Effects of celastrol on human cervical cancer cells as revealed by ion-trap gas chromatography–mass spectrometry based metabolic profiling

Biochimica et Biophysica Acta 1830 (2013) 2779–2789

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbagen

Effects of celastrol on human cervical cancer cells as revealed by ion-trap gas chromatography–mass spectrometry based metabolic profiling Yongsheng Hu a, b, c, 1, Yunpeng Qi a, b, 1, Hua Liu a, b, Guorong Fan a, b,⁎, Yifeng Chai a, b,⁎⁎ a b c

Department of Pharmaceutical Analysis, School of Pharmacy, Second Military Medical University, Shanghai 200433, China Shanghai Key Laboratory for Pharmaceutical Metabolite Research, Second Military Medical University, Shanghai 200433, China Department of Pharmacy, the 118th Hospital of PLA, Wenzhou 325000, China

a r t i c l e

i n f o

Article history: Received 19 April 2012 Received in revised form 4 October 2012 Accepted 29 October 2012 Available online 8 November 2012 Keywords: Celastrol Human cervical cancer cell Apoptosis Gas chromatography–mass spectrometry Metabolic profiling Metabolomics

a b s t r a c t Background: Celastrol, a quinine methide triterpene extracted from a Chinese medicine (Trypterygium wilfordii Hook F.), has the potential to become an anticancer drug with promising prospects. Cell culture metabolomics has been a powerful method to study metabolic profiles in cell line after drug treatment, which can be used for discovery of drug targets and investigation of drug effects. Methods: We analyzed the metabolic modifications induced by celastrol treatment in human cervical cancer cells, using an ion-trap gas chromatography–mass spectrometry based metabolomics combined with multivariate statistical analysis, which allows simultaneous screening of multiple characteristic metabolic pathways related to celastrol treatment. Three representative apoptosis-inducing cytotoxic agents, namely cisplatin, doxorubicin hydrochloride and paclitaxel, were selected as positive control drugs to validate reasonableness and accuracy of our metabolomic investigation on celastrol. Results: Anti-proliferation and apoptotic effects of celastrol were demonstrated by CCK-8 assay, Annexin-V/PI staining method, mitochondrial membrane potential (ΔΨm) assay and caspase-3 assay. Several significant metabolites involved in energy, amino acid and nucleic acid metabolism in HeLa cells induced by celastrol and positive drugs were reported. Our method is proved to be effective and robust to provide new evidence of pharmacological mechanism of celastrol. Conclusions: The metabolic alterations induced by drug treatment showed the impaired physiological activity of HeLa cells, which also indicated anti-proliferative and apoptotic effects of celastrol and these positive drugs. General significance: GC/MS-based metabolomic approach applied to cell culture could give valuable information on the systemic effects of celastrol in vitro and help us to further study its anticancer mechanism. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Traditional medicine represents a cornucopia of plant-derived remedies to discover novel lead molecules for the development of new drugs. Celastrol (structure see Fig. 1A), a quinine methide triterpene extracted from a typical Chinese medicine (Trypterygium wilfordii Hook F.), has been extensively investigated as a promising drug for the treatment of autoimmune diseases, asthma, chronic inflammation, and neurodegenerative disease [1,2]. In 2006, celastrol is reported for the first time to be a natural proteasome inhibitor and

⁎ Correspondence to: G. Fan, Department of Pharmaceutical Analysis, School of Pharmacy, Second Military Medical University, Shanghai 200433, China. Tel./fax: +86 21 81871260. ⁎⁎ Correspondence to: Y. Chai, Department of Pharmaceutical Analysis, School of Pharmacy, Second Military Medical University, Shanghai 200433, China. Tel./fax: +86 21 81871201. E-mail addresses: [email protected] (Y. Hu), [email protected] (Y. Qi), [email protected] (H. Liu), [email protected] (G. Fan), [email protected] (Y. Chai). 1 These authors contributed equally to this work. 0304-4165/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbagen.2012.10.024

has exhibited a great potential for cancer prevention and treatment [3]. From then on, investigation on therapeutic efficacy of celastrol against various cancer cells has become a hot spot [4–9]. Celastrol can inhibit the proliferation of wide variety of human tumor cells, and prevent their malignant tissue invasion and block angiogenesis [3,4,10,11]. When used in combination therapy, it can also sensitize resistant melanoma cell to temozolomide treatment, and potentiate radiotherapy in prostate cancer cells [12,13]. These studies show that celastrol has the potential to become an anticancer drug with promising prospects. Several molecular targets of celastrol have been characterized, including heat shock protein (HSP), reactive oxygen species (ROS), vascular endothelial growth receptor (VEGFR), nuclear factor-κB (NF-κB) and so on [14,15]. Interestingly, many of them are centered on the function of IκB kinase enzyme (IKK) complex and NF-κB system [1,11], which is the key regulator in cancer disease [16,17]. However, the NF-κB system is highly integrated with other signaling pathways via a variety of protein kinases [18,19], which makes it difficult to explain the mechanism of celastrol's therapeutic effects. Hence, although

2780

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

Fig. 1. Chemical structures of celastrol (A), cisplatin (B), doxorubicin hydrochloride (C) and paclitaxel (D).

investigations focusing on celastrol's effects on specific cellular pathways have revealed a number of targets in a diverse array of in vitro models [14,15], there is still lack of a thorough insight into its anticancer effects from a global view. Metabolomics seeks to characterize the metabolic profile of a biological system. Since the panel of metabolites with relatively low molecular weight are downstream products of biomolecular processes, their identity and concentration in living biological system can provide biochemical signatures for globally tracking the physiological effects, and exploring the drug effects [20–22]. Particularly, because cancer cells have several specific metabolic features, such as high enzyme activities, high phosphometabolite levels and high energy metabolism (for an overview see http://www.metabolic-database.com/ html/tumor_metabolome_overview.html), cell culture metabolomics has been instrumental in finding further susceptible biomarkers for cancer diagnosis or drug treatment [23–26]. Apoptosis is an important phenomenon in cancer therapy and represents a common mechanism of drug effect [27]. Therefore, investigation on biomarkers indicative of early apoptosis is crucial in theranostics of cancer therapy [28]. Importantly, celastrol has been reported to induce apoptosis in many cancer cells. However, the metabolic intervention of celastrol on cancer cells has not been revealed, whereas this is clearly very meaningful for exploring its mechanism of action in preventing and treating cancer. These considerations prompted us to study the metabolic modifications induced by celastrol treatment in cancer cells. Comparing to LC-MS and NMR, GC–MS remains a good choice for metabolomic study, since it has been proved of high selectivity and reproducibility with relatively low cost, and a number of structure databases are available [29]. The present study aimed to design a fast, robust and reliable GC–MS analysis system for metabolite measurements in cancer cells, for the purpose of providing a global view of celastrol's effects. We report for the first time several metabolites indicative for early apoptotic processes in HeLa cells culture induced by celastrol using ion-trap gas chromatography–mass spectrometry. Meanwhile, in order to validate reasonableness and accuracy of our metabolomic investigation on celastrol, we selected three representative apoptosis-inducing cytotoxic agents, namely cisplatin, doxorubicin hydrochloride and paclitaxel (structures see Fig. 1B to D) as positive control drugs. In our study, cell fate and apoptosis were determined

