Assessment of Cisplatin-Induced Nephrotoxicity by Microarray Technology

Qihong Huang*,1, Robert T. Dunn, II{dagger}, Supriya Jayadev*, Olimpia DiSorbo*, Franklin D. Pack*, Spencer B. Farr{dagger}, Raymond E. Stoll* and Kerry T. Blanchard*

* Department of Toxicology, Boehringer Ingelheim Pharmaceuticals, Inc., 900 Ridgebury Road, P.O. Box 368, Ridgefield, Connecticut 06877-0368; {dagger} Phase-1 Molecular Toxicology, Inc., Santa Fe, New Mexico 87505

Received May 10, 2001; accepted July 31, 2001


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Microarrays are a new technology used to study global gene expression and to decipher biological pathways. In the current study, microarrays were used to examine gene expression patterns associated with cisplatin-mediated nephrotoxicity. Sprague-Dawley rats received either single or seven daily ip doses of cisplatin (0.5 or 1 mg/kg/day) or the inactive isomer transplatin (1 or 3 mg/kg/day). Histopathological evaluation revealed renal proximal tubular necrosis in animals that received cisplatin for 7 days, but no hepatotoxic findings. Microarray analyses were performed using rat specific arrays containing 250 toxicity-related genes. Prominent gene expression changes were observed only in the kidneys of rats that received cisplatin for 7 days. Mechanistically, the gene expression pattern elicited by cisplatin (e.g., Bax{uparrow} and SMP-30{downarrow}) suggested the occurrence of apoptosis and the perturbation of intracellular calcium homeostasis. The induction of multidrug resistance genes (MDR1{uparrow}, P-gp{uparrow}) and tissue remodeling proteins (clusterin{uparrow}, IGFBP-1{uparrow}, and TIMP-1{uparrow}) indicated the development of cisplatin resistance and tissue regeneration. Select gene expression changes were further confirmed by TaqMan® analyses. Gene expression changes were not observed in the liver following cisplatin administration. In contrast to these in vivofindings, studies using NRK-52E kidney epithelial cells and clone-9 liver cells suggested that liver cells were more sensitive to cisplatin treatment. The discrepancies between thein vivoandin vitroresults suggest that caution should be taken when extrapolating data fromin vivotoin vitrosystems. Nonetheless, the current study elucidates the biochemical pathways involved in cisplatin toxicity and demonstrates the utility of microarrays in toxicological studies.

Key Words: microarray; toxicogenomics; cisplatin; nephrotoxicity; apoptosis; real-time PCR.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
The emergence of new technologies such as expressed sequence tag (EST), combinatorial chemistry, and high throughput screening (HTS) have facilitated drug development in the pharmaceutical industry. The resultant increase in the number of developmental compounds presents a challenge to traditional, time-consuming toxicity studies. Microarrays, introduced in 1995 (Schena et al., 1995Go), are a powerful tool useful for examining global gene expression changes associated with development (Tanaka et al., 2000Go), disease (Elek et al., 2000Go; Heller et al., 1997Go; Wang et al., 1999Go), and chemical exposure (Puga et al., 2000Go; Zimmermann et al., 2000Go). This new technology, which allows us to study thousands of genes simultaneously, is a backbone of toxicogenomics, a new subdiscipline combining toxicology with emerging technologies of genomics (Farr and Dunn, 1999Go; Medlin, 1999Go; Nuwaysir et al., 1999Go). In microarray-based toxicogenomics, gene expression changes induced by classical toxic compounds are studied using gene arrays; functional genomic profiles are established; and informative, mechanistic markers of toxicities are discovered. Subsequently, the gene expression signatures from drug candidates are compared against the preestablished learning set of known toxic agents. This type of comparative approach will allow us to predict a deleterious outcome prior to undertaking laborious studies (Braxton and Bedilion, 1998Go; Farr and Dunn, 1999Go). Microarrays enable toxicologists to view the effects of a compound at the genomic level and provide insight into the mechanism of toxicity. Such insights are beyond the scope of descriptive histopathology. Furthermore, microarray analyses can be extended to future clinical diagnostic tests and patient management (Snijders et al., 2000Go).

Although high-density gene arrays containing more than 10,000 genes and EST sequences are increasingly available and utilized, efforts have been made to generate lower-density, targeted arrays such as toxicity specific arrays (Bartosiewicz et al., 2000Go; Farr and Dunn, 1999Go; Nuwaysir et al., 1999Go). Two advantages of "tox arrays" over larger arrays are decreased cost and more approachable mechanistic data interpretation due to the use of genes with well-understood roles in toxicology (Farr and Dunn, 1999Go).

In the current study, a 250-gene tox array was used to determine gene expression patterns elicited by cisplatin [cis-dichlorodiammine-platinum(II)] and transplatin [trans-dichlorodiammine-platinum(II)] in tissues and cell lines. Cisplatin is an antineoplastic agent effective in the treatment of various solid tumors, yet its clinical application is limited due to its severe renal toxicity. Cisplatin is actively uptaken by probenecid-inhibitable organic ion transporters (Ban et al., 1994Go; Safirstein et al., 1984Go), resulting in an enrichment of cisplatin in proximal tubular epithelial cells. At least two mechanisms contribute to cell death following cisplatin treatment: inhibition of DNA synthesis (Howle and Gale, 1970Go) and glutathione depletion–induced oxidative stress (Kuhlmann et al., 1997Go). Transplatin, the inactive isomer, is taken up by the same site in kidney, yet fails to induce any nephrotoxicity at equivalent doses. Herein, we report a gene expression study on the topic of cisplatin-induced renal toxicity using both in vivo and in vitro systems. Our microarray results not only confirmed some of the findings in the literature, but also identified numerous novel gene expression changes associated with cisplatin treatment. These results increase our understanding of the mechanism of cisplatin-mediated nephrotoxicity. In addition, the expression profiles contribute to the building of a toxicogenomic database of known toxic compounds. Taken together, the current study provides a good example for the application of microarrays in toxicological studies.


    MATERIALS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Materials.
Cisplatin (CAS # 15663-27-1) and transplatin (CAS # 14913-33-8) were obtained from Sigma (St. Louis, MO). Cell culture reagents were obtained from Life Technologies (Rockville, MD). Rat Tox Microarrays were purchased from Phase-1 Molecular Toxicology (Santa Fe, NM). Microarrays consisted of 250 rat cDNAs with four replicates of each sequence spotted across the array.

