Continuous Water Infusion Enhances Atmospheric Pressure Chemical

Aug 23, 2014 - ... University Hospital Regensburg, Franz-Josef-Strauss-Allee 11, .... and isooctane from BDH Prolabo (VWR International, Vienna, Austr...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Continuous Water Infusion Enhances Atmospheric Pressure Chemical Ionization of Methyl Chloroformate Derivatives in Gas Chromatography Coupled to Time-of-Flight Mass SpectrometryBased Metabolomics Christian J. Wachsmuth,*,† Katja Dettmer,† Sven A. Lang,‡ Maria E. Mycielska,‡ and Peter J. Oefner† †

Institute of Functional Genomics, University of Regensburg, Josef-Engert-Strasse 9, 93053 Regensburg, Germany Department of Surgery, University Hospital Regensburg, Franz-Josef-Strauss-Allee 11, 93053 Regensburg, Germany

‡

S Supporting Information *

ABSTRACT: The effects of continuous water infusion on efficiency and repeatability of atmospheric pressure chemical ionization of both methyl chloroformate (MCF) and methoxime-trimethylsilyl (MO-TMS) derivatives of metabolites were evaluated using gas chromatography−time-of-flight mass spectrometry. Water infusion at a flow-rate of 0.4 mL/h yielded not only an average 16.6-fold increase in intensity of the quasimolecular ion for 20 MCF-derivatized metabolite standards through suppression of in-source fragmentation but also the most repeatable peak area integrals. The impact of water infusion was the greatest for dicarboxylic acids and the least for (hetero-) aromatic compounds. Water infusion also improved the ability to detect reliably fold changes as small as 1.33-fold for the same 20 MCF-derivatized metabolite standards spiked into a human serum extract. On the other hand, MOTMS derivatives were not significantly affected by water infusion, neither in their fragmentation patterns nor with regard to the detection of differentially regulated compounds. As a proof of principle, we applied MCF derivatization and GC-APCI-TOFMS to the detection of changes in abundance of metabolites in pancreatic cancer cells upon treatment with 17-DMAG. Water infusion increased not only the number of metabolites identified via their quasimolecular ion but also the reproducibility of peak areas, thereby almost doubling the number of significantly regulated metabolites (false discovery rate < 0.05) to a total of 23. etabolic fingerprinting aims at the measurement of the largest possible number of metabolites in a single analysis with the objectives of distinguishing sample classes and identifying discriminating metabolites to unravel the molecular effects of both genetic and nongenetic factors on biological systems. Gas chromatography−electron ionization−mass spectrometry (GC-EI-MS) has proven highly valuable in metabolite fingerprinting in large part due to high analytical reproducibility, low detection limits, and the ready availability of large mass spectral libraries for trimethylsilyl (TMS) derivatives. For the identification of unknown metabolites, soft ionization techniques, such as chemical ionization1,2 (CI) and atmospheric pressure chemical ionization3−5 (APCI) have gained in importance, because they allow in combination with highresolution mass spectrometry the calculation of elemental formulas from the accurate masses of quasimolecular ([M + H]+) ions. The analytical potential of GC-APCI-TOFMS, originally introduced in the 1970s,6,7 has been increasingly utilized in recent years with the availability of commercial APCI sources.8,9 Successful applications of GC-APCI-(Q)TOFMS range from metabolomics3−5,10 to the analysis of foodstuffs,11 pesticides,12

M

© 2014 American Chemical Society

high-molecular-weight phthalates,13 impurities in pharmaceutical process development,14 and sterols.15 Technical advances include source miniaturization16 and the introduction of a webbased GC-APCI-QqTOF spectral library.17 In a recent study,4 we could demonstrate the complementarity of GC-APCITOFMS and comprehensive two-dimensional gas chromatography (GC × GC)-EI-TOFMS for metabolic fingerprinting of Escherichia coli strains including the successful identification of discriminating metabolites missed by the latter technique.18 It has been recognized as early as 1976 that water promotes the formation of protonated molecular ions in APCI.19 Notable improvements in sensitivity and selectivity achieved by the infusion of water into the APCI source have been demonstrated recently for gas chromatography−tandem mass spectrometry of underivatized pyrethroid insecticides12 and GC-APCI-TOFMS analysis of dissolved standards of methyl stearate, benzophenone, methyl palmitate, and cocaine.15 Whether such improveReceived: June 10, 2014 Accepted: August 23, 2014 Published: August 23, 2014 9186