by CCK-8 assay, Annexin-V/PI staining method, mitochondrial membrane potential assay and caspase-3 assay. 2. Materials and methods 2.1. Reagents Dulbecco's modified Eagle's medium (DMEM) was purchased from HyClone Thermo scientific (Beijing, China). Fetal bovine serum (FBS) was obtained from GIBCO BRL (Grand Island, NY, US). Celastrol, cisplatin, doxorubicin hydrochloride and paclitaxel were all purchased from Sigma-Aldrich (St. Louis, MO, US). Celastrol and positive control drugs were prepared from stock solutions in dimethyl sulfoxide (DMSO). The stock solutions were kept frozen in aliquot at −20 °C and thawed immediately prior to each experiment. Methoxylamine hydrochloride, N-methyl-N-(trimethylsilyl)-trifluoracetamide (MSTFA), pyridine, trimethyl-chlorosilane (TMCS) and ribitol (used as internal standard) were purchased from Sigma-Aldrich (St Louis, MO, US). 2.2. Cell culture Human cervical cancer HeLa cells line was purchased from Chinese Academy of Sciences (Shanghai, China). Cells were routinely cultured in DMEM, supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/mL), streptomycin (0.1 mg/mL) and were maintained in a humidified atmosphere of 5% CO2 at 37 °C. Cells were passaged every 3–4 days. For drug treatment, appropriate amounts of celastrol and positive control drugs were added to culture medium to achieve the appropriate concentrations and then incubated for the indicated time periods. 2.3. CCK-8 assay to determine cell viability To evaluate the percentage of viable cells after different treatments, the Cell Counting Kit-8 (CCK-8) assay (Beyotime Biotech, China) was performed. Cells were seeded in 96-well plates at a density of 2000 cells/well. On the next day, cells were incubated with different concentrations of celastrol. After appropriate incubation time, 10 μL CCK-8 was added to each well. After another 1 h of incubation at 37 °C, absorbance was measured at 480 nm (A480) with the Synergy™ 4 Mulit-Detection

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

Microplate Reader (BioTek, US). The percentage of viable cells was calculated by the formula: the percentage of viable cells (%) = 100 − [(A480, control − A480, drug) / A480, control] × 100. 2.4. Apoptosis analysis by flow cytometry Cell apoptosis was examined by the Annexin-V/PI method. HeLa cells were seeded at a density of 1 × 105 cells/well into a 6-well plate. After being treated with celastrol (2.5, 5, and 10 μM) and incubated for 24 h, the HeLa cells were collected and apoptosis was examined by using an Annexin V-FITC Apoptosis Detection Kit (KeyGEN, Nanjing, China), which detects phosphatidylserine exposed on the outer surface of the cell membrane. The cells were harvested with trypsin (without EDTA) and washed in phosphate-buffered saline (PBS, pH 7.4). After centrifugation at 2000 rpm, the supernatant was removed and then the cells were suspended in stain containing Annexin V-FITC and propidium iodide (PI). This was mixed well and incubated at room temperature for 15 min in the dark. The cells were analyzed by MACSQuant ™ flow cytometry (Miltenyi Biotec, Germany) within 1 h of staining. Apoptotic cells were defined as annexin V-FITC-positive cells.

2.5. Mitochondria membrane potential (ΔΨm) assay The ΔΨm was determined by using the mitochondrial membrane potential assay kit with JC-1 (Beyotime Biotech, China). JC-1 is capable of selectively entering mitochondria, where it forms monomers and emits green fluorescence when ΔΨm is relatively low. At a high ΔΨm, JC-1 aggregates and gives red fluorescence [30]. Briefly, HeLa cells were seeded in 6-well plates. After celastrol treatment, cells were collected, and then suspended in 1 mL staining dye (culture medium: JC-1 working dye= 1:1) and incubated at 37 °C for 20 min, 5% CO2. After this, cells were washed twice with cold JC-1 staining buffer, and examined with the FACSCalibur flow cytometry (BD, New York, USA). The depolarization of ΔΨm was represented by the percent of R2 (percent of JC-1 monomer).

2.6. Caspase-3 assay to detect apoptosis For detection of apoptosis cells were grown in culture dish (10 cm). The enzymatic activity of caspase-3 was determined using the Caspase-3 Colorimetric Assay Kit (KeyGEN Biotech, China). After treatment of 5 or 10 μM celastrol for 6, 12 and 24 h, the cells were collected to measure the caspase-3 activity according to manufacturer's instructions. Cell extracts were incubated with 5 μL caspase-3 substrate at 37 °C for 4 h. The reaction was measured at 405 nm in the Synergy™ 4 Mulit-Detection Microplate Reader. Celastrol treated samples were normalized to the caspase activity of the untreated sample, which was set to 1.0. Fold of increases in caspase activities were presented.

2.7. Cell quenching For metabolite measurements, HeLa cells were cultured in culture dish (10 cm) to approximately 70% confluence, and then incubated with celastrol and positive control drugs for 12 h. After the culture medium was removed from the culture dish, cells were rapidly washed twice with 37 °C PBS. The residual PBS was removed by vacuum. Cells were then quenched using 1.5 mL −80 °C HPLC grade methanol (Sigma-Aldrich, St. Louis, MO). Next, cells were quickly detached from the culture dish using a cell lifter (Fisher Scientific, Suwanee, GA). The methanol solution containing the quenched cells was pipetted into a 2 mL centrifuge tube and frozen in liquid nitrogen until extraction. In addition, five parallel dishes of cells were trypsinized and counted, subsequent metabolite measurements were normalized to cell count.