Animal treatment.
Male Sprague-Dawley VAF+ albino rats (CRL:CD(SD) BR; Charles River, Kingston, NY) approximately 5–7 weeks old were maintained on certified rodent chow (PMI Feeds, Inc., Brentwood, MO) ad libitum in individual stainless steel wire bottom cages suspended on racks. The animals were kept under carefully controlled conditions of 12-h light:dark cycle, 72° ± 5° F and 50 ± 20% relative humidity. The animals were acclimated to this environment for 4–7 days prior to the start of the study. Rats were randomly assigned to treatment groups (2–5/group) and received ip doses of vehicle (saline-glucose), cisplatin (0.5 mg/kg/day or 1 mg/kg/day), or transplatin (1 mg/kg/day or 3 mg/kg/day) for 24 h or 7 days. The doses of cisplatin were chosen to produce nephrotoxicity without lethality after a 7-day exposure period. Following treatment, animals were sacrificed, kidneys and livers were examined macroscopically, and cross sections of kidneys and livers were collected in 10% neutral buffered formalin for histopathological evaluation. The remaining portions of the kidney and liver were snap-frozen in liquid nitrogen for RNA isolation. Experiments were performed according to the guidelines established in the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Histopathology.
The tissues collected in formalin at necropsy were processed, embedded in paraffin, sectioned at 5 microns, and stained with hematoxylin and eosin (H&E). Histopathologic examinations of the tissue sections were conducted by a veterinary pathologist and peer reviewed.

Cell culture and compound treatment.
Normal rat renal epithelial NRK-52E cells and normal rat hepatocyte clone-9 cells were obtained from the American Type Culture Collection (Manassas, VA). NRK-52E cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) (high glucose with sodium pyruvate) containing 10% calf serum. F12 medium supplemented with 10% fetal bovine serum was used as growth medium for clone-9 cells. Cells were grown in a humidified incubator containing 5% CO2 at 37°C. Cells were seeded in 100-mm dishes and grown to confluence. Cells (two dishes/dose) were then treated for 24 h with increasing concentrations (0, 1, 2.5, and 5 µg/ml) of cisplatin dissolved in culture medium.

RNA isolation and probe labeling.
Total RNA from tissues and cells was isolated using QIAGEN RNeasy kits (QIAGEN Inc., Valencia, CA). RNA from individual experimental samples was reverse transcribed in the presence of Cy 3-dCTP. Pooled RNA from naive animals (six rats) or cells was reverse transcribed in the presence of Cy5-dCTP and used as a reference for the experimental samples. Briefly, total RNA (20 µg/sample) was primed with oligo d(T) (0.25 µg/µl) at 70°C for 10 min. This was followed by the addition of reaction mix containing dNTP mix, 0.1 M dithiothreitol, and first-strand reaction buffer. For experimental RNA, dNTP mix constituted 0.5 mM dATP/dGTP/dTTP, 0.125 mM dCTP, and 0.125 mM Cy3-dCTP. For reference RNA, dNTP mix contained 0.5 mM dATP/dGTP/dTTP, 0.15 mM dCTP, and 0.1 mM Cy5-dCTP. Superscript II Rnase H reverse transcriptase (Life Technologies) was added and the reaction was carried out at 45°C for 2 h. Labeled cDNA probes were purified using Wizard® Series 9600TM DNA Purification Resin (Promega Co., Madison, WI), and the remaining fluorescence density was determined using a Wallac Victor2 1420 Plate Reader (Perkin Elmer Wallac Inc., Gaithersburg, MD).

Hybridization, scanning, and quantitation.
Cy5- and Cy3- labeled probes were combined and hybridized to microarrays (250 genes in four replicates) overnight at 42°C. Sample from each animal/culture dish was hybridized to a single array. After hybridization, microarrays were washed, dried, and scanned using an Axon GenePix 4000A® scanner (Axon Instrument Inc., Foster City, CA). Fluorescence intensities of the Cy3 and Cy5 channels were quantitated using GenePix Pro 3.0® software (Axon Instrument Inc.). Gene expression data were analyzed as follows: the Cy3/Cy5 ratios of each gene replicate were averaged and normalized to the total fluorescence of Cy3/Cy5 across the slide. Normalized ratios were transformed by base-2 logarithms.

Real-time PCR.
PCR primers and TaqMan® probes (Table 1Go) were designed using Primer Express V.1.5 software (Applied Biosystems Inc., Foster City, CA) based on the sequences from GenBank, and purchased from Applied Biosystems. TaqMan® probes were labeled with 6-carboxy-fluorescein (FAM) as the reporter dye and 6-carboxy-tetramethyl-rhodamine (TAMRA) as the quencher dye. The Rodent GAPDH amplicon (as supplied by Applied Biosystems Inc.) was used as an endogenous control to normalize all the data. The GAPDH probe was labeled with VIC as the reporter dye and TAMRA as the quencher dye. Real-time PCR was performed in a two-step process. In the first step, sample RNA (0.1 µg) or pooled reference RNA (0.01–0.5 µg) were reverse transcribed in a volume of 100 µl containing TaqMan® RT buffer, 5.5 mM MgCl2, 500 µM of each dNTP, 2.5 µM random hexamers, 0.4 U/µl RNase inhibitor, and 1.25 U/µl MultiScribe Reverse Transcriptase at 25°C for 10 min, 48°C for 30 min, and 95°C for 5 min. In the second step, real-time PCR was carried out in a MicroAmp Optical 96-well plate using TaqMan® Gold PCR Reagents (Applied Biosystems Inc.). Each well contained 5 µl of reverse-transcribed cDNA, TaqMan® buffer A, 5.5 mM MgCl2, 200 µM each of dATP/dCTP/dGTP, 400 µM dUTP, 900 nM each of forward and reverse primers (100 nM each for GAPDH), 250 nM TaqMan® probe (200 nM for GAPDH), 0.01 U/µl AmpErase UNG, and 0.025 U/µl AmpliTaq Gold DNA polymerase in a total volume of 50 µl. The thermal cycling conditions for real-time PCR were a) 50°C for 2 min, b) 95°C for 10 min, and c) 40 cycles of melting (95°C, 15 s) and annealing/extension (60°C, 60 s). PCR reactions were monitored in real time using the ABI PRISM 7700 Sequence Detector (Applied Biosystems Inc.). A standard curve for each target gene was generated with pooled reference RNA. Relative quantitation of gene expression was determined using the standard curve method as described in ABI's User Bulletin #2.