dx.doi.org/10.1021/ac502133r | Anal. Chem. 2014, 86, 9186−9195

Analytical Chemistry

Article

ments also hold true for derivatized compounds was the objective of the present study. Many metabolites contain polar functional groups that require derivatization prior to GC-MS analysis. Typically, the derivatization method of choice for metabolic fingerprinting is methoximation followed by silylation (MO-TMS).20 More recently, alkylation/esterification with alkyl chloroformates (CFs) has emerged as a simpler and faster alternative for a steadily increasing range of compounds.21−25 In 2011, VillasBôas et al.26 compared MO-TMS with methyl CF (MCF) derivatization and found the latter to form more consistently stable derivatives that proved beneficial in the discrimination of culture media samples from different bacterial strains. The primary aim of this work was to study the effects of continuous water infusion on APCI-TOF mass spectra of MCF and MO-TMS derivatives of 20 standard compounds and the ability of GC-APCI-TOFMS to detect differences in metabolite concentration levels in a human serum extract. Subsequently, MCF derivatization and GC-APCI-TOFMS were applied to comparative metabolic fingerprinting of extracts of pancreatic cancer cells that had been cultivated in the absence and presence of the heat-shock protein 90 (Hsp90) inhibitor 17dimethylaminoethylamino-17-demethoxygeldanamycin (17DMAG).

■

Information) contained 20 compounds at a concentration of 1 mM each in MeOH. Cell Culture. MiaPaCa-2 pancreatic cancer cells were grown in 75 cm2 cell culture flasks (PAA Laboratories GmbH) in 25 mL of DMEM medium supplemented with 15% (v/v) fetal bovine serum and enriched with 2 mM glutamine, 25 mM glucose, and standard penicillin/streptomycin solution. Cells were cultured in a humidified incubator at 37 °C with 5% CO2. For assessment of the effect of 17-DMAG on cell metabolism, 1 × 106 cells were plated into 10 cm Petri dishes and treated with either medium containing 100 nM 17-DMAG or control medium only for 72 h (N = 5 biological replicates). Spike-in Experiment. Ten microliter aliquots of human serum were extracted with 50 μL of MeOH as previously described,27 without adding extraction standards. Extracts were pooled and then split into aliquots for the preparation of spikein samples. Amounts of 960 and 160 μL each were taken from the pool for MCF and MO-TMS derivatization, respectively. Six different spike-in levels were prepared (N = 5 derivatization replicates), ranging in final concentrations of the spike-in compounds from 133 μM to 200 μM, 267 μM, 333 μM, 400 μM, and 467 μM, respectively, for MCF derivatization, and 67 μM to 100 μM, 133 μM, 167 μM, 200 μM, and 233 μM, respectively, for MO-TMS derivatization. Extraction of Metabolites. Cells were washed thrice with cold phosphate-buffered saline (PBS). For quenching of cell metabolism as well as combined cell harvesting and extraction of metabolites, direct solvent scraping in MeOH/H2O (80:20, v/v)28 was performed on ice and with cold solutions. Dried extracts were redissolved in 200 μL of MeOH/H2O (80:20, v/ v), and 60 μL was taken for MCF derivatization. Determination of Protein Content. For the determination of cellular protein content according to the instructions provided with the FluoroProfile kit from Sigma-Aldrich, protein pellets were dissolved in 1 mL of a 20 mM sodium dihydrogen phosphate buffer supplemented with 1% (v/v) sodium dodecyl sulfate and diluted 1:50 (v/v) with water. Derivatization. On the basis of the desired final standard concentration, the respective volume of standard stock solution, spike-in mixture, or an appropriate dilution was transferred to a 2 mL glass vial together with 20 μL of fatty acid mixture (MCF derivatization), respectively, to a 2 mL glass vial with a 400-μL glass insert (MO-TMS derivatization). For example, 120 μL and 20 μL, respectively, of a 0.5 mM standard stock solution were taken for preparation of 200 μM MCF- and 100 μM MOTMS-derivatized standard samples in initial experiments and for optimizing the water infusion rate. Depending on the sample set, cell extract or an aliquot of the serum extract pool was added as well, and the entire sample was then dried using a vacuum evaporator (CombiDancer, Hettich AG, Bäch, Switzerland). MCF derivatization was performed based on a modified MeOH/MCF protocol published previously.23 Briefly, 275 μL of H2O, 167 μL of MeOH, and 34 μL of pyridine were added to the dried samples. Then 20 μL of MCF was added twice, followed by vortexing for 10 s after each addition. After 30 s, MCF derivatives were extracted with 300 μL of chloroform and analyzed. In the case of cell extracts, 250 μL of organic phase was carefully dried and reconstituted in 50 μL of chloroform. Starting from the dried samples, MO-TMS derivatization was performed as described by Wachsmuth et al.,4 which included the addition of 10 μL of fatty acid mixture prior to silylation. Instrumentation. A model 450-GC (Bruker Daltonics GmbH, Bremen, Germany) with an autosampler (model PAL

