Chem. Res. Toxicol. 2004, 17, 463-470
463
Acute Molecular Markers of Rodent Hepatic Carcinogenesis Identified by Transcription Profiling Jeffrey A. Kramer,*,† Sandra W. Curtiss,† Kyle L. Kolaja,‡,§ Carl L. Alden,†,| Eric A. G. Blomme,‡,⊥ William C. Curtiss,†,# Julio C. Davila,† Carmen J. Jackson,†,+ and Roderick T. Bunch‡,@ Pfizer Corporation, World Wide Safety Science, 800 North Lindbergh Boulevard, St. Louis, Missouri 63167, and 4901 Searle Parkway, Skokie, Illinois 60077 Received November 25, 2003
Currently, the only way to identify nongenotoxic hepatocarcinogens is through long-term repeat dose studies such as the 2 year rodent carcinogenicity assay. Such assays are both time consuming and expensive and require large amounts of active pharmaceutical or chemical ingredients. Thus, the results of the 2 year assay are not known until very late in the discovery and development process for new pharmaceutical entities. Although in many cases nongenotoxic carcinogenicity in rodents is considered to be irrelevant for humans, a positive finding in a 2 year carcinogenicity assay may increase the number of studies to demonstrate the lack of relevance to humans, delay final submission and subsequent registration of a product, and may result in a “black box” carcinogenicity warning on the label. To develop early identifiers of carcinogenicity, we applied transcription profiling using several prototype rodent genotoxic and nongenotoxic carcinogens, as well as two noncarcinogenic hepatotoxicants, in a 5 day repeat dose in vivo toxicology study. Fluorescent-labeled probes generated from liver mRNA prepared from male Sprague-Dawley rats treated with one of three dose levels of bemitradine, clofibrate, doxylamine, methapyrilene, phenobarbital, tamoxifen, 2-acetylaminofluorene, 4-acetylaminofluorene, or isoniazid were hybridized against rat cDNA microarrays. Correlation of the resulting data with an estimated carcinogenic potential of each compound and dose level identified several candidate molecular markers of rodent nongenotoxic carcinogenicity, including transforming growth factor-β stimulated clone 22 and NAD(P)H cytochrome P450 oxidoreductase.
Introduction Two year rodent carcinogenicity studies must be performed prior to registration of many new pharmaceutical agents intended for chronic or intermittent use over 6 months of duration (1). These studies are typically the most expensive studies performed for preclinical safety assessment and are often development rate limiting. They require a significant expenditure of both capital and active pharmaceutical ingredient and the use of large numbers of animals. As such, these studies typically do not occur until very late in the development process, usually long after clinical trials have begun. However, conducting these studies so late in the development of a new chemical entity increases the risk and complexity of the development process. A positive finding in a carcinogenicity study comes after a significant amount * To whom correspondence should be addressed. Tel: 314-274-2738. E-mail:
[email protected]. † Pfizer Corporation, St. Louis, Missouri. ‡ Pfizer Corporation, Skokie, Illinois. § Current address: Iconix Pharmaceuticals, 325 E. Middlefield Road, Mountain View, CA 94043. | Current address: Millennium Pharmaceuticals, 45 Sidney St., Cambridge, MA 02139. ⊥ Current address: Abbott Laboratories, 100 Abbott Park Rd., Abbott Park, IL 60064. # Current address: Tripos Inc., 1699 South Hanley Road, St. Louis, MO 63144. + Current address: TAP Pharmaceuticals Inc., Lake Forest, IL 60045. @ Current address: Amgen Inc., 1 Amgen Center Dr., Thousand Oaks, CA 91320.
