Improved Data-Dependent Acquisition for Untargeted Metabolomics

Article pubs.acs.org/ac

Improved Data-Dependent Acquisition for Untargeted Metabolomics Using Gas-Phase Fractionation with Staggered Mass Range Zhixiang Yan and Ru Yan* State Key Laboratory of Quality Research in Chinese Medicine, Institute of Chinese Medical Sciences, University of Macau, Taipa, Macao 999078, China S Supporting Information *

ABSTRACT: The inherently limited MS2 rate of the mass spectrometer is still restricting the performance of data-dependent acquisition (DDA) for untargeted metabolomics. When dealing with the complex metabolome ocean, top-n-based DDA is just scratching the surface as only a small fraction of the ions could be selected for MS2. Here, we report an improved DDA method for untargeted metabolomics using gas-phase fractionation with staggered mass range (sGPF). Unlike the single m/z segment for conventional GPF, the m/z segments for sGPF were narrower, multiplex, and discrete to allow more homogeneous selection of precursor ions in low, medium, and high m/z ranges. This was achieved indirectly by predefining an inclusion list containing multiple discontinuous m/z ranges. Five fraction levels (2, 4, 6, 8, and 10) and two staggering strategies (staggered wide and narrow subsegments (sGPFa and sGPFb)) were compared for characterizing the human urinary metabolites. For both targeted and untargeted comparison, the highest MS2 coverage was obtained by sGPFb8. Targeted comparison of 60 metabolites indicated sGPFb performed the best for 2, 4, 6, and 8 fractions with an increased MS2 triggering rate of 15.0−36.6% over GPF and 6.6− 11.7% over sGPFa. For untargeted screening of phase II metabolites and carboxylates, the best performance achieved by sGPFb8 exhibited a 46.9% increase over GPF8 with the increase evenly distributed in glucuronides (54 vs 38), sulfates (55 vs 41), and carboxylates (31 vs 16). Such superiority of sGPF over GPF is mainly due to the reduced number of concurrent precursor ions and increased relative ion intensity ranks.

T

significantly increased cycle time so that single MS and multiple MS2 could be simultaneously collected. The general DDA criteria include MS1 intensity threshold, dynamic exclusion, dynamic background subtraction, and top n most-intense precursor ions subjected to MS2. As an inherent limitation, the top n strategy fails to trigger low-abundance ions and therefore leads to insufficient identification. To select the ions of interest, more specific criteria available for DDA can be employed such as isotope pattern,10 mass defect,11 inclusion and exclusion lists.12 Compared with data-independent acquisition (DIA), using much wider precursor ion selection windows (from 25 amu as in SWATH,13 to extremely broad range as in MSE14), DDA acquires qualitatively better MS2 spectra but with a lower MS2 acquisition hit rate.15 Recently, as a straightforward but effective approach, gasphase fractionation (GPF), dividing the full MS1 window into several smaller m/z segments to reduce the number of concurrent precursor ions has been demonstrated to offer enhanced untargeted metabolome identification.16 In this study,

here has been tremendous interest in using metabolomics for disease-specific biomarkers discovering drug toxicity and therapeutic evaluation.1−5 Metabolomics analysis with LCMS has gone mainstream because of its unmatched sensitivity and minimal sample preparation. Untargeted metabolomics which aim to determine all possible metabolites differentially expressed under different biological or experimental conditions have been widely used. Compared to targeted quantification, the untargeted metabolomics requires a time-consuming workflow. The samples are first screened by full-scan to obtain the mass-to-charge (m/z) and relative abundance of all ion features detected. Then discriminatory ions between groups are extracted using bioinformatic software. Multiple subsequent tandem MS (MS2) of interested precursor ions is carried out to elucidate and confirm the chemical structures of potential biomarkers by comparing the data with standards or public MS2 databases such as MassBank,6 Metlin,7 and HMDB.8 Recently, an accelerated workflow for untargeted metabolomics has been proposed by matching the experimental MS2 data with MS2 data in the Metlin database in an automated fashion.9 In untargeted metabolites profiling, MS2 spectra are usually collected through data-dependent acquisition (DDA), which uses a narrow precursor isolation window, typically 1−3 amu wide. It enables “on the fly” acquisition of MS2 without © XXXX American Chemical Society

