Induction of cytochrome P-450c in hematopoietic cells of fetal liver

The transplacental inductive effect of beta-naphthoflavone (beta NF) on cytochrome P-450 isozymes was studied in separate hematopoietic and hepatocyte cells from fetal rat liver. Two fractions of dispersed fetal liver cells were isolated by Ficoll-Paque centrifugation and shown by histologic examination to be enriched in erythroblasts and hepatocytes, respectively. beta NF treatment increased ethoxyresorufin-O-deethylase activity 250-fold in both erythroblast and hepatocyte cell fractions. Polyacrylamide gel electrophoresis and immunostaining techniques showed the induction of cytochrome P-450c, but not P-450d, in erythroblast and hepatocyte fractions.     

Changes in the temporal and spatial expression of H beta 58 during formation and maturation of the chorioallantoic placenta in the Rat

Cloning and sequencing of a cDNA amplified by RNA fingerprinting at the implantation site of pregnant rats revealed 80% similarity with H beta 58, previously shown to be essential for formation of the chorioallantoic placenta in the mouse. H beta 58 mRNA was detected in the endometrium of hormonally sensitized rats stimulated to undergo decidualization and in the contralateral uterine horns lacking a decidual stimulus, indicating that uterine expression of H beta 58 mRNA did not require decidualization or the presence of a blastocyst. Immunodetection in the early postimplantation uterus (Days 6-8 of pregnancy) showed H beta 58 localized in the luminal and glandular epithelia and some stromal cells. Decidual cells at Day 6 of pregnancy expressed H beta 58, and by Day 9 of pregnancy, the protein localized throughout the maternal decidua. The temporal and spatial distribution of H beta 58 in the developing chorioallantoic placenta was assessed at Days 10, 12, and 14 of pregnancy. Immunoreactive H beta 58 localized to erythroid cells within the developing fetal vasculature of the chorioallantoic primordia at Day 10 of pregnancy. By Day 12, the fetal vasculature extended into the placental labyrinth, and the erythroid stem cells continued to strongly express H beta 58. At Day 14 of pregnancy, immunoreactivity became evident in the trophoblast giant cells and syncytiotrophoblast of the fetal placenta. As the chorioallantoic placenta matured (Day 18), H beta 58 mRNA was 3.6-fold higher in the labyrinth compared with the junctional region. Stable cell lines (HRP/LRP) isolated from the rat labyrinthine placenta expressed H beta 58 mRNA and protein. The expression pattern of H beta maternal and fetal placental tissues and its early expression in fetal erythroid stem cells during formation and maturation of the chorioallantoic placenta suggest that H beta 58 plays key roles in the regulatory networks that control hematopoietic development and placentation.     

DNA--benzo[a]pyrene adducts formed in a Salmonella typhimurium mutagenesis assay system.

The DNA adducts formed in Salmonella typhimurium when bacteria are incubated with radioactive benzo[a]pyrene and liver microsomal enzymes from several sources has been investigated. When enzyme preparations from Aroclor I254 or 3-methylcholanthrene induced C57BL/6N (B6) mice were used to mediate activation, the predominant product was an adduct between the 10 position of 7 beta, 8 alpha-dihydroxy-9 alpha, 10 alpha epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene and the N-2 position of deoxyguanosine. Similar results were obtained with human liver and with Aroclor-induced rat-liver enzyme preparations. This adduct is also the major DNA product previously found when human tissues or certain rodent cells were incubated with benzo[a]pyrene. On the other hand, when activation of benzo[a]pyrene was mediated by a phenobarbital-induced B6 mouse-liver enzyme preparation, the extent of binding was quite low and the profile of DNA adducts in S. typhimurium DNA was quite different. Thus, under appropriate conditions, the activation and DNA binding of benzo[a]pyrene inthe microsome mediated S. typhimurium mutagenesis assay generally resembles that seen in intact mammalian cells. Caution must be exercised, however, in the choice of microsome-activation systems.     

