For Dioxin-induced Birth Defects, Mouse or Human CYP1A2 in Maternal Liver Protects whereas Mouse CYP1A1 and CYP1B1 Are Inconsequential*

Dioxin (2,3,7,8-tetrachlorodibenzo-p-dioxin) induces cleft palate and hydronephrosis in mice, when exposed in utero; these effects are mediated by the aryl hydrocarbon receptor. The Cyp1a1, Cyp1a2, and Cyp1b1 genes are up-regulated by the aryl hydrocarbon receptor. To elucidate their roles in dioxin-induced teratogenesis, we compared Cyp1a1(-/-), Cyp1a2(-/-), and Cyp1b1(-/-) knock-out mice with Cyp1(+/+) wild-type mice. Dioxin was administered (25 μg/kg, gavage) on gestational day 10, and embryos were examined on gestational day 18. The incidence of cleft palate and hydronephrosis was not significantly different in fetuses from Cyp1a1(-/-), Cyp1b1(-/-), and Cyp1(+/+) wild-type mice. To fetuses carried by Cyp1a2(-/-) dams, however, this dose of dioxin was lethal; this effect was absolutely dependent on the maternal Cyp1a2 genotype and independent of the embryonic Cyp1a2 genotype. Dioxin levels were highest in adipose tissue, mammary gland, and circulating blood of Cyp1a2(-/-) mothers, compared with that in the Cyp1(+/+) mothers, who showed highest dioxin levels in liver. More dioxin reached the embryos from Cyp1a2(-/-) dams, compared with that from Cyp1(+/+) dams. Fetuses from Cyp1a2(-/-) dams exhibited a ∼6-fold increased sensitivity to cleft palate, hydronephrosis, and lethality. Using the humanized hCYP1A1_1A2 transgenic mouse (expressing the human CYP1A1 and CYP1A2 genes in the absence of mouse Cyp1a2 gene), the teratogenic effects of dioxin reverted to the wild-type phenotype. These data indicate that maternal mouse hepatic CYP1A2, by sequestering dioxin and thus altering the pharmacokinetics, protects the embryos from toxicity and birth defects; substitution of the human CYP1A2 trans-gene provides the same protection. In contrast, neither CYP1A1 nor CYP1B1 appears to play a role in dioxin-mediated teratogenesis.

TCDD is a potent teratogen. In mice, the most dramatic teratogenic effects, with TCDD exposure on or before gestational day 11 (GD11), are cleft palate, hydronephrosis, and open eye (14). Ahr(Ϫ/Ϫ) knock-out mice are protected against in utero or lactational TCDD exposure (15)(16)(17)(18). Interestingly, Ahr(ϩ/Ϫ) heterozygous pups carried by Ahr(Ϫ/Ϫ) dams are 5-fold more sensitive to maternal TCDD exposure, compared with those carried by Ahr(ϩ/ϩ) dams; this effect may be explained in part by the patent ductus venosus, which exists in female Ahr(Ϫ/Ϫ) mice (19). Thus, venous shunting by the liver eliminates first-pass elimination of many compounds that would otherwise be metabolized in liver. Indeed, pups carried by Ahr(Ϫ/Ϫ) dams are more sensitive to glucocorticoid exposure, and glucocorticoid metabolism is not affected by AHR-mediated pathways. Such first-pass elimination, however, likely does not explain enhanced TCDD toxicity, because TCDD is very poorly metabolized with a half-life of Ͼ30 -180 days in mice (20,21).
It is well established that binding of TCDD to the AHR is essential for its toxic actions (22)(23)(24). AHR, without its ligand, is present in the cytosol, and forms a complex with several chaperones: two heat shock protein ␣ 90-kDa, class A member 2 (official name HSP90AA2; also called Hsp90) molecules, a small protein (p23), and an AHR-interacting protein AIP (also called XAP2) (25). When TCDD binds the AHR, the complex translocates from the cytoplasm to the nucleus, and AHR switches partners from HSP90AA2 to aryl hydrocarbon receptor nuclear translocator (ARNT) (26). Then, this heterodimer binds a specific DNA motif, named the AHR response element (also called XRE) in the promoter region of many target genes (27). The AHR also interacts with other transcription factors, such as retinoblastoma protein-1 and NF-␤; for these interactions, ARNT and/or AHR response element binding may not be necessary (28 -30).
It has been shown that AHR directly or indirectly mediates the transcriptional activation of many genes, including those of the [Ahr] gene battery (11). This group includes three Cyp1 family members, the Cyp1a1, Cyp1a2, and Cyp1b1 genes; CYP1A1 and CYP1B1 are central to the metabolism of polycyclic aromatic hydrocarbons (PAHs), whereas CYP1A2 substrates are mostly N-heterocyclics and arylamines. Emerging evidence from knock-out mice has demonstrated that intestinal and hepatic CYP1A1 and CYP1A2 are more important in detoxication (31)(32)(33), whereas spleen and bone marrow CYP1B1 is responsible for metabolically potentiating the toxicity of several PAHs (33)(34)(35).
