Metabonomic Study of Aristolochic Acid-Induced Nephrotoxicity in Rats Minjun Chen,†,‡ Mingming Su,† Liping Zhao,§ Jian Jiang,‡ Ping Liu,‡ Jiye Cheng,† Yijiang Lai,| Yumin Liu,| and Wei Jia*,†,‡ School of Pharmacy, Shanghai Jiao Tong University, Shanghai 200030, China, Shanghai University of Traditional Chinese Medicine, Shanghai 201203, China, Shanghai Institute for Systems Biology, Shanghai 200030, China, Center for Instrumental Analysis, Shanghai Jiao Tong University, Shanghai 200030, China Received November 15, 2005
Abstract: This paper describes a metabonomic study characterizing the nephrotoxicity induced by aristolochic acid (AA), a suspected kidney toxicant. For these studies, we examined the biochemical compositions of AA-treated rat urine using LC-MS and pattern recognition methods. The biochemical and histological patterns of rat groups treated with different AA sources showed distinct differences from those of the control group. Certain metabolic pathways, such as homocysteine formation and the folate cycle were significantly accelerated, while others, including arachidonic acid biosynthesis, were decreased. A subset-validation procedure using linear discriminant analysis (LDA) and selected predictive variables indicated that approximately 95% of the treated and nontreated rat urine samples were classified correctly into their respective treatment groups. The results suggested that this metabonomic approach is a promising methodology for the rapid in vivo screening of nephrotoxicity associated with ingesting multi-ingredient medicinal herb supplements, and provides a valid method for comprehending the chemical-induced perturbations in the metabolic network and the networked lesions. Keywords: metabonomics • LC-MS • nephrotoxicity • aristolochic acid • Caulis Aristolochiae manshuriensis • traditional Chinese medicine
Introduction Metabonomics is a new platform for studying systems biology leading to high throughput screening processes in the pharmaceutical industry and clinical diagnosis.1-3 To date, the use of high-field Nuclear Magnetic Resonance (NMR) in metabonomic studies has been well-documented and extensively developed to provide a more efficient assessment of toxicity, possible disease onset mechanisms and gut microflora alterations.4-8 Recently, electrospray ionization mass * To whom correspondence should be addressed. Tel: 86-21-6293-2292. Fax: 86-21-6294-5529. E-mail:
[email protected]. † School of Pharmacy, Shanghai Jiao Tong University. ‡ Shanghai University of Traditional Chinese Medicine. § Shanghai Institute for Systems Biology. | Center for Instrumental Analysis, Shanghai Jiao Tong University. 10.1021/pr050404w CCC: $33.50
2006 American Chemical Society
spectrometry coupled with high-performance liquid chromatography (HPLC-ESI-MS), either alone or combined with other analytical devices, such as NMR or GC-MS, can detect and profile endogenous metabolites in biofluids.9-15 These hyphenated LC-MS or GC-MS approaches are likely viable analytical techniques for metabonomic analysis, owing to the high chromatographic resolution, sensitivity and reproducibility. Additionally, classical multivariate statistical analysis, including principal components analysis (PCA), partial leastsquares (PLS), artificial neural networks (ANN) and ANOVAsimultaneous component analysis (ASCA), can suggest or identify disease biomarkers.16-19 Aristolochic acid (AA) is a chemical compound present in plants of the Aristolochia, Bragantia, and Asarum species. Since a number of Belgian women who had followed a slimming regimen with some medicinal herbs containing AA developed acute renal failure (ARF), AA has been under great scrutiny and suspicion as being responsible for ARF associated with interstitial fibrosis, hyperproteinemia, severe anaemia, uremia and carcinoma.20-22 Additionally, some AA-containing herbs, such as Caulis Aristolochiae manshuriensis (CAM, Figure 1), reportedly induce similar ARF effects at high doses.23,24 Some studies indicated that AA-induced nephrotoxicity is distinct from that induced by other toxicological substances, including heavy metals.24 Furthermore, AA likely induces end-stage renal disease (ESRD) characterized by hyperhomocysteinemia in exposed patients.24-26 The nephrotoxicity of AA remains in dispute, and may occur through multiple mechanisms, such as direct cytotoxicity or oncogene activation and hypersensitivity, the latter of which can be confirmed by detecting AA-DNA adducts (7-(deoxyadenosin-N6-yl) aristolactam I or II) in renal biopsies.23-25 As a result, AA and AA-containing herbs have come under pharmacological scrutiny and concern (Figure 1).30-33 The purpose of the study is to investigate the use of a metabonomic method to identify the characteristic metabolic profile associated with AA-induced nephrotoxicity. In the present work, we used LC-MS analysis combined with pattern recognition techniques to identify biomarkers of ARF. The dynamic disease course, including toxic onset, progression, and recovery of kidney lesions was observed using LC-MS analysis of rat urine samples at these different phases, and PCA for differentiating the healthy and the AA-dosed rats. Additionally, we used linear discriminant analysis (LDA), for predicting the Journal of Proteome Research 2006, 5, 995-1002
995
Published on Web 03/15/2006
technical notes
Aristolochic Acid-Induced Nephrotoxicity in Rats
Figure 1. Photos of the material medica of Aristolochiae manshuriensis.
potential toxicology of AA-induced ARF based on the metabolic profiling. We used the urine samples from rats treated with CAM for validating the model.
Experimental Section Chemicals and Material Process. Methanol (HPLC grade) was purchased from Sigma-Aldrich (USA). Aristolochic acid authentic standard was obtained from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). Water was prepared from a Milli-Q system (Millipore, USA). CAM was provided by Shanghai Tong Han Chun Tang Chinese Herbal Medicine Factory (Shanghai, China), and was authenticated by Dr. Mengyue Wang, Associate Professor in pharmacognostic research at School of Pharmacy, Shanghai Jiao Tong University. A 600-g sample of dried and pulverized plant materials (CAM) was refluxed with 2000 mL of water for 60 min twice. After cooling, the solution was filtered through a glass filter covered with filter paper. The supernatant liquid was washed with 200 mL of water. The solution was evaporated under vacuum to roughly 150 mL, and then diluted to 200 mL with water in a volumetric flask. The extraction for experiment use was condensed to 3 g CAM/mL, containing 0.62% of AA determined by HPLC. Animal Handling and Sample Collection. Fifteen male Wistar rats, weighing 150 ( 10 g, were purchased from Shanghai SLAC Laboratory Animal Co. Ltd. (Shanghai, China), and were housed separately for approximately 2 weeks in stainless steel wire-mesh cages. They were fed with a certified standard diet and tap water ad libitum. Temperature and humidity were regulated at 22 ( 1 °C and 45 ( 15%, respectively. A light cycle of 12 h on/12 off was established. After 2 weeks of acclimatization, the rats were randomly divided into three groups (n ) 5/group) as follows: AA Group (AAG), oral gavage with AA authentic standard at a single dose of 50 mg/ kg on day 0; Healthy Control Group (HCG), oral gavage with the same volume of 0.9% saline solution as AAG; CAM Group 996
Journal of Proteome Research • Vol. 5, No. 4, 2006
(CAMG), oral gavage with water extraction of CAM at a dose of 30 g/kg/day (an equivalent dose of AA 96 mg/kg/day) from days 0 to 3. After dosing, the rats were allowed to eat for 1 h before they were put into the metabolic cages. The urine samples were collected at various time intervals (-12∼0 h pre-dose, 1∼12 h, 12∼24 h, 48∼60 h, 84∼96 h, and 132∼144 h post-dose), during which the rats were deprived of food to avoid solid debris pollution, but were allowed tap water ad libidum. The fresh urine samples were immediately centrifuged at 3500 rpm for 10 min at room temperature, to remove particle contaminants, and the supernatants were stored at -70 °C pending LC-MS analysis. The urine volumes and body weights for each rat were recorded daily during the entire experiment course. Blood samples were collected 3 d before and 7 d after the administration of AA or CAM to test for renal function. These samples were obtained from the retro-orbital venous plexus under light diethyl ether anaesthesia. Plasma creatinine and urea concentrations (enzymatic UV test) were determined. Histopathology. At experiment termination, the rats were euthanized. Each kidney was fixed in 10% formalin for 12 h and embedded in paraffin wax. Four-micrometer histologic sections of the paraffin-embedded tissues were stained with hematoxylin-eosin and then prepared for light microscopy. LC-MS Analysis. Urinary samples were prepared by centrifugation at 16 000 rpm for 15 min, and the supernatant was transferred into a 2 mL auto-sampler vial equipped with a conical low-volume insert for analysis. In a typical experiment, a 20 µL aliquot of urine was injected into a 4.6 mm × 15 cm Extend-C18 5µm column, using an Agilent 1100 Series for LCMS (Agilent, USA). The column was maintained at 40 °C and eluted with a linear gradient of 30-100% B at a flow rate of 800 µL/min (where A ) water and B ) methanol) for 0-12 min. After holding the solvent content to 100% methanol for 3 min, the column was returned to its starting condition. The column elution was split so that approximately 250 µL/min elute was introduced to the ESI-MS.
technical notes
Chen et al.
Figure 2. Flowchart of metabonomic analysis used in the study.
Mass spectra were obtained on a full-scan operation in positive ion mode. The capillary voltage was set at 3.0 kV, and the cone voltage was optimized at 30 V, respectively. The source temperature of 110 °C, a desolvation gas temperature of 350 °C, and a nebulization gas flow of 9.0 L/min were used. Data profiling of positive ions from m/z 100 to 500 were recorded at a speed of 1s/scan during analysis. The tune mixture solution (Agilent, USA) was employed as the lockmass (m/z ) 118.09, 622.05, or 922.02) at a flow rate of 30 µL/min, via a lock spray interface for accurate mass measurement. Peak Resolution and Alignment. The typical total ion current (TIC) chromatograms were unsuitable for pattern recognition, due to overlapping peak profiles, although significant visual differences were observed among the nondosed and dosed rats. The overall data analysis process is illustrated in the flowchart (Figure 2). Prior to peak resolving or alignment, the LC-MS data, which comprised a two-way matrix (retention time × mass-to-charge ratio) for each sample, were exported and stored as AIA (Analytical Instrumental Association) format via LC-MSD ChemStation (Rev.A.09.03, Agilent, USA). Each file was extracted subsequently by MATLAB 7.0 software (The MathWorks, Inc., USA). The resulting output data were stored in a two-dimensional matrix, and included retention time in one direction and mass-to-charge ratio (m/z) in another direction. Chromatograms extracted from each m/z channel (100∼500) were obtained in a sequence, and the peaks were detected
using a classical algorithm based on first and second derivatives.34 In general, peaks with signal-to-noise (S/N) values lower than 10 were rejected to avoid the disturbance of noise. Herein, the peak intensity at 2000 was set as the ‘noise threshold level’. After processing the raw LC/MS data matrix of a sample, a list of the peaks combined from all m/z channels (100∼500) associated with their corresponding retention times was summarized, and the intensity of each peak was recorded. To align the peaks of each sample, the peak list of a typical sample from HC group was selected as a contrast one, and the data from different samples were combined into a single matrix with the same mass/retention time pair together, along with their associated intensities. The obtained matrix was then employed for pattern recognition methods. Pattern Recognition. Pattern recognition methods such as PCA and LDA were employed herein to uncover the biochemical pattern of rat urine induced by AA, and to suggest variables that may be useful biomarkers of toxicity. Prior to performing PCA or LDA, additional data preprocessing procedures were required. The log -transformed data were used to eliminate the disturbances of high concentration components and the heteroscedasticity of noise structure.35 Since the responses of some variable (pair of m/z and retention) were zero, a default constant (half of the least response) was filled up before the data transformation. The PCA score plots were generated from data for different days and between groups (HCG and AAG). In addition, the LDA analysis (SPSS 13.0 for windows, The SPSS Journal of Proteome Research • Vol. 5, No. 4, 2006 997
technical notes
Aristolochic Acid-Induced Nephrotoxicity in Rats
Univariate ANOVAs, and Wilks’ lambda from SPSS software packages were chosen to evaluate the discriminant model obtained under the stepwise method. Box’s M is used to test the assumption of equality of covariances across groups. Wilks’ lambda is used to measure how well each discriminant function separates all the cases into different groups.
Results and Discussion
Figure 3. Body weight comparison between pre-dose and endpoint.
Inc.) was used to construct models for the classification between HCG and AAG, and to make independent predictions of renal failure caused by CAM. For this reason, about 70% samples were randomly selected out of the 30 urinary samples obtained from 48 h post-dose within the HC and AA groups as a training set. The predicted set, combination with 14 samples from CAM group and 30% remained samples from HC and AA groups, was used to validate the model results. Box’s M,
General Information about the Animal Experiment. As indicated in Figure 3, the body weights increased slightly among the HCG rats as anticipated, whereas significant body weight decreases were observed for the AA- and CAM-treated rats. Compared to the AA group, the CAM members lost more body weight, as shown in Figure 3, which might be due to the higher AA dose. The volumes of rat urine in the AA and CAM groups decreased rapidly on day 3, and were associated with a darkened urine color, which returned to normal on day 5. On day 7, after administering AA or CAM, we noted that the plasma creatinine and urea levels in both groups were distinctly higher than those in the control group (Table 1). Compared with the AA group, the CAM rats exhibited slightly higher blood plasma creatinine and urea levels. The histological examinations of rats treated with AA (Figure 4B) or CAM (Figure 4C) showed evidence of nephrosis. Compared with those treated with AA, the CAM-treated rats were more severely affected (Figure 4C), indicating the higher toxic potential of CAM. Extensive tubular necrosis occurred across
Figure 4. Photomicrographs of renal cortex of rats on day 7 after administration. (A) for a control rat, no abnormalities were noted. HE stain. Magnification, ×100; (B) for a high-dose AA-treated (50 mg/kg body wt) rat and (C) for a high-dose Caulis Aristolochiae manshuriensis-treated (30 g/kg, an equivalent dose of AA 96 mg/kg) rat, necrosis of the proximal tubules were observed. HE stain. Magnification, ×100; 998
Journal of Proteome Research • Vol. 5, No. 4, 2006
technical notes
Chen et al.
Figure 5. TIC chromatograms of urine samples obtained from (A) a healthy control rat, (B) a AA treated rat at 132∼144h post-dose, and (C) a CAM treated rat at 132∼144 h post-dose. Table 1. Clinical Chemistry Parameters Obtained from AA-treated Rats, CAM-treated Rats and Controlsa blood chemistry
day
HCG
AAG
CAMG
Creatinine (µmol/L) Urea (mmol/L)
-3 7 -3 7
63.2 ( 5.43 61.3 ( 4.33 8.6 ( 0.87 7.9 ( 0.63
64.5 ( 3.05 73.4 ( 7.17b 8.9 ( 0.72 11.3 ( 2.51c
60.4 ( 4.29 97.0 ( 25.43c 9.1 ( 0.84 12.8 ( 3.36c
a
Means ( standard deviation, n ) 5. b p < 0.05. c p < 0.01)
the cortex, and a significant quantity of epithelial cell debris was exfoliated and gathered within the renal tubular. These abnormalities, that are distinct from those in HCG, corresponded with the rapidly progressive renal failure induced by AA or CAM reported in the literature.20,23 LC/MS Fingerprinting of Rat Urine Since urine contains thousands of endogenous metabolites, including amino acids, fatty acids, sugars, lipids, hormones, and sulfate conjuncts, there are no universal analytical techniques to analyze these compounds simultaneously. In this study, however, the LCESI-MS detected numerous metabolites of AA in the urine extracts. Preliminary experiments proved that methanol was more appropriate than acetonitrile for our experiment. The elution time was reduced to 15 min, due to the fact that few
compounds with reasonable peak intensities were obtained after a 15 min retention time. More quantitative information was obtained from the positive ion mode than that collected under the negative ion mode. Therefore, considering the sensitivity on the single mode, full-scan detection was set as positive ion mode. Cone voltage was also optimized, so that molecular ions [M+H]+ accounted for the majority of the mass spectrum. As shown in Figure 5, there were significant visual differences in the TIC, especially from 6 to 10 min among the groups. The difference between the HCG (Figure 5A) and administration groups was more salient than that between the two dosed groups, indicating that some endogenous metabolic pathways were greatly perturbed by AA or CAM (Figure 5, parts B and C). Simultaneously, the chromatographic profiles of the AA and CAM groups were comparable, which suggested that the biochemical patterns induced by these two AA sources (Figure 5, parts B and C) are similar and unique. Multivariate Analysis. Approximately 1000 peaks (defined by a pair of m/z value and RT) were resolved for each urine sample using the above method. Reportedly, the information involved is useful and important to explain the pathological Journal of Proteome Research • Vol. 5, No. 4, 2006 999
technical notes
Aristolochic Acid-Induced Nephrotoxicity in Rats
Figure 6. (A) Scores plot of PCA performed on all the AAG samples. Pre-dose samples are separated from the dosed rats collected from 0 to 24 h and after 48 h, respectively. (B) A mean-centered PCA scores plot obtained from AAG samples. Table 2. Results of Linear Discriminant Analysis predicted group membership group
validation
healthy nephrotoxic total
training set HCG cross-validated 11/11 prediction set AAG subset-validation 0/0 HCG 4/4 AAG 0/0 CAMG 1/0
Figure 7. 3-D PCA scores plot obtained from rat urine samples collected after 48 h post-dose within healthy control group (X) and AA group (O).
changes and seek the potential biomarkers, although not all the peaks detected here could be identified.35 The PCA score plot of AAG (Figure 6A) could be readily divided into three clusters: -12∼0 h (pre-dose), 1∼24 h (postdose), and after 48 h (post-dose), suggesting that AA-induced renal morbidity seemed to progress in a stepwise fashion. A mean-centered PCA score plot (Figure 6B) was generated in the study, providing a time-dependent tendency of nephrotoxicity. On the basis of the above results, the three-dimensional PCA score plots were obtained from these samples of HCG and AAG collected at the time intervals of 48-60 h, 84-96 h and 132144 h. It appeared that the distance between the two groups (Figure 7) was distinct. One of the AAG samples collected at 48-60h from rat 2, was inappropriately clustered into the HCG group. The daily records showed that the body weight of this rat at that time period was significantly higher than those of the other rats in the AA group, suggesting that this rat might be somewhat resistant to developing AA nephrotoxicity. Additionally, the HCG samples appeared to concentrate into a “normal zone” with less deviation, whereas the AA-dosed samples were more scattered, reflecting the individual com1000
Journal of Proteome Research • Vol. 5, No. 4, 2006
0/0 12/12 0/0 3/3 13/14
23 21
plexity differences of the biochemical responses to the metabolic network perturbation in these animals. Independent T tests (P < 0.05) were used to select suitable variables for the LDA analyses, thus reducing the number of peaks to 82 unique variables. On the basis of 23 random selected samples (or cases) in the training set, the LDA employed a stepwise method and offered six significant latent components, the retention-m/z pairs of which were (12.69, 461), (2.66, 102), (12.77, 417), (3.96, 266), (12.41, 318), and (2.74, 136). From the LDA results (Table 2), all training set members were correctly classified using a cross-validation procedure. Furthermore, the subset validation procedure was accomplished by classifying the prediction group consisting of the cases unselected to model. Approximately 95% of the cases from the HCG and AAG groups were classified correctly. The canonical correlation parameter and the Wilks’ lambda value are 0.986 and 0.029, respectively, suggesting that the discriminant model fits the data well. Therefore, this method can generate reliable predictions with robust accuracy, and enables one to screen for potential nephrotoxicity induced by AAcontaining herbs. Discussion. The metabonomic analysis conducted in the present study reveals AA-induced changes in the ratio profiles of certain endogenous metabolites. Moreover, the histopathology and plasma chemistry analyses confirmed the presence of substantial kidney damage after administration of high AA doses. In Table 3, the AA and CAM groups display a similar biochemical trend. Some up-regulated metabolites induced by AA include L-leucine, creatine, creatinine, D-serine, homocysteine (Hcy), L-aspartyl-P, adenosine, 5-CH3-tetrahydrofolate (5-CH3-THF) and L-arginine. The metabolites that were downregulated by AA include valeric acid, caprylic acid, arachidonic acid, L-methionine, and carbamoyl phosphate. With the excep-
technical notes
Chen et al.
Table 3. Some Endogenous Urinary Metabolites of Rats Perturbed at the Time Interval of 132∼144 h Post-Dose as Measured by LC-MSa retention time
measured mass/Da
actual mass/Da
elemental composition of [M+H]+
postulated identity
HCG
AAG
CAMG
2.20 2.77 2.18 2.74 6.99 3.96 5.24 12.14 2.66 2.37 6.56 3.37 4.82
132 114 106 136 214 267 461 175 102 145 305 150 142
132.2 114.1 106.2 136.2 214.2 267.1 460.7 175.2 102.1 145.2 305.5 150.2 142.2
C6H15NO2 C4H8N3O C3H8NO3 C4H10NO2S C4H9NO7P C10H13N5O4 C20H26N7O6 C6H15N4O2 C5H10O2 C8H17O2 C20H33O2 C5H12NO2S CH5NO5P
L-Leucine Creatinine D-Serine Homocysteine L-aspartyl-P Adenosine 5-CH3-THFd L-Arginine Valeric acid Caprylic acid Arachidonic acid L-Methionine Carbamoyl phosphate
3.98 ( 0.13 3.43 ( 0.21 3.69 ( 0.08 4.90 ( 0.21 3.95 ( 0.22 3.94 ( 0.12 3.29 ( 0.31 3.97 ( 0.11 4.58 ( 0.10 4.23 ( 0.17 4.42 ( 0.17 3.83 ( 0.13 3.65 ( 0.30
4.52 ( 0.32c 3.64 ( 0.41 3.83 ( 0.19b 5.31 ( 0.12b 4.14 ( 0.20b 4.06 ( 0.13b 4.35 ( 0.11c 4.07 ( 0.09b 4.05 ( 0.28c 3.92 ( 0.16b 4.19 ( 0.24b 3.71 ( 0.15b 3.42 ( 0.40
4.35 ( 0.22c 3.27 ( 0.55 3.94 ( 0.23c 5.54 ( 0.09c 3.83 ( 0.23 3.98 ( 0.09 4.01 ( 0.16c 4.16 ( 0.15c 3.94 ( 0.40c 3.77 ( 0.10c 4.09 ( 0.07c 3.64 ( 0.22b 3.45 ( 0.25b
a The relative intensities of metabolites in the HG, AAG and CAMG are expressed with log -transformed value of their peak heights. Values are represented as means(standard deviation (n ) 5), significant difference between the AAG/CAMG and HCG is based on a two-tailed student’s T-test. b p < 0.05. c p < 0.01). d THF: tetrahydrofolate.
tion of Hcy and creatininie, which were confirmed with authentic references from Sigma, all of the other metabolites were identified by their measured mass-to-charge ratios (Table 3). Hcy is a well-established biomarker of renal function. As its concentration increases in kidney tissues, renal function worsens.36,37 Hcy is formed by the demethylation of methionine, and plays an important role in the activated methyl and folate cycles. Methionine can be converted to S-adenosylmethionine intracellularly, and its demethylated S-adenosyl homocysteine product, a thioether, is readily hydrolyzed to Hcy and adenosine by S-adenosyl homocysteine hydrolase. The clearance of Hcy includes two metabolic pathways. In the transsulfuration pathway, Hcy condenses with serine to cystathionine, which is further metabolized to cysteine and sulfate. Alternately, the remethylation of Hcy to methionine occurs through a methyl donation from 5-methyltetrahydrofolate (5-CH3-THF), by means of methionine synthase. 5-CH3-THF is demethylated to tetrahydrofolate (THF), which is reconverted to 5-CH3-THF in a multiple-step folate cycle with serine glycine conversion via methylene tetradydrofolate reductase activity. Herein, the elevated Hcy and adenosine levels suggest the accelerated formation of Hcy, and the observed increase of serine and 5-CH3-THF metabolites indicates participation of the fasted folate cycle. Meanwhile, the relative decrease in methinione levels indicates that Hcy remethylation kidney is markedly decreased.38 Remethylation of Hey is vital to Hcy clearance, and the observed decrease in this modification is consistent with many clinical renal failure reports, especially among patients with end-stage renal disease (ESRD).39,40 D-Serine administered at high doses could cause selective necrosis in the straight portion of the rat renal proximal tubules,41 which is consistently accompanied by proteinuria, glucosuria and amino aciduria. Williams and colleagues42 reported that the metabolism of D-serine by D-amino acid oxidase (D-AAO) may be involved in nephrotoxicity. D-amino acid oxidase (D-AAO), an enzyme that is localized within the peroxisomes of renal tubular epithelial cells,43 is localized within the peroxisomes of renal tubular epithelial cells. This is also the site of D-serine reabsorption. The concentration of D-serine reabsorbed in the pars recta of renal tubules,44 where D-AAO is highly expressed, could account for the selective renal toxicity observed herein. The up-regulated D-serine might suggest that
the metabolism of nephrotoxity.
D-serine
is disturbed during AA-induced
It is also reported that the acute tubular necrosis and acute renal failure might relate to enzymes involved eicosanoids synthesis. The adverse effects of nonsteroidal anti-inflammatory drugs are known to be mediated via inhibition of eicosanoids synthesis from arachidonic acid by nonspecific blocking of the enzyme cyclooxygenase leading to vasoconstriction and reversible mild renal impairment in volume contracted states. When unopposed, this may lead to acute tubular necrosis and acute renal failure.45 Aristolochic acid have effects on the enzymes involved in the release of eicosanoids such as the phospholipase A(2), cyclooxygenase and lipooxygenas pathways.46 Among them, phospholipase A(2) mediated hydrolysis of fatty acids, espcifically arachidonate, from sn-2 position of membrane phospholipids, and it play an important role in this ratelimiting step for the biosynthesis of arachidonic acid.47 Aristolochic acid could directly block phospholipase A(2) catalyzed release of arachidonic acid. The result is also in agreement with the depleted arachidonic acid observed in AA-induced rats.
Conclusion We observed consistent differences among the urinary metabolite profiles of rats treated with AA or CAM, using a LCMS-based metabonomic strategy. This study has demonstrated that the AA-induced changes in the metabolic patterns were associated with rapidly progressive renal failure, and could be identified by a metabonomic method based on LC-MS analysis and multivariate statistics. The PCA score plots suggested distinct differences between the nontreated and AA-dosed rats. The metabolic signature changes suggested the involvement of specific pathways. The formation of Hcy, and the folate cycle were accelerated, while the remethylation of Hcy and the biosysnthis of arachidonic acid were decreased. Six specific variables were used in the data analysis that successfully classified the HCG and AAG groups, and predicted the nephrotoxicity of the CAM group. The results strongly suggested that herbs containing AA might share a similar biochemical metabolite pattern that differs greatly from the normal metabolic process. The high prediction accuracy obtained from these analyses also suggested that these patterns could be utilized to screen the potential nephrotoxicity in related herbs. PerJournal of Proteome Research • Vol. 5, No. 4, 2006 1001
technical notes
Aristolochic Acid-Induced Nephrotoxicity in Rats
spective studies using more sophisticated analytical devices and authentic standards are required to verify the nephrotoxicity predictability potential of these six variables. This preliminary study indicated that metabonomics, one of the most important systems biology platforms, is a promising new tool for identifying and characterizing biochemical responses to toxicity. Therefore, this strategy offers a practical method for conducting toxicity assessments of traditional Chinese medicines (TCM) in the future.
Acknowledgment. This study was financially supported by Shanghai Leading Academic Discipline Project, Project Number, T0301. References (1) Nicholson, J. K.; Holmes, E.; Lindon, J. C.; Wilson, I. D. Nat. Biotechnol. 2004, 22, 1268-1274. (2) Pognan, F. Pharmacogenomics 2004, 5, 879-893. (3) van der Greef, J.; Stroobant, P.; van der Heijden, R. Curr. Opin. Chem. Biol. 2004, 8, 559-565. (4) Nicholson, J. K.; Lindon, J. C.; Holmes, E. Xenobiotica 1999, 29, 1181-1189. (5) Robertson, D. G.; Reily, M. D.; Sigler, R. E.; Wells, D. F.; Paterson, D. A.; Braden, T. K. Toxicol. Sci. 2000, 57, 326-337. (6) Lindon, J. C.; Holmes, E.; Nicholson, J. K. Prog. Nucl. Magn. Reson. Sp. 2004, 45, 109-143. (7) Bollard, M. E.; Stanley, E. G.; Lindon, J. C.; Nicholson, J. K.; Holmes, E. NMR Biomed. 2005, 18, 143-62. (8) Nicholson, J. K.; Holmes, E.; Wilson, I. D. Nat. Rev. Microbiol. 2005, 3, 431-438. (9) Yang, J.; Xu, G.; Zheng, Y.; Kong, H.; Pang, T.; Lv, S.; Yang, Q. J. Chromatogr. B. 2004, 813, 59-65. (10) Wilson, I. D.; Plumb, R.; Granger, J.; Major, H.; Williams, R.; Lenz, E. M. J. Chromatogr. B. 2005, 817, 67-76. (11) van Elswijk, D. A.; Diefenbach, O.; van der Berg, S.; Irth, H.; Tjaden, U. R.; van der Greef, J. J. Chromatogr. A. 2003, 1020, 4558. (12) Castro-Perez, J.; Plumb, R.; Liang, L.; Yang, E. Rapid Commun. Mass Spectrom. 2005, 19, 798-804. (13) Williams, R. E.; Lenz, E. M.; Evans, J. A.; Wilson, I. D.; Granger, J. H.; Plumb, R. S.; Stumpf, C. L. J. Pharm. Biomed. Anal. 2005, 38, 465-471. (14) Fiehn, O.; Kopka, J.; Trethewey, R. N.; Willmitzer, L. Anal. Chem. 2000, 72, 3573-3580. (15) Jonsson, P.; Johansson, A. I.; Gullberg, J.; Trygg, J. A J.; Grung, B.; Marklund, S.; Sjostrom, M.; Antti, H.; Moritz, T. Anal. Chem. 2005, 77, 5635-5642. (16) Jackson, J. E. A User’s Guide to Principal Components; J. Wiley & Sons: New York, 1991. (17) Bro, R. J. Chemom. 1996, 10, 47-62. (18) Yang, J.; Xu, G.; Kong, H.; Zheng, Y.; Pang, T.; Yang, Q. J. Chromatogr. B 2002, 780, 27-33. (19) Smilde, A. K.; Jansen, J. J.; Hoefsloot, H. C.; Lamers, R. J.; van der Greef, J. Bioinformatics 2005, 21, 3043-3048. (20) Mengs, U.; Stotzem, C. D. Arth. Toxicol. 1993, 67, 307-311. (21) Qiu, Q.; Liu, Z. H.; Chen, H. P. Kidney Dis. Kidney Transplant Mag. 1999, 8, 15-18.
1002
Journal of Proteome Research • Vol. 5, No. 4, 2006
(22) Vanherweghem, J. L.; Depierreux, M.; Tielemans, C.; Abramowicz, D.; Dratwa, M.; Jadoul, M.; Richard, C.; Vandervelde, D.; Verbeelen, D.; Vanhaelen-Fastre, R. Lancet 1993, 341, 387-391. (23) Ma, H. M.; Zhang, B. L.; Xu, Z. P. China Pharm. Mag. 2002, 37, 456-458. (24) Isnard Bagnis, C.; Deray, G.; Baumelou, A.; Le Quintrec, M.; Vanherweghem, J. L. Am. J. Kidney Dis. 2004, 44, 1-11. (25) Taes, Y. E.; Delanghe, J. R.; De Vriese, A. S.; Rombaut, R.; Van Camp, J.; Lameire, N. H. Kidney Int. 2003, 64, 1331-1337. (26) Selhub, J. Annu Rev Nutr. 1999, 19, 217-46. (27) Nortier, J.; Schmeiser, H. H.; Muniz Martinez, M. C.; Arlt, V. M.; Vervaet, C.; Garbar, C. H.; Daelemans, P.; Vanherweghem, J. L. Nephrol. Dial. Transplant. 2003, 18, 426-8. (28) Schmeiser, H. H.; Bieler, C. A.; Wiessler, M.; van Ypersele de Strihou, C.; Cosyns, J. P. Cancer Res. 1996, 56, 2025-28. (29) Bieler, C. A.; Stiborova, M.; Wiessler, M.; Cosyns, J. P.; van Ypersele de Strihou, C.; Schmeiser, H. H. Carcinogenesis 1997, 18, 106367. (30) Debelle, F. D.; Nortier, J. L.; De Prez, E. G.; Garbar, C. H.; Vienne, A. R.; Salmon, I. J.; Deschodt-Lanckman, M. M.; Vanherweghem, J. L. J. Am. Soc. Nephrol. 2002, 13, 431-436. (31) Gillerot, G.; Goffin, E.; Moulin, P.; Arlt, V. M.; Phillips, D. H.; Cosyns, J. P.; Devuyst, O. Kidney Int. 2003, 64, 1883-1892. (32) Balachandran, P.; Wei, F.; Lin, R. C.; Khan, I. A.; Pasco, D. S. Kidney Int. 2005, 67, 1797-1805. (33) Jia, W.; Zhao, A.; Chen M. J. Phytotherapy Res., in press. (34) Miao, H. J.; Hu S. X. Chin. J. Anal. Chem. 1994, 22 (3), 247-250. (35) Idborg-Bjorkman, H.; Edlund, P. O.; Kvalheim, O. M.; SchuppeKoistinen, I.; Jacobsson, S. P. Anal. Chem. 2003, 75, 4784-4792. (36) Taes, Y. E.; Delanghe, J. R.; De Vriese, A. S.; Rombaut, R.; Van Camp, J.; Lameire, N. H. Kidney Int. 2003, 64, 1331-1337. (37) Zoccali, C. Curr. Opin. Nephrol. Hypertens. 2005, 14, 532-537. (38) Van Guldence; Stehouwer, C. D. Semin. Vase. Med. 2005, 5, 211218. (39) Herrmann, W.; Schorr, H.; Obeid, R.; Makowski, J.; Fowler, B.; Kuhlmann, M. K. Clin. Chem. 2005, 51, 891-897. (40) Tessari, P.; Coracina, A.; Kiwanuka, E.; Vedovato, M.; Vettore, M.; Valerio, A.; Zaramella, M.; Garibotto, G. Diabetes 2005, 54, 29682976. (41) Ganote, C. E.; Peterson, D. R.; Carone, F. A. The nature of D-serinesinduced nephrotoxicity. Am. J. Pathol. 1974, 77 (2), 269-282. (42) Williams, R. E.; Lock, E. A. Sodium benzoate attenuates D-serine induced nephrotoxicity in the rat. Toxicology 2005, 207 (1), 3548. (43) Pilone, M. S. D-Amino acid oxidase: new findings. Cell Mol. Life Sci. 2000, 57 (12), 1732-1747. (44) Silbernagl, S.; Volker, K.; Dantzler, W. H. D-Serine is reabsorbed in rat renal pars recta. Am. J. Physiol. 1999, 276 (6 Pt 2), F857863. (45) Ejaz, P.; Bhojani, K.; Joshi, V. R. NSAIDs and kidney. J. Assoc. Phys. India 2004, 52, 632-640. (46) Scorrano, L.; Penzo, D.; Petronilli, V.; Pagano, F.; Bernardi, P. Arachidonic acid causes cell death through the mitochondrial permeability transition. Implications for tumor necrosis factoralpha aopototic signaling. J. Biol. Chem. 2001, 276 (15), 1203512040. (47) Moreno, J. J. Effect of aristolochic acid on arachidonic acid cascade and in vivo models of inflammation. Immunopharmacology 1993, 26 (1), 1-9.
PR050404W
