Analysis of LC-MS Data for Characterizing the Metabolic Changes in Response to Radiation

Transcription

1 Analysis of LC-MS Data for Characterizing the Metabolic Changes in Response to Radiation Rency S. Varghese,, Amrita Cheema,, Prabhdeep Cheema, Marc Bourbeau, Leepika Tuli, Bin Zhou, Mira Jung, Anatoly Dritschilo,*,, and Habtom W. Ressom*, Department of Oncology, Department of Radiation Medicine, Lombardi Comprehensive Cancer Center, Georgetown University Medical Center, Washington, DC, and Hartwick College, Oneonta, New York Received March 1, 2010 Abstract: Recent advances in mass spectrometry-based metabolomics have created the potential to measure the levels of hundreds of metabolites that are the end products of cellular regulatory processes. In this study, we investigate the metabolic changes in genetically engineered cell lines in response to radiation exposure. Shrinkage t statistic and partial least-squares-discriminant analysis methods are utilized to identify peaks whose signal intensities were significantly altered by radiation. This is accomplished through pairwise comparison of radiation treated cell lines at various time points following radiation against untreated cell lines. A pathway analysis is performed following identification of the metabolites represented by the selected peaks. The results indicate an ATM regulated induction of major pathways in response to radiation treatment. Keywords: metabolomics LC-MS SIMCA shrinkage t statistic PLS-DA Introduction Genomics and proteomics have been used extensively to study radiation-induced molecular events. However, little is known about the alterations in the levels of small molecule metabolites in biological systems in response to radiation. Metabolomics, a methodology for measuring small-molecule metabolite profiles and fluxes in biological matrices, following genetic modification or exogenous changes, has become an important component of systems biology, complementing genomics, transcriptomics, and proteomics. 1,2 While still a relatively new member of the omics family, metabolomics is already finding applications in a wide variety of research fields including disease diagnosis, 3 drug toxicity studies, 4 and functional genomics. 5 Metabolomics is the large-scale analysis of metabolites and as such requires bioinformatics tools for data analysis, visualization, and integration. Typically hundreds to thousands of compounds or spectral features can be seen in a typical high throughput metabolomic assay. * To whom correspondence should be addressed. H.W. Ressom, hwr@georgetown.edu. A. Dritschilo, dritscha@georgetown.edu. These authors contributed equally to this manuscript. Department of Oncology, Georgetown University Medical Center. Hartwick College. Department of Radiation Medicine, Georgetown University Medical Center. Multiple analytical techniques are being currently used for resolving and detecting thousands of metabolites each characterized by its own set of advantages and disadvantages. Choice of analytical technique is based on metabolites of interest and the subsequent methods of their analysis. Most commonly used platforms for metabolite profiling include mass spectrometry (MS), nuclear magnetic resonance (NMR), liquid chromatography coupled with electrochemical coulometric array detection (LCECA), Fourier transform infrared (FTIR) spectrometer, and imaging technologies. 4 Since no single method is best suited for detecting and identifying all metabolites (intra- and/or extra-cellular), a combination of techniques is needed to encompass most, if not all, of them. With advances in instrumentation, MS is the technique of choice for metabolomic studies. Small as well as large metabolites can be detected with high sensitivity and specificity. Following sample preparation and chromatographic separation, the ionized metabolites are fragmented and analyzed for their molecular composition and key substructures. 4 With tremendous progress being made in MS-based metabolomics, a variety of choices are available for chromatographic separation, mass analyzers, and use of spectral information for metabolite identification. Gas chromatography coupled with mass spectrometry (GC-MS) is an established technique well suited for thermally stable metabolites including compounds such as sugars, fatty acids, amino acids, aromatic amines, and sugar alcohols. 6 While volatile analytes can be directly separated and quantified by GC-MS, polar metabolites require chemical derivatization at the functional group to increase volatility and thermal stability prior to separation. 6 An alternative to GC-MS is liquid chromatography coupled with mass spectrometry (LC-MS), which not only provides flexibility in terms of combination of different separation mechanisms but also precludes the need for metabolite derivatization. 6 Since LC is not limited to one separation mode, it can be easily customized to a certain class of metabolites using reverse-phase, normal phase, size exclusion, hydrophilic interaction chromatography, etc. 7 It is applicable to compounds with varying polarity such as lipids, peptides, nucleotides, etc., making it more suitable to global metabolomics. Ultra performance liquid chromatography (UPLC) further enhances chromatographic resolution and peak capacity while enabling low detection limits. 7 Following chromatographic separation, a mass spectrometer separates ions based on their mass-to-charge ratio (m/z) ina /pr100185b XXXX American Chemical Society Journal of Proteome Research XXXX, xxx, 000 A