Cyp1 expression is chronically activated by planar polyhalogenated aromatic dioxins such as TCDD; chronic activation occurs via the AHR and is the consequence of the poor metabolism and very slow elimination of this class of compounds.
It has been hypothesized that TCDD-mediated toxicity is the consequence of chronic activation of the direct transcriptional targets of the AHR/ARNT heterodimer (23,36,37). In this regard, both Cyp1a1(Ϫ/Ϫ) and Cyp1a2(Ϫ/Ϫ) knock-out mice show some protection against high dose TCDD toxicity (7,38,39), and Cyp1a2(Ϫ/Ϫ) mice show protection against planar polychlorinated biphenyl-induced hepatocarcinogenesis (40,41). In all cases, genetic absence of either the CYP1A2 or CYP1A1 enzyme results in the dramatic lowering of risk or the complete ablation of uroporphyria; uroporphyria is likely to be important for protective effects in the liver. Many of the toxic manifestations of TCDD persist in Cyp1a1(Ϫ/Ϫ) or Cyp1a2(Ϫ/Ϫ) mice, suggesting that TCDD-mediated toxicity relies on cell type-specific alterations in the transcriptome (7): in some cell types, such as the hepatocyte, Cyp1 genes play a role in TCDD-mediated toxicity, and in other cell types they do not. In the present study we have examined the roles of CYP1A1, CYP1A2, and CYP1B1 during TCDD-induced teratogenesis.
Animals-Mice were housed in a pathogen-free facility with food and water ad libitum. Wild-type Cyp1(ϩ/ϩ) (C57BL/6J) breeders were originally obtained from The Jackson Laboratory (Bar Harbor, ME), and the breeders were maintained in the Nebert animal colony. The Cyp1a1(Ϫ/Ϫ) (42), Cyp1a2(Ϫ/Ϫ) (43), and Cyp1b1(Ϫ/Ϫ) (44) genotypes have all been backcrossed onto the C57BL6/J background for ten generations, meaning that the genomes of these knock-out lines are Ͼ99.9% C57BL6/J. The original Cyp1b1(Ϫ/Ϫ) knock-out mouse was generously provided by Frank J. Gonzalez (Bethesda, MD). All animal experiments were approved and conducted in accordance with the National Institutes of Health standards and the University of Cincinnati Medical Center Institutional Animal Care and Use Committee.
Biohazard Precaution-TCDD is a very toxic compound and a probable human carcinogen; all personnel were therefore instructed in safe handling procedures. Lab coats, gloves, and masks were worn at all times, and contaminated materials were collected separately for disposal by the Hazardous Waste Unit or by independent contractors. TCDD-pretreated mice were housed separately, and their carcasses were treated as contaminated biological materials.
Treatment of Animals-Mice, 10 -12 weeks old, were used in these studies. Female mice of the genotype indicated were mated to male mice of the indicated genotype. Gestational day zero (GD0) was the day on which a vaginal plug was detected. Unless otherwise indicated, pregnant females received, by gavage, either vehicle alone (corn oil, 100 l) or the indicated dose of TCDD in 100 l of corn oil, prepared as previously described (14,45).
To avoid confusion caused by interference with genomic DNA, all Cyp1 primers span an intron of their respective genes. The ␤-actin primers spanned intron 5 of the ACTB gene. Detection of RT-PCR products required reverse transcription, and product analysis in all cases was consistent with cDNA detection.
Assessment of TCDD-induced Teratogenesis-Each group started with a minimum of five pregnant females. TCDD or vehicle alone was given on GD10, and on GD18, the maternal-fetal unit was weighed, and the dam was anesthetized by CO 2 inhalation and euthanized by cervical dislocation. The maternal liver and intact uterus were removed and weighed. The fetuses were removed from the uterus, weighed, and placed on ice. The parameters evaluated for each litter included the number of implantation sites and the number of live or dead fetuses. Teratogenesis was evaluated principally by the incidence of cleft palate (number of fetuses with cleft palate/total number of live fetuses) and the incidence and severity of hydronephrosis; presence and severity of hydronephrosis in each kidney was examined by microscopy, as previously described (48), using severity scores ranging from 0 to 3ϩ (0 ϭ normal kidney; 1ϩ ϭ slight decrease in length of papilla; 2ϩ ϭ marked decrease in length of papilla with some loss of renal parenchyma; and 3ϩ ϭ complete absence of papilla, shell of kidney remaining with only a small amount of renal parenchyma). Fetuses were decapitated, and their mandible removed, to examine the palate. Dissected fetal head, liver, and kidney were placed in 4% paraformaldehyde, post-fixed with 1% osmium tetroxide, dehydrated in ethanol and propylene oxide, embedded in Spurr's resin, and then sectioned for histological analysis.
