The ciliary protein C2cd3 is required for mandibular musculoskeletal tissue patterning

Abstract

The mandible is composed of several musculoskeletal tissues including bone, cartilage, and tendon that require precise patterning to ensure structural and functional integrity. Interestingly, most of these tissues are derived from one multipotent cell population called cranial neural crest cells (CNCCs). How CNCCs are properly instructed to differentiate into various tissue types remains nebulous. To better understand the mechanisms necessary for the patterning of mandibular musculoskeletal tissues we utilized the avian mutant talpid² (ta²) which presents with several malformations of the facial skeleton including dysplastic tendons, mispatterned musculature, and bilateral ectopic cartilaginous processes extending off Meckel’s cartilage. We found an ectopic epithelial BMP signaling domain in the ta² mandibular prominence (MNP) that correlated with the subsequent expansion of SOX9+ cartilage precursors. These findings were validated with conditional murine models suggesting an evolutionarily conserved mechanism for CNCC-derived musculoskeletal patterning. Collectively, these data support a model in which cilia are required to define epithelial signal centers essential for proper musculoskeletal patterning of CNCC-derived mesenchyme.

Introduction

The mandible is an anatomically complex structure composed of numerous tissue types, including cartilage, bone, muscle, tendons, ligaments, nerves, and vasculature, that allows for essential behaviors such as communication and mastication (Klingenberg and Navarro, 2012). Each tissue component of the mandible must be temporally and spatially patterned, and differentiated in concert, to ensure structural and functional integrity. Disruptions and/or variations in mandibular development can result in several musculoskeletal conditions and negatively impact patients’ ability to speak, breath and chew.

Development of the mandible begins when multipotent cranial neural crest cells (CNCCs) are generated in the dorsal neural tube. CNCCs delaminate from the neural tube and subsequently migrate into the mandibular prominence (MNP) (Couly et al., 1998; Köntges and Lumsden, 1996). As CNCCs migrate into the MNP, they are exposed to numerous signals from the surrounding epithelium including Sonic hedgehog (Shh) from the pharyngeal endoderm and oral ectoderm, Bone morphogenetic protein 4 (Bmp4) from the distal aboral ectoderm, and Fibroblast growth factor 8 (Fgf8) from the proximal aboral ectoderm (Bitgood and McMahon, 1995; Tucker et al., 1999; Vainio et al., 1993; Xu et al., 2019). These instructive signals regulate the expression of numerous downstream transcription factors that ultimately control the differentiation of CNCCs into numerous musculoskeletal tissues (e.g., bone, cartilage, tendon). Shh is necessary for the survival of post-migratory CNCCs and the subsequent differentiation of chondrocytes (Billmyre and Klingensmith, 2015). CNCCs that receive a Bmp signal express SRY-box transcription factor 9 (Sox9) and subsequently differentiate into chondrocytes to form Meckel’s cartilage, the structural scaffold for the developing mandible (Bi et al., 1999; Hu et al., 2008; Svandova et al., 2020). CNCCs that receive an Fgf signal will express the transcription factor Scleraxis (Scx) and will differentiate into tenocytes and ligamentocytes (Bobzin et al., 2021; Edom-Vovard et al., 2002; Havis et al., 2016). Despite having knowledge of the initial signals that prompt CNCC differentiation into numerous musculoskeletal lineages, a complete understanding of how these signals integrate to generate a functional mandible, and how disruptions in these signaling pathways cause syndromes affecting mandibular development, is incomplete.

Mandibular development is frequently disrupted in a class of diseases called ciliopathies, which arise from disruptions in the structure and function of the primary cilium - a microtubule-based cellular extension that functions as a centralized hub for molecular signal transduction (Elliott and Brugmann, 2019; Goetz and Anderson, 2010). Ciliopathies affect as many as 1 in 800 human patients and are characterized by numerous pleiotropic phenotypes including skeletal dysplasias and craniofacial anomalies (Baker and Beales, 2009; Reiter and Leroux, 2017). Approximately 30% of all ciliopathies can be classified by their craniofacial phenotypes, including cleft lip/palate, craniosynostosis, and micrognathia (Schock and Brugmann, 2017). Human patients and animal models that present with craniofacial ciliopathies have an array of mandibular phenotypes such as micrognathia, dysmorphic skeletal elements, and reduced bone density (Adel Al-Lami et al., 2016; Bonatto Paese et al., 2021a; Cela et al., 2018; Ferrante et al., 2009; Kantaputra et al., 2023; Kitamura et al., 2020; Kolpakova-Hart et al., 2007; Zhang et al., 2011). While significant progress has been made in understanding the molecular etiologies of craniofacial ciliopathies, comprehensive knowledge of how individual ciliary proteins regulate the spatial patterns of early molecular signals that contribute to musculoskeletal tissue patterning is currently lacking.

Oral-facial-digital syndrome (OFD) is an umbrella term for at least 20 distinctive ciliopathic disorders characterized by anomalies in the oral cavity, facial structures, and digits (Bruel et al., 2023; Franco and Thauvin-Robinet, 2016; Iturrate et al., 2022). Several types of OFD present with mandibular phenotypes including micrognathia, retrognathia, hypo- or aglossia, hamartomas, and ankyloglossia. OFD14 is caused by mutations in C2 domain containing 3 centriole elongation regulator (C2CD3), which is localized to the distal centrioles and acts as a critical elongation factor during ciliogenesis (Boczek et al., 2018; Cortés et al., 2016; Hoover et al., 2008; Thauvin-Robinet et al., 2014; Ye et al., 2014). To study the molecular etiologies of phenotypes associated with human OFD14, we utilized the talpid² (ta²). The ta² is a naturally occurring avian mutant that is characterized by its numerous craniofacial phenotypes as well as its polydactylous limbs (Abbott et al., 1960, 1959; Bonatto Paese et al., 2022, 2021a; Brugmann et al., 2010; Caruccio et al., 1999; Chang et al., 2014; Harris et al., 2006; Schneider et al., 1999; Schock et al., 2015). We previously reported that the ta² phenotype was the result of a 19 bp deletion in C2CD3 (Chang et al., 2014). Interestingly, this C2CD3 variant faithfully phenocopies human OFD14 (Bonatto Paese et al., 2021a; Chang et al., 2014; Harris et al., 2006; Schock et al., 2015), presenting with strikingly similar craniofacial, neural, and limb phenotypes, making the ta² an excellent model for understanding the molecular etiology for several craniofacial and musculoskeletal phenotypes.

Herein, we characterized previously unreported ectopic cartilaginous processes extending off Meckel’s cartilage in ta² embryos. The generation of these structures correlated with an ectopic, epithelial BMP signaling center and increased expression of BMP target genes. Together, these results suggest a novel cellular and molecular mechanism for musculoskeletal craniofacial phenotypes present in OFD14.

Materials and Methods

Avian embryo collection, processing, and genotyping

Fertilized control^+/+ and ta² eggs were supplied by the University of California, Davis Avian Facility. Eggs were incubated at 38.8°C for 2 – 13 days in a rocking incubator with humidity control. Embryos were staged according to the Hamburger-Hamilton (HH) staging system (Hamburger and Hamilton, 1951), harvested in cold diethyl pyrocarbonate-treated phosphate-buffered solution (DEPC PBS), and fixed in either 10% neutral buffered formalin (NBF) overnight at room temperature or 4% paraformaldehyde (PFA) in DEPC PBS overnight at 4°C. Embryos were genotyped as previously described (Chang et al., 2014).

