Research Article

Received: 21 June 2010, Revised: 16 August 2010, Accepted: 16 August 2010 Published online in Wiley Online Library: 10 December 2010

(wileyonlinelibrary.com) DOI 10.1002/jat.1591

GC-MS-based metabolomics reveals mechanism of action for hydrazine induced hepatotoxicity in rats
Kiyoko Bando,a,b Takeshi Kunimatsu,b Jun Sakai,c Juki Kimura,b Hitoshi Funabashi,b Takaki Seki,b Takeshi Bambaa and Eiichiro Fukusakia*
ABSTRACT: Gas chromatographymass spectrometry (GC-MS) has great advantages for analyzing organic/amino acids, which are often targets in ecacy and/or toxicity studies. Although GC-MS has been used for the detection of many metabolic disorders, applications of GC-MS-based metabolomics in pharmacology/toxicology are relatively underdeveloped. We intended to investigate applicability of a GC-MS-based metabolomics approach for toxicological evaluation, and tried to elucidate the mechanism of hydrazine-induced hepatotoxicity. Rats were administered hydrazine chloride orally (120 and 240 mg kg-1), and urine, plasma and liver samples were collected at 24 or 48 h post-dosing. Conventional clinical chemistry and liver histopathology were performed, urine and plasma were analyzed by GC-MS, and metabolic proles were assessed using chemometric techniques. Principal component analysis score plots showed clear separation of the groups, indicating dose-dependent toxicity and recovery. The mechanism of toxicity was investigated based on semi-quantication data of identied metabolites. Amino acid precursors of glutathione (cystein, glutamate and glycine) and a product of glutathione metabolism (5-oxoproline) were elevated dose-dependently, accompanied with elevation of ascorbate levels. In addition, intermediates of the TCA cycle were decreased, whereas participants of the urea cycle and other amino acids were increased. These alterations were associated with histopathological changes such as fatty degeneration and glycogen accumulation. Application of GC-MS-based metabolomics revealed that oxidative stress and GSH consumption play important roles in the etiology of hydrazine-induced hepatotoxicity, demonstrating that this approach is a useful tool in pharmacology and toxicology for screening, elucidating mode of action and biomarker discovery. Copyright 2010 John Wiley & Sons, Ltd. Keywords: hydrazine; hepatotoxicity; metabolomics; GC-MS; oxidative stress

INTRODUCTION
Hydrazine it has been used widely as an intermediate in industrial synthetic chemistry and as a rocket fuel (ONeil, 2006). Hydrazine has also been found to be a metabolite of some important drugs, such as the antihypertensive drug hydralazine (Blair et al., 1985; Timbrell and Harland, 1979) and the antituberculosis drug isoniazid (Blair et al., 1985; Iguchi et al., 1977). It has been reported that hydrazine induces hepatotoxicity (Scales and Timbrell, 1982; Timbrell et al., 1982; Watereld et al., 1993), carcinogenicity (IARC, 1999), mutagenicity (IARC, 1999), teratology (Toth, 1993) and neurotoxicity (Moloney and Prough, 1983). Hydrazine-induced steatosis has been extensively studied in animal models; however, the denitive mechanism of toxicity has not been understood to date. Metabolomic investigations attempt to detect and prole changes in metabolites, which reect changes in metabolic pathways and may provide information concerning a disease state or the biological stress of an organism (Lindon et al., 2004; Weckwerth and Morgenthal, 2005). Metabolomics is increasingly being applied to pharmacology and toxicology studies, and in particular, high-resolution 1H nuclear magnetic resonance (NMR) spectrometry-based metabolome analysis study is a wellestablished technique for detection of the endogenous metabolic changes in biouids such as plasma and urine caused by drug toxicity or disease processes (Clarke and Haselden, 2008;

Goldsmith et al., 2010; Lindon et al., 2007; Powers, 2009; Serkova and Niemann, 2006). In fact, NMR-based metabolomics approaches were previously used to investigate hydrazine toxicity of animals, and it has already been reported that hydrazine induces alterations of endogenous metabolites (Bollard et al., 2005; Garrod et al., 2005; Kleno et al., 2004; Nicholls et al., 2001; Wang et al., 2003). However, the majority of metabolic proling studies using combined gas chromatographymass spectrometry (GC-MS) and chemometric techniques reside in the eld of plant metabolomics

*Correspondence to: E. Fukusaki, Department of Biotechnology, Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: fukusaki@bio.eng.osaka-u.ac.jp
a Department of Biotechnology, Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan b Safety Research Laboratories, Dainippon Sumitomo Pharma Co. Ltd, 3-1-98 Kasugade-naka, Konohana-ku, Osaka 554-0022, Japan c Genomic Science Laboratories, Dainippon Sumitomo Pharma Co. Ltd, 3-1-98 Kasugade-naka, Konohana-ku, Osaka 554-0022, Japan The study represents a portion of the dissertation submitted by Kiyoko Bando to Osaka University in partial fulllment of the requirement for her Ph.D.

524

Copyright 2010 John Wiley & Sons, Ltd.

J. Appl. Toxicol. 2011; 31: 524535

GC-MS-based metabolomics of hydrazine toxicity (Bino et al., 2004; Fiehn et al., 2000; Glinski and Weckwerth, 2006). Although GC-MS has been used for the detection of many metabolic disorders since the 1970s (Chace, 2001), the application of GC-MS-based metabolomics for pharmacology/toxicology is relatively underdeveloped compared with NMR or LC-MS. Advantages of GC-MS include high resolution and reproducibility, as well as the availability of a reliable electron impact (EI) mass spectral library database for metabolite identication or conrmation. Liquid chromatography (LC/MS) mass spectral libraries are, in most cases, instrument-dependent, and therefore standard reference LC/MS libraries are unavailable for general use, making the identication eciency even lower for metabolites (Bino et al., 2004). Additionally, GC-MS has great advantages for analyzing organic acids and amino acids, which are often targets in ecacy and/or toxicity studies. Therefore, metabolic proling using GC-MS has potential as a powerful tool in toxicological evaluations, providing a comprehensive understanding of the response of biological systems to xenobiotic intervention (Lee et al., 2009; Lee et al., 2007; Pasikanti et al., 2008). In the present study, we investigated the applicability of a GC-MS-based metabolomics approach for toxicological evaluation, and tried to elucidate the mechanism of toxicity of hydrazine using metabolic proles derived from GC-MS. 24.0 2.0 C, relative humidity 55.0 15.0%) with a 12 h light dark cycle (uorescent lighting from 8:00 to 20:00) and more than 10 air changes per hour. The rats received pelleted basal diet (CRF-1; Oriental Yeast Co. Ltd, Chiba, Japan) and tap water ad libitum. However, diet was removed during urine collection. After quarantine/acclimatization for 7 days, hydrazine was administered to rats at 7 weeks of age (body weight 228278 g at 7 weeks of age).