Tissue Distribution of TCDD-The amount of [ 3 H]TCDD (10 Ci/mmol) used to spike the usual dose (TCDD, 25 g/kg) was 20 Ci. For the Cyp1(ϩ/ϩ) and Cyp1a2(Ϫ/Ϫ) genotypes, three pregnant female mice of each were given an intraperitoneal injection of the spiked TCDD on GD15. Twenty-four hours later, on GD16, each individual fetus and each placenta was harvested from the uterus. The fetuses, placentas, and maternal liver, thymus, kidney, spleen, heart, brain, adipose tissue, mammary gland, and lung were weighed and added to hyamine hydroxide (1 ml/100 mg) in glass scintillation vials. Maternal and fetal blood was similarly digested with hyamine hydroxide (1 ml/100 l). The ratio of hyamine hydroxide to tissue was based on the manufacturer's recommendations (ICN Radiochemicals, Irvine, CA). After 5 days of digestion, 5 ml of Cytoscint (ICN Radiochemicals) was added to the samples, and the radioactivity was determined.
Statistical Analysis-Data were analyzed using Sigmastat software. When comparing control and TCDD-treated mice, the values were compared using analysis of variance analysis. A p value of Ͻ0.05 was considered significant.

RESULTS
Three Cyp1 Knock-out Mouse Lines-In mammals, the CYP1 gene family contains three members: CYP1A1, CYP1A2, and CYP1B1; targeted null mutant mice have been described for each of these genes (42)(43)(44). Each of these knock-out lines shows normal fecundity and no overt phenotype. In our mouse colony, all three Cyp1 knock-out lines have been bred into the C57BL6/J (B6) inbred mouse strain (Ͼ99.9%) background; consequently, B6 inbred mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and used as the wild-type Cyp1(ϩ/ϩ) control. This inbred strain and the three knock-out lines thus express the high affinity Ahr(b1) allele. Our initial hypothesis was that loss of one or more of the CYP1 enzymes would protect mice against TCDD-induced teratogenesis. Therefore, to the pregnant dam at GD10 we administered a dose of TCDD (25 g/kg) by gavage; this dose is known to induce a high incidence of birth defects in wild-type mice (14,18,48,51).
The mRNA data are further supported by the hepatic enzyme assays. For benzo[a]pyrene hydroxylase (Fig. 1D), which predominantly reflects CYP1A1 and CYP1B1 activity, increases in this hydroxylase were found for all TCDD-treated mice, except for the Cyp1a1(Ϫ/Ϫ) knock-out line. The constitutive benzo[a]pyrene hydroxylase activity shown in Fig. 1D reflects a CYP2C enzyme, which shows hepatic benzo[a]pyrene hydroxylase activity in untreated mice (52). For acetanilide 4-hydroxylase (Fig. 1E), which predominantly reflects CYP1A2 activity, increases in this activity were found for all TCDD-treated mice, except for the Cyp1a2(Ϫ/Ϫ) mouse. The constitutive acetanilide 4-hydroxylase, which is detected in all groups except Cyp1a2(Ϫ/Ϫ) mice, parallels the basal CYP1A2 mRNA levels found in all untreated groups except Cyp1a2(Ϫ/Ϫ) mice.

Effect of the Three CYP1 Family Members on TCDD-induced Fetal
Death-High doses of TCDD are known to cause wasting, which is a hallmark of acute TCDD toxicity (7). The dose of TCDD that we used (25 g/kg) had no effect on the body weight of dams of the four genotypes ( Fig. 2A), suggesting the absence of TCDD toxicity; alternatively, because the experiment was terminated after only 8 days of TCDD exposure, it is probably too soon to see any manifestation of significant wasting. Similarly, no significant effect on litter size was seen among the four groups (Fig. 2B). Because TCDD gavage was performed on GD10, which is well after implantation (GD3.5), it is not surprising that we saw no effect on the number of implantation sites, which was assessed on GD18. Thus, this dose of TCDD on GD10 did not alter the number of implanted fetuses between control and TCDD-treated wild-type, as well as any of the three Cyp1 knock-out lines.
We conclude that loss of any of the Cyp1 family members does not protect mice against TCDD-induced teratogenesis or fetal lethality. Importantly, neither CYP1A1 nor CYP1B1 plays any substantial role in protection against, or potentiation of, these TCDD-induced effects. Loss of CYP1A2, on the other hand, sensitizes fetuses to TCDD-induced lethality and teratogenesis; for the remainder of this report, we will address the mechanism by which CYP1A2 protects against these effects.