Mouse strains

All mouse strains used in this study have been previously described: AP2-Cre (Macatee et al., 2003), C2cd3^ex4-5flox (Chang et al., 2021), Crect (Reid et al., 2011), Scx-Cre (Blitz et al., 2009), and Wnt1-Cre2 (Lewis et al., 2013). Both male and female mice were used. A maximum of 4 adult mice were housed per cage and breeding cages housed one male paired with up to two females. All mouse usage was approved by the Institutional Animal Care and Use Committee (IACUC) and monitored daily by the Division of Veterinary Services at Cincinnati Children’s Hospital Medical Center.

Mouse embryo collection and genotyping

Timed matings were performed, with noon of the day a vaginal plug was discovered designated as E0.5. Embryos were harvested via Caesarian section after isoflurane-induced asphyxiation and cervical dislocation and were dissected in cold DEPC PBS. Samples were fixed in 10% NBF overnight at room temperature or 4% PFA in DEPC PBS overnight at 4°C. For all experiments, Cre-negative littermates were used as controls. Genotyping was performed using the primers in Table 1.

Table 1.

Genotyping Primers

Primer Set Name	Forward Sequence	Reverse Sequence	Source
C2CD3 ta 2	GAGGAAAGCAGTGAGGTTCTAAA	CTTCCAGTGGGTCTTTGTGG	CF Chang, et al., 2014.
C2cd3 ex4-5flox	CGTGTGAAGGCAGCTCTACT	TAACTCACGTGGGTCACAGA	CF Chang, et al., 2021.
Generic Cre	TGCTCTGTCCGTTTGCCG	ACTGTGTCCAGACCAGGC	KH Elliott, et al., 2020.

Wholemount skeletal staining and morphometrics

HH30-32 avian embryos were stained with .05% Alcian blue solution to detect cartilage and cleared in 1:2 benzyl alcohol:benzyl benzoate solution as previously described (Nagy et al., 2009). HH34-39 avian embryos and e15.5-e17.5 mouse embryos were first immersed in hot water to facilitate the manual removal of skin and soft tissues and were stained with .015% Alcian blue solution to detect cartilage. e17.5 mouse embryos were further stained for bone with 0.005% Alizarin red in 1% KOH solution. Samples were then cleared in graded 1% KOH/glycerol solutions and imaged in 50% glycerol as previously described (Elliott et al., 2020). All samples were imaged using a Leica M165 FC stereo microscope system. Area measurements were performed by tracing the outline of Meckel’s cartilage on HH30 and HH36 avian embryos via the freehand tool in FIJI (Schindelin et al., 2012). To approximate the location of the ectopic cartilaginous process along Meckel’s cartilage in mutant embryos, length measurements of HH36 ta² and e15.5 C2cd3^ex4-5fl/fl; AP2-Cre Meckel’s cartilages from the mandibular symphysis to the articular cartilage and from the mandibular symphysis to the ectopic cartilaginous process were performed using FIJI. A ratio of the ectopic cartilaginous process measurement to the total Meckel’s cartilage length measurement was calculated and the resulting ratio was binned into anterior, intermediate anterior, intermediate posterior, and posterior classifications of equal size based on previous reports (Fig. S1C-D) (Pitirri et al., 2022; Svandova et al., 2020).

Wholemount hybridization chain reaction

The Hybridization Chain Reaction (HCR) assay from Molecular Instruments, Inc. was carried out as previously described (Choi et al., 2018, 2016). HCR on HH14-20 avian MNPs and e10.5 mouse MNPs was performed using the manufacturer’s instructions for chicken and mouse embryos, respectively. For HCR assays performed on HH32 avian mandibular prominences, all PBS with 0.1% Tween (PBST) and 5× SSC with 0.1% Tween (SSCT) washes were increased to 10 min, and proteinase K treatment was increased to 20 μg/mL for 15 min. For HCR assays performed on HH36 avian mandibles, several modifications were made. Samples were quickly blanched in a 55°C water bath shortly after fixation to facilitate the dissection of excess keratinized epithelium to enhance reagent penetration, proteinase K solution concentration was increased to 50 μg/mL, and all PBST and 5× SSCT washes were increased to 10 min. All probes used in this study were designed by Molecular Instruments and can be found in Table 2. All samples were counterstained with 1:10000 dilution of Hoescht 33342 (Invitrogen H3570) in 5× SSCT for 5 min at room temperature then washed five times in 5× SSCT. Samples were then cleared in multiple changes of refractive index matching solution (RIMS) (Muntifering et al., 2018) at room temperature on a nutator until samples were transparent. Samples were mounted in RIMS and imaged on either a Nikon FN1 upright confocal microscope system or a Nikon A1 inverted confocal microscope system. Images were denoised in Nikon NIS-Elements software and then loaded into Imaris 10.0.0 software for analysis and visualization. Three-dimensional renderings of avian and murine triplex SHH/BMP4/FGF8 and avian duplex BMP4/SOX9 expression data were generated using Imaris 10.0.0 software via the Surfaces Tool. Surface areas of SHH, BMP4, FGF8, and the total MNP were calculated using the Imaris Surfaces Tool.

Table 2.

HCR Probes

Probe Name	Species	Channel	GenBank Identifier	Lot Number
BMP4	Gallus gallus (chicken)	B3	NM_205237.4	PRO952
COL2A1	Gallus gallus (chicken)	B5	NM_204426.2	RTD418
FGF8	Gallus gallus (chicken)	B1	NM_001012767.2	PRK963
MYOG	Gallus gallus (chicken)	B4	NM_204184.2	RTG675
SCX	Gallus gallus (chicken)	B1	NM_204253.2	PRO152
SHH	Gallus gallus (chicken)	B2	NM_204821.1	RTE050
Bmp4	Mus musculus (mouse)	B3	NM_007554.3	RTC282
Fgf8	Mus musculus (mouse)	B2	NM_010205	RTC512
Shh	Mus musculus (mouse)	B1	NM_009170.3	RTC686

Paraffin embedding and sectioning

Samples were dehydrated in a graded ethanol series, washed in xylene, embedded in paraffin, and sectioned at 6 – 8 μm thickness using a Sakura Accu-Cut SRM 200 microtome. Sections were dried overnight at room temperature.

RNAscope in situ hybridization

The RNAscope^® assay was carried out as previously described (Bonatto Paese et al., 2021a; Wang et al., 2012). Briefly, transcripts were detected using the RNAscope Multiplex Fluorescent V2 kit per manufacturer’s instructions. All probes to detect transcripts used in this study were designed by Advanced Cell Diagnostics and can be found in Table 3. Signals for the transcripts were detected using fluorescein (Akoya Biosciences NEL741001KT) and Cyanine 3 (Akoya Biosciences NEL744001KT) diluted 1:500 in RNAscope Multiplex TSA Buffer. All samples were counterstained with RNAscope DAPI (4’,6-diamidino-2-phenylindole) solution and imaged on a Leica DM5000B upright microscope system or a Nikon A1 inverted confocal microscope system.

Table 3.