View this table:
[in this window]
[in a new window]
 
TABLE 1 Sequences of PCR Primers and Taqman® Probes
 
Statistical analysis.
The results of microarray and TaqMan® analyses from in vivo studies were examined for statistical significance with the Dunnett's test. A p value less than 0.05 was accepted as significant.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Histopathology
Rats dosed ip with cisplatin (0.5 mg/kg/day or 1 mg/kg/day) or its inactive isomer transplatin (1 mg/kg/day or 3 mg/kg/day) exhibited no drug-related histopathology in kidneys or livers at 24 h (data not shown). However, in the 7-day cisplatin-treated kidneys, necrosis occurred in the proximal tubular epithelial cells of the S3 segment located in the outer stripe of the outer medulla (Figs. 1C and 1DGoGo). Necrosis was also accompanied by scattered apoptosis, as evidenced by diminished cell size, reduced cytoplasmic volume with condensed eosinophilic cytoplasm, condensed nuclear chromatin, and karyorrhexis (data not shown). These changes were accompanied variably by formation of apoptotic bodies in individual tubular epithelial cells (data not shown). Variable interstitial lymphoplasmacytic inflammatory infiltrates were observed. Moreover, some dilated tubules were lined with flattened, hyperchromatic epithelium (Figs. 1C and 1DGoGo). Mitotic figures were occasionally noted in regenerative epithelium. In contrast to the nephrotoxicity, no hepatotoxicity was found in cisplatin-treated livers (data not shown). Transplatin treatment for 7 days had no toxic effect on either kidney (Figs. 1E and 1FGoGo) or liver as expected (data not shown).



View larger version (165K):
[in this window]
[in a new window]
 
FIG. 1. Light micrographs comparing kidneys from control male rats with kidneys from male rats treated with cisplatin or transplatin. (A) Kidney from control male showing normal morphology of the S3 segment of the proximal tubule located in the outer stripe of the outer medulla. Original magnification x100. (B) Same section at higher magnification. Original magnification x400. (C) Kidney from male rat treated with 1.0 mg/kg/day cisplatin for 7 days. Original magnification x100. Note the widespread necrosis and sloughing of proximal tubular epithelial cells lining the S3 segment (arrow). (D) Higher magnification of the previous section showing overt necrosis and sloughing of proximal tubular epithelial cells of the S3 segment. Affected tubules were frequently lined with flattened hyperchromatic epithelium (asterisks). Original magnification x400. (E) Kidney from male rat treated with 3.0 mg/kg/day transplatin for 7 days. Original magnification x100. No drug-related histopathologic changes were observed. (F) Same area at higher magnification. Original magnification x400.

 
Differential Gene Expression in Cisplatin-Treated Kidney in Vivo
Microarray analyses: significant outliers.
To study the underlying mechanism of cisplatin-mediated nephrotoxicity, kidney samples from cisplatin- or transplatin-treated rats were analyzed on 250 gene tox arrays. Reverse-transcribed samples from individual animals were hybridized to a single array. Each group contained 2–4 animals; therefore, each dose group was represented by 2–4 arrays. This type of approach, in contrast to running replicate arrays of sample from the same animal, took into account biological as well as methodological variability.

Dunnett's test, a commonly used statistical test in toxicology, was applied to the microarray data to determine statistical significance between compound-treated groups and control groups. The Dunnett's test and a 95% confidence cutoff were chosen in order to minimize false-negative results. With the use of an in vivo model system (which generally exhibits higher variance than in vitro systems) and a low-density array (250 genes), high stringency analyses would miss potentially significant changes. Using the parameters above on the 7-day cisplatin kidney data, the expression of 76 genes was found to be significantly different compared to controls (Table 2 Go). The 76 genes could be subdivided into three subsets based on their absolute log ratios, as follows: 1) absolute log ratio >= 1 (>= 2-fold change); 2) absolute log ratio between 0.5 and 1 (between 1.4- and 2-fold change); and 3) absolute log ratio < 0.5 (< 1.4-fold change).


View this table:
[in this window]
[in a new window]
 
TABLE 2 Statistical Significance of Differential Gene Expression in Seven-Day Cisplatin-Treated Kidney
 
The first subset (>= 2-fold change) contained 22 genes, which could be subdivided into eight functional categories (Fig. 2Go): 1) calcium homeostasis genes, 2) oxidative stress genes, 3) cell cycle genes, 4) transporter genes, 5) tissue remodeling genes, 6) inflammatory genes, 7) metabolism genes, and 8) cancer-related genes. The most robust changes (approximately 2.5 log ratio/7-fold increase) were observed in the multidrug resistance genes MDR1 and p-glycoprotein (P-gp; Fig. 2Go). These proteins are involved in efflux of drugs, thus the upregulation of the mRNA for MDR1 and P-gp is consistent with drug exposure. Secondary to the changes in the MDR genes, the most significant changes were observed in the expression levels of cyclin G, clusterin, insulin-like growth factor binding protein 1 (IGFBP-1), and tissue inhibitors of metalloproteinases-1 (TIMP-1; Fig. 2Go). The high magnitude changes in these genes were likely due to toxicity-related tissue regeneration. Repression of senescence marker protein-30 (SMP-30; Fig. 2Go) and catalase (Fig. 2Go) indicated perturbation of calcium homeostasis and oxidative status. Furthermore, the expression levels of genes involved in organic ion transporters, inflammation, and phase II metabolism also exhibited significant changes (Fig. 2Go). These changes were also consistent with the toxic mechanism of cisplatin or with secondary compensatory responses to cisplatin toxicity (Fig. 2Go; see discussion).



