Ciliary Hedgehog signaling regulates cell survival to build the facial midline

Craniofacial defects are among the most common phenotypes caused by ciliopathies, yet the developmental and molecular etiology of these defects is poorly understood. We investigated multiple mouse models of human ciliopathies (including Tctn2, Cc2d2a, and Tmem231 mutants) and discovered that each displays hypotelorism, a narrowing of the midface. As early in development as the end of gastrulation, Tctn2 mutants displayed reduced activation of the Hedgehog (HH) pathway in the prechordal plate, the head organizer. This prechordal plate defect preceded a reduction of HH pathway activation and Shh expression in the adjacent neurectoderm. Concomitant with the reduction of HH pathway activity, Tctn2 mutants exhibited increased cell death in the neurectoderm and facial ectoderm, culminating in a collapse of the facial midline. Enhancing HH signaling by decreasing the gene dosage of a negative regulator of the pathway, Ptch1, decreased cell death and rescued the midface defect in both Tctn2 and Cc2d2a mutants. These results reveal that ciliary HH signaling mediates communication between the prechordal plate and the neurectoderm to provide cellular survival cues essential for development of the facial midline.


Introduction
Primary cilia are microtubule-based organelles present on diverse vertebrate cell types and critical for development. Primary cilia function as specialized cellular signaling organelles that coordinate multiple signaling pathways, including the Hedgehog (HH) pathway (Zaghloul and Brugmann, 2011). Defects in the structure or signaling functions of cilia cause a group of human syndromes, collectively referred to as ciliopathies, which can manifest in diverse phenotypes including cystic kidneys, retinal degeneration, cognitive impairment, respiratory defects, left-right patterning defects, polydactyly, and skeletal defects (Baker and Beales, 2009;Hildebrandt et al., 2011;Tobin and Beales, 2009). In addition to these phenotypes, craniofacial defects including cleft lip/palate, high-arched palate, jaw disorders, midface dysplasia, craniosynostosis, tongue abnormalities, abnormal dentition, and tooth number and exencephaly are observed in approximately one-third of individuals with ciliopathies (Brugmann et al., 2010b;Zaghloul and Brugmann, 2011). The molecular and developmental etiology of these craniofacial abnormalities remains poorly understood.
We investigated the molecular underpinnings of forebrain and midface defects in ciliopathies utilizing multiple mouse mutants affecting the transition zone. The mutants exhibited forebrain and midface defects by E9.5, which persisted throughout development. In these mutants, the prechordal plate, an organizer of anterior head development, displayed defects in HH pathway activation at E8.0. These early prechordal plate defects attenuated Shh expression in the adjacent ventral forebrain. Decreased HH signaling increased apoptosis in the ventral neurectoderm and facial ectoderm. Surprisingly, reducing Ptch1 gene dosage rescued the apoptosis and its corresponding midface defect. Thus, investigating the function of the transition zone has revealed a key role of prechordal plate-activated HH signaling in forebrain and midface cell survival. Moreover, our genetic results reveal that inhibition of PTCH1 can prevent ciliopathy-associated midface defects. Based on these mouse genetic models, we propose that the etiology of hypotelorism in human ciliopathies is a failure of the prechordal plate to induce SHH expression in the overlying ventral neuroectoderm.

Results
The ciliary MKS transition zone complex is essential for midline facial development Individuals affected by developmental ciliopathies, such as Meckel, orofaciodigital, and Joubert syndromes, often display craniofacial phenotypes including holoprosencephaly and hypotelorism (Baker and Beales, 2009;Dowdle et al., 2011;Garcia-Gonzalo et al., 2011). To explore the etiology of these craniofacial defects, we examined the craniofacial development in Tctn2 mouse mutants Shaheen et al., 2011). TCTN2 is a component of the MKS transition zone complex critical for ciliary localization of several ciliary membrane proteins, including SMO, a key ciliary mediator of HH signaling (Chih et al., 2011;Corbit et al., 2005;Garcia-Gonzalo et al., 2011). Mutations in human TCTN2 cause Meckel and Joubert syndromes (Huppke et al., 2015;Sang et al., 2011;Shaheen et al., 2011).