Animal Study Design Animals were randomly assigned to six groups (eight animals/ group), and three of the groups were allocated to a sampling group for sampling at 24 h post-dosing, while the other three groups were allocated to a sampling group for sampling at 48 h post-dosing. The dosage levels of hydrazine dihydrochloride were 0, 120 and 240 mg kg-1. The high dosage was selected as a dosage level expected to show marked hepatotoxicity, and a half dose of the high dose was set as a low dose to observe the dose-dependence of toxicity. Hydrazine dihydrochloride was dissolved in saline, and dosing formulation volumes were calculated based on body weights of animals (5 ml kg-1). Saline was administered to rats of the 0 mg kg-1 groups (control groups). The dosing formulations were drawn into polypropylene syringes, and were promptly administered once orally via intubation. Figure 1 shows a urine and plasma sampling design. Animals were housed in metabolic cages for 8 h from 16 to 24 h postdosing and from 40 to 48 h post-dosing, and then urine samples were collected. During urine pooling in metabolic cages, the urine collection tubes were frozen on dry ice. Pooled urine samples were thawed in a refrigerator after the collection period, all urine samples were centrifuged at 540g for 5 min at 5 C, and a portion of supernatant was used for measurement of creatinine concentration. Creatinine concentrations of urine samples were measured by a sarcosine oxidaseperoxidase method (JCABM1650, JEOL, Japan). Remaining urine samples were divided into sub-samples, frozen rapidly using liquid N2, and stored in a deep freezer (-80 C) until GC-MS analysis. Blood samples were drawn via the abdominal aorta under isoflurane anesthesia from animals at 24 and 48 h post-dosing. Blood samples were divided into two sub-samples. One was treated with EDTA-2K and the other was treated with sodium heparin as anticoagulant, then both were centrifuged at 2150g for 10 min at 5 C, and supernatants (plasma) were divided into sub-samples, frozen rapidly using liquid N2, and stored in a deep freezer (-80 C) until analysis.

MATERIALS AND METHODS

Chemicals Hydrazine dihydrochloride and EDTA-2K and were purchased from Kanto Chemical Co. Inc. (Tokyo, Japan). Sodium heparin (Novo-Heparin for injectionTM) was purchased from Mochida Pharmaceutical Co. Ltd. (Tokyo, Japan). Saline, a Japanese Pharmacopoeia standard product (Saline PL Fuso), was purchased from Fuso Pharmaceutical Industries Ltd (Osaka, Japan). Isourane (IsouTM) was obtained from Dainippon Sumitomo Pharma Co. Ltd (Osaka, Japan). Chemicals used in GC-MS analysis were as follows: methoxyamine hydrochloride from Sigma Aldrich Corp. (St Louis, MO, USA), pyridine from Wako Pure Chemical Industries Ltd (Osaka, Japan), chloroform and methanol from Kishida Chemical Co. Ltd (Osaka, Japan), and N-methyl-N-(trimethylsilyl)triuoroacetamide (MSTFA) from GL Science Inc. (Tokyo, Japan). Animals Male Crl:CD(SD) rats (SPF) were purchased from Charles River Japan Inc. (Hino Breeding Facility, Japan). Two rats per cage were housed in aluminum wire mesh cages or rats were housed individually in metabolic cages during the collection of urine, and maintained in an air-conditioned animal room (temperature

Figure 1. Two sampling groups were designed: 24 h and 48 h post-dosing groups. Urine samples were pooled for 8 h from 16 to 24 h post-dosing and from 40 to 48 h post-dosing. Just after pooling urine, blood was drawn from animals, and plasma samples were obtained. Then, animals were necropsied and liver samples were obtained.

525

J. App. Toxicol. 2011; 31: 524535

Copyright 2010 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jat

K. Bando et al. The present study was conducted in compliance with the Law Concerning the Protection and Control of Animals, with Partial Amendments (Law No. 68, Jun. 22, 2005, Japan) and in-house guidance. (EI) energy of 70 kV, and 20 scans s-1 were recorded over the mass range of m/z 85500. The acceleration voltage was applied after a 428 and 250 s solvent delay on urine and plasma samples, respectively.

Blood Biochemistry Plasma, obtained by treatment with heparin, was used for determining the owing blood biochemistry parameters: aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), lactate dehydrogenase (LDH), total bilirubin (T-Bil), g -glutamyl transpeptidase(g -GTP), creatine kinase (CK), total cholesterol (T-Cho), phospholipids (PL), triglycerides (TG), glucose (Glu), urea nitrogen (BUN), creatinine (Cre), inorganic phosphorus (P), calcium (Ca), sodium (Na), potassium (K) and chloride (Cl). Measurements were performed using a clinical biochemistry analyzer (JCA-BM1650, JEOL, Japan).

Analysis of GC-MS Data Baseline correction and automatic peak detection with ChromaTOF (LECO, St Joseph, MI, USA) software were performed with a peak width of 1.333. Peaks with signal-to-noise (S/N) ratios lower than 10 were rejected, and raw chromatographic data (Pegasus le, *.peg) were converted into ANDI les (Analytical Data Interchange Protocol, *.cdf ). ANDI formatted data can be converted and transferred among dierent mass spectral equipment. The ANDI formatted data were read into LineUp (Infometrix Inc., Bothell, WA, USA) and PiroTrans (GL Science Inc. Tokyo, Japan), where alignment of chromatogram and normalization of intensity were performed on the basis of ribitol peak in plasma samples. Meanwhile, chromatogram alignment was performed on the basis of a particular peak (retention time: around 8.8 min), which had appeared in chromatograms of all urine samples (probably a peak derived from derivatization reagents), and intensities of each peak were normalized by creatinine concentrations of each animal, because 2-aminoadipate, which was remarkably elevated in urine by hydrazine treatment, co-eluted with ribitol in urine samples. The generated peak lists were imported into SIMCA-P+ (version 12; Umetrics, Ume/Malm, Sweden) for multivariate statistical analysis. Principal component analysis (PCA) was conducted to nd alterations of metabolic proling induced by hydrazine treatment.

Histopathology All animals were necropsied, and liver was removed and xed in 10% neutral buered formalin. Liver specimens were trimmed, embedded in paran wax, sectioned and stained with hematoxylin and eosin (H&E) for histopathological examination. The sections were also stained by periodic acidSchi (PAS) method with or without diastase digestion.