of resources have been expended in support of a compound. Left unaddressed, a positive finding in the 2 year rodent carcinogenicity assay could result in a “black box” carcinogenicity warning on the label of the drug and/or affect the approval of a compound. For this reason, a nongenotoxic carcinogenicity finding can be a major setback to a program and may cause a significant increase in time to market or even the discontinuation of a compound or entire project in some therapeutic areas (2, 3). At present, several efforts have been initiated with the aim of using specific transgenic or knockout mouse lines to replace the standard 2 year mouse carcinogenicity studies with shorter duration studies (4-6). However, even these shorter studies are expensive. None is less than 6 months in duration, and they are only alternatives to the mouse 2 year bioassay and do not afford relief from conducting the bioassay in rat. The application of DNA microarrays and other expression profiling technologies to preclinical safety assessment holds great promise for elucidating mechanisms of intoxication and identifying predictive molecular markers of toxicity (7-9). There have been several reports of the use of microarrays to delineate mechanisms of toxicity for a number of prototypical toxicants (10-12). Additionally, the transcription profile of a target organ in response to treatment with a novel compound may be used to categorize toxicants by toxicologic and/or pharmacologic mechanistic class (13-15). We have applied toxicogenomics to a 5 day repeat dose toxicity study to identify
10.1021/tx034244j CCC: $27.50 © 2004 American Chemical Society Published on Web 03/19/2004
464
Chem. Res. Toxicol., Vol. 17, No. 4, 2004
Kramer et al.
Table 1. Compound Doses and Estimated Carcinogenicity low dose
mid-dose
high dose
compound
mpka
P. Carc
mpk
P. Carc
mpk
P. Carc
Bem Clo Dox MP PB Tam 2-AAF 4-AAF Iso
450 200 35 35 60 25 8 8 35
0.235 0.067 NDb 0.238 0.310 0.052 0.103 0.000 ND
700 400 70 70 100 100 50 50 70
0.353 0.333 ND 0.557 0.361 0.455 0.911 0.000 ND
1000 800 140 140 200 200 100 100 150
0.400 0.667 0.566 0.680 0.800 0.500 1.000 0.000 0.022
a
mpk, mg/kg/day. b ND, not determined.
candidate molecular markers that may predict hepatic carcinogenesis induced by either nongenotoxic or genotoxic compounds. Although there are likely to be numerous mechanisms by which hepatic rodent carcinogenesis is mediated, two broad mechanisms have been proposed. Specifically, a role for oxidative stress over a long duration has been postulated to result in an eventual overwhelming of the organism’s ability to cope, resulting in eventual tumorigenesis (16). Other researchers have suggested that nongenotoxic carcinogens act via a disruption in the balance between proliferation and apoptosis (17, 18). We hypothesized that there might exist a small number of critical genes whose expression may be predictive of the early events of carcinogenesis initiated by multiple mechanisms. Our analyses have identified several candidate genes that can implicate a compound with the potential to produce a positive result in a 2 year rat carcinogenicity study after a short-term repeat dose in vivo toxicology study. This battery of genes has the ability to allow toxicologists to address potential nongenotoxic carcinogenicity issues with appropriate mechanistic studies extremely early in the drug development process, significantly reducing time and money lost as a result of a positive finding.