To assess whether this narrowing of the facial midline is specific to TCTN2, we examined the possible involvement of two other components of the MKS complex, TMEM231 and CC2D2A, in craniofacial development. Human CC2D2A mutations cause Meckel and Joubert syndromes, and TMEM231 mutations cause Meckel, Joubert, and Orofaciodigital syndromes (Noor et al., 2008;Roberson et al., 2015;Shaheen et al., 2013;Srour et al., 2012;Tallila et al., 2008). Similar to Tctn2 mutants, both E10.5 Cc2d2a and Tmem231 mutant embryos exhibited decreased infranasal distance ( Figure 1B and C, respectively). The similarity of the midline hypoplasia in all three transition zone mutants suggested a common etiology.
We also examined the involvement of a fourth member of the MKS complex, TMEM67, in craniofacial development. Human mutations in TMEM67 also cause Meckel and Joubert syndromes (Otto et al., 2009;Smith et al., 2006). Mutation of mouse Tmem67 causes phenotypes that are less severe than Tctn2, Tmem231, or Cc2d2a (Garcia-Gonzalo et al., 2011). The mild phenotype of Tmem67 mutants may be attributable to its dispensability for ciliary accumulation of HH pathway activator SMO . Unlike Tctn2, Tmem231, and Cc2d2a mutants, Tmem67 mutants did not exhibit altered infranasal distance ( Figure 1D). Thus, some, but not all, MKS components are critical for early facial midline development.
Given the central role of the neural crest in craniofacial development as the main source of craniofacial mesenchyme (Santagati and Rijli, 2003), we tested whether transition zone disruption in neural , Tmem231 (C) wild-type and null embryos at embryonic day (E)10.5 and E11.5. Tmem67 null embryos (D) display normal midface width at E11.5. Quantification of midface width (denoted by yellow brackets) at respective timepoints was measured via one-way ANOVA followed by Tukey's multiple comparisons test. Data are expressed as mean, and error bars represent the standard deviation (SD) with individual data points (N) representing biological replicates (biologically distinct samples). Scale bar indicates 500 μm. ns = not significant.
The online version of this article includes the following figure supplement(s) for figure 1:  crest can account for the midline hypoplasia observed in Tctn2 mutants. More specifically, we generated E11.5 Wnt1 cre ;Tctn2 fl/embryos and quantified midface width (Figure 1-figure supplement 2A). Wnt1 cre induces recombination throughout the neural crest, and conditional deletion of Tctn2 in the neural crest abrogated ARL13B localization to cilia (Figure 1-figure supplement 2B,C). Interestingly, removing TCTN2 from the neural crest did not cause hypotelorism. These results indicate that altered transition zone function in the neural crest is not the etiology of midline hypoplasia. Therefore, we investigated functions of TCTN2 in the prechordal plate, an early organizing center also critical for the development of anterior head structures.
Tctn2 mutants exhibit defects in prechordal plate differentiation soon after gastrulation As Tctn2, Cc2d2a, and Tmem231 mutants all displayed facial midline defects at midgestation, we hypothesized that they shared a role in a patterning event early in craniofacial development. One organizing center critical for early forebrain and craniofacial development is the prechordal plate (Camus et al., 2000;Kiecker and Niehrs, 2001;Muenke and Beachy, 2000;Rubenstein and Beachy, 1998;Som et al., 2014). The prechordal plate is the anterior-most midline mesendoderm, immediately anterior to the notochord and in contact with the overlying ectoderm. The homeobox gene Goosecoid (Gsc) is specifically expressed in the prechordal plate at E8.0 and is a marker of differentiation in this organizing center (Belo et al., 1998;Izpisúa-Belmonte et al., 1993). In contrast, Shh and Brachyury (T) are expressed at E8.0 in both the prechordal plate and notochord and are critical for prechordal plate induction (Aoto et al., 2009). Previous work demonstrated that surgical removal of the rat prechordal plate results in midface defects (Aoto et al., 2009) that seemed similar to those of the mouse Tctn2, Cc2d2a, and Tmem231 mouse mutants.