RNAscope Probes

Probe Name	Species	Channel	Catalog Number
Gg-BMP4	Gallus gallus (chicken)	C1, C2	558411, 558411 – C2
Gg-COL1A1	Gallus gallus (chicken)	C2	571551 - C2
Gg-FGF8	Gallus gallus (chicken)	C1	868491
Gg-FGFR2	Gallus gallus (chicken)	C3	1106541 - C3
Gg-LHX6	Gallus gallus (chicken)	C3	128220 - C3
Gg-MSX1	Gallus gallus (chicken)	C2	480091 - C2
Gg-PTCH1	Gallus gallus (chicken)	C2	551571 - C2
Gg-SHH	Gallus gallus (chicken)	C1	551581
Gg-SOX9	Gallus gallus (chicken)	C1	1075381 - C1
Mm-Bmp4	Mus musculus (mouse)	C1	401301
Mm-Fgf8	Mus musculus (mouse)	C2	313411 - C2
Mm-Lhx6	Mus musculus (mouse)	C1	422791
Mm-Msx1	Mus musculus (mouse)	C3	421841 - C3
Mm-Ptch1	Mus musculus (mouse)	C1	402811
Mm-Shh	Mus musculus (mouse)	C2	314361 - C2

Avian embryo paralysis

Decamethonium bromide (DBr) (Sigma-Aldrich D1260) was diluted to 10 mg/ml in Hank’s Buffered Sterile Saline (HBSS) and filter sterilized using a 0.22 μm filter. HH29 embryos were treated with 0.5 mL DBr via injection into the albumin with a syringe through a pinhole made in the eggshell as previously described (Hall, 1975; Woronowicz et al., 2018). Embryos were assayed for paralysis via hindlimb extension before collecting for Alcian blue staining.

Single-cell RNA-sequencing

Three MNPs per group from HH26 control and talpid² embryos were quickly dissected in ice-cold DEPC PBS. Any regions rich in blood were manually removed. The MNPs were then minced to a fine paste. Cells were dissociated into a single-cell suspension and sequenced using NovaSeq 6000 and the S2 flow cell. 12.5 mg of tissue was placed in a sterile 1.5 mL tube containing 0.5 mL protease solution containing 125 U/mL DNase and 3 mg/mL Bacillus Licheniformis. The samples were incubated on ice for a total of 10 min, with gentile trituration using a wide boar pipette tip every minute after the first two. Protease was inactivated using ice-cold DEPC PBS containing 0.02% bovine serum albumin and filtered using 30 μM filter. The cells were pelleted by centrifugation at 200G for 4 min at room temperature and resuspended in 50 μL Dulbecco’s Modified Eagle Medium with 10% Fetal Bovine Serum. Cell number and viability were assessed using a hemocytometer and trypan blue staining. 9,600 cells were loaded onto a well on a 10x Chromium Single-Cell instrument (10X Genomics) to target sequencing of 6,000 cells. Barcoding, cDNA amplification, and library construction were performed using the Chromium Single-Cell 3’ Library and Gel Bead Kit v3. Post cDNA amplification and cleanup was performed using SPRI select reagent (Beckman Coulter B23318). Post cDNA amplification and post-library construction quality control was performed using the Agilent Bioanalyzer High Sensitivity kit (Agilent 5067–4626). Libraries were sequenced using a NovaSeq 6000 and the S2 flow cell. Sequencing parameters used were: Read 1, 28 cycles; Index i7, eight cycles; Read 2, 91 cycles, producing about 300 million reads. Sequenced reads were mapped to the Ensembl build of the chicken genome GRCg6a using CellRanger (http://10xgenomics.com) to obtain a gene-cell data matrix.

Analysis of single-cell RNA-sequencing data

Analysis of the data was conducted using the Seurat v4.4.0 package (Hao et al., 2021). Control and ta² cells were filtered for quality by unique feature counts (less than 200 or greater than 4000) and mitochondrial gene expression (greater than 15%) to remove any low-quality cells. Filtered cells were then normalized and cell cycle genes were removed. “RunPCA,” “RunUMAP,” and “FindNeighbors” were then performed using 20 dimensions along with “FindClusters” with a resolution of 1 to generate the working R objects. The control and ta² datasets were then integrated by “FindIntegrationAnchors” and “IntegrateData” and the same filtering criteria were applied to generate the working integrated R object. Dot plots for BMP family member expression were generated using the “DotPlot” function on the working integrated R object in Seurat.

CellChat analysis

The standard CellChat workflow was used to assay signaling interactions as previously described (Jin et al., 2021). In brief, the individual Seurat objects were used as inputs for CellChat. These objects were processed by the standard parameters using the human CellChat database as the reference. Once the top pathways list was generated, the functions “plotGeneExpression”, “netAnalysis_computeCentrality”, “netAnalysis_signalingRole_network”, and “netVisual_aggregate” were used to generate heatmaps and circle plots of the pathways of interest.

Section histology

Safranin-O staining was performed on frontal sections from NBF-fixed e13.5 mouse heads as previously described (Chang et al., 2016).

Micro-computed tomography of murine embryo heads

Heads from e17.5 C2cd3^ex4-5fl/fl and C2cd3^ex4-5fl/fl; AP2-Cre embryos were harvested in cold DEPC PBS and fixed overnight in 4% PFA at 4°C. Microcomputed tomography (microCT) was performed using a Siemens Inveon PET/SPECT/CT scanner in the Preclincal Imaging Core of the University of Cincinnati Vontz Center for Molecular Studies. The cone-beam CT parameters were as follows: 360° rotation, 1080 projections, 1300 ms exposure time, 1500 ms settle time, 80 kVp voltage, 500 μA current, and effective pixel size 17.67 μm. Acquisitions were reconstructed using a Feldkamp algorithm with slight noise reduction, 3D matrix size 1024x1024x1536, using manufacturer-provided software. Protocol-specific Hounsfield Unit (HU) calibration factor was applied. DICOM files were loaded into 3D Slicer 5.6.0 software (Fedorov et al., 2012). Three-dimensional renderings of the head were constructed, surface smoothing was executed, and the mandible was segmented via the Segment Editor function. Mandibular length measurements were taken from the mandibular symphysis to the condylar process as previously described (Roberts et al., 2019) on both sides of the mandible and averaged. Ectopic mandibular skeletal elements were manually pseudocolored in Adobe Illustrator.

Statistical methods

Unpaired two-tailed, Student’s t-test were used in all statistical analyses performed at the 5% significance level. P-values less than .05 (p < .05) were considered statistically significant.

Results

ta² embryos presented with ectopic, bilateral cartilaginous processes extending off Meckel’s cartilage

Previous reports from our lab and others have documented the craniofacial and skeletal phenotypes present in ta² embryos (Abbott et al., 1960, 1959; Bonatto Paese et al., 2022, 2021b, 2021a; Brooks et al., 2021; Brugmann et al., 2010; Caruccio et al., 1999; Chang et al., 2014; Dvorak and Fallon, 1991; Harris et al., 2006; Schneider et al., 1999; Schock et al., 2015). To better understand the molecular etiology of these phenotypes, we performed a temporal analysis of mandibular development, initially focused on Meckel’s cartilage. Alcian blue staining and measurements of HH30 control and ta² Meckel’s cartilage revealed no discernable phenotypic or size difference (Fig. 1A-B, Fig. S1A). Between HH32-34, however, ectopic, bilateral cartilaginous processes were consistently observed extending from the more medial anterior aspect of Meckel’s cartilage (Fig. 1C-F). By HH36, these ectopic processes were prominent (Fig. 1G-H) and contributed to an increase in the total area of Meckel’s cartilage (Fig. S1B). The ectopic processes were consistently observed with a conserved shape and location in the intermediate anterior portion of Meckel’s cartilage until embryonic lethality at HH39 (Fig. 1I-J; Fig. S1C-D).

