A method for in vitro culture of rat Zymbal gland: use in mechanistic studies of benzene carcinogenesis in combination with 32P-postlabeling.

Zymbal glands were excised bilaterally from the ear ducts of female Sprague-Dawley rats (three/group), minced into approximately four fragments per gland, and transferred into a microtiter plate containing 1.5 mL per well of Waymouth's tissue culture medium supplemented with fetal calf serum, hydrocortisone, insulin, and gentamicin. After addition of a test compound or solvent vehicle, plates were incubated for 6, 24, 48, or 96 hr at 37 degrees C in a humidified atmosphere of 5% CO2 in air. Tissue in culture for 6 hr was histologically indistinguishable from the freshly excised tissue, while that in culture for 24, 48, and 96 hr showed a progressive deterioration often with necrosis and/or squamous metaplasia. More pronounced deterioration was noted in samples treated with 750 or 1500 micrograms/mL of benzene. Using a nuclease P1-enhanced 32P-postlabeling assay, aromatic DNA adducts were detected in cultured Zymbal glands exposed for 48 hr to benzene and its derivatives, as well as to 7,12-dimethylbenzanthracene (DMBA) and 2-acetylaminofluorene (AAF). Benzene produced very low levels of adducts (0.5 adducts per 10(9) nucleotides), whereas its congeners produced relatively high levels of adducts (50-2000 lesions per 10(9) nucleotides), which decreased in the order benzoquinone greater than hydroquinone greater than phenol greater than benzenetriol greater than catechol. Each adduct profile overall was characteristic for the compound studied, suggesting the formation of compound-specific electrophiles. AAF and DMBA adducts were identical to those formed in vivo in animals. Our results show that the Zymbal glands are capable of metabolizing different carcinogens to DNA-reactive intermediates, a process that may be causally associated with tumor formation in vivo in this organ.


Introduction
Rat Zymbal glands, located at the base of the external auditory canal, have a distinctive pink-yellow color, weigh about 7 to 8 mg, and have a diameter of 3 to 5 mm. These glands, also present in other rodents and insectivores, but not in humans, have a typical sebaceous morphology, which is characterized by rounded sacs of plump, fatcontaining cells (1). These cells subsequently break down, releasing fat droplets into the external ear canal (1). Maltoni et al. (2) and the National Toxicology Program (3) have reported a dose-dependent fonnation of solld tumors in Zymbal glands of rats following administration of repeated high doses of benzene. The mechanism by which benzene induces this cancer is unknown, but it has been suggested from in vitro and in vivo metabolism studies that benzene is enzymatically activated to reactive intermediates that can bind to DNA (4), thereby forming covalent adducts, which, in turn, may causally be associated with mutation and the initiation of carcinogenesis (5).
The liver is generally considered to be the primary organ for benzene metabolism, but the incidence of liver tumors (hepatomas) has been reported to be three times lower than that of Zymbal gland carcinomas in rats after daily administration of 500 mg/kg benzene for 104 weeks (2). This difference in the susceptibility to carcinogenesis may, in part, derive from the unique metabolic capabilities of Zymbal gland tissue.
In order to facilitate the study of Zymbal gland metabolism, we have devised a culture system to maintain fragments of Zymbal glands for several hours after excision from rats. The capability of the cultured glands to metabolize benzene and its derivatives [catechol, 1,2,4-benzenetriol (BT), phenol, hydroquinone (HQ), and benzoquinone (BQ)], as well as 7,12-dimethylbenzanthracene (DMBA) and 2-acetylaminofluorene (AAF), was then assessed by measuring covalent binding of these compounds to DNA using a highly sensitive 32P-postlabeling assay (6,7). DNA adducts were readily detectable with all compounds tested, except benzene, which produced relatively low levels of adducts. DMBA-and AAF-DNA adduct profiles were similar to those formed in vivo in the epidermis and Zymbal gland, respectively, of treated animals.

Materials and Methods
Benzene was purchased from American Burdick and Jackson, a subsidiary of American Hospital Supply Corporation (Muskegon, MI). Catechol, BT, HQ, and BQ were obtained from Aldrich Chemical Company. Phenol (Mallinckrodt, Cat. No. 0025-2.5) was distilled prior to use. DMBA, AAF, and calf thymus DNA were from Sigma Chemical Company. Waymouth's culture medium and calf serum were from Gibco Laboratories. Carrierfree y-32P-ATP (about 4000 Ci/mmole) was synthesized from 32Pi (8) or purchased from ICN Radiochemicals (Cat. No. 35020). Polyethyleneimine (PEI)cellulose plates were prepared (9) in the laboratory. The sources of all other materials needed for 32P-postlabeling analysis of adducts have been documented previously (6)(7)(8).