Therefore, we analyzed the prechordal plate of Tctn2 mutants by examining the expression of prechordal plate differentiation marker Gsc and induction markers Shh and T. Gsc is expressed specifically in the prechordal mesoderm, while Shh and T are expressed in both the prechordal mesoderm and notochord (Dale et al., 1997;Herrmann, 1991;Schulte-Merker et al., 1994). In situ hybridization analysis revealed that in Tctn2 mutants, Shh and T expression in the prechordal plate and notochord were unaffected ( Figure 2A). Therefore, TCTN2 is not essential for prechordal plate specification. In contrast, Tctn2 mutants exhibited abrogated Gsc expression in the prechordal plate ( Figure 2B), indicating that TCTN2 is critical for prechordal plate differentiation.
The transition zone is critical for HH signaling, and one HH protein, SHH, is essential for Gsc expression in the prechordal plate (Aoto et al., 2009;Chih et al., 2011;Garcia-Gonzalo et al., 2011). Therefore, we investigated whether TCTN2 is required for HH signaling in the prechordal plate by examining the expression of the transcriptional target Gli1. Tctn2 mutants exhibited reduced Gli1 expression throughout the axial mesendoderm, including the prechordal plate ( Figure 2B). These results indicate that TCTN2 is dispensable for the formation of the prechordal plate, but is required for midline signaling by SHH to induce Gsc expression in this organizing center.
TCTN2 and other members of the MKS complex are required for proper cilia formation in some tissues but not in others . Therefore, we examined whether TCTN2 is required for ciliogenesis in the prechordal plate. The prechordal plate expresses FOXA2 (Jin et al., 2001; Figure 2C). Co-immunostaining embryos for FOXA2 and acetylated tubulin (TUB Ac ), a marker of cilia, revealed that Tctn2 mutants did not display decreased ciliogenesis in the E8.5 prechordal plate ( Figure 2C, middle panel). Previous studies have shown that Tctn1 is expressed in the ventral epithelium of the node of a six-somite stage embryo and in the neural tube, notochord, gut epithelium, and somites at E9.5 (Reiter, 2006). Tctn2 is similarly broadly expressed during mouse development (Diez-Roux et al., 2011;Magdaleno et al., 2006), and CC2D2A is broadly expressed during human development (Mougou-Zerelli et al., 2009).
In cell types in which the MKS complex is dispensable for ciliogenesis, like neural progenitors, it is required for localization of ARL13B to cilia . Therefore, we examined ARL13B localization in E8.0 control embryos and Tctn2 mutants and discovered that ARL13B localization to prechordal plate cilia was attenuated without TCTN2 ( Figure 2C, bottom panel). Thus, TCTN2 is not required for ciliogenesis in the prechordal plate, but does control ciliary composition.

Figure 2 continued on next page
Tctn2 mutants display decreased HH signaling in the ventral telencephalon The axial mesendoderm helps pattern the overlying neurectoderm (Anderson and Stern, 2016;Rubenstein and Beachy, 1998). In the rostral embryo, the prechordal plate patterns the overlying ventral telencephalon via SHH (Chiang et al., 1996;Xavier et al., 2016). As extirpation of the prechordal plate results in decreased SHH activity in the basal telencephalon (Aoto et al., 2009;Aoto and Trainor, 2015), we investigated whether the prechordal plate defects observed in Tctn2 mutants result in mispatterning of the ventral telencephalon. Although Shh expression in the notochord was unaffected in Tctn2 mutants at E8.75, it was severely reduced in the ventral telencephalon ( Figure 3A). This reduced expression of Shh in the ventral telencephalon persisted at E9.5 ( Figure 3A). These results are concordant with previous findings that mutations in genes encoding other transition zone Middle panel in C is magnified region in top panel highlighted by dotted rectangle and rotated 90 degrees. Scale bar in A-B indicates 0.2 mm, C (top panel) is 100 μm, C (middle and bottom panels) is 10 μm. components disrupt brain development, resulting in holoprosencephaly, reduced telencephalon size, or exencephaly (Dowdle et al., 2011;Garcia-Gonzalo et al., 2011;Reiter, 2006).

Figure 2 continued
To assess whether HH pathway activity was compromised by the absence of TCTN2, we assessed the expression of the HH transcriptional targets Gli1 and Ptch1. Whole mount in situ hybridization (WM-ISH) of Tctn2 mutants revealed dramatically reduced or absent expression of both Gli1 and Ptch1, especially in the basal forebrain ( Figure 3B). Consistent with the WM-ISH data, quantitative real-time polymerase chain reaction (RT-qPCR) analysis of E8.5 ( Figure 3C) and E9.5 ( Figure 3D) Tctn2 mutant heads also revealed decreased expression of Shh, Ptch1, and Gli1, revealing that Tctn2 mutants exhibit both an early defect in prechordal plate differentiation and a defect in HH signaling in the adjacent neurectoderm.