Mutagenicity of benzo[a]pyrene 7,8-dihydrodiol and 7,12-dimethylbenz[a]anthracene 3,4-dihydrodiol in S. typhimurium mediated by microsomes from rat liver and mouse skin

The mutagenic activities of trans-7,8-dihydro-7,8-dihydroxybenzo[a]-pyrene (BP 7,8-diol) and of trans-3,4-dihydroxy-7,12-dimethylbenz[a]-anthracene (DMBA 3,4-diol) towards S. typhimurium TA100 were measured in assays that were carried out on a micro-scale in liquid medium in the presence of microsomal fractions prepared from mouse skin or rat liver. In the presence of an NADPH-generating system, microsomal enzymes converted both diols into mutagens that were probably the respective 'bay-region' diol-epoxides. The rate of the enzyme-catalysed conversion of the BP 7,8-diol into mutagens by microsomal preparations from mouse epidermis was similar to that occurring with microsomes from rat liver. Pretreatment of mice by the topical application of benz[a]anthracene (BA) or 7,12-dimethylbenz[a]-anthracene (DMBA) increased the mutagenic activity of BP 7,8-diol mediated by mouse skin microsomal preparations by 2-fold and this was paralleled by a 4-fold increase in epidermal aryl hydrocarbon (benzo[a]pyrene) hydroxylase (AHH) activity. The results are discussed in relation to the high susceptibility of mouse skin to polycyclic aromatic hydrocarbon (PAH) carcinogenesis.     

Metabolism of benzo(a)pyrene by isolated hepatocytes and factors affecting covalent binding of benzo(a)pyrene metabolites to DNA in hepatocyte and microsomal systems

Covalent binding of benzo(a)pyrene (BP) metabolites to DNA was investigated in hepatocytes and liver microsomes (MC-microsomes) isolated from 3-methylcholanthrene-treated rats. The major DNA adducts formed during BP metabolism in both hepatocytes and incubations of calf thymus DNA with MC-microsomes were adducts of anti and syn isomers of trans-7,8,-dihydroxy-9,10-epoxy-7,8,9,10-tetrahydrobenzo(a)pyrene (diol-epoxides) and of epoxide derivatives of BP-9-phenol (phenol-oxides). Diol-epoxide adducts predominated over phenol-oxide adducts in hepatocytes, while the reverse was found in microsomal incubations. In hepatocytes, both diol-epoxide and phenol-oxide adducts increased with increasing BP concentration; the ratio of diol-epoxide adduct to phenol-oxide adduct decreased from 6:1 to 3:1 between 30 and 100 μm BP. In microsomal incubations, decreases in DNA concentration or addition of the hepatocyte L15 medium produced larger decreases in phenol-oxide adducts than in diol-epoxide adducts. The effects of the inhibitors salicylamide, diethylmaleate, and 3,3,3,-trichloropropene oxide on formation of BP-DNA adducts are interpreted in terms of changes in precursor formation and metabolism and reductions in hepatocyte glutathione levels. Addition of 1.5 mg/ml exogenous DNA to hepatocyte incubations produced no change in covalent binding to cellular DNA, even though extracellular BP-DNA adducts accounted for 97% of the total adducts formed. Both the relative amounts of diol-epoxide and phenol-oxide adducts and the total adducts per milligram of DNA were indistinguishable with respect to extracellular and intracellular DNA. Modification of extracellular DNA by diol-epoxides was at least as efficient as modification of calf thymus DNA in incubations with MC-microsomes. It is concluded that BP diol-epoxides and phenol-oxides can leave the cell or enter the nucleus with equal facility but are more effective in binding to DNA in the cell in which they are generated.     

Ubiquitous binding of benzo[a]pyrene metabolites to DNA and protein in tissues of the mouse and rabbit

The in vivo formation of benzo[alpha]pyrene (BP) metabolite-DNA adducts in several tissues of mice and rabbits was examined. Included were tissues with widely divergent xenobiotic metabolizing capabilities such as liver and brain. The major adduct identified in each tissue was the (+)-7 beta,8 alpha-dihydroxy-9 alpha,10 alpha-epoxy-7,8,9,10-tetrahydro-BP (BPDEI)-deoxyguanosine adduct. A 7 beta,8 alpha-dihydroxy-9 beta,10 beta-epoxy-7,8,9,10-tetrahydro-BP (BPDEII)-deoxyguanosine adduct, a (-)-BPDEI-deoxyguanosine adduct and an unidentified adduct were also observed. These adducts were present in all of the tissues of the mice and in the lungs of the rabbits; only BPDEI and BPDEII were seen in the rest of the rabbit tissues. In all of the tissues studied, the DNA adduct levels were unexpectedly similar. For example, the BPDEI-DNA adduct levels in muscle and brain of mice were approx. 50% of those in lung and liver at each oral BP dose examined. After an i.v. dose of BP in rabbits, the BPDEI adduct levels in lung were three times those in brain or liver and twice those in muscle. The binding of BP metabolites to protein was also determined in these tissues. The tissue-to-tissue variation in protein binding levels of BP metabolites was greater than that for BPDEI-DNA adducts. There are several possible explanations for the in vivo binding of BP metabolites to DNA and protein of various tissues. First, oxidative metabolism of BP in each of the examined tissues might account for the observed binding. Second, reactive metabolites could be formed in tissues such as liver and lung and be transported to cells in tissues such as muscle and brain where they bind to DNA and protein. In any case, the tissue-to-tissue variations in protein and DNA binding of BP-derived radioactivity do not correlate with differences in cytochrome P-450 activity.     

