talpid² (ta²) heterozygous carriers were mated, eggs were collected and shipped from the UC Davis Avian Sciences Department. Embryos were incubated at 37°C for approximately 5–7 days when embryos reached Hamburger Hamilton stage 25–31 (HH25-31).

FNPs, MXPs, and MNPs were harvested from day 5 chick embryos. mRNA was prepared with TRIzol reagent (Thermo Fisher Scientific), and then converted to complementary DNA through reverse transcription reaction (High-Capacity cDNA Reverse Transcription Kit, Applied Biosystems™). Different amounts of cDNA (40, 20, 10 and 5 ng) was used for quantitative PCR to test PCR efficiency and a linear range of duplication (SsoAdvanced™ Universal SYBR® Green Supermix, BioRad). Expression of genes known to play a role in craniofacial development were examined with the following primer sets: ALX4 (105 bp) F: GTTACGGTAAGGAGAGCAGTTT, wordR: CTTTCACTCCAGCCTCCTTC, BMPR1A (100 bp) F: GTGCTGTCGGACTGATTTCT, wordR: TGCCATCCAACGAATGCT, WIF1 (100 bp) F: CAACCTGTTTCAATGGAGGAAC, R: GGCTGATGGCATTTACTGATTT, OSR2 (140 pb) F: CCACTTCACCAAGTCCTACAA, R: TCTCTTTGGAATGGAT GTACCG. The statistical significance of the data was evaluated through two-tailed Student's t-test. p-values less than or equal to 0.05 (95% confidence level) were considered as statistically significant differences.

FNPs, MXPs, or MNPs were pooled and lysed in RIPA buffer containing protease inhibitors and phosphatase inhibitors (1 mM Phenylmethylsulfonyl fluoride, 1 mM NaVO₄, 1X complete protease inhibitor cocktail, EDTA-free), and slightly sonicated with a microprobe to recover chromosome bound GLI2 and GLI3 proteins. BCA assay (Pierce) was used to measure protein concentration of cell lysates. Proteins were boiled with 1X Laemmli sample buffer and run on 6% SDS-PAGE for GLI2 and GLI3, or 12% SDS-PAGE for GAPDH, which later were wet-transferred to Polyvinylidene difluoride (PVDF) membrane. Anti-GLI2, anti-GLI3 (Polyclonal goat IgG 1:500, R&D systems) and anti-GAPDH (FL-335, polyclonal rabbit IgG 1:4000, Santa Cruz Biotechnology) were prepared in 1X TBST (0.1% Tween-20)/6% nonfat milk, as well as secondary antibodies (anti-goat and anti-rabbit, 1:10,000). Proteins were detected by Electrochemiluminescence assay (Amersham ECL Prime, GE Healthcare Life Science).

PATCHED 1 promoter oligonucleotides were designed according to the location of a Gli binding region (GBR) at −2549 from the TSS site (GGAAGAAGTGTCAGTGTAAGAGTCTCCACGTGGGTGGTCAAGGCCATGGCTGCCTCACGG). 100 pmole of biotin-conjugated positive oligonucleotides and complementary oligonucleotides (Integrated DNA Technologies) were annealed in 1X TE/50 mM NaCl buffer in a PCR cycler and incubated with Dynabeads Streptavidin (M280, Invitrogen). FNPs, MXPs, or MNPs from day 5 control or ta² embryos were pooled and processed as described for Western blotting. Cell lysates were incubated with oligonucleotides-bound Dynabeads at 4°C for 2 h. Beads were washed with 1 ml RIPA buffer three times and processed for Western Blotting analysis.

