Animal experiments were approved by the Research Ethics Office at the University of Alberta (Animal Care and Use Committee, AUP1149, AUP2527) in compliance with guidelines set by the Canadian Council of Animal Care. Mice on a C57Bl/6 background were housed in the animal facility at the University of Alberta. Ankrd11^{TM 1a(EUCOMM)Wtsi//}IcsOrl mice, in which exon 7 was floxed, were rederived from sperm (EM:07651, the European Mouse Mutant Archive – Infrafrontier) and crossed with Flp recombinase expressing (Jackson Labs) mice to generate Ankrd11^{TM 1c} conditional ready mice (Ankrd11^fl/fl) as described (Skarnes et al., 2011). Exon 7 is located within the Ankyrin repeat domain. Ankrd11^fl/fl mice were crossed with B6.Cg-E2f1Tg(Wnt1-cre)^2Sor/J (Jackson Labs) (Lewis et al., 2013) mice for neural crest-specific deletion of Ankrd11 (Ankrd11^ncko). Deletion of exon 7 results in out-of-frame splicing of exon 6 to exon 8, leading to a pre-mature stop and truncation of Ankrd11. Genotyping of mice was performed using Taq DNA Polymerase 2x Master Mix RED (Ampliqon, Denmark) from ear-notch or tissue biopsies. The following primers and PCR conditions were used: (1) for Ankrd11: forward, 5′