Figure 1. ta² mandibles present with numerous musculoskeletal anomalies reminiscent of a developing enthesis.

(A-J) Dorsal views of Alcian blue-stained HH30 (A-B), HH32 (C-D), HH34 (E-F), HH36 (G-H), and HH39 (I-J) control^+/+ (A, C, E, G, I) and ta² (B, D, F, H, J) Meckel’s cartilages. (K-N’) Ventral views of wholemount HCR assays for COL2A1 (red) and SCX (white) (K-L’) or COL2A1 (red) and MYOG (yellow) (M-N’) in HH36 control^+/+ (K-K’, M-M’) and ta² (L-L’, N-N’) mandibles (n = 3 per group). (O-P’) Section RNAscope in situ hybridization assays for SOX9 (red) and COL1A1 (green) in frontal sections of HH29 control^+/+ (O-O’) and ta² (P-P’) mandibular prominences (n = 3 per group). Dotted lines in O’ and P’ denote the outline of Meckel’s cartilage as denoted by SOX9 expression. (Q) Quantification of the number of SOX9+/COL1A1+ cells as a percentage of total SOX9+ cells in control^+/+ and ta² mandibular prominences. Asterisk denotes p < 0.05. Arrows in D, F, H, and J denote ta² ectopic cartilaginous process. Dotted boxes in K-P denote areas of higher magnification. crt: central raphe tendon, ECP: ectopic cartilaginous process, im: intermandibular muscle, it: intermandibular tendon, MC: Meckel’s cartilage. Scale bars: 1mm (A-K, L, M, N); 500μm (K’, L’, O, P); 300μm (M’, N’); 50μm (O’, P’).

The presence of these ectopic cartilaginous processes was surprising, as the ta² has been studied for nearly 7 decades without mention of such structures. As such, we searched the literature to see if similar structures had previously been reported. Interestingly, Runx2-deficient mice were reported to have two ectopic cartilaginous processes on Meckel’s cartilage which attached to the digastric and mylohyoid muscles (Shibata et al., 2004). These data, coupled with our previous findings that mandibular skeletal elements were hypoplastic in the ta² embryos (Bonatto Paese et al., 2021a), prompted the examination of other musculoskeletal tissues. Using wholemount hybridization chain reaction (HCR; Choi et al., 2018, 2016), we examined the organization of tendons, muscle, and cartilage in HH36 embryos via the expression of SCLERAXIS (SCX) (Cserjesi et al., 1995; Schweitzer et al., 2001), MYOGENIN (MYOG) (Wright et al., 1989) and COLLAGEN TYPE II ALPHA 1 CHAIN (COL2A1) (Kosher et al., 1986), respectively. In control embryos, the intermandibular tendons ran parallel to, and were immediately adjacent to Meckel’s cartilage (Fig. 1K-K’); however, in ta² mandibles, they appeared to insert directly into the ectopic cartilaginous processes (Fig. 1L-L’). Similarly, the intermandibular muscle, which spans the entire length of Meckel’s cartilage (Fig. 1M-M’; Robson, 1993), was disorganized, hypoplastic, and also appeared to insert directly into the ectopic cartilaginous processes (Fig. 1N-N’). Thus, in addition to the presence of ectopic cartilaginous processes, the ta² mandible presented with a musculoskeletal disorganization similar to other mutations that impacted skeletal development of CNCCs.

The appearance and organization of ectopic cartilaginous processes and accompanying ectopic musculoskeletal attachments in the ta² mandible phenotypically resembled a secondary cartilage, or a cartilage that is independent from the primary cartilaginous skeleton and arises on existing dermal bones at sites of articulations and insertions (Hall, 2015; Murray, 1963). Secondary cartilages can be induced through mechanical forces and express both FIBROBLAST GROWTH FACTOR RECEPTOR 2 (FGFR2) and BMP4 (de Beer and Barrington, 1934; Solem et al., 2011). Given the observation that the intermandibular muscle appeared to be inserted into the ectopic cartilaginous processes, we hypothesized that they could be secondary cartilages caused by muscle contraction. To test this hypothesis, ta² embryos were paralyzed via decamethonium bromide (DBr) treatment and characterized molecularly. Neither the presentation nor size of the ectopic cartilaginous processes was alleviated via paralysis (Fig. S2A-D). Furthermore, the processes did not express BMP4 or FGFR2 (Fig. S2E-F’). Thus, these results suggested that ectopic cartilaginous processes observed in ta² embryos were not secondary cartilages.

Given that our molecular analysis did not support the identity of the ectopic cartilaginous processes as secondary cartilages and that the intermandibular muscle was attached to these structures, we next tested the hypothesis that they were precursors of entheses, or tendon-bone interfaces. In the stages preceding enthesis specification and mineralization, the musculature attaches to the cartilaginous template of the developing skeleton (Schweitzer et al., 2010; Sunadome et al., 2023). Molecularly, an enthesis develops from a cell population that co-expresses factors associated with cartilage (SOX9) and tendon (SCX, COL1A1) development (Blitz et al., 2013; Bobzin et al., 2021; Sugimoto et al., 2013). Utilizing RNAscope for SOX9 and COL1A1 in HH29 control mandibular prominences (MNPs), we found distinct areas of either SOX9+ or COL1A1+ cells that did not intermix (Fig. 1O-O’). Conversely, in ta² embryos, the domain of SOX9+ cells expanded into the COL1A1+ area resulting in a significant increase in the number of SOX9+/COL1A1+ cells (Fig. 1P-Q). Thus, taken together, these analyses suggested that the ectopic cartilaginous processes present on Meckel’s cartilage molecularly and anatomically resembled an enthesis during mandibular musculoskeletal development. Considering these findings, we next sought to understand the molecular etiology of these ectopic processes in the ta².

Ectopic cartilage formation in ta² mandibles correlated with expanded BMP signaling activity

The generation of the ectopic cartilaginous processes in the ta² mandible prompted us to investigate the mechanisms that induce CNCCs to differentiate into various mandibular musculoskeletal tissues. As CNCCs migrate into the MNP, they experience signals from compartmentalized epithelial domains including BMP4 in the aboral distal ectoderm, FGF8 in the aboral proximal ectoderm, and SHH in the pharyngeal endoderm and oral ectoderm (Jeong et al., 2004; Shigetani et al., 2000; Tucker et al., 1999). These factors antagonize each other during early mandibular development to ensure proper mesenchymal gene expression and subsequent musculoskeletal tissue differentiation (Haworth et al., 2007, 2004; Neubüser et al., 1997; Tucker et al., 1998; Xu et al., 2019). Previous studies demonstrated that the combinatorial action of FGF8 and SHH was required to drive cartilaginous outgrowth in the developing craniofacial complex (Abzhanov and Tabin, 2004; Hu et al., 2003), while other studies indicated that exogenous BMP4 expression could lead to ectopic craniofacial chondrogenesis (Barlow and Francis-West, 1997; Hu et al., 2008; Nonaka et al., 1999; Semba et al., 2000). With the knowledge that primary cilia regulate signal transduction and the ta² embryo has dysregulated Hedgehog signaling in the developing craniofacial complex (Brooks et al., 2021; Brugmann et al., 2010; Chang et al., 2014), we sought to examine the relative expression patterns of BMP4, FGF8 and SHH in control and ta² MNPs.