FNPs, MXPs, or MNPs from day 5 control or ta² embryos were harvested, pooled and crosslinked in 1% formaldehyde. Tissue was homogenized in RIPA buffer and sonicated (Bioruptor®; Pico, Diagenode) at 5 cycles of 30 s on/45 s off. Sheared DNA was distributed around 0.3 kb to 1 kb on 1% agarose gel. Cell lysates were pre-cleaned with Dynabeads protein G (ThermoFisher Scientific) and quantified by BCA assay. Dynabeads for immunoprecipitation were blocked with 20 μg/ml Glycogen, 20 μg/ml BSA, 20 μg/ml yeast RNA in RIPA buffer at 4°C for an hour. GLI3 antibody (AF3690, R&D systems) and pre-blocked Dynabeads Protein G were incubated with 90% of cell lysates at 4°C overnight. 10% of cell lysates were kept as Input. Beads were washed with IP wash buffer I (low salt; 50 mM HEPES-KOH pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% sodium deoxycholate, 1% Triton X-100), IP wash buffer II (high salt; 50 mM HEPES-KOH pH 7.5, 500 mM NaCl, 1 mM EDTA, 0.1% sodium deoxycholate, 1% Triton X-100), IP Wash Buffer III (LiCl containing buffer; 10 mM Tris-Cl pH 8.0, 250 mM LiCl, 1 mM EDTA, 0.5% Sodium deoxycholate, 0.5% NP-40) and TE buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA). ChIP samples were reverse-crosslinked by boiling with 10% Chelex-100 (BioRad), and treated with 0.2 mg/ml Proteinase K at 55°C for 30 min. Immunoprecipitated DNA samples were analyzed with quantitative real-time PCR (BioRad) with primers to GBR of target genes. Error bars in all figures represent standard error of the mean (S.E.M.) from five to seven independent experiments. The statistical significance of the data was evaluated through two-tailed Student's t-test. p-values less than or equal to 0.05 (95% confidence level) were considered as statistically significant differences.

Sequences for GBRs from previous publications (Vokes et al., 2007, 2008) were used with a custom perl script to search for all the exact matches of the various possible sequences of the consensus GBRs in the chicken genome. Acquired positions of the motifs in the genome were run through a second custom perl script to search for genes that encompass these motif sites at a distance of 20 kb from either end. Potential GLI targets were confirmed using chromatin-immunoprecipitation (ChIP) assays.

To understand the transcriptional networks affected in craniofacial ciliopathies we analyzed the talpid² (ta²) mutant, a naturally occurring avian ciliopathic mutant that has been established as a model for the human craniofacial ciliopathy, Oral-facial-digital syndrome 14 (Chang et al., 2014; Schock et al., 2015). The ta² craniofacial phenotype characterized by facial and palatal clefting, micrognathia, and hypoglossia, is fully evident at day 7 (Figure 1). Although our previous work has characterized this phenotype (Chang et al., 2014; Schock et al., 2015), to determine the transcriptional networks that contribute, we first needed to identify when phenotypic onset occurred. At day 7 the frontonasal prominence (FNP) is shorter and wider and frequently does not fuse to adjacent prominences (Figures 1A–B'). Two days earlier at day 5, the MXPs were medially rotated, the nasal pits were larger, thus preventing the proper juxtapositioning of the FNP with adjacent prominences (Figures 1C–D'). Palatal views showed increased patency of the naturally cleft avian secondary palate in ta² embryos relative to controls at day 7 (Figures 1E–F', dotted white lines). Two days earlier, at day 5, the dysmorphology and malposition of the MXPs is just becoming apparent (Figures 1G–H'). Dorsal views of the developing lower beak showed the MNP of ta² embryos failed to fuse completely and had a hypoplastic tongue (hypoglossia, dotted black line) (Figures 1I–J', white arrow). At day 5 there was little difference in mandibular growth between control and ta² MNPs (Figures 1K–L'). From phenotypic evaluations, taken together with the fact that these anomalies were not readily identifiable at day 4, we determined that craniofacial anomalies were initiated at approximately day 5 of development. Thus, to determine the molecular networks responsible for these phenotypes we carried out our analyses in the facial prominences of day 5 embryos.

