Targeted Disruption of the Microsomal Epoxide Hydrolase Gene

Microsomal epoxide hydrolase (mEH) is a conserved enzyme that is known to hydrolyze many drugs and carcinogens, and a few endogenous steroids and bile acids. mEH-null mice were produced and found to be fertile and have no phenotypic abnormalities thus indicating that mEH is not critical for reproduction and physiological homeostasis. mEH has also been implicated in participating in the metabolic activation of polycyclic aromatic hydrocarbon carcinogens. Embryonic fibroblast derived from the mEH-null mice were unable to produce the proximate carcinogenic metabolite of 7,12-dimethylbenz[a]anthracene (DMBA), a widely studied experimental prototype for the polycylic aromatic hydrocarbon class of chemical carcinogens. They were also resistant to DMBA-mediated toxicity. Using the two-stage initiation-promotion skin cancer bioassay, the mEH-null mice were found to be highly resistant to DMBA-induced carcinogenesis. In a complete carcinogenesis bioassay, the mEH mice were totally resistant to tumorigenesis. These data establish in an intact animal model that mEH is a key genetic determinant in DMBA carcinogenesis through its role in production of the ultimate carcinogenic metabolite of DMBA, the 3,4-diol-1,2-epoxide.

Microsomal epoxide hydrolase (mEH) 1 is a critical phase I biotransformation enzyme that catalyzes hydrolysis of a large number of epoxide intermediates (1,2). mEH is highly conserved in different mammalian species, is expressed in the embryo (3)(4)(5) and multiple organs, and is active toward some endogenous epoxy-steroids (6) and bile acids (7) thus suggest-ing that it plays a critical physiological role. mEH is usually thought to play a pivotal role in protection against the toxicity of reactive epoxide intermediates, because metabolism of epoxides by this enzyme results in the production of less reactive and less toxic dihydrodiol intermediates of drugs such as phenytoin and carbamazepine (8,9) and epoxides of environmental toxins (10,11). In contrast to this protective effect, mEH is thought to be required for the metabolic activation of the potent carcinogen 7,12-dimethylbenz[a]anthracene (DMBA), a widely studied experimental prototype for the polycylic aromatic hydrocarbon class of chemical carcinogens (12). P450s and mEH metabolize DMBA to both inert metabolites and metabolites that are electrophilic and capable of producing DNA adducts (Fig. 1). Cytochrome P450 CYP1B1 oxidizes DMBA to the 3,4epoxide (13). This is followed by hydrolysis of the epoxide by mEH to the proximate carcinogenic metabolite, DMBA-3,4diol. This metabolite can be further oxidized by either CYP1A1 or CYP1B1 to the principal ultimate carcinogenic metabolite, DMBA-3,4-diol-1,2-epoxide, that is capable of producing DNA adducts (14 -17). Other ring hydroxylations and methyl hydroxylations of DMBA result in inactive metabolites that do not bind DNA. Based on this scheme, mEH should be a critical enzyme in the pathway leading to the carcinogenic activity of DMBA. However a role for mEH in DMBA carcinogenesis has not been established in an intact animal model. DMBA carcinogenesis bioassays have been established using mouse skin with either the two stage initiation-promotion protocol in which DMBA is initially applied followed by repeated applications of 12-O-tetradecanoylphorbol-13-acetate (TPA), or the complete carcinogenesis assay where DMBA is continuously applied (18 -20). In the two stage carcinogenesis model, metabolites of DMBA cause DNA adducts and TPA administration results in cell proliferation that serves to fix the mutations in cell cycle control genes leading to selection of rapidly dividing cancer cells. In the complete carcinogenesis scheme, DMBA serves as both an initiator and promoter of cell growth. It is not known whether the mechanism of carcinogenesis differs between the two protocols. The mouse skin cancer bioassays can be used to determine the effects of genes on DMBA carcinogenesis.
In this report, mEH-null mice were produced and found to have no phenotype, indicating that in mice, mEH is not essential for reproduction and physiological homeostasis. Using the two widely used skin cancer bioassays discussed above, mEHnull mice were found to be highly resistant to DMBA-induced carcinogenesis.

