESI-TOF-MS system. Full scan data was acquired in ESI positive mode (scan range 50-1300 m/z). An additional run in MS/MS mode was performed, acquiring full scan MS/MS spectra for a precursor mass of m/z 1171. Sodium formate solution was injected as external calibration standard at beginning of each chromatographic run. Sum formula suggestions were calculated using DataAnalysisTM 4.0 SR2 (Bruker Daltonik GmbH) applying different settings.
ResultsFull scan mass spectra on the micrOTOF-Q II gave an exact mass value of m/z 1170.7112 for the bacterial metabolite with a mass accuracy of 5 ppm according to instrument specification for external mass calibration. Additional MS signals of lower intensity allowed for the characterization of the observed ion as ammonium adduct (presence of [M+H]+, sodium and potassium adducts, Fig. 2a). Allowing only C, H, N, O as elements 30 sum formula suggestions were calculated using the 5 ppm mass accuracy window (Fig. 3a).
If additional elements were allowed, the number of theoretical suggestions increased rapidly to several hundred, even if rules were applied, which filter the results for example for validity of the nitrogen rule, electron configuration, reasonable C/H ratios or a minimum number of C atoms according to the intensity of the second isotope (Fig. 3b).
An additional similar analysis using an UHR-TOF-MS system (maXis) showed the benefit of enhanced mass accuracy (1 ppm for internal mass calibration), since this approach only left one sum formula in addition to the cereulide formula for combined mass, isotope and fragment information (Fig. 6).
Fig. 2: Fullscan (a) and MSMS spectrum (b) of the metabolite. The adduct pattern observed in the fullscan spectrum allows for a clear assignment of the [M+H]+
signal.
The data evaluation software allowed the rating of results according to the matching of experimental and theoretical isotope patterns. Taking isotope pattern information into account the number of meaningful suggestions was reduced to about 12 (Fig. 3a).
Acknowledgment
Many thanks to Roland Perz (CVUA Stuttgart, Germany) for providing this sample!
fragment sum formulae of the selected suggestion
MS/MS-peaklist
sum formula suggestions for the precursor ion after
combination of fullscan and MS/MS information
MS peaklist(precursor
region only)
Fig. 4: Principle of SmartFormula3DTM and result formicrOTOF-Q data (5 ppm calculation window): only sixsum formula candidates left.
• Combination of information from MS and MS/MS spectra (accurate mass and isotope pattern) together with a query in online available databases allows for sum formula generation and identification of compounds even at high m/z values > 1000.
The additional tool Smart Formula 3DTM allowed for the verification of sum formula suggestions by combining full scan MS and MS/MS data, taking also the fragments' accurate mass and isotope pattern information into account. Only sum formulae remain, which can explain a specified minimum number of fragment signals. This reduced the sum formula candidate list to six (Fig. 4).
Only one of these five sum formulae could be found in online available databases (like ChemSpider), using the CompoundCrawler software tool (Fig. 5). Thus the bacterial metabolite could putatively be identified as cereulide.
fragment sum formulae of the selected suggestion
MS/MS-peaklist
sum formula suggestions for the precursor ion after
combination of fullscan and MS/MS information
MS peaklist(precursor
region only)
-25-
Metabolomics Data evaluation, Identification of unknowns
Metabolomics Posterbook 2011
Conclusions
University of Granada & Bruker Daltonics
Introduction
Summary
Rapid and automated identification of phenolic compounds in food using LC/ESI-Qq-TOF-MS/MS and library search
HPLC-ESI-Qq-TOF
• A library of dietary phenolic compounds was successfully created using data acquired with a micrOTOF-Q mass spectro-meter.
• The proposed LC-Auto/MS/MS method is fast, sensitive, enabling the characterization of even low-abundance compounds, and provides mass accuracy and true isotopic pattern for both precursor and product ions.
• Positive identifications of dietary phenolics were obtained in a variety of samples analyzed under the same conditions from those of the reference spectra.