View larger version (33K):
[in this window]
[in a new window]
 
FIG. 2. Microarray analysis of differential gene expression in the kidney following 7 days of cisplatin treatment. Rats received ip cisplatin at low (0.5 mg/kg/day) and high (1 mg/kg/day) doses; kidney RNA was isolated and subjected to microarray analyses. Genes undergoing statistically significant changes with >= 2-fold induction or repression (absolute log ratio >= 1) were plotted. Genes are classified into eight functional categories: calcium homeostasis: SMP-30; oxidative stress: catalase; cell cycle: c-myc, cyclin G, waf1; transporter: MDR1, P-gp, OAT3, OAT-K1, OATP-1, OCT2; tissue remodeling: clusterin, IGFBP-1, TIMP-1; inflammation: ceruloplasmin, IP-10; metabolism: {gamma}-GT, HST-a, S-adenosylmethionine synthetase (SAMS); and cancer: CD44 metastasis suppressor gene (CD-44), thymosin beta-10 (T-beta-10). Data are presented as mean ± SE. (n = three rats) *Indicates significantly different from control at p < 0.05.

 
The second subset of significantly induced genes contained 15 genes with a median magnitude change in gene expression (between 1.4- and 2-fold change). These genes were identified as significant changes compared to controls due to low variance between replicates. The third subset of significant genes (< 1.4-fold change) exhibited only slight changes in gene expression relative to controls. For this subset of genes, caution should be taken and one should not solely rely on Dunnett's test to determine their biological significance. Higher-order statistical models that account for multiple sources of variance are necessary to draw definitive conclusions (see discussion).

Most of the genes exhibited a larger number of significant expression changes with the higher dose treatment, a finding consistent with the observed histopathology, suggesting the changes were real and cisplatin dependent. However, only subsets 1 and 2 exhibited dose-responsive changes. Three genes in subset 1 and six genes in subset 2 (Table 2Go) were found to be significantly induced in response to low-dose cisplatin and induced to a greater extent with high-dose treatment. Even in the case where low dose did not result in statistically significant change in expression levels, the trend from low dose to high dose was consistent with a dose-responsive effect. In contrast to subsets 1 and 2, subset 3 did not show obvious dose responsiveness. Because subset 3 genes exhibited only slight gene expression changes even with high-dose treatment, it is difficult to assess whether these genes were dose responsive or not.

Using the same statistical criteria, the remaining experimental conditions were also evaluated. In the 7-day transplatin-treated kidney data, 23 genes were determined to be statistically significant compared to controls. However, of the 23, only 2 genes were changed by greater than 2-fold at the high dose: CRP (log ratio = 1.2) and liver fatty acid binding protein (log ratio = 1.2). The remaining genes fell in subset 3, with changes less than 1.4-fold. As transplatin is considered an inactive isomer of cisplatin with no significant histopathological effects on either the kidney or liver, the lack of many large gene changes in transplatin kidneys is not surprising. Similarly, in the 1-day cisplatin group, minimal gene expression changes were observed compared to the 7-day treatment. Forty genes were identified as significantly different in the 1-day cisplatin-treated kidneys by the Dunnett's test. Of the 40 significant genes, only three genes (fibrinogen gamma chain {uparrow}, transthyretin {uparrow}, UDP- glucuronosyltransferase 2B {uparrow}) fell into gene subset 1 (>= 2-fold change), compared to 22 genes at the 7-day time point. Two of the 40 genes (p55CDC {downarrow}, liver fatty acid binding protein {uparrow}) were in the second subset (between 1.4- and 2-fold change) compared to 15 genes at the 7-day time point. The majority of the genes (35) that were statistically significant at the 1-day time point fell into subset 3 (< 1.4-fold change). The fewer large gene changes observed at 1 day compared to 7 day were consistent with the lack of histopathological findings in single-dose cisplatin-treated kidneys. As nephrotoxicity became apparent histopathologically, more genes were regulated and changed to a larger extent.

Finally, few gene expression changes were observed in cisplatin- or transplatin-treated livers (data not shown), consistent with the fact that neither cisplatin nor transplatin had any histopathological effect (even after 7 days of treatment) on the liver at the doses used in the current study. Based on the high correlation between histopathology and microarray data, we are confident that the overall gene expression change is a specific effect and not an artifact generated from the experimental procedure.

Validation with real-time quantitative PCR.
Microarrays are powerful tools for investigating global gene expression; however, as with any developing technology, a second quantitative measure is necessary to validate critical findings. Real-time, quantitative PCR (TaqMan® assay) has become widely accepted for this purpose. In the current study, the TaqMan® assay was used to validate seven interesting gene changes identified by microarray analyses (MDR1, cyclin G, SMP-30, IGFBP-1, catalase, {gamma}-glutamyl-transpeptidase ({gamma}-GT), and P-gp). These seven genes selected for TaqMan® analyses spanned the key functional categories presented in Figure 2Go.

The TaqMan® assay yielded results that were in agreement with the gene array results (Fig. 3Go). According to both analyses, the expression of MDR1, cyclin G, SMP-30, and IGFBP-1 were significantly induced in the 7-day cisplatin-treated kidney samples. TaqMan® determined changes in catalase, {gamma}-GT, and P-gp were less prominent compared to those observed with gene array analyses (Table 3Go), albeit the changes were still statistically significant. The lack of consensus in the magnitude of change between gene arrays and TaqMan® analyses was not surprising. Two reasons may have contributed to the difference in magnitude. First, differences could have been due to differences intrinsic to the methodologies. Array data were normalized to total fluorescence across the entire slide. TaqMan® data were normalized to a single housekeeping gene (GAPDH in the current study). Large differences in the normalization value would result in large variances in the fold change. Furthermore, whereas cDNA gene arrays are sensitive to saturation at high fluorescence levels, TaqMan® is not. Thus, large changes in array data could be partially masked by saturation. A second factor that may have accounted for the magnitude differences observed with the two techniques was sequence difference. Although the same genes were analyzed using arrays and TaqMan®, the specific subsection of sequence assessed could differ. The arrays contain a 500 base pair region of each cDNA, which may not necessarily overlap the approximately 100 base pair amplicon probed by TaqMan®. Without considering the differences in magnitude, it is apparent that the TaqMan® results confirmed the results observed with gene array.



View larger version (16K):
[in this window]
[in a new window]
 
FIG. 3. Quantitation of gene expression by real-time PCR. Rats received ip cisplatin (1 mg/kg/day) or transplatin (3 mg/kg/day) for 7 days. The expression of MDR1, cyclin G, SMP-30, IGFBP-1, catalase, {gamma}-GT, and P-gp in the kidney was quantitated by real-time PCR. Relative quantities, normalized to GAPDH (see Materials and Methods section), are shown. Data are presented as mean ± SE. (n = three rats) *Indicates significantly different from control at p < 0.05.