Our previous work determined that aberrant ciliogenesis in ta² embryos disrupted the production of GLI proteins in such a manner that there was increased GLI_FL production and decreased GLI_T production (Chang et al., 2014). The GLI_FL isoform typically goes on to function as an activator, whereas the GLI_T isoform goes on the function as a repressor. The excessive production of GLI_FL is an extraordinarily common molecular phenotype in ciliopathies, including those with craniofacial phenotypes (Huangfu and Anderson, 2005; Davey et al., 2006; Tran et al., 2008; Tabler et al., 2013). To carefully exam the differences of GLI_FL and GLI_T protein levels in control and ta² mutants, we performed Western blot analysis of GLI2 and GLI3 in the three facial prominences affected in the ta² at day 5. We detected a very low level of GLI2_FL and a substantial amount of GLI2_T isoforms in the control facial prominences (Figure 2A). The loss of cilia in ta² embryos disrupted GLI processing and altered the production of GLI2 protein isoforms. In ta² facial prominences, we detected dramatically increased levels of GLI2_FL, and low levels of GLI2_T, relative to control prominences. We next examined the production of GLI3 isoforms. Western blot analyses showed that, similar to GLI2_FL, GLI3_FL was increased in ta² prominences relative to controls. Contrary to what was observed with the GLI2_T, GLI3_T was readily detectable in both control and ta² facial prominences. Specifically, ta² MNP had less GLI3_T than control MNP. These data suggested that disrupted ciliogenesis affected the processing of GLI2 and GLI3 in distinct manners, yet the net result was an increase in GLI_FL production. We did not observe a change in GLI1 protein levels between control and mutant embryos (Data not shown).

Since the GLI_FL isoform typically goes on to act as an activator, these results suggest that craniofacial phenotypes in the ta² mutant could be caused by increased GLI activator function leading to increased expression of GLI target genes. To test this hypothesis, we utilized an in silico approach to look for potential GLI target genes within the avian genome. Using previously published sequences of GBRs (Vokes et al., 2007, 2008) we scanned the chicken genome for possible GLI targets. GBR positions were run through a custom perl script to search for genes that encompass these motif sites at a distance of 100 kb from the 5′ or 3′ end of the gene. From this list we identified several genes known to play a role in craniofacial development (Supplemental Table 1). Confirmation of our in silico approach was carried out on selected genes using ChIP assays (Supplemental Figures 1 A–D). Through these analyses, we selected four GLI3 target genes known to be expressed during, and have a role in, craniofacial development: ALX4 (Beverdam et al., 2001; Mavrogiannis et al., 2001), BMPR1A (Li et al., 2011, 2013; Saito et al., 2012), OSR2 (Lan et al., 2001), and WIF1 (Hsieh et al., 1999; Darnell et al., 2007). PTCH1, a known GLI target, was used as a positive control (Supplemental Figure 1E) and IgG IP as an antibody background control. The data were all normalized to IgG IP percentage, and the genes with relative enrichment >1 were considered positive for GLI3 binding. To determine if increased production of GLI3_FL in ta² facial prominences correlated with increased expression of these target genes, we performed quantitative RT-PCR (qPCR) with mRNA from facial prominences of day 5 control and ta² embryos (Figures 2B–E). There was not an increase in ALX4 expression in any of the facial prominences (Figure 2B); however, we detected a significant decrease in ALX4 expression in the MXP. BMPR1A expression was also not significantly increased in any of the developing prominences (Figure 2C). No significant changes in OSR2 expression were detected in the FNP or MNP; however, a significant increase in expression was observed in the MXP (Figure 2D). Finally, WIF1 expression was not changed in the FNP, yet was significantly increased in the MXP and significantly decreased in the MNP (Figure 2E). Taken together, these data do not support the idea that increased production of GLI3_FL directly and uniformly results in increased expression of GLI targets throughout the facial prominences. Additionally, these data suggest that each facial prominence interprets aberrant GLI production in a unique manner.

For GLI_FL to function as an activator, it has to occupy the regulatory regions of GLI targets. We wondered if lack of uniform increases of target gene expression in ta² embryos was due to failure of GLI_FL to recognize and occupy GBRs of target genes. To test this hypothesis, we performed an in vitro DNA binding assay using the PATCHED 1 (PTCH1) promoter. We synthesized a 60 base pair biotin-labeled oligonucleotide of the PTCH1 promoter containing an endogenous GLI binding motif found at position −2549 proximally upstream of the transcription start site (TSS). Through high affinity of Biotin-Streptavidin interaction, we were able to evaluate the DNA binding ability of GLI3 isoforms by Western blot. Pre-incubation of non-labeled PTCH1 oligonucleotides depleted GLI3 protein signals, which confirmed the specificity of the GLI3-GBR interaction. Under the same exposure, the affinity based pull-down assay showed that GLI3_T predominantly bound to the biotin-labeled oligonucleotides in both control and ta² mutants. Interestingly, despite the high level of GLI3_FL production in ta² embryos, GLI3_FL failed to bind to the biotin-labeled oligonucleotides (Figures 3A,B). On the contrary, comparison between control and ta² embryos indicated that the amount of GLI3_T binding to the oligonucleotides correlates with the protein concentration (Figures 3A,B). Taken together these data suggest that despite increased GLI3_FL production, target gene expression is not increased via increased GLI activator function because GLI3_FL does not occupy the GBRs of target genes. In addition, the predominant binding of the GLI3_T repressor at GBRs supports the idea of the threshold repression model, in which GLI targets underwent de-repression due to the removal of GLI_T repressor.