Quantitation and mapping of aflatoxin B1-induced DNA damage in genomic DNA using aflatoxin B1-8,9-epoxide and microsomal activation systems

Aflatoxin B1 (AFB1) is a mutagenic and carcinogenic mycotoxin which may play a role in the etiology of human liver cancer. In vitro studies have shown that AFB1 adducts form primarily at the N7 position of guanine. Using quantitative PCR (QPCR) and ligation-mediated PCR (LMPCR), we have mapped total AFB1 adducts in genomic DNA treated with AFB1-8,9-epoxide and in hepatocytes exposed to AFB1 activated by rat liver microsomes or human liver and enterocyte microsomal preparations. The p53 gene-specific adduct frequencies in DNA, modified in cells with 40-400 microM AFB1, were 0.07-0.74 adducts per kilobase (kb). In vitro modification with 0. 1-4 ng AFB1-8,9-epoxide per microgram DNA produced 0.03-0.58 lesions per kb. The adduct patterns obtained with the epoxide and the different microsomal systems were virtually identical indicating that adducts form with a similar sequence-specificity in vitro and in vivo. The lesions were detected exclusively at guanines with a preference towards GpG and methylated CpG sequences. The methods utilizing QPCR and LMPCR thus provide means to assess gene-specific and sequence-specific AFB1 damage. The results also prove that microsomally-mediated damage is a suitable method for avoiding manipulations with very unstable DNA-reactive metabolites and that this damage can be detected by QPCR and LMPCR.     

Localization and secretion of inhibins in the equine fetal ovaries

To clarify the source of inhibins in equine female fetuses, concentrations of immunoreactive (ir-) inhibin, inhibin pro-alphaC, and inhibin A in both fetal and maternal circulation and in fetal ovaries were measured. In addition, the localization of inhibin alpha and inhibin/activin beta(A), and beta(B) subunits and the expression of inhibin alpha(A) and inhibin/activin beta(A) subunit mRNA in fetal ovaries were investigated using immunohistochemistry and in situ hybridization. Concentrations of circulating ir-inhibin, inhibin pro-alphaC, and inhibin A were remarkably more elevated in the fetal than in the maternal circulation between Days 100 and 250 of gestation. Fetal ovaries contained large amounts of ir-inhibin, inhibin pro-alphaC, and inhibin A. In contrast, these inhibin forms were undetectable in both the maternal ovaries and placenta. The inhibin alpha and inhibin/activin beta(A) and beta(B) subunit proteins were localized to enlarged interstitial cells of the equine fetal ovary. Expression of inhibin alpha and inhibin/activin beta(A) subunit mRNAs were also observed in the interstitial cells. We conclude that the main source of large amounts of inhibins in fetal circulation is interstitial cells of fetal ovary and is not of maternal origin. Furthermore, these inhibins may play some important physiological roles in the development of gonads in the equine fetus.     

Effect of iron deficiency on placental cytokine expression and fetal growth in the pregnant rat.

