Cell-autonomous retinoic acid receptor signaling has stage-specific effects on mouse enteric nervous system

Retinoic acid (RA) signaling is essential for enteric nervous system (ENS) development, since vitamin A deficiency or mutations in RA signaling profoundly reduce bowel colonization by ENS precursors. These RA effects could occur because of RA activity within the ENS lineage or via RA activity in other cell types. To define cell-autonomous roles for retinoid signaling within the ENS lineage at distinct developmental time points, we activated a potent floxed dominant-negative RA receptor α (RarαDN) in the ENS using diverse CRE recombinase–expressing mouse lines. This strategy enabled us to block RA signaling at premigratory, migratory, and postmigratory stages for ENS precursors. We found that cell-autonomous loss of RA receptor (RAR) signaling dramatically affected ENS development. CRE activation of RarαDN expression at premigratory or migratory stages caused severe intestinal aganglionosis, but at later stages, RarαDN induced a broad range of phenotypes including hypoganglionosis, submucosal plexus loss, and abnormal neural differentiation. RNA sequencing highlighted distinct RA-regulated gene sets at different developmental stages. These studies show complicated context-dependent RA-mediated regulation of ENS development.


Introduction
The enteric nervous system (ENS) is a complex network of neurons and glia that resides in the bowel wall and is essential for intestinal function (1,2). These ENS cells arise primarily from vagal enteric neural crestderived cell (ENCDC) precursors that divide rapidly and colonize the bowel in a rostral to caudal direction from E9 to E13.5 in mice (3)(4)(5)(6). In addition to vagal ENCDC, ENS precursors include sacral neural crest (7), mesenteric neural crest (8), sympatho-enteric precursors (9), Schwann cells (10), and perhaps bowel epithelial cells (11). As ENCDC colonize the bowel, they differentiate into about 20 neuron types, and many types of glia that form extensive networks to control most aspects of bowel function (2,(12)(13)(14)(15)(16)(17). Maturation of the ENS continues during fetal development, and remodeling continues after birth (18)(19)(20). Retinoic acid (RA), the active metabolite of vitamin A, is an important morphogen with an integral role in ENCDC migration, proliferation, and differentiation (21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32). RA functions mainly as a transcriptional regulator that binds to and activates RA receptor/retinoid x receptor (RAR/RXR) heterodimers (33). RAR/RXR heterodimers bind DNA at RA response elements (RAREs) to regulate transcriptional activity. Several studies demonstrate vital roles for vitamin A and its active metabolite RA in ENS development. In rat and mouse primary cell culture, RA increases ENCDC proliferation and neuronal differentiation while decreasing neurite length (23,27). In vivo, vitamin A-deficient mice (Rbp4 -/mice on a vitamin A-deficient diet) develop distal bowel aganglionosis (26), confirming that vitamin A is needed for bowel colonization by ENCDC. The extent of bowel aganglionosis dramatically increased when Ret heterozygosity was combined with vitamin A deficiency, suggesting potent gene-environment interactions in mice (26). Consistent with these observations, loss of the primary RA synthesis enzyme retinaldehyde dehydrogenase 2 (Raldh2) also causes severe intestinal aganglionosis in mice (29), with more minor effects of murine Raldh1 and Raldh3 mutations (25). Furthermore, maternal vitamin A deficiency or excess can cause intestinal hypoganglionosis without aganglionosis in rats and mice (22,28,32). While these results clearly show that RA signaling is needed for ENS development, many questions remain. First, because RAR and RXR receptors are expressed in diverse cell types (27), it is not clear whether RA acts directly on ENCDC or via effects on neighboring cells. Second, effects of RA signaling at distinct developmental stages remain elusive. Third, the RA-targeted gene network in ENCDC in vivo is not yet established. To address these questions, we employed a very potent Retinoic acid (RA) signaling is essential for enteric nervous system (ENS) development, since vitamin A deficiency or mutations in RA signaling profoundly reduce bowel colonization by ENS precursors. These RA effects could occur because of RA activity within the ENS lineage or via RA activity in other cell types. To define cell-autonomous roles for retinoid signaling within the ENS lineage at distinct developmental time points, we activated a potent floxed dominant-negative RA receptor α (RarαDN) in the ENS using diverse CRE recombinase-expressing mouse lines. This strategy enabled us to block RA signaling at premigratory, migratory, and postmigratory stages for ENS precursors. We found that cell-autonomous loss of RA receptor (RAR) signaling dramatically affected ENS development. CRE activation of RarαDN expression at premigratory or migratory stages caused severe intestinal aganglionosis, but at later stages, RarαDN induced a broad range of phenotypes including hypoganglionosis, submucosal plexus loss, and abnormal neural differentiation. RNA sequencing highlighted distinct RA-regulated gene sets at different developmental stages. These studies show complicated context-dependent RA-mediated regulation of ENS development.  (34) in combination with fluorescence-based lineage tracing in mice. We discovered that RA signaling regulates ENS development in a cell-autonomous manner, with distinct effects on different developmental stages. Furthermore, gene expression profiling showed stage-specific effects of blocking RAR signaling in developing ENCDC. These results suggest that vitamin A deficiency or excess could alter ENS structure and function in many ways during intrauterine and postnatal periods, contributing to human bowel motility disorders.

