Prediction of clinically relevant safety signals of nephrotoxicity through plasma metabolite profiling.

Addressing safety concerns such as drug-induced kidney injury (DIKI) early in the drug pharmaceutical development process ensures both patient safety and efficient clinical development. We describe a unique adjunct to standard safety assessment wherein the metabolite profile of treated animals is compared with the MetaMap Tox metabolomics database in order to predict the potential for a wide variety of adverse events, including DIKI. To examine this approach, a study of five compounds (phenytoin, cyclosporin A, doxorubicin, captopril, and lisinopril) was initiated by the Technology Evaluation Consortium under the auspices of the Drug Safety Executive Council (DSEC). The metabolite profiles for rats treated with these compounds matched established reference patterns in the MetaMap Tox metabolomics database indicative of each compound's well-described clinical toxicities. For example, the DIKI associated with cyclosporine A and doxorubicin was correctly predicted by metabolite profiling, while no evidence for DIKI was found for phenytoin, consistent with its clinical picture. In some cases the clinical toxicity (hepatotoxicity), not generally seen in animal studies, was detected with MetaMap Tox. Thus metabolite profiling coupled with the MetaMap Tox metabolomics database offers a unique and powerful approach for augmenting safety assessment and avoiding clinical adverse events such as DIKI.

Tissue and species distribution of the glutathione pathway transcriptome.

The goal of this study was to compare and contrast the basal gene expression levels of the various enzymes involved in glutathione metabolism among tissues and genders of the rat, mouse and canine. The approach taken was to use Affymetrix GeneChip microarray data for rat, mouse and canine tissues, comparing intensity levels for individual probes between tissues and genders. As was hypothesized, the relative expression in liver, lung, heart, kidney and testis varied from gene to gene, with differences of expression between tissues sometimes greater than a 1000-fold. The pattern of differential expression was usually similar between male and female animals, but varied greatly between the three species. Gstp1 appears to be expressed at high levels in male mouse liver, reasonable levels in canine liver, but very low levels in male rat liver. In all species examined, Gstp1 expression was below detectable levels in testis. Gsta3/Yc2 expression appeared high in rodent liver and female canine liver, but not male canine liver. Finally, Mgst1 and Gpx3 expression appeared to be lower in canine heart and testis than seen in rodents. Given the critical role of the glutathione pathway in the detoxification of many drugs and xenobiotics, the observed differences in basal tissue distribution among mouse, rat and canine has far-reaching implications in comparing responses of these species in safety testing.

Comparison of mutant frequencies at the transgenic lambda LacI and cII/cI loci in control and ENU-treated Big Blue mice.

We compared the lambda cII/cI transgenic mutation assay described by Jakubczak et al. [(1996): Proc Natl Acad Sci USA 93:9073-9078] to the previously established Big Blue assay. Genomic DNA isolated from liver, spleen, and lung tissue of control or ethylnitrosourea (ENU)-treated Big Blue mice (100 mg/kg i.p., single dose) was packaged into phage (five animals, two packagings per DNA sample) which were simultaneously plated for lacI and cII/cI mutant frequency (MF) and titer. Mean MF of control animals was higher for cII/cI than lacI for all three tissues examined (spontaneous cII/cI MF divided by spontaneous lacI MF = 2.9, 3.1, and 1.7 for liver, spleen, and lung, respectively). The differences were statistically significant for liver and spleen, but not lung. The ENU-induced MF measured by subtracting control MFs from ENU-treated MFs was higher in the cII/cI assay than lacI (liver = 23.0 x 10(-5) for cII/cI vs. 15.1 x 10(-5) for lacI; spleen = 64.8 x 10(-5) for cII/cI vs. 36.1 x 10(-5) for lacI; lung = 17.1 x 10(-5) for cII/cI vs. 15.8 x 10(-5) for lacI). Fold increase over control values measured by dividing MF of ENU-treated animals by appropriate control values was higher for lacI than cII/cI (liver = 4.4-fold for lacI vs. 2.7 for cII/cI; spleen = 13.1-fold for lacI vs. 8.4 for cII/cI; and lung = 5.6-fold for lacI vs. 4.0 for cII/cI). Despite these differences, overall results were similar for the two mutational endpoints. These results suggest that the cII/cI assay may be an acceptable alternative to lacI where transgenic mutation studies are indicated.