Maternal CYP1A2 Protects against TCDD-induced Fetal Lethality-To distinguish between the protective capacity of the maternal versus FIGURE 2. Effect of TCDD, given at GD10, on parameters of the dam and fetus. Treatment and group sizes of the four genotypes were the same as in Fig. 1. Mothers were first weighed (A) on GD18, following which litter size (B), average fetal weights (C ), and percent lethality (D) were determined. The number of viable fetuses that were weighed in 3 to 5 litters/group ranged between 35 and 48 pups total; on the other hand, due to extensive lethality, only four TCDD-treated Cyp1a2(Ϫ/Ϫ) pups survived to be weighed (Fig. 2C ). *, significantly different ( p Ͻ 0.001) from controls of the same genotype. **, significantly different ( p Ͻ 0.001) from the other three TCDD-treated groups (as well as from controls of the same genotype). paternal Cyp1a2 genotype, reciprocal crosses were conducted between Cyp1(ϩ/ϩ) wild-type and Cyp1a2(Ϫ/Ϫ) mice (Table 1). In either cross with the female Cyp1a2(Ϫ/Ϫ), TCDD-induced fetal lethality was almost 100% and was not different, whether the male genotype was Cyp1(ϩ/ϩ) or Cyp1a2(Ϫ/Ϫ). Thus, presence of paternal CYP1A2 is not protective. In either cross with the female Cyp1(ϩ/ϩ), TCDD-induced fetal lethality was only slightly different when the male genotype was Cyp1a2(Ϫ/Ϫ) (12%), compared with when the male genotype was Cyp1(ϩ/ϩ) (2%). Thus, it is very striking that the maternal Cyp1a2 genotype is necessary and sufficient to protect the fetus from TCDDinduced lethality.
In the Cyp1a2(Ϫ/Ϫ) maternal-fetal unit, neither the mother nor the fetuses had (basal or inducible) CYP1A2; in the wild-type maternal-fetal unit, both did. In the reciprocal crosses (Table 1, last two rows), all embryos are Cyp1a2(ϩ/Ϫ). The results from reciprocal crosses therefore suggest that embryonic CYP1A2 plays little or no role in protection, in the context of a Cyp1a2(Ϫ/Ϫ) mother. We conducted crosses between female Cyp1a2(ϩ/Ϫ) and male Cyp1a2(Ϫ/Ϫ) mice; fetuses in this cross are predicted to be 50% Cyp1a2(ϩ/Ϫ) and 50% Cyp1a2(Ϫ/Ϫ). The heterozygous mother was found to be as protective as the Cyp1(ϩ/ϩ) homozygous mother (data not shown). We therefore conclude that the fetal Cyp1a2 genotype has little or no influence on TCDD-induced fetal lethality. Moreover, even one maternal Cyp1a2 allele appears to be sufficient to protect the fetus from TCDD-induced lethality.
Maternal CYP1A2 Protects the Fetus from TCDD Exposure-To shed light on the mechanism by which the maternal Cyp1a2 genotype protects against TCDD-induced fetal lethality and teratogenesis, we examined the distribution of [ 3 H]TCDD in maternal and fetal tissues. In Cyp1(ϩ/ϩ) dams carrying Cyp1(ϩ/ϩ) fetuses, TCDD was sequestered at highest concentrations in liver (Fig. 4A). In the absence of CYP1A2 in the dam and fetuses, hepatic TCDD sequestration was dramatically lower, and TCDD levels in adipose and mammary gland were substantially increased. Thus, as previously shown (7,38,53,54), large amounts of CYP1A2 cause a redistribution of TCDD to liver, at the expense of TCDD sequestered in tissues with high levels of fat, which include adipose tissue and mammary gland. TCDD from these latter sources may be a source for easily available TCDD, because redistribution of TCDD from liver to fat also results in higher levels of circulating TCDD in the blood (Fig. 4B). Most notably, absence of maternal hepatic CYP1A2 resulted in ϳ2-fold higher levels of TCDD in the fetal whole body and liver (Fig. 4C) and ϳ30-fold higher levels in fetal blood (Fig. 4D).
Maternal CYP1A2 Protects the Fetus from TCDD-induced Mitochondrial Changes-The one or more mechanisms of TCDD-induced pathologies are poorly characterized. Several reports have shown that TCDD produces oxidative stress (55)(56)(57)(58). Low dose TCDD causes a sustained oxidative stress in liver that is dependent on the AHR but independent of the presence of CYP1A1 or CYP1A2 (49, 59, 60). Reac-tive oxygen species are apparently mitochondrial in origin, as the result of alterations in the mitochondrial respiratory chain and glutathione levels (50,61,62). In vitro, TCDD-induced mitochondrial reactive oxygen can be measured as succinate-dependent H 2 O 2 liberation. Using a treatment regimen similar to the one used to delineate TCDD tissue distribution, we measured succinate-stimulated H 2 O 2 liberated from fetal hepatic mitochondria (Fig. 5). Fetuses carried by TCDD-treated Cyp1(ϩ/ϩ) dams showed a trend (not significant, p Ͼ 0.064) toward enhanced hepatic mitochondrial H 2 O 2 production, compared with control dams treated with corn oil alone. Fetuses carried by Cyp1a2(Ϫ/Ϫ) dams showed significant Ͼ2-fold increases in succinatestimulated H 2 O 2 liberated from hepatic mitochondria, compared with fetuses from corn oil-treated controls of either genotype or fetuses from TCDD-treated Cyp1(ϩ/ϩ) dams. The magnitude of this response (ϳ2.5-fold) is similar to that previously observed for increases in succinate-stimulated H 2 O 2 production in adult mouse liver (62). Although we are unable yet to appreciate the complete relationship between mitochondrial H 2 O 2 production and fetal pathology, these data demonstrate the role of CYP1A2 (likely maternal hepatic CYP1A2) in protection against a potentially important end-point of TCDD toxicity in the fetus. Moreover, this finding, along with the TCDD distribution data described above, emphasize the role of maternal hepatic CYP1A2 in protection against fetal TCDD exposure.