Received: November 19, 2014 Accepted: February 5, 2015

A

DOI: 10.1021/ac504325x Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Mass segmenting principles of GPF (A), sGPFa (B), and sGPFb (C). Density maps of the ion features detected in human urine in terms of retention time, m/z, and ion intensity using RPLC/ESI− (D) and HILIC/ESI+ (E).

acetic acid in water and (B) acetonitrile. The RPLC at a flow rate of 400 μL min−1 was programmed as follows: 0−2 min, 5% B; 2−4 min, 5%−35% B; 4−10 min, 35%−70% B; 10−12 min, 70%−100% B; 12−15 min, 100% B; 15−17 min, 100%−5% B; 17−21 min, 5% B. The HILIC at a flow rate of 300 μL min−1 was programmed as follows: 0−2 min, 95% B; 2−4 min, 95%− 65% B; 4−10 min, 65%−55% B; 10−12 min, 55%−50% B; 12− 15 min, 50% B; 15−17 min, 50%−95% B; 17−22 min, 95% B. The mobile phase flow was diverted to waste after 15 min to prevent contamination of mass spectrometer. Mass Spectrometry. The MS system used was a 4000 QTRAP System (Applied Biosystems, Foster City, CA) controlled by Analyst 1.5.1 and equipped with a Turbo V ion source. The ESI source was set using the following parameters: the curtain gas and gases 1 and 2 set to 10, 20, and 30 psi, respectively; the CAD gas, high; the ion source voltage, +5500 and −4500 V for positive (ESI+) and negative (ESI-) ionization modes, respectively; the source temperature, 250 °C; the declustering potential (DP), −50 or +40 V for ESI+ and ESImodes, respectively. A survey MS1 scan was acquired from m/z 100−800 in at 1000 Da s−1. The four most intense precursor ions in different GPF and sGPF ranges exceeding 8000 counts per second were selected for two data-dependent product ion acquisitions at 4000 Da s−1 (step size 0.08 Da) followed by dynamic exclusion of 15 s. Dynamic fill time was selected. The CE was +24 and −30 V with a spread of 12 V and −15 V for ESI+ and ESI−, respectively. One microscan was acquired for each MS1 and MS2 spectrum. Generation of Inclusion Lists for GPF and sGPF. GPF and sGPF were achieved indirectly by predefining an inclusion list for DDA. A new form of inclusion list was used in this study which covered selected mass ranges rather than targeted specific ion masses in the traditional way. Thus, MS2 acquisition was only triggered for precursor ions within that m/z range. The step size of the listed m/z was 0.1 amu (e.g., 100, 100.1, 100.2...) and a mass tolerance of 0.25 amu was used so that all the m/z in the range was covered. The step size and mass tolerance could be adjusted according to the resolution of the employed instrument. Retention time and window width for all listed m/z values were the same. In this case, they were

we demonstrated that the conventional GPF, based on a single m/z segment in each analysis, is not the most efficient DDA for untargeted metabolomics due to the different number and density of metabolite precursor ions between segments. We investigated the possibility of using multiple staggered narrower m/z segments to increase the MS2 depth of GPF for untargeted metabolomics. This strategy, however, is not directly supported by current instrument controlling software, which only allows one m/z segment being selected for DDA. To circumvent this limitation, we employed a novel version of inclusion list which contains discrete m/z ranges rather than the specific m/z values. This successfully restricted the DDA precursor selection from predefined staggered mass ranges. By targeted and untargeted comparison of the DDA of human urinary metabolome, we demonstrated the staggered GPF (sGPF) could significantly increase the MS2 depth of GPF by 20% to 50%.