 

View this table:
[in this window]
[in a new window]
 
TABLE 3 Comparison of Microarray and TaqMan® Results of Seven-Day Cisplatin-Treated Kidney
 
Differential Gene Expression in Cisplatin-Treated Liver and Kidney Cell Lines
Animal studies tend to be labor intensive and time consuming, and can be very expensive. Ideally, an in vitro system would allow for a more rapid, higher throughput assay. However, validation of an in vitro system is essential to the success of such an approach. Thus, we assessed whether fingerprint patterns obtained with in vivo studies could be reproduced using in vitro cell lines. Cisplatin's site of action in the kidney is at the proximal tubules. NRK-52E cells are a rat tubular epithelial cell line, and the closest available cell line representative of proximal tubular epithelium. Because NRK-52E cells are an established in vitro system used to study cisplatin nephrotoxicity (Fukuishi and Gemba, 1989Go), we have used them to profile gene expression changes. As with the in vivo studies, we compared kidney with liver. The clone-9 cell line derived from normal Sprague-Dawley liver was used for comparison.

Both clone-9 and NRK-52E cells were treated with cisplatin at multiple doses (0, 1, 2.5, and 5 µg/ml) for 24 h, and RNA was subjected to microarray analyses. Although no statistical analysis was conducted due to the lack of sufficient replicates, gene expression profiles obtained from in vitro studies were distinct from those obtained from in vivo studies. There was a dose-response effect in gene expression changes in clone-9 cells (data not shown), whereas gene expression changes (absolute log ratio >= 1, 2-fold change) only occurred at the highest dose (5 µg/ml) in NRK-52 cells. High-dose cisplatin (5 µg/ml) induced 33 and 9 genes differentially (absolute log ratio >= 1, 2-fold change) in clone-9 and NRK-52E cells, respectively (Table 4Go). Consistently, cell shrinkage was observed in liver cells dosed with 2.5 and 5 µg/ml (data not shown), whereas no morphological changes were observed in kidney cells at any dosing concentration (data not shown). These data suggest that liver cells were more responsive to cisplatin-mediated toxicity than kidney cells. The converse was true in the in vivo studies, where no gene expression changes were detected in rat liver. Furthermore, the magnitude of change in the apoptosis-related genes Bax, c-fos, and Gadd153 were much greater in vitro than observed in vivo (Table 4Go). Although discrepancies exist, we did see some correlation in gene changes between in vitro and in vivo systems. C-myc, clusterin, CD44 metastasis suppressor gene, cyclin G, hydroxysteroid sulfotransferase a (HST-a), IGFBP-1, interferon inducible protein 10 (IP-10), MDR1, P-gp, TIMP-1, and waf1 were differentially expressed both in the cell lines and in vivo. These results suggest that in vitro to in vivo extrapolations only yield part of the picture.


View this table:
[in this window]
[in a new window]
 
TABLE 4 Comparison of in Vivoandin VitroDifferential Gene Expression in Response to Cisplatin Treatment
 

    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
The clinical usage of cisplatin is limited due to its renal toxicity. Establishing the gene expression patterns and understanding the mechanism of cisplatin-induced toxicity should allow earlier identification of clinically relevant toxicological findings in compound screening, and aid in the development of therapeutics to reduce nephrotoxicity. Using microarrays in conjunction with statistical analyses, the current study has identified a number of differentially expressed genes associated with cisplatin-mediated nephrotoxicity (Table 2Go). Twelve of the genes identified here (Table 2Go) have previously been associated with cisplatin treatment, and most of the changes shown before were consistent with current findings in terms of induction or repression. The majority of the genes we have identified are newly associated with cisplatin activity.

Genes that were significantly regulated by cisplatin treatment fell into approximately 8–10 major functional categories (Fig. 2Go and discussion below). These genes could be associated with the toxic mechanism of cisplatin (apoptosis and/or necrosis) or with secondary compensatory responses (inflammation, resistance development, and tissue regeneration). Intracellular uptake of cisplatin by organic ion transporters plays a key role in cisplatin-mediated nephrotoxicity (Kuhlmann et al., 1997Go). The major intracellular toxic mechanisms of cisplatin include inhibition of DNA synthesis (Howle and Gale, 1970Go), oxidative stress (Kuhlmann et al., 1997Go), and perturbation of calcium homeostasis (De Witt et al., 1988Go). These changes could be responsible for downstream effects such as cell death and inflammation in cisplatin-treated kidneys (Fig. 4Go). Significant gene expression changes identified by microarray analyses allow us to tie the above mechanisms and consequences together (Fig. 4Go).



View larger version (44K):
[in this window]
[in a new window]
 
FIG. 4. Putative cellular pathways of cisplatin-mediated nephrotoxicity. The diagram depicts the putative pathways involved in cisplatin-mediated nephrotoxicity. This schematic was assembled using the gene expression data from 7 days cisplatin (1 mg/kg/day, ip) treatment and previous literature. Up- or downregulation is denoted by {uparrow} or {downarrow}. Mechanisms such as oxidative stress and perturbation of calcium homeostasis might contribute to cisplatin toxicity. Secondary compensatory mechanisms were evident as increases in tissue regeneration and development of cisplatin resistance.

 
One of the major histopathological findings in the 7-day cisplatin treated kidneys was necrosis of the proximal tubular epithelial cells (Fig. 1Go), with scattered apoptosis observed in the affected tubules. Consistent with this, development of apoptosis has been associated with cisplatin-induced acute renal failure (Zhou et al., 1999Go). In the current study, pro-apoptotic proteins annexin V, Bax, and Gadd153 were mildly induced in 7-day cisplatin-treated kidneys (Table 2Go), consistent with previous reports of cisplatin's apoptotic mechanism. (Delmastro et al., 1997Go; Gately et al., 1994Go; Henkels and Turchi, 1999Go; Los et al., 1999Go).

