Retinoic Acid Drives Aryl Hydrocarbon Receptor Expression and Is Instrumental to Dioxin-Induced Toxicity during Palate Development

Background: Palate development depends on complex events and is very sensitive to disruption. Accordingly, clefts are the most common congenital malformations worldwide, and a connection is proposed with fetal exposure to toxic factors or environmental contaminants, such as dioxins. There is increasing evidence that dioxin interferes with all-trans-retinoic acid (atRA), a hormone-like signal derived from vitamin A, which plays an essential role during embryonic development. Although similarities have been described between dioxin-induced toxicity and the outcome of altered atRA signaling during palate development, their relationship needs to be clarified. Objectives: We used a genetic approach to understand the interaction between atRA and dioxin and to identify the cell type targeted by dioxin toxicity during secondary palate formation in mice. Methods: We analyzed the phenotype of mouse embryos harboring an atRA-sensitive reporter transgene or bearing null mutations for atRA-synthesizing enzymes (RALDH) or atRA receptors (RAR) and maternally exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) at gestation day 10.5. Results: We found that an intact atRA signal was required to enable TCDD to induce cleft palate. This mandatory atRA signal was generated through the activity of RALDH3 in the nasal epithelium and was transduced by RARγ (RARG) in the nasal mesenchyme, where it notably controlled aryl hydrocarbon receptor (Ahr) transcript levels. TCDD also did not alter the developmental pattern of atRA signaling during palate formation. Conclusions: TCDD-induced alteration of secondary palate development in the mouse appears to depend on atRA signaling, which controls AHR expression. This mechanism is likely conserved throughout vertebrate evolution and may therefore be relevant in humans.

volume 119 | number 11 | November 2011 • Environmental Health Perspectives Research Embryonic development of the mamma lian face relies on a sequence of complex and interdependent molecular, cellular, and tis sue interactions. Disruption of these develop mental processes often results in orofacial defects, which are the most common of all congenital disorders in humans. The second most frequent non syndromic orofacial defect is isolated cleft palate, which has a frequency of approximately 1:2,500 in European new borns (Cobourne 2004). Its effect on speech, hearing, appearance, and cognition leads to longlasting adverse outcomes impair ing social integration (Vallino et al. 2008). Isolated cleft palate is currently considered to have a multifactorial etiology in which a deleterious genetic background is combined with environmental factors. Gene linkage and association studies indicate that the patho genesis of non syndromic isolated cleft palate relies on a combination of multiple mutations in different genes (Lidral et al. 2008), and epidemiological studies have revealed a cor relation between increased risk for oral cleft and exposure to dioxinlike chemicals during pregnancy (Leite et al. 2002;Murray 2002). In laboratory animals, notably in mice, expo sure to 2,3,7,8tetrachloro dibenzopdioxin (TCDD) during organogenesis causes cleft palate (Courtney and Moore 1971;Couture et al. 1990). Because developing palatal shelves of human, rat, and mouse respond similarly to TCDD exposure in organ cultures (Abbott et al. 1999;Couture et al. 1990), the mouse is an ideal in vivo model for studying teratogenic effects of dioxins.
At the molecular level, TCDD alters gene expression by activating the aryl hydrocarbon receptor (AHR). Unliganded AHR resides in the cytoplasm in an inactive complex with heatshock molecules. Binding of TCDD to AHR causes the dissociation of the receptor from the complex and its translocation into the nucleus, where it dimerizes with the AHR nuclear translocator (ARNT). The AHR/ ARNT heterodimer functions as a transcrip tional activator by binding to specific DNA sequences called dioxin response elements (DREs) located in the regulatory regions of AHRresponsive genes (Beischlag et al. 2008;Pohjanvirta et al. 2011). Mice with a homozy gous ablation of the Ahr gene suffer from vari ous agerelated pathologies; this suggests that AHR exerts important physiological functions (FernandezSalguero et al. 1995). Thus, under standing the molecular mechanisms through which TCDD exposure results in a cleft palate may provide clues not only to the mechanisms of TCDD teratogenicity but also to the nature of homeostatic AHR functions.