■

EXPERIMENTAL SECTION Reagents and Materials. Acetonitrile, acetic acid, ammonium acetate, and formic acid were from Merck (Darmstadt, Germany). Standards of adenine, choline, hippuric acid, L-agmatine, L-asparagine, L-carnitine, L-glutamine, Lhistidine, L-leucine, L-lysine, L-phenylalanine, L-taurine, Lthreonine, L-tryptophan, and L-tyrosine were purchased from Sigma-Aldrich (St. Louis, MO, U.S.A.). Urine Sample Collection and Preparation. Pooled urine samples were collected from two healthy volunteers in our laboratory, aliquoted and stored at −80 °C immediately after collection. Sample preparation was methanol precipitation (1:1 vol/vol) with the supernatant further diluted with pure water (1:1 vol/vol) and filtered through a 0.22 μm membrane before LC-MS analysis. Liquid Chromatography. The chromatographic experiments were performed using an Agilent 1200 LC system with a Kinete XB-C18 (2.1 × 50 mm, 1.7 μm) or a Kinete HILIC (2.1 × 100 mm, 1.7 μm). Column temperature was 35 °C, injection volume was 4 μL, and autosampler temperature was 4 °C. The mobile phase for positive/negative ion modes consisted of (A) 10 mM ammonium acetate and 0.1% formic acid/0.0125% B


Article

Analytical Chemistry Table 1a. m/z Ranges in Inclusion Lists for DDA-MS2 in GPF, sGPFa, and sGPFb

7.5 min and 900 s, respectively to cover the entire LC-MS profile (15 min). The inclusion lists were readily generated

using Microsoft Excel which took about 2 min for each and 60 min for the complete GPF and sGPF lists tested in this study. C


Article

Analytical Chemistry Table 1b

They were exported as tables (.csv file format) that can be read by the MS instrument software and imported to the DDA inclusion functionality table. Once these lists were created, they could be used as a template for all the similar LC-MS based metabolomics. The full MS range was divided into 2, 4, 6, 8, and 10 sequential (Figure 1A) and staggered (Figure 1B,C) fractions for GPF and sGPF, respectively (Table 1a, 1b). For sGPF, the fraction is divided into multiple subsegments, and two sGPF strategies were further compared: one being increasing the width of each subsegment with reduced segmenting frequency (sGPFa, Figure 1B), and the other being increasing the number of staggered subsegments with each segment unit spanning narrower m/z range (sGPFb, Figure 1C). As a demonstration, Table S1 (Supporting Information) provided the inclusion list for sGPFa10 (10−1). Data Processing and Analysis. All DDA-MS2 spectra were extracted and viewed using the IDA-explorer of Analyst 1.5.1. Neutral loss and product ion filtering of MS2 spectra (acquired from 1.0 to 12.0 min) were carried out using a script, IDAtractor.17