2 technical notes mass analyzer. For example, the hybrid quadropole-time-offlight (Q-TOF), a high mass resolution ( ) analyzer, filters ions in the quadropole based on mass trajectory. Ions of interest having amplitudes less than half the radius of the rods are retained within the rods. The Q-TOF can select precursor ions for fragmentation after traversing through the collision cell and offers a mass accuracy in the 5-10 ppm range. 10 Additionally, Q-TOF is easily maintained and can be efficiently coupled to LC. Based on prevalent practices, metabolomic analysis by LC-MS can be categorized as either bottom-up (targeted) or top-down (nontargeted) profiling. 11 The bottom-up approach is a targeted profiling scheme where metabolites are identified prior to quantification. Experimental MS/MS data are compared to a spectral library of reference compounds; however accuracy can only be ensured if the experimental MS/MS data are collected and processed in a similar way to the ones in the reference library. Quantification can be achieved by comparing peak intensities with internal standards of known concentrations. The limitation of this approach is the lack of large spectral libraries currently available which can skew metabolite identification and interpretation. 11 In the top-down approach, peak intensities are analyzed without necessarily identifying all metabolites. With this approach, the focus is to look at all metabolites at the same time rather than quantifying the ones that are already known. Since no identification is required at the initial level, only MS scans are utilized. The benefit of this holistic approach is balanced by the drawback of requiring further validation of components of interest through comparison of their fragmentation pattern to pure compounds. Also, computational challenges exist in using top-down metabolomic analysis of LC-MS data for identification of metabolic changes between biological groups. The data analysis begins with data preprocessing procedures including filtering, peak detection, peak alignment, and normalization. Filtering methods process the raw measurement signal to remove effects like measurement noise or background signals. Peak detection is used to transform the continuous spectral data into discrete peak list where each peak is represented by a triplet of m/z, RT, and ion abundance (peak intensity). This transformation can bring two advantages: (1) part of the noise in the continuous data is removed; (2) the feature space of the sample is reduced as most of the variation in data can be captured by a relatively small number of peaks. Peak alignment is necessary since the retention time of the same ion can have significant drift in different runs. This drift is generally nonuniform across the retention time range and is a result of various instrumental, experimental and environmental factors which cannot be completely eliminated. 15 Peak alignment methods cluster measurements across different samples and correct the drifts in retention time. One approach is to add reference compounds in the sample and use these compounds as landmarks to guide the alignment for the retention time. 7 Peak alignment approaches without relying on reference compounds were also proposed The alignment is performed based on either total ion chromatograms (TICs) or the three-dimensional LC-MS maps. A critical assessment of several popular alignment software tools have been previously reported, 19 where it was concluded that the performance of various methods is largely based on user experience and the proper choice of parameters. Finally, normalization is needed to reduce the systematic bias and to make the peaks from multiple runs comparable. Several software tools have been Varghese et al. developed to relieve the burden of LC-MS metabolomic data preprocessing from users. These include MarkerLynx, Met- Align, 23 XCMS, 24 and MZmine. 25 Other packages, some of them specific for LC-MS-based metabolomics, have been reviewed. 12 A comparison on MZmine and XCMS for peak detection and alignment have been reported. 26 The comparison suggests that both software packages can be used with proper parameter adjustments. Following data preprocessing, statistical methods are typically used to identify significant differences in metabolic changes between distinct biological groups. Since preprocessed LC-MS and microarray data share common characteristics, statistical methods previously developed for microarray data analysis can be utilized for difference detection. For example, Gene Expression Dynamics Inspector, 28 which was originally developed for microarray data analysis, was used to identify the subsets of dose-dependent metabolites in a study of cellular response to radiation. 