To uncover the potential molecular basis for ectopic cartilaginous process formation in ta² embryos, we began our analysis at HH14, a developmental stage in which the MNP is being patterned and precedes the onset of gross ta² craniofacial anomalies (Abramyan and Richman, 2018). In dorsal views of HH14 control MNPs, SHH and BMP4 were expressed in complementary domains across the oral-aboral axis (Fig. 2A) (Xu et al., 2019). However, the SHH domain was diminished and the BMP4 domain was expanded into the oral aspect of the ta² MNP (Figure 2B). Sagittal views confirmed that BMP4 expression was expanded into the oropharyngeal domain of HH14 ta² MNPs while the SHH domain was diminished when compared to stage-matched controls (Figure 2C-D). Quantification of the expression domain surface area confirmed a significant decrease in the SHH domain and a significant increase in the BMP4 domain (Figure 2E). A similar expansion of BMP4 and reduction of SHH was observed in HH16 ta² MNPs compared to controls (Fig. 2F-J). We then analyzed the expression of SHH, BMP4, and FGF8 in HH20 control and ta² MNPs. While the SHH domain was significantly decreased and the BMP4 domain was significantly increased in HH20 ta² MNPs, FGF8 was maintained in the proximal epithelial domain with no change in the domain size (Fig. 2K-O).

Figure 2. Expression of BMP signaling factors is expanded in ta² MNPs.

(A-D) Three-dimensional renderings of wholemount HCR expression data for SHH (red) and BMP4 (purple) in HH14 control^+/+ (A, C, n = 5) and ta² (B, D, n = 3) mandibular prominences. (E) Quantification of three-dimensional surface renderings of SHH and BMP4 in HH14 control^+/+ and ta² MNPs as a percentage of total mandibular surface area. (F-I) Three-dimensional renderings of wholemount HCR expression for SHH (red) and BMP4 (purple) in HH16 control^+/+ (D-D’, n = 5) and ta² (E-E’, n = 5) MNPs. (J) Quantification of three-dimensional surface renderings of SHH and BMP4 in HH16 control^+/+ and ta² MNPs as a percentage of total mandibular surface area. (K-N) Three-dimensional renderings of wholemount HCR expression data for SHH (red), FGF8 (green), and BMP4 (purple) in HH20 control^+/+ (K, M, n = 4) and ta² (L, N, n = 4) MNPs. (O) Quantification of three-dimensional surface renderings of SHH, FGF8, and BMP4 in HH20 control^+/+ and ta² MNPs as a percentage of total mandibular surface area. (P-U) RNAscope in situ hybridization for SHH and PTCH1 (P-Q), BMP4 and MSX1 (R-S), and FGF8 and LHX6 (T-U) on HH20 control^+/+ (P, R, T) and ta² (Q, S, U) sagittal MNP sections. n = 3 for all RNAscope experiments. The dotted lines in C, D, H, I, M and N denote the midpoint of the oral-aboral axis of the MNP. Scale bars: 100μm (A-D, F-I, P-U), 200μm (K-N). Asterisk denotes p < 0.05, n.s. signifies no significance.

With the observation that the spatial organization of epithelial BMP4 and SHH expression was perturbed in the ta² MNP, we next examined the expression of downstream targets in the underlying mesenchyme utilizing RNAscope in situ hybridization in HH20 MNPs. In HH20 control MNPs, SHH and its downstream transcriptional target PATCHED1 (PTCH1; Goodrich et al., 1996; Marigo et al., 1996) were expressed in the oropharyngeal endoderm and adjacent mesenchyme, respectively (Fig. 2P). In stage-matched ta² MNPs, however, SHH expression was reduced and PTCH1 was diffuse within the mesenchyme (Fig. 2Q). Concordantly, BMP4 and the downstream transcriptional target MUSCLE SEGMENT HOMEOBOX 1 (MSX1; Barlow and Francis-West, 1997) were expanded into the oral mandibular epithelium and mesenchyme, respectively when compared to stage-matched controls (Fig. 2R-S). Additionally, there was not a significant change in ectodermal FGF8 expression or mesenchymal expression of its downstream transcriptional target LIM HOMEOBOX 6 (LHX6; Grigoriou et al., 1998) (Fig. 2T-U).

Following the observation that BMP4 and its downstream target MSX1 were upregulated in the early ta² MNP, we next sought to examine BMP pathway activity immediately prior to the onset of chondrogenesis. We performed scRNA-seq (FaceBase: FB00001245) on HH26 control and ta² MNPs and utilized this dataset to run CellChat analysis to quantitatively analyze signal transduction pathways and infer cellular communication networks (Jin et al., 2021). In the control MNP, one epithelial cluster (cluster 1, outlined in green in Fig. S3A) expressed BMP pathway genes, signaled to itself, as well as to numerous neural crest mesenchyme and epithelial clusters (Fig. 3A-B). In the ta² MNP, however, a second, ectopic epithelial cell population (cluster 15, outlined in red in Fig. 3A, S3A) was identified in addition to that which was found in the control MNP (Fig. 3A-B). Interestingly, many neural crest mesenchyme and epithelial cell populations in the ta² MNP had an increased role for permitting (or acting as an “influencer” of) BMP signaling pathway activity (Fig. 3A). Notably, applying CellChat analytics to the Hedgehog and FGF signaling pathways revealed no significant changes between control and ta² MNPs as only a singular epithelial cluster (cluster 3, outlined in purple in Fig. S3A) sent and received a Hedgehog signal (Fig. S3B-C) and the FGF pathway was not detectable. Cluster 15 both received a BMP signal from cluster 1 and signaled to the same epithelial and mesenchymal cell populations as cluster 1 (Fig. 3B). To confirm these observations, we examined expression of various BMP pathway components. ta² clusters 1 and 15 had increased expression of various BMP ligands and clusters 15, 5, and 8 had increased expression of two downstream transcriptional targets of BMP signaling, SMAD1 and MSX1 (Fig. 3C). Additionally, the increased expression of BMP ligands that was originally identified at HH14 and validated through single-cell analyses were maintained through HH34, a stage at which the ectopic cartilaginous processes were prominent (Fig. 3D-K’). These results suggested that ectopic cartilaginous processes in the ta² MNP correlated with and localized to domains of ectopic BMP signaling activity.

Figure 3. BMP signaling dynamics are augmented in the ta² MNP during musculoskeletal tissue differentiation.