Sample Preparation for GC-MS Analysis Mixed solvent (250 ml; CHCl3:CH3OH:H2O, 1 : 2.5 : 1, v/v) was mixed into 50 ml of urine or plasma samples with 90 ml of 0.2 mg ml-1 solution of ribitol, which was used as an internal standard. The samples were shaken for 30 min at 37 C and centrifuged at 16 000g for 3 min at 4 C. A 250 ml portion of the supernatant was transferred to an Eppendorf tube with a pierced cap. The samples were dried rst in a vacuum centrifuge dryer and then a freeze dryer. For derivatization, 100 ml of methoxylamine hydrochloride in pyridine (20 mg ml-1) the rst derivatizing agent was added to the samples. The mixture was incubated at 30 C for 90 min. The second derivatizing agent, i.e. 50 ml N-methyl-N(trimethylsilyl)triuoroacetamide (MSTFA), was added, and the mixture was incubated at 37 C for 30 min. A 1 ml sample was injected in split mode (urine, 25 : 1, v/v; plasma, 10 : 1, v/v).

Identication of Metabolites and Semi-quantication Remarkably altered metabolites and associated metabolites induced by hydrazine were identied and relative intensities were calculated. Metabolites were identied by mass spectral patterns and retention times on the basis of an in-house chemical library; 82 and 67 metabolites were identied in the urine and plasma samples, respectively. Total intensities of each metabolite peak in plasma were normalized by the total intensity of the ribitol peak, while total intensities of each metabolite peak in urine were normalized by the intensity of unique mass (m/z 319) of ribitol peak and creatinine concentration. The unique mass (m/z 319) of ribitol was not included in the mass spectrum of 2-aminoadipate, which co-eluted with ribitol in urine samples of the hydrazine treatment groups, and therefore the intensity of the unique mass was considered an appropriate normalization factor.

GC-MS Analysis Chromatography was performed on an 6890N gas chromatograph system (Agilent Co., Palo Alto, CA, USA) equipped with a fused-silica capillary column (internal diameter, i.d., 30 m 0.25 mm) coated with a CP-SIL 8 CB low-bleed lm (thickness, 0.25 mm; Varian Inc., Palo Alto, CA, USA) that was connected to a Pegasus III TOF mass spectrometer (LECO, St Joseph, MI, USA) equipped with an autosampler 7683B series injector (Agilent Co., Palo Alto, CA, USA). The injection temperature was 230 C. The helium gas ow rate through the column was 1 ml min-1. The column temperature was isothermally maintained at 80 C for 2 min and then raised to 330 C at the rate of 15 C min-1; the temperature was isothermally maintained at 330 C for 6 min. The temperatures of the transfer line and the ion source were 250 and 200 C, respectively. Ions were generated at an electron impact

RESULTS
Toxicological Examination Table 1 shows prominent alterations in parameters of the blood biochemistry in rats treated with hydrazine. Several parameters remarkably changed in the hydrazine-treated groups; these changes were dose-dependent at 24 and 48 h post-dosing. Changes at 48 h post-dosing were less than those at 24 h post-dosing, which indicated that 48 h post-dosing was on the

526

wileyonlinelibrary.com/journal/jat

Copyright 2010 John Wiley & Sons, Ltd.

J. App. Toxicol. 2011; 31: 524535

GC-MS-based metabolomics of hydrazine toxicity

Table 1. Alterations in the parameters of blood biochemistry in rats treated with hydrazine 24 h Dose (mg kg-1) 120 10.6 2.4 0.12 8.7 8.3 23.0 1.5 27 18 4.9 49 103 29 14 5.5** 2.1** 0.12 9.8** 14.4* 11.4* 3.5* 21 20 4.5 31 77 37 16 48 h Dose (mg kg-1) 120 36.4 2.5 0.25 8.7 15.4 39.7 1.1 60 16 5.0 65 112 38 11 10.3 2.2 0.22 8.0 13.9 27.8 1.9 41 18 4.8 43 96 23 15

0 AST (U/L) ALT (U/L) TP (g dl-1) TC (mg dl-1) PL (mg dl-1) TG (mg dl-1) BUN (mg dl-1) 67 25 5.0 68 120 53 10

240 2.9** 6.7 0.23** 10.3** 18.0** 19.4 2.0** 78 26 5.2 70 127 76 11

240 11.4 5.2 0.16 16.3 21.6 8.4 4.9

AST, aspartate aminotransferase; ALT, alanine aminotransferase; TP, total protein; TC, total cholesterol; PL, phospholipid; TG, triglyceride; BUN, urea nitrogen. *Signicantly dierent from the 0 mg kg-1 group (24 h) (P < 0.05); **signicantly dierent from the 0 mg kg-1 group (24 h) (P < 0.01). Signicantly dierent from the 0 mg kg-1 group (48 h) (P < 0.05); signicantly dierent from the 0 mg kg-1 group (48 h) (P < 0.01).

Table 2. Treatment-related histopathological ndings of the liver in rats treated with hydrazine 0 mg kg-1 24a 48 8b 8 0c 0 0 0 0 0 0 8 0 0 0 0 0 0 0 8 120 mg kg-1 24 48 8 8 3 2 0 1 2 3 0 2 3 0 0 0 0 0 0 5 240 mg kg-1 24 48 8 8 2 3 1 5 2 3 0 0 2 2 1 2 2 3 2 0

Groups Histopathological ndings/grade Fatty degeneration, midzonal Minimal Mild/slight Moderate Anisonucleosis Minimal Single cell necrosis Minimal Increase, Pas-positive granules, centrilobular Minimal Mild/slight No abnormality detected
a

Hours after dosing. Number of animals examined. c Number of animals bearing the lesion.
b

convalescent phase. Typical conventional hepatotoxicity biomarkers, aspartate aminotransferase (AST) and alanine aminotransferase (ALT), decreased in the hydrazine-treated group. These phenomena were previously reported (Watereld et al., 1993). Hydrazine and its analogs have been shown to lead to sequestration of the aminotransferase cofactor pyridoxal 5-phosphate (PLP), inhibiting aminotransferase activity (Cornish, 1969; Lightcap and Silverman, 1996), which may explain the reduced levels of ALT and AST in the rat, despite the observation of liver toxicity, thus conventional hepatotoxicity biomarkers cannot predict hydrazine induced hepatotoxicity. A few lipid parameters (i.e. total cholesterol, phospolipid, and triglyceride) and total protein decreased dose-dependently, which may suggest metabolic turnover of lipid and protein were lower in the liver by hydrazine treatment. In addition, the level of BUN slightly increased at 24 h post-dosing in a dose dependent manner, which may suggest kidney hypofunction. The incidences of the histopathological ndings in the liver are presented in Table 2, and representative photographs are pro-