29 Other methods, such as Significant Analysis of Microarray (SAM) or Empirical Bayesian Analysis of Microarray (EBAM), have been incorporated into the metabolomics data analysis software MetaboAnalyst. 27 Other tools such as MeltDB 30 provide various methods including t test, principle component analysis (PCA), independent component analysis (ICA), clustering, etc. PCA and partial least-squaresdiscriminant analysis (PLS-DA) methods are widely used to reduce the dimension of metabolomics data and identify relevant peaks Finally, the metabolites represented by the selected peaks are identified and verified. Database search is the most popular metabolite identification method for nontargeted metabolomics study. For each ion of interest, its m/z value is used to retrieve those molecules in databases having monoisotopic weights within prespecified mass measurement accuracy, ionization mode, and modification of metabolites. Several databases have been assembled to retrieve putative identifications The limitation of such mass-based database search is that the identification result is rarely unique. It is shown that even with an m/z accuracy of 1 ppm, it is still insufficient for unambiguous metabolite identification. 38 Thus, it is important that the putative identifications are further verified by comparing their MS/MS spectra with those of authentic compounds. In this paper, we present analytical and computational methods to characterize the metabolic changes that occur in response to radiation exposure by a top-down metabolomic analysis. The goal is to understand the differential response to radiation exposure in two isogenic cell lines namely AT5BIVA and ATCL8 at the metabolic level owing to the differential expression status of ATM and in turn correlate this to the fate of the cells. ATCL8 contains an introduced ATM gene (mutated in ataxia telangiectasia (A-T)). A-T is a genetic disorder characterized by cerebellar degeneration, immunodeficiency, defective recognition of damage to DNA, radiosensitivity, cell cycle checkpoint defects, and cancer predisposition. 39 A-T cells in culture are also characterized by slow growth rate, extreme radiosensitivity, chromosomal instability, specific translocations, defective repair of a subtype of DNA strand breaks, and defective cell cycle checkpoint activation. 39 The AT5BIVA cells used in this study were derived from a patient with A-T exhibiting extreme radiosensitivity. 40 The ATCL8 cell line was derived by transfecting AT5BIVA cells with the wild type, full length ATM gene which resulted in correction of radiosensitivity. B

3 Metabolic Changes in Response to Radiation Figure 1. Overview of our top-down metabolomic analysis. We used XCMS to perform peak detection and peak alignment. Shrinkage t statistic was used to select differentially expressed metabolites between radiation treated and untreated cell lines. A multivariate analysis was also performed using the SIMCA-P+ (Umetrics, Kinnelon, NJ) software. The metabolites corresponding to the selected peaks were identified using massbased databases. The Ingenuity Pathway Analysis (IPA) tool (Ingenuity Systems, Redwood City, CA) was utilized to search for the pathways regulated by the identified metabolites. Some of the selected metabolites are further validated by MS/MS analysis. Figure 1 depicts an overview of a top-down metabolomic analysis we performed for the characterization of AT5BIVA and ATCL8 cells and identification of metabolic changes in the cells in response to radiation. Materials and Methods Samples. Ataxia telangiectasia fibroblast cells AT5BIVA were obtained from the National Institute of General Medical Sciences (NIGMS). AT5BIVA have been previously characterized as complementation Groups A, C, and D (3), and the others are unknown. Cells were maintained in modified Eagle s medium with 20% fetal bovine serum, 100 U/penicillin, and 100 pg/ml streptomycin. For this study the cells were grown to 80% confluence and serum starved for 24 h and subjected to radiation dose of 5Gy. The control cells were not given radiation treatment. The cells were allowed to recover for predetermined periods of time (30 min, 1, 2, 3, 6, and 24 h) by technical notes incubating at 37 C. The cells were then washed twice with chilled phosphate buffered saline (PBS) and harvested by scraping and centrifuging. Total cell count was carried out using a hemocytometer and equal number of cells (10 7 ) from each time point was aliquoted for further processing for metabolomic analysis. The AT5BIVA cells used in this study were derived from a patient with A-T exhibiting extreme radiation sensitivity. 