reaches saturation, and many of them are likely to be untriggered due to the limited acquisition rate of a mass spectrometer. On the contrary, at the low coeluting region of LC-MS profiles, most precursors will be subjected to MS2, but the instrument’s capacity is not fully employed. To increase the DDA-MS2 efficiency, we used multiple staggered narrow m/z segments for sGPF, which featured more uniform selection of precursor ions by encompassing low, medium, and high m/z ranges within a single experiment (Table 1a,1b, Figure 1B,C). Because only one mass range could be selected for GPF by the instrument software, we used the inclusion list to restrict the DDA precursor ion selection from multiple staggered m/z ranges. The widely known advantage of GPF over conventional DDA was not evaluated. As shown in Figure 1A, a large number of metabolites with abundant ion features at m/z 300−550 eluted within the retention time of 6−10 min in RPLC/ESImode; thus, GPF at this segment led to a large portion of untriggered precursor ions. As exemplified in Figure S1 (Supporting Information), only 2 MS2 acquisition was triggered for the ion at m/z 441.2 using GPF10−4 (targeting m/z 380−450), whereas 5 MS2 were triggered using sGPFa10 and sGPFb10. Because identical instrument settings were used for GPF and sGPFa and sGPFb with the only exception being the precursor-ion inclusion list, essentially consistent numbers of MS2 spectra were acquired across different methods, and their different DDA performances were reasonably attributed to the different m/z segments employed for GPF and sGPF. We further investigated why MS2 of m/z 441.2 at 6.56 min was only triggered by sGPFb10. This precursor ion with an intensity of 1.0 e6 was below the top 15 in GPF10, increased to the top 10 in sGPFa10, and boosted to the third most abundant ion in sGPFb10, where a new order of ion ranks in the candidates for MS2 were established (Figure S2, Supporting Information). Apparently, there were far less precursor ions in other mass ranges at this moment of chromatographic time,

■

RESULTS AND DISCUSSION Principle of sGPF. Although each segment in conventional GPF covers an equal m/z range, the covered precursors were distributed unevenly in either RPLC-MS- or HILIC-MS-based urinary metabolites profiling even with optimized gradients (Figure 1D and E). The heterogeneous distribution of serum metabolites were also observed in RPLC/ESI+ and RPLC/ESIMS.16 This is because each kind of submetabolome generally has similar retention time and resides in the same mass range. For instance, amino acids (100−250 amu) generally elute within the first few minutes of the gradient on RP C18 columns, where fatty acids (200−350 amu) and lysophospholipids (400−600 amu) elute in the second half of the run.16 For metabolites in the high coeluting zone, DDA-MS2 easily D


Article


Figure 2. Comparison of the DDA of 60 identified metabolites using GPF, sGPFa, and sGPFb. DDA-MS2 spectra that generated appropriate precursor and product ion information for a metabolite are indicated by a green box, and those that did not contain the data are indicated by a red rectangle. * Metabolite identity was confirmed by standard.

and many precursor ions in these ranges (e.g., m/z 100−150 and 600−800) do not have a large enough intensity to obtain informative fragmentation. This means when GPF spanning these “barren” mass ranges, the instrument will be partially idling and most obtained MS2 spectra will be of rather limited quality or useless for identification, whereas many other more abundant and productive precursor ions in GPF (m/z 300− 550), at this point in the gradient, are still suffering from insufficient MS2. The sGPF methods such as sGPFa10 and sGPFb10 shrinks the segment width to 10 and 5 amu and picks segments at an interval of 100 and 50 amu, respectively. This guarantees ion representation throughout the entire m/z range (Table 1a,1b) and homogenizes ion intensities so that each analytical run could deal with less abundant precursor ions in the high coeluting region and more abundant ions in the low coeluting region, therefore increasing the DDA-MS2 efficiency. Comparison of Targeted DDA-MS2. In this study, five levels (2, 4, 6, 8, and 10) of fraction were compared for GPF and sGPF. Because the entire comparison involved hundreds of injections, each sample aliquot was analyzed for at most 16 h to obtain the best comparability by minimizing the variation of MS1 survey spectra due to potential sample degradation. We first evaluated the DDA of 60 identified metabolites (Figure 2) chosen at random in HILIC/ESI+ MS mode (Table S2, Supporting Information) covering different retention time and