In vitro DNA binding affinity assays suggested that there was not an increase of ectopic GLI3_FL at GBRs of target genes in ta² mutants. To determine if GLI3 binding to GBRs was altered in vivo in ta² mutants compared to control embryos, we next performed ChIP-qPCR (Figure 4). In ta² mutants, GLI3 enrichment at GBRs associated with the craniofacial genes ALX4 and OSR2 was reduced in all facial prominences, yet only significantly in the MXP and MNP (Figures 4A,B). GLI3 enrichment at the GBRs associated with BMPR1A and WIF1, was also overwhelmingly reduced in facial prominences, yet only significantly in the ta² FNP and MXP (Figures 4C,D). GLI3 binding in the MNP of WIF1 was below detectable levels in controls, and thus could not be evaluated. These data suggested that GLI3 occupancy at the GBR of target genes is decreased. Our in vitro data suggested that GLI3_FL did not bind at target gene GBRs (Figure 3). Thus, taken together these data indicate that the observed reduction of GLI3 binding is indicative of reduced GLI3_T binding at the regulatory regions of target genes.

Reduced GLI3 binding at GBRs of GLI target genes appeared to be a general trend in ta² mutants, as several other target genes expressed in the craniofacial complex also exhibited reduced enrichment of GLI3 at associated GBRs (Supplemental Figure 2). Thus, these data suggested that altered expression of GLI targets could be caused by aberrant GLI3_T binding in ta² mutants. Interestingly, there was not always a direct correlation with loss of GLI3_T repressor binding and increased gene expression. Furthermore, different results among the three facial prominences examined pointed to context—and tissue-specific regulation of GLI targets (Supplemental Figures 1A–D). Several studies have proposed the possibility that GLI proteins work together with co-regulators to influence expression of targets genes (Brewster et al., 1998; Koyabu et al., 2001; Mizugishi et al., 2001; Lee et al., 2010; Peterson et al., 2012). To determine if target gene expression in the craniofacial complex was influenced by the action of GLI co-regulators, we next examined the proximity of GBRs to motifs for potential GLI co-regulators.

GBRs have previously been identified in close proximity to binding motifs for other transcription factors, including members of the bHLH, SP, and Sox families (Vokes et al., 2008; Peterson et al., 2012; Aberger and Ruiz I Altaba, 2014). Furthermore, the co-occupancy and cooperativity of GLI with SOX transcription factors was previously shown to be essential for activating neural gene expression signatures (Peterson et al., 2012). To determine if expression of our identified GLI targets could be influenced by the presence or absence of GLI co-regulators, we first examined the genomic sequence around the GBRs of our target genes for E-box, SP, and SOX binding motifs. (Figure 5, Supplemental Figure 3). We defined the area < 1 kb upstream of the transcription start site (TSS) as the promoter region, < 20 kb away from the TSS as proximal upstream or downstream, and < 100 kb away from the TSS as distal upstream or downstream. We analyzed the sequence surrounding GBRs in our four selected GLI target genes for sequences predictive of E-box, SP, and SOX binding. All four of our identified GLI target genes contained at least one motif cluster containing a GBR, E-box, SP, and SOX binding site within 1 kb of each other (Figures 5A–D). Several other clusters containing three of the four motifs were also identified (Figures 5A–C; red box). The close proximity of these binding motifs suggested that expression of GLI targets in the developing craniofacial complex could be influenced by the cooperative function of GLI isoforms and co-regulator proteins.