There is increasing evidence that environ mental pollutants such as dioxinlike com pounds interfere with alltransretinoic acid (atRA) signaling (Novák et al. 2008). atRA is a pleiotropic, paracrine or autocrine signaling molecule produced from vitamin A through oxidative reactions carried out by the cyto solic retinaldehyde dehydro genases RALDH1, RALDH2, and RALDH3 (Duester 2008). atRA acts as a hormone by binding to and activating α, β, and γ isotypes of atRA recep tors (RARs; RARA, RARB, RARG, respec tively), which belong to the nuclear hormone receptor superfamily and function as ligand dependent transcription factors inter acting with regulatory regions located in target genes. In a large variety of tissues, RARs act by forming heterodimers with one of the rexinoid receptors, retinoid X receptor (RXR; α, β, or γ isotypes), which bind 9cis but not atRA (Mark et al. 2006). Similarities between dioxin toxicity and atRA deficiency or excess have often been pointed out (Nilsson and Håkansson 2002;Novák et al. 2008). Accordingly, atRA excess induces a cleft palate (Abbott et al. 1989), as does TCDD expo sure (Courtney and Moore 1971;Couture et al. 1990). In many instances, the effects of TCDD on atRAcontrolled processes in vitro appear to be mediated by AHR either interfer ing positively or negatively with atRA signal ing in certain cell types or changing activity of the enzymes responsible for transformation of retinoids (Novák et al. 2008). However, further investigation is needed to confirm that the mechanisms shown to operate in vitro are indeed mediating TCDDinduced defects in vivo.
In this study, we used mouse embryos har boring null mutations in the genes coding for RALDH3, RARA, or RARG to unravel the possible interaction between atRA and TCDD during palate development and to reassess the etiology of TCDDinduced cleft palate. We demonstrate that TCDD does not alter the pattern of atRA signaling in the embryonic face. However, we show that an atRA signal generated through the activity of RALDH3 in the nasal epithelium and transduced by RARG in the nasal mesenchyme is mandatory to enable TCDD to induce cleft palate when administered at gestation day (GD) 10.5, notably through controlling the levels of Ahr expression. In addition, our results suggest that TCDD acts not directly on the develop ing palatal shelves, but on the mesenchyme adjacent to the nasal epithelium.

Materials and Methods
Animal use. Mice were housed in an animal facility licensed by the French Ministry of Agriculture (agreement B672185). Animal experiments were supervised by one of the authors who is qualified for experimenting with mice, in compliance with the European legislation on care and use of laboratory ani mals (agreement 67205). The mice were treated humanely and with regard for allevia tion of suffering.
Mice genotyping and treatments. The trans genic line Tg(RARE-Hspa1b/lacZ)12Jrt and the lines carrying the Ahr tm1Bra , Rara tm3.1Ipc , Rarg tm1Ipc , and Aldh1a3 tm1.1Pcn null alleles were genotyped as previously described (Chapellier et al. 2002;Dupé et al. 2003;Lohnes et al. 1993;Rossant et al. 1991;Schmidt et al. 1996). Noon of the day a vaginal plug was observed was considered GD0.5. At GD10.5, pregnant mice were given a single dose of 100 mg/kg atRA (Biomol International, Plymouth Meeting, PA, USA) or 30 μg/kg TCDD (Wellington Laboratories, Guelph, ON, Canada) dissolved in sunflower oil (Sigma, Lyon, France) by oral gavage. The number of fetuses and litters analyzed and an over view of cleft palate occurrence as a function of treatments and genotypes are presented in Supplemental Material, Phenotype analysis. We stained skeletons with Alcian blue and Alizarin red as previously described (Lufkin et al. 1992). For detection of βgalactosidase activity, we performed 5bromo4chloro3indolylbetadgalacto pyranoside (XGal)based staining (Rossant et al. 1991) and embryos were post fixed in Bouin's fluid, embedded in paraffin, seri ally sectioned, and then counterstained with eosin. Wholemount in situ RNA hybridiza tion was performed as previously described (Wendling et al. 2001). In situ hybridization and immuno histochemistry on cryosections were also performed as previously described (Vernet et al. 2006), using embryos that were fixed for 1 hr in 4% (wt/vol) phosphate buffered paraformaldehyde at 4°C. RNA analysis. We prepared transverse slices of the nasopalatal region from GD11.5 embryos (n ≥ 3 for each condition) from which the eyes and the maxillary component of first bran chial arches were removed. Wildtype (WT) or RARdeficient (Rara -/-/Rarb -/-/Rarg -/-) mouse embryonic fibroblasts (MEFs) were maintained as previously described (Epping et al. 2007). At subconfluency, we added cycloheximide (10 -6 M) 1 hr before atRA (10 -6 M) and har vested cells 6 hr after atRA. We extracted total RNA using Trizol reagent (Invitrogen, Life Technologies, VillebonsurYvette, France), converted it to cDNA, and analyzed it by real time polymerase chain reaction performed in a Realplex Mastercycler (Eppendorf, Le Pecq, France). We normalized the transcript levels relative to that of Rplp0 (ribosomal protein, large, P0) transcript (MGI:1927636), whose expression is not altered in mutant mice or in atRA or TCDDtreated fetuses. We analyzed each sample in triplicate and assessed results using Student's ttest.