Maternal CYP1A2 Protects against TCDD-induced Teratogenesis-We have shown that, of the three CYP1 family members, only CYP1A2 has an impact on TCDD-induced fetal viability and teratogenesis. To understand the potential importance of this effect, we treated Cyp1(ϩ/ϩ) and Cyp1a2(Ϫ/Ϫ) dams, carrying like-genotype fetuses, with increasing doses of TCDD and assessed fetal lethality and terato- versus 12% is not regarded as statistically significant ( p Ͻ 0.05). TCDD (25 g/kg) was given by gavage on GD10, and neonatal viability was assessed on GD18.

FIGURE 4. Distribution of radiolabeled TCDD in maternal liver, adipose tissue, and mammary gland (A) and maternal blood (B), and in fetal liver, whole body carcass with liver removed, and placenta (C) and fetal blood (D), 24 h after a single TCDD dose.
TCDD was given to the pregnant dams on GD15, and the analysis of tissue TCDD burden was carried out 24 h later on GD16. n ϭ 3 Cyp1(ϩ/ϩ) and n ϭ 3 Cyp1a2(Ϫ/Ϫ) maternal-fetal units. Cyp1(ϩ/ϩ) dams had significantly ( p Ͻ 0.0001) more TCDD in liver than did Cyp1a2(Ϫ/Ϫ) dams. Cyp1a2(Ϫ/Ϫ) dams had significantly (p Ͻ 0.001) more TCDD in adipose tissue, mammary gland, and circulating blood than did Cyp1(ϩ/ϩ) dams. Fetuses from Cyp1a2(Ϫ/Ϫ) dams showed ϳ2.5-fold more TCDD in whole body and liver and ϳ30-fold more TCDD in blood than fetuses from Cyp1(ϩ/ϩ) dams. *, significantly different ( p Ͻ 0.001) from wild-type mice receiving same dose of TCDD, except for panel C where Cyp1a2(Ϫ/Ϫ) liver is significantly different (p Ͻ 0.05) from wild-type mice receiving same dose of TCDD. genesis (Fig. 6). As noted above, TCDD treatment reproducibly causes very high rates of hydronephrosis and cleft palate, as a consequence of TCDD exposure to the fetus. Almost all fetuses carried by Cyp1a2(Ϫ/Ϫ) dams die prior to GD18, when administered TCDD at doses of 25 g/kg and greater; therefore, teratogenesis was difficult to quantify in Cyp1a2(Ϫ/Ϫ) fetuses at these doses. Fetal lethality was the least sensitive end-point of TCDD toxicity. Fetuses carried by Cyp1(ϩ/ϩ) dams were protected against a ϳ6-fold higher dose of TCDD, compared with fetuses carried by Cyp1a2(Ϫ/Ϫ) dams (Fig. 6A). TCDD-induced cleft palate (Fig. 6B) and hydronephrosis (Fig. 6C) show similar dose-response relationships that are dependent on CYP1A2. Fetuses carried by Cyp1(ϩ/ϩ) dams were protected against a ϳ9-fold higher dose of TCDD for both of these birth defects, compared with fetuses carried by Cyp1a2(Ϫ/Ϫ) dams. Human CYP1A2, in Place of Mouse CYP1A2, Protects Fetuses against TCDD-induced Teratogenesis-We have shown that mouse CYP1A2 functions to protect the fetus against TCDD. The CYP1A2 ortholog is conserved among mammals, but it should be noted that its evolutionarily closest relative, CYP1A1, does not function to sequester TCDD in liver (7); furthermore, as shown above, CYP1A1 does not protect the fetus against TCDD. Is hepatic sequestration of TCDD a conserved function of CYP1A2? Is human CYP1A2 able to serve this function as well as mouse CYP1A2? We have generated and characterized transgenic mice that contain a human bacterial artificial chromosome (BAC), expressing both the human CYP1A1 and CYP1A2 genes, which are oriented head-to-head and separated by 23.3 kb (63,64); this hCYP1A1_1A2 BAC-transgenic mouse line is on a C57BL6/J and Cyp1a2(Ϫ/Ϫ) background. Using these mice, we performed a doseresponse experiment similar to that shown in Fig. 6, and the data collected using the hCYP1A1_1A2 line are compared with the Cyp1a2(Ϫ/Ϫ) data in Fig. 7. When considering fetal lethality, cleft palate, or hydronephrosis, we found that the human BAC functionally replaced the missing mouse Cyp1a2 gene, reverting the Cyp1a2(Ϫ/Ϫ) mouse to the wild-type phenotype.