For the organ culture experiments, Zymbal glands were aseptically excised from both the ear ducts of female Sprague-Dawley rats (4 months of age, weighing approximately 280 g), minced into four to six fragments per gland, and placed into Waymouth's MB752/1 tissue culture medium supplemented with 10% fetal calf serum (heat inactivated at 570C for 30 min), 350 Ag/mL glutamine, 5 ;Ag/mL hydrocortisone, 5 ,ug/mL insulin, and 0.1% gentamicin (10). The gland fragments from three rats were transferred into a 24-well microtiter plate (1.5 mL of the above medium per well), to which was added a test chemical or solvent vehicle (1%, v/v). DMBA and AAF were dissolved in DMSO, BT in ethanol, and others in water. Final concentrations in the incubation media were: DMBA, 5 Ag/mL; AAF, 40 ,ug/mL; benzene, 7500 Ag/mL (solubility in water is 1800 Ag/mL); others 750 and/or 1500 Ag/mL. The plates were incubated for 6, 24, 48, or 96 hr at 370C in a humidified atmosphere at 5% CO2 in air. For the histopathological examination, freshly excised Zymbal glands, as well as the treated tissues, from each of the time points were fixed in neutral formalin, embedded in paraffin, cut at 6 iAm thick, and stained with hematoxylin and eosin. For the adduct study, tissues were collected at 12 hr (DMBA) or 48 hr (others) after treatment. In vivo-modified AAF-DNA was prepared from the Zymbal glands of rats at 24 hr after daily oral gavage doses of 40 mg/kg AAF in ethanol/DMSO/olive oil (1:2:14, v/v, 3 ml/kg) for 5 days; control animals received solvent vehicle alone.
DNA was isolated from the Zymbal glands by treatment with ribonucleases (A, T1) and proteinase K, fol-lowed by a solvent extraction as described previously (11), except that the volumes of reagents were scaled down by 3-fold. An average yield of DNA from pooled Zymbal glands in several extractions corresponded to 17 (±6) ,ug DNA per rat. The presence of adducts in DNA was analyzed using a nuclease Pi-enhanced postlabeling assay (7,8). Briefly, DNA (10-15 dig) was enzymatically digested to 3'-deoxyribomonucleotides (12), which were then treated with nuclease Pi, which dephosphorylates nonnal nucleotides but not aromatic adducted nucleotides (8). The latter were 32P-labeled using y-32P-ATP and polynucleotide kinase and resolved by PEI-cellulose TLC according to the following two methods.
In method A, which utilizes the techniques previously described (13)(14)(15), the adducted 32P-nucleotides of AAF, benzene, and its derivatives were purified on a PEIcellulose thin-layer plate ( Fig. 1) in direction Dl in 2.3 M sodium phosphate, pH 5.77 (15), in situ magnet transferred to a second 20 x 12.5-cm size PE I-cellulose sheet (15), and resolved by a three-directional development (15) of the sheet in the following solvents: D3, 2. In method B, which was specially designed to permit the recovery of additional adducts more polar than those isolated in method A, the labeled solution was applied to D4,D5 -_-2.5 FIGURE 1. Scheme of PEI-cellulose TLC for purification and separation of adducted 32P-nucleotides (method A). This scheme is essentially the same as the one previously described by Lu et al. (15) with a few minor modifications. 32P-Labeled DNA digests were applied to the origins (a to g) of a PEI-cellulose sheet that had been attached to a Whatman paper No. 1 wick by stapling (A). The sheet was then developed in direction Dl to remove leftover y-32P-ATP and 32P-labeled contaminants. Each origin area (shaded area) containing purified aromatic adducts was cut out and attached to a second PEI-cellulose sheet (B) with the aid of button-type magnets for contact-transfer of adducts (15). The sheet was then developed along D3, D4, and D5 with solvents given in the text to transfer and resolve adducts. Measurements are in centimeters.  