Phenolic compounds have great importance in the nutritional, organoleptic and commercial properties of plant-derived food and beverages. Furthermore, their consumption has been associated with positive health benefits such as antioxidant, antiviral, antiallergenic, cardioprotective, and anticarcinogenic effects. The health effects of polyphenols depend on the amount consumed and their bioavailability. Detailed knowledge of the phenolic composition of food and beverages will contribute to a better understanding of their influence on biological properties. A method for the automated and robust identification of phenolic compounds in foods based on accurate mass and library search has been developed.
María Gómez-Romero1, Birgit Schneider2,, Antonio Segura-Carretero1, Alberto Fernández-Gutiérrez1, Gabriela Zurek2, Carsten Baessmann21Department of Analytical Chemistry, Faculty of Sciences, University of Granada, Granada, Spain2Bruker Daltonik GmbH, Bremen, Germany
•acquisition in automated MS/MS mode (1 precursor ion)• scan range 50-800 m/z• mass scale calibrations with sodium formate acetate clusters
Library generation. MS and MS/MS mass spectra of 39 phenolic compounds (all commercial standards) were added to the spectral library using DataAnalysisTM
and LibraryEditorTM software (Bruker Daltonik, GmbH). Chemical information was collected from company and public websites.
Sample preparation. Juices: filtered and diluted (1:1). Phenolic concentrates (obtained at low temperature from vegetables rich in phenolic compounds): mixed with methanol for 30 min, filtered and diluted.
We have built up a library of common phenolic compounds using retention time (RT), accurate MS and MS/MS spectra. Conditions were optimized to detect phenolic acids, flavonoids and their glycosides. During spectra review for library generation, special emphasis was on the mass accuracy and isotopic pattern reflected in the sigma-value for both MS and MS/MS spectra. As proof of principle, a blueberry sample was spiked with a mixture of nine phenolic compounds.
Subsequently, fruit juice and phenolic concentrate real samples were analyzed. Library search was performed against the created ESI-Qq-TOF and an existing ion trap library, containing more than 1500 spectra. The automated library search method comprises recalibration, peak detection and settings for the library search itself. The method combines the search of RT, MS and MS/MS spectra improving identification of compounds considerably while reducing ambiguities and false positive hits.The base peak chromatogram (BPC) of spiked blueberry juice as well as the extracted ion chromatograms (EIC) of the identified compounds are shown in figure 1.1. A representative compound and its search results are shown in figure 1.2. The identification of the compounds can be confirmed using their mass accuracy and isotopic pattern information for formula generation. The final list of identified compounds in the juice is shown in figure 1.3. The quality of the identification is given by an effective purity score reflecting a combined score of MS and MS/MS data as well as acquisition parameters (max. value = 1000). Peaks #9 myricetin and #11 quercetin were identified by RT and MS only due to acquisition in AutoMSn with one precursor ion. Even low-abundance compounds can be identified using the library search approach as it is demonstrated in figure 2.Table 1 shows the summary of all samples analyzed in ESI negative mode. Compounds were identified by the library search with good purity scores as well as by formula generation.
Results
-MS
-MS/MS
2
3
1
BPC (black) and EIC (color) of identified compounds. Blue numbers: spiked compounds.
Identification of peak #7: m-coumaric acid. Search results (matched library spectra) and formula generation.
3
1
2
Library search results: final list of identified compounds.Blue boxes: spiked compounds.