Results
Cell-autonomous RAR signaling in neural crest derivatives is required for craniofacial and ocular development. To characterize cell-autonomous roles for RAR signaling within the ENS lineage, we bred RarαDN LoxP/+ to Wnt1Cre + mice. RarαDN LoxP/+ produce RARαDN after CRE-mediated DNA recombination. Wnt1Cre express Cre recombinase in the CNS and many neural crest derivatives, including essentially all fetal ENS precursors (35)(36)(37). RarαDN LoxP/+ ; Wnt1Cre + mice are viable at E12.5, but they die by E14.5 (Table 1). E12.5 RarαDN LoxP/+ ; Wnt1Cre + have major malformations of neural crest-derived facial structures ( Figure 1, A and B) with absent facial cartilage ( Figure 1, C and D). Sectioning showed failure of nasomedial process fusion at the midline and a wide frontonasal region (Figure 1, E and F). In contrast, dorsal root ganglia (DRG), another crest-derived structure, appeared fairly normal even though Wnt1Cre induced recombination of an EYFP reporter in DRG (Figure 1, G-L). These data highlight distinct RA roles in different neural crest-derived tissues during development. Our primary goal was to investigate RA signaling effects in the ENS.
ENS development requires cell autonomous RAR signaling. At E12.5 in Wnt1Cre + (control) mice, ENCDC had colonized the esophagus, stomach, small intestine, and half of the colon, as seen by TuJ1 (neuron specific β3-tubulin) antibody staining ( Figure 2A). In contrast, E12.5 RarαDN LoxP/+ ; Wnt1Cre + mice only had TuJ1 + cells in the esophagus and stomach ( Figure 2B). Recognizing that the absence of TuJ1 staining might reflect impaired neuronal differentiation in RarαDN LoxP/+ ; Wnt1Cre + mice or could reflect absent ENS, we bred to R26R-TdTomato lineage reporter mice so that cells undergoing CRE-mediated DNA recombination are unambiguously marked. While TdTomato + ENS and HuC/D + (neuronal RNA binding protein) neurons were readily detected in stomach, small bowel, and proximal colon of control animals ( Figure 2, C-E and I), the RarαDN LoxP/+ ; Wnt1Cre + ; R26R-TdTomato + mice did not have TdTomato + or HuC/D + cells in the small bowel or colon and had fewer ENS cells in stomach than controls (Figure 2, F-H and J). These analyses confirmed that blocking cell-autonomous RAR signaling in the Wnt1Cre lineage completely prevented these ENS precursors from colonizing the small bowel and colon.
Blocking RAR signaling causes defects in ENCDC migration and differentiation by E10.5. To determine if RarαDN expression within neural crest-derived ENS precursors acts at earlier developmental stages, we examined E10.5 whole embryo via 3DISCO tissue clearing (38) and confocal microscopy ( Figure 3, A and B). TuJ1 and SOX10 antibody staining showed many ENCDC in the esophagus and stomach in control mice (Figure 3, C-F) but very few stained ENCDC in the proximal bowel of RarαDN LoxP/+ ; Wnt1Cre + mice ( Figure 3, G-J), consistent with a defect in early stages of bowel colonization. More specifically, while there were some SOX10 + ENCDC near the vagus in mutant mice (Figure 3, G-J), the control mice had many SOX10 + ENCDC that had migrated far beyond the vagus and into the stomach (Figure 3, C-F). Furthermore, in control mice, many ENCDC were TuJ1 immunoreactive, suggesting early neuronal differentiation (Figure 3, C and E), but there were almost no TuJ1 + cells in the esophagus or stomach of RarαDN LoxP/+ ; Wnt1Cre + mice ( Figure 3, G and I). Collectively, these data suggest that cell-autonomous RAR signaling is needed for ENCDC to populate the bowel and differentiate into neurons.