2). It was then developed overnight (16-17 hr) in direction Dl in 1.7 M sodium phosphate, pH 6.0, onto a Whatman No. 1 paper wick stapled to the top of the sheet. The wet sheet was cut below the paper wick, which was discarded and soaked twice in 1 L of deionized water for 5 min each. After drying, it was developed in D3, the same direction as the preceding development, with water to 1 cm from the bottom edge, then with 1.9 M lithium formate, 3.8 M urea, pH 3.5, to the top of the sheet. The sheet was soaked in water as before, dried, and developed in D4, perpendicular to the preceding development, in water to 1 cm, followed by 0.36 M sodium phosphate, 0.23 M Tris-HCl, pH 8.0, to the top, and dried. 32P-adducts were detected by screen-enhanced autoradiography (6,7). Adduct radioactivity was determined by excising the spots from the chromatograms and Cerenkov counting (6,7). An estimation of adduct levels, expressed as relative adduct labeling (RAL), was made from the values of adduct count rates, adjusted for background radioactivity, and the specific activity of y-32P-ATP determined by measuring the incorporation of 32p into a known amount of dAp (7,8). any test substance revealed that the glands at 6 hr (not shown) were histologically indistinguishable from those at 0 hr, i.e., the freshly removed tissue (Fig. 3A), while the ones at 24 (not shown), 48 (Fig. 3B), and 96 hr (Fig.   3C) exhibited a progressive deterioration often with necrosis and/or squamous metaplasia.

Analysis of DMBA and AAF Adducts
Using the nuclease Pi-version of the 32P-postlabeling assay (7,8), we detected DNA adducts in the Zymbal glands treated in culture with DMBA (5 ,ug/mL) and AAF (40 1g/gL), both being used in the present study as positive controls (Fig. 4). DMBA produced two major (spots 1 and 8) and five minor adducts (Fig. 4B) that were not detected in control DNA (Fig. 4A). The adduct pattern, except for an additional product 8, was qualitatively similar to that seen previously (7,8,16,17) in the epidermal DNA of mice that had been treated by topical application with DMBA. Cerenkov counting revealed that adducts 1 and 8 together comprised 70% of total DNA modifications, which corresponded to 25 adducts in 109 DNA nucleotides. Schmeiser et al. (18) have recently characterized adduct 1 as the major syn dihydrodiol epoxide deoxyguanosine (dG) product, and adducts 3 and 4 as the major anti dihydrodiol epoxide derivatives of deoxyadenosine (dA) and dG, respectively. AAF produced four adduct spots (Fig. 4D) that were chromatographically identical to those obtained in vivo in Zymbal gland DNA of rats dosed orally with AAF (Fig. 4E). Because some aromatic amine adducts are not recovered completely by the nuclease Pi-enhanced 32Pprocedure (7,8), AAF adducts were also evaluated by an alternate procedure involving butanol extraction of DNA digests prior to 32P-labeling (19). More radioactivity was incorporated into spot 1 after butanol extraction (Fig. 4G, 4H) when compared with that after exposure to nuclease P1 (Fig. 4D, 4E), suggesting the loss of this adduct by enzymic 3 '-dephosphorylation. The recovery of adduct 1 after nuclease Piwas 4% of that after butanol extraction; adduct 2 was recovered similarly with and without the enzyme (data not shown). Adduct 1, which has been characterized previously as dpGp-C8-(N2-AF) (13), composed 89% of total modifications, corresponding to 31 adducts per 109 DNA nucleotides. Modification of the in vivo sample was about 540 lesions per 109 nucleotides, with adduct 1 representing 97% of the total. Adduct 2 has previously been identified as dpGp-N2-(C3-AAF) (13).