EXPERIMENTAL SECTION

Materials. Compounds included in the standard mix, as well as citric acid, aminoadipic acid, cis-aconitic acid, malic acid, lactic acid, tryptophan, lysine, threonine, cysteine, histidine, tyrosine, arginine, asparagine, lauric acid, myristic acid, stearic acid, serine, ornithine, glycine, palmitic acid, N-acetyl-L-aspartic acid, odd-numbered, saturated straight-chain fatty acids (C9− C19), pyridine, methyl chloroformate, methoxylamine hydrochloride, 3-picoline, propyl chloroformate, fetal bovine serum, L-glutamine, sodium dodecyl sulfate, and the FluoroProfile kit for determination of protein content in cell pellets, were from Sigma-Aldrich/Fluka (Taufkirchen, Germany). N-Methyl-Ntrimethylsilyl trifluoroacetamide (MSTFA) was purchased from Macherey-Nagel (Dueren, Germany), sodium dihydrogen phosphate monohydrate and chloroform (analytical grade) from Merck KGaA (Darmstadt, Germany), and [U-13C, U‑15N] cell-free amino acid mix from (Euriso-top) (Saint-Aubin Cedex, France). Methanol (MeOH; LC-MS grade), ethyl acetate (analytical grade), and propanol (LC-MS grade) were from Fisher Scientific GmbH (Ulm, Germany), and isooctane from BDH Prolabo (VWR International, Vienna, Austria). Purified water from a PURELAB Plus system (ELGA LabWater, Celle, Germany) was used in all experiments. MiaPaCa-2 pancreatic cancer cells were obtained from ATCC (Manassas, VA). DMEM medium, glucose, and phosphate-buffered saline (PBS) were purchased from PAA Laboratories GmbH (Coelbe, Germany), and penicillin/streptomycin from Invitrogen (Karlsruhe, Germany). The Hsp90 inhibitor 17-DMAG was obtained from Invivogen (Toulouse, France). For each standard compound, an individual stock solution was prepared in either methanol, water, or a mixture thereof, except for the even- (C12−C18) and odd-numbered (C9− C19), saturated straight-chain fatty acids, which were dissolved in propanol and isooctane, respectively. The latter, hereinafter referred to as fatty acid mixture (1 mM of each compound), were used as retention index markers and for recalibration of the mass scale. The spike-in mixture (Table S3, Supporting 9187