Several transcription factors synergistically cooperate with GLI proteins to influence GLI target gene expression (Aberger and Ruiz I Altaba, 2014). Their co-occupancy at the promoter of GLI targets is required for the optimal activation/repressor. ChIP-based, high-throughput analyses uncovered several transcription factor motifs located close to GBRs in the cis-regulatory modules of GLI targets. Specifically, binding sequences for Sox (Peterson et al., 2012), bHLH (Lee et al., 2010), and SP proteins (Vokes et al., 2008) have been shown to exist in very close proximity to GBRs. Our in silico analyses confirmed these sequences exist in near GBRs in four previously identified GLI-targets. We hypothesized that differential expression of these co-regulators could contribute to the differential gene expression of GLI targets in both control and ta² mutant embryos. We first investigated the expression of genes that could bind to motifs found in close proximity to GBRs, specifically SOX8, SP3, and HAND2 (Figures 6A–C). SOX8 expression was significantly reduced in the FNP and MXP of ta² mutants, yet was not significantly altered in the MNP (Figure 6A). SP3 expression was significantly increased in ta² MNP, yet not changed in the FNP and MXP (Figure 6B). HAND2 expression was reduced in the MXP, but not significantly changed between control and ta² FNPs and MNPs (Figure 6C). Thus, differential expression of GLI co-regulators could possibly contribute to altered target gene expression in ta² mutants and explain why changes in GLI_FL and GLI_T isoforms do not uniformly or directly correlate with changes in target gene expression.

Our previous analyses indicate that GLI target gene expression changed in a prominence specific manner (Chang et al., 2014). We wondered if differential expression of potential GLI co-regulators could contribute to this prominence-specific expression changes. SOX8 was robustly expressed within the FNP, with levels significantly higher than those in the MXP or MNP (Figure 6A'). SP3 was more robustly expressed in the MXP and MNP, relative to the FNP (Figure 6B'). HAND2 was robustly and exclusively expressed in the MNP (Figure 6C'). Thus, each of these potential co-regulators has a prominences specific expression pattern that could differentially influence GLI target gene expression. Collectively, the proximity of binding motifs, coupled with the differential expression of these co-regulators could possibly contribute to altered expression of GLI targets in ta² embryos.

Work in other species has identified GBRs and examined the role of GLI proteins as transcription factors that affect gene expression in numerous signaling pathways (Vokes et al., 2007, 2008). Three models currently exist to explain GLI_FL/GLI_T function (Falkenstein and Vokes, 2014). First the GLI_FL::GLI_T ratio sensing model suggest that the relative levels of GLI_FL to GLI_T is integrated and results in graded levels of transcription of targets. This model suggests that response to changes in ratio, rather than concentration, affect target gene expression. The threshold activation model suggests that, rather than ratio of GLI_FL::GLI_T, a threshold-specific concentration of GLI_FL activates target gene expression (Oosterveen et al., 2012). The threshold repression model depends on the removal of the GLI_T repressor from target genes, allowing for transcriptional activation by other transcription factors to initiate gene expression. From our results, mRNA expression of most target genes was not dramatically altered despite excessive GLI_FL production in ta² mutant, indicating that the ratio sensing model cannot explain our observation in ta² mutant. Secondly, the accumulated GLI3_FL in the nucleus was expected to bind and activate target gene in the threshold activation model; however, instead we found reduced occupancy of GLI3 at target genes, along with lack of detectable GLI_FL binding to GBRs (Figures 3, 4), suggesting that a dysfunctional GLI_FL that fails to promote GLI_A-mediated activation is produced under ciliopathic conditions. Thus, these data do not support the threshold activation model for GLI_FL function. Conversely, our results showed that reduced GLI_T production could lead to a de-repression of target genes and potentially a gene activation, when the required transcription factors are available. The threshold repression model seems to be more applicable to the observation owing to less GLI3 occupancy and the failure of GLI_FL to GLI_T conversion. A more in depth understanding of GLI mechanisms of action will require tools with the ability to definitively decipher between GLI_FL and GLI_T in vivo.