TCDD administration and excess atRA at GD10.5 induce identical cleft palates.
To com pare the morphological outcomes of TCDD and atRA treatments on palatal develop ment, we analyzed skeletons of 34 GD18.5 fetuses. An oral dose of TCDD (30 μg/kg) to pregnant WT mice at GD10.5 always (n = 27 fetuses) inhibited the development of the palatal pro cesses of the maxillary bones, which were hypoplastic, as well as those of the palatine bones, which were agenic ( Figure 1B). In con trast, other parts of these bones (e.g., alveo lar, orbital, and zygomatic processes) were normal [see Supplemental Material, Figure 1 (http:// dx.doi.org/10.1289/ehp.1003075)]. Treatment of pregnant WT mice with atRA (100 mg/kg) at GD10.5 also systematically induced a cleft palate (n = 7 fetuses), which was indistinguish able from its TCDDinduced counterpart ( Figure 1C; see also Supplemental Material, Figure 1) and was not accompanied by other craniofacial defects. Therefore, both TCDD exposure and atRA excess at GD10.5 induce a cleft palate through inhibition of palatal shelf development. This finding raised the possibility either that atRA activates AHR or that TCDD mimics the effects of atRA excess through acti vating this pathway.
Excess atRA does not produce cleft palate through activating AHR, and reciprocally, TCDD does not modify the developmental pattern of atRA signaling in the face. To test for these hypotheses, we first intercrossed Ahr +/mice and then fed the pregnant females atRA at GD10.5 and examined their prog eny at GD18.5. We observed a cleft palate in all Ahr -/fetuses (n = 5), thus ruling out the possibility that atRA was activating AHR to induce a cleft palate [see Supplemental Material, Figure 2 (http://dx.doi.org/10.1289/ ehp.1003075)]. Then, to test whether TCDD activated atRA signaling, we analyzed its effects on embryos harboring the Tg(RARE-Hspa1b/ lacZ)12Jrt transgene (Rossant et al. 1991), which is used to monitor variations in atRA signaling (Duester 2008). We analyzed its activity at GD10.75, GD11.5, and GD12.5 (n = 5 embryos at each developmental age) and found that control and TCDDtreated fetuses displayed identical patterns of XGal staining in the frontonasal region (Figure 2), indicating that the developmental pattern of atRA signaling remained unaltered upon TCDD treatment. To extend these observa tions, we quantified the mRNA levels of lacZ and endogenous atRAtarget genes, including Crabp2 (cellular retinoic acid binding protein), Rara, and Rarg (Balmer and Blomhoff 2002).
In agreement with the lack of effect of TCDD on transgene activity, we did not detect a sig nificant difference for any of these four genes in the nasopalatal region of TCDDtreated (n = 3) or control (n = 3) embryos. In addi tion, we verified that Rxra mRNA level was not altered, ruling out the possibility that TCDD acted through modulating expression of the RAR partner ( Figure 3). Importantly, we confirmed the efficiency of TCDD treat ment by the induction of Cyp1a1 (cytochrome P450 1A1), a wellestablished target gene of TCDDactivated AHR (Abbott et al. 1999).