2781

2.8. Preparation of intracellular extracts for GC–MS analysis The tube containing the quenched cells were thawed in an ice bath for 10 min, spiked with the internal standard (5 μL of 1 mg/mL of ribitol), vortexed vigorously, and incubated on ice for 10 min. After that, the tube was frozen in liquid nitrogen for 10 min, thawed in ice bath for 10 min, and briefly vortexed. The tube was centrifuged at 4 °C and 12,000 rpm for 5 min and the supernatant was transferred to a new tube. The cell pellet was re-extracted twice with 500 μL cold 80% methanol (20% water). The combined extracts was transferred into a GC vial and evaporated to dryness under N2 stream at room temperature. The derivatization was performed using methoxyamine pyridine (75 μL; 15 mg/mL) at 60 °C for 1 h, followed by MSTFA (75 μL) with 1% TMCS as catalyst at 60 °C for 1 h. 2.9. GC–MS analysis The derivatized samples for GC–MS were analyzed on a Thermo Scientific ITQ 1100™ GC/MS n (ThermoFisher Electron Corporation, USA). A 1.0 μL of sample solution was injected with splitless mode to TR-5MS column, 30 m × 0.25 mm ID × 0.25 μm film thickness (ThermoFisher Electron Corporation, USA), with helium as the carrier gas at a flow of 0.6 mL/min. The initial oven temperature was set at 50 °C, ramped to 300 °C by 8 °C/min, and held for 10 min. The temperatures of injector, ion source and transfer line were all set at 280 °C. The electron energy was 70 eV. The mass spectrometer was operated in full scan mode from 35 to 600 m/z with a scan time of 0.5 s. The solvent delay was set at 5 min. Identification of the metabolites in the GC–MS spectra was performed by searching the NIST (National Institute of Standards and Technology) database installed in the ITQ 1100™ GC–MS n system. To ensure the stability of the GC–MS system, an equal volume of each sample was pooled together to generate a pooled quality control (QC) sample [31,32]. This QC sample was processed in the same way as the samples and then analyzed randomly through the analytical batch. 2.10. Data processing and multivariate statistical analysis The acquired GC–MS data was first converted into CDF format. XCMSOnline (https://xcmsonline.scripps.edu/) was used for nonlinear alignment of the data in the time domain and automatic integration and extraction of the peak intensities, using default GC/Single Quad parameters. The XCMS output data containing 4378 ion peaks was preprocessed using the Microsoft Excel software (Microsoft, Redmond, WA), where the IS peaks, and impurity peaks from column bleeds and derivatization procedures were excluded, and the variables presenting in at least 80% of either group were extracted. Then, the remaining ion features were normalized to the internal standard (m/z 217.1, the most abundance fragment ion for the silylation derivative of ribitol). Next, the most abundant fragment ion with the same retention time (the time bin is 0.01 min) was remained and the other ions were excluded [33]. The processed data matrix with intensities of 353 ion peaks was further subjected to statistical analysis. Pattern recognition methods including PLS-DA (partial least squares-discriminate analysis) and PCA (Principal Component Analysis) (using SIMCA-P, version 11, Umetrics) and Heatmap (using MetATT, http://metatt.metabolomics.ca/MetATT/) were established to investigate the intracellular metabolic profiles of the negative control, celastrol and positive control treated groups. The data were mean-centered and unit variance (UV)-scaled before multivariate statistical analysis. The discriminating metabolites were obtained using a statistically significant threshold of variable influence on projection (VIP) values obtained from the PLS-DA model and two-tailed Student's t test (P value) on the normalized raw data at univariate analysis level, where the metabolites with VIP values larger than

2782

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

3. Results and discussion 3.1. Effects of celastrol treatment on cell viability, apoptosis, mitochondria, ΔΨm and caspase-3 expression

Fig. 2. HeLa cells were seeded in 96-well plate and treated with 2.5 μM, 5 μM and 10 μM celastrol for the indicated time points. The cell proliferation was determined by the CCK-8 assay. Data were presented as the percentage of proliferation relative to DMSO-treated control. All measurements were performed in triplicate. Significant differences were compared with the control at ** p b 0.01 by Student's t-test.

1.0 and P values less than 0.05 were selected. Fold changes were calculated as the average mass response (area) ratio between two groups.

At first, the effects of celastrol on proliferation, apoptosis, ΔΨm and caspase-3 expression of HeLa cells were investigated, in order to evaluate the efficacy of celastrol as well as choose the best conditions for the metabolomic analysis. To test whether celastrol inhibits HeLa cells growth, we incubated the cells in various concentrations of celastrol for 24, 48, and 72 h, and then performed CCK-8 assay. As shown in Fig. 2, celastrol inhibited the growth of HeLa cells in a time- and dose-dependent manner (with increasing concentrations from 2.5 to 10 μM) and showed significant inhibition at concentrations of 5 and 10 μM after celastrol treatments for 24, 48 and 72 h (p b 0.05). The ability of celastrol to induce apoptosis in HeLa cells was assessed using the Annexin-V/PI method. Celastrol induced a significant concentration-dependent apoptosis in HeLa cells (Fig. 3). The apoptosis rates were 4.28% in the control group, 7.22%, 37.3%, and 34.8% in the celastrol (2.5, 5 and 10 μM) treated groups, respectively.

Fig. 3. Effect of celastrol on apoptosis of HeLa cells. HeLa cells treated with DMSO or celastrol for 24 h were co-stained with annexin V–FITC and PI and then examined for apoptosis by flow cytometry. The HeLa cells were treated for 24 h with celastrol 2.5 μM (B), 5 μM (C) and 10 μM (D). Control cells were treated with DMSO (A). The celastrol-induced apoptosis rate of HeLa cells increased in a dose dependent manner as shown in the figure.

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

2783

Fig. 4. Mitochondrial dysfunction induced by celastrol treatment. HeLa cells were treated for 24 h with celastrol 2.5 μM (B), 5 μM (C) and 10 μM (D). Control cells were treated with DMSO (A), and then stained with JC-1 dye, incubated with cells for 20 min at 37 °C, 5% CO2 and examined by flow cytometry at the emission wavelength of 530 nm (green, R2). The depolarization of ΔΨm was represented by the percentage of R2 (percentage of JC-1 monomer).

Rapid loss of ΔΨm is believed to be the hallmarks of mitochondrial dysfunction, which may induce the activation of caspase and lead to apoptotic cell death. To determine whether or not apoptotic cell death was triggered by mitochondrial dysfunction, mitochondrial membrane potential assay was performed. As shown in Fig. 4, celastrol induced a significant concentrationdependent depolarization of ΔΨm in HeLa cells. The percentages of R2 (JC-1 monomer) were 0.99% in the control group, 4.64%, 28.58%, and 87.12% in the celastrol (2.5, 5 and 10 μM) treated groups, respectively. To estimate the extent of apoptosis induced in the HeLa cell lines by celastrol treatment, caspase-3 activity was analyzed. Values were normalized to the caspase-3 activity of control group. As shown in Fig. 5, the enzymatic activity of the caspase-3 significantly increased after celastrol treatment. The HeLa cells showed 1.5 and 3.5 fold of increases in caspase-3 enzyme activity as compared to untreated controls after 6 h celastrol (5 and 10 μM) treatment (Pb 0.001).