40 The ATCL8 cell line was derived by transfecting AT5BIVA cells with the wild type, full length ATM gene which resulted in correction of radiation sensitivity. In this paper, we used these two isogenic cell lines (AT5BIVA and ATCL8) as a model system for characterizing the metabolic changes that occur in response to radiation exposure in a time-dependent manner. Sample Preparation and LC-MS Data Acquisition. Sample Preparation. The sample preparation for small molecule metabolite profiling using LC-MS was carried out using the method described by Sana et al. 41 Briefly, the cells were lysed using chilled water and sequential metabolite extraction was done using methanol and chloroform. One micromolar Debrisoquine was added to the extraction buffer as internal standard. LC-MS Data Acquisition. An aliquot of each sample supernatant was deposited in an autosampler vial, and 5 µl was separated on a 50 mm 2.1 mm Acquity 1.7 µm C18 column using an Acquity UPLC system online with Q-TOF Premier (Waters Corp, Milford, MA). The method for data acquisition is described by Patterson et al. 29 The gradient mobile phase consisted of 0.1% formic acid (A) and acetonitrile containing 0.1% formic acid (B). A typical 10 min sample run consisted of 0.5 min of 100% solvent A followed by an incremental increase of solvent B up to 100% for the remaining 9.5 min. The flow rate was set to 0.6 ml/min. The eluent was introduced by electrospray ionization into the Q-TOF Premier operating in positive ionization mode. The capillary and sampling cone voltages were set to 3000 and 30 V, respectively. Source and desolvation temperatures were set to 120 and 350 C, respectively, and the cone and desolvation gas flows were set to 50.0 and L/h, respectively. To maintain mass accuracy, sulfadimethoxine ([M + H] + ) ) at a concentration of 500 pg/µl in 50% acetonitrile was used as a lock mass and injected at a rate of 0.08 µl/min. For MS scanning, data were acquired in centroid mode from 50 to 850 m/z, and for MS/ MS, the collision energy was ramped from 5 to 35 V. LC-MS Data Preprocessing. The raw data obtained from the UPLC-QTOF machine were converted into Network Common Data Form (NetCDF) format using the MassLynx software (Waters Corp, Milford, MA). We used XCMS to preprocess the LC-MS data. The XCMS tool uses several algorithms to preprocess the data. 24 The first step is to filter and detect the peaks. In this step, the ion chromatograms are extracted and a model peak matched filter is applied. The peak detection algorithm is based on cutting the LC-MS data into slices, a fraction of a mass unit wide, and then operating on those individual slices in the chromatographic time domain. After detecting peaks in individual samples, the peaks are matched across samples to allow calculation of retention time deviations and relative ion intensity comparison. This is accomplished using a group method, which creates bins to group peaks in the mass domain. The peak matching algorithm in XCMS takes into account the two-dimensional anisotropic nature of LC-MS data. These groups are then used to identify and correct drifts in retention time from run to run. For every group, the median retention C

4 technical notes Varghese et al. Figure 2. S-plot of a pairwise comparison between 0 and 24 h for ATCL8 cell line. We selected the peaks marked with 0. time and the deviation from median for every sample in that group is calculated. To account for the missing peak data, a mathematical function is used to approximate the difference between deviation and interpolate in sections where no peak groups are present. Missing peaks can occur because they could be missed during the peak detection step or because an analyte was not present in a sample. This creates a data matrix of ion abundance for each pair of m/z and retention time. The data matrix obtained from XCMS was normalized using the quantile normalization. Quantile normalization has been applied for normalization of microarray data. 42 The method assumes an underlying common distribution of intensities exists across samples. Peak Selection. Shrinkage t Statistic Analysis. We used the shrinkage t statistic and correlation adjusted t-score to identify peaks whose intensities are significantly different in a pairwise comparison. Shrinkage t statistic is based on a novel and model-free shrinkage estimate of the variance vector across peaks. It is a further development of the quasi-likelihood approach of Strimmer 43 but with additional regularization. It is developed in the framework of James-Stein-type analytic shrinkage. 44 This approach offers a highly efficient means for regularized inference, both in the statistical and computational sense. It is complementary to more well-known alternatives such as Bayesian and penalized likelihood inference. 45 It is similar to the standard t statistic with the replacement of the sample variances by corresponding shrinkage estimates. These are derived in a distribution-free fashion and with little a priori assumptions. The cat ( correlation-adjusted t ) score is the product of the square root of the inverse correlation matrix with a vector of t-scores. The cat score describes the contribution of each individual feature in separating the two groups, after removing the effect of all other features. The cat score is a natural Table 1. Number of Statistically Significant Peaks cell line 0 vs 30 min 0 vs 1 h 0 vs 2 h 0 vs 3 h 0 vs 6 h 0 vs 24 h Number of peaks selected by the shrinkage t statistic analysis ATCL AT5BIVA Number of peaks selected by PLS-DA ATCL AT5BIVA Number of peaks selected by both shrinkage t statistic and PLS-DA methods ATCL AT5BIVA Number of peaks used for pathway analysis by IPA a ATCL AT5BIVA a The peaks were selected by shrinkage t statistic and/or PLS-DA. Metabolites were identified by METLIN, MMCD, and HMDB. criterion to rank features in the presence of correlation. 