Changes in oxidative state, calcium homeostasis, and cell cycle could be responsible for downstream cell death (Chao, 1996Go; Kowaltowski et al., 2001Go). In agreement with such a mechanism, we observed changes in the oxidative stress response genes catalase ({downarrow}), superoxide dismutase ({downarrow}), metallothionein 1 ({uparrow}), and {gamma}-GT ({downarrow}) in 7-day cisplatin-treated kidneys (Table 2Go). Glutathione depletion and oxidative stress might precede apoptosis, as glutathione depletion has been illustrated in cisplatin-mediated nephrotoxicity (Bompart, 1989Go), and glutathione supplementation prevents cisplatin-induced apoptosis (Okuda et al., 2000Go). We also found a significant decrease in the expression of SMP-30 (Fig. 2Go), a regulator of intracellular calcium concentration, which expression has previously been found to decrease following cisplatin administration (Kurota and Yamaguchi, 1995Go). Furthermore, an increase in expression of the cell cycle inhibitor waf-1 (Fig. 2Go) suggested the arrest of cell cycle following cisplatin treatment.

Inflammation and tissue regeneration usually occur secondary to tissue necrosis. In fact, histopathological examination revealed that kidney injury induced by repeated cisplatin treatment was accompanied by interstitial lymphoplasmacytic infiltrates and regenerative epithelium (Fig. 1Go). Because ceruloplasmin and IP-10 were strongly upregulated (Fig. 2Go), the inflammatory process observed in 7-day cisplatin-treated kidneys involved at least these two genes. The microscopic observation of a regenerative process was substantiated at the gene level as changes in clusterin ({uparrow}), EGF ({downarrow}), TIMP-1 ({uparrow}), and IGFBP-1 ({uparrow}) (Fig. 2Go), all of which have been associated with kidney remodeling after acute renal failure (Bohe et al., 1998Go; Gobe et al., 1995Go; Lee et al., 1997Go; Leonard et al., 1994Go; Silkensen et al., 1997Go).

Cell proliferation, a key component in tissue regeneration, was evidenced as increased levels of cyclin G and c-myc (Fig. 2Go). Ectopic cyclin G expression has been shown to result in increased sensitivity to cisplatin cytotoxicity (Smith et al., 1997Go), suggesting that the proliferative response is not protective in nature. It is interesting to note that increases in cyclin G and c-myc are compatible with apoptotic response and may in fact act as mediators of the apoptotic response (Hoffman and Liebermann, 1998Go; Okamoto and Prives, 1999Go). Whether the cell cycle changes were mediators of toxicity or compensatory responses to toxicity remains to be determined.

A key compensatory mechanism initiated in response to cisplatin treatment was the development of resistance mechanisms. We observed a decrease in {gamma}-GT levels (Fig. 2Go). As inhibition of {gamma}-GT activity blocks the nephrotoxicity of cisplatin (Hanigan et al., 1996Go), downregulation of renal {gamma}-GT expression probably signaled a protective response. In addition, two closely related sequences of the p-glycoprotein family, MDR1 and P-gp (Fig. 2Go), were strongly induced in kidneys following 7 days of cisplatin treatment. In fact, the expression changes in these two genes were the most robust changes observed in the current study. The p-glycoprotein family of multidrug resistance proteins are responsible for efflux of various toxic compounds and chemotherapeutic agents. Upregulation of these transporters could be a mechanism to increase cisplatin efflux and thereby decrease cellular toxicity.

Expression levels of organic ion transporters were also changed in kidneys following 7 days of cisplatin treatment. In contrast to the increases in multidrug resistance transporters, we found decreases in four organic ion transporters with cisplatin treatment. The expression of three organic anion transporters, organic anion transporter 3 (OAT3), organic anion transporter K1 (OAT-K1), and organic anion transporting polypeptide 1 (OATP-1), and one organic cation transporter, organic cation transporter 2 (OCT2), were decreased significantly following cisplatin exposure (Fig. 2Go). The significance of these changes to cisplatin toxicity remains elusive; however, it is intriguing to speculate that these transporters may play a role in the uptake of cisplatin and that their decrease would result in decreased uptake of cisplatin.

The current study emphasizes the importance of using statistical analyses in assessing microarray data. By applying the Dunnett's test and a 95% confidence level to the data, we were able to pull out a handful of cisplatin response indicators. If we had simply applied an arbitrary cutoff value (e.g., 2-fold or greater expression change) to distinguish significance, we may have generated a number of false-positive and false-negative results. Using statistical parameters allows us to distinguish between the noise of the system and significant changes. For example, catechol-O-methyltransferase (COMT), which exhibited > 2-fold expression change in the 7-day cisplatin-treated samples, failed to be statistically significant. Historical control data generated by our facility suggests that several genes, including COMT, have a large dynamic range in basal expression. Thus, to find significance in COMT gene expression change, a much greater than 2-fold/1 log ratio change would need to be observed. Moreover, using the 2-fold cutoff, the second subset of genes in Table 2Go would have been completely missed, illustrating the necessity of replicates and statistics in determining specific treatment-dependent genes. As for the genes in the third subset of Table 2Go whose expression only changed slightly, additional factors should be taken into account. Small changes in gene expression could be significant in the context of a heterogeneous tissue where only a specific cell type exhibits a particular response. In such a situation, microdissection of the affected cells and analysis of their gene expression changes could yield a large-fold change that is lost when assessing the heterogeneous cell population. In cases where low gene expression changes are observed, higher-order statistical models are necessary to arrive at a definitive answer. Such mixed models, which take into account sources of variability across experiments, samples, etc., are currently under development for microarray data analyses (Wolfinger, personal communication). Moreover, viewing the global gene expression as a whole and depicting the interrelationship between genes within the same functional group facilitates interpretation. Thus, specific gene changes can be evaluated based on the context of other gene changes associated with the same pathways.

Finally, by comparing gene expression profiles elicited by cisplatin treatment in vivo versus in vitro, we found one system was not necessarily predictive of the other. In vivo, kidney, not liver, was the target organ of cisplatin-mediated toxicity. In vitro, liver cells were more sensitive than kidney cells to cisplatin-induced gene expression changes. In vivo, the vulnerability of the kidney to cisplatin is related to its role as the primary route for cisplatin excretion (Safirstein et al., 1986Go). Platinum concentrations in kidney following cisplatin injection can reach 6-time higher levels than in liver (Suzuki and Cherian, 1990Go). Such kinetics does not exist in vitro. Bathing cells in cisplatin presumably exposes cultured cells to considerable amounts of cisplatin irrespective of whether they are liver cells or kidney cells. Exposing liver cells to the appropriate physiological dose (e.g., six times less than used on kidney cells) may result in sensitivity that is more comparable with the in vivo system.