Fig. 1 Blueberry juice spiked with chrysin, apigenin, luteolin, catechin, genistein, isosakuranetin, m, o and p-coumaric acids analyzed in negative polarity ESI
MethodsChromatographic separation. HPLC Agilent 1100 series (Santa Clara, CA) with Phenomenex Synergi Fusion-RP 100A, 50x2 mm (2.5 µm) column & precolumn.• flow rate 0.5 mL/min (use of splitter)• injection volume 5 µL• column temperature 35 ºC• solvent A: water; solvent B: acetonitrile both with 0.1% (v/v) formic acid • gradient: 1-100% B, 0-9 min; 100-1% B 9-9.5 min
MS conditions. MS-instrument: micrOTOF-QTM ESI-Qq-TOF mass spectrometer (Bruker Daltonik GmbH, Bremen, Germany). • positive and negative ion polarity ESI• nebulizer 2 bar; dry gas flow 7 L/min; dry gas heater 200ºC• spectra rate 1Hz
ESI-Qq-TOF mass-spectral library
Analysis of food samples
Analysis of phenolic compounds
Adding spectra to library
Sample preparation
Identification of phenolics by library
search approach
Establishment of LC- and Auto-MS/MS-conditions
MS MS/MS
RT
Review of spectra
1
Fig. 2 Cowberry juice (C). BPC (black), EIC (red) and identification result of a low-abundance compound
Identified by library search:1: MS 2: MS/MS
Table. 1 Identified compounds in the analyzed samples. Juices:apple (A), red grape (B), cowberry (C), blueberry (D). Concentrates:green tea (E), grape (F), propolis (G)
Library development and identification workflow
The author MGR gratefully acknowledges Grant AP-2005-2985 from Ministerio de Educación y Ciencia, Bruker Daltonik GmbH and Verbionat SA.
For research use only. Not for use in diagnostic procedures.
-24-
Metabolomics Data evaluation, Identification of unknowns
Metabolomics Posterbook 2011
Conclusions
Bruker Daltonics
Structure ElucidationUHR-TOF
Deconvolution and Automatic Formula Assignment of Alternating MS and Broad-Band CID Analyses
Peter Sander, Ilmari Krebs, Sebastian Goetz, and Birgit Schneider, Aiko BarschBruker Daltonik GmbH, Fahrenheitstraße 4, 28359 Bremen, Germany
IntroductionTypical metabolite profiling applications are faced with complex samples – partly overlapping chromatographic peaks of co-eluting compounds resulting in many potentially relevant precursors to be isolated and fragmented to create a product ion spectrum. Alternating full-scan-MS and broad-band-CID (bbCID) is an alternative acquisition mode for generating fragment data of complex samples. A mathematical algorithm unequivocally assigns the observed product ion peaks to its precursor ion peaks. Beside of accurate mass measurements and correct isotopic pattern this fragmentation information is key for assigning an elemental composition to fragment and precursor ions enabling the correct identification of potential impurities or metabolites.The automated workflow of formulae assignment is exemplified on a mixture of 6 antibiotics.
Methods
• UHPLC system coupled to UHR-Q-TOF (maXisTM)
• Data acquisition: Alternating MS and broad-band CID (bbCID) spectra; Broad-band fragmentation in the collision cell is achieved by setting a higher collision energy (25 eV) without precursor isolation, whereas the precursor spectrum is taken at a low collision energy (8 eV) with basically no fragmentation (Fig. 1).
• Deconvolution of complex mixtures
• “Clean” compound spectra of MS and bbCID reconstructed
• Clear assignment of fragments to a precursor
• Automatic formula assignment based on precursor and all fragment signals
• Confidence of formula assignment raised by Scoring of precursor incl. fragment signals
Results• A mixture of 6 antibiotics has been analyzed in alternating MS and bbCID mode. Precursor and corresponding fragment ions were deconvoluted by the Dissect algorithm. All antibiotics have been assigned reliably by a good combined score Fig.2. The details of the formula assignment and scoring is demonstrated on Ampicillin.
• The list of possible sum formulae for the precursor is greatly reduced by combining it with the list of possible sum formulae for the fragments and by requiring the most intense fragments to be explained.
• The combined score finally ranks the correct formula topmost.
• An extracted human blood serum spiked with pharmaceuticals has been analyzed and evaluated with the same method (Fig.3).The complexity of such a sample is shown on the retention time vs. m/z plot.
• The details and performance of the Dissectdeconvolution algorithm is demonstrated on the Rifamycine signal. Although overlapped by another more abundant pharmaceutical and other background and matrix signals, the Rifamycine-related precursor and fragment signals have been retained cleanly.
Fig. 3: Deconvolution of a LC MS/bbCID analysis of a complex mixture (extracted human blood serum spiked with pharmaceuticals) with Dissect. Overlapping compounds are cleanly deconvoluted (compared to averaged spectra).