A number of disorders identified as ciliopathies have craniofacial abnormalities including Oral-facial-digital syndrome, Joubert syndrome, Bardet-Biedl syndrome, Meckel-Gruber syndrome, Ellis-van Creveld syndrome (Zaghloul and Brugmann, 2011). The phenotypes for syndromes such as these, while not identical, do have several phenotypes indicative of aberrant SHH signaling including widening of the midface, cleft lip/palate, micrognathia, craniosynostosis and oral/dental anomalies (Zaghloul and Brugmann, 2011). Increased production of GLI_FL isoforms has been observed in several ciliary mutants (Huangfu and Anderson, 2005; Liu et al., 2005; Humke et al., 2010), thus a common interpretation of these data is that ciliopathic phenotypes were due to either increased activator function or skewed GLI_FL::GLI_T ratio in favor of GLI_FL. Despite these common interpretation regarding the molecular mechanism causing ciliopathic phenotypes, examination of subsequent levels of GLI processing, binding, and transcriptional function in ciliary mutants has not been performed.

Our studies are among the first to examine how and if GLI proteins function in the developing craniofacial complex of ciliary mutants; however, some questions remain regarding the mechanistic reasons as to why GLI_FL appears not to function in ciliary mutants. Prior to processing, GLI proteins associate with Suppressor of Fused (SUFU), a conserved protein known to regulate the activity of GLI proteins via modulating GLI processing, stabilization and subcellular localization (Barnfield et al., 2005; Humke et al., 2010; Tukachinsky et al., 2010; Wang et al., 2010). In the presence of a SHH signal, the SUFU-GLI_FL complex traffics through the cilium (Eggenschwiler and Anderson, 2007). Activated ciliary SMO then works through KIF7 to promote the dissociation of the inhibitory SUFU-GLI_FL complex (Humke et al., 2010; Tukachinsky et al., 2010; Li et al., 2012). Free GLI_FL is then processed into an activator and moves to the nucleus to activate downstream targets. In the absence of the SHH ligand, SMO is not translocated into the cilium and thus cannot antagonize SUFU. SUFU remains in complex with GLI_FL, GLI_FL is proteolytically processed into GLI_T and the SUFU-GLI_T complex moves to the nucleus where it recruits the Sap18-Sin3 co-repressor complex to repress GLI target genes (Ding et al., 1999; Kogerman et al., 1999; Cheng and Bishop, 2002; Paces-Fessy et al., 2004). Furthermore, our recent work with the murine ciliary mutant, Kif3^fl/fl;Wnt1-Cre shows increased SUFU production and nuclear localization, as well as enhanced association of SUFU with GLI3 (Chang et al., in press). We hypothesize a similar mechanism is at play in ta² mutants. Specifically, excessive GLI_FL produced in ta² mutants cannot dissociate from SUFU because the complex cannot undergo ciliary trafficking. We further hypothesize that lack of dissociation prevents GLI_FL from occupying GBRs and directly activating target gene transcription. Our future studies will address the association of GLI_FL and SUFU in ta² mutants and determine if their maintained association contributes to ciliopathic phenotypes.

Frequently the craniofacial complex is thought of as a singular organ system; however, the prominences that make up the face develop independently prior to their fusion. There is evidence to support that these prominences have distinct molecular profiles and develop as separate developmental fields (Brugmann et al., 2007). Our ChIP-qPCR results detected differential GLI binding to target genes in facial prominences of control embryos (Figures 2B–D). Further, the expression of co-regulators also followed a prominence specific pattern in both control and ta² embryos (Figures 6A'–C'). These data supported the hypothesis that during normal development, as well as when cilia are lost, each facial prominence uses a unique mechanism to transduce a SHH-dependent GLI signal. Based on our examination of co-regulators, we hypothesize that there is a combinatorial code of GLI isoforms and co-regulators that work together to precisely regulate target gene expression. Thus, when cilia are lost and GLI production is altered, target gene expression is dictated by how the combinatorial code of remaining GLI isoforms and co-regulators function together.

In sum, our data suggest that the increased or decreased production of GLI isoforms alone is not sufficient to explain how target gene expression will be altered. To understand the molecular mechanisms responsible for ciliopathic, GLI-mediated phenotypes, future studies will have to account for tissue specificity, the presence or absence of co-regulators and the mode of GLI function (activation or de-repression) to begin to address this process.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.