Lack of response to multiple genotoxic agents at the hprt locus in peripheral blood T-lymphocytes of cynomolgus monkeys.

We tested the ability of a series of known genotoxic agents to cause mutations at the hprt locus in peripheral blood T-lymphocytes of cynomolgus monkeys as measured by the ability to form clones in the presence of 6-thioguanine. Ethylmethane sulfonate (EMS, 300 mg/kg i.p.), chloroethylmethane sulfonate (CI-EMS, 35 or 50 mg/kg i.p.), and the Pharmacia & Upjohn antitumor agents adozelesin (1.6, 4, 6, or 8 microg/kg i.v.) and CC-1065 (6 microg/kg i.v.) were all negative in the hprt mutation test. Results with cyclophosphamide (CP, 75 mg/kg i.v.) were equivocal. Adozelesin, CC-1065, and CI-EMS treatments increased the percentage of T-lymphocytes with chromosome aberrations, as well as inducing types of aberrations not seen in control cells. EMS and CP were not tested for chromosome aberrations. We have previously shown that treatment of monkeys with 77 mg/kg ENU substantially increased the hprt mutant frequency, with a lag time of approximately 77 days between treatment and peak MF values. The results of the present study suggest a low sensitivity of the hprt mutation assay to certain classes of genotoxic agents in cynomolgus monkeys.

Spontaneous mutation spectrum at the lambda cII locus in liver, lung, and spleen tissue of Big Blue transgenic mice.

Big Blue mice harbor a recoverable transgene in a lambda/LIZ shuttle vector. In the standard assay, in vivo mutations are measured in the bacterial lacI gene using a labor-intensive color plaque assay. Applying a simpler assay [Jakubczak et al. (1996): Proc Natl Acad Sci USA 93:9073-9078], we measured mutations in the lambda cII gene portion of the transgene. Spontaneous clear plaque mutants were analyzed from liver, lung, and spleen of five untreated mice. Of 314 mutants, 182 (58%) had independent mutations, 74 (23.5%) appeared clonal, and 58 (18.5%) showed no cII mutations. Of 182 independent cII mutations, 156 (85.7%) were base substitutions, 20 (10.9%) were frameshifts, and 6 (3.2%) were multiple substitutions and one deletion. G:C --> A:T transitions were the predominant base substitution (78% of these at CpG sites). The major mutation hotspot, a six G run and its 3' flanking T at bases 179 to 185, comprised 18.7% of the independent mutations. Other hotspots were positions 103, 196, and 212. The in vivo cII spectrum had a significantly higher proportion of G --> A and G --> T mutations and fewer frameshifts than reported in vitro. The cII and published lacI spectra are similar, though G --> A transitions and deletions were fewer in the cII gene. The cI gene was sequenced in 48 mutants with no cII mutations and most had cI mutations: 81.3% base substitutions and 18.7% frameshifts. We conclude that the cII/cI system is insensitive to deletion events, but is useful for detecting point mutations.

Effect of plating medium and phage storage on mutant frequency and titer in the lambda cII transgenic mutation assay.