Fig.2: Automatic formulae assignment for MS precursor and bbCID fragment spectrum of Ampicillin with SmartFormula3D.
Manually Averaged Spectra
MS
bbCID
MS
bbCID
TIC
“clean” Rifamycine spectraoverlapping signals of Terfenadine and Rifamycine
Dissect Compound Chromatograms
m/z 823.4: Rifamycine
RT vs. m/z RT vs. I
Dissect Deconvoluted Spectra
calculatecombined scores
MS
bbCID
Ampicillin
Name Formula M [M+H]+ Error [mDa] Combined Score
Apramycine C21H41N5O11 540.2881 0.4 95
Ampicillin C16H19N3O4S 350.1175 -0.5 97
Vancomycine C66H75Cl2N9O24 1448.4380 0.6 82
Clarithromycine C38H69NO13 748.4847 -1.1 73
Erythromycine C37H67NO13 734.4691 -1.2 74
Rifamycine C43H58N4O12 823.4129 -1.3 52
matching precursorby accurate mass
and isotope pattern
matching fragmentsby accurate mass
and isotope pattern
Filtering:combining all matches
and explainmost intense fragments
UHR-TOF LC-MS/bbCID is a viable tool for structure elucidation.
For research use only. Not for use in diagnostic procedures.
• Deconvolute LC-MS/bbCID analysis with Dissectalgorithm producing MS and bbCID compound spectra of accurately co-eluting signals which are fully background subtracted and deconvoluted from overlapping chromatographic peaks or matrix signals. Also, the isotope pattern of each ion species is fully preserved.SmartFormula3DAssigns a chemical formula to the most intense peaks of the precursor mass spectrum in combination with the related product ion spectrum by matching the exact mass and isotope pattern of a potential precursor ion with potential fragment ions.
• The final combined score is calculated as average score for the precursor ion and the scores for the fragment ions.Each score is calculated by matching the exact mass and isotope pattern and normalized to 100 for the best matching formula.
Source
Dual Ion Funnel
Quadrupole Ion Cooler
Collision cellbbCID
maXis Schematic
Fig.1: Schematic of maXis UHR-Q-TOF
Conclusions
Bruker Daltonics
MALDI ImagingFT-ICR MS
Small Molecule Imaging Mass Spectrometry using the solariX-MIJens Fuchser and Michael Becker
Bruker Daltonik GmbH, Bremen, Germany
IntroductionImage analysis of small molecules in tissue with MALDI TOF systems typically requires the acquisition of a separate MS-MS spectrum for each target ion at each pixel location. Such procedures are necessary to confirm target ions among a dense chemical background. The high mass-resolution and measurement accuracy of FTICR MS offers an alternate single-scan strategy for imaging small molecules in less time. Moreover, the MS-mode acquisition can yield images for thousands of ions, providing greater depth of information and the ability to monitor many more than a few selected compounds.
Methods
Tissue sections were frozen on a cryostat steel plate using as little OCT (Optimal Cutting Temperature) polymer as possible to avoid contamination of the section with OCT. Sections (10 μm thick) were cut with microtome (Leica Microsystems CM1900 UV, Wetzlar, Germany) and mounted onto conductive glass slides (Bruker Daltonics). Prior to matrix application, slides were marked with dots of liquid white-out, and native sections were scanned using a Coolscan V ED (Nikon GmbH, Düsseldorf, Germany).Subsequently HCCA matrix was applied using the ImagePrep (a piezo-driven aerosol generator, Bruker Daltonics)[1] matrix solution consisted of 7g/kg HCCA matrix in water/acetonitrile/TFA (tri-fluoro acetic acid) = 49.9/49.9/0.2 (all high pure reagents from Sigma).
MALDI images have been acquired with a solariX 12 T FT-ICR mass spectrometer (Bruker) equipped with a Dual ionization source with Smartbeam-II laserTM
technology[2] in positive ion mode (Figure 1).
External calibration provided a measurement accuracy of <1ppm over the m/z range of interest, 140 - 1000 Da. Potential ion compositions were computed using SmartFormulaTM which uses the measurement accuracy as well as the isotopic pattern as tolerance, typically resulting in one composition only.