Cell autonomous RAR signaling is required for RET and PHOX2B expression in ENCDC. The defect in bowel colonization by ENCDC of RarαDN LoxP/+ ; Wnt1Cre + mice closely resembles the phenotype in Ret and Phox2b null mice  TdTomato mice, however, was not statistically different from controls ( Figure 5, A-F and G); therefore, it is not clear that reduced proliferation within stomach ENCDC fully accounts for the phenotype. Collectively, these data suggest that RAR signaling is needed in the Wnt1Cre ENS lineage for Ret and Phox2b expression, further suggesting that loss of either RET or PHOX2B could cause this type of extensive intestinal aganglionosis (39,40). Loss of RAR signaling causes defective vagal nerve development. In parallel with the loss of ENCDC in fetal stomach, E11.5 RarαDN LoxP/+ ; Wnt1Cre + mice had smaller vagus nerves than control littermates ( Figure 6, A-C). The reduction in vagal nerve fibers in the stomach appears to be non-cell autonomous because TuJ1 + vagal nerve fibers are TdTomato-negative in Wnt1Cre + ; TdTomato + and RarαDN + ; Wnt1Cre + ; TdTomato + mice ( Figure 6D, Supplemental Figure 1, and Supplemental Videos 1 and 2; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.145854DS1). Enlarged images also demonstrate many TdTomato + cells migrating along the TuJ1 + vagal nerve fibers in Wnt1Cre + ; TdTomato and RarαDN; Wnt1Cre + ; TdTomato mice (Supplemental Figure 1, E, F, K, and L). Furthermore, while ENCDC migrate far beyond the vagus in control animals, TuJ1 + -and SOX10-labeled ENCDC remained close to vagal fibers in E11.5 RarαDN LoxP/+ ; Wnt1Cre + mice ( Figure 3, D and H, and Figure 6, A and B). These observations suggest synergistic interactions between growing vagal nerve fibers and migrating ENCDC that colonize the bowel to form ENS.
Loss of RAR signaling in SOX10 lineage of ENCDC causes aganglionosis. To evaluate cell-autonomous effects of RAR beyond E12.5 when RarαDN LoxP/+ ; Wnt1Cre + die, we generated RarαDN LoxP/+ ; SOX10Cre + mice. These mice express Cre from Sox10 regulatory elements, beginning at E8.5 when ENCDC migrate through the paraxial mesoderm to foregut (41). To confirm CRE activation in the ENS lineage, we examined SOX10Cre; R26R-TdTomato  Figure 2). Unlike controls ( Figure 7A) and like RarαDN LoxP/+ ; Wnt1Cre + , the E12.5 RarαDN LoxP/+ ; SOX10Cre + had TuJ1 + ENCDC predominantly confined to the stomach at E12.5 (Figure 7, B-E), with a few sparsely distributed TuJ1 + ENCDC in the proximal small intestine ( Figure 7D). By E15.5, RarαDN LoxP/+ ; SOX10Cre + had obvious eye and craniofacial defects but remained viable ( Figure 7F). The E15.5 ENS network was dense in the esophagus and stomach but was sparse in the proximal small intestine ( . These data are consistent with the hypothesis that RAR signaling is required as ENCDC migrate from the neural tube to bowel and that it is not needed by ENS precursors before E8.5 in the neural tube. Blocking RAR signaling in the TyrCre lineage reduces enteric neurons and alters ENS patterning. We next examined the ENS in mice that express Cre in ENCDC from the tyrosinase promoter (RarαDN LoxP/+ ; TyrCre + ) starting at E10.5 (41, 42) when ENCDC normally reach the midgut (3). In these mice, the extent of bowel colonization by To determine if blocking RAR impacted neuron subtype ratios, we counted NOS1 and HuC/D double-labeled cells and discovered that more HuC/D + neurons were NOS1 + in RarαDN LoxP/+ ; TyrCre + ENS (Figure 8, Q-S) compared with controls. Collectively, these results show that cell-autonomous RAR signaling is important after E10.5 for ENS patterning, to increase neuron number, and for neuron subtype specification.