dx.doi.org/10.1021/ac502133r | Anal. Chem. 2014, 86, 9186−9195

Analytical Chemistry

Article

in-house Visual Basic script loaded into DataAnalysis.4 Retention times and m/z values of the [M + H]+ ions for the fatty acid derivatives are listed in Table S1, Supporting Information. Calculation of Elemental Formulas and Compound Identification. The recalibrated masses of [M + H]+ ions were subjected to the SmartFormula Manually routine included in DataAnalysis accepting a maximum deviation of 5 mDa for the calculated formulas. Further calculation parameters are given with Table S6 in Supporting Information. For revealing the identity of unknown features, calculated elemental formulas with a corresponding mSigma value below 50 were further considered. For each formula, groups introduced by derivatization and the proton had to be eliminated for obtaining the sum formula of the native compound, which was then searched in the Human Metabolome Database.33 For tentatively identified metabolites, standards were analyzed and mass spectra and retention indices based on odd-numbered fatty acids were compared for final annotation. For details see Wachsmuth et al.4 Feature Extraction and Alignment. For the analysis of data from the spike-in experiment, optimized Find Molecular Features (FMF) algorithm and bucketing parameters for feature extraction and alignment within ProfileAnalysis were as follows for MCF/MO-TMS derivatives, with deviating parameters for MO-TMS derivatives in brackets: S/N threshold, 2 (5); correlation coefficient threshold, 0.6 (0.7); minimum compound length, 8 (10); smoothing width, 2 (3); additional smoothing, enabled; proteomics, enabled; bucketing basis, M + H. Retention time range, 8−25 min (8−17 min); mass range, m/z 100−250 (m/z 100−500); advanced bucketing tolerance parameters, 0.1 min (0.4 min) and 10 mDa for retention time and mass, respectively; split buckets with multiple compounds, disabled; value count of group attribute (spike level) within bucket as bucket filter, enabled at least three; other parameters, always none or disabled. Each of the four different bucket tables (0.0 mL/h H2O/ MCF, 0.4 mL/h H2O/MCF, 0.0 mL/h H2O/MO-TMS, and 0.4 mL/h H2O/MO-TMS) listed features in rows with accompanying information on retention time and m/z value as well as the feature intensities over all spike-in samples from entire spike-in levels. To reduce the number of zero values within the bucket table, bucket filtering was applied: at least three out of five area integrals had to be available for each spikein level; otherwise the respective feature was excluded. Further data analysis proceeded in Excel as described in Results and Discussion. For metabolite fingerprinting in cell extracts, individual data matrices were generated from GC-APCI/+H2O-TOFMS and GC-APCI/−H2O-TOFMS measurements of MCF-derivatized cell extract samples of control and 17-DMAG-treated groups, applying with a few exceptions the same FMF and bucketing parameters as for the MCF spike-in experiment. Exceptions were as follows: the mass range for bucketing was extended to m/z 100−300, advanced bucketing tolerance parameter for retention time was increased to 0.15 min, and bucket filtering criteria allowed only the inclusion of features into the bucket table, for which at least 3/5 area integrals were available in at least one group. Further data analysis proceeded in Excel as described in Results and Discussion.

COMBI-xt from CTC Analytics, Zwingen, Switzerland) for sample injection with a 10-μL Hamilton syringe was coupled to a microTOF orthogonal acceleration TOF mass spectrometer (Bruker Daltonics) via an APCI source. For separation of analytes, a Phenomenex ZB-AAA column (15 m × 0.25 mm i.d. × 0.1 μm film thickness, Torrence, CA) was used. The oven temperature program started at 50 °C and held for 1 min, and temperature was ramped at 8 K/min to 300 °C and held for 15 min. A sample volume of 1 μL was introduced by means of an 1177 split/splitless injector at 280 °C employing splitless mode with a splitless time of 1 min. Helium served as the carrier gas at a constant flow rate of 1 mL/min. The transfer line to the MS instrument was maintained at 300 °C, and the protrusion of the GC capillary from the exit of the transfer line was approximately 1 mm. The APCI source was operated as follows: ionization mode, positive; drying gas (nitrogen) temperature, 160 °C; drying gas flow rate, 2.0 L/min; APCI vaporizer temperature, 300 °C; nebulizer gas (nitrogen) pressure, 4.0 bar; current of the corona discharge needle, +3000 nA; capillary voltage, −2900 V; end-plate offset, 0 V. Water was introduced through the top ESI inlet into the APCI source by means of a syringe pump (KD Scientific Inc., Holliston, MA). Mass range for acquisition of spectra with a rate of 3 spectra/s was from 50 to 1000 m/z. Initial external mass calibration was performed with an electrospray ionization tuning mix (Agilent) and an ESI source, with an additional postacquisition processing routine being performed for internal recalibration of each mass spectrum (see below). Spike-in samples along with blanks and cell extracts were measured in random order to avoid systematic error. A blank sample was always run between changes in water flow rates for equilibration of APCI source conditions. Cross-Validation of Amino Acids. See Supporting Information. Data Analysis. Bruker DataAnalysis V4.1 (Bruker Daltonics, Bremen, Germany) was employed for manual inspection of chromatograms and mass spectra, internal recalibration of mass spectra and calculation of elemental formulas. For peak integration, data files were loaded into MassLynx V4.1 from Waters Inc. (Waters, Milford, MA), with m/z of either [M + H]+ for standard compounds (see Table S1 in Supporting Information) or m/z of a particular feature used as quantifier mass. Generation of bucket tables, feature extraction, and alignment were performed in ProfileAnalysis V2.1 (Bruker Daltonics). Bucket tables were then exported into Excel (Microsoft Corporation, Redmond, WA). The statistical computing package R was used to perform principal component analysis (PCA), the Kolmogorov−Smirnov test, the Wilcoxon signed-rank test, ANOVA followed by post hoc tests using LIMMA,29 to impute missing values by MISSMDA,30 and to calculate false discovery rates (FDRs) according to Benjamini and Hochberg31 using MULTTEST.32 The partition coefficients of the MCF and MO-TMS derivatives of the investigated standard compounds and fatty acids were estimated using ACD/Laboratories V12.01 (Advanced Chemistry Development Inc., Toronto, Canada) and expressed as the log Poctanol/water, i.e., the logarithm of the ratio of the concentrations of the un-ionized solute in octanol and water, respectively. Processing of APCI Mass Spectra. Prior to calculation of elemental formulas, mass spectra were recalibrated using the m/z values of the [M + H]+ ions of fatty acid derivatives and an 9188