We examined several experimental parameters of the lambda cI/cII transgenic mutation assay. In the assay, clear plaque lambda phage mutants are identified in a positive selection scheme following rescue of the lambda/LIZ shuttle vector from frozen tissues of Big Blue" transgenic mice. Mutant frequency and titer of phage from various tissues of control and ENU-treated animals was essentially the same on LB or TB1 plating medium, and storage of isolated DNA at 4 degrees C for up to 4 months did not affect either mutant frequency or titer. Storage of packaged phage for 28 days at 4 degrees C did not affect titer. The mean mutant frequency of packaged phage stored 28 days at 4 degrees C was consistently higher than phage plated the same day as packaging (day 0), though the difference was statistically significant in only two of the four samples tested. Reconstruction experiments in which numerically defined titers of known cII mutants were plated on both G1217 and G1225 E. coli strains and incubated at 37 degrees C or 24 degrees C showed highest titers on G1217 at 37 degrees C. The fraction of the G1217, 37 degrees C titer seen in the other strains and conditions varied widely with the cII mutation.

Quantitative reverse transcriptase/PCR assay for the measurement of induction in cultured hepatocytes.

Hepatic enzyme induction is often evoked as the cause for a variety of effects in animal studies, e.g. hepatic hypertrophy and secondary thyroid neoplasms in rodents. In clinical practice, enzyme induction often enhances drug clearance and may result in reduced efficacy. For example, carbamazepine or rifampin treatment induces P450 3A in humans, and as a result, dramatically reduces the efficacy of midazolam or cyclosporine. Due to species differences in substrate specificity and the regulation of various drug-metabolizing enzymes, assessing enzyme induction in human tissues is a desirable goal. Since induction often occurs as a result of increased synthesis of mRNA coding for a particular enzyme, induction may be quantified by measuring specific mRNA levels. We describe an approach to quantifying mRNA levels using reverse transcriptase-polymerase chain reaction (RT-PCR). This approach makes use of either radiolabeled PCR primers or fluorimetric quantification of product and does not require the synthesis of a competitor RNA or DNA molecule. Thus, Cyp2B1/2 mRNA can be shown to be induced 17-fold in the H4IIE rat hepatoma cell line. Likewise, Cyp3A and Cyp2A6 mRNAs can be shown to be induced in primary human hepatocytes cultured on collagen-coated plates and treated with rifampin for 72 h. By contrast, mRNA levels for Cyp1A1 and Cyp2E1 were not increased by rifampin treatment. This report demonstrates the potential of this approach for examining a number of mRNAs from drug-treated cultured cells, as a means of assessing metabolic enzyme induction.

Contribution of serum protein association to discrepancy between the in vivo and in vitro UDS results for 6,7-dimethyl-2,4-di-1-pyrrolidinyl-7H-pyrrolo[2,3-d]pyrimidine (U-89843).

U-89843 has been shown to undergo biotransformation, both in vitro and in vivo, to form U-97924 as a major primary metabolite. U-89843 was found to be positive in an in vitro UDS mutagenesis screen conducted with primary rat hepatocytes in serum-free media. In contrast to in vitro results, no evidence of genetic toxicity of U-89843 was observed in rats in the in vivo/in vitro version of the UDS test with single oral doses up to 1400 mg/kg. The negative results may be related to more robust in vivo detoxification mechanisms or relatively lower exposure to reactive metabolites formed by bioactivation of U-89843 as compared to that observed in the serum-free in vitro hepatocyte test system. Further studies showed rat serum suppressed the in vitro metabolism of U-89843 as well as the formation of the corresponding hydroxylated metabolite, U-97924, the putative precursor of proposed reactive electrophilic metabolite. The measured in vivo systemic clearance of U-89843 (0.53 l/h/kg) in rats was about 1000-fold slower than the in vitro intrinsic clearance (606 l/h/kg) estimated by measuring the formation of U-97924 in rat liver microsomal incubations. Since U-89843 is extensively associated with serum proteins a poor extraction ratio into the liver may account for the slower biotransformation of U-89843 in vivo as compared to that exhibited in in vitro serum-free hepatocyte incubations. Addition of bovine serum albumin (1-40 mg/ml) to the in vitro UDS assay medium decreased the UDS mean net grains per nucleus response of U-89843. These results suggest that the effect of serum protein should be considered when comparing serum-free in vitro UDS and in vivo UDS results for highly serum protein bound compounds.

Excision of DNA adducts of nitrogen mustards by bacterial and mammalian 3-methyladenine-DNA glycosylases.