EXPERIMENTAL PROCEDURES
Isolation and Sequencing of the cDNA and Gene-The mouse mEH cDNA was isolated from a liver cDNA library (Stratagene, La Jolla, CA) using the rat mEH as a probe (21). Mouse soluble EH (sEH) cDNA was * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  isolated by reverse transcription-polymerase chain reaction. Total RNA was isolated from livers of 8-week-old male C57BL/6 mice using the Ultraspec RNA reagent according to the manufacturer's protocol (Biotecx Laboratories, Houston, TX). Primers were generated from the 5Ј (GGGAATTCATGGCGCTGCGTGTAGCCGCG) and 3Ј (GGGAATTC-CTAAATCTTGGAGGTCACTGA) ends of the sEH cDNA sequence (22) and were used to amplify the cDNA using the cDNA Cycle Kit (Invitrogen, Carlsbad, CA). The resultant cDNA was completely sequenced using an Applied Biosystems Model 377 sequencer (Foster City, CA). The mEH genomic clones were obtained from a mouse 129/Sv Lambda FIX genomic library (Stratagene) using the mouse mEH cDNA as a probe.
Construction of the Targeting Vector-A 5.5-kbp BamHI fragment containing exon 2 was isolated from the genomic clone and subcloned into pUC18. To disrupt the gene, an ApaI fragment containing a part of exon 2 was deleted and replaced with the bacterial phosphoribosyltransferase II gene cassette (TK-NEO) derived from the pMC1Neo poly(A) vector including ApaI adapters (Stratagene). A herpes simplex virus thymidine kinase gene was inserted at the 3Ј end of mEH gene for use in negative selection with gancyclovir.
Production of Chimeric Mice-The plasmid DNA used for targeting was purified by banding twice on cesium chloride gradients, linearized with SalI and introduced into embryonic stem (ES) cells purchased from Genome Systems Inc. (St. Louis, MO) by electroporation at 250 V, 250 F capacitance using a Bio-Rad gene pulser. One day after electroporation, the ES cells were incubated in medium containing G418 (300 g/ml) and gancyclovir (5 M) as positive and negative selectable markers, respectively. The surviving ES cell clones were analyzed by Southern blotting using a probe corresponding to exon 3 and exon 4 of the mEH gene. The probes were derived from the mEH cDNA. The TK-NEO gene was used to identify specific homologous recombinants and eliminate clones having other nonspecific integration of the targeting construct. The ES cell clones heterozygous for the disrupted mEH gene were injected into C57BL/6 blastocysts to generate chimeric founder mice as described previously (23). The injected blastocysts were transferred into the uterus of pseudopregnant recipient NIH Swiss females to produce chimeric mice. Male chimeras were bred with C57BL/6 females to determine whether germ line transmission had occurred.
The heterozygous mice, having germ line transmission of the targeted mEH allele, were interbred to generate homozygotes.
Genotyping of the ES Cells and Mice-Genomic DNA was isolated from ES cells and mouse tails (24) and digested with XbaI and EcoRI. The digested DNA was subjected to electrophoresis in 0.6% agarose gels, transferred to a Gene Screen Plus nylon membrane (NEN Life Science Products) using 0.4 N NaOH, and hybridized with the mEH probe.
Analysis of mEH Expression-Mouse liver tissue was homogenized in Ultraspec RNA reagent (Biotecx Laboratories). Total RNA (10 g) was denatured and subjected to electrophoresis in 1% agarose gels containing 2.2 M formaldehyde and blotted to Gene Screen Plus nylon membranes. mEH and sEH cDNAs were labeled with [␣-32 P]dCTP using a random primer polymerase kit (Amersham Pharmacia Biotech). The conditions used for prehybridization, hybridization, and washing were previously described (23). Liver microsomal protein (10 g) was subjected to SDS-polyacrylamide gel electrophoresis. Protein was electroblotted to nitrocellulose membrane (Schleicher & Schuell), and the immune complexes were detected by use of ECL reagent (Amersham Pharmacia Biotech). Rabbit antibody prepared against rat mEH was kindly provided by Dr. James P. Hardwick (Northeastern Ohio University College of Medicine).