(A-B) BMP signaling dynamics in HH26 control and ta² mandibular prominences as visualized via heatmap (A) and circle plot (B). Red boxes in panel A denote the absence and presence of BMP signaling from cluster 15. “Mediator” refers to the ability of a cell population to gatekeep information flow within a signaling pathway while “influencer” refers to the ability of a cell population to regulate a signaling pathway’s information flow. (C) Dot plot for the expression of numerous BMP ligands and effectors in HH26 control^+/+ and ta² mandibular prominence. Green dotted boxes represent changes in gene expression for the selected markers and clusters. (D-I) Three-dimensional renderings of wholemount HCR assays for SOX9 and BMP4 in HH32 control^+/+ (D, F, H) and ta² (E, G, I) mandibular prominences presented in dorsal (D-E), anterior (F-G), and sagittal (H-I) views. n = 3 for both groups. (J-K’) RNAscope in situ hybridization for SOX9 and BMP4 on sagittal sections of HH34 control^+/+ (J-J’) and ta² (K-K’) mandibles. n = 3 for both groups. ECP: ectopic cartilaginous process, FNP: frontonasal prominence, MNP: mandibular prominence. Scale bars: 1 mm (D-E), 500 μm (F-G), 750 μm (H-K); 250 μm (J’, K’).

Conditional deletion of C2cd3 in murine models histologically and molecularly recapitulated ta² skeletal phenotypes

Although our expression data and CellChat analysis suggested a role for epithelial signaling in the emergence of ectopic cartilaginous processes, the avian model system was not easily amenable to definitively testing this hypothesis. Since the ta² C2CD3 mutation was hypomorphic and affected all tissues (Chang et al., 2014), we transitioned to a murine model system and utilized three distinct Cre drivers to conditionally ablate C2cd3 in tissues that contribute to the developing mandible. First, the Wnt1-Cre2 driver recombined within CNCCs that comprise the craniofacial mesenchyme and differentiate into musculoskeletal tissues without the accompanying ectopic Wnt1 expression of the widely used Wnt1-Cre (Lewis et al., 2013). Second, the Crect driver recombined in the surface and oral ectoderm (Reid et al., 2011). Third, the AP2-Cre driver recombined in the surface and oral ectoderm, the CNCC-derived mesenchyme, and the neuroectoderm (Macatee et al., 2003). To determine the tissue-specific requirement for C2cd3 function, we utilized previously generated mice with a C2cd3^ex4-5flox allele (Chang et al., 2021) in combination with these Cre drivers (Fig. 4A-D) and assayed the resulting Meckel’s cartilage phenotype.

Figure 4. Conditional loss of C2cd3 in murine epithelium and mesenchyme recapitulates ta² craniofacial phenotypes.

(A-D) Schematization of spatial domains of Cre recombination (blue) for control (A), Wnt1-Cre2 (B), Crect (C), and AP2-Cre (D) in e11.5 murine embryo faces. (E-H’) Wholemount Alcian blue staining for cartilage of e15.5 C2cd3^ex4-5f/f (E, E’), C2cd3^ex4-5f/f; Wnt1-Cre2 (F, F’), C2cd3^ex4-5f/f; Crect (G, G’), and C2cd3^ex4-5f/f; AP2-Cre (H, H’) mandibles. Arrows in F, G, and H denote the presence of ectopic cartilaginous protuberances with images of higher magnification in F’, G’, and H’. (I) Bar chart of distribution of the number of ectopic cartilaginous protuberances observed in C2cd3 murine mutants. (J-K’) Safranin O staining on e13.5 frontal sections of C2cd3^ex4-5fl/fl (J-J’) and C2cd3^ex4-5fl/fl; AP2-Cre (K-K’) MNPs to detect cartilage with lower magnification images (J’, K’) shown in the inset. White dotted circles in panels J and K outline Safranin O-positive staining for cartilage. Dotted black boxes in J’ and K’ denote areas of high magnification shown in J and K. n = 3 for both groups. (L-M) Sagittal views of microCT-scanned E17.5 C2cd3^ex4-5fl/fl (L, n = 3) and C2cd3^ex4-5fl/fl; AP2-Cre (M, n = 3) heads with the mandible highlighted in blue. (N-O’) Dorsal views of E17.5 C2cd3^ex4-5fl/fl (N-N’) and C2cd3^ex4-5fl/fl; AP2-Cre (O-O’) mandibles from microCT scans (N, O) and Alizarin red staining (N’, O’). Purple shading in panel O and white arrowheads in O’ denote ectopic skeletal elements. (P) Morphometric analysis of mandibular bone lengths from microCT-scanned C2cd3^ex4-5fl/fl and C2cd3^ex4-5fl/fl; AP2-Cre mandibles. (Q-R’) Ventral views of E17.5 C2cd3^ex4-5fl/fl (Q-Q’) and C2cd3^ex4-5fl/fl; AP2-Cre (R-R’) palates from microCT scans (Q, R) and Alizarin red staining (Q’, R’). bo: basioccipital, bs: basisphenoid, f: frontal bone, ip: interparietal bone, mx: maxilla, ncc: neural crest cells, oe: oral epithelium, p: parietal bone, pb: palatine bone, pmx: premaxilla, pp: pterygoid process, ppm: palatine process of the maxilla, se: surface ectoderm, so: supraoccipital. Asterisk denotes p < 0.05. Scale bars: 1 mm (E, F, G, H, N’, O’, Q’, R’), 500 μm (E’, F’, G’, H’), 200 μm (J, K).

Alcian blue staining of e15.5 C2cd3^ex4-5fl/fl; Wnt1-Cre2 mandibles revealed that 48.9% (n = 3/7) of embryos presented with bilateral, ectopic processes on Meckel’s cartilage that resembled those present in the ta² (Fig. 4E-F’, I). Furthermore, 28.6% (n = 2/7) embryos presented with one, unilateral ectopic process and 28.6% (n = 2/7) embryos did not present with any ectopic process (Fig. 4I, Fig. S4A-B). C2cd3^ex4-5fl/fl; Crect embryos presented with bilateral and unilateral ectopic processes in 12.5% (n = 1/8) and 50% (n = 4/8) of embryos, respectively (Fig. 4G-G’, I, Fig. S4C). The remaining C2cd3^ex4-5fl/fl; Crect embryos (37.5%, n = 3/8) did not present with ectopic cartilaginous processes (Fig. S4D). Interestingly, C2cd3^ex4-5fl/fl; AP2-Cre embryos presented with ectopic cartilaginous processes, either bilateral (60%, n = 6/10) or unilateral (40%, n = 4/10; Fig. 4H-I, Fig. S4E), at the highest level of penetrance of all Cre drivers. We found that these ectopic cartilaginous processes were consistently within the intermediate anterior portion of Meckel’s cartilages as observed in ta² avian mutants (Fig. S4F). Additionally, we first detected ectopic chondrogenesis in C2cd3^ex4-5fl/fl; AP2-Cre MNPs at e13.5, the first stage in which cartilaginous matrix is secreted by chondroblasts in Meckel’s cartilage (Svandova et al., 2020) through Safranin O staining (Fig. 4J-K’). Thus, these data suggested an essential role for C2cd3, and cilia, in both epithelial and CNCC-derived mesenchymal tissues. Of note, loss of C2cd3 in the Scx-lineage did not result in ectopic cartilage (Fig. S5), further suggesting that the onset of ectopic cartilage was a result of early aberrant mandibular patterning rather than altered fate decisions.