TCTN2 protects neurectoderm and facial ectoderm from apoptosis
In the developing craniofacial complex, SHH induces cell proliferation (Hu and Helms, 1999;Hu et al., 2015). Therefore, we assessed if the reduction in facial midline width in Tctn2 mutants was due to decreased cell proliferation. More specifically, we measured cell proliferation by quantitating phospho-histone H3 (pHH3) in the components of the craniofacial complex -the forebrain, hindbrain, facial ectoderm, and mesenchyme ( Figure 4A and B). Tctn2 mutants showed no differences in amount or spatial distribution of cell proliferation.
In other developmental contexts, HH signaling promotes cell survival (Ahlgren and Bronner-Fraser, 1999;Aoto et al., 2009;Aoto and Trainor, 2015;Litingtung and Chiang, 2000). Therefore, we assessed apoptosis in the craniofacial complex of Tctn2 mutants. Quantification of TUNEL staining revealed that apoptosis was increased in the neurectoderm, facial ectoderm, and mesenchyme of Tctn2 mutants compared to controls, and most dramatically in the ventral telencephalon ( Figure 4C and D). To further test whether apoptosis is increased in the absence of TCTN2, we stained for activators of the intrinsic apoptotic pathway, cleaved-caspase-3 and caspase-9 (activated CASP3 and CASP9). Both activated CASP3 and CASP9 were increased in the ventral telencephalon, facial ectoderm, and mesenchyme of Tctn2 mutants at E9.5 ( Figure 4E). These data indicate that TCTN2 is required to protect against cell death, but does not affect proliferation, in the neurectoderm, nonneural ectoderm and neural crest mesenchyme. As SHH also protects neurectoderm from apoptosis (Thibert et al., 2003), we propose that TCTN2 mediates cell survival by promoting HH signaling and that the increase in cell death in the neurectoderm, mesenchyme and facial ectoderm underlies the midline hypoplasia in transition zone mutants.
Our finding that selective disruption of transition zone function in the neural crest did not contribute to midline growth ( Similarly, we investigated the function of TCTN2 in the facial ectoderm and telencephalon. Tcfap2a Cre and Foxg1 Cre activated recombination in the facial ectoderm or telencephalon and ectoderm, respectively (Figure 4-figure supplement 1E, F). Both Tcfap2a Cre ;Tctn2 fl/and Foxg1 Cre ;Tctn2 fl/embryos displayed no decrease in midline width at E11.5 (Figure 4-figure supplement 1G, H).
Analysis of ARL13B localization to cilia at E8.5 and E9.5 in the prechordal plate and neurectoderm in Isl1 Cre ;Tctn2 fl/and Sox1 Cre ;Tctn2 fl/embryos, respectively, revealed that there was no alteration in ciliogenesis or ciliary localization of ARL13B (Figure 4-figure supplement 2A and B). Analysis of ciliary ARL13B localization in the ectoderm of E11.5 Tcfap2a Cre ;Tctn2 fl/mutants similarly revealed persistent ARL13B localization to cilia (Figure 4-figure supplement 2C). In contrast, Foxg1 Cre ;Tctn2 fl/mutants exhibited loss of ARL13B ciliary localization in the basal telencephalon at E11.5 (Figure 4-figure  supplement 2D). These results indicate that Tctn2 conditional deletion in either the prechordal plate, neurectoderm, facial ectoderm, or forebrain individually fails to recapitulate the midline narrowing observed in Tctn2 -/embryos. However, the persistence of ARL13B at cilia suggests that TCTN2 function persists for some time after recombination or TCTN2 functions in the prechordal plate, neurectoderm and facial ectoderm may be important for facial development. To assess whether decreased HH signaling is not just correlated with midfacial hypoplasia in transition zone mutants, but is causative, we investigated whether modulating the HH pathway could rescue the midface defects. We employed a strategy targeting Ptch1, a negative regulator of the HH pathway, by generating Tctn2 -/-;Ptch1 +/embryos and comparing them to Tctn2 -/-;Ptch1 +/+ embryos. Surprisingly, removing a single allele of Ptch1 in Tctn2 mutants restored midface width at E11.5 ( Figure 5A and B).