46 If there is no correlation, the cat score reduces to the usual t-score. We used the st package in R which implements the shrinkage t statistic and a shrinkage estimate of the cat score. For determining p-values and false discovery rate (FDR), a twocomponent mixture model is fitted to the observed cat scores. 47 An example for ranking of metabolomics markers of prostate cancer using shrinkage t and cat score is shown in Zuber and Strimmer. 46 A comparison of methods for assigning significance to cat scores is also discussed. For computing FDR, we used the fdrtool software package in R. Partial Least-Squares-Discriminant Analysis (PLS-DA). We used the SIMCA-P+ software to generate PLS-DA two-component models. PLS-DA is a supervised method that uses multiple linear regression technique to find the direction of maximum D

5 Metabolic Changes in Response to Radiation technical notes Figure 3. Heatmap of ion intensities of peaks selected by Shrinkage t statistic and/or PLS-DA in ATCL8 cells. Figure 4. Heatmap of ion intensities of peaks selected by Shrinkage t statistic and/or PLS-DA in AT5BIVA cells. covariance between a data set and the class labels. It sharpens the separation between groups of observations by rotating PCA components such that a maximum separation among classes is obtained, and to understand which variables carry the class separating information. PLS components are built by trying to find a proper compromise between two purposes: describing the set of explanatory variables and predicting the response ones. Mass-Based Metabolite Identification. We used the METLIN, Madison Metabolomics Consortium Database (MMCD), and the Human Metabolite DataBase (HMDB) for mass-based metabolite E

6 technical notes Figure 5. Canonical pathways associated to the metabolites that showed significant changes in expression in response to radiation exposure of ATCL8 cells. Figure 6. Canonical pathways associated to the metabolites that showed significant changes in expression in response to radiation exposure of AT5BIVA cells. identification. METLIN is a web-based database that has previously been developed by the Scripps Research Institute to facilitate the identification of metabolites using accurate mass data. It includes an annotated list of structural information for known metabolites. MMCD is a database maintained by the National Magnetic Resonance Facility at Madison, WI. HMDB focuses on metabolites found in human body fluid. These databases have links to other public databases such as the Kyoto Encyclopedia of Genes and Genomes (KEGG) and PubChem. Pathway Analysis. Following identification of metabolites by METLIN, MMCD, and HMDB, we introduced the KEGG, HMDB, or PubChem IDs of the identified metabolites into the IPA. IPA-Metabolomics is a capability within IPA that extracts rich pathway information from metabolomics data. For a given data set, IPA automatically generates the pathways that are related to those metabolites. Results and Discussion LC-MS metabolic profiles from the two isogenic cell lines, AT5BIVA and ATCL8 were analyzed before and after radiation at 30 min, 1 h, 2hrs, 3hrs, 6hrs and 24hrs. Each time point consisted of five samples. Thus, the data matrix consisted of 35 samples including five controls that were not exposed to radiation. It uses two criteria (ratio and significance) to evaluate the potential association of the biomarkers to these pathways. The ratio is calculated by dividing the number of candidate biomarkers in a given pathway by the total number of biomolecules that make up the pathway. The significance of a given pathway (p-value) is a measure of the likelihood that the pathway is associated with the list of candidate biomarkers by chance. The null hypothesis is that there is no association. A very small p-value may be an indication that the pathway is more likely to explain the observed phenotype. While the ratio Varghese et al. gives a good idea of percentage of candidate biomarkers in a pathway, the significance addresses the question if the observed association is merely by chance. The raw LC-MS data were converted into NetCDF format using MassLynx. Filtering and peak detection were performed initially using the XCMS software package. To enable further analysis and visualization of data, all m/z values were binned to fixed m/z values with a bin size of 100 ppm. As a result, the data were transformed into a two-dimensional matrix of ion abundances. Rows of the matrix represent the ions with specific RT and m/z values and columns represent the samples. A list of all peaks in each sample was compiled. After identifying peaks in individual samples, the peaks were aligned across samples to allow calculation of retention time deviations and relative ion intensity comparison. The peak-matching algorithm in XCMS that takes into account the two-dimensional, anisotropic nature of LC-MS data was used. After preprocessing, 2729 peaks for AT5BIVA and 2719 peaks for ATCL8 were detected. A quantile normalization method was then used to remove systematic bias. We performed statistical data analysis to identify differentially expressed metabolites in pairwise comparisons of 0 vs 30 min, 0 vs 1 h, 0 vs 2 h, 0 vs 3 h, 0 vs 6 h, and 0 vs 24 h for each of the two cell lines. Specifically, we used the shrinkage t statistic and cat score to select the differentially expressed metabolites with p-value less than 0.05 and FDR less than 10% for each pairwise comparison. To select individual metabolites according to their relative effect on group separation, we ranked the metabolites according to the magnitude of their respective shrinkage t statistic. For determining p-values and FDR values, a twocomponent mixture model was fitted to the observed cat scores. 