relative intensities. Out of the three methods discussed, GPF yielded the lowest triggering rate at all levels while sGPFb performed the best for 2, 4, 6, and 8 fractions with an increased triggering rate of 15.0−36.6% over GPF and 6.6%−11.7% over sGPFa (Figure 3A). Among the five fraction levels, sGPFb8 offered the highest MS2 coverage of 91.7%. These results indicate employing more staggered segments with reduced m/z window improves the performance of sGPF. At 10 fractions level, sGPF only showed slightly increased MS2 coverage over GPF. It should be noted that repeat analysis did not increase MS2 triggering rate, especially for low-abundance ions, as it could not fundamentally change the ion populations competing for DDA. Indeed, the improved MS2 coverage of sGPF mainly resided in low-abundance metabolites such as M2, 27, and 51 (Figure 2) whose intensity were ranked below the top 20 in full MS1. The superiority of sGPF over GPF could be explained by the reduced number of concurrent precursor ions and most importantly the increased ion intensity ranks (Figure 4) due to the homogeneous inclusion of precursor ions. The limitation of GPF is especially evident when analyzing M27, because there was no effective reduction of concurrent precursor ions for all 5 GPF fractions. Furthermore, unlike sGPF, the reduction of precursors in GPF generally did not contribute to a proportional increase in the ion intensity ranks. As exemplified E


Article


Untargeted Comparison. To further evaluate the sGPF performance, we next compared the DDA in RPLC/ESI- MS containing a large number of phase II metabolites mainly including glucuronides and sulfates and fewer mercapturic acids (MA), which afford the characteristic neutral loss of 176, 80, and 129 Da, respectively. Exceptions are aliphatic sulfates which predominantly generate a product ion at m/z 97 (HSO4− at m/ z 96.959). Standards of these kinds of metabolites are usually scarce or unavailable, and their reference MS2 spectra are not included in public MS2 database. Consequently, an automatic library search will bring out low identification rate. ESI− mode also affords many carboxylates which readily lose a molecule of CO2 (44 amu). To eliminate the bias caused by targeted comparison, we used neutral loss filtering (±0.5 amu) of 176, 80 (in combination with product ion filtering of 97), 129, and 44 amu to compare the untargeted detection of glucuronides, sulfates, mercapturic acids, and carboxylates, respectively, regardless of their exact identities. Intriguingly, similar to the targeted comparison, the best performance of untargeted comparison with respect to potential phase II metabolites and carboxylates was also achieved by sGPFb8, a 46.9% increase over GPF8 (Figure 3B). The increased coverage was evenly distributed in glucuronides (54 vs 38), sulfates (55 vs 41), and carboxylates (31 vs 16), suggesting a comprehensive improvement rather than skewed contribution to certain subgroups of metabolites. For the GPF and sGPFa methods, the highest number of phase II metabolites and carboxylates was triggered by GPF10 (118) and sGPFa10 (125) at the expense of 2 additional injections but was still less than (19.5% and 12.8%) that of sGPFb8 (141). On the other hand, some metabolites were solely triggered by GPF10 or sGPFa10 (Figure 3C), indicating there were still more precursor ions in sGPFb8 than that can be fragmented by the instrument. With staggered m/z segments in sGPF, overwhelming concurrent precursor ions for DDA happens less frequently, but it is still difficult to avoid without corresponding increase in the MS2 acquisition rate. Furthermore, the increase offered by some sGPF methods in

Figure 3. Targeted comparison of the MS2 triggering rate of 60 identified metabolites in Figure 2 (A). Histogram representation of aggregated nonredundant number of potential sulfates affording neutral loss of 80 (NL80) and/or product ion at m/z 97 (PI97), glucuronides affording neutral loss of 176 (NL176), and carboxylates affording neutral loss of 44 (NL44) detected by GPF, sGPFa, and sGPFb (B). The increased percentage (>5%) of metabolites characterized by sGPFa and sGPFb over GPF is indicated on the top of column. Venn diagram of glucuronides, sulfates, and carboxylates triggered by sGPFa10, sGPFb8, and GPF10 (C).

by M2 in Figure 4, GPF4, 6, 8, 10 merely offered a 0.04-, 0.6-, 0.6-, and 1.0-fold increase in intensity ranks of targeted ion while reducing the concurrent precursors by 75.5, 83.0, 83.0, and 86.2%, respectively. These results indicate that the conventional GPF fractionation strategy based on a single segment even as narrow as 70 m/z in GPF10 is still insufficient in reducing precursor-ion complexity due to the uneven distribution of precursor ions and this problem could be mitigated using multiple staggered m/z ranges for sGPF.