Furthermore, the fingerprint of cisplatin-responsive gene changes in vivo versus in vitro was significantly different. This is not surprising, in light of the fact that only one cell type is represented in vitro, whereas the in vivo system is composed of a heterogeneous population of cells. So far, in vitro cell lines have been the major system applied to microarray studies (Burczynski et al., 2000Go), probably due to issues of cost and approachability. Based on the data presented here, we would suggest that cultured cell lines may not be the best system for extrapolating in vivo response. Perhaps a better predictor of in vivo response would be primary cell cultures or tissue sections containing a mixed population of cells. No matter what the system, it is important to consider that in vitro models may not necessarily substitute for or be predictive of in vivo systems.

Toxicogenomics assumes that toxicity is always accompanied by gene expression changes; hence, the nature and extent of toxicity can be deduced by examining gene expression. Although interpretation of our microarray data supports the above assumption, there will certainly be cases that do not fit this hypothesis. A generic approach to understanding all compounds will only be accomplished when a variety of molecular strategies are integrated, such as genomics, proteomics, metabonomics, etc. It is critical to point out that generating this mass of data is prone to overinterpretation. Once specific patterns of gene expression are elucidated from the extensive microarray data, one can apply the learning set to risk assessment during drug development. However, presently, a complete interpretation of microarray data (especially in the presence of numerous undefined ESTs) for a regulatory decision and risk assessment is not possible. Finally, the global gene expression pattern, rather than one single gene, provides the most information regarding the toxicity of a given compound. Attention should be focused on the whole picture, and the application of statistical analyses is strongly recommended to establish key patterns. Furthermore, relying solely on in vitro data should be approached with caution until confirmed in vivo.


    NOTES
 
1 Fax: (203) 798-5797. E-mail: qhuang{at}rdg.boehringer-ingelheim.com. Back


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Ban, M., Hettich, D., and Huguet, M. (1994). Nephrotoxicity mechanism of cis-platinum (II) diamine dichloride in mice. Toxicol. Lett. 71, 161–168.[ISI][Medline]

Bartosiewicz, M., Trounstine, M., Barker, D., Johnston, R., and Buckpitt, A. (2000). Development of a toxicological gene array and quantitative assessment of this technology. Arch. Biochem. Biophys. 376, 66–73.[ISI][Medline]

Bohe, J., Ding, H., Qing, D. P., Yoon, K, Hirschberg, R., Wolfgang, G. H., and Kopple, J. D. (1998). IGF-I binding proteins, IGF-I binding protein mRNA and IGF-I receptor mRNA in rats with acute renal failure given IGF-I. Kidney Int. 54, 1070–1082.[ISI][Medline]

Bompart, G. (1989). Cisplatin-induced changes in cytochrome P-450, lipid peroxidation and drug-metabolizing enzyme activities in rat kidney cortex. Toxicol. Lett. 48, 193–199.[ISI][Medline]

Braxton, S., and Bedilion, T. (1998). The integration of microarray information in the drug development process. Curr. Opin. Biotechnol. 9, 643–649.[ISI][Medline]

Burczynski, M. E., McMillian, M., Ciervo, J., Li, L., Parker, J. B., Dunn, II, R. T., Hicken, S., Farr, S., and Johnson, M. D. (2000). Toxicogenomics-based discrimination of toxic mechanism in HepG2 human hepatoma cells. Toxicol. Sci. 58, 399–415.[Abstract/Free Full Text]

Chao, C. C. (1996). Molecular basis of cis-diamminedichloroplatinum(II) resistance: A review. J. Formos. Med. Assoc. 95, 893–900.[ISI][Medline]

Delmastro, D. A., Li, J., Vaisman, A., Solle, M., and Chaney, S. G. (1997). DNA damage inducible-gene expression following platinum treatment in human ovarian carcinoma cell lines. Cancer Chemother. Pharmacol. 39, 245–253.[ISI][Medline]

De Witt, L. M., Jones, T. W., and Moore, L. (1988). Stimulation of the renal endoplasmic reticulum calcium pump: A possible biomarker for platinum toxicity. Toxicol. Appl. Pharmacol. 92, 157–169.[ISI][Medline]

Elek, J., Park, K., and Narayanan, R. (2000). Microarray-based expression profiling in prostate tumors. In Vivo 14, 173–182.[ISI][Medline]

Farr, S., and Dunn, R. T., II (1999). Concise review: Gene expression applied to toxicology. Toxicol. Sci. 50, 1–9.[Free Full Text]

Fukuishi, N., and Gemba, M. (1989). Use of cultured renal epithelial cells for the study of cisplatin toxicity. Jpn. J. Pharmacol. 50, 247–249.[ISI][Medline]

Gately, D. P., Jones, J. A., Christen, R., Barton, R. M., Los, G., and Howell, S. B. (1994). Induction of the growth arrest and DNA damage-inducible gene GADD153 by cisplatin in vitro and in vivo. Br. J. Cancer. 70, 1102–1106.[ISI][Medline]

Gobe, G. C., Buttyan, R., Wyburn, K. R., Etheridge, M. R., and Smith, P. J. (1995). Clusterin expression and apoptosis in tissue remodeling associated with renal regeneration. Kidney Int. 4, 411–420.

Hanigan, M. H., Gallagher, B. C., and Taylor, P. T., Jr. (1996). Cisplatin nephrotoxicity: Inhibition of gamma-glutamyl transpeptidase blocks the nephrotoxicity of cisplatin without reducing platinum concentrations in the kidney. Am. J. Obstet. Gynecol. 175, 270–274.[ISI][Medline]

Heller, R. A., Schena, M., Chai, A., Shalon, D., Bedilion, T., Gilmore, J., Woolley, D. E., and Davis, R. W. (1997). Discovery and analysis of inflammatory disease-related genes using cDNA microarrays. Proc. Natl. Acad. Sci. 94, 2150–2155.[Abstract/Free Full Text]

Henkels, K. M., and Turchi, J. J. (1999). Cisplatin-induced apoptosis proceeds by caspase-3-dependent and -independent pathways in cisplatin-resistant and -sensitive human ovarian cancer cell lines. Cancer Res. 59, 3077–3083.[Abstract/Free Full Text]

Hoffman, B., and Liebermann, D. A. (1998). The proto-oncogene c-myc and apoptosis. Oncogene 17, 3351–3357.[ISI][Medline]

Howle, J. A., and Gale, G. R. (1970). Cis-dichlorodiammineplatinum (II). Persistent and selective inhibition of deoxyribonucleic acid synthesis in vivo. Biochem. Pharmacol. 19, 2757–2762.