Inactivation of RAR signaling in the RET lineage causes distal bowel hypoganglionosis. Sparse ENS in the preceding models might reflect inadequate RET, making it difficult to identify other roles for RAR signaling in ENC-DC. Our prior studies suggest that RA is not needed to sustain Ret expression in ENCDC that already express Ret (26,27). We therefore decided to use mice expressing CRE-ERT2 from the Ret locus to activate RarαDN expression in ENCDC that already express Ret. For these experiments, we also needed a CRE-dependent fluorescent reporter (EYFP) to track CRE activity. Because Ret is on mouse chromosome 6 near the ROSA26 locus that drives expression of RarαDN, as well as most fluorescent reporters, we bred RarαDN LoxP/+ mice to RET-CreERT2-EYFP Tandem mice to generate RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem and RarαDN LoxP/+ controls. The  (Figure 9, A, C, E, and G) but had hypoganglionic ENS in the colon, with thick chains of ENS cells (Figure 9,C and I). We confirmed high levels of CRE-activation in the ENS after tamoxifen treatment and almost no CRE activation in the absence of tamoxifen in these RETCreERT2-EYFP Tandem mice (Supplemental Figure 3). Because tamoxifen effects may take 12-18 hours, we also tamoxifen treated at E8.5 but found similar phenotypes (Figure 9, J-M). RET and PHOX2B were readily detected in EYFP + ENCDC of E13.5 RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem mice, consistent with our prior studies (26,27). These data suggest that RAR is needed to activate RET and PHOX2B expression in ENCDC (Figure 4, D-I, and Figure  5, A-F and H) but not to maintain expression once regulatory elements are activated (Figure 9, N-Q). The unusual patterning in the distal colon of tamoxifen-treated E13.5 RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem mice suggests that RAR regulates additional ENS patterning genes in ENCDC.
Ret lineage loss of RAR signaling profoundly reduced submucosal neuron density and altered cell identity. RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem treated with tamoxifen at E10.5 ( Figure 10) were born at a normal Mendelian ratio, grew normally, and had a normal-appearing bowel ( Figure 10B), permitting analysis of adult ENS. We therefore stained the bowel of 2-month-old mice with HuC/D and TuJ1 antibodies ( Figure  10, C-N). Quantitative analysis demonstrated a 50%-65% reduction in myenteric neurons in small bowel and colon and a 90% loss of submucosal neurons in RarαDN-expressing mutant mice (Figure 10, O-T). In contrast to neurons, adult enteric glia marked by SOX10 antibody appeared similar in abundance in tamoxifen-treated RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem and control mice (Supplemental Figure 4A). Interestingly, E10.5 tamoxifen treatment also led to a dramatic increase in NOS1 + submucosal neurons in RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem small bowel (Supplemental Figure 4B), and a mild increase in NOS1 + myenteric neurons. These studies confirm distinct age-dependent effects of cell-autonomous RAR signaling in the ENS.
One initially surprising feature at E13.5 was that many hemoglobin genes were less abundant in flow-sorted cells from RarαDN; RETCreERT2-EYFP Tandem compared with controls. We suspect that this occurred because Ret is expressed in the hematopoietic stem cell (HSC) lineage, where RAR supports erythropoiesis (43,44) and hemoglobin genes are normally expressed at high levels. Consistent with this hypothesis, some RETCreERT2-EY-FP Tandem -lineage cells in the colon were stained with TER119 (erythroid lineage) and CD31 (endothelial, platelet, and leucocyte lineage) antibodies (Supplemental Figure 6). To define a gene set clearly linked to the ENS, we  Table 3). Of these, a core set of 115 genes were regulated in the same direction by RARαDN in ENCDC at E11.5 and E13.5 (Supplemental Tables 4 and 5). Gene enrichment pathway analysis of the aforementioned 2 sets of genes showed many pathways related to neuron development (Supplemental Figure 7). To validate RNA-seq data, we selected 2 differentially expressed genes of ENCDC at E11.5 (Stmn2 and Pax3) and performed quantitative PCR (qPCR), which confirmed differences predicted by RNA-seq (Supplemental Figure 8). The RNA-seq data for Ret and Phox2B in E11.5 stomach also correlate well with our IHC for RET and PHOX2B protein (Figure 4, Figure 5, and Supplemental Figure 9).