Iron deficiency anemia is the most common nutritional disorder in the world. Anemia is especially serious during pregnancy, with deleterious consequences for both the mother and her developing fetus. We have developed a model to investigate the mechanisms whereby fetal growth and development are affected by maternal anemia. Weanling rats were fed a control or iron-deficient diet before and throughout pregnancy and were killed at Day 21. Dams on the deficient diet had lower hematocrits, serum iron concentrations, and liver iron levels. Similar results were recorded in the fetus, except that the degree of deficiency was markedly less, indicating compensation by the placenta. No effect was observed on maternal weight or the number and viability of fetuses. The fetuses from iron-deficient dams, however, were smaller than controls, with higher placental:fetal ratios and relatively smaller livers. Iron deficiency increased levels of tumor necrosis factor alpha (TNFalpha) only in the trophoblast giant cells of the placenta. In contrast, levels of type 1 TNFalpha receptor increased significantly in giant cells, labyrinth, cytotrophoblast, and fetal vessels. Leptin levels increased significantly in labyrinth and marginally (P = 0.054) in trophoblast giant cells. No change was observed in leptin receptor levels in any region of the placentas from iron-deficient dams. The data show that iron deficiency not only has direct effects on iron levels and metabolism but also on other regulators of growth and development, such as placental cytokines, and that these changes may, in part at least, explain the deleterious consequences of maternal iron deficiency during pregnancy.     

Relationship between mutagenicity and DNA adduct formation in mammalian cells for fjord- and bay-region diol-epoxides of polycyclic aromatic hydrocarbons

Chinese hamster V79 cells were treated with the anti- and syn-diastereomers of the bay- or fjord-region diol-epoxides of four polycyclic aromatic hydrocarbons, namely benzo[a]pyrene (BP), benzo[c]chrysene (BcC), benzo[g]chrysene (BgC) and benzo[c]phenanthrene (BcPh). The frequency of induction of 6-thioguanine-resistant mutations was determined, and the extent of formation of DNA adducts was measured by 32P-postlabelling. When expressed as mutation frequency per nanomoles compound per millilitre incubation medium, this group of chemicals expressed a 160-fold range in potency. In agreement with previous experimental studies, the anti-diol-epoxide of BcC was highly mutagenic, inducing in excess of 3 x 10(4) mutations/10(6) cells per nmol compound/ml. The mutagenic activities of the anti- and syn-diol-epoxides of BP were 10- and 100-fold lower, respectively. Both diol-epoxides of BgC, the syn-BcC and the anti-BcPh derivatives were also highly mutagenic, and only the syn-BcPh diol-epoxide was less mutagenic than the anti-diol-epoxide of BP. Determination of the levels of DNA adducts formed by the diol-epoxides indicated that the most mutagenic compounds were the most DNA reactive, although the fjord-region diol-epoxides gave rise to more complex patterns of adducts than those of the BP diol-epoxides. When the mutagenicity results were expressed as mutations per femtomoles total adducts formed, all compounds showed similar activities. Thus the potent mutagenicity of the fjord region diol-epoxides appears to be due to the high frequency with which they form DNA adducts in V79 cells, rather than to formation of adducts with greater mutagenic potential.     

Comparison of the formation of benzo[a]pyrene diolepoxide-DNA adducts in vitro by rat and human microsomes: evidence for the involvement of P-450IA1 and P-450IA2

The involvement of cytochrome P-450 isozymes in the activation of benzo[a]pyrene (BP) by human placental and liver microsomes was studied in vitro using monoclonal antibodies (Mab) toward the major 3-methylcholanthrene (MC)-inducible and phenobarbital-inductible rat liver P-450 isozymes (Mab 1-7-1 and Mab 2-66-3, respectively). Microsomes from human placenta and liver and rat liver were incubated with BP and DNA, and BP-diolepoxide-DNA (BPDE-DNA) adducts were measured by synchronous fluorescence spectrophotometry (SFS). The only BP metabolite giving the same fluorescence peak as chemically modified BPDE-DNA was BP-7,8-dihydrodiol. Five (smokers) out of 29 human placentas (smokers and nonsmokers), and five out of nine human livers were able to metabolically activate BP to BPDE-DNA adducts in this system. The Mab 1-7-1 totally inhibited the formation of BPDE-DNA adducts in placental microsomal incubations. Inhibition using rat or human liver microsomes was 50-60% and about 90%, respectively. The Mab 2-66-3 had no effect in any of the microsome types. Adduct formation was inhibited more strongly and at lower concentrations of Mab 1-7-1 compared with the inhibition of AHH activity. This study is a clear indication of the major role of P-450IA1 (P-450c) in human placenta and probably P-450IA2 (P-450d) in human liver in BP activation, while other isozymes also take part in the activation in rat liver. Furthermore, this clearly indicates that AHH activity and BP activation are not necessarily associated.     