Nitrogen mustards are among the DNA alkylating agents most widely used in chemotherapy. The homogeneous Escherichia coli AlkA protein (3-methyladenine-DNA glycosylase II) is shown to excise damaged guanine and adenine bases from DNA modified by mechlorethamine, uracil mustard, phenylalanine mustard and chlorambucil, and less efficiently acridine mustard adducts. Homogeneous recombinant human and rat 3-methyladenine-DNA glycosylases excise adducts formed by nitrogen mustards less efficiently than the AlkA protein. In addition to the in vitro excision of adducts, the AlkA protein eliminates cytotoxic mechlorethamine adducts from DNA in vivo.

Correlation of frameshift mutagenicity with DNA intercalation by CGS 20928A using an in vitro DNA unwinding assay.

A compound's mutagenicity in different Salmonella tester strains can suggest its mechanism of reaction with DNA. Clear confirmation of such a mechanism, however, requires a direct test of the compound's reaction with DNA, often relying on specific in vitro studies. We report the use of a rapid in vitro test designed to measure DNA unwinding, a characteristic of DNA intercalators and many frameshift mutagens. CGS 20928A, an adenosine antagonist, produced a significant (> 2-fold) increase in revertants only for Salmonella tester strain TA1537, and only without metabolic activation. These data indicated that the compound was a direct acting frameshift mutagen and possibly intercalated into DNA. Our DNA unwinding assay indicated that at concentrations of > 0.1 mM CGS 20928A behaved like known intercalating compounds in that it unwound DNA. These concentrations of compound are comparable to those found mutagenic to TA1537. By comparison, the frameshift mutagen and known intercalating compound 9-aminoacridine unwound DNA in this assay in a concentration dependent fashion between 6-12 microM. ICR-191, another acridine frameshift mutagen, also unwound DNA. A compound structurally related to CGS 20928A, which was not mutagenic in Salmonella tester strains, did not produce any DNA unwinding even at 10 mM. Because the assay uses microgram quantities of material, it should be ideal for screening small amounts of congeneric series suspected of frameshift mutagenicity.

An application of 3D-QSAR to the analysis of the sequence specificity of DNA alkylation by uracil mustard.

The sequence specificity of DNA alkylation by uracil mustard was examined using a novel three-dimensional QSAR method known as HASL, or the hypothetical active site lattice. The structures of a variety of 4-mer sequences obtained from pBR322 and SV40 were related to their degree of guanine-N7 alkylation by uracil mustard. The resulting correlations were found to point to a significant contribution from bases on the 3' side of the target guanine nucleotide. The HASL models derived from the analysis of 52 guanine-containing 4-mer sequences were used to highlight those atomic features in the favored TGCC sequence that were found most important in determining specificity. It was found that the NH2-O systems present in the two GC base pairs on the 3' side of the target guanine were significantly correlated to the degree of alkylation by uracil mustard. This finding is consistent with a prealkylation binding event occurring between these sites along the major groove and the uracil mustard O2/O4 system.

Use of [8-3H]guanine-labeled deoxyribonucleic acid to study alkylating agent reaction kinetics and stability.

Alkylation at the N7 position of guanine in DNA renders the C8-hydrogen acidic. This serves as the basis for an assay of guanine N7 alkylation using [8-3H]-guanine-labeled DNA. I modified the assay by preparing a high specific activity substrate in vitro and by replacing the distillation step with charcoal adsorption of substrate. Using the appearance of noncharcoal-adsorbable label as a measure of guanine-N7 alkylation I examined the reaction of DNA with dimethyl sulfate and mechlorethamine. The rate of reaction of dimethyl sulfate with the N7 position of guanine in DNA was constant over time, i.e., loss of label from DNA proceeded linearly with time. On the other hand, the rate of reaction of mechlorethamine with DNA increased with time, consistent with the initial formation of the reactive aziridinium ion. The assay can also be used to compare the reaction rates of various alkylating agents with DNA. Thus, the acridine mustards ICR-170 and quinacrine mustard were far more potent alkylating agents than mechlorethamine. Furthermore the assay may be used to determine the alkylating potency and stability of various alkylating agent preparations: while frozen solutions of acridine mustards in organic solvents retained alkylating activity for several months, different commercial preparations of quinacrine mustard had little or no alkylating activity.

alpha-Naphthyl butyrate carboxylesterase activity in human and rat nasal tissue.