With the observation that conditional deletion of C2cd3 in CNCC mesenchyme and craniofacial ectoderm in murine embryos phenocopied the ta² ectopic cartilaginous process phenotype, we further characterized C2cd3^ex4-5fl/fl; AP2-Cre embryos to understand if this model recapitulated additional ta² cranial skeletal phenotypes. microCT scans of C2cd3^ex4-5fl/fl; AP2-Cre embryos revealed micrognathia as well as an ectopic mandibular skeletal element similar to that previously reported in ta² embryos (Fig. 4L-P) (Bonatto Paese et al., 2021a). Furthermore, C2cd3^ex4-5fl/fl; AP2-Cre embryos presented with palatal clefting characterized by hypoplastic maxillary, premaxillary, and palatine bones, an absent palatine process, and a dysmorphic pterygoid process, similar to ta² embryos (Fig. 4Q-R’; Bonatto Paese et al., 2022; Chang et al., 2014; Schock et al., 2015). These results further validated the importance of C2cd3 and cilia in tissues of the craniofacial complex and suggested a conserved role between species.

With the ectopic cartilaginous processes in C2cd3^ex4-5fl/fl; AP2-Cre resembling those in ta² embryos, we performed wholemount HCR to determine if the molecular basis for these phenotypes were conserved between species. As expected in e10.5 control MNPs, Shh was expressed in the oropharyngeal ectoderm, Fgf8 was expressed in the proximal aboral ectoderm, and Bmp4 was expressed in the distal aboral ectoderm (Fig. 5A). Similar to the ta², Bmp4 expression was expanded into the oral aspect, whereas Shh was diminished, and Fgf8 was unchanged in e10.5 C2cd3^ex4-5fl/fl; AP2-Cre MNPs (Fig. 5B). Sagittal views and quantification of expression domains of both C2cd3^ex4-5fl/fl and C2cd3^ex4-5fl/fl; AP2-Cre confirmed expansion of Bmp4 into the oral ectoderm and reduction of Shh expression, with no significant change to Fgf8 expression (Fig. 5C-E). Associated target gene expression was commensurately altered in CNCCs of C2cd3^ex4-5fl/fl; AP2-Cre MNPs (Fig. 5F-K). These results ultimately suggested a conserved mechanism for the induction of ectopic cartilaginous processes between avian and murine embryos.

Figure 5. Conditional loss of C2cd3 in epithelium and mesenchyme recapitulates ta² molecular mandibular patterning.

(A-D’) Three-dimensional renderings of wholemount HCR expression data for Shh (red), Fgf8 (green), and Bmp4 (purple) in e10.5 C2cd3^ex4-5fl/fl (A, C, n = 3) and C2cd3^ex4-5fl/fl; AP2-Cre (B, D, n = 3) MNPs. (E) Quantitation of three-dimensional surface renderings of Shh, Fgf8, and Bmp4 in e10.5 C2cd3^ex4-5fl/fl and C2cd3^ex4-5fl/fl; AP2-Cre MNPs as a percentage of total mandibular surface area. (F-K) Section RNAscope in situ hybridization for Shh and Ptch1 (F-G), Bmp4 and Msx1 (H-I), and Fgf8 and Lhx6 (J-K) on sagittal sections of e10.5 C2cd3^ex4-5fl/fl (F, H, J, n = 3) and C2cd3^ex4-5fl/fl; AP2-Cre (G, I, K, n = 3) MNPs. The dotted lines in C and D denote the midpoint of the oral-aboral axis of the MNP. Scale bars: 200μm (A-D), 100μm (F-K). Asterisk denotes p < 0.05, n.s. signifies no significance.

Discussion

Mandibular dysmorphologies are a common feature of craniofacial ciliopathies. However, the mechanisms that govern mandibular musculoskeletal tissue patterning remain nebulous. Herein, we utilized both avian and murine model systems with mutations in C2cd3 to understand how disruptions in ciliary function impacted signaling events that guide CNCC differentiation into musculoskeletal tissues. We found that loss of C2cd3, and subsequent loss of ciliary extension, led to an ectopic epithelial BMP signaling center that propagated a BMP signal into the mandibular CNCC mesenchyme, resulting in ectopic cartilaginous processes (Fig. 6). This study further expands our understanding of the necessity of primary cilia in regulating the spatial organization of signals that govern mandibular patterning.

Figure 6. Summary of Meckel’s cartilage development in control and C2cd3 avian and murine mutant embryos.

(A) In control conditions with normal C2cd3 expression and ciliogenesis, Bmp4, Shh, and Fgf8 expression is normal in the developing MNP. Normal patterning of the MNP results in normal differentiation of tendon and cartilage. (B) When C2cd3 is mutated, such as in ta² avian and C2cd3^ex4-5fl/fl; AP2-Cre murine mutants, ciliogenesis is perturbed. There is a subsequent decrease in Shh expression and an increase in Bmp4 expression in the developing mandibular prominence. Augmented Bmp4 expression correlates with ectopic chondrogenesis during mandibular musculoskeletal differentiation. Cilium diagrams were created using BioRender.com. APs: attachment progenitors.

The primary cilium as a signaling hub for Hedgehog and BMP pathways

The cilium has long been associated with transduction of the Hh pathway due to localization of several pathway members to the cilium itself and the similarities of ciliopathic and Hedgehog-related phenotypes (Goetz and Anderson, 2010; Huangfu et al., 2003; Huangfu and Anderson, 2005). Furthermore, the observed trafficking of Gli transcription factors through the cilium, which is essential for proper Hh signal transduction, has also been definitively and repeatedly documented (Haycraft et al., 2005; Kim et al., 2009; Wen et al., 2010). Interestingly, some but not all, of the mandibular (ectopic bone and expanded cartilage) and molecular (decreased Hh activity) readouts observed in both ta² and murine C2cd3 mutant embryos were previously reported in conditional knockouts of Intraflagellar protein 88 (Ift88), and the Hh pathway transducer Smoothened (Kitamura et al., 2020; Xu et al., 2019). Our RNAscope gene expression analysis supports the suggestion that reduced canonical Hh signaling may indeed contribute to the ectopic cartilages observed in our mutants; however, CellChat analysis did not validate that as the primary causal mechanism. Based on these data, we suggest that the unique phenotype may be associated with either a mutation in a protein localized to the centriole versus one localized to the axoneme, or additional aberrant signaling.

While the evidence for the relationship between the cilium and Hh has been observed and published by multiple groups, the evidence for if and how other molecular pathways (Wnt, FGF, BMP, PDGF, etc.) utilize the cilium has been less clear. In the case of BMP, current data suggests that TGFβ/BMP receptors are recruited to the cilium in order to activate SMADs (Anvarian et al., 2019). In addition to the localization of receptors to the ciliary membrane, proteins necessary for receptor recycling, pathway phosphorylation, and pathway feedback have all been reported to localize to the ciliary base/subdistal appendages (Clement et al., 2013; Miyazawa and Miyazono, 2017; Westlake et al., 2011). Since both RNAscope and CellChat analyses support increased BMP expression and signaling, we hypothesize that the combinatorial impact of reduced Hh and increased BMP is causal for the ectopic cartilaginous processes, as well as accompanying musculoskeletal malformations observed in ta² and murine C2cd3 mutant embryos. However, the exact mechanism of this combinatorial molecular insult is still unclear. It is possible that ectopic BMP expression and subsequent increased activity is secondary to reduced Hh activity since there is a well-documented antagonistic relationship between the two pathways (Bastida et al., 2009; Patten and Placzek, 2002; Xu et al., 2019). It is also possible that both pathways are directly impacted by a C2cd3 mutation. While we cannot eliminate the possibility of a BMP ligand being transduced through the cilium via a currently unknown mechanism, our molecular expression studies, in conjunction with other work utilizing ciliopathic animal models, favor a mechanism by which the cilium fine-tunes the antagonistic relationship between the BMP and Hh pathways during tissue patterning (Horner and Caspary, 2011; Kitamura et al., 2020). Future work into understanding if and how downstream factors of the Hh pathway mediate this antagonistic relationship, and potentially interact with phospho-Smad transcription factors, could provide this mechanistic insight.