Analysis of Adducts from Benzene and Its Derivatives
When DNA samples isolated from the Zymbal glands that had been treated in culture with various derivatives of benzene (1500 mg/mL), as well as benzene itself (r-1800 pg/mL), were analyzed by the nuclease Pi-amplified 32ppostlabeling assay, the autoradiograms shown in Figure  5 were obtained. DNA adducts were readily detectable Continued on next page C following exposure to different derivatives of benzene, with unique adduct profiles for each compound. Catechol produced one main adduct (spot a) and several minor ones (spots b to m), some of the latter migrated in areas where the DNA derivatives of BT migrate. Phenol produced one major (spot 1) and seven minor adducts; HQ induced an adduct pattern of 3 major spots (spots 5 to 7) and many minor adducts. With BQ-DNA, one major (spot 6) and a few minor alterations were detected. The major adduct was found to be a guanine derivative, as a chromatographically identical product was formed when dGp was reacted in vitro with BQ and 32 P-labeled (data not shown). With benzene-DNA, no adducts were detected when the chromatogram was exposed to X-ray film for the same time period as that employed for the detection of metabolite-DNA adducts; however, upon five times longer film exposure, two faint adduct spots in the area where the major products of BQ and phenol migrate were seen when compared with the control map left in contact with the film for the same period. Adducts exhibiting similar mobilities on the 32p maps were tentatively identified by overlapping the autoradiograms. Unequivocal evidence for the identity of adducts of interest would, however, require chromatography of a mixture of 32plabeled DNA digests in additional solvents and using different TLC systems (20). As shown in Figure 6, when 32plabeled digests were chromatographed according to  . Autoradiograms of TLC maps of 32P-labeled DNA adducts derived from benzene and its derivatives. DNA samples isolated from Zymbal glands exposed in culture for 48 hr to the solvent vehicle (control) and the indicated compounds were digested and 32P-labeled. Labeled adducts in DNA digests were purified, contact-transferred, and resolved by two-dimensional TLC according to method A outlined in Fig. 1. D3 development was in the vertical direction, and D4 and D5 developments were in the horizontal direction. Autoradiographic conditions were: control (top row), catechol and BT, -800C for 8 hr; phenol and HQ, -800C for 2 hr; BQ, 230C for 2 hr; control (bottom row) and benzene, -800C for 36 hr. Note that the sensitivity of film detection is increased 3to 4-fold at -800C relative to 230C. Chromatographically corresponding adducts on maps of catechol and BT, as well as those of phenol, HQ, BQ, and benzene, were identified with the same letters or numbers. method B, which permits recovery of derivatives more polar than those detected above (Fig. 5), one additional adduct (no. 23) with phenol-DNA and two (nos. 23

Discussion
In this report, we show for the first time that rat Zymbal glands can be maintained in culture for a period long enough to study early effects of genotoxicants, e.g., DNA adduct formation. Employing the nuclease Pi-enhanced 32P-postlabeling assay, DNA adducts have been detected in Zymbal glands cultured in the presence of DMBA, AAF, benzene, and its derivatives. Detection of DNA adducts with the indirect carcinogens, DMBA and AAF, indicates that Zymbal glands in culture are capable of metabolically activating these carcinogens to electrophilic intermediates that can bind to DNA. The activation pathways in the cultured Zymbal glands appear to be similar to those occurring in vivo, in as much as the adduct profiles (Fig. 4) were qualitatively the same as those seen in vivo in DMBA-treated mouse epidermal DNA (8,(16)(17)(18) and in AAF-treated rat Zymbal glands in the present studies. DMBA, however, induced an additional product (no. 8) in cultured Zymbal gland, which has not been detected in vivo in mouse skin (8,16,17), suggesting the formation of some tissue-specific reactive intermediates. DMBA and AAF metabo lsm is mediated by a mixed-function oxidase system (22,23). Adduct detection in Zymbal glands is consistent with the observation by Pohl and Fouts (24), who have shown the presence of cytochrome P-450-dependent aryl hydrocarbon hydroxylase activity in rat Zymbal gland homogenates.
The comparison of adduct patterns among different benzene metabolites studied (Fig. 5) enables the prediction of possible electrophiles formed from the individual compounds. Metabolism studies with liver and bone marrow (4,25,26) have shown that phenol can be oxidized to catechol, HQ, and biphenol by cytochrome P-450-dependent monooxygenase, peroxidase, or active oxygen released during the oxidative burst. Our findings that the major derivatives of phenol (Fig. 5, adducts 1 and 6) were different from those of catechol and overlapped with a fraction of HQ adducts suggest, but do not prove, that, in cultured Zymbal glands, conversion of phenol to catechol is negligible, while its conversion to HQ represents a minor pathway. This further suggests that phenol may be predominantly metabolized to biphenols, which, following oxidation to their semi-quinone and quinone entities, can bind to DNA.