47 The output list table is then ordered by p-value with the most significant metabolites listed first. In addition, we used the SIMCA-P+ software to fit a PLS- DA model for each pairwise comparison. The software generates an S-plot, which is a scatter plot that can be utilized to explain the variable influence on the model. By combining covariance (x-axis, p 1 ) and correlation loading profiles (y-axis, p(corr) 1 ), the s-plot can be utilized for the extraction of putative metabolites (Figure 2). The upper right quadrant of the S-plot shows those components which are elevated in the control group, while the lower left quadrant shows components elevated in the treated group. The farther along the x-axis the greater the contribution to the variance between the groups, while, the farther the y-axis the higher the reliability of the analytical result. Figure 2 presents the S-plot derived from a pairwise comparison (0 vs 24 h) for ATCL8 cell line by PLS- DA. For this comparison, we chose the peaks with p 1 > and p(corr) 1 > 0.6. The figure shows the selected peaks. Table 1 summarizes the number of peaks that were selected using the shrinkage t statistic and PLS-DA. The table presents the peaks selected by these methods separately and the number of overlapping peaks selected by both methods. A mass-based database search was performed using METLIN, MMCD, and HMDB databases for identification of the metabolites represented by the peaks selected through shrinkage t statistic and/or PLS-DA methods. The table presents the number of peaks for which the corresponding metabolites were identified at each time point. Overall, the database search led to identification of about 25% of the peaks. Specifically, 130 differentially abundant peaks were identified in ACTL8 cells across all time points combined, while 140 peaks were identified in AT5BIVA cells. These metabolites were used for the pathway analysis F

7 Metabolic Changes in Response to Radiation technical notes Figure 7. Determination of the chemical structure of an upregulated metabolite in ATCL8 by tandem mass spectrometry. Top panel shows the positive ion MS/MS fragmentation of the m/z from ATCL8 cell extracts. The bottom panel shows the positive ion mode fragmentation spectra for synthetic guanine. using IPA. Figures 3 and 4 depict heatmaps of the ion intensities of these metabolites for ATCL8 and AT5BIVA cells, respectively, with the metabolites annotated by their most enriched canonical pathways. Figures 5 and 6 present the top 12 canonical pathways enriched in the metabolites identified for ATCL8 and AT5BIVA cells, respectively. The figures show remarkable differences in metabolic responses to radiation treatment under ATM proficient (ATCL8) or deficient (AT5BIVA) conditions. While ATCL8 shows significant enrichment of metabolites involved in purine metabolism, linoleic acid metabolism, pentose and glucurronate interconversions, fructose and mannose metabolism, etc., AT5BIVA shows a predominance of glycerophospholipd metabolism and phospholipid degradation. A number of metabolites involved in purine metabolism were affected due to radiation. This is consistent with our previous proteomics analysis where we found enzymes involved with purine metabolism to be differentially changed in the two cell lines. 48 We observed that ATCL8 showed more enrichment of metabolites involved in purine metabolism than AT5BIVA. Taken together, these results show that the presence of ATM in the ATCL8 cells elicits a normal radiation response leading to inhibition of cell growth and proliferation and increased DNA repair. Future mechanistic studies will be needed in order to correlate the role of these pathways with respect to ATM functionality. We verified some of the identified metabolites using MS/ MS analysis. For example, we verified by MS/MS analysis the identity of guanine (involved in purine metabolism), which was up regulated nearly at all time points in ATCL8. Figure 7 depicts the positive ion MS/MS fragmentation of the m/z from ATCL8 cell extracts (top panel) and the positive ion mode fragmentation spectra for synthetic guanine (bottom panel). Conclusion and Future Work The study of metabolism at the global or -omics level, referred to as metabolomics, is a new but rapidly growing field that has the potential to impact our understanding of molecular mechanisms of disease. In our study we used two isogenic cell lines (AT5BIVA and ATCL8) as a model system for