Experimental Procedures Compound and Dose Selection. The study utilized three dose levels of a number of well-characterized compounds (Table 1), including five nongenotoxic rodent carcinogens (Bem,1 Clo, Dox, MP, and PB), one genotoxic carcinogen (2-AAF), one carcinogen that may act via genotoxicity (Tam), a mitogen (Iso), and a noncarcinogenic toxicant (4-AAF). Bem, a diuretic antihypertensive agent previously under development at G. D. Searle & Company, had its development stopped due in part to a finding of nongenotoxic carcinogenicity (2). A dose of 450 mg/ kg/day increased the incidence of hepatic neoplasia in a 2 year bioassay in rats, while dose-limiting cardiotoxicity was observed at doses of 1000 mg/kg/day and above. Clo, a peroxisome proliferator and nongenotoxic rodent carcinogen, induced liver tumors in Fisher rats at 200 and 400 mg/kg/day (19). Shorter repeat dose studies have indicated that 1 Abbreviations: 2-AAF, 2-acetylaminofluorene; 4-AAF, 4-acetylaminofluorene; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Bem, bemitradine; BDE, balanced differential expression; bP2, balanced Cy5 (treated) probe signal intensity; CD:IGS, cesarean-derived international gold standard; Clo, clofibrate; Con, control; Cy3, cyanine 3 dye; Cy5, cyanine 5 dye; CYP-R, NAD(P)H cytochrome P450 reductase; Dox, doxylamine; dpp, decapentaplegic; fBDE, fractional balanced differential expression; FBW, final body weight; GEM, gene expression microarray; Iso, isoniazid; L2B, liver to body weight; MP, methapyrilene; P1, Cy3 (control) probe signal intensity; PB, phenobarbital; P. Carc, predicted carcinogenic potential; Q-PCR, quantitative polymerase chain reaction; STB, signal to background; Tam, tamoxifen; TGF-β, transforming growth factor-β; TSC22, transforming growth factor-β stimulated clone 22.
higher doses were well-tolerated, so the top dose selected in this study was 800 mg/kg/day. Dox, a nongenotoxic hepatocarcinogen, induced hepatocellular adenomas and carcinomas at doses of 2000 ppm (∼140 mg/kg/day), while doses of 500 and 1000 ppm did not increase hepatocellular tumor incidence (20). Doses of 35 and 140 mg/kg/day of the nongenotoxic carcinogen MP promoted the growth of preneoplastic hepatic focal lesions, and the 140 mg/kg/day dose also led to overt hepatotoxicity (21). The nongenotoxic rodent carcinogen PB induced liver tumors in about 30% of male rats dosed at 30-60 mg/kg/day either in the drinking water or in the diet (22). Previous in-house studies indicated that the maximally tolerated dose was approximately 200 mg/kg/day (via oral gavage). Tam, considered a trans species genotoxic carcinogen (23-25), induced liver tumors at all doses tested (5-35 mg/kg/day). At 25 mg/kg/day, Tam induced hepatocellular carcinoma in rats, while shorter duration studies have measured DNA adducts at doses up to 100 mg/kg/day. 2-AAF is a trans species mutagenic carcinogen (26). In rat bioassays, 2-AAF dosed in the diet induced liver tumors at an incidence approaching 100% at doses up to 8 mg/kg/day in some rat strains (27). Shorter term studies have measured DNA damage at doses up to 100 mg/kg/day. 4-AAF is considered a nongenotoxic noncarcinogen but is a liver promoter (28, 29). It induced no statistically significant increase in liver tumors in a rat bioassay where the top dose was 8 mg/kg/day. Acute studies indicate that doses up to 200 mg/kg/day are reasonably well-tolerated. To compare genetic expression results between 2-AAF and 4-AAF, identical doses were selected for these two compounds. Iso is a nongenotoxic nonhepatocarcinogen but is an acute liver toxicant (30). Doses of 35 mg/kg/day lead to hepatotoxicity in rat, while doses of 100 and 150 mg/kg/day did not increase hepatocarcinogenicity over 2 years of exposure. In Vivo Experiments. All experimental procedures were approved by the Institutional Animal Care and Use Committee and were performed in compliance with laws regarding humane treatment of laboratory animals. Appropriate amounts of test article were suspended in 0.5% methylcellulose (w/v) plus 0.1% polysorbate 80 (v/v) in distilled water. Prior to treatment, male Charles River CD:IGS rats, aged 6-7 weeks, were acclimated for 1 week and then randomly assigned to dosage groups. All animals were housed individually and fed ad libidum in