Conditional knockouts of C2cd3 suggest cartilage development requires tissue-specific signaling contributions

Single-cell analysis suggested a role for epithelial signaling in the emergence of ectopic cartilaginous processes. To examine specific tissue contributions, we utilized conditional mouse models to ablate C2cd3 in tissues that contribute to the developing mandible (ectoderm only, mesenchyme only, and ectoderm and mesenchyme). We observed variations in the frequency, number, and presentation of ectopic cartilaginous processes in our conditional murine mutants. We detected a higher incidence of bilateral ectopic cartilaginous processes on Meckel’s cartilage in our AP2-Cre (ectoderm and mesenchyme) conditional mutants compared to our Wnt1-Cre2 (mesenchyme only) and Crect (ectoderm only) conditional mutants. These results suggested that signals from both the ectoderm and the CNCC-derived mesenchyme contributed to cartilage development. Furthermore, it is possible that these adjacent tissue populations must engage in tissue-tissue crosstalk for proper cartilage development. The generation of unilateral ectopic cartilaginous processes in a subset of conditional murine mutants is not an unsurprising finding. Several craniofacial anomalies present unilaterally including hemifacial microsomia, unilateral cleft lip, and unilateral coronal craniosynostosis. There are currently no widely accepted molecular mechanisms for why these conditions present unilaterally (Bishara et al., 1994; Chen et al., 2018; Dixon et al., 2011; Marbate et al., 2022). Since the ta² mutation is a hypomorphic mutation and the C2cd3^ex4-5fl/fl conditional deletion can generate a truncated C2CD3 protein isoform (Chang et al., 2021, 2014), we surmise that the generation of unilateral ectopic cartilaginous processes could be due to the variability in the percent of cilia loss in individual embryos.

Is ectopic cartilage in talpid² associated with atavism?

Previous studies characterizing ta² embryos revealed integumentary outgrowths on both the lower and upper jaws that resemble first-generation teeth from crocodilians as a result of shifts in the oral-aboral patterning molecules SHH and BMP4 (Harris et al., 2006). The location of the ectopic cartilaginous processes in the distal portion of the mandible coincides with the emergence of archosaurian-like teeth in the ta². Interestingly, secondary cartilages have been shown to emerge in sites of tooth eruption in the fossils of non-avian dinosaur hatchlings (Bailleul et al., 2013, 2012). While we are limited in our analysis of the ta² due to embryonic lethality, the altered musculoskeletal anatomy in the intermediate anterior mandible at the ectopic cartilaginous processes does not exclude the possibility that they are associated with ectopic tooth formation.

Taking the presence of archosaurian-like teeth, together with the more recently discovered skeletal patterning in the limb that resembled ancestral tetrapods (Bonatto Paese et al., 2021b), our finding of ectopic cartilaginous processes that resemble entheses solidifies the ta² as an experimental model to dissect the molecular mechanisms underlying the emergence of atavisms, or recurrent ancestral characteristics, during skeletal development and evolution in comparison to embryos from reptilian and other avian species. Additional phenotyping efforts and examination of accompanying molecular changes in the ta² may reveal critical insights into the evolution of jaw, limb, and other musculoskeletal structures.

Biomedical relevance of ectopic cartilaginous outgrowths in ta² for human ciliopathies

Craniofacial anomalies, including those present in ciliopathies, are among the most common congenital conditions affecting humans. Furthermore, they represent a significant biomedical burden of approximately $700 million per year. Surgical repair of craniofacial conditions is difficult and often requires a large source of skeletal tissue to reconstruct the facial skeleton. Depending upon the skeletal deficit, distinct treatment options are utilized. Common treatments for craniofacial skeletal anomalies include bone grafting with growth factor (typically BMP) supplementation and distraction osteogenesis. Bone grafting involves removal of bone from one part of the body to repair the anomaly/injury. This is the preferred treatment when the bone deficit is relatively small. Distraction osteogenesis is a surgical procedure which involves cutting and slowly separating bone, allowing the bone healing process to fill in the gap (Ilizarov, 1988; McCarthy et al., 1992). This eliminates the need for taking bone from elsewhere but does require healthy soft tissue. Both approaches, however, have variable outcomes and can cause damaging off-target effects (e.g., relapse, nerve damage, infection, device failure, etc.) (Holloway et al., 2014; Kahn, 2014). Thus, there is a clear need for an improved approach towards surgical repair of craniofacial anomalies as well as metrics to determine which strategy is best.

Some ciliopathies have been reported to present with ectopic skeletal structures (Bonatto Paese et al., 2021a; Bredrup et al., 2011; Cela et al., 2018; Kitamura et al., 2020; Tabler et al., 2013; Zhang et al., 2011). Of those, bony protrusions, or tori particularly on the palate (torus palatinus) (Bredrup et al., 2011) or mandible (torus mandibularis) have been observed. Despite the common occurrence of these structures, there has been a significant knowledge gap in understanding their origin. Recently, one hypothesis put forward suggested mandibular tori originate embryonically from a bending of Meckel’s cartilage that undergoes endochondral ossification at the level of the mental foramen (Rodríguez-Vázquez et al., 2013). A separate study examining the success of mandibular advancement devices, like those used in distraction osteogenesis, suggested that the presence of torus mandibularis almost tripled the likelihood of a successful response to distraction osteogenesis (Diaz de Teran et al., 2022). These data prompted two questions: 1) are the ectopic cartilaginous processes we observed embryonic precursors of mandibular tori, and 2) would the presence of ectopic skeletogenic processes in ciliopathy patients with micrognathia make them good candidates for distraction osteogenesis? While the early embryonic lethality of the ta² prevents us from addressing the former question, the latter may be addressed by careful phenotyping and documentation of outcome measures in ciliopathy patients. Such steps should be taken in an effort to best move forward with appropriate treatment options for this growing class of diseases.

Highlights.

• Ciliopathies present with malformations of the mandibular skeleton

• The talpid² avian mutant presents with bilateral outgrowths on Meckel’s cartilage

• Ectopic cartilage in talpid² mandibles resembles developing muscle attachment sites

• Ectopic mandibular cartilage outgrowth correlates with ectopic BMP signaling

• Deletion of C2cd3 in murine neural crest and ectoderm phenocopies talpid² phenotype

Data Availability

The single-cell transcriptomic dataset used in this manuscript can be found on FaceBase.org under the accession code FB00001245.