It should be noted that the hCYP1A1_1A2 transgenic mouse contains a single copy of the human BAC, which expresses both the human CYP1A1 and CYP1A2 genes. Thus, hCYP1A1_1A2 mice carry three copies (two mouse and one human) of CYP1A1 and one copy of the human CYP1A2 gene. Because mouse CYP1A1 plays no discernable role in TCDD-induced fetal pathology, we believe it is extremely unlikely that human CYP1A1 would play any such role. As a consequence, we think it is very likely that protection against TCDD-induced fetal pathology in the hCYP1A1_1A2 transgenic mouse is the consequence of human CYP1A2 gene expression, which, like mouse Cyp1a2 gene expression, is constitutively expressed at high levels in mouse liver. Moreover, as shown above, heterozygous Cyp1a2(ϩ/Ϫ) maternal liver

FIGURE 6. Semi-log dose-response curves of TCDD-induced fetal lethality (A), incidence of cleft palate (B), and incidence of hydronephrosis (C) in Cyp1(؉/؉) versus Cyp1a2(؊/؊) mice.
On GD10, pregnant Cyp1(ϩ/ϩ) and Cyp1a2(Ϫ/Ϫ) dams were given TCDD by gavage at doses of 0, 1, 2.5, 6.5, 12.5, 25, 35, 50, 100, or 200 g/kg. At GD18 the embryos were harvested and scored for lethality and the two birth defects. Each data point represents pooled data from at least three litters derived from three separate dams. Symbols and brackets denote mean Ϯ S.E., respectively; in some cases, no brackets can be seen because the S.E. is smaller than the symbol. Curves were fit to a sigmoidal dose-response curve, using nonlinear regression.
protects as well as homozygous Cyp1(ϩ/ϩ) maternal liver, suggesting that one human CYP1A2 allele might also be sufficient. We cannot at present rule out with absolute certainty, however, the possibility that the human CYP1A1 might play a role in TCDD-induced fetal pathology.

DISCUSSION
Many studies have suggested that CYP1 metabolism is required for mediating TCDD-induced pathology: TCDD is the prototype of several environmentally ubiquitous planar teratogenic, carcinogenic, and toxic polyhalogenated aromatic hydrocarbon (PHAH) contaminants. Toxicity within this class of compounds is correlated with AHR-binding affinity and rate of clearance (6,65,66). Therefore, dioxins, dibenzofurans, polychlorinated biphenyls, and polybrominated biphenyls, which adopt planar molecular architecture consistent with the AHR-ligand binding site and resist monoxygenation due to the halogen pattern whereby no two adjacent carbon atoms are without a halogen atom, are, in general, the most toxic. Such compounds persistently activate the AHR, and such activation is essential for teratogenic, carcinogenic, and toxic endpoints (10,25,(67)(68)(69).
The AHR controls transcription of the Cyp1 family of monooxygenase genes; these genes, although not evolutionarily related to the AHR gene, encode enzymes that seemingly contain ligand-binding domains that are similar to the AHR ligand-binding domain. Thus, planar PAHs and PHAHs that bind and activate the AHR are oxygenated by CYP1 enzymes and prepared for phase II elimination. The notion that CYP1 genes would be downstream mediators of PHAH-mediated AHRdriven pathology stems from in vitro experiments that demonstrate an interaction between planar PHAHs and CYP1 enzymes (58,62,70). Because these compounds resist oxygenation, oxygen, reduced by the CYP1-PHAH complex, is liberated as reactive oxygen, either as superoxide or H 2 O 2 , and is very likely to be important in toxicity caused by PHAHs. Although this reaction may indeed occur in the intact animal, the present study and our previous studies suggest that CYP1 family members play a limited role in mediating PHAH toxicity (7,38,41); an exception to this statement is high dose TCDD-mediated hepatotoxicity, which is greatly decreased in Cyp1a1(Ϫ/Ϫ) or Cyp1a2(Ϫ/Ϫ) mice (7,38). In the liver of these mice, TCDD fails to elicit the uroporphyric response, which is arguably the cause of hepatotoxicity in this model. Thus, chronic up-regulation of CYP1 enzymes may cause toxicity, under certain conditions and in certain cell types, but, unlike the AHR, these enzymes may not play a generalized role in TCDD-mediated pathology. To the contrary, as shown herein, CYP1A2 functions to protect the developing fetus against TCDD-induced teratogenesis and lethality, whereas CYP1A1 and CYP1B1 play no discernable role in these processes.