characterizing the metabolic changes that occur in response to radiation exposure. The goal was to understand the underlying differences in metabolic changes because of absence of ATM (AT5BIVA) or presence of ATM (ATCL8). Shrinkage t statistic and partial least-squares-discriminant analysis methods are performed to select peaks with significant change in signal intensities between radiation-treated and untreated cell lines. The metabolites represented by the selected peaks are identified through mass-based database search. A pathway analysis is done to identify the pathways these metabolites are involved in. A limitation of this analysis is that only a quarter of the peaks selected were involved in the pathway analysis due to lack of metabolite identification by mass spectrometry. The data analysis algorithms used in this study have successfully revealed interesting molecular differences in the response to radiation exposure in the presence and absence of ATM. Future studies will entail further validation of our findings with respect to the identification of the metabolites and their actual cellular levels. Identifying metabolites is a difficult task. Efficient identification tools are critical because of the large number of metabolites in any given metabolomics study. Extensive library of experimental MS/MS data is needed for a comprehensive characterization and identification of metabolites. Acknowledgment. This work was supported in part by the National Science Foundation Grant (IIS ) and National Institute of Health PO1 CA We acknowledge the contribution of Alfredo Velena for culturing the cell lines. References (1) Nicholson, J. K.; Wilson, I. D. Opinion: understanding global systems biology: metabonomics and the continuum of metabolism. Nat. Rev. Drug Discovery 2003, 2 (8), (2) Griffin, J. L. The Cinderella story of metabolic profiling: does metabolomics get to go to the functional genomics ball. Philos. Trans. R. Soc. Lond., B: Biol. Sci. 2006, 361 (1465), (3) Moolenaar, S. H.; Engelke, U. F.; Wevers, R. A. Proton nuclear magnetic resonance spectroscopy of body fluids in the field of inborn errors of metabolism. Ann. Clin. Biochem. 2003, 40 (Pt 1), G

8 technical notes (4) Kaddurah-Daouk, R.; Kristal, B. S.; Weinshilboum, R. M. Metabolomics: a global biochemical approach to drug response and disease. Annu. Rev. Pharmacol. Toxicol. 2008, 48, (5) Bino, R. J.; Hall, R. D.; Fiehn, O.; Kopka, J.; Saito, K.; Draper, J.; Nikolau, B. J.; Mendes, P.; Roessner-Tunali, U.; Beale, M. H.; Trethewey, R. N.; Lange, B. M.; Wurtele, E. S.; Sumner, L. W. Potential of metabolomics as a functional genomics tool. Trends Plant Sci. 2004, 9 (9), (6) Shulaev, V. Metabolomics technology and bioinformatics. Brief. Bioinform. 2006, 7 (2), (7) Dettmer, K.; Aronov, P. A.; Hammock, B. D. Mass spectrometrybased metabolomics. Mass Spectrom. Rev. 2007, 26 (1), (8) Yates, J. R.; Ruse, C. I.; Nakorchevsky, A. Proteomics by mass spectrometry: approaches, advances, and applications. Annu. Rev. Biomed. Eng. 2009, 11, (9) Lu, W.; Bennett, B. D.; Rabinowitz, J. D. Analytical strategies for LC-MS-based targeted metabolomics. J. Chromatogr., B: Analyt. Technol. Biomed. Life Sci. 2008, 871 (2), (10) Dunn, W. B. Current trends and future requirements for the mass spectrometric investigation of microbial, mammalian and plant metabolomes. Phys. Biol. 2008, 5 (1), (11) Wishart, D. S. Current progress in computational metabolomics. Brief. Bioinform. 2007, 8 (5), (12) Katajamaa, M.; Oresic, M. Data processing for mass spectrometrybased metabolomics. J. Chromatogr., A 2007, 1158 (1-2), (13) Du, P.; Kibbe, W. A.; Lin, S. M. Improved peak detection in mass spectrum by incorporating continuous wavelet transform-based pattern matching. Bioinformatics 2006, 22 (17), (14) Yang, C.; He, Z.; Yu, W. Comparison of public peak detection algorithms for MALDI mass spectrometry data analysis. BMC Bioinformatics 2009, 10 (1), 4. (15) Mathias, V.; Li-Thiao-Té, S.; Hans-Michael, K.; Runxuan, Z.; Tero, A.; Benno, S. Alignment of LC-MS images, with applications to biomarker discovery and protein identification. Proteomics 2008, 8 (4), (16) Kong, X.; Reilly, C. A Bayesian approach to the alignment of mass spectra. Bioinformatics 2009, 25 (24), (17) Befekadu, G. K.; Tadesse, M. G.; Ressom, H. W. In A bayesian based functional mixed-effects model for analysis of LC-MS data, Engineering in Medicine and Biology Society; Annual International Conference of the IEEE; IEEE Press Books: New York, 2009; pp (18) Hoffmann, N.; Stoye, J. ChromA: signal-based retention time alignment for chromatography-mass spectrometry data. Bioinformatics 2009, 25 (16), (19) Lange, E.; Tautenhahn, R.; Neumann, S.; Gropl, C. Critical assessment of alignment procedures for LC-MS proteomics and metabolomics measurements. BMC Bioinform. 2008, 9, 375. (20) Redestig, H.; Fukushima, A.; Stenlund, H.; Moritz, T.; Arita, M.; Saito, K.; Kusano, M. Compensation for Systematic Cross- Contribution Improves Normalization of Mass Spectrometry Based Metabolomics Data. Anal. Chem. 2009, 81 (19), (21) Sysi-Aho, M.; Katajamaa, M.; Yetukuri, L.; Oresic, M. Normalization method for metabolomics data using optimal selection of multiple internal standards. BMC Bioinform. 