rooms set to maintain 72 ( 5 °F (22 ( 3 °C) and 40% humidity with a 12 h light, 12 h dark cycle. Compounds were administered once daily for 5 days by oral gavage at 10 mL/kg. Animals were not fasted prior to sacrifice. Animals were weighed prior to day 1, on day 3, and on the day of necropsy. Prior to sacrifice on day 5, blood was collected from the abdominal aorta of animals anesthetized with CO2-O2. Changes in serum ALT and AST levels were assessed as a part of a clinical chemistry panel. Livers were collected from each animal and weighed, and then, sections of each were frozen in liquid nitrogen for transcription profiling. Treatment effects were evaluated using analysis of variance on the log transformed data followed by Dunnett’s (onetailed) test for multiple comparisons against a control. The family wise error rate for each compound was held at 5% (R ) 0.05). Transcription Profiling. Total RNA was prepared from frozen livers using RNA-Stat reagent and protocol (Leedo Medical Laboratories, Houston, TX). Poly(A)+ RNA was isolated using micro poly(A)plus kits (Ambion, Austin, TX). Equal amounts of liver mRNA from each control animal were used to prepare a pooled control sample. Aliquots of 300 ng of mRNA from pooled vehicle treated control rats were used to generate complementary DNA labeled with Cy3 dye using Incyte GEM Brite probe labeling kits (Incyte Pharmaceuticals, Palo Alto, CA). Aliquots of 300 ng of mRNA from individual and pooled treated animals were used to prepare Cy5-labeled cDNA probes. Additionally, individual control animal mRNA was labeled with Cy5 and compared to the pooled Cy3-labeled control samples to identify basal animal-to-animal variability. Labeled probes were separated from unincorporated primer using Chromaspin TE-30 columns (Clontech, Palo Alto, CA). Probe hybridization
Molecular Markers of Carcinogenicity
Chem. Res. Toxicol., Vol. 17, No. 4, 2004 465
Table 2. RT-PCR Primer and Probe Sequences gene
forward primer
probe
reverse primer
cyclophilin TSC-22 CYP-R
agagaaatttgaggatgagaacttcat tgcagacagctaggaggctt aaccccccttcgatgctaaaatccatt
tgccatggacaagatgccaggacc catcgccactggtctacaccat cctggctgctgtcaccg
ccttctttcaccttcccaaagac gtctcagttggtaaactcct cgctcagtgccttggttca
was performed at Incyte Pharmaceuticals using RatGEM 1.0, containing ∼7800 cloned rat cDNAs, as previously described (31). Over the course of several days, each individual high dose animal was hybridized onto an array at least once, and many were run more than once (Clo three times, MP, PB, 2-AAF, and 4-AAF twice). Individual low and mid-dose group animals for Clo (run twice), MP, PB, 2-AAF, and 4-AAF were also run. Similarly, pooled samples from the high dose group of each compound were run at least once, and pooled low and mid-dose group animals for most compounds were also run. This resulted in 99 hybridizations of individual or pooled treated rat liver samples and six hybridizations of individual control rat liver samples all compared to the same pooled control. Microarray hybridization data were balanced using total average signal intensity to generate a balance coefficient. A BDE value was generated for every element that met minimum element quality control criteria. Elements with signal intensities P1 + P2 e 500, STB values P1STB + P2STB e 10, or area of coverage less than 40% were recorded as absent values and were omitted from further consideration. Signal intensity was corrected for global background. BDE values were calculated using eq 1
if bP2 < P1, then BDE ) -1 × (P1/bP2), or else BDE ) (bP2/P1) (1) where P1 represents the signal intensity of the Cy3-labeled control sample and bP2 represents the balanced signal intensity of the Cy5-labeled treated sample. A fBDE value was also calculated using eq 2.
fBDE ) (bP2 - P1)/MIN (P1,bP2)
(2)
When calculating a fBDE, the signal intensity in the Cy3 channel (P1, control sample) was subtracted from the balanced signal intensity in the Cy5 (bP2, treated sample) channel. This value was then divided by the smaller of the two signal intensities. Data Analysis. To identify those genes whose expression may correlate with hepatic carcinogenicity, we applied a correlation score method that has been previously described (32). The correlation score was determined for estimated carcinogenic potential using eq 3.