vided in Figure 2. Figure 2(A and B) shows centrilobular to midzonal areas of the liver in control and hydrazine-treated rats, respectively. In Fig. 2(A), normal hepatocytes surrounding the central vein, which is located in the center of the picture, can be observed. In hydrazine-treated rats, fatty degenerations characterized by ne cytoplasmic vacuoles in hepatocytes were seen in the midzonal areas (Fig. 2B). In addition, anisonucleosis and single cell necrosis of the hepatocytes were observed (Fig. 2C, D). Figure 2(C) shows the nuclei of hepatocytes which are irregular in size and Fig. 2(D) shows isolated extracellular, eosinophilic bodies, which are characteristic in single cell necrosis. These ndings were seen to occur in a dose dependent manner and more prominent at 24 h post-dosing than at 48 h post-dosing, which indicated 48 h post-dosing was on the convalescent phase, in the same manner as with blood biochemistry parameters. Moreover, hydrazine-treated rats showed minimal to mild increases of PASpositive granules in hepatocytes of the centrilobular areas (Fig. 2E, F). These granules were digested by diastase (data not shown), revealing that they consisted of glycogen. All the

527

J. App. Toxicol. 2011; 31: 524535

Copyright 2010 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jat

K. Bando et al.

Figure 2. Histopathology of the liver 24 h after dosing. No abnormality was detected in control rats (0 mg kg-1) (A). In hydrazine-treated rats (240 mg kg-1), fatty degenerations in the midzonal areas (B), anisonucleosis (C) and single cell necrosis (arrowhead) (D) were observed. H&E. In contrast to the control rats (0 mg kg-1) (E), hydrazine-treated rats (240 mg kg-1) showed increases in PAS-positive granules in hepatocytes of the centrilobular areas (F).

histopathological ndings indicated the toxic eects of hydrazine, which induced steatosis, as in preliminary reports (Scales and Timbrell, 1982; Timbrell et al., 1982).

Metabolic Proling of Urine and Plasma after Treatment with Hydrazine First, metabolome analysis was performed via non-target approach; all peak data of GC-MS chromatograms were used for multivariate analysis after alignment and normalization. Figure 3(A) shows a score plot of principal component analysis (PCA) of urine data. Each group was clearly clustered depending on dosages or sampling time after hydrazine treatment; spots of 240 mg kg-1 groups were farther from those of 0 mg kg-1 groups than were those of 120 mg kg-1 groups. Furthermore, spots of 48 h post-dosing groups were closer to those of 0 mg kg-1 groups than were those of 24 h post-dosing groups; in particular, spots of 120 mg kg-1 group at 48 h post-dosing group were plotted at the same area of those of 0 mg kg-1 groups. These results suggested that metabolic proling showed dose-dependent toxicity and recovery induced by hydrazine. The corresponding loading plot revealed metabolites that were responsible for group separation (Fig. 3C), and those metabolites were identied by mass spectra and retention times on the basis of an in-house chemical library. Highly contributing metabolites found to be elevated in the hydrazine-treated groups were 2-aminoadipate and b-alanine,

whereas highly contributing metabolites found to be decreased in the hydrazine-treated groups were 2-oxoglutarate and citrate. The PCA score plot of plasma metabolic proling showed dose and time dependency as did urine metabolic proling (Fig. 3B); however, group separation was less clear than in those of plasma. The corresponding loading plot revealed glucose decreased in the hydrazine treatment group (Fig. 3D); however, individual variation in glucose levels was large, which may have led to less group separation on the score plot. Highly contributing metabolites other than glucose found to be elevated in the hydrazinetreated groups were lactate and tyrosine.

Alterations of Metabolites Associated with Glutathione Metabolism Next, remarkably altered metabolites and associated metabolites induced by hydrazine were identied, and relative intensities of each metabolite were calculated. The mapping of these data into general biochemical pathways as illustrated in the Kyoto Encyclopedia of Genes and Genomes (KEGG) revealed signicant biological impact of alterations of metabolic proling by hydrazine treatment. We found that the glutathione metabolism pathway was the most noteworthy in relation to hydrazine-induced toxicity. Figure 4 shows glutathione metabolism pathway and relative intensities of associated metabolites. Intermediate metabolites of glutathione, namely cystein, glutamate and glycine in the urine

528

wileyonlinelibrary.com/journal/jat

Copyright 2010 John Wiley & Sons, Ltd.

J. App. Toxicol. 2011; 31: 524535

GC-MS-based metabolomics of hydrazine toxicity

Figure 3. Score plots and the corresponding loading plots of principal component analysis. PCA score plot derived from GC-MS spectra of urine (A) and plasma (B) samples. Corresponding loading plots of PC 1 and PC 2 in urine (C) and plasma (D). Twenty-four hour groups: 0 mg kg-1, solid circles; 120 mg kg-1, solid triangles; 240 mg kg-1, solid squares. Forty-eight hour groups: 0 mg kg-1, open circles; 120 mg kg-1, open triangles; 240 mg kg-1, open squares.

and/or plasma were elevated by hydrazine treatment at 24 h post-dosing. 5-Oxoproline, a product of glutathione metabolism, also slightly increased, which is an indicator for glutathione biosynthesis ux and oxidative stress (Lu, 2009; Metges et al., 1999, 2000; Yu et al., 2002). These alterations indicated hydrazineinduced oxidative stress, although glutathione (GSH) and glutathione disulde (GSSG) could not be directly measured. At 48 h post-dosing, levels of the metabolites had recovered: particularly, the levels in the120 mg kg-1 group were generally comparable to those in the 0 mg kg-1 group. In addition, ascorbate, a typical anti-oxidant (Foyer, 2001; Sies, 1997) was also remarkably elevated in the urine of 240 mg kg-1 groups, and the levels at 48 h post-dosing were almost same as those at 24 h post-dosing. This may suggest continuous response to hydrazine-induced oxidative stress. The levels of putrescine also were elevated, which may relate with alterations in the urea cycle, as described below.

shows the TCA cycle and alterations of relative intensities of intermediate metabolites. At 24 h post-dosing, almost intermediate citrate, 2-oxoglutarate, succinate and fumarate, were found to be decreased dose-dependently in urine and/or plasma. Changes of some metabolites, which were detected in both urine and plasma, were more dramatic in urine than in plasma. These results indicated that hydrazine may cause alteration in the TCA cycle. On the contrary, the levels of pyruvate and lactate in plasma and urinary glucose were elevated in the hydrazine treatment groups. At 48 h post-dosing, levels of the metabolites had recovered: particularly, the levels in the 120 mg kg-1 group were generally comparable to those in the 0 mg kg-1 group. Alterations of Metabolites Associated with Urea Cycle Participants in the urea cycle were at higher levels after hydrazine treatment when compared with 0 mg kg-1 groups, including citrulline and ornithine in urine or plasma at 24 h post-dosing (Fig. 6). Accordingly, the levels of associate metabolites, namely fumarate and urea, also increased. These results indicated up-regulation of the urea cycle. The levels of putrescine and GABA showed increases by hydrazine treatment.