CYP1A2 Is an Abundant Hepatic PHAH-binding Protein-CYP1A2 is inducible in several tissues, although it does not display the ubiquitous pattern of inducible expression as CYP1A1 does (35,69). In contrast to CYP1A1, however, the hepatic CYP1A2 protein is present at substantial levels in the absence of any inducer. Strong evidence that mouse CYP1A2 is a hepatic TCDD-binding protein was provided by the demonstration that protein-bound radiolabeled TCDD co-purified with a protein that reacted with antibodies, which recognized CYP1A proteins (65). Confirmation that CYP1A2 is a TCDD-binding protein was then gleaned using Cyp1a2(Ϫ/Ϫ) mice (7,38,66,71).
Herein we have shown that maternal CYP1A2 protects against fetal TCDD-induced teratogenesis and lethality. In agreement with its role in binding TCDD, Cyp1a2(Ϫ/Ϫ) mice have low levels of hepatic TCDD, which lead to much higher levels of maternal blood TCDD levels, which are passed to the fetus. As a consequence of the unequivocal identification of CYP1A2 as the hepatic TCDD-binding protein, earlier studies may now clearly be interpreted to demonstrate that CYP1A2, like the AHR, binds planar PAHs, such as benzo[a]pyrene, a known teratogen and carcinogen, in addition to planar PHAHs that include PCBs, polybrominated biphenyls, dioxins, and dibenzofurans (65,72,73). Thus, CYP1A2 functions to anchor such compounds in a nonlabile pool, thereby limiting bioavailability. Cyp1a2(Ϫ/Ϫ) dams were given TCDD by gavage at doses of 0, 12.5, 25, 50, 100, or 200 g/kg. At GD18, the embryos were harvested and scored for lethality and the two birth defects. Each data point represents pooled data from at least three litters derived from three separate dams. Symbols and brackets denote mean Ϯ S.E., respectively; in some cases, no brackets can be seen because the S.E. is smaller than the symbol. Curves were fit to a sigmoidal dose-response curve, using nonlinear regression.
The hepatocyte is uniquely adapted to serve as host for this function, because it possesses both strong metabolic and excretory properties. In the case of many PHAHs, metabolism is very slow, and binding to CYP1A2 is therefore likely to be long term. In the case of PAHs such as benzo [a]pyrene, binding to CYP1A2 may function as a hepatic anchor, while the appropriate metabolic enzyme, CYP1A1, in this case, becomes induced.
Body Pools of TCDD-Maternal hepatic CYP1A2 binds to TCDD and protects the fetus. In the absence of CYP1A2, TCDD distributes to adipose tissue and, as a consequence of this, becomes more bioavailable. It is important, therefore, to consider the trafficking of TCDD, from the time it is ingested to the time it is deposited at its site of storage. Although there is little information regarding TCDD uptake, other lipophilic contaminants that resist metabolism in the small intestine are similar to lipid molecules themselves, packaged into chilomicrons, deposited into the lymphatics, which dump into the bloodstream (74,75). Some lipophilic compounds such as retinal ester are packaged into the lipophilic chilomicron core. Lipophilic compounds that exist in the chilomicron core are eventually taken up by liver, where the chilomicron is converted by multiple enzymatic events to chilomicron remnants. If TCDD is trafficked in this manner, its delivery to liver would be anticipated. However, some lipophilic pollutants, such as the planar small molecule hexachlorobenzene, are associated with chilomicrons but are rapidly distributed to tissues (particularly adipose) and are not enriched in chilomicron remnants taken up by liver. It is thus possible that TCDD is initially distributed to adipose tissue, and then makes its way to liver; a carrier lipid particle for such trafficking is not yet known.
From the present study, it seems likely that fetal protection against TCDD-induced pathology is a consequence of hepatic CYP1A2-binding in Cyp1(ϩ/ϩ) wild-type mice, as opposed to the more labile adipose-associated TCDD that is present in Cyp1a2(Ϫ/Ϫ) knock-out mice; this is reflected by the dramatic inverse relationship of blood levels of TCDD on CYP1A2 (Fig. 4). The labile nature of lipophilic contaminants stored in adipose has been demonstrated in rats (75). Indeed, such compounds become remarkably more bioavailable during usage of dietary fat stores. There is evidence that a similar process occurs in humans (76). It seems reasonable to speculate that TCDD, bound to CYP1A2, would not be sensitive to dietary changes, because, if TCDD were liberated from CYP1A2, it would activate new synthesis of hepatic CYP1A2 protein via the AHR. Thus, the AHR and CYP1A2 likely participate in a regulatory loop, which sequesters TCDD (and, likely, other planar PHAHs) in the liver. Thus, broadly speaking, CYP1A2 with its high constitutive hepatic-binding capacity represents a mechanism regulated by the AHR in hepatocytes to protect the AHR in peripheral or developing tissues from illicit activation.