DMBA Toxicity and Metabolism in Embryo-derived Fibroblasts-Mice at gestation day 14 were euthanized by carbon dioxide asphyxiation, and the embryos were placed in phosphate-buffered saline (pH 7.4). Following removal of internal organs and head, the remaining torsos were minced and placed into 2 ml of 0.25% trypsin for 45 min at 37°C. The reaction was stopped by addition of incubation medium (Dulbecco's modified Eagle's medium with 10% fetal bovine serum, 100 g/ml penicillin, 100 g/ml streptomycin, 250 g/ml amphotericin B and 2 mM glutamine). The cells were grown at 37°C in 5% carbon dioxide until confluent, trypsinized, and stored at Ϫ80°C until use. The cells were thawed before use, grown for 1 day, trypsinized, and seeded into 96-well plates at a density of 3,000 cells/well in 100 l of medium. Following incubation for 1 day, a series of dilutions of DMBA in 100 l of medium were added to the cells and incubation continued for 3 days. The cells were fixed with 10% trichloroacetic acid, washed, and stained with sulforhodamine B (25). mCi/mmol), purified by reversed-phase high performance liquid chromatography to remove impurities and dissolved in Me 2 SO, was incubated with 1.4 ϫ 10 6 embryo-derived fibroblast (EF) cells in 5 ml of medium (the final concentration of DMBA was 0.3 M) for 24 h. The cells were mechanically harvested and extracted with 1 volume of cold acetone, and the aqueous phase was extracted twice with 10 ml of ethyl acetate. The combined organic phase was dried, and the residues were analyzed by reversed-phase high performance liquid chromatography with a radio-detector (26). Comparing retention times with authentic compounds identified the metabolites formed. 12-Hydroxymethyl-7-methylbenz[a]anthracene (7M,12OH-DMBA); DMBA-3,4-dihydrodiol (3,4-diol); 2-,3-, and 4-hydroxybenz[a]anthracene (DMBA 2-, 3-, 4-phenols) were obtained from the NCI Chemical Carcinogen Repository (Frederick, MD). DMBA 8, was formed by incubation of DMBA with human recombinant CYP1A1 (Gentest Corp., Woburn, MA) (26,27).
Skin Carcinogenesis-The skin tumor-induction experiments were carried out using the initiation-promotion protocol as described (18 -20). Female mEH-null and wild-type mice, aged 8 -11 weeks were shaved with electric clippers 2 days before treatment with DMBA (Sigma). The mice were initiated by treating once with 25 g of DMBA in 200 l of acetone, and 1 week later, they were treated with 5 g of TPA (Mallinckrodt Baker, Inc., Paris, KY) in 200 l of acetone. TPA was administered three times weekly for 25 weeks. As a control, wild-type mice were initiated with only acetone and promoted with TPA. For the complete carcinogenesis experiments, the mice received weekly applications of 12.8 g of DMBA in 200 l of acetone for 25 weeks. Skin papillomas greater than 1 mm in diameter were counted and the incidence was monitored weekly.

RESULTS
Generation and Characterization of mEH-Null Mice-To investigate the developmental and physiological function of mEH and its role in chemical carcinogenesis, mEH-null mice were generated using the classical strategy of homologous recombination in ES cells. The mouse mEH gene was isolated and characterized and found to span 29 kbp ( Fig. 2A) and have 9 exons (Table I), similar to the human and rat mEH genes (28,29). The splice junctions and nucleotide sequences in each exon were highly conserved between the human and mouse mEH genes. The mouse mEH cDNA was isolated from a mouse liver cDNA library, and the nucleotide sequences of the exon portions in the mouse mEH gene were identical to those of the cDNA.
A 5.5-kbp BamHI fragment containing exon 2 was used to make a vector for targeted gene disruption. A 0.7-kbp ApaI fragment containing a part of exon 2 was replaced with 1.1-kbp TK-NEO cassette encoding the NEO selectable marker (Fig.  2B). To allow the use of gancyclovir for negative selection, the herpes simplex virus thymidine kinase gene expression cassette was inserted into 3Ј end of the targeting vector. The linearized targeting construct was electroporated into ES cells, and four ES cell homologous recombinant clones were found out of the 220 G418-and gancyclovir-resistant clones analyzed. Southern blot analysis demonstrated that heterozygous ES clones were obtained as indicated by the presence of an additional 7.7-kbp fragment obtained upon double digestion with XbaI and EcoRI (data not shown). When these clones were hybridized with the TK-NEO gene as a probe, only a single hybridizing band of 7.7-kbp with XbaI-EcoRI double-digested DNA was detected, demonstrating that these clones did not contain any additional random integrations of the targeting construct. One of these homologous recombinant ES cell clones was injected into C57BL/6 blastocysts and seven male chimeric mice were generated; one of these yielded germline transmission. Heterozygous (mEHϩ/Ϫ) mice were normal and used for interbreeding to generate homozygous (mEHϪ/Ϫ or mEH-null) mice. Breeding of the heterozygous mice produced offspring at the expected Mendelian distribution indicating that no in utero lethality occurs as a result of loss of the functional mEH gene. The mEHϩ/Ϫ and mEHϪ/Ϫ mice DNAs were analyzed for the presence of mutant alleles of the mEH gene (Fig. 2C). The absence of a diagnostic 10.5-kbp band and the presence of a 7.7-kbp band in the offspring clearly indicated that the mEHnull mice were born and viable. mEH-null mice exhibited no differences in weight, development, fertility and behavior when compared with their wildtype littermate controls. The finding of no effect on fertility and reproduction was quite surprising, because mEH is expressed in the ovary and testis (30). Histological examination of several other organs, including liver, lung, kidney, intestine, spleen, thymus, heart, and brain, revealed no difference between mEH-null and wild-type mice, indicating that mEH is not a critical requirement for normal development and physiology. This finding of no apparent role in development and physiological homeostasis was of interest, because the enzyme is highly conserved at the level of amino acid sequence similarity (31)(32)(33) and substrate specificity in different mammalian species, and it is expressed in the embryo and multiple organs (3)(4)(5).