Alterations of Metabolites Associated with the Tricarboxylic acid (TCA) Cycle One of the dramatic changes observed in the present study was the reduction of levels of the TCA cycle metabolites. Figure 5

529

J. App. Toxicol. 2011; 31: 524535

Copyright 2010 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jat

K. Bando et al.

Figure 4. Alterations of metabolites in the glutathione and associated metabolism pathway. Vertical axis of graphs is relative intensity (mean SD) of each metabolite peak. Total intensities of each metabolite peak in plasma were normalized by the total intensity of ribitol peak, while those in urine were normalized by the intensity of unique mass (m/z 319) of ribitol peak and creatinine concentration. C, 0 mg kg-1; L, 120 mg kg-1; H, 240 mg kg-1.

Alterations of Metabolites Associated with Lysine Metabolism Amino acid metabolisms were altered by hydrazine treatment: among them, the most remarkable changes were observed in the lysine metabolism (Fig. 7). The levels of lysine and cadaverine were increased by hydrazine treatment in urine: these changes persisted at 48 h post-dosing. Furthermore, the level of 2-aminoadipate was remarkably increased in both plasma and urine, although it was not detected in urine or plasma of 0 mg kg-1 groups. This result suggested that lysine excretion and degradation were accelerated.

DISCUSSION
We investigated the applicability of a GC-MS-based metabolomics approach for toxicological evaluation. Before identication of metabolites, all peak data of chromatograms obtained by GC-MS was included in principal component analysis, a typical unsupervised statistical method. The result showed clear group separation of spots, depending on dosages of hydrazine and sampling times after treatment. In particular, results for a group that showed most prominent toxicity in the toxicological examination, the 240 mg kg-1 group at 24 h post-dosing, were spotted in

530

wileyonlinelibrary.com/journal/jat

Copyright 2010 John Wiley & Sons, Ltd.

J. App. Toxicol. 2011; 31: 524535

GC-MS-based metabolomics of hydrazine toxicity

Figure 5. Alterations of metabolites in the TCA cycle and associated metabolism pathway. Vertical axis of graphs is relative intensity (mean SD) of each metabolite peak. Total intensities of peak of each metabolite in plasma were normalized by the total intensity of ribitol peak, while those in urine were normalized by the intensity of unique mass (m/z 319) of ribitol peak and creatinine concentration. C, 0 mg kg-1; L, 120 mg kg-1; H, 240 mg kg-1.

the farthest area from 0 mg kg-1 groups. On the contrary, results for a group that showed less toxicity (indications of recovery) in the toxicological examination, the 120 mg kg-1 group at 48 h post-dosing, were spotted in the almost same area of 0 mg kg-1 groups. GC-MS analysis data reected hydrazine-induced toxicity with extreme sensitivity and related well to toxicological parameters. The results of the present study suggested that nontargeted ngerprinting derived from chromatogram data of GC-MS has good potential for applicability in ecacy and toxicity screening, classication, or ranking. Previous metabolic proling using NMR (Bollard et al., 2005; Kleno et al., 2004; Nicholls et al., 2001; Wang et al., 2003) showed results with similar tendency, but metabolic proling by GC-MS revealed toxicity and recovery more clearly in the present study. One of the possible reasons is that GC-MS has higher sensitivity and selectivity than NMR, and that GC-MS has great advantages

for analyzing organic acids and amino acids, which dramatically changed with hydrazine treatment. The latter point is important for applicability of metabolomics to pharmacology and toxicology. In fact, we were able to nd that 2-aminoadipate and b-alanine increased, and that 2-oxoglutarate and citrate decreased in the urine using the corresponding loading plot, and thus we could understand alterations of organic and amino acid metabolism and obtain possible biomarkers that can predict hydrazine-induced toxicity. We indentied compounds by mass spectral patterns and retention times on the basis of an in-house chemical library with high accuracy, but even without an in-house library, compounds can be identied by comparison with an open mass spectral library database that can be commercially obtained, such as the National Institute of Standards and Technology GC library. Data collections for LC-MS mass spectral libraries are in progress, such

531

J. App. Toxicol. 2011; 31: 524535

Copyright 2010 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jat

K. Bando et al.

Figure 6. Alterations of metabolites in the urea cycle and associated metabolism pathway. Vertical axis of graphs is relative intensity (mean SD) of each metabolite peak. Total intensities of peak of each metabolite in plasma were normalized by the total intensity of ribitol peak, while those in urine were normalized by the intensity of the unique mass (m/z 319) of ribitol peak and creatinine concentration. C, 0 mg kg-1; L, 120 mg kg-1; H, 240 mg kg-1.

Figure 7. Alterations of lysine degradation pathway. Vertical axis of graphs is relative intensity (mean SD) of each metabolite peak. Total intensities of the peak of each metabolite in plasma were normalized by the total intensity of ribitol peak, while those in urine were normalized by the intensity of unique mass (m/z 319) of ribitol peak and creatinine concentration. C, 0 mg kg-1; L, 120 mg kg-1; H, 240 mg kg-1.

532

wileyonlinelibrary.com/journal/jat

Copyright 2010 John Wiley & Sons, Ltd.