Does Human CYP1A2 Protect against Birth Defects?-We have demonstrated a role for mouse maternal hepatic CYP1A2 in protection against TCDD-induced teratogenesis and lethality. Moreover, mouse CYP1A2 tightly binds a myriad of planar PHAHs and PAHs (71), suggesting that it might perform a similar role in protection against an array of environmental contaminants. We have previously generated a hCYP1A1_1A2 BAC-transgenic mouse line (63). The tissue-specific expression pattern and inducibility of the human CYP1A genes in these mice suggest that sufficient genomic regulatory sequences are present in the 180-kb BAC to direct authentic expression of these genes, similar to what is observed with the human or mouse endogenous CYP1A genes (77)(78)(79)(80). We generated these mice primarily because of reported catalytic differences between rodent and human CYP1A2, and in response to the need for a better model for human xenobiotic metabolism and risk assessment in laboratory animals. Indeed, we have shown that metabolism of both the anti-asthmatic drug theophylline (64) and the procarcinogen 2-amino-1-methyl-6-phenylimidazo [4,5-b]pyridine display a humanized metabolite profile in these transgenic mice (64,81).
Although there is evidence that planar PHAHs in humans are sequestered in liver (82), the mechanism has never been determined. Because hCYP1A1_1A2 transgenic dams on the mouse Cyp1a2(Ϫ/Ϫ) background protect fetuses against TCDD-induced teratogenesis and lethality (Fig. 7), similar to that seen in Cyp1(ϩ/ϩ) wild-type mice, we think it is extremely likely that human CYP1A2 binds and sequesters TCDD (and likely other planar PHAHs) similar to mouse CYP1A2. Thus, human CYP1A2, like mouse CYP1A2, may function to protect the developing fetus against planar PHAH exposure.
With the exception of CYP2E1, which metabolizes several dozen small drugs and many small environmental chemicals, and CYP3A4, which metabolizes Ͼ50% of all commonly prescribed drugs, CYP1A2 is considered the next most abundant constitutive CYP in human liver (83). It must be noted, however, that the constitutive expression of hepatic CYP1A2 is extremely variable in humans, with Ͼ60-fold differences between high and low expressing individuals (69,84,85). Thus far, no single-nucleotide polymorphisms within or near the human CYP1A2 gene or other genes explain these differences (86).
Is a polymorphic human AHR responsible for these differences? Across common animal model species, AHR allelic variants that drastically alter function have been reported. In mouse, the Ala-375 3 Val mutation in the ligand-binding domain of the AHR decreases ligandbinding affinity by almost 10-fold, leading to large differences in responses to many natural and environmental chemicals (11,87,88). In rat, a single-nucleotide polymorphism in the portion of the Ahr gene encoding the C-terminal trans-activation domain produces an AHR that activates gene expression seemingly normally but does not cause toxicity in response to TCDD (reviewed in Ref. 55). In humans, several AHR variants have been discovered (89 -91), but, thus far, none of these variants has been shown to be related to the Ͼ12-fold difference in K d values obtained from Scatchard analysis of [ 3 H]TCDD binding to cytosol from 115 placental samples (69). There are also well documented differences in CYP1 inducibility within human populations (89 -93); the molecular basis for this is unknown, although CYP1 inducibility has been linked to the AHR locus (94).
Based on data from mouse (see above), constitutive as well as inducible Cyp1a2 expression is controlled by the AHR. If this is true in both humans and mice, then DNA tests (when they become available) to determine the AHR gene polymorphism, as well as the human CYP1A2 polymorphism, should be considered important, when trying to determine the genetic basis for hepatic CYP1A2 levels.
In humans, could the combination of exposure to planar PHAHs, and high degree of variability in hepatic CYP1A2 levels, be important in risk assessment of birth defects? The answer to this question is not yet known, but the present study provides the basis for speculation about an "at-risk" population. First, the genotype of the affected fetus may not be a determinant of teratogenic risk; rather, instead, the maternal genotype appears to be more important. Second, the highest risk individual is likely to be one whose mother has genetically very low CYP1A2 activity, combined with a fetus who expresses the highest AHR inducibility. This notion follows from the observation that, although fetal CYP1A2 does not contribute to teratogenesis (see above), perhaps because it is not expressed in the fetus (95), the fetal AHR is essential for PHAH teratogenesis (15,18). It should be noted, of course, that determination of the precise level of environmental PHAH exposure would be expected to be a confounding factor in such genotype-phenotype association studies; in addition, no single-nucleotide polymorphism or haplotype in or near either the human CYP1A2 or AHR gene has unequivocally been shown to reflect variations in the CYP1A2 or AHR phenotype, to date (69,86).
In summary, in the mouse, maternal hepatic CYP1A2 binds and sequesters TCDD, which, in turn, protects the developing embryo from teratogenesis and lethality. Human CYP1A2 also is capable of this same sequestering function. Because CYP1A2 is known to bind to many planar PHAHs, its role in protection may extend well beyond protection against TCDD. The fact that human populations show Ͼ60-fold differences in hepatic basal CYP1A2 levels might be very important in risk assessment for planar PHAH-induced birth defects.