Carboxylesterase activity in the rat nasal mucosa plays an important role in the response of this tissue to certain toxic inhaled esters. We have examined this activity in extracts of both Fischer-344 rat and human nasal tissue using the substrate alpha-naphthyl butyrate, the same substrate as that used for histochemical analysis of this activity. We find substantially higher activity in rat nasal extracts than in human nasal extracts. The Michaelis constant (Km), however, is approximately the same for both activities and is significantly less than that reported for rat nasal carboxylesterase activity using dibasic esters as a substrate. p-Nitrophenyl butyrate is a competitive inhibitor of the rat alpha-naphthyl butyrate esterase, but surprisingly has no effect on the human activity. The assay reported here should prove a powerful tool in the development of a valid in vitro nasal toxicity assay system that uses cultured rat or human cells expressing carboxylesterase activity.

Lesion selectivity in blockage of lambda exonuclease by DNA damage.

Various kinds of DNA damage block the 3' to 5' exonuclease action of both E. coli exonuclease III and T4 DNA polymerase. This study shows that a variety of DNA damage likewise inhibits DNA digestion by lambda exonuclease, a 5' to 3' exonuclease. The processive degradation of DNA by the enzyme is blocked if the substrate DNA is treated with ultraviolet irradiation, anthramycin, distamycin, or benzo[a]-pyrene diol epoxide. Furthermore, as with the 3' to 5' exonucleases, the enzyme stops at discrete sites which are different for different DNA damaging agents. On the other hand, digestion of treated DNA by lambda exonuclease is only transiently inhibited at guanine residues alkylated with the acridine mustard ICR-170. The enzyme does not bypass benzo[a]-pyrene diol epoxide or anthramycin lesions even after extensive incubation. While both benzo[a]-pyrene diol epoxide and ICR-170 alkylate the guanine N-7 position, only benzo[a]-pyrene diol epoxide also reacts with the guanine N-2 position in the minor groove of DNA. Anthramycin and distamycin bind exclusively to sites in the minor groove of DNA. Thus lambda exonuclease may be particularly sensitive to obstructions in the minor groove of DNA; alternatively, the enzyme may be blocked by some local helix distortion caused by these adducts, but not by alkylation at guanine N-7 sites.

DNA damage and sequence specificity of DNA binding of the new anti-cancer agent 1,4-bis(2'-chloroethyl)-1,4-diazabicyclo-[2.2.1] heptane dimaleate (Dabis maleate)

The DNA damage and the sequence specificity of guanine-N7 alkylation produced by the novel, positively charged, antineoplastic agent 1,4-bis(2'-chloroethyl)-1,4-diazabicyclo-[2.2.1] heptane dimaleate (Dabis maleate) and its uncharged tertiary amine analogue 1,4-bis(2'-chloroethyl)-1,4-diazacyclohexane (Dabis analogue) were investigated in L1210 cells and isolated DNA. Both compounds are cytotoxic in vitro causing an arrest of L1210 cells in G2/M phase of the cell cycle. In isolated DNA, Dabis maleate alkylates guanine at the N7-position with some differences in specificity compared to other alkylating agents (e.g. nitrogen mustard). Significant differences are also evident between Dabis maleate and Dabis analogue, suggesting that Dabis analogue is not the sole alkylating species of Dabis maleate. Using the alkaline elution technique a moderate number of DNA interstrand cross-links were detected in L1210 cells treated with both compounds, which were completely repaired within 24 h. Dabis maleate and Dabis analogue do not cause DNA single strand breaks or DNA protein cross-links at the doses at which DNA interstrand cross-links were detected.

