Rescue of Holoprosencephaly in Fetal Alcohol-Exposed Cdon Mutant Mice by Reduced Gene Dosage of Ptch1

Holoprosencephaly (HPE) is a commonly occurring developmental defect in which midline patterning of the forebrain and midface is disrupted. Sonic hedgehog (SHH) signaling is required during multiple stages of rostroventral midline development, and heterozygous mutations in SHH pathway components are associated with HPE. However, clinical presentation of HPE is highly variable, and carriers of heterozygous mutations often lack apparent defects. It is therefore thought that such mutations must interact with more common modifiers, genetic and/or environmental. We have modeled this scenario in mice. Cdon mutant mice have a largely subthreshold defect in SHH signaling, rendering them sensitive to a wide spectrum of HPE phenotypes by additional hits that are themselves insufficient to produce HPE, including transient in utero exposure to ethanol. These variable HPE phenotypes may arise in embryos that fail to reach a threshold level of SHH signaling at a specific developmental stage. To provide evidence for this possibility, here we tested the effect of removing one copy of the negative regulator Ptch1 from Cdon−/− embryos and compared their response to ethanol with that of Cdon−/−;Ptch1+/+ embryos. Ptch1 heterozygosity decreased the penetrance of HPE in this system by >75%. The major effect of reduced Ptch1 gene dosage was on penetrance, as those Cdon−/−;Ptch1+/− embryos that displayed HPE did not show major differences in phenotype from Cdon−/−;Ptch1+/+ embryos with ethanol-induced HPE. Our findings are consistent with the notion that even in an etiologically complex model of HPE, the level of SHH pathway activity is rate-limiting. Furthermore, the clinical outcome of an individual carrying a SHH pathway mutation will likely reflect the sum effect of both deleterious and protective modifier alleles and their interaction with non-genetic risk factors like fetal alcohol exposure.


Introduction
Holoprosencephaly (HPE), a common developmental defect of the forebrain and midface, results from disruption in patterning of the rostroventral midline [1]. HPE occurs with high frequency, ,1:250 human conceptions, the great majority of which are lost during pregnancy [2,3]. Both genetic and environmental factors are implicated in the etiology of HPE [4][5][6]. Heterozygous mutations in components of the Sonic hedgehog (SHH) signaling pathway are found in both inherited and sporadic forms of HPE [5]. These include the SHH ligand itself and the receptors, PTCH1, CDON, and GAS1 [5,7,8]. Among the non-genetic factors suggested to elevate the risk of HPE are preexisting maternal diabetes and fetal alcohol exposure [9,10].
Experiments with mice and other vertebrate model organisms indicate that development of the rostroventral midline occurs by a progressive patterning mechanism whereby SHH produced by one midline structure induces Shh expression in a successive midline structure. SHH produced by the prechordal mesendoderm (PCM) is a key early step in development of the midline of the forebrain and midface [1,[11][12][13]. SHH produced by the PCM induces expression of pathway target genes (including Shh itself) in the rostral diencephalon ventral midline of the developing forebrain, which in turn leads to Shh expression in the ventral telencephalon [13][14][15][16][17][18]. As development proceeds, SHH produced by the forebrain induces expression of Shh in the ectoderm of the frontonasal and maxillary processes of the developing face, helping to pattern the craniofacial midline [17,[19][20][21].
Clinical presentation of HPE is extremely variable, and the range of defects falls within a continuum known as the HPE spectrum [4,22]. The most severe form, alobar HPE, is characterized by complete failure to partition the forebrain into left and right hemispheres; this is usually associated with the most severe midfacial phenotypes, including cyclopia. The spectrum encompasses progressively less severe forebrain defects (semilobar and lobar HPE) and also mild facial midline abnormalities (socalled HPE microforms) that can occur in the absence of brain malformations.
A full spectrum of HPE phenotypes is seen both in sporadic and familial cases [23]. Furthermore, many mutation carriers in pedigrees are without apparent clinical manifestation, and mutations in many sporadic HPE patients are inherited from an apparently unaffected parent [23,24]. Statistical analysis of these observations and gene mutation frequencies have led to an ''autosomal dominant with modifier'' model, in which the phenotypic outcome associated with a heterozygous mutation is influenced by more common genetic variants and/or environmental exposures [25]. It is important, therefore, to identify interactions between bona fide mutations and additional factors that grade penetrance and expressivity, resulting in the wide range of defects encompassed by the HPE spectrum. Animal models are well placed to do so. We have established Cdon mutant mice as such a model. Mice lacking CDON on a 129S6 genetic background have a largely subthreshold defect in SHH signaling that renders them sensitive to induction of HPE by second hits that are, themselves, insufficient to produce HPE (e.g., removal of one copy of Shh; gene dosage-sensitive removal of the Cdon paralog, Boc) [26][27][28].