dx.doi.org/10.1021/ac502133r | Anal. Chem. 2014, 86, 9186−9195

Analytical Chemistry

Article

Figure 1. GC-APCI-TOFMS mass spectra of the MCF derivative of suberic acid (concn = 200 μM) acquired without (A) and with (B) infusion of water at 0.4 mL/h. Neutral losses yield a series of fragments in the case of APCI/−H2O as annotated in the left spectrum.

■

RESULTS AND DISCUSSION Effects of Continuous Water Infusion on APCI Mass Spectra of MCF and MO-TMS Derivatives. Initial experiments on the effect of continuous water infusion on APCI mass spectra of MCF and MO-TMS derivatives were performed at an infusion rate of 0.4 mL/h. Table S1 lists the most prominent ions observed in APCI/+H2O mass spectra of MCF and MOTMS derivatives of 20 standard compounds as compared to spectra acquired without water infusion (APCI/−H2O). Compounds included in the standard mixture were selected as to cover a variety of chemical classes, including organic acids, amino acids, and analytes containing amide and thiol functional groups, according to the three performance classes introduced by Koek et al.20 for silylated derivatives. APCI/−H2O mass spectra of MCF derivatives were typically dominated by fragment ions. The [M + H]+ ion constituted only a minor component with a relative intensity of 20 for corresponding peaks across all samples of at least one group. For visualization, principal component analysis was performed based on the final bucket tables. While a group separation between control and 17-DMAG-treated cell extracts along PC1 was obtained based on both APCI/−H2O- and APCI/+H2OTOFMS data, variability within the two biological groups was distinctly reduced in the latter case, as presented in Figure 5B,C. Among the top loadings of PC1, features corresponding to cysteine, threonine, alanine, glycine, and proline were always found, irrespective of water infusion. The identities of discriminating features were determined whenever possible. One has to keep in mind that the number of detected features does not necessarily equal that of metabolites, as features corresponding to isotopes, adduct, and fragment ions might have been included as well. Eventually, 27 out of 35 (APCI/−H2O) and 41 out of 64 (APCI/+H2O) significant features could be assigned to 13 and 23 metabolites, respectively. Table 1 lists all discriminating metabolites along with figures of merit from the identification procedure. Because, according to Sumner et al.,37 at least two independent and orthogonal data derived from a standard are required for the definite (level 1) identification of an unknown, we compared van den Dool retention indices (RIs), in addition to the mass spectra. Differences in RIs (ΔRI) were between 0 and 6 index

malate, which eluted at 14.9 min in the chromatogram depicted in Figure 4B (peak no. 9). Twenty of the metabolites identified in extracts of untreated pancreatic cancer cells are represented in Figure 5A, with mean [M + H]+ areas ± SD given in the graph. With the exception of the fatty acid derivatives, which formed water adducts at the expense of [M + H]+ ions under water infusion, all metabolites showed an increase in mean [M + H]+ area. Furthermore, standard deviations were lower for most metabolites. Importantly, of the four metabolites that could solely be identified by water infusion (Figure 5A), malate, serine, and aminoadipate were found to show significant discrimination between control and 17-DMAG-treated cell extracts. Comparative Analysis of Control and 17-DMAGTreated Sample Groups. Following extraction and alignment of features for both control and 17-DMAG-treated cells that had been acquired by APCI/−H2O and APCI/+H2O, respectively, peak area integrals were normalized against protein content followed by log 2 transformation. Further, FDRs according to Benjamini and Hochberg31 assuming equal variances in control and treated groups were calculated. A total of 18 and 77 significant features (FDR < 0.05) that distinguished control from 17-DMAG-treated cells were obtained by APCI/−H2O and APCI/+H2O, respectively. To minimize false positive features, areas of significant features were reintegrated manually using MassLynx to exclude incorrect feature integration by the automated data analysis routine. Furthermore, GC-MS [M + H]+ features for amino acids that were also independently quantified by HPLC-MS/ MS (compare Table S7 in the Supporting Information) as well as [M + H]+ features of metabolites displayed in Figure 5A 9193