High resolution and mass accuracycombined with superior SmartBeam-IItechnology render FTICR MS as the idealsolution for small molecule MALDIimaging.
Summary- High mass accuracy and resolution are a pre-requisite for successful small molecule imaging.- Direct identification of compounds in MS mode is possible with FTICR MALDI Imaging- Detection of physiological amounts of neurotrans-mitters is achieved due to superior SmartBeam-II technology - Tissue can be classified by statistical methods
Literature[1] Bruker Daltonics technical note 18 “ImagePrep“ .[2] Holle, A., Haase, A., Kayser, M., Höhndorf, J. “Optimizing UV laser focus profiles for improved MALDI performance”. Mass Spectrom. 2006 Jun;41(6): 705-16.
AcknowledgementThe whole body rat was kindly provided by Jonathan Stauber, ImaBiotech, Villeneuve d’Ascq, France
ResultsMALDI Imaging experiments are providing huge numbers of signals due to the complexity of tissue in general. Thousands of peaks can be resolved by FTICR MS in a MALDI spectrum, giving rise to unrivalled amount of data in a single experiment.Nevertheless, this intense chemical background can mask compounds of interest. Displaying different mass traces throughout the tissue shows the spatial distribution of masses (Figure 1, whole body mouse). The overlay of the mass traces with the optical picture of the tissue reveals molecular features of certain tissue areas.
The detection of neurotransmitters in a sagittally cut rat brain is shown in figure 2a, figure 2b represents a single mass spectrum (30µm Laser spot size). One can see the dynamic range needed for small molecule tissue imaging. Figure 2c shows a part of the rat brain (cerebellum) and the identification of a compound using SmartFormulaTM. Here dipalmitoylglycerol is readily identified with a mass error of 0.5 ppm.
Figure 2 a) Display of different masses showing the detection of neurotransmitters at physiological levels; b) Single spectrum with higher spatial resolution demonstrating the dynamic range within a spot of 30 µm; c) Example of identification of a compound using SmartFormulaTM
Figure 1. Fading between the optical and the mass spectrometric image of a whole body mouse, spatial resolution 400 µm
a)
b)
c)
Acetylcholine (blue)m/z =146.11757, 0.1ppm
C7H15NO4Na (green)m/z = 200.08926, 0.3 ppm
184.07334
228.00573
265.96161
379.09247
455.00421
534.29564
713.45224
772.52525
961.56725
200 400 600 800 1000 1200 m/z0
1
2
3
4
7x10Intens.
1: +MS
146.10886146.11282
146.11756
146.13225146.14290
146.11 146.12 146.13 146.14 146.15 146.16 146.17 146.18 146.19 m/z
2
4
6
5x10Intens.
1: +MS
Single spectrum of 30 µm diameter
146.11758
518.2636
551.5031 (Dipalmitoylglycerol)
Optical image is removed to give a sharper image (see Figure 3c).
C6H17NO4P (red), m/z = 198.0890, 0.4 ppm
ResultsClassification of tissue
Using statistical methods tissue can be analysed on grounds of the presence and intensity of different mass signals. The resulting statistical model can be used to classify tissue. Figure 3a shows the selection of regions for building the statistical model. Figure 3b is a screenshot of ClinProToolsTM, Bruker Daltonics` solution for statistical analysis of MALDI Imaging data.Model generation was carried out using a genetic algorithm. Figure 3c shows the classification of the mass spectra according to the two classes being generated by the model for the entire data set.
b)
Rat brain, HCCA, pos, 30 µm spatial resolutionPart of cerebellum, overlay of classification and optical image
Based on the generatedModel the entire measured area can be classified
a)Two regions are pickedto build a model for classification
c)
Figure 3a) Regions selected for model generation; b) Screenshot of ClinProToolsTM showing the model generation; c) Applying the model to the entire tissue results in an accurate representation of the classified feature.
Rat Brain, sagittally cut
Whole body mouse
For research use only. Not for use in diagnostic procedures.
-26- -27-
Metabolomics Data evaluation, Identification of unknowns Small Molecule Imaging