We recently reported use of these mice to develop a model of gene-environment interaction in HPE. In utero ethanol exposure at E7.0 had little or no effect on wild type 129S6 embryos, but produced a wide range of HPE phenotypes with .75% penetrance in Cdon 2/2 embryos [29]. The spectrum of HPE phenotypes produced in ethanol-treated Cdon 2/2 mice ranged from alobar HPE with an undivided eye field, to lobar HPE with single nostril, to microform HPE and defects in palatogenesis [29]. A model consistent with production of phenotypes at both the severe and mild ends of the spectrum in ethanol-treated 129S6.Cdon 2/2 mice is as follows: ethanol initiates defects in midline patterning specifically in genetically sensitized embryos; the variable severity of the HPE phenotype may arise as a consequence of individual embryos failing to reach a threshold level of SHH signaling at specific developmental stages, the stage of the deficit occurring stochastically and dictating phenotypic severity [29]. This hypothesis argues that pathway signaling strength is an important determinant of phenotypic outcome. PTCH1, the primary SHH receptor, functions as a negative regulator of pathway signaling [30]. To test the hypothesis, we removed one copy of Ptch1 from 129S6.Cdon 2/2 embryos and assessed ethanol-induced production of HPE. As predicted by the model, Ptch1 heterozygosity effectively rescued such embryos.

Results
Rescue of HPE in fetal alcohol-exposed Cdon mutant mice by reduced Ptch1 gene dosage Loss of Cdon or in utero ethanol exposure in 129S6 mice gave little or no phenotype individually, but together produced defects in midline patterning, inhibition of SHH signaling in the developing forebrain, and a broad spectrum of HPE phenotypes [29]. When analyzed at E10.0, ,13.5% of ethanol-treated 129S6.Cdon 2/2 embryos displayed severe phenotypes that included alobar HPE and failure to divide the eye field. These embryos died in utero and were resorbed by E11.0. When analyzed at E14.0, ,70% of 129S6.Cdon 2/2 embryos displayed phenotypes ranging from lobar through microform HPE, and variously included single nostril, defective palatogenesis, diminished nasal septal cartilage, and rudimentary vomeronasal organs. In asking whether removal of one copy of Ptch1 could rescue HPE in these animals, we analyzed E14.0 embryos to take advantage of the high penetrance of the phenotypes observed at this stage.
Taken together, there may be a small diminution in HPE phenotypic expressivity in ethanol-treated Cdon 2/2 ;Ptch1 +/2 embryos, relative to similarly treated Cdon 2/2 ;Ptch1 +/+ embryos. However, in most cases, embryos of either genotype that displayed external HPE features showed roughly similar defects in forebrain and midfacial regions. These results suggest that the major effect of removing one copy of Ptch1 is to decrease the penetrance of HPE in ethanol-treated Cdon 2/2 mice, rather than to selectively alter the severity of some HPE phenotypes and not others.
Effect of Ptch1 gene dosage on Shh and Nkx2.1 expression in the ventral forebrain of ethanol-treated Cdon 2/2 mice We next tested the expression of Shh and a direct SHH pathway target gene, Nkx2.1, in the forebrain by whole-mount in situ hybridization at E10.25. We previously found that about half of the ethanol-treated Cdon 2/2 embryos at this stage had significantly reduced expression of these genes in the developing ventral telencephalon [29]. In this set, we observed that three of eight ethanol-treated Cdon 2/2 ;Ptch1 +/+ embryos had an obvious diminution in Shh expression and five of nine embryos of this genotype had an obvious reduction in Nkx2.1 expression (Figure 3). In contrast, one of eight ethanol-treated Cdon 2/2 ;Ptch1 +/2 embryos showed a noticeable change in the Shh expression pattern and no Cdon 2/2 ;Ptch1 +/2 embryos had alterations in the expression pattern of Nkx2.1, as compared to controls (n = 7) (Figures 3 and  S3).