Fig. 5. Effect of celastrol on activation of caspase-3. HeLa cells were treated with celastrol for 6, 12 and 24 h. Data values were expressed as mean ± SD of triplicate determinations. Significant differences were compared with the control at ** p b 0.01 and *** p b 0.001 by Student's t-test.

2784

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

Fig. 6. The typical total ions current chromatograms (TICs) of HeLa cells intracellular extract. (A) Negative control group; (B) celastrol treated group; (C) cisplatin treated group; (D) doxorubicin hydrochloride treated group; and (E) paclitaxel treated group.

Fig. 7. Score plot from PLS-DA model. (A) Score plot of Celastrol treated group (blue triangle) and negative control group (red square); (B) Score plot of cisplatin treated group (green open triangle) and negative control group (red square); (C) score plot of doxorubicin hydrochloride treated group (violet star) and negative control group (red square); (D) score plot of paclitaxel treated group (brown open inverted triangle) and negative control group (red square). The PLS discriminant model was validated and found to be predictive (A: R2X=0.822, R2Y=0.99 and Q2 =0.987; B: R2X=0.749, R2Y=0.992 and Q2 =0.981; C: R2X=0.801, R2Y=0.994 and Q2 =0.993; D: R2X=0.819, R2Y=0.998 and Q2 =0.984). (For interpretation of the references to color in this figure legend, the reader is referred to the web of this article.)

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

3.2. Dose selection, cell quenching and metabolites extraction Based on the results in Section 3.1, in this investigation, we used a concentration of 5 μM celastrol treatment in confluent cultures for 12 h; this dose could cause mostly cell apoptosis but only little necrosis. The doses of positive control drugs were referred to previous studies (20 μM cisplatin, 4.8 μM doxorubicin hydrochloride and 0.5 μM paclitaxel) [34–36]. The goal of metabolomics is to analyze all or, at least, as many as possible different metabolites without selectivity for any particular molecular type and/or characteristic. Cell quenching and metabolite extraction are the two crucial steps during the cell culture metabolomic sample preparation. Over the past few years, several optimized protocols have been developed for the preparation of intracellular metabolites [37–39]. According to these protocols, we used cold methanol to quench the cells and directly harvest the cells using a cell lifter to avoid the change of metabolites. After that, in order to extract the metabolites as complete as possible, we used methanol–water (80:20) mixture to extract metabolites from frozen-thawed cells [38]. Our method is proved to be rapid, effective and robust compared to the conventional method. 3.3. Metabolic profiling of celastrol treated HeLa cells Typical total ion current chromatograms (TICs) of cell extracts were shown in Fig. 6. For the QC samples, the relative standard deviation (RSD) ranged from 0.04% to 0.52% for the retention times and ranged from 1.58% to 4.96% for the peak areas, which demonstrate the robustness of the method. In order to explore clustering of the negative control and the celastrol treated group, PLS-DA, a method derived from PLS analysis where the Y matrix was set as a dummy descriptor, was used. The goodness of the fit and prediction ability of the model were validated (R 2X = 0.822, R 2Y = 0.99 and Q 2 = 0.987). As could be observed in the PLS-DA score plot (Fig. 7A), separation between these two groups was clearly seen, indicating that biochemical perturbation significantly happened due to celastrol treatment. According to the loading plot of this PLS-DA model, and using the above-stated statistically significant threshold, 14 metabolites with VIP-values greater than 1.0 and P values less than 0.05 were finally revealed to be significant in differentiating the celastrol treated and negative control groups, in which D-lactic acid, D-fructose, 1,4-butanediamine, inosine and guanosine were elevated in celastrol treated group, and other metabolites including serine, L-cysteine, glycine, malic acid, α-ketoglutarate, isocitric acid, octadecanoic acid, D-ribofuranose and α-D-galactopyranoside were reduced in this group (Table 1). The alteration in levels of these metabolites in celastrol treated group manifested the characteristics of metabolic profile in cell response to celastrol's apoptotic effect, which will be explained later. 3.4. Biological explanation of the marker metabolites in celastrol treated group The enzymes of the citrate cycle are mostly located in the mitochondrial matrix, hence mitochondrial dysfunction can reduce the efficiency of citrate cycle [40]. Our study indicated that celastrol led to mitochondrial dysfunction as it induces rapid loss of ΔΨm in HeLa cells. Accordingly, we observed the decline of a couple of key intermediates in citrate cycle (isocitric acid, α-ketoglutarate and malic acid) in celastrol treated group, which also implied the reduced citrate cycle flux (for an overview see http://www.nutritionreview.org/library/krebs.php). Decrease in oxidative metabolism (citrate cycle) could be compensated by increased glycolysis from supplement of pyruvate [41]. In term of the cancer cells, the well-known Warburg effect has revealed that they have increased aerobic glycolysis producing more lactate [42]. In our study, fluctuation of several metabolites in glycolysis

2785

Table 1 Summary of the significant metabolites revealed in this study. Fold change was calculated as the average mass response (area) of celastrol or positive drugs treated group to the negative control group. Abbreviation: Cel (celastrol treated group), Cis (cisplatin treated group), Dox (doxorubicin hydrochloride treated group), Pac (paclitaxel treated group) and NA (not significant). Metabolites