As genes encoding transition zone MKS components are epistatic to Ptch1 (Reiter, 2006), we pondered how reducing Ptch1 gene dosage restored facial midline development to transition zone MKS component mutants. The best studied role for PTCH1 is in repression of the HH signal transduction pathway. Therefore, we examined HH signal transduction pathway activity in Tctn2 -/-;Ptch1 +/+ and Tctn2 -/-;Ptch1 +/embryos. WM-ISH revealed that expression of neither Shh nor Gli1 was increased in the ventral telencephalons of Tctn2 -/-;Ptch1 +/in comparison to Tctn2 -/-;Ptch1 +/+ embryos ( Figure 5G). Thus, the restoration of midface growth by reduction of Ptch1 gene dosage is not due to a restoration of HH signal transduction.
In addition to its role in regulating HH signal transduction, PTCH1 exhibits pro-apoptotic activity in vitro (Thibert et al., 2003). In the absence of ligand, PTCH1 C-terminal domain is cleaved and binds scaffolding proteins TUCAN1 and DRAL to recruit caspase-9 and activate caspase-3, resulting in apoptosis. The colocalization of another PTCH1-binding protein that regulates apoptosis, X-linked inhibitory apoptosis protein (XIAP), with PTCH1 at cilia (Aoto and Trainor, 2015), raises the possibility that PTCH1 cleavage occurs at the primary cilium to induce apoptosis. In addition, PTCH1 may induce apoptosis at the plasma membrane. As reducing Ptch1 gene dosage reduces apoptosis without increasing HH signal transduction in Tctn2 mutants, our data supports a model in which the PTCH1mediated death of the midline neurectoderm, facial ectoderm, and neural crest-derived mesenchyme, and not alterations in HH signal transduction within those cells, is the etiology of midface defects in transition zone mutants. Taken together, these results suggest a working model for how ciliary HH signaling regulates midface development.
In wild-type embryos, HH signaling within the prechordal plate is critical for Gsc expression and the induction of Shh in the adjacent neurectoderm and inhibition of apoptosis ( Figure 6A). In transition zone mutants, defects in prechordal plate signaling cause reduced SHH in the neurectoderm, resulting in increased PTCH1-mediated cell death and midline collapse ( Figure 6B). In transition zone mutants lacking a single allele of Ptch1, reduced SHH in the neurectoderm persists, but the attenuated PTCH1 is no longer sufficient to induce extensive cell death, allowing for normal midline facial development ( Figure 6C).

Discussion
Using a combination of genetic, developmental, and biochemical techniques, we have identified a mechanism by which disruption of MKS transition zone proteins (TCTN2, CC2D2A, and TMEM231) results in midline hypoplasia and hypotelorism. We traced the origin of the molecular defect contributing to the midline phenotype to the prechordal plate, defects in which resulted in reduced HH pathway activation and cell survival in the adjacent neurectoderm and facial midline collapse. We uncovered Ptch1 gene dosage as a key mediator of cell survival in the facial midline of transition zone mutants, as loss of a single allele of Ptch1 rescued cell survival and midline development in Tctn2 and Cc2d2a mutants. Together, these results reveal a new paradigm whereby primary cilia mediate signal crosstalk from the prechordal plate to the adjacent neurectoderm to promote cell survival, without which the facial midline collapses and hypotelorism results.
Different ciliopathies are associated with either narrowing or expansion of the facial midline (hypotelorism or hypertelorism) (Schock et al., 2015;Zaghloul and Brugmann, 2011). Severe ciliopathies associated with perinatal lethality, such as MKS, can present with hypotelorism or hypertelorism while other ciliopathies such as Joubert syndrome typically present with hypertelorism (Brugmann et al., 2010b;MacRae et al., 1972;Schock and Brugmann, 2017). How disruption of ciliary functions can give rise to these opposing phenotypes has been an active area of interest. Hypertelorism is attributable to roles for cilia in promoting GLI3 repressor formation in neural crest cells (Brugmann et al., 2010a;Chang et al., 2014;Chang et al., 2016;Liu et al., 2014). Our work implicates a distinct etiology of hypotelorism: rather than involving neural crest, midline hypoplasia can be caused by defects in the ciliary transition zone in the prechordal plate.