Kowaltowski, A. J., Castilho, R. F., and Vercesi, A. E. (2001). Mitochondrial permeability transition and oxidative stress. FEBS Lett. 495, 12–15.[ISI][Medline]

Kuhlmann, M. K., Burkhart, G., and Köhler, H. (1997). Insight into potential cellular mechanisms of cisplatin nephrotoxicity and their clinical application. Nephrol. Dial. Transplant 12, 2478–2480.[Free Full Text]

Kurota, H., and Yamaguchi, M. (1995). Suppressed expression of calcium-binding protein regucalcin mRNA in the renal cortex of rats with chemically induced kidney damage. Mol. Cell Biochem. 151, 55–60.[ISI][Medline]

Lee, P. D., Giudice, L. C., Conover, C. A., and Powell, D. R. (1997). Insulin-like growth factor binding protein-1: Recent findings and new directions. Proc. Soc. Exp. Biol. Med. 216, 319–357.[Abstract]

Leonard, I., Zanen, J., Nonclercq, D., Toubeau, G., Heuson-Stiennon, J. A., Beckers, J. F., Falmagne, P., Schaudies, R. P., and Laurent, G. (1994). Modification of immunoreactive EGF and EGF receptor after acute tubular necrosis induced by tobramycin or cisplatin. Ren. Fail. 16, 583–608.[ISI][Medline]

Los, G., Benbatoul, K., Gately, D. P., Barton, R., Christen, R., Robbins, K. T., Vicario, D., Kirmani, S., Orloff, L. A., Weisman, R., and Howell, S. B. (1999). Quantitation of the change in GADD153 messenger RNA level as a molecular marker of tumor response in head and neck cancer. Clin. Cancer Res. 5, 1610–1618.[Abstract/Free Full Text]

Medlin, J. F. (1999). Timely toxicology. Environ. Health Perspect. 107, A256–A258.[ISI][Medline]

Nuwaysir, E. F., Bittner, M., Trent, J., Barrett, J. C., and Afshari, C. A. (1999). Microarrays and toxicology: the advent of toxicogenomics. Mol. Carcinog. 24, 153–159.[ISI][Medline]

Okamoto, K., and Prives, C. (1999). A role of cyclin G in the process of apoptosis. Oncogene 18, 4606–4615.[ISI][Medline]

Okuda, M., Masaki, K., Fukatsu, S., Hashimoto, Y., and Inui, K. (2000). Role of apoptosis in cisplatin-induced toxicity in the renal epithelial cell line LLC-PK1. Implication of the functions of apical membranes. Biochem. Pharmacol . 59, 195–201.

Puga, A., Maier, A., and Medvedovic, M. (2000). The transcriptional signature of dioxin in human hepatoma HepG2 cells. Biochem. Pharmacol. 60, 1129–1142.[ISI][Medline]

Safirstein, R. L., Miller, P., Guttenplan, J. B. (1984). Uptake and metabolism of cisplatin by rat kidney. Kidney. Int. 25, 753–758.[ISI][Medline]

Safirstein, R., Winston, J., Goldstein, M., Moel, D., Dikman, S., and Guttenplan, J. (1986). Cisplatin nephrotoxicity. Am. J. Kidney Dis. 8, 356–367.[ISI][Medline]

Schena, M., Shalon, D., Davis, R. W., and Brown, P. O. (1995). Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270, 467–470.[Abstract]

Silkensen, J. R., Agarwal, A., Nath, K. A., Manivel, J. C., and Rosenberg, M. E. (1997). Temporal induction of clusterin in cisplatin nephrotoxicity. J. Am. Soc. Nephrol. 8, 302–305.[Abstract]

Smith, M. L., Kontny, H. U., Bortnick, R., and Fornace, A. J., Jr. (1997). The p53-regulated cyclin G gene promotes cell growth: p53 downstream effectors cyclin G and Gadd45 exert different effects on cisplatin chemosensitivity. Exp. Cell Res. 230, 61–68.

Snijders, A. M., Meijer, G. A., Brakenhoff, R. H., van den Brule, A. J., and van Diest, P. J. (2000). Microarray techniques in pathology: Tool or toy? Mol. Pathol. 53, 289–294.[Abstract/Free Full Text]

Suzuki, C. A., and Cherian, M. G. (1990). The interactions of cis-diamminedichloroplatinum with metallothionein and glutathione in rat liver and kidney. Toxicology 64, 113–127.[ISI][Medline]

Tanaka, T. S., Jaradat, S. A., Lim, M. K., Kargul, G. J., Wang, X., Grahovac, M. J., Pantano, S., Sano, Y., Piao, Y., Nagaraja, R., Doi, H., Wood, W. H., III, Becker, K. G., Ko, M. S. (2000). Genome-wide expression profiling of mid-gestation placenta and embryo using a 15,000 mouse developmental cDNA microarray. Proc. Natl. Acad. Sci. 97, 9127–9132.[Abstract/Free Full Text]

Wang, K., Gan, L., Jeffery, E., Gayle, M., Gown, A. M., Skelly, M., Nelson, P. S., Ng, W. V., Schummer, M., Hood, L., and Mulligan, J. (1999). Monitoring gene expression profile changes in ovarian carcinomas using cDNA microarray. Gene 229, 101–108.[ISI][Medline]

Zhou, H., Miyaji, T., Kato, A., Fujigaki, Y., Sano, K., and Hishida, A. (1999). Attenuation of cisplatin-induced acute renal failure is associated with less apoptotic cell death. J. Lab. Clin. Med. 134, 649–658.[ISI][Medline]

Zimmermann, J., Erdmann, D., Lalande, I., Grossenbacher, R., Noorani, M., and Fürst, P. (2000). Proteasome inhibitor induced gene expression profiles reveal overexpression of transcriptional regulators ATF3, GADD153 and MAD1. Oncogene 19, 2913–2920.[ISI][Medline]