Figure 4. Precursor ion information on metabolites 27, 2, and 51 (M27, M2, M51) selected from Figure 2. Number of concurrent precursor ions competing for DDA-MS2 (A) and relative intensity rank of targeted precursor ion (B) at the retention time of interested metabolite. Compared with GPF, the reduction of precursors in sGPFa and sGPFb results in more pronounced increase in the ion intensity ranks. Top 1, 2, 3... indicate the first, second, third... most abundant precursor ions, respectively. F


Article


sulfates characterized in our study have not been reported in these three previous studies, suggesting both GPF and sGPF could effectively improve the MS/MS coverage with sGPF digging dipper into the complex metabolome. Limitations of sGPF. The current study suggests that the best MS2 coverage of untargeted urinary metabolomics using a 4000 QTRAP could be achieved by sGPFb8. Because metabolomic workflows require repetitive analysis of the same sample to minimize experimental variability, the sGPF implemented by eight injections can be applied to a study design with two biological replicates and four technical replicates. However, this could be a limitation in the case of valuable samples. Furthermore, for large-scale metabolomic studies involving hundreds of samples, sample degradation should be considered because sGPFb8 greatly increases the analysis time. It should also be noted that for untargeted comparison, the sGPF and GPF performed roughly equivalently at some faction levels. Our further study will evaluate the performance of sGPF for untargeted metabolomics using other tandem mass spectrometers including Q-Tof and LTQorbitrap with different MS2 acquisition rates.

untargeted comparison was not as significant as the increase in targeted comparison. Although the above comparisons may not give the best numeric representation of the performance of GPF and sGPF, our results clearly indicate higher information content could be extracted from sGPF-DDA due to its increased MS2 efficiency. The significance of informative MS2 spectra lies in that metabolites with a common structural moiety and thus diagnostic fragment ion or neutral loss could be used to monitor the class of interest (i.e., phase II metabolites as discussed below). Characterization of metabolites at class level may be a reasonable approach because a large number of metabolites are absent from current MS/MS databases with this deficit being particularly acute for lowabundance and unknown metabolites. Only several potential MAs were detected using neutral loss filtering of 129 for DDA of GPF and sGPF. This was not surprising because MA derives from conjugates of reactive endogenous or exogenous compounds with glutathione (GSH).18 Our results indicated that in normal conditions there are negligible reactive metabolites in vivo. However, smoking, oxidative stress and damage, and environmental exposure, could lead to increased MA excretion.19 Importantly, one reported MA, 1,4-dihydroxy-2(E)-nonene mercapturic acid (DHN-MA), a major metabolite of 4-hydroxy-2-nonenal (HNE),20 was triggered by sGPFb4, GPF6 and sGPFa6. HNE is the end products of lipid peroxidation as a result of the formation of reactive oxygen species (ROS). DHN-MA could represent a specific and noninvasive biomarker of lipid peroxidation in vivo. The MS2 spectra of this metabolite has only been reported by Scholz et al.21 using multiple reaction monitoring triggered MS2 and is not included in public MS2 databases. To our best knowledge, this is the first time that this metabolite was detected by full scan triggered DDA and the experimental MS2 spectrum of DHN-MA was provided in Figure S3 (Supporting Information). Product ion filtering of 97 revealed both singly and doubly charged precursor ions. Those singly charged precursors generated a dominant product ion at m/z 97 (HSO4−). One precursor ion at m/z 471.0 was tentatively identified as one of the three isomers, ursodeoxycholic acid 3-sulfate (UDCA3S), deoxycholic acid 3-sulfate (DCA3S), chenodeoxycholic acid 3sulfate (CDCA3S).22 Two doubly charged precursor ions at m/ z 280.7 and m/z 288.7 were identified as tauro-lithocholic acid 3-sulfate and one of the three isomers, tauro-UDCA3S, tauroDCA3S, and tauro-CDCA3S, respectively.22 They produced the product ion [M-H-98 (H2SO4)]− and fragment at m/z 97 which is in accordance with the reported fragmentation behavior of taurine conjugated bile acid 3-sulfates (Figure S6, Supporting Information).22 A series of doubly charged ions [M2H]2− affording product ion [M-H-75]− and fragments at m/z 97 and 74 were assigned as glycine conjugated bile acid 3sulfates (Figure S4, Supporting Information). Among them, the ion at m/z 263.7 was identified as one the three isomers, glycoUDCA3S, glyco-DCA3S, and glyco-CDCA3S.22 These conjugated or unconjugated bile acid 3-sulfates were also absent in MS2 databases. The above results suggest manual data-mining is indispensable to extract more useful information from DDAMS2. Also, including doubly charged precursors in DDA criteria could increase the sampling depth for metabolomics. To the best of our knowledge, the most comprehensive untargeted metabolites profiling of human urine have been carried out by Zhang et al.,23 Roux et al.,24 and Zhang et al.,25 where only traditional DDA was employed. DHN-MA and bile acid 3-