dx.doi.org/10.1021/ac502133r | Anal. Chem. 2014, 86, 9186−9195

Analytical Chemistry

Article

units corresponding to relative ΔRIs of 0−0.5%. This was below the established threshold of 1.0% for transferring RIs between different methods by Strehmel et al.,38 who had found ΔRI to increase with retention time and, therefore, had suggested to use relative rather than absolute RIs values. Significant features from APCI/+H2O analysis were either not detectable with APCI/−H2O or not extracted by the FMF routine despite a S/N threshold of 2. Further reasons for the smaller number of significant features from APCI/−H2O analysis included missing values within group attributes of extracted features, which entailed exclusion of the respective features, as well as insufficient reproducibility of peak areas. With a rapid turnover of intracellular metabolites, biological variances are increased, thus putting even higher requirements on analytical precision. Water infusion made it possible to extract features with overall better reproducibility, e.g., RSDs for peak areas of lysine (m/z [M + H]+: 277.1394 Da) in control and treated samples improved from 32% and 53% to 21% and 24%, respectively, thus enabling its identification as a metabolite differing in abundance between untreated and treated cells. The same held true for ornithine and methionine with fold changes < 2, that had been missed by APCI/−H2O. Finally, the identity of four significant metabolites could not have been revealed in the case of APCI/−H2O. Malate, serine, and aminoadipate only formed an [M + H]+ ion under water infusion, whereas for ornithine, mSigma values > 50 obtained from APCI/−H2O mass spectra would have entailed exclusion of the correct elemental formula during the identification procedure. Cross-Validation of Discriminating Amino Acids. To confirm that amino acids identified by metabolite fingerprinting to discriminate between control and 17-DMAG-treated cells reflect true differences in abundance of those amino acids, targeted analysis by means of HPLC-ESI(+)-MS/MS was performed using stable isotope-labeled internal standards. Concentration levels were normalized by the protein content of the respective cell pellet, which was 3.1 ± 0.2 mg and 1.5 ± 0.4 mg for the control and the 17-DMAG group, respectively. As displayed in Table S7, Supporting Information, significant amino acids in metabolic fingerprinting were confirmed in all cases by HPLC-MS/MS, except for valine and cysteine, whose content could not be determined by the targeted approach. In addition to the discriminating amino acids already identified by the fingerprinting approach, HPLC-MS/MS revealed arginine and asparagine as significantly regulated. In the case of arginine this came as no surprise, as it is thermally instable and, therefore, not amenable to GC-MS. Furthermore, the 1.3-fold difference in abundance of asparagine between the two groups detected by the untargeted approach was too small to reach significance due to the lower reproducibility of GC-APCI/ +H2O-TOFMS in the absence of stable isotope-labeled internal standards. The FCs obtained by GC-APCI/+H2O-TOFMS deviated less than 20% from HPLC-MS/MS results for 15 out of 17 amino acids (Table S7, Supporting Information), emphasizing that our GC-MS approach was capable of reproducing fold changes in a quantitative manner.

different behavior of the MCF and MO-TMS derivatives is the more polar nature of the former, which facilitates the formation of analyte−water ion clusters. Overall, the beneficial effect of water infusion was most pronounced for carboxylic acids, less so for amino acids, and the least for (hetero) aromatic compounds. Water infusion was also found to improve the ability of GC-APCI-TOFMS to detect, determine, and identify metabolites, whose abundance changed significantly upon treatment of cancer cells with the Hsp90-inhibitor 17-DMAG. This was due to reduced in-source fragmentation of MCFderivatized metabolites and a concomitant increase in [M + H]+ ions, that improved not only detection sensitivity but also facilitated elucidation of elemental composition of discriminating metabolites, which also benefitted from the improved isotope patterns seen upon water infusion. Finally, the ability of metabolite fingerprinting based on MCF derivatization and GC-APCI/+H2 O-TOFMS to detect true differences in metabolite abundance could be confirmed by targeted HPLCMS/MS analysis employing stable isotope-labeled internal standards. In summary, APCI/+H2O provides ionization softer than that of conventional APCI, with reproducible formation of [M + H]+ being favored over fragment ions for MCF derivatives, especially in the case of organic acids, thus extending the range of compounds amenable to metabolic fingerprinting.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +49-941-943-5006; fax: +49-941-943-5020; e-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Funding by DFG (KFO 262) is gratefully acknowledged. The authors thank Dr. Alexander Riechers for fruitful discussions throughout the study.