Folder changes R.T. (min) of metabolites

Urea D-Lactic acid L-Leucine

6.02 7.66 9.47

L-Isoleucine

9.86

L-Valine

10.30

D-Mannose

11.37

L-Threonine

11.77

Glycine

11.88

Serine

12.71

Malic acid L-Aspartic acid

14.78 15.30

L-Cysteine

15.85

L-Alanine

16.07

α-Ketoglutarate L-Glutamine

16.25 16.77

1,4-Butanediamine

18.36

Ornithine

18.71

Isocitric acid D-Fructose

19.44 19.93

Hexadecanoic acid Octadecanoic acid Inosine

22.55 24.79 26.12

Guanosine

29.80

D-Ribofuranose

30.29

Cel

α-D-Galactopyranoside 34.40

Cis

Dox

Pathway names Pac

NA NA 7.13 3.23 Pyrimidine metabolism 6.52 5.10 6.30 3.75 Pyruvate metabolism NA 0.36 0.66 0.66 Valine, leucine and isoleucine biosynthesis NA 0.41 0.70 0.72 Valine, leucine and isoleucine biosynthesis NA 0.38 0.50 0.60 Valine, leucine and isoleucine biosynthesis NA 1.79 4.03 3.21 Fructose and mannose metabolism NA 0.37 NA NA Valine, leucine and isoleucine biosynthesis 0.64 0.45 0.23 0.26 Glycine, serine and threonine metabolism 0.48 0.33 0.61 0.52 Glycine, serine and threonine metabolism 0.60 0.17 0.37 0.53 Citrate cycle (TCA cycle) NA 0.24 NA 0.56 Alanine, aspartate and glutamate metabolism 0.45 NA NA NA Cysteine and methionine metabolism NA NA 0.53 0.45 Alanine, aspartate and glutamate metabolism 0.56 0.12 0.60 0.29 Citrate cycle (TCA cycle) NA 0.46 0.43 0.34 Alanine, aspartate and glutamate metabolism 3.51 1.99 1.66 2.24 Glutathione metabolism NA 2.16 4.39 3.08 Glutathione metabolism 0.50 0.24 0.31 0.43 Citrate cycle (TCA cycle) 2.81 3.31 6.18 2.93 Fructose and mannose metabolism NA 0.59 0.71 0.57 Fatty acid biosynthesis 0.32 0.64 0.73 0.62 Fatty acid biosynthesis 5.21 1.40 6.08 3.39 Inosine monophosphate biosynthesis 2.97 2.09 6.37 3.94 Nucleic acid metabolism 0.43 NA NA NA Pentose phosphate pathway 0.46 NA NA NA Unknown

coincides well with the above established knowledge. As the main intermediates of glycolysis, fructose in celastrol treated group is found to increase significantly by 2.81 times compared to the negative control group, which implied up-regulated glycolysis (for an overview see http://themedicalbiochemistrypage.org/non-glucose-sugar-metabolism. php). Lactic acid is another important substance as end product of glycolysis. The dramatically increase of lactic acid (reaching 6.52 times compared to the negative control group) may indicated an augmented consumption of pyruvate, its precursor. All together, the increase of fructose and lactic acid in celastrol treated group implied that celastrol may aggravate Warburg effect of cancer cells, and the insufficiency of pyruvate may partly lead to the reduced citrate cycle [43]. In this study, levels of serine, cysteine and glycine decreases in celastrol treated group. Among them glycine and serine are involved in glycine, serine and threonine metabolism [44,45] and cysteine is involved in cysteine and methionine metabolism [46,47]. The downregulation of these amino acid metabolisms may result from the reduced energy supply, which can also down-regulate protein biosynthesis because of shortage of the metabolism pool [48].

2786

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

For inosine and guanosine, 5.21 and 2.97 fold increase was found in celastrol treated group compared to negative control group, respectively. Inosine and guanosine are key nucleic acid metabolites involved in nucleic acid metabolism, cellular energy metabolism and protein biosynthesis. The reduced nucleic acid metabolism, cellular energy metabolism and amino acid metabolism may lead to reduced consumption of inosine and guanosine. Meanwhile, inosine can improve activity of a variety of enzymes, in particular, coenzyme A and pyruvate oxidase, which can up-regulate glycolysis of the cells under anaerobic conditions [49,50]. Therefore, the increased inosine probably enhances glycolysis. As shown in Fig. 8, the changes of metabolic profiles in the celastrol treated HeLa cells reflect the cell's physiological state. Celastrol can induce apoptosis in HeLa cells by loss of ΔΨm and caspase-3 activation. Mitochondrial dysfunction reduced the efficiency of citrate cycle, and glycolysis is increased to compensate reduced citrate cycle. Concomitantly, the synthesis of amino acids, protein biosynthesis and nucleic acid metabolism may be hampered with the lower energy supply. These metabolic alterations induced by celastrol treatment showed the impaired HeLa cell's physiological activity, and can also provide new insights into celastrol's action mechanisms of anti-proliferation and apoptotic effects. 3.5. Validation of the metabolomic assessment on celastrol using some representative cytotoxic agents In order to validate the reasonability and accuracy of the present metabolomic assessment on celastrol, we proceeded to investigate the effects of three widely used representative cytotoxic agents,

namely cisplatin, doxorubicin hydrochloride and paclitaxel on cervical cancer HeLa cells, which have all been reported to induce apoptosis in HeLa cell [34–36]. Among them, cisplatin is known to cause both DNA damage and apoptosis by binding the cis-[Pt(NH3)2] unit to DNA [51]. Doxorubicin is an antibiotic agent that inhibits DNA topoisomerase and can also induce DNA damage and apoptosis [52]. Paclitaxel, currently the most successful microtubule-targeted chemotherapeutic agent, can cause both mitotic arrest and apoptotic cell death [53]. Three PLS-DA models were established to characterize the metabolic profiles of HeLa cells treated by each of these positive drugs. As shown in the PLS-DA score plots (Fig. 7B to D), obvious separation between the negative control and the positive drug treated groups was seen, indicating that all the drugs intervened metabolic profiles of HeLa cells. Using the screening threshold again (VIP>1.0 and Pb 0.05), we subsequently found several specific metabolites associated with activity of the corresponding positive drugs (Table 1). These metabolites, together with the 14 significant metabolites revealed in Section 3.4, are capable of expanding a metabolic network that is associated with the apoptosis-inducing effects of these drugs, and may possibly extend and strengthen our understanding on celastrol's mechanism of action. To facilitate observing and comparing the metabolic characteristics of all these groups, a heatmap exhibiting levels of all the significant metabolites listed in Table 1 was then constructed (Fig. 9). As shown in Fig. 9, quite a few metabolites have similar trends in celastrol and positive drug treated groups. For example, the down-regulation of several amino acids (glycine and serine) and the citrate cycle intermediates (isocitric acid, α-ketoglutarate and malic acid), and the buildup of intermediates in glycolysis (fructose and lactic acid), glutathione metabolism (1,4-butanediamine) and two synthesis precursors of

Fig. 8. Schematic overview of metabolic perturbation induced by celastrol and positive drugs in HeLa cells. Column value in histograms is expressed as mean ± SD. Abbreviation: NC (negative control group), Cel (celastrol), Cis (cisplatin), Dox (doxorubicin hydrochloride) and Pac (paclitaxel).