■

CONCLUSION In this study, by comparing the single segment, staggered wide segments (sGPFa), and staggered narrow segments (sGPFb) fractionating strategies, we demonstrated sGPF using discrete narrower m/z segments (sGPFb) significantly increased the DDA-MS2 depth for untargeted metabolomics. This sGPFb establishes a new order of ion ranks in the precursor candidates for MS2, allowing more homogeneous selection of precursor ions, producing more informative MS2, and picking additional minor precursors for MS2. Since inclusion ion functionality for DDA is available for most tandem MS instruments, the broader application of this sGPF method is possible.

■

ASSOCIATED CONTENT

S Supporting Information *

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

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: + 853-88224682. Fax: 85328841358. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was financially supported by the National Natural Science Foundation (ref. no: 81473281), University of Macao (ref. no.: MYRG162), and the Science and Technology Development Fund of Macao (ref. no.: FDCT043/2011/A2).

■

REFERENCES

(1) Sreekumar, A.; et al. Nature 2009, 457, 910−914. (2) Spratlin, J. L.; Serkova, N. J.; Eckhardt, S. G. Clin. Cancer Res. 2009, 15, 431−440. (3) Pasikanti, K. K.; et al. J. Proteome Res. 2010, 9, 2988−2995. (4) Nordstrom, A.; Lewensohn, R. J. NeuroImmune Pharmacol. 2010, 5, 4−17. (5) Keun, H. C. Pharmacol. Ther. 2006, 109, 92−106. (6) Horai, H.; Suwa, K.; Arita, M.; Nihei, Y.; Nishioka, T. Mass Spectral Database for Metabolome Analysis. Proceedings of the 55th G