The putative reactive metabolites derived from HQ are likely to be p-benzosemiquinone and/or BQ, which can be formed either by autoxidation of HQ at physiological pH in the presence of oxygen or by enzyme-mediated oxidation of HQ (4,25,26). In addition, HQ can be converted to BT, which can autoxidize to electrophiles such as ahydroxy-p-benzosemiquinone and a-hydroxy-p-benzoquinone (4,25,26). Since only a minor fraction of HQ adducts (Fig. 5, adducts 6 and 8, composing < 20%) corresponded to BQand BT-derived modifications (Fig. 5,  adducts 6, 8, and b), it appears that in Zymbal gland culture, BQ and BT are minor reactive intermediates formed from HQ, while the p-benzosemiquinone may predominate.
The DNA-reactive species derived from catechol may be either a mixture of o-benzosemiquinone and obenzoquinone formed by enzyme mediated-oxidation or a mixture of semiquinone and quinone species of BT (4,25,26). About 20% of total modifications produced by catechol exhibited chromatographic properties similar to those derived from BT (Table 1), indicating that the conversion of catechol to BT is a minor pathway. Therefore, the alternate pathway leading to the formation of the semiquinone and quinone is probably the major one. The unique 32P-fingerprints for each metabolite tested (Figs. 5 and 6) are consistent with the data of Rushmore et al. (27), who have observed different Sephadex LH-20 column proflles of dG adducts in 3H-GMP-prelabeled DNA of rabbit bone marrow mitoplasts exposed in vitro to HQ, BQ, phenol, catechol, and BT.
Although phenol or other metabolites of benzene induced relatively high levels of adducts, benzene itself produced adducts at levels barely detectable by the 32Ppostlabeling assay. This could result from a) less efficient conversion of benzene to phenol in the Zymbal gland cultures; b) the high volatility of benzene, reducing the effective dose reaching Zymbal glands in the culture medium; and/or c) the metabolism of benzene predominantly to a ring-opened derivative (28), which may form polar adducts not detectable by postlabeling under the fThe recovery of this adduct varied markedly. The value represents a maximum estimate. experimental conditions described. Currently, none of these possibilities can be dismissed. Reaction of BQ or HQ with dG has been shown by Jowa et al. (29) to generate (3 'OH)benzetheno(N1,N2)dG as the major derivative on the basis of HPLC and NMR analysis. In comparison with this data, adduct 6, which was detected as the major product of the interaction of BQ with DNA or dGp (Fig. 5), may be (3'OH)-benzetheno-(N1,N2)deoxyguanosine 3',5'-1[32P]bisphosphate. The detection of multiple 32P-postlabeled adducts with HQ-DNA is in accordance with the multiple HPLC peaks obtained from the hydrolysate of calf thymus DNA modified in vitro by 14C-HQ (29).
Analysis of AAF adducts here (Fig. 4) and those of 4-aminobiphenyl previously (7,8) indicate that the C8-guanine derivatives are susceptible to 3'-dephosphorylation by nuclease Pi, while the N2-guanine derivatives are not. Adducts of benzo[a]pyrene, safrole, and mitomycin C, which predominantly bind at the N2 position of guanine, have also been shown to be resistant to the enzyme activity (7,8). Taken together, the data suggest that an adducted 3'-nucleotide with a carcinogen bound to a base at the exocyclic position, but not at the endocyclic position, is resistant to nuclease P -mediated dephosphorylation. When the attachment is simultaneously at the exocyclic and endocyclic positions, for example, a cyclic adduct, as in the case of the major adduct of BQ-DNA above, the adduct may not be susceptible to the enzyme activity (21). Such differential enzyme activity may be useful in adduct characterization.
Our results demonstrate the utility of postlabeling techniques to determine the capacity of tissues or cells to metabolize a mutagen/carcinogen of interest by measuring DNA adduct fonnation. Because of high sensitivity and the need for only a few micrograms of DNA, the 32p postlabeling assay affords adduct measurement in a small number of cells, as evident from the present studies on Zymbal glands weighing only about 15 mg per rat. The 32P-postlabeling protocol, with further refinement to reduce background noise, may potentially be applicable to monitor the formation of DNA adducts in blood cells of humans exposed to benzene and thereby provide useful information for risk assessment and management.
By combining the highly sensitive postlabeling assay with the newly devised Zymbal culture system, we have begun to elucidate the metabolic pathways and macromolecular interactions in this target tissue. Insights gained from this work will subsequently be applied to the design and interpretation of analogous in vivo studies, with the ultimate goal of better understanding the mechanisms involved in the carcinogenic process.