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

2787

Fig. 9. Heatmap of significant metabolites detected in HeLa cells intracellular extract metabolomics analysis. Rows: metabolites; columns: sample. Label: NC (negative control group), Cel (celastrol treated group), Cis (cisplatin treated group), Dox (doxorubicin hydrochloride treated group) and Pac (paclitaxel treated group). Color key indicates metabolite expression value, blue: lowest, red: highest. (For interpretation of the references to color in this figure legend, the reader is referred to the web of this article.)

nucleic acid (inosine and guanosine). These findings showed that similar to the positive control drugs, celastrol destructed HeLa cell's physiological activity through affecting the above related pathways (Fig. 8). More specifically, glycine level decreased in all the positive control groups as well as celastrol treated group. According to a recent study [54], glycine uptake and glycine mitochondrial biosynthesis are antagonized in HeLa cells by the cytotoxic drugs. Subsequently, proliferation of HeLa cells could be impaired due to shortage of the metabolism pool as a substrate for synthesis of proteins, nucleic acids and other substances [48]. From the heatmap (Fig. 9), we also found serine level decreased after celastrol and positive drugs treatments. A recent study indicated that targeting serine synthesis pathway may be therapeutically valuable in breast cancers with elevated phosphoglycerate dehydrogenase (PHGDH) expression or PHGDH amplification [55]. From our observation on serine, we supposed that serine synthesis pathway may also be a target of these anticancer drugs in HeLa cells. It remains to be investigated, though, whether PHGDH, a key enzyme in serine synthesis pathway is the target of celastrol. Contents of inosine and guanosine increased in all the drug treated HeLa cells. As is known, inosine and guanosine are employed to produce DNA and RNA. As DNA-targeted drugs [51–53], cisplatin, doxorubicin and paclitaxel could hamper DNA and RNA synthesis, which possibly resulted in the excess of inosine and guanosine. This suggested that celastrol may also play its role by affecting nucleotide synthesis, similar to the positive control drugs. Moreover, similar trends of the metabolites in citrate cycle and glycolysis

in the drug treatment groups indicated that all of them influence these pathways. Interestingly, cysteine level in celastrol treated group decreased significantly whereas it remains unchanged or slightly increased in the positive drug treated groups. Previous studies have proved that celastrol contains electrophilic sites within the rings of quinone methide structure in positions C2 (ring A) and C6 (ring B), enabling its reaction with the nucleophilic thiol groups of cysteine residues to form covalent Michael adducts. This seems to be the major mechanism by which celastrol affects the functions of a series of proteins and exhibits a wide variety of pharmacological activities [14,15,56]. As the cysteine level decreased only in celastrol treated HeLa cells, we suspected that this may result from the covalent binding of thiol group in cysteine to celastrol, being an new evidence of the speculative pharmacological mechanism of celastrol [57]. Although celastrol and the three positive drugs were proved to impair HeLa cell's physiological activity and induce apoptosis in HeLa cells, they act via different mechanisms [3,51–53]. Hence, after treatment on HeLa cells, these drugs are likely to induce various metabolic characteristics. Therefore, we finally used the data of the significant metabolites in all the groups to construct an unsupervised PCA model, to observe the clustering of groups treated with drugs of different mechanisms of action. As shown in the PCA score plot (Fig. 10), separation among the negative control group, celastrol treated group and positive control groups was clearly seen. The negative control group was located in bottom left of the figure, and all the drug treated groups were located

2788

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789

Fig. 10. Score plot from PCA model of significant metabolites revealed in this study. Negative control group (red square), celastrol treated group (blue triangle), cisplatin treated group (green open triangle), doxorubicin hydrochloride treated group (violet star) and paclitaxel treated group (brown open inverted triangle). The PCA model was validated and found to be predictive (R2X= 0.864 and Q2 = 0.708). (For interpretation of the references to color in this figure legend, the reader is referred to the web of this article.)

in the right or up of the figure, which shows different metabolic profile of HeLa cells after treatment of these four cytotoxic agents. In all, characterization of metabolic perturbations signatures in HeLa cells due to drug treatment may help to elucidate drug effects and provide new insights into drug action mechanism. 4. Conclusion In this study, we applied a GC/MS-based metabolomic approach to investigate the metabolic system affected by celastrol treatment in HeLa cells, and for the first time several metabolites indicative for early apoptotic processes in HeLa cells culture induced by celastrol were reported. Three representative apoptosis-inducing cytotoxic agents were selected as positive control drugs to validate reasonableness and accuracy of our metabolomic investigation on celastrol. We found that celastrol and positive control drugs affected citrate cycle, glycolysis, amino acid metabolism and protein biosynthesis of HeLa cells. Moreover, from our results celastrol was likely to bind to cysteine, which provided new evidence of the speculative pharmacological mechanism of celastrol. In conclusion, metabolomic approach applied to cell culture metabolomics could give valuable information on the systemic effects of celastrol treatment and help us to further study the anticancer mechanism of celastrol. Acknowledgements This work was supported by Platform on Research of Metabolism Technology of Traditional Chinese Medicine funded by Science & Technology Department of Shanghai, China (09DZ1975100) and the National Science & Technology Major Special Project for “Major New Drugs Innovation and Development” of China (2009ZX09301-011). Our special thanks go to Professor Junping Zhang (Department of Biochemical Pharmacy, Second Military Medical University) for his helpful advice. References [1] J.H. Lee, T.H. Koo, H. Yoon, H.S. Jung, H.Z. Jin, K. Lee, Y.S. Hong, J.J. Lee, Inhibition of NF-kappa B activation through targeting I kappa B kinase by celastrol, a quinone methide triterpenoid, Biochem. Pharmacol. 72 (2006) 1311–1321. [2] H. Sassa, K. Kogure, Y. Takaishi, H. Terada, Structural basis of potent antiperoxidation activity of the triterpene celastrol in mitochondria: effect of negative membrane surface charge on lipid peroxidation, Free Radic. Biol. Med. 17 (1994) 201–207.