GC-rich regions in genomes as targets for DNA alkylation.

For many DNA-damaging agents, the extent of damage at any given base site is influenced by the DNA sequence surrounding that site. Most agents that alkylate the guanine N7 position, including mechlorethamine (nitrogen mustard) and benzo[a]pyrene diol epoxide, alkylate oligo-guanine sequences preferentially. Since these data suggest that guanine-cytosine(GC)-rich regions in genes could be preferred sites of damage by these agents, GenBank was searched for genes containing 30 bp sequences of greater than 90% GC (GC runs). While primate, rodent, other mammalian, vertebrate and animal virus genes constituted 57% of the annotated entries, they included 90% of the entries with the GC runs. In addition, the percentage of oncogenes in the group of the entries with GC runs was higher than that in the overall database. One gene of interest containing GC runs was the human c-Ha-ras oncogene. All seven GC runs in the c-Ha-ras gene are in the 5'-flanking region, rather than in the coding sequences. In fact, some of the GC runs are contained in Sp1-binding enhancer sequences. Gel analysis of the alkylation of cloned c-Ha-ras DNA by several carcinogenic alkylating agents strongly suggest that in this gene GC runs can be preferred sites of damage. These observations suggest mechanisms by which DNA damage at sites other than oncogene coding sequences may play a role in carcinogenesis and/or chemotherapy.

DNA sequence specificity of guanine N7-alkylations for a series of structurally related triazenes.

The base sequence selectivity for reaction at the guanine-N7 position was examined for a series of structurally related triazenes by a modification of a standard DNA sequencing method. The monomethyl and monochloroethyl triazenes alkylate guanines extensively at the N7 position with a general preference for runs of contiguous guanines, similar to, but not as striking as that observed previously for the chloroethylnitrosoureas. In contrast to the nitrosoureas, the triazenes had patterns of base sequence selectivity that differed somewhat from agent to agent, with the monochloro-ethylphenyltriazene having the pattern most different from the others in the series. Thus, the nature of the nonalkylating portion of the molecule can influence the ultimate alkylation preference. The monoethylating analogues alkylated weakly with little sequence preference, and the dimethyl analogues were essentially unreactive in this system.

DNA sequence specificity of antitumor agents. Oncogenes as possible targets for cancer therapy.

An examination of the DNA sequence specificity of guanine-N7 alkylation for nitrogen mustards and chlorethylnitrosoureas revealed that large variations in alkylation intensities existed among different guanines in the DNA sequence. The most striking finding was that most agents reacted preferentially at runs of G's, the degree of preference being much greater than would be expected from the number of G's alone. This correlated with the molecular electrostatic potential induced at the guanine-N7 position by the nearest neighbor base pairs. Uracil and quinacrine mustards, however, showed distinctly different reaction patterns from other mustards and a detailed examination has led to structural hypotheses to account for the differences. Certain regions of the genome including regions in some oncogenes and the Epstein-Barr virus have unusually high GC contents (greater than 80% GC) which suggests that the antitumor effectiveness of alkylating agents may in part be due to selective reaction at certain regions in the genome. In fact certain mustards have been shown to exhibit enhanced reactivities with such regions in DNA fragments derived from the c-H-ras oncogene. The above findings point to the possibility of design of alkylating agents to optimise the selectivity of reaction with critical DNA regions. An alternative approach presently under investigation has emerged from an understanding of the characteristics of the sequence specific interaction of the natural oligopeptide antibiotics netropsin and distamycin in the minor groove of DNA. This has led to the synthesis of novel agents (lexitropsins) in which the binding specificity can be shifted from (AT)n in (GC)n in a predictable fashion. Thus the rational design of DNA sequence specific vectors linked to DNA reactive groups, such as alkylating or cleaving agents, could enable DNA damage to be delivered selectively to predetermined critical sites on the genome.