To evaluate the level of mEH mRNA, Northern blot analysis was performed (Fig. 3). As a control, sEH mRNA was also examined. sEH mRNA levels were not different among the mEH-null, mEHϩ/Ϫ, and wild-type mice. In the mEH-null mouse liver, a lower abundance RNA band (that was of slightly higher mobility than that seen in the wild-type mice) was detected using the mEH cDNA as a probe (Fig. 3). This transcript is probably derived from the disrupted allele although it is not certain how the TK-NEO cassette would alter splicing to produce a smaller mRNA. The lower level of expression of this transcript is not surprising, because mRNAs that do not encode a normal protein are typically unstable and rapidly degraded (34,35). To further confirm that mEH protein is not expressed in the mEH-null mice, immunoblot analysis was performed using rabbit antibody against rat mEH. mEH protein was not detected in the mEH-null mouse liver microsomes (Fig. 4). The data confirm that the mEH is not expressed in the mEH-null mice.
The Role of mEH in DMBA Metabolism and Toxicity-To determine whether lack of the mEH affects the ability to metabolically activate DMBA, EF were prepared from wild-type and mEH-null mice and subjected to incubation with increasing concentrations of DMBA (Fig. 5). Cells that metabolically activate DMBA to electrophilic metabolites will be killed. Wildtype EF were killed by DMBA at less than 1 M. In contrast, mEH-null EF were resistant to up to 10 M of the carcinogen. These data suggest that cells lacking mEH are unable to catalyze the metabolic activation of DMBA. This was confirmed by direct analysis of DMBA metabolism by EF in situ. EF from wild-type mice were able to form the 3,4-diol metabolite that results from P450 oxidation to the 3,4-epoxide followed by hydrolysis to the diol by mEH (Table II). In contrast, no detectable DMBA-3,4-diol was formed in cells from mEH-null mice. These EF cells also failed to produce the 8,9-diol and 10,11-diol, whereas they were able to generate the phenol metabolites that result from hydrolysis of epoxides by water. These data indicate that cells from mEH-null mice are unable to produce the proximate carcinogenic metabolite of DMBA, the 3,4-diol, that FIG. 3. Northern blot analysis of liver RNA. Total liver RNAs (10 g) were subjected to Northern blotting. The mouse mEH and sEH cDNAs were used as probes. ϩ/ϩ, wild-type mice; ϩ/Ϫ, heterozygous mice; Ϫ/Ϫ, homozygous mEH-null mice.

FIG. 4. Western blot analysis of liver microsomes.