J. App. Toxicol. 2011; 31: 524535

GC-MS-based metabolomics of hydrazine toxicity as the Mass Bank (http://www.massbank.jp/), the KNApSAcK (http://kanaya.naist.jp/KNApSAcK/) and the Human Metabolome Database (HMDB) (http://www.hmdb.ca/), while LC-MS libraries are not yet available for general use, because mass spectral data sometimes depend on instruments or analysis conditions. Therefore, in terms of abundance and availability of mass spectral library data for compound identication, compared with LC-MS, GC-MS has a greater advantage for applicability of metabolomics to the pharmacology or toxicology, which need biological interpretation and discussion of results of metabolome analysis. Hydrazine is a typical hepatotoxicity compound; however, its toxicity mechanism has not been completely elucidated. In the present study, we could discuss mechanism of hydrazineinduced toxicity based on semi-quantication data of many identied metabolites. Most notable changes were observed in the glutathione metabolism pathway. Glutathione (GSH) is a tripeptide (L-g-glutamyl-L-cysteinyl-glycine) that is synthesized in the cytosol from the precursor amino acids glutamate, cysteine and glycine. It is the most abundant redox molecule and is critical in maintaining redox status in cells. GSH plays a key role in the detoxication of reactive oxygen species, reactive nitrogen species and xenobiotic compounds in cells. In liver, depletion of GSH has been shown to induce apoptosis of hepatocytes. GSH homeostasis is an important mechanism in mediating the pathogenesis of many liver diseases (Han et al., 2006; Yuan and Kaplowitz, 2009). While glutathione was not directly measured in the plasma and urine samples in the present study (it is typically detected in cell or tissue samples), higher levels of metabolites in the glutathione biosynthetic pathway (i.e. glutamate, cysteine, and glycine) and associated metabolites were found in the present study. This indicated that this pathway was accelerated, most likely due to higher cellular demand for glutathione to combat oxidative stress. The synthesis of GSH is up-regulated during oxidative stress and inammation (Biswas and Rahman, 2009). In addition, higher levels of ascorbate were observed. Ascorbate participates in a variety of enzymatic reactions (e.g. collagen and catecholamine synthesis) as an electron donor and it is one of the most important watersoluble antioxidants of mammalian tissues (Levine and Morita, 1985; Levine, 1986; Meister, 1994; Rose and Bode, 1993; Winkler et al., 1994). The GSH and ascorbate can mutually regulate the synthesis of each other (Banhegyi et al., 1997): GSH deciency produced by the in vivo administration of buthionine sulfoximine increases ascorbate synthesis in mice (Martensson and Meister, 1992). Additionally, the glutathione-depleting agents investigated (buthionine sulfoximine, menadione, diamide and acetaminophen) increased glycogen breakdown and ascorbate production in isolated murine hepatocytes (Braun et al., 1996). Based on these ndings, oxidative stress and GSH deciency are considered to be primary modes of action of hydrazine-induced toxicity. Additionally, many TCA cycle intermediates decreased after hydrazine treatment. It is indicated that hydrazine inhibited activity of enzymes that mediate reactions in the TCA cycle, such as citrate synthase (E.C. 2.3.3.1), isocitrate dehydrogenase (E.C. 1.1.1.4.2) and succinate dehydrogenase (E.C. 1.3.5.1). Ojima et al. reported that the activity of citrate synthase and mRNA expressions of the enzymes in TCA cycle were repressed by about half under oxidative stress conditions, as compared with those under normal conditions (Ojima et al., 2008). It is considered that hydrazine-induced-oxidative stress caused down-regulation of the TCA cycle. In the present study, plasma pyruvate and lactate levels also increased. Lactate is produced through anaerobic glycolysis in the liver tissue. The buildup of lactate in cells can be caused by lactate overproduction or underutilization (Luft, 2001). Overproduction of lactate, also termed type A lactic acidosis, occurs when the body must regenerate ATP without oxygen (tissue hypoxia). Underutilization involves removal of lactate by oxidation or conversion to glucose. Liver disease, inhibition of gluconeogenesis, pyruvate dehydrogenase deciency and uncoupling of oxidative phosphorilation are the most common causes of underutilization. Ojima et al. also suggested that the oxidativestress-suering cells switch the metabolic pathway into a suppressed aerobiosis, possibly for lowering the generation of reactive oxygen species (Ojima et al., 2008). Therefore, downregulation of the TCA cycle may lead to underutilization of pyruvate and lactate. Thus, general down-regulation of energy metabolism might lead to alterations of sugar or lipid metabolism, which might cause histopathological changes (i.e. fatty degeneration and glycogen accumulation) in the liver. Increase of urinary glucose was observed in the hydrazine treatment group, although glucose levels in the plasma showed little change. Under normal conditions, urinary glucose is not detected, because glomerular ltrated glucose is reabsorbed in the renal tubule in the kidney (Guyton and Hall, 2006). In addition, the level of BUN, a conventional biomarker of kidney function, increased at 24 h post-dosing in blood biochemistry. Therefore, elevation of urinary glucose levels suggested kidney malfunction, rather than TCA cycle alterations. Furthermore, up-regulation of the urea cycle was observed in this study. The urea cycle is a cycle of biochemical reactions that produces urea from ammonia in the liver. Biosynthesis of amino acids was up-regulated by hydrazine treatment, glutathione metabolism, and lysine metabolism, so the urea cycle might be expected to be up-regulated in order to remove excess ammonia derived from elevated amino acids. In addition, the amino groups of amino acids that have been used as metabolic fuel are converted into urea through the urea cycle, which occurs mainly in the liver. The levels of putrescine showed increases by hydrazine treatment. It is plausible that the putrescine elevation is accompanied by up-regulation of urea cycle. On the other hand, previous studies showed that putrescine exhibits a protective eect against acute liver injury caused by other hepatotoxins such as carbon tetrachloride, D-galactosamine and cadmium (Nagoshi et al., 1994; Nishiguchi et al., 1990; Tzirogiannis et al., 2004). Polyamines have been reported to inhibit lipid peroxidation and oxidative damage caused by ferric iron and free radicals by acting as scavengers, to stabilize biological membranes and to modulate the production of secondary messengers and inammatory mediators as well as the activity of enzymes and ionic carriers (Heby, 1981; Schuber, 1989; Tadolini et al., 1984). Thus, another explanation for elevated putrescine may be an adaptive cellular response for damaged hepatocytes. Hydrazine treatment had very dramatic impacts on amino acids and amino acid metabolism; in particular, lysine metabolism was remarkably aected. The levels of 2-aminoadipate increased in both urine and plasma, as previously reported in NMR-based metabolomics studies (Holmes et al., 2000; Nicholls et al., 2001; Perry et al., 1981). It is likely that hydrazine inhibits activity of 2-aminoadipate aminotransferase (EC 2.6.1.39), which is responsible for the catalysis of the reversible transamination of 2-aminoadipate and 2-oxoglutarate to form 2-oxoadipate and