Discussion
RA regulates activity of the RAR/RXR transcription factor family to alter gene expression and influence many aspects of development (33). Prior studies using constitutive KO mice, RAR antagonist, stem cells, or vitamin A depletion in mice and rats show critical roles for RA in the developing ENS (21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32). For example, loss of the RA biosynthetic enzyme RALDH2 causes extensive bowel aganglionosis in Raldh2 -/mice  (25). The Raldh2 -/phenotype probably occurs because RA is needed in vagal paraxial mesoderm starting at E9 in mice (or E1.5 in avians) to induce Ret expression in ENCDC as these cells migrate from neural tube to bowel (24). In contrast, when RAR signaling was blocked by the chemical pan-RAR antagonist BMS493 at E11.5 in organ culture (26) or at E12.5 in dissociated cell culture (27), RET expression in ENCDC appeared unaffected by RAR inhibition. Nonetheless, RAR inhibition at these later stages impaired ENCDC proliferation, bowel colonization, neurite growth, neuronal differentiation, and ENS patterning (23,26,27). Consistent with these observations, nutritional deficiency in the RA precursor vitamin A causes distal bowel aganglionosis in mice, mimicking human Hirschsprung disease (HSCR) (26), and gene variants in people with HSCR may alter RA signaling (45,46). These studies suggest that RA signaling could impact ENS development in many ways, depending on the timing and severity of the RA signaling defect.
Prior strategies did not define the cell types influenced by RA signaling. To determine how cellautonomous RAR signaling affects the ENS lineage, we used a potent CRE-inducible dominant-negative RAR transgene (RARaT403) that blocks all 3 RAR receptors (i.e., RarαDN LoxP/+ mice) (47). By breeding to several different CRE-expressing mouse lines, we induced RARaT403 (RARαDN) expression selectively within ENCDC at specific times during development. Our RarαDN LoxP/+ ; Wnt1Cre + studies show that cellautonomous RAR signaling within ENCDC is required to activate Ret expression early in development, consistent with data from zebrafish (21) and avian models (24). RAR signaling is also needed to activate Ret in the ureteric bud, suggesting that similar cell-autonomous mechanisms may control Ret expression in the kidney and developing ENS (34,48). In contrast, when CRE expression is driven by Ret regulatory elements, RarαDN  induction did not lead to loss of RET protein in ENCDC, suggesting that RA is needed to turn on Ret but not to maintain Ret expression, consistent with our prior BMS493 data (26,27). We were surprised to discover that PHOX2B protein was also undetectable in ENCDC of RarαDN LoxP/+ ; Wnt1Cre + mice. Loss of PHOX2B could, by itself, explain the loss of RET in RarαDN LoxP/+ ; Wnt1Cre + ENCDC (40), but we are unable to find evidence that PHOX2B regulates RET in the kidney, suggesting that Ret is regulated by RA via additional PHOX2B-independent mechanisms, as supported by prior studies (30,45,49).
There were many other interesting observations. First, the thick chains of enteric neurons seen in the distal bowel of RarαDN LoxP/+ ; TyrCre + and RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem mice closely resemble the colon ENS patterning defect we observed in BMS493-treated fetal gut organ cultures (26). This suggests that the normal dispersion of enteric neurons into small colon ganglia is RA dependent. Second, in fetal RarαDN LoxP/+ ; SOX10Cre + mice, we saw some clustered small bowel enteric neurons near extrinsic nerve fibers but far from the more proximal ENS cells. These clusters closely resemble Schwann cell-derived ENS described by Uesaka et al., but the ENCDC we observed were restricted to small regions of fetal bowel (10). Alternatively, these cells might originate from the "mesenteric neural crest cells" recently described by Yu et al. that they hypothesize contribute to human skip segment HSCR (8). Third, vagus nerves at E11.5 in RarαDN LoxP/+ ; Wnt1Cre + mice occupied a smaller area of the stomach compared with control animals. Since vagus nerve fibers were not TdTomato labeled in Wnt1Cre; R26R-TdTomato mice, this suggests reciprocal interactions between migrating Wnt1Cre lineage ENC-DC and growing vagal fibers. Fourth, in addition to approximately 80% reduction in total neuron number in the ENS of RarαDN LoxP/+ ; TyrCre + mice, there was a striking increase in the percentage of enteric neurons that express NOS1 (nitric oxide synthase). We confirmed that these NOS1 + neurons had expressed CRE using an R26R-TdTomato reporter. This suggests that NOS1 + neuron differentiation is less dependent on RAR signaling than other neuron subtypes or that RAR turns off Nos1 expression in some enteric neuron subtypes. Finally, we found an almost complete loss of submucosal neurons in RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem mice treated with tamoxifen at E10.5, suggesting a critical role for RA signaling in radial migration of ENCDC to form submucosal plexus. The normal postnatal growth of these tamoxifen-treated RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem mice and normal appearance of the adult bowel suggests that a loss of approximately 90% of submucosal neurons is well tolerated, at least in mice. This is interesting, in part, because we know little about mechanisms controlling radial migration of ENS precursors to the submucosal plexus, with prior studies implicating only GDNF and netrin/DCC signaling in this process (50,51). These observations highlight the remarkable range of ENS abnormalities that may occur when RAR signaling is inadequate and are consistent with the observation that low serum vitamin A is associated with increased constipation in children with autism (52).