Magnesium ions affect the quantitative but not the qualitative microsome mediated binding of benzo[alpha]-pyrene to DNA

Five distinct hydrocarbon-deoxyribonucleoside adducts are separated by high pressure liquid chromatography after reaction of benzo[alpha]pyrene with calf thymus DNA in the presence of liver microsomes from 3-methylcholanthrene treated rats. The two major adducts co-chromatography with deoxyribonucleoside adducts obtained after hydrolysis of calf thymus DNA previously reacted with liver microsomal metabolically activated 9-hydroxy-benzo[alpha]pyrene or trans-7,8-dihydro-7,8-dihydroxybenzo[alpha]pyrene. High magnesium ion concentrations in the microsomal incubations cause a significant decrease in the covalent binding of the hydrocarbon to DNA but do not affect the qualitative distribution of the individual benzo[alpha]pyrene-deoxyribonucleoside adducts.     

The fetal erythroblast is not the optimal target for non-invasive prenatal diagnosis: preliminary results.

Fetal cells, present in the blood of pregnant women, are potential targets for non-invasive prenatal diagnosis. The fetal erythroblast has been the favorite target cell type. We investigated four methods of enrichment for fetal erythroblasts, identifying only three fetal erythroblasts in 573 ml of maternal blood. This is much less than the expected two to six fetal cells per ml of maternal blood. Hamada and Krabchi used a cell type-independent marker, i.e., the Y chromosome in maternal blood from male pregnancies after Carnoy fixation, leaving the nuclei for hybridization with X-and Y-chromosome-specific probes. We found with a similar technique 28 fetal cells in 15 ml of maternal blood. The fetal origin of cells was confirmed by hybridizing the nuclei with X- and Y-chromosome-specific probes, using two consecutive hybridizations with the two probes in opposite colors (reverse FISH). Candidate fetal cells were inspected after each hybridization. Only cells that were found to change the color of both probe signals from first to second hybridization were diagnosed as fetal. To reduce the labor-intensive slide screening load, we used semiautomated scanning microscopy to search for candidate cells. We conclude that erythroblasts form only a small fraction of fetal cells present in maternal blood.     

Up-regulation of alpha-inhibin expression in the fetal ovary of estrogen-suppressed baboons is associated with impaired fetal ovarian folliculogenesis

We recently demonstrated that the number of primordial follicles was significantly reduced in the ovaries of near-term baboon fetuses deprived of estrogen in utero and restored to normal in animals administered estradiol. Although the baboon fetal ovary expressed estrogen receptors alpha and beta, the mechanism(s) of estrogen action remains to be determined. It is well established that inhibin and activins function as autocrine/paracrine factors that impact adult ovarian function. However, our understanding of the expression of these factors in the primate fetal ovary is incomplete. Therefore, we determined the expression of alpha-inhibin, activin beta(A), activin beta(B), and activin receptors in fetal ovaries obtained at mid and late gestation from untreated baboons and at late gestation from animals in which fetal estrogen levels were reduced by >95% by maternal administration of the aromatase inhibitor CGS 20267 or restored to 30% of normal by treatment with CGS 20267 and estradiol benzoate. Immunocytochemical expression of alpha-inhibin was minimal to nondetectable in fetal ovaries from untreated baboons. In contrast, in baboons depleted of estrogen, alpha-inhibin was abundantly expressed in pregranulosa cells of interfollicular nests and granulosa cells of primordial follicles. Thus, the number (mean +/- SEM) per 0.08 mm2 of fetal ovarian cells expressing alpha-inhibin, determined by image analysis, was similar at mid and late gestation and increased approximately 8-fold (P < 0.01) near term in baboons treated with CGS 20267 and was restored (P < 0.01) to normal in baboons treated with CGS 20267 plus estradiol. Activin beta(A) was detected in oocytes and pregranulosa cells at midgestation and in oocytes and granulosa cells of primordial follicles at late gestation. Activin beta(B) was also expressed in pregranulosa cells and granulosa cells at mid and late gestation, respectively, but was not detected in oocytes. Neither the pattern nor the apparent level of expression of activin beta(A) or beta(B) were altered in fetal ovaries of baboons treated with CGS 20267 or CGS 20267 and estrogen. Activin receptors IA, IB, IIA, and IIB were detected by Western blot analysis in fetal ovaries at mid and late gestation, and expression was not altered by treatment with CGS 20267 or CGS 20267 and estrogen. Activin receptors IB and IIA were localized to oocytes and pregranulosa cells at midgestation and to granulosa cells and oocytes of primordial follicles at late gestation. Thus, the decrease in the number of follicles in the primate fetal ovary of baboons deprived of estrogen in utero was associated with increased expression of alpha-inhibin. Therefore, we propose that estrogen regulates fetal ovarian follicular development by controlling alpha-inhibin expression and, thus, the intraovarian inhibin:activin ratio.     