533

J. App. Toxicol. 2011; 31: 524535

Copyright 2010 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jat

K. Bando et al. glutamate, because hydrazine reduces the activity of aminotransferases by removal cofactors, as previously described. Elevation of 2-aminoadipate may be related to hydrazine-induced neurotoxicity, because 2-aminoadipate causes apoptosis of primary astrocytes in the brain or interferes with neurotransmitters (Chang et al., 1997; Haugstad and Langmoen, 1997; Nishimura et al., 2000).
Foyer CH. 2001. Prospects for enhancement of the soluble antioxidants, ascorbate and glutathione. Biofactors 15(24): 7578. Garrod S, Bollard ME, Nicholls AW, Connor SC, Connelly J, Nicholson JK, Holmes E. 2005. Integrated metabonomic analysis of the multiorgan eects of hydrazine toxicity in the rat. Chem. Res. Toxicol. 18(2): 115 122. Glinski M, Weckwerth W. 2006. The role of mass spectrometry in plant systems biology. Mass Spectrom. Rev. 25(2): 173214. Goldsmith P, Fenton H, Morris-Sti G, Ahmad N, Fisher J, Prasad KR. 2010. Metabonomics: a useful tool for the future surgeon. J. Surg. Res. 160(1): 122132. Guyton AC, Hall JE. 2006. Textbook of Medical Physiology. Elsevier/ Saunders: Philadelphia, PA. Han D, Hanawa N, Saberi B, Kaplowitz N. 2006. Mechanisms of liver injury. III. Role of glutathione redox status in liver injury. Am. J. Physiol. Gastrointest. Liver Physiol. 291(1): G17. Haugstad TS, Langmoen IA. 1997. L-Alpha-aminoadipate reduces glutamate release from brain tissue exposed to combined oxygen and glucose deprivation. J. Cereb. Blood Flow Metab. 17(5): 567570. Heby O. 1981. Role of polyamines in the control of cell proliferation and dierentiation. Dierentiation 19(1): 120. Holmes E, Nicholls AW, Lindon JC, Connor SC, Connelly JC, Haselden JN, Damment SJ, Spraul M, Neidig P, Nicholson JK. 2000. Chemometric models for toxicity classication based on NMR spectra of biouids. Chem. Res. Toxicol. 13(6): 471478. IARC. 1999. Re-evaluation of Some Organic Chemicals, Hydrazine and Hydrogen Peroxide. World Health Organization: Lyon. Iguchi S, Goromaru T, Noda A, Matsuyama K, Sogabe K. 1977. Quantitative determination of hydrazines derived from isoniazid in man. Chem. Pharm. Bull. (Tokyo) 25(10): 27962800. Kleno TG, Kiehr B, Baunsgaard D, Sidelmann UG. 2004. Combination of omics data to investigate the mechanism(s) of hydrazine-induced hepatotoxicity in rats and to identify potential biomarkers. Biomarkers 9(2): 116138. Lee MS, Jung BH, Chung BC, Cho SH, Kim KY, Kwon OS, Nugraha B, Lee YJ. 2009. Metabolomics study with gas chromatographymass spectrometry for predicting valproic acid-induced hepatotoxicity and discovery of novel biomarkers in rat urine. Int. J. Toxicol. 28(5): 392404. Lee SH, Woo HM, Jung BH, Lee J, Kwon OS, Pyo HS, Choi MH, Chung BC. 2007. Metabolomic approach to evaluate the toxicological eects of nonylphenol with rat urine. Anal. Chem 79(16): 61026110. Levine M. 1986. New concepts in the biology and biochemistry of ascorbic acid. New Engl. J. Med. 314(14): 892902. Levine M, Morita K. 1985. Ascorbic acid in endocrine systems. Vitam. Horm. 42: 164. Lightcap ES, Silverman RB. 1996. Slow-binding inhibition of gammaaminobutyric acid aminotransferase by hydrazine analogues. J. Med. Chem. 39(3): 686694. Lindon JC, Holmes E, Bollard ME, Stanley EG, Nicholson JK. 2004. Metabonomics technologies and their applications in physiological monitoring, drug safety assessment and disease diagnosis. Biomarkers 9(1): 131. Lindon JC, Holmes E, Nicholson JK. 2007. Metabonomics in pharmaceutical R&D. FEBS J. 274(5): 11401151. Lu SC. 2009. Regulation of glutathione synthesis. Mol. Asp. Med. 30(12): 4259. Luft FC. 2001. Lactic acidosis update for critical care clinicians. J. Am. Soc. Nephrol. 12(suppl. 17): S1519. Martensson J, Meister A. 1992. Glutathione deciency increases hepatic ascorbic acid synthesis in adult mice. Proc. Natl Acad. Sci. USA 89(23): 1156611568. Meister A. 1994. Glutathione-ascorbic acid antioxidant system in animals. J. Biol. Chem. 269(13): 93979400. Metges CC, Yu YM, Cai W, Lu XM, Wong S, Ajami AM, Young VR. 1999. Plasma L-5-oxoproline carbon and nitrogen kinetics in healthy young adults. J. Nutr. 129(11): 19982004. Metges CC, Yu YM, Cai W, Lu XM, Wong S, Regan MM, Ajami A, Young VR. 2000. Oxoproline kinetics and oxoproline urinary excretion during glycine- or sulfur amino acid-free diets in humans. Am. J. Physiol. Endocrinol. Metab. 278(5): E868876. Moloney SJ, Prough RA. 1983. Biochemical Toxicology of Hydrazine, Hodgson E, Philpot RM (eds). Elsevier Science: New York; 313. Nagoshi S, Ohta Y, Matsui A, Fujiwara K. 1994. Protective action of putrescine against rat liver injury. Scand. J. Gastroenterol. 29(2): 166 171.

CONCLUSION
We demonstrated that a GC-MS-based metabolomics approach could reveal alterations of metabolic proles, which related well to dose-dependent toxicity and recovery induced by hydrazine treatment. Additionally, many identied metabolites could be discussed in terms of the biochemical metabolic pathway based on the alterations, and from this we found that oxidative stress plays an important role in the etiology of hydrazine-induced hepatotoxicity. Many other biochemical pathways, such as energy metabolism and amino acid metabolism, can be associated with liver malfunction, and thus we can simultaneously overviewed living organisms through large amounts of detailed information about biochemical metabolism. The GC-MS-based metabolomic approach will be a useful tool for pharmacology and toxicology, in screening, elucidating modes of action, and biomarker discovery. Acknowledgment The authors would like to thank K. Horii, S. Mikami and E. Saijo of Dainippon Sumitomo Pharma. Co. Ltd (Japan), and M. Oyanagi of Sumika Technoservice Co. Ltd (Japan) for conducting animal experiments.