■

REFERENCES

(1) Abate, S.; Ahn, Y. G.; Kind, T.; Cataldi, T. R.; Fiehn, O. Rapid Commun. Mass Spectrom. 2010, 24, 1172−1180. (2) Kumari, S.; Stevens, D.; Kind, T.; Denkert, C.; Fiehn, O. Anal. Chem. 2011, 83, 5895−5902. (3) Carrasco-Pancorbo, A.; Nevedomskaya, E.; Arthen-Engeland, T.; Zey, T.; Zurek, G.; Baessmann, C.; Deelder, A. M.; Mayboroda, O. A. Anal. Chem. 2009, 81, 10071−10079. (4) Wachsmuth, C. J.; Almstetter, M. F.; Waldhier, M. C.; Gruber, M. A.; Nürnberger, N.; Oefner, P. J.; Dettmer, K. Anal. Chem. 2011, 83, 7514−7522. (5) Strehmel, N.; Kopka, J.; Scheel, D.; Böttcher, C. Metabolomics 2014, 10, 324−336. (6) Horning, E. C.; Horning, M. G.; Carroll, D. I.; Dzidic, I.; Stillwell, R. N. Anal. Chem. 1973, 45, 936−943. (7) Horning, E. C.; Carroll, D. I.; Dzidic, I.; Haegele, K. D.; Lin, S.; Oertli, C. U.; Stillwell, R. N. Clin Chem. 1977, 23, 13−21. (8) McEwen, C. N.; McKay, R. G. J. Am. Soc. Mass Spectrom. 2005, 16, 1730−1738. (9) Schiewek, R.; Lorenz, M.; Giese, R.; Brockmann, K.; Benter, T.; Gab, S.; Schmitz, O. J. Anal. Bioanal. Chem. 2008, 392, 87−96.

■

CONCLUSIONS The present study showed that the continuous infusion of water enhanced not only the efficiency but also the reproducibility of APCI of MCF-derivatized metabolites, while exerting almost no or at times even a deleterious effect on APCI of MO-TMS derivatives. A likely explanation for the 9194