Stereoselectivity of the epoxidation of 7,8-dihydrobenzo[a]pyrene by prostaglandin H synthase and cytochrome P-450 determined by the identification of polyguanylic acid adducts

The stereoselectivity of the oxidation of 7,8-dihydrobenzo[a]pyrene (H2BP) to 9,10-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene (H4BP-epoxide) by prostaglandin H (PGH) synthase and cytochrome P-450 has been studied using microsomal preparations from ram seminal vesicles and rat liver. Incubations were performed in the presence of polyguanylic acid and the adducts formed between H4BP-epoxide and guanosine were isolated following the recovery and hydrolysis of the poly(G). When (+/-)-H4BP-epoxide was reacted with poly(G), four diastereomeric adducts were formed by the cis and trans addition of the exocyclic amino group of guanine to the benzylic carbon of the epoxide enantiomers. Each diastereomer was identified by a combination of ultraviolet, nuclear magnetic resonance, circular dichroism, and mass spectroscopy. Under comparable conditions, ram seminal vesicle microsomes in the presence of arachidonic acid triggered the binding of H2BP to poly(G) to a greater extent than rat liver microsomes from untreated and phenobarbital- and methylcholanthrene pretreated animals in the presence of NADPH. Quantitation of the (-)-cis- and (+)-cis-guanosine adducts revealed the degree of stereoselectivity of epoxidation. The ratio of (-)/(+) adducts was 54:46 for PGH synthase and 89:11 (control), 62:38 (phenobarbital), and 69:31 (methylcholanthrene) for cytochrome P-450-catalyzed reactions. PGH synthase catalyzed the epoxidation of H2BP with little or no stereoselectivity in contrast to cytochrome P-450. The utility of the poly(G) binding technique for the elucidation of the stereoselective generation of chiral electrophiles is discussed along with the mechanistic implications of the results.     

Fetal cells in mother rats contribute to the remodeling of liver and kidney after injury

Fetal microchimerism indicates a mixture of cells of maternal and fetal origin seen in maternal tissues during and after pregnancy. Controversy exists about whether persistent fetal microchimerism is related with some autoimmune disorders occurring during and after pregnancy. In the current experiment, an animal model in which EGFP positive cells were taken as fetal-origin cells was designed to detect the fetal microchimerism in various maternal organs. Ethanol drinking and gentamicin injection were adopted to induce liver and kidney injury simultaneously. EGFP positive cells were engrafted not only in the maternal circulation and bone marrow, but also in the liver and kidney as hepatocytes and tubular cells, respectively. These results indicate that fetal cells are engrafted to maternal hematopoietic system without apparent injury and they also contribute to the repairing process of maternal liver and kidney.     

Formation and removal of benzo[a]pyrene metabolites--DNA adducts in cultured normal human fibroblasts using a microsome-activating system

The rate of removal of DNA adducts of several benzo[a]pyrene metabolites from nuclear DNA was compared by introducing a microsome-activating system in human fibroblast cells. Confluent human fibroblasts were exposed to benzo[a]pyrene in the presence of a microsomal activating system and DNA adducts were formed in the nuclear DNA. The adducts present in DNA were determined after 1 h of incubation and 48 h later. There was no difference in the rate of removal between 7S- and 7R -N2-[10-(7 beta, 8 alpha-trihydroxy-7,8,9,10- tetrahydrobenzo[a]pyrene)yl]deoxyguanosine, 7R -N2-[10(7beta, 8 alpha, 9 beta-trihydroxy-7,8,9,10-tetrahydrobenzo[a]pyrene)yl]deoxyguanosine and the covalent adduct of 9-hydroxybenzo[a]pyrene-4,5-epoxide to guanosine. This finding does not agree with the idea that metabolites forming 'persistent DNA adducts' are always responsible for the carcinogenicity of their parent compound.