REFERENCES
Banhegyi G, Braun L, Csala M, Puskas F, Mandl J. 1997. Ascorbate metabolism and its regulation in animals. Free Radic. Biol. Med. 23(5): 793 803. Bino RJ, Hall RD, Fiehn O, Kopka J, Saito K, Draper J, Nikolau BJ, Mendes P, Roessner-Tunali U, Beale MH and others. 2004. Potential of metabolomics as a functional genomics tool. Trends Plant. Sci. 9(9): 418425. Biswas SK, Rahman I. 2009. Environmental toxicity, redox signaling and lung inammation: the role of glutathione. Mol. Asp. Med. 30(12): 6076. Blair IA, Mansilla Tinoco R, Brodie MJ, Clare RA, Dollery CT, Timbrell JA, Beever IA. 1985. Plasma hydrazine concentrations in man after isoniazid and hydralazine administration. Hum. Toxicol. 4(2): 195202. Bollard ME, Keun HC, Beckonert O, Ebbels TM, Antti H, Nicholls AW, Shockcor JP, Cantor GH, Stevens G, Lindon JC and others. 2005. Comparative metabonomics of dierential hydrazine toxicity in the rat and mouse. Toxicol. Appl. Pharmacol. 204(2): 135151. Braun L, Csala M, Poussu A, Garzo T, Mandl J, Banhegyi G. 1996. Glutathione depletion induces glycogenolysis dependent ascorbate synthesis in isolated murine hepatocytes. FEBS Lett. 388(23): 173176. Chace DH. 2001. Mass spectrometry in the clinical laboratory. Chem. Rev. 101(2): 445477. Chang YF, Cauley RK, Chang JD, Rao VV. 1997. L-Alpha-aminoadipate inhibits kynurenate synthesis in rat brain hippocampus and tissue culture. Neurochem. Res. 22(7): 825829. Clarke CJ, Haselden JN. 2008. Metabolic proling as a tool for understanding mechanisms of toxicity. Toxicol. Pathol. 36(1): 140147. Cornish HH. 1969. The role of vitamin B6 in the toxicity of hydrazines. Ann. NY Acad. Sci. 166(1): 136145. Fiehn O, Kopka J, Dormann P, Altmann T, Trethewey RN, Willmitzer L. 2000. Metabolite proling for plant functional genomics. Nat. Biotechnol. 18(11): 11571161.

534

wileyonlinelibrary.com/journal/jat

Copyright 2010 John Wiley & Sons, Ltd.

J. App. Toxicol. 2011; 31: 524535

GC-MS-based metabolomics of hydrazine toxicity

Nicholls AW, Holmes E, Lindon JC, Shockcor JP, Farrant RD, Haselden JN, Damment SJ, Watereld CJ, Nicholson JK. 2001. Metabonomic investigations into hydrazine toxicity in the rat. Chem. Res. Toxicol. 14(8): 975987. Nishiguchi S, Kuroki T, Takeda T, Nakajima S, Shiomi S, Seki S, MatsuiYuasa I, Otani S, Kobayashi K. 1990. Eects of putrescine on D-galactosamine-induced acute liver failure in rats. Hepatology 12(2): 348353. Nishimura RN, Santos D, Fu ST, Dwyer BE. 2000. Induction of cell death by L-alpha-aminoadipic acid exposure in cultured rat astrocytes: relationship to protein synthesis. Neurotoxicology 21(3): 313320. Ojima Y, Nishioka M, Taya M. 2008. Metabolic alternations in SODdecient Escherichia coli cells when cultivated under oxidative stress from photoexcited titanium dioxide. Biotechnol. Lett. 30(6): 1107 1113. ONeil MJ. 2006. The Merck Index: an Encyclopedia of Chemicals, Drugs, and Biologicals. Merck: Whitehouse Station, NJ. Pasikanti KK, Ho PC, Chan EC. 2008. Gas chromatography/mass spectrometry in metabolic proling of biological uids. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 871(2): 202211. Perry TL, Kish SJ, Hansen S, Wright JM, Wall RA, Dunn WL, Bellward GD. 1981. Elevation of brain GABA content by chronic low-dosage administration of hydrazine, a metabolite of isoniazid. J. Neurochem. 37(1): 3239. Powers R. 2009. NMR metabolomics and drug discovery. Magn. Reson. Chem. 47(suppl. 1): S211. Rose RC, Bode AM. 1993. Biology of free radical scavengers: an evaluation of ascorbate. FASEB J. 7(12): 11351142. Scales MD, Timbrell JA. 1982. Studies on hydrazine hepatotoxicity. 1. Pathological ndings. J. Toxicol. Environ. Health 10(6): 941953. Schuber F. 1989. Inuence of polyamines on membrane functions. Biochem. J. 260(1): 110. Serkova NJ, Niemann CU. 2006. Pattern recognition and biomarker validation using quantitative 1H-NMR-based metabolomics. Expert. Rev. Mol. Diagn. 6(5): 717731. Sies H. 1997. Oxidative stress: oxidants and antioxidants. Exp. Physiol. 82(2): 291295. Tadolini B, Cabrini L, Landi L, Varani E, Pasquali P. 1984. Polyamine binding to phospholipid vesicles and inhibition of lipid peroxidation. Biochem. Biophys. Res. Commun. 122(2): 550555. Timbrell JA, Harland SJ. 1979. Identication and quantitation of hydrazine in the urine of patients treated with hydralazine. Clin. Pharmacol. Ther. 26(1): 8188. Timbrell JA, Scales MD, Streeter AJ. 1982. Studies on hydrazine hepatotoxicity. 2. Biochemical ndings. J. Toxicol. Environ. Health 10(6): 955 968. Toth B. 1993. Teratogenic hydrazines: a review. In Vivo 7(1): 101 110. Tzirogiannis KN, Panoutsopoulos GI, Demonakou MD, Papadimas GK, Kondyli VG, Kourentzi KT, Hereti RI, Mykoniatis MG. 2004. The hepatoprotective eect of putrescine against cadmium-induced acute liver injury. Arch. Toxicol. 78(6): 321329. Wang Y, Bollard ME, Keun H, Antti H, Beckonert O, Ebbels TM, Lindon JC, Holmes E, Tang H, Nicholson JK. 2003. Spectral editing and pattern recognition methods applied to high-resolution magic-angle spinning 1H nuclear magnetic resonance spectroscopy of liver tissues. Anal. Biochem. 323(1): 2632. Watereld CJ, Turton JA, Scales MD, Timbrell JA. 1993. Investigations into the eects of various hepatotoxic compounds on urinary and liver taurine levels in rats. Arch. Toxicol. 67(4): 244254. Weckwerth W, Morgenthal K. 2005. Metabolomics: from pattern recognition to biological interpretation. Drug Discov. Today 10(22): 1551 1558. Winkler BS, Orselli SM, Rex TS. 1994. The redox couple between glutathione and ascorbic acid: a chemical and physiological perspective. Free Radic. Biol. Med. 17(4): 333349. Yu YM, Ryan CM, Fei ZW, Lu XM, Castillo L, Schultz JT, Tompkins RG, Young VR. 2002. Plasma L-5-oxoproline kinetics and whole blood glutathione synthesis rates in severely burned adult humans. Am. J. Physiol. Endocrinol. Metab. 282(2): E247258. Yuan L, Kaplowitz N. 2009. Glutathione in liver diseases and hepatotoxicity. Mol. Asp. Med. 30(12): 2941.

535

J. App. Toxicol. 2011; 31: 524535

Copyright 2010 John Wiley & Sons, Ltd.

wileyonlinelibrary.com/journal/jat