Together, these results demonstrate that the ability of TCDD to induce a cleft palate can not be accounted for by increased or ectopi cally activated atRA signaling. TCDD is unable to induce a cleft palate when atRA signaling is impaired. The above results did not exclude the possibility that endogenous, atRAdependent events are required to allow TCDDinduced inhibition of palatal shelf development. Between GD10.5 and GD12.5, we detected atRA signals in the frontonasal region but not in the palate per se ( Figure 4B). During this period of develop ment, RALDH3 is the sole RAsynthesizing enzyme expressed in the nasopalatal region as assessed by in situ hybridization ( Figure 4A) and by the complete disappearance of Tg(RARE-Hspa1b/lacZ)12Jrt activity in Aldh1a3 -/fetuses ( Figure 4C). RARA and RARG are the sole nuclear receptors driving atRA activity during craniofacial development (Lohnes et al. 1994). To test the possibility that TCDD relies on endogenous atRA to induce cleft palate, we impaired atRA signaling in vivo through ablation of genes coding for atRA receptors or atRAsynthesizing enzymes. Thus, mutant mice carrying Rara, Rarg, or Aldh1a3null alleles were mated, and we dosed pregnant females at GD10.5 with TCDD or vehicle [see Supplemental Material, "Material and Methods" (http://dx.doi.org/10.1289/ ehp.1003075)]. We analyzed skulls from GD18.5 null mutants (i.e., Rara -/-, Rarg -/-, and Aldh1a3 -/-) and control WT littermates (n ≥ 3 for each genotype and treatment). All Rara -/mutants exposed to TCDD consis tently displayed a cleft palate ( Figure 4D). In contrast, the palates of TCDDtreated Rarg -/and Aldh1a3 -/mutants were invariably closed ( Figure 4E,F). Analysis of dissected bones indicated that the palatal processes of maxil lary and palatine bones developed normally in Aldh1a3 -/mutants (see Supplemental Material, Figure 1). Most important, the ability of TCDD to induce cleft palate was restored in Aldh1a3 -/embryos rescued for atRA signaling through the repeated adminis tration of a low dose of atRA (2 mg/kg), which was not otherwise teratogenic for palate devel opment ( Figure 4G-I). This demonstrates that atRA is the missing signal impairing the action of TCDD in the Aldh1a3null genetic back ground. Thus, our results show that inhibi tion of palatal shelf development by TCDD actually requires atRAdependent signal(s) generated through the activity of RALDH3 and mediated by RARG. AHR expression depends on RALDH3 in the frontonasal region. The most straightfor ward explanation for the abovementioned results would be that expression of AHR relies upon atRA synthesized by RALDH3 within or close to the developing palate. To test for this possibility, we compared expression of Ahr in the nasopalatal regions isolated from GD11.5 WT (n = 3) and Aldh1a3 -/-(n = 3) embryos. We found a significantly decreased steadystate level of Ahr mRNA in mutants lacking RALDH3. Importantly, Ahr mRNA was restored to normal levels in Aldh1a3 -/embryos (n = 3) rescued for atRA signaling through the repeated administration of low, nonteratogenic doses of atRA ( Figure 5A). To further assess the effect of atRA on Ahr expression, we examined the level of Ahr mRNA in WT MEFs cultured with cyclo heximide, an inhibitor of protein translation, and found a 5fold increase upon atRA stimu lation ( Figure 5B). This indicated that atRA activated RAR controlled AHR expression without the need for intermediate protein synthesis, suggesting a direct effect of RAR on Ahr gene. Accordingly, atRA did not increase Ahr mRNA levels in RARdeficient MEFs ( Figure 5B). Together, these results support the idea that Ahr expression requires atRA generated by RALDH3 and can explain why palate formation was unaffected by TCDD treatment in Aldh1a3 -/mutants.
Together, our findings that a) RARG, but not RARA, is mandatory for TCDD to induce cleft palate and b) RAR directly controls Ahr gene expression imply that the AHRexpressing cells at the origin of the mal formation are necessarily distributed within the domain where RARG is operational. Using immunohistochemistry, we found that RARG was expressed in the mesenchyme adjacent to nasal epithelium ( Figure 5C,D) and overlapped with an active atRA signaling only in the mesenchyme distributed lateral to the nasal epithelium (compare Figures 2E and  5C). Thus, TCDD likely acts through activat ing AHR present in the mesenchyme localized lateral to the nasal epithelium.