Liver microsomal proteins (10 g) were subjected to SDS-polyacrylamide gel electrophoresis, transferred to nylon membranes, incubated with rabbit antibody, and the bands visualized using the ECL reagent. The ECL reagent was used to develop the blots. ϩ/ϩ, wild-type mice; ϩ/Ϫ, heterozygous mice; Ϫ/Ϫ, homozygous mEH-null mice.  is the precursor to the 3,4-diol-1,2-epoxide ultimate carcinogenic metabolite of DMBA. The Role of mEH in Susceptibility to Skin Cancer-To test the role of mEH in DMBA-induced carcinogenesis, skin cancer bioassays were carried out comparing the mEH-null with wildtype mice (Fig. 6). The mEH-null allele is on a 129/Sv genetic background, and this stain is known to be sensitive to DMBAinduced skin carcinogenesis (36). In a standard initiation-promotion protocol, wild-type mice started to develop papillomas after 6 weeks of TPA treatment, and the percentage of mice with tumors increased to 70% by 15 weeks (Fig. 6A). This value is within the range of that found by others (37). The mean number of papillomas per mouse at 15 weeks of treatment was 4 (Fig. 6B). In contrast, papillomas first appeared in mEH-null mice at 9 -10 weeks and the percentage of mice with tumors reached 20% at 17 weeks. The mean number of tumors per mouse in this group was 0.5 (Fig. 6B). These data indicate that lack of mEH results in a marked resistance to DMBA-induced skin cancer thus demonstrating in an intact animal model that mEH is a major determinant for susceptibility to DMBA-induced cancer. In contrast to the initiation-promotion assay, mEH-null mice were completely resistant to skin cancer when a complete carcinogenesis assay was performed in which only DMBA was applied. After 25 weeks of treatment, almost 80% of wild-type mice had papillomas, whereas none of the mEH-null mice had even a single detectable nodule (Fig. 6, C and D). This is strong evidence that there are significant mechanistic differences between the initiation-promotion skin cancer bioassay and the complete carcinogenesis assay, which involve metabolic activation of DMBA. DISCUSSION There are a large number of enzymes that are involved in the detoxification of carcinogens including microsomal and soluble epoxide hydrolases, UDP-glucuronosyltransferases, glutathione S-transferases, and diaphorase. Although mEH is gener-FIG. 6. Skin cancer induction by DMBA. Skin cancer bioassays with wild-type (E) and mEH-null (q) mice. In the DMBA initiation and TPA promotion study (panels A and B), the number of mice per group was 24 for wild-type, 26 for mEH-null, and 20 for control. Controls (‚) were wild-type mice that were not initiated with DMBA. In the DMBA complete carcinogenesis study (panels C and D), the number of mice in each group is 24. Panels A and C show the percentage of mice with papillomas versus time of treatment in weeks. Panels B and D show the mean number of papillomas per mouse versus time in weeks. Each point represents the mean Ϯ S.E. ally considered a detoxification enzyme for arene oxides, its involvement in the hydrolysis of chemical intermediates, leading to the ultimate carcinogenic metabolite, has only been demonstrated in vitro and has not previously been demonstrated in an intact animal model (38). In the complete carcinogenesis protocol, no cancers were found in the mEH-null mice thus establishing that mEH is required for the carcinogenicity of polycyclic aromatic hydrocarbons such as DMBA in vivo. In the initiation-promotion carcinogen bioassay, the null mice did develop papillomas albeit at levels significantly lower than in wild-type mice. These studies indicate a possible mechanistic difference between the complete carcinogenesis and the TPA initiation-promotion bioassays. It is well established that mEH is absolutely required for production of the bay region diol-epoxide of DMBA that is highly mutagenic (12). However, the fact that tumors were found in the mEH-null mice treated with TPA suggests the existence of a second pathway for activation of DMBA that is dependent on TPA. DMBA is known to cause elevated hydrogen peroxide and oxidized bases, particularly 8-hydroxy-2Ј-deoxyguanosine in epidermal DNA, although it is unknown whether this is dependent on P450 oxidation (39). Perhaps the active metabolites derived from this, or another P450 and mEH-independent pathway of DMBA activation such as that mediated by mutagenic sulfate esters (40), result in DNA adducts that are more readily repaired than those adducts produced by the DMBA diol-epoxide metabolite. Fixation of these adducts into mutations may require a strong hyperplastic response such as that found with TPA. These possibilities can now be investigated using the P450 and mEH null mouse models.
Recent studies with the CYP1B1-null mouse have established that CYP1B1 is critical for DMBA carcinogenesis in vivo using an oral route of administration that results in lymphomas in wild-type mice (41). These data support the in vitro evidence showing that CYP1B1 is considerably more active than CYP1A1 in producing the DMBA-3,4-diol proximate carcinogenic metabolite (13). These results establish that mEH is also required for DMBA carcinogenesis and validates the in vivo relevance of the bay region diol-epoxide pathway for metabolic activation of polycyclic aromatic hydrocarbon carcinogens. These studies suggests that genetics differences in expression of P450s and mEH may be critical determinants of cancer susceptibility. In humans, hepatic mEH exhibits a wide range of expression levels between different people (42). Polymorphic allelic variants in the mEH gene have also been described in humans (43,44). Thus, it is thought that interindividual variation in mEH expression in the human population could play a role in cancer susceptibility, teratogenesis, and cytotoxicity (45)(46)(47). These studies will provide a mechanisticbased framework in an intact animal model for future efforts in molecular epidemiology to determine whether genetic polymorphisms in human mEH confer altered sensitivities to the toxic and carcinogenic effects of xenobiotics that damage cells through epoxide intermediates.