score ) (nQC∧|fBDE|/n) ‚ |r|/P
(3)
This score takes into account three factors, the fraction of chips that met the QC cutoffs described above for the gene in question, and of those, the fraction of the treated rats that displayed changes in expression at least 70% (equivalent to a fold change of (1.7) above or below the pooled control rats (nQC∧|fBDE|/n), the absolute value of the Pearson correlation coefficient (|r|), and the P value associated with the correlation (P). Carcinogenic potential was estimated based upon dosage and published 2 year rodent carcinogenicity assays (33). The incidence (number of animals that developed tumors at the end of the 2 year treatment period divided by the total number of animals in a given study) was expressed as a value between 0 (0% incidence, noncarcinogenic) and 1.0 (100% incidence, most carcinogenic) for the highest dose, an intermediate dose, and the lowest dose reported in 2 year carcinogenicity assays. When possible, studies using male Sprague-Dawley rats were used to calculate the value. The resulting carcinogenic potential values were applied to the high (maximum tolerated), intermediate (minimally toxic), and low (minimally efficacious, or NOAEL) dose levels, respectively, which were utilized in the present study (Table 1). Although the doses used for each compound in the present
study were quite high relative to those used in published 2 year rodent carcinogenicity studies, they were commensurate with doses typical of early, short-term, repeat dose toxicity studies. Upon determining a correlation score for each array element as it related to estimated carcinogenic potential, the genes were sorted by that score. A correlation score was also calculated with respect to each phenotype for the control elements, consisting of empty elements, complex genomic DNA, or yeast control RNA, as for all of the rat cDNA elements. On the assumption that a good correlation would not be expected between the biological phenotype and the control elements, only those genes that scored greater than 1000-fold higher than the highest scoring control element were considered to be relevant. Quantitative Real-Time PCR Analysis. Quantitative real time PCR was performed as previously described (34) using the ABI Prism 7700 sequence detector (Applera Corp., Foster City, CA) to validate changes in the expression of TSC-22 and CYP-R (accession numbers L25785 and M12516, respectively), relative to rat cyclophilin (accession number NM_017101). Aliquots of 400 ng of RNA from the livers of each rat were converted to cDNA in 20 µL reaction volumes using 200 U Superscript II RNase H- Reverse Transcriptase in a 1X buffer (Invitrogen, Carlsbad, CA) containing 10 mM dithiothreitol, 0.5 nM dNTPs, and 2.5 µM oligo d(T)15 primer at 37 °C for 1 h. Amplification was performed in duplicate in 25 µL reaction volumes using 50 ng of cDNA. Cyclophilin and TSC-22 were carried out using 1X PCR buffer A, 2 mM MgCl2, 300 nM forward and reverse primers, 100 nM probe, 200 µM dNTP, and 1 U Amplitaq Gold. CYP-R amplification was performed using 1X PCR buffer A, 3 mM MgCl2, 600 nM forward and reverse primers, 100 nM probe, 200 µM dNTP, and 1 U Amplitaq Gold. Cycle conditions included a 10 min 95 °C hot start followed by 40 cycles consisting of a 15 s 95 °C melting step followed by a 1 min 57 °C annealing and elongation step. Primer and probe sequences are indicated in Table 2.
Results In Vivo Study Results. In general, most compounds were well-tolerated at all doses. One high dose (200 mg/ kg) PB animal died on day 2, but no other mortality was observed. Few clinical signs were noted during the study except reduced activity in PB-treated animals and altered gait in high dose Clo-treated animals. No treatmentrelated gross abnormalities were noted at necropsy. Increased liver/body weight ratios were observed in a dose responsive manner for Bem, Clo, and PB and for the high dose 4-AAF-treated rats (Table 3). Most of the compounds caused modest (