Article

Analytical Chemistry ASMS Conference on Mass Spectrometry and Allied Topics, Indianapolis, Indiana, June 3−7, 2007. (7) Smith, C. A.; O’maille, G.; Want, E. J.; Qin, C.; Trauger, S. A.; Brandon, T. R.; Custodio, D. E.; Abagyan, R.; Siuzdak, G. Ther. Drug Monit. 2005, 27, 747−751. (8) Wishart, D. S.; Tzur, D.; Knox, C.; Eisner, R.; Guo, A. C.; Young, N.; Cheng, D.; Jewell, K.; Arndt, D.; Sawhney, S.; Fung, C.; Nikolai, L.; Lewis, M.; Coutouly, M. A.; Forsythe, I.; Tang, P.; Shrivastava, S.; Jeroncic, K.; Stothard, P.; Amegbey, G.; Block, D.; Hau, D. D.; Wagner, J.; Miniaci, J.; Clements, M.; Gebremedhin, M.; Guo, N.; Zhang, Y.; Duggan, G. E.; Macinnis, G. D.; Weljie, A. M.; Dowlatabadi, R.; Bamforth, F.; Clive, D.; Greiner, R.; Li, L.; Marrie, T.; Sykes, B. D.; Vogel, H. J.; Querengesser, L. Nucleic Acids Res. 2007, 35, D521− D526. (9) Tautenhahn, R.; Cho, K.; Uritboonthai, W.; Zhu, Z.; Patti, G. J.; Siuzdak, G. Nat. Biotechnol. 2012, 30, 826−828. (10) Lim, H.-K.; Chen, J.; Cook, K.; Sensenhauser, C.; Silva, J.; Evans, D. C. Rapid Commun. Mass Spectrom. 2008, 22, 1295−1311. (11) Ramagiri, S.; Miller, J.; Ghobarah, H. Simultaneous Pharmacokinetic Profiling and Automated Metabolite Identification using the AB Sciex TripleTOF 5600 System and MetabolitePilot Software, 2010. AB Sciex Web site. http://www.absciex.com/ Documents/Downloads/Literature/mass-spectrometry-BuspironePK1270210.pdf (accessed Sept 20, 2013). (12) Hoffmann, T.; Krug, D.; Hüttel, S.; Müller, R. Anal. Chem. 2014, 86, 10780−10788. (13) Hopfgartner, G.; Tonoli, D.; Varesio, E. Anal. Bioanal. Chem. 2012, 402, 2587−2596. (14) Plumb, R. S.; Johnson, K. A.; Rainville, P.; Smith, B. W.; Wilson, I. D.; Castro-Perez, J. M.; Nicholson, J. K. Rapid Commun. Mass Spectrom. 2006, 20, 1989−1994. (15) Zhu, X.; Chen, Y.; Subramanian, R. Anal. Chem. 2014, 86, 1202−1209. (16) Calderón-Santiago, M.; Priego-Capote, F.; Luque de Castro, M. D. Anal. Chem. 2014, 86, 7558−7565. (17) Yan, Z.; Lin, G.; Ye, Y.; Wang, Y.; Yan, R. J. Am. Soc. Mass Spectrom. 2014, 25, 955−965. (18) Haufroid, V.; Lison, D. Int. Arch. Occup. Environ. Health 2005, 78, 343−354. (19) Wagner, S.; Scholz, K.; Donegan, M.; Burton, L.; Wingate, J.; Völkel, W. Anal. Chem. 2006, 78, 1296−1305. (20) Alary, J.; Debrauwer, L.; Fernandez, Y.; Cravedi, J. P.; Rao, D.; Bories, G. Chem. Res. Toxicol. 1998, 11, 130−135. (21) Scholz, K.; Dekant, W.; Völkel, W.; Pähler, A. J. Am. Soc. Mass. Spectrom. 2005, 16, 1976−1984. (22) Goto, T.; Myint, K. T.; Sato, K.; Wada, O.; Kakiyama, G.; Iida, T.; Hishinuma, T.; Mano, N.; Goto, J. J. Chromatogr. B 2007, 846, 69− 77. (23) Zhang, X.; Clausen, M. R.; Zhao, X.; Zheng, H.; Bertram, H. C. Anal. Chem. 2012, 84, 7785−7792. (24) Roux, A.; Xu, Y.; Heilier, J. F.; Olivier, M. F.; Ezan, E.; Tabet, J. C.; Junot, C. Anal. Chem. 2012, 84, 6429−6437. (25) Zhang, T.; Creek, D. J.; Barrett, M. P.; Blackburn, G.; Watson, D. G. Anal. Chem. 2012, 84, 1994−2001.

H


Improved Data-Dependent Acquisition for Untargeted Metabolomics

Recommend Documents