The earliest alteration we detected in Tctn2 -/signaling centers that regulate craniofacial development was in the prechordal plate at the end of gastrulation. TCTN2 was dispensable for the expression of Shh and T in the prechordal plate, indicating that induction and specification of the prechordal plate were not dependent on transition zone function. In contrast, TCTN2 was essential for prechordal plate expression of Gli1 and Gsc. In tissues such as the limb bud, TCTN2 is dispensable for ciliogenesis but critical for ciliary HH signaling and the induction of HH target genes such as Gli1 (Chih et al., 2011;Dowdle et al., 2011;Garcia-Gonzalo et al., 2011). We found that, similarly in the prechordal plate, TCTN2 is dispensable for ciliogenesis but critical for induction of Gli1.
In many developmental events, such as limb patterning, SHH signals to neighboring cells to induce a pattern (Panman and Zeller, 2003;Zhulyn et al., 2014). In other developmental events, such as notochord to neural tube signaling, SHH signals produced by the notochord induce the expression of Shh in the overlaying neural tube (Fuccillo et al., 2006). SHH produced by the prechordal plate may fall into the latter category, as the absence of Shh expression in Tctn2 -/embryos presages reduced Shh expression and reduced expression of HH pathway transcriptional targets Gli1 and Ptch1 in the region of the basal forebrain sometimes referred to as the rostral diencephalon ventral midline. Thus, in the posterior midline, the notochord induces Shh expression in overlying neuroectoderm, and in the anterior midline, the prechordal plate induces Shh expression in the overlying neuroectoderm. In the failure of the prechordal plate to induce Shh expression in the overlying neurectoderm, Tctn2 mutants recapitulate previous observations of Lrp2 mutants which display attenuated responses to SHH (Christ et al., 2012). One possible mechanism by which SHH may activate expression of Shh in the basal forebrain is via the induction of the transcription factor SIX3. SIX3 is regulated by HH signaling and required for the induction of Shh in the developing forebrain Jeong et al., 2008). However, our observation that Six3 expression is unaltered in the forebrains of Tctn2 mutants diminishes support for this hypothesis.
In caudal neural tube and limb patterning, HH signals induce patterning. In hair follicles and cerebellar granule cells, HH signaling promotes proliferation. In addition to roles in patterning and proliferation, HH signals can bind to PTCH1 to inhibit apoptosis (Aoto and Trainor, 2015;Borycki et al., 1999;Thibert et al., 2003). Our data are consistent with a role of PTCH1 in promoting apoptosis in the neuroectoderm and facial ectoderm which is inhibited by prechordal plate-produced SHH. In the absence of TCTN2, the ventral telencephalon does not produce SHH, releasing PTCH1 to promote apoptosis in the midline and resulting in midface hypoplasia. This model is consistent with previous data demonstrating that surgical ablation of the prechordal plate reduces the forebrain (Aoto et al., 2009). Increased cleaved-caspase-3 and caspase-9 staining in the basal forebrain and facial ectoderm of E9.5 Tctn2 mutants provides further support for apoptosis contributing to the associated midline hypoplasia.
We speculate that in tissues, such as the developing spinal cord where cilia are required for SHH expression in the floor plate, ciliary dysfunction will cause increased apoptosis. In other tissues, such as the limb bud where cilia are not required for SHH expression in the zone of proliferating activity, we predict that ciliary dysfunction will not cause increased apoptosis. Thus, ciliary dependence of SHH expression may etermine which tissues, like the craniofacial midline, increase apoptosis upon ciliary dysfunction.
To try to narrow down the tissues in which transition zone function is critical for midline facial development, we used conditional mouse genetics to delete Tctn2 in different tissues that comprise the craniofacial complex. Deletion of Tctn2 in the prechordal plate (via Isl1 Cre ) or the neurectoderm (via Sox1 Cre ) did not recapitulate the midline hypoplasia observed in Tctn2 -/embryos ( Figure 4-figure  supplement 1). Similarly, deletion of Tctn2 in the facial ectoderm (via Tcfap2a Cre ) or in the forebrain and facial ectoderm (via Foxg1 Cre ) also failed to result in midline hypoplasia (Figure 4-figure supplement 1). These results could reflect the dispensibility of TCTN2 in these tissues for facial development, or could reflect residual TCTN2 function after Tctn2 recombination. This latter possibility is supported by persistent ARL13B localization at the cilium in several tissues after conditional deletion of Tctn2. As the half-life of TCTN2 and the TCTN2 level required to sustain transition zone activity are unclear, residual TCTN2 activity may support normal ciliary signaling even after conditional gene ablation.