Discussion
HPE is etiologically complex, involving interactions between the timing and strength of developmental signaling pathways, genetic variation, and exposure to environmental agents [4,22]. The preponderance of the evidence is consistent with an ''autosomal dominant plus modifier'' model, wherein inherited or de novo heterozygous mutations in one of at least 12 genes interact with more common modifier loci and/or environmental exposures to generate a wide spectrum of rostroventral midline patterning defects [25]. The mutations are found in genes encoding components of the SHH pathway or in genes that, directly or indirectly, regulate SHH expression or pathway activity in the early midline or forebrain [4,5,16,30,32,33]. In contrast, only ,15% of EtOH-treated Cdon 2/2 ;Ptch1 +/2 embryos had external features of HPE. Two ethanol-treated Cdon 2/2 ;Ptch1 +/2 embryos are shown, one that did not have external HPE features (D) and one that did (E; note the single nostril (arrowhead) and smooth, pointed philtrum (arrow)). (F-J) H&E-stained coronal sections of E14.0 embryos. Note that the ethanol-treated Cdon 2/2 ;Ptch1 +/+ embryo (H) displays continuity across the ventral midline of the forebrain (arrow), indicative of lobar HPE. Two ethanol-treated Cdon 2/2 ;Ptch1 +/2 embryos are shown, one that did not have external HPE features (I) and one that did (J). Note that only the embryo that had external HPE (J) also had continuity across the ventral midline of the forebrain (arrow). See Figure S1 for saline-treated control embryos. See Table 1  Mouse models of HPE provide support for this model. We have shown that mice lacking the SHH co-receptor CDON display HPE with strain-dependent penetrance and severity, consistent with an important role for silent modifier genes in grading the spectrum of phenotypes [27,28]. CDON is partially redundant with other co-receptors [27,34,35], and 129S6.Cdon 2/2 animals Note that the ethanol-treated Cdon 2/2 ;Ptch1 +/+ embryo (C) displays defective development of the palatal shelves (asterisks). Two ethanol-treated Cdon 2/2 ;Ptch1 +/2 embryos are shown, one that did not have external HPE features (D) and one that did (E). Note that only the embryo that had external HPE (E) also had defective palatogenesis (asterisks). (F-J) The nasal septal cartilage and vomeronasal organ are denoted in (F) by an arrowhead and arrow, respectively. Note that the ethanol-treated Cdon 2/2 ;Ptch1 +/+ embryo (H) displays diminished nasal septal cartilage and vomeronasal organ (arrow). Two ethanol-treated Cdon 2/2 ;Ptch1 +/2 embryos are shown, one that did not have external HPE features (I) and one that did (J). Note that only the embryo that had external HPE (J) also had reduced nasal septal cartilage and vomeronasal organ (arrow). (K-O) Note that the ethanol-treated Cdon 2/2 ;Ptch1 +/+ embryo (M) displays a narrow midfacial region with additional presumptive mesenchyme (asterisk). Two ethanol-treated Cdon 2/2 ;Ptch1 +/2 embryos are shown, one that did not have external HPE features (N) and one that did (O). Note that only the embryo that had external HPE (O) also had a narrow midfacial region with additional presumptive mesenchyme (asterisk). See Figure S2 for saline-treated control embryos. See Table 1   have only a substhreshold defect in SHH signaling that renders them sensitive to induction of HPE by additional insults, genetic or environmental, that are themselves insufficient to produce HPE [26][27][28][29]34,35].
Epidemiological evidence suggests that fetal alcohol exposure raises the likelihood of HPE [9]. We recently showed that in utero exposure to ethanol produces a nearly complete spectrum of HPE phenotypes in 129S6.Cdon 2/2 embryos, but is not teratogenic to 129S6.Cdon +/+ embryos [29]. We hypothesized that a brief exposure to ethanol during gastrulation in a genetically sensitized embryo sets up a situation whereby the variable severity of HPE arises as individual embryos fail to reach a threshold level of SHH signaling at a specific developmental stage, with the stage of the deficit occurring stochastically and determining the severity of the outcome. This is consistent with observations in model systems demonstrating that SHH pathway activity is required during multiple stages of rostroventral midline patterning, including forebrain partitioning, bilateral division of the eye field and nonneural facial structures, and development of the primary and secondary palates [19,36,37]. This hypothesis argues that the level of SHH pathway signaling activity is an important determinant of phenotype. To provide experimental evidence for this possibility we tested the effect of removing one copy of the negative regulator Ptch1 from Cdon 2/2 embryos and compared their response to ethanol-induced HPE with that of Cdon 2/2 ;Ptch1 +/+ embryos. As predicted, reduced gene dosage of Ptch1 substantially reduced the penetrance of HPE in this system, from ,65% in Cdon 2/2 ;Ptch1 +/+ embryos to ,15% in Cdon 2/2 ;Ptch1 +/2 embryos, a reduction of .75%. The major effect of Ptch1 heterozygosity in this system was on penetrance, as those Cdon 2/2 ;Ptch1 +/2 embryos that displayed HPE did not show major differences in phenotypic spectrum or selectivity from Cdon 2/2 ;Ptch1 +/+ embryos with ethanol-induced HPE.