[3] H. Yang, D. Chen, Q.C. Cui, X. Yuan, Q.P. Dou, Celastrol, a triterpene extracted from the Chinese “Thunder of God Vine,” is a potent proteasome inhibitor and suppresses human prostate cancer growth in nude mice, Cancer Res. 66 (2006) 4758–4765. [4] Y.X. Zhou, Y.L. Huang, Antiangiogenic effect of celastrol on the growth of human glioma: an in vitro and in vivo study, Chin. Med. J. (Engl) 122 (2009) 1666–1673. [5] H.S. Yang, J.Y. Kim, J.H. Lee, B.W. Lee, K.H. Park, K.H. Shim, M.K. Lee, K.I. Seo, Celastrol isolated from Tripterygium regelii induces apoptosis through both caspase-dependent and -independent pathways in human breast cancer cells, Food Chem. Toxicol. 49 (2011) 527–532. [6] S.Y. Jang, S.W. Jang, J. Ko, Celastrol inhibits the growth of estrogen positive human breast cancer cells through modulation of estrogen receptor alpha, Cancer Lett. 300 (2011) 57–65. [7] D. Zhang, L. Xu, F. Cao, T. Wei, C. Yang, G. Uzan, B. Peng, Celastrol regulates multiple nuclear transcription factors belonging to HSP90's clients in a dose- and cell type-dependent way, Cell Stress Chaperones 15 (2010) 939–946. [8] W.B. Wang, L.X. Feng, Q.X. Yue, W.Y. Wu, S.H. Guan, B.H. Jiang, M. Yang, X. Liu, D.A. Guo, Paraptosis accompanied by autophagy and apoptosis was induced by celastrol, a natural compound with influence on proteasome, ER stress and Hsp90, J. Cell. Physiol. 227 (2012) 2196–2206. [9] J. Hansen, J. Palmfeldt, S. Vang, T.J. Corydon, N. Gregersen, P. Bross, Quantitative proteomics reveals cellular targets of celastrol, PLoS One 6 (2011) e26634. [10] A. Petronelli, G. Pannitteri, U. Testa, Triterpenoids as new promising anticancer drugs, Anticancer Drugs 20 (2009) 880–892. [11] G. Sethi, K.S. Ahn, M.K. Pandey, B.B. Aggarwal, Celastrol, a novel triterpene, potentiates TNF-induced apoptosis and suppresses invasion of tumor cells by inhibiting NFkappaB-regulated gene products and TAK1-mediated NF-kappaB activation, Blood 109 (2007) 2727–2735. [12] M. Chen, A.E. Rose, N. Doudican, I. Osman, S.J. Orlow, Celastrol synergistically enhances temozolomide cytotoxicity in melanoma cells, Mol. Cancer Res. 7 (2009) 1946–1953. [13] Y. Dai, J.T. DeSano, Y. Meng, Q. Ji, M. Ljungman, T.S. Lawrence, L. Xu, Celastrol potentiates radiotherapy by impairment of DNA damage processing in human prostate cancer, Int. J. Radiat. Oncol. Biol. Phys. 74 (2009) 1217–1225. [14] R. Kannaiyan, M.K. Shanmugam, G. Sethi, Molecular targets of celastrol derived from Thunder of God Vine: potential role in the treatment of inflammatory disorders and cancer, Cancer Lett. 303 (2011) 9–20. [15] A. Salminen, M. Lehtonen, T. Paimela, K. Kaarniranta, Celastrol: Molecular targets of Thunder God Vine, Biochem. Biophys. Res. Commun. 394 (2010) 439–442. [16] M. Karin, Nuclear factor-kappaB in cancer development and progression, Nature 441 (2006) 431–436. [17] V. Baud, M. Karin, Is NF-kappaB a good target for cancer therapy? Hopes and pitfalls, Nat. Rev. Drug Discov. 8 (2009) 33–40. [18] M.S. Hayden, S. Ghosh, Signaling to NF-kappaB, Genes Dev. 18 (2004) 2195–2224. [19] N.D. Perkins, Integrating cell-signalling pathways with NF-kappaB and IKK function, Nat. Rev. Mol. Cell Biol. 8 (2007) 49–62. [20] J.A. Loeb, Identifying targets for preventing epilepsy using systems biology, Neurosci. Lett. 497 (2011) 205–212. [21] O. Fiehn, Combining genomics, metabolome analysis, and biochemical modelling to understand metabolic networks, Comp. Funct. Genomics 2 (2001) 155–168. [22] M. Massimi, A. Tomassini, F. Sciubba, A.P. Sobolev, L.C. Devirgiliis, A. Miccheli, Effects of resveratrol on HepG2 cells as revealed by (1)H-NMR based metabolic profiling, Biochim. Biophys. Acta 1820 (2012) 1–8. [23] K. Kim, P. Aronov, S.O. Zakharkin, D. Anderson, B. Perroud, I.M. Thompson, R.H. Weiss, Urine metabolomics analysis for kidney cancer detection and biomarker discovery, Mol. Cell. Proteomics 8 (2009) 558–570.

Y. Hu et al. / Biochimica et Biophysica Acta 1830 (2013) 2779–2789 [24] A. Sreekumar, L.M. Poisson, T.M. Rajendiran, A.P. Khan, Q. Cao, J. Yu, B. Laxman, R. Mehra, R.J. Lonigro, Y. Li, M.K. Nyati, A. Ahsan, S. Kalyana-Sundaram, B. Han, X. Cao, J. Byun, G.S. Omenn, D. Ghosh, S. Pennathur, D.C. Alexander, A. Berger, J.R. Shuster, J.T. Wei, S. Varambally, C. Beecher, A.M. Chinnaiyan, Metabolomic profiles delineate potential role for sarcosine in prostate cancer progression, Nature 457 (2009) 910–914. [25] J.L. Griffin, J.P. Shockcor, Metabolic profiles of cancer cells, Nat. Rev. Cancer 4 (2004) 551–561. [26] A.T. Miccheli, A. Miccheli, R. Di Clemente, M. Valerio, P. Coluccia, M. Bizzarri, F. Conti, NMR-based metabolic profiling of human hepatoma cells in relation to cell growth by culture media analysis, Biochim. Biophys. Acta 1760 (2006) 1723–1731. [27] G.S. Salomons, L.A. Smets, M. Verwijs-Janssen, A.A. Hart, E.G. Haarman, G.J. Kaspers, E.V. Wering, A.V. Der Does-Van Den, W.A. Kamps Berg, Bcl-2 family members in childhood acute lymphoblastic leukemia: relationships with features at presentation, in vitro and in vivo drug response and long-term clinical outcome, Leukemia 13 (1999) 1574–1580. [28] F.J. Picard, M.G. Bergeron, Rapid molecular theranostics in infectious diseases, Drug Discov. Today 7 (2002) 1092–1101. [29] A. Huenerbein, M.A. Sipoli Marques, S. Pereira Ados, F.R. de Aquino Neto, Improvement in steroid screening for doping control with special emphasis on stanozolol, J. Chromatogr. A 985 (2003) 375–386. [30] A. Cossarizza, M. Baccarani-Contri, G. Kalashnikova, C. Franceschi, A new method for the cytofluorimetric analysis of mitochondrial membrane potential using the J-aggregate forming lipophilic cation 5,5',6,6'-tetrachloro-1,1',3,3'tetraethylbenzimidazolcarbocyanine iodide (JC-1), Biochem. Biophys. Res. Commun. 197 (1993) 40–45. [31] T. Sangster, H. Major, R. Plumb, A.J. Wilson, I.D. Wilson, A pragmatic and readily implemented quality control strategy for HPLC-MS and GC–MS-based metabonomic analysis, Analyst 131 (2006) 1075–1078. [32] H.G. Gika, G.A. Theodoridis, J.E. Wingate, I.D. Wilson, Within-day reproducibility of an HPLC-MS-based method for metabonomic analysis: application to human urine, J. Proteome Res. 6 (2007) 3291–3303. [33] G. Tan, Z. Lou, W. Liao, Z. Zhu, X. Dong, W. Zhang, W. Li, Y. Chai, Potential biomarkers in mouse myocardium of doxorubicin-induced cardiomyopathy: a metabonomic method and its application, PLoS One 6 (2011) e27683. [34] T. Luo, J. Yu, J. Nguyen, C.R. Wang, R.G. Bristow, D.A. Jaffray, X.Z. Zhou, K.P. Lu, Q.B. Lu, Electron transfer-based combination therapy of cisplatin with tetramethyl-pphenylenediamine for ovarian, cervical, and lung cancers, Proc. Natl. Acad. Sci. U. S. A. 109 (2012) 10175–10180. [35] A.A. Aljabali, S. Shukla, G.P. Lomonossoff, N.F. Steinmetz, D.J. Evans, CPMV-DOX Delivers, Mol. Pharm. (2012), http://dx.doi.org/10.1021/mp3002057. [36] C. Feng, L.Y. Zhou, T. Yu, G. Xu, H.L. Tian, J.J. Xu, H.X. Xu, K.Q. Luo, A new anticancer compound, Oblongifolin C, inhibits tumor growth and promotes apoptosis in HeLa cells through bax activation, Int. J. Cancer 131 (2012) 1445–1454. [37] Q. Teng, W. Huang, T. Collette, D. Ekman, C. Tan, A direct cell quenching method for cell-culture based metabolomics, Metabolomics 5 (2009) 199–208. [38] K. Dettmer, N. Nurnberger, H. Kaspar, M.A. Gruber, M.F. Almstetter, P.J. Oefner, Metabolite extraction from adherently growing mammalian cells for metabolomics studies: optimization of harvesting and extraction protocols, Anal. Bioanal. Chem. 399 (2011) 1127–1139. [39] S. Dietmair, N.E. Timmins, P.P. Gray, L.K. Nielsen, J.O. Kromer, Towards quantitative metabolomics of mammalian cells: development of a metabolite extraction protocol, Anal. Biochem. 404 (2010) 155–164.