Discussion
Several experiments underline the impact of TCDD on retinoid homeostasis, but whether it yields states of functional vitamin A deficiency or excess remains a matter for further clarification. Overall, findings on TCDDinduced growth retardation, abnor mal immune function, and developmental defects reminiscent of vitamin A-deficient states suggest that dioxinlike compounds reduce atRA signaling (Murphy et al. 2007). Several mecha nisms supporting this scenario have been proposed. First, TCDD exposure mobilizes retinyl esters from liver stores and may thus rapidly exhaust them (Nilsson and Håkansson 2002). Second, TCDD may impair RAR functioning as it inhibits both atRA binding and induction of target genes in cultured cells (Lorick et al. 1998;Weston et al. 1995). Third, TCDD may enhance atRA catabolism through inducing enzymes such as CYP1A1 and CYP2S1 that can transform atRA into less active metabolites (Lampen et al. 2000;Smith et al. 2003).
An opposite set of data suggests that dioxinlike compounds may enhance atRA signaling, particularly with regard to bone and developmental defects (Nilsson and Håkansson 2002). Different scenarios have been proposed to explain the enhancement of Rara -/mutants (D) displayed cleft palate, whereas Rarg -/-(E) and Aldh1a3 -/-(F) mutants were resistant to this malformation. Administration of low doses of atRA (2 mg/kg every 12 hr from GD8.5 to GD12.5) failed to induce cleft palate in WT (G) and Aldh1a3 -/mutant (H) embryos but was sufficient to restore TCDD toxicity for palate development in Aldh1a3 -/mutants (I). Abbreviations: i, incisive bone; if, incisive foramen; lng, lacrimonasal groove; m, maxillary bone; nr, nasal region; p, palatine bone; ppi, palatal process of incisive bone; ppm, palatal process of maxillary bone; ppp, palatal process of palatine bone; ps, presphenoid bone; v, vomer bone. Bar in I = 300 μm for A-C and 1 mm for D-I.  (Schmidt et al. 2003) because DRE have been identified in Aldh1a2 gene coding for RALDH2, the most potent atRA synthesizing enzyme (Wang et al. 2001) and because AHRinduced CYP1A1 can par ticipate in atRA synthesis (Chen et al. 2000;Tomita et al. 1996). Second, in vitro experi ments indicate that TCDD induces expres sion of Rarg and Rrxa (Murphy et al. 2004) and that activated AHR is able to divert RARΑ from its corepressor, thereby allowing the receptor to become transcriptionally active in the absence of atRA (Widerak et al. 2006).
Regarding development of the second ary palate, the present study demonstrates that TCDD and excess atRA share similar teratogenic properties when administered at GD10.5 as their effects are morphologically indistinguishable from one another, as previ ously reported (Abbott et al. 1989;Moore et al. 1973). Furthermore, a combination of low doses of each compound synergistically induces this defect ). Thus, an interaction exists between dioxin induced and atRAdependent events involved in this developmental defect. The possibil ity that a decrease in atRA signaling could account for TCDDinduced effects appears unlikely because the palatal malformations generated upon functional vitamin A defi ciency or Rar gene ablations a) affect essen tially the development of the primary palate and b) are never isolated, but instead occur as part of a holoprosencephaly syndrome (Lohnes et al. 1994). These features are clearly distinct from those of dioxininduced cleft palates. We therefore explored the possibility that TCDD exposure during gestation may enhance atRA signaling through increasing either atRA levels or RAR activity. Our exper iments using mice harboring an atRAsensitive reporter transgene, as well as the quantitative analysis of atRAtarget gene expression in the palatal region, revealed that TCDD does not modify the pattern of atRA signaling in the facial region during palate development.