At what developmental stage(s) might the rescue of HPE by Ptch1 haploinsufficiency occur? The PCM is required for induction of the ventral forebrain and subsequent partitioning of the forebrain and eye field, as well as patterning of the midfacial midline [1,[11][12][13]. The PCM serves as a source of SHH early during this multistep process, but SHH is also required for normal development of the PCM itself [1,13,14]. Cdon is expressed in the PCM but not in the ventral forebrain, even though the latter structure is affected in Cdon 2/2 mice; this observation and additional results argue that loss of Cdon function in the PCM is the most likely cause of HPE in such mice [26][27][28]38]. CDON and PTCH1 bind to one another as components of the SHH receptor complex [7,35] and, like Cdon, Ptch1 is expressed in the PCM [14]. Therefore, it seems likely that rescue of HPE phenotypes in ethanol-treated Cdon 2/2 ;Ptch1 +/2 embryos may occur as a consequence of enhanced SHH signaling function in the PCM itself. Alternatively, enhanced responsiveness to PCM-derived SHH may occur in the Ptch1 +/2 ventral forebrain, exclusive of coexpression of Cdon and Ptch1. Finally, Cdon and Ptch1 are also coexpressed in the developing facial mesenchyme, and restoration of SHH signaling there may contribute to rescue of facial midline phenotypes. These possibilities are not mutually exclusive, and conditional mutagenesis will be required to address this issue. The mechanism whereby ethanol synergizes with Cdon mutation to inhibit SHH signaling is also not fully resolved but must involve early patterning events between E7.0 -E8.0, as ethanol is no longer detectable by E8.0 [29].
Studies on individuals with HPE and animal models conclusively demonstrate that HPE is associated with loss of SHH pathway function. The fact that haploinsufficiency for Ptch1 rescues HPE phenotypes in ethanol-treated Cdon 2/2 embryos is consistent with PTCH1 functioning as a negative regulator of SHH signaling. However, heterozygous, germline missense mutations in PTCH1 have been identified in HPE patients [39,40]. PTCH1 functions in the absence of Hedgehog ligand to inhibit the activity of a second membrane protein, Smoothened (SMO). Binding of SHH to PTCH1 relieves this inhibition, and SMO signals to activate pathway target genes [41]. Because HPE is caused by loss, not gain, of SHH pathway function, it is highly unlikely that the PTCH1 mutations found in HPE are complete loss-of-function mutations. In fact, heterozygous, germline loss-offunction mutations in PTCH1 result in a different disorder, Basal Cell Nevus Syndrome (also called Gorlin syndrome) [42]. Rather, such PTCH1 variants would be predicted to maintain the ability to inhibit SMO activity but to be insensitive to ligand-induced pathway activation. A synthetically constructed deletion mutant with such properties is the widely used PTCH1 Dloop2 variant [43].
Our findings are consistent with the notion that even in an etiologically complex animal model of HPE (one that models the complexities of human HPE), the level of SHH pathway activity is rate-limiting. These findings are highly relevant in the emerging era of human genomics and soon-to-be widespread personal access to genome sequences. Systematic analyses of common variants in SHH pathway components may reveal not only alleles that predispose offspring of heterozygous HPE mutation carriers to severe phenotypes but also alleles with protective effects. With a core database of variants -which will need to be functionally validated -it may be possible to predict the sum effect of deleterious and beneficial alleles on penetrance and expressivity of bona fide mutations and, conceivably, their interaction with nongenetic risk factors like fetal alcohol exposure.

Ethics Statement
All animal work was approved by the Icahn School of Medicine at Mount Sinai Institutional Animal Care and Use Committee (IACUC). Our animal facility is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC).

Mice
STOCK.Ptch1 tm1Mps/J mice [44] were purchased from the Jackson Laboratory and transferred onto the 129S6/SvEvTac (129S6) background with the Speed Congenics Program at Taconic, employing back-crossing and SNP arrays. Mice used for these experiments were estimated to be .98% 129S6. 129S6.Ptch1 +/2 mice were crossed with 129S6.Cdon +/2 animals to generate 129S6.Cdon +/2 ;Ptch1 +/2 double mutant mice. These animals were then used to generate Cdon 2/2 ;Ptch1 +/2 , Cdon +/ 2 ;Ptch1 +/2 , and Cdon +/2 ;Ptch1 +/+ mice, which were then variously crossed for ethanol treatment as previously described [29]. Briefly, two to three month-old mice were mated for one hour in the dark. The time of the plug was designated as embryonic day (E) 0.0. Pregnant female mice were injected intraperitoneally twice with 15 ml per g body weight of a solution of 30% ethanol in saline (3.48 g/kg), first at E7.0 and again 4 hr later. Saline injections were used as a control.