dx.doi.org/10.1021/ac502133r | Anal. Chem. 2014, 86, 9186−9195

Analytical Chemistry

Article

(10) Pacchiarotta, T.; Nevedomskaya, E.; Carrasco-Pancorbo, A.; Deelder, A. M.; Mayboroda, O. A. J. Biomol. Tech. 2010, 21, 205−213. (11) Garcia-Villalba, R.; Pacchiarotta, T.; Carrasco-Pancorbo, A.; Segura-Carretero, A.; Fernandez-Gutierrez, A.; Deelder, A. M.; Mayboroda, O. A. J. Chromatogr. A 2011, 1218, 959−971. (12) Portoles, T.; Mol, J. G.; Sancho, J. V.; Hernandez, F. Anal. Chem. 2012, 84, 9802−9810. (13) David, F.; Sandra, P.; Hancock, P. LC-GC Eur. 2011, 24, 16−19. (14) Bristow, T.; Harrison, M.; Sims, M. Rapid Commun. Mass Spectrom. 2010, 24, 1673−1681. (15) Matysik, S.; Schmitz, G.; Bauer, S.; Kiermaier, J.; Matysik, F. M. Biochem. Biophys. Res. Commun. 2014, 446, 751−755. (16) Ö stman, P.; Luosujärvi, L.; Haapala, M.; Grigoras, K.; Ketola, R. A.; Kotiaho, T.; Franssila, S.; Kostiainen, R. Anal. Chem. 2006, 78, 3027−3031. (17) Pacchiarotta, T.; Derks, R. J.; Hurtado-Fernandez, E.; van Bezooijen, P.; Henneman, A.; Schiewek, R.; Fernandez-Gutierrez, A.; Carrasco-Pancorbo, A.; Deelder, A. M.; Mayboroda, O. A. Bioanalysis 2013, 5, 1515−1525. (18) Almstetter, M. F.; Appel, I. J.; Gruber, M. A.; Lottaz, C.; Timischl, B.; Spang, R.; Dettmer, K.; Oefner, P. J. Anal. Chem. 2009, 81, 5731−5739. (19) Dzidic, I.; Carroll, D. I.; Stillwell, R. N.; Horning, E. C. Anal. Chem. 1976, 48, 1763−1768. (20) Koek, M. M.; Jellema, R. H.; van der Greef, J.; Tas, A. C.; Hankemeier, T. Metabolomics 2011, 7, 307−328. (21) Kvitvang, H. F.; Andreassen, T.; Adam, T.; Villas-Boas, S. G.; Bruheim, P. Anal. Chem. 2011, 83, 2705−2711. (22) Husek, P. LC-GC Int. 1992, 5, 43−49. (23) Waldhier, M. C.; Dettmer, K.; Gruber, M. A.; Oefner, P. J. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2010, 878, 1103−1112. (24) Husek, P. J. Chromatogr. B: Biomed. Sci. Appl. 1998, 717, 57−91. (25) Villas-Boas, S. G.; Delicado, D. G.; Akesson, M.; Nielsen, J. Anal. Biochem. 2003, 322, 134−138. (26) Villas-Bôas, S. G.; Smart, K. F.; Sivakumaran, S.; Lane, G. A. Metabolites 2011, I, 3−20. (27) Dettmer, K.; Almstetter, M. F.; Appel, I. J.; Nürnberger, N.; Schlamberger, G.; Gronwald, W.; Meyer, H. H. D.; Oefner, P. J. Electrophoresis 2010, 31, 2365−2373. (28) Dettmer, K.; Nürnberger, N.; Kaspar, H.; Gruber, M. A.; Almstetter, M. F.; Oefner, P. J. Anal. Bioanal. Chem. 2011, 399, 1127− 1139. (29) Smyth, G. K. Stat. Appl. Genet. Mol. Biol. 2004, 3, Article 3. (30) Josse, J.; Husson, F. J. SFdS 2012, 153, 79−99. (31) Benjamini, Y.; Hochberg, Y. J. R. Stat. Soc. Ser. B: Stat. Methodol. 1995, 57, 289−300. (32) Pollard, K. S.; Dudoit, S.; van der Laan, M. J. MULTTEST Multiple Testing Procedures and Applications to Genomics; Division of Biostatistics, University of California: Berkeley, CA, 2004. (33) Wishart, D. S.; Knox, C.; Guo, A. C.; Eisner, R.; Young, N.; Gautam, B.; Hau, D. D.; Psychogios, N.; Dong, E.; Bouatra, S.; Mandal, R.; Sinelnikov, I.; Xia, J.; Jia, L.; Cruz, J. A.; Lim, E.; Sobsey, C. A.; Shrivastava, S.; Huang, P.; Liu, P.; Fang, L.; Peng, J.; Fradette, R.; Cheng, D.; Tzur, D.; Clements, M.; Lewis, A.; De Souza, A.; Zuniga, A.; Dawe, M.; Xiong, Y.; Clive, D.; Greiner, R.; Nazyrova, A.; Shaykhutdinov, R.; Li, L.; Vogel, H. J.; Forsythe, I. Nucleic Acids Res. 2009, 37, D603−610. (34) Shahin, M. M. J. Chem. Phys. 1966, 45, 2600−2605. (35) Warren, C. R. Metabolomics 2013, 9, S110−S120. (36) Sunner, J.; Nicol, G.; Kebarle, P. Anal. Chem. 1988, 60, 1300− 1307. (37) Sumner, L. W.; Amberg, A.; Barrett, D.; Beale, M. H.; Beger, R.; Daykin, C. A.; Fan, T. W.; Fiehn, O.; Goodacre, R.; Griffin, J. L.; Hankemeier, T.; Hardy, N.; Harnly, J.; Higashi, R.; Kopka, J.; Lane, A. N.; Lindon, J. C.; Marriott, P.; Nicholls, A. W.; Reily, M. D.; Thaden, J. J.; Viant, M. R. Metabolomics 2007, 3, 211−221. (38) Strehmel, N.; Hummel, J.; Erban, A.; Strassburg, K.; Kopka, J. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2008, 871, 182−190.

9195

dx.doi.org/10.1021/ac502133r | Anal. Chem. 2014, 86, 9186−9195