In contrast, because Aldh1a3 -/and Rarg -/mutants are resistant to TCDD induced clefts, our study provides evidence that a functional, intact atRA signal origi nating from RALDH3, which is the sole atRAsynthesizing enzyme expressed in the palatal region and mediated by RARG in the facial mesenchyme, is required for TCDD to exert teratogenic effects on palate devel opment. Accordingly, we provide evidence that the atRA signaling pathway is instrumen tal in AHR signaling in the palatal region, notably through controlling Ahr transcript levels [see Supplemental Material, Figure 3 (http://dx.doi.org/10.1289/ehp.1003075)]. Even though no atRAresponse element is characterized in this gene, we show a direct involvement of atRAactivated RAR in this process. Interestingly, a role of atRA and RAR in the control of AHR expression has been previously demonstrated in medaka fish (Hayashida et al. 2004). Because RARs exert their functions heterodimerized with RXRA during mouse development (Mark et al. 2006), it is likely that RARG acts with RXRA to control AHR expression. This hypothe sis, however, cannot be confirmed because Rxra -/mutants die before the completion of palate closure (reviewed in Mark et al. 2006). Nonetheless, our finding that Rxra expression was unchanged after TCDD exposure and the fact that RXRA is not activated by atRA (Duester 2008) together support the view that RXRA plays no major role aside from its partnership with RARG. Importantly, the fact that Ahr -/mice developed cleft palates upon atRA excess allows us to rule out the possibil ity that atRA binds to and activates AHR, as opposed to other retinoids (Gambone et al. 2002;Soprano and Soprano 2003;Soprano et al. 2001). We also excluded the possibil ity that TCDD acts through binding to, and activating, RARG because ablation of Aldh1a3 was sufficient to prevent TCDD induced cleft palate.
AHR expression has been detected in the developing mouse at several sites and distinct time points (Abbott et al. 1995). In addi tion, the temporal and spatial context of AHR activation after TCDD exposure in vivo has been determined in a transgenic mouse model (Willey et al. 1998). Together, these studies, which focused mainly on elevation and fusion of palatal shelves, have highlighted roles of AHR in mesenchyme and epithelium at GD14.5 but did not identify the cell type(s) that were targeted by TCDD at GD10.5. Interestingly, our study indicates that AHR is required at GD10.5 in the mesenchyme lateral to the nasal epithelium. Therefore, at the beginning of palato genesis, TCDD does not act directly on nascent, firstarch-derived palatal shelves, but rather on the frontonasal mesenchyme distrib uted lateral to the nasal epithelium.

Conclusions
Our results indicate that atRAactivated RARG controls the expression of AHR at GD10.5 in the developing palate, which in turn appears necessary for TCDD to induce cleft palate. TCDD, however, does not alter the pattern of atRAsignaling in the devel oping face. These findings provide evidence about the molecular mechanism through which TCDD exposure at GD10.5 can result in a cleft palate and thereby clarify a possible mechanism of action for TCDD. The eti ology of cleft palate induced by TCDD at later developmental stages (e.g., GD12.5) may stem from another mechanism Birnbaum et al. 1989). Nonetheless, because both AHR and RAR are universal signaling systems conserved across vertebrate species, including humans (Campo Paysaa et al. 2008;Hahn 2002), it is possible that the mechanistic findings in the present study are of general relevance. Accordingly, Figure 5. AHR expression depends upon RALDH3 in the frontonasal region. (A and B) Relative Ahr mRNA levels in GD11.5 nasopalatal regions of WT, Aldh1a3 -/and atRA-rescued (2 mg/kg) Aldh1a3 -/embryos (A) and in WT or RAR-deficient MEFs treated with vehicle or atRA (B). Data are the mean + SD of triplicates from at least three samples (nasopalatal regions or cell cultures) in each experimental condition. (C and D) Immunohistochemistry showing RARG protein localization in the mesenchyme surrounding the nasal epithelium in WT embryo (C). An Rarg -/embryo was used as a negative control for immunostaining (D). Abbreviations: f, forebrain; lm, lateral mesenchyme of the frontonasal region; lng, lacrimonasal groove; mm, medial mesenchyme of the frontonasal region; mx, maxillary prominence of first branchial arch; ne, nasal epithelium. Bar = 200 μm. *p < 0.05 in ligand-treated versus vehicle-treated samples. Ahr expression is controlled by atRAactivated RAR both in mouse (present study) and in medaka fish (Hayashida et al. 2004). In this context, intact atRA signaling may be manda tory to enable the AHR message not only in the developing palate at GD10.5 but also in other organ systems.