One concern is that we did not evaluate ENS biology in every possible control group, so we cannot exclude some effects of tamoxifen, Cre alleles, fluorescent reporters, or the RarαDN LoxP/+ allele in isolation. For example, a recent study clearly shows that Wnt1Cre + ; R26R-TdTomato + increases the severity of the Ednrb -/-ENS phenotype (8). We also were underpowered to examine the effect of sex on ENS phenotypes, an issue that could be important for some of the milder phenotypes we examined.
One advantage of our strategy is that CRE-dependent expression of RARαDN and a fluorescent reporter in the same cells facilitated flow sorting and RNA-seq to identify RAR targets. Our analyses identified thousands of differentially expressed genes in RARαDN-expressing versus control ENCDC at E11.5 and E13.5 with 115 genes regulated in the same way by RARαDN at each age. Interpreting these data is complicated, since RARαDN induces changes in cell type ratios, in addition to changes in gene expression in individual cells. Furthermore, it is not clear if the induced changes occur because of direct effects on RAREs or if they reflect more global effects on the differentiation state of the sequenced cells. For example, loss of PHOX2B in E11.5 ENCDC should change the expression of many genes, independent of effects on RAR activity.
As one strategy to further define the role of RAR signaling, we narrowed our lists to include only differentially expressed genes (FDR < 0.2) where expression changed in the same direction at E11.5 and E13.5 for RARαDN + cells (Supplemental Tables 4 and 5). At E11.5, there were 55 genes expressed at higher levels in WT ENCDC than in RARαDN + ENCDC. Thirty-five of these genes had easily identified functions in neurons. Thirteen genes at E11.5 were more abundant in RARαDN + ENCDC than in WT ENCDC. This list included Rest, an epigenetic master negative regulator of neurogenesis (53,54), and Bmp4, a gene with complex roles in ENS patterning (55)(56)(57)(58)(59)(60)(61)(62). At E13.5, our list included 44 genes more abundant in WT than in RARαDN + cells. Twenty-nine of these genes had easily identified roles in neurons. Genes more abundant in RARαDN + than in WT cells at E13.5 included Rest, Sox10, and Ets1, consistent with a role for RAR in ENS neurogenesis. The SOX10 transcription factor is essential for ENS development (63), at least in part because SOX10 activates RET expression (64,65). However, as multipotent ENCDC differentiate into enteric neurons, SOX10 expression is lost, while enteric glia continue to express SOX10 through adulthood (3). ETS1 enhances Sox10 expression and is essential to make radial glia in Xenopus (66). Consistent with these mRNA findings, SOX10 + cells were abundant in myenteric and submucosal plexus of adult RarαDN-expressing mice, even though RarαDN LoxP/+ ; RETCreERT2-EYFP Tandem had an approximately 90% reduction in submucosal neurons and the enteric glia are derived from RET + ENCDC (67). In addition, many Wnt ligands (Wnt1, Wnt2, Wnt4, and Wnt10a) were highly differentially expressed in E11.5 ENCDC of RarαDN LoxP/+ ; Wnt1Cre + versus control. Since Wnt signaling plays important roles in many aspects of neural development (68)(69)(70), RAR might alter Wnt signaling to regulate ENCDC development. Collectively, these data suggest that cell-autonomous RAR signaling directs neurogenesis in the ENS from multipotent ENCDC and that cell-autonomous RAR signaling has distinct effects at many stages of ENS development. Our observations further suggest that maternal retinoid status during pregnancy and postnatal vitamin A deficiency or excess could have long-term effects on ENS structure and function. This may be important because vitamin A deficiency is a common problem in many regions of the world (71).  (72). Mouse Ret and Rosa26 loci are close together (about 5 million bp), but RETCreERT2-EYFP Tandem recombined so that they are on the same chromosome. Ret TGM/TGM (Ret -/allele Ret tm1Jmi ) on C57BL/6J were previously described (73). A complete list