2789

[40] E.O. Akgul, M. Toygar, Y. Poyrazoglu, Evaluating the mitochondrial function, Diabetol. Metab. Syndr. 4 (2012) 11. [41] A.M. Puzio-Kuter, The Role of p53 in Metabolic Regulation, Genes Cancer 2 (2011) 385–391. [42] M.G. Vander Heiden, L.C. Cantley, C.B. Thompson, Understanding the Warburg effect: the metabolic requirements of cell proliferation, Science 324 (2009) 1029–1033. [43] J. Wei, G. Xie, Z. Zhou, P. Shi, Y. Qiu, X. Zheng, T. Chen, M. Su, A. Zhao, W. Jia, Salivary metabolite signatures of oral cancer and leukoplakia, Int. J. Cancer 129 (2010) 2207–2217. [44] A.U. Kuhlmann, E. Bremer, Osmotically regulated synthesis of the compatible solute ectoine in Bacillus pasteurii and related Bacillus spp, Appl. Environ. Microbiol. 68 (2002) 772–783. [45] D. Canovas, N. Borges, C. Vargas, A. Ventosa, J.J. Nieto, H. Santos, Role of Ngamma-acetyldiaminobutyrate as an enzyme stabilizer and an intermediate in the biosynthesis of hydroxyectoine, Appl. Environ. Microbiol. 65 (1999) 3774–3779. [46] A. Sekowska, V. Denervaud, H. Ashida, K. Michoud, D. Haas, A. Yokota, A. Danchin, Bacterial variations on the methionine salvage pathway, BMC Microbiol. 4 (2004) 9. [47] A. Goyer, E. Collakova, Y. Shachar-Hill, A.D. Hanson, Functional characterization of a methionine gamma-lyase in Arabidopsis and its implication in an alternative to the reverse trans-sulfuration pathway, Plant Cell Physiol. 48 (2007) 232–242. [48] J. Fan, J. Hong, J.D. Hu, J.L. Chen, Ion chromatography based urine amino Acid profiling applied for diagnosis of gastric cancer, Gastroenterol. Res. Pract. 2012 (2012) 474907. [49] E.D. Lewandowski, D.L. Johnston, R. Roberts, Effects of inosine on glycolysis and contracture during myocardial ischemia, Circ. Res. 68 (1991) 578–587. [50] G. Hasko, M.V. Sitkovsky, C. Szabo, Immunomodulatory and neuroprotective effects of inosine, Trends Pharmacol. Sci. 25 (2004) 152–157. [51] S. Ramachandran, A.P. Quist, S. Kumar, R. Lal, Cisplatin nanoliposomes for cancer therapy: AFM and fluorescence imaging of cisplatin encapsulation, stability, cellular uptake, and toxicity, Langmuir 22 (2006) 8156–8162. [52] S.C. Huelsenbeck, A. Schorr, W.P. Roos, J. Huelsenbeck, C. Henninger, B. Kaina, G. Fritz, Rac1 signaling is required for DNA damage response stimulated by topoisomerase II poisons, J. Biol. Chem. 287 (2012) 38590–38599. [53] M.T. Branham, S.B. Nadin, L.M. Vargas-Roig, D.R. Ciocca, DNA damage induced by paclitaxel and DNA repair capability of peripheral blood lymphocytes as evaluated by the alkaline comet assay, Mutat. Res. 560 (2004) 11–17. [54] M. Jain, R. Nilsson, S. Sharma, N. Madhusudhan, T. Kitami, A.L. Souza, R. Kafri, M.W. Kirschner, C.B. Clish, V.K. Mootha, Metabolite profiling identifies a key role for glycine in rapid cancer cell proliferation, Science 336 (2012) 1040–1044. [55] R. Possemato, K.M. Marks, Y.D. Shaul, M.E. Pacold, D. Kim, K. Birsoy, S. Sethumadhavan, H.K. Woo, H.G. Jang, A.K. Jha, W.W. Chen, F.G. Barrett, N. Stransky, Z.Y. Tsun, G.S. Cowley, J. Barretina, N.Y. Kalaany, P.P. Hsu, K. Ottina, A.M. Chan, B. Yuan, L.A. Garraway, D.E. Root, M. Mino-Kenudson, E.F. Brachtel, E.M. Driggers, D.M. Sabatini, Functional genomics reveal that the serine synthesis pathway is essential in breast cancer, Nature 476 (2011) 346–350. [56] S. Sreeramulu, S.L. Gande, M. Gobel, H. Schwalbe, Molecular mechanism of inhibition of the human protein complex Hsp90-Cdc37, a kinome chaperone-cochaperone, by triterpene celastrol, Angew. Chem. Int. Ed Engl. 48 (2009) 5853–5855. [57] L. Klaic, P.C. Trippier, R.K. Mishra, R.I. Morimoto, R.B. Silverman, Remarkable stereospecific conjugate additions to the Hsp90 inhibitor celastrol, J. Am. Chem. Soc. 133 (2011) 19634–19637.