2007, 8 (1), 93. (22) Warrack, B. M.; Hnatyshyn, S.; Ott, K.-H.; Reily, M. D.; Sanders, M.; Zhang, H.; Drexler, D. M. Normalization strategies for metabonomic analysis of urine samples. J. Chromatogr., B 2009, 877 (5-6), (23) Tikunov, Y.; Lommen, A.; de Vos, C. H.; Verhoeven, H. A.; Bino, R. J.; Hall, R. D.; Bovy, A. G. A novel approach for nontargeted data analysis for metabolomics. Large-scale profiling of tomato fruit volatiles. Plant Physiol. 2005, 139 (3), (24) Smith, C. A.; Want, E. J.; O Maille, G.; Abagyan, R.; Siuzdak, G. XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal. Chem. 2006, 78 (3), (25) Katajamaa, M.; Oresic, M. Processing methods for differential analysis of LC/MS profile data. BMC Bioinform. 2005, 6, 179. (26) Kind, T.; Tolstikov, V.; Fiehn, O.; Weiss, R. H. A comprehensive urinary metabolomic approach for identifying kidney cancerr. Anal. Biochem. 2007, 363 (2), (27) Xia, J.; Psychogios, N.; Young, N.; Wishart, D. S. MetaboAnalyst: a web server for metabolomic data analysis and interpretation. Nucleic Acids Res. 2009, 37 (Suppl 2), W (28) Eichler, G. S.; Huang, S.; Ingber, D. E. Gene Expression Dynamics Inspector (GEDI): for integrative analysis of expression profiles. Bioinformatics 2003, 19 (17), (29) Patterson, A. D.; Li, H.; Eichler, G. S.; Krausz, K. W.; Weinstein, J. N.; Fornace, A. J.; Gonzalez, F. J.; Idle, J. R. UPLC-ESI-TOFMS- Based Metabolomics and Gene Expression Dynamics Inspector Self-Organizing Metabolomic Maps as Tools for Understanding the Cellular Response to Ionizing Radiation. Anal. Chem. 2008, 80 (3), (30) Neuweger, H.; Albaum, S. P.; Dondrup, M.; Persicke, M.; Watt, T.; Niehaus, K.; Stoye, J.; Goesmann, A. MeltDB: a software platform for the analysis and integration of metabolomics experiment data. Bioinformatics 2008, 24 (23), (31) Ramautar, R.; van der Plas, A. A.; Nevedomskaya, E.; Derks, R. J. E.; Somsen, G. W.; de Jong, G. J.; van Hilten, J. J.; Deelder, A. M.; Mayboroda, O. A. Explorative Analysis of Urine by Capillary Electrophoresis-Mass Spectrometry in Chronic Patients with Complex Regional Pain Syndrome. J. Proteome Res. 2009, 8 (12), (32) Bao, Y.; Zhao, T.; Wang, X.; Qiu, Y.; Su, M.; Jia, W.; Jia, W. Metabonomic Variations in the Drug-Treated Type 2 Diabetes Mellitus Patients and Healthy Volunteers. J. Proteome Res. 2009, 8 (4), (33) Yin, P.; Zhao, X.; Li, Q.; Wang, J.; Li, J.; Xu, G. Metabonomics Study of Intestinal Fistulas Based on Ultraperformance Liquid Chromatography Coupled with Q-TOF Mass Spectrometry (UPLC/Q-TOF MS). J. Proteome Res. 2006, 5 (9), (34) Kim, K.; Aronov, P.; Zakharkin, S. O.; Anderson, D.; Perroud, B.; Thompson, I. M.; Weiss, R. H. Urine Metabolomics Analysis for Kidney Cancer Detection and Biomarker Discovery. Mol Cell Proteomics 2009, 8 (3), (35) 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. HMDB: the Human Metabolome Database. Nucleic Acids Res. 2007, 35 (Database issue), D (36) Cui, Q.; Lewis, I. A.; Hegeman, A. D.; Anderson, M. E.; Li, J.; Schulte, C. F.; Westler, W. M.; Eghbalnia, H. R.; Sussman, M. R.; Markley, J. L. Metabolite identification via the Madison Metabolomics Consortium Database. Nat. Biotechnol. 2008, 26 (2), (37) Smith, C. A.; Maille, G. O.; Want, E. J.; Qin, C.; Trauger, S. A.; Brandon, T. R.; Custodio, D. E.; Abagyan, R.; Siuzdak, G. METLIN: A Metabolite Mass Spectral Database. Ther. Drug Monit. 2005, 27 (6), (38) Kind, T.; Fiehn, O. Metabolomics database annotations via query of elemental compositions: Mass accuracy is insufficient even at less than 1 ppm. BMC Bioinform. 2006, 7, 234. (39) Lavin, M. F.; Shiloh, Y. The genetic defect in ataxia-telangiectasia. Annu. Rev. Immunol. 1997, 15, (40) Jung, M.; Zhang, Y.; Lee, S.; Dritschilo, A. Correction of radiation sensitivity in ataxia telangiectasia cells by a truncated I kappa B-alpha. Science 1995, 268 (5217), (41) Sana, T. R.; Waddell, K.; Fischer, S. M. A sample extraction and chromatographic strategy for increasing LC/MS detection coverage of the erythrocyte metabolome. J. Chromatogr., B: Analyt. Technol. Biomed. Life Sci. 2008, 871 (2), (42) Bolstad, B. M.; Irizarry, R. A.; Astrand, M.; Speed, T. P. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 2003, 19 (2), (43) Strimmer, K. Modeling gene expression measurement error: a quasi-likelihood approach. BMC Bioinform. 2003, 4, 10. (44) Schafer, J.; Strimmer, K. A shrinkage approach to large-scale covariance matrix estimation and implications for functional genomics. Stat. Appl. Genet. Mol. Biol. 2005, 4, Article32. (45) Opgen-Rhein, R.; Strimmer, K. Accurate ranking of differentially expressed genes by a distribution-free shrinkage approach. Stat Appl. Genet. Mol. Biol. 2007, 6, Article9. (46) Zuber, V.; Strimmer, K. Gene ranking and biomarker discovery under correlation. Bioinformatics 2009, 25 (20), (47) Efron, B. Microarrays, empirical Bayes, and the two-groups model. Statist. Sci. 2008, 23, (48) Hu, Z. Z.; Huang, H.; Cheema, A.; Jung, M.; Dritschilo, A.; Wu, C. H. Integrated Bioinformatics for Radiation-Induced Pathway Analysis from Proteomics and Microarray Data. J. Proteomics Bioinform. 2008, 1 (2), PR100185B Varghese et al. H