Surprisingly, removing one allele of Ptch1 fully rescues the midface defect in both Tctn2 and Cc2d2a transition zone cilia mutants. Even more surprisingly, this phenotypic rescue is not associated with restoration of either Shh expression or HH pathway activation in the basal forebrain. We propose that reducing Ptch1 levels attenuates the PTCH1-mediated pro-apoptotic program normally attenuated by SHH in the basal forebrain. However, it remains possible that reducing PTCH1 levels activates GLI effectors to induce an anti-apoptotic program that does not include general HH target genes or Shh.
In summary, we have identified the primary cilia transition zone as a critical regulator of facial midline development. The transition zone component TCTN2 was critical for SHH signaling in the prechordal plate and uncovered a signaling paradigm whereby the transition zone promotes cell survival by mediating crosstalk between the prechordal plate and neurectoderm to promote HH pathway activation. These results provide insights into how primary cilia mediate cell survival to promote facial development.  Sang et al., 2011). More specifically, we removed the puromycin resistance cassette of Tctn2 tm1aReit by Cre-mediated recombination, leaving two loxP sites flanking the first three Tctn2 exons to generate Tctn2 fl . All mice were maintained on a C57BL/6 J background. For timed matings, noon on the day a copulation plug was detected was considered to be 0.5 days postcoitus. Genotyping primers for all mouse strains used in this study can be found in Supplementary file 1.

RT-qPCR
For gene expression studies, RNA was extracted from E8.5/E9.5 embryo heads using an RNAeasy Micro Kit (QIAGEN), and cDNA synthesis was performed using the iScript cDNA synthesis kit (BioRad). RT-qPCR was performed using EXPRESS Sybr GreenER 2 × master mix with ROX (Invitrogen) and primers homologous to Shh, Ptch1, or Gli1 on an ABI 7900HT RT-PCR machine. Expression levels were normalized to the geometric mean of three control genes (Actb, Hprt and Ubc), average normalized Ct values for control and experimental groups determined, and relative expression levels determined by ΔΔCt. The RT-qPCR of each RNA sample was performed in quadruplicate with a minimum of N = 3 biological replicates (biologically distinct samples) per genotype. All primers used for RT-qPCR experiments can be found in Supplementary file 1.

Embryo processing for midface imaging
Embryos were harvested in ice-cold PBS, staged by counting somite number, and fixed o/n at 4 degrees in 4%PFA/PBS. Embryo heads were removed and stained in 0.01 % ethidium bromide in PBS at room temperature for 15 min. Embryos were positioned using glass beads in PBS and imaged on a Leica MZ16 F fluorescence stereomicroscope.

Image Quantification
For 2D midface width quantification, the infranasal distance was measured using FIJI software by drawing a line between the center of each nasal pit. For quantification of cell death and proliferation assays, a minimum of two sections per embryo were quantified over three biological replicates (biologically distinct samples). Staining with epithelial marker E-cadherin was used for quantification of facial ectoderm while nuclear morphology was used to separate mesenchyme, hindbrain, and forebrain tissue compartments. For quantification, threshold was first set for each image followed by binary watershed separation to obtain accurate nuclei counts. The percentage of TUNEL + or pHH3 + nuclei were compared between Tctn2 mutant and control samples.

Additional information
Competing interests Jeremy F Reiter: Reviewing editor, eLife. The other author declares that no competing interests exist. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Additional files
Supplementary files • Supplementary file 1. Primer sets used for genotyping and quantitative real-time polymerase chain reaction (RT-qPCR). File contains all sequencing primer pairs used for genotyping of all mouse strains used in this study. In addition, file contains primer sets used for RT-qPCR experiments. Inquiries regarding primers should be sent to Jeremy Reiter.
• Transparent reporting form

Data availability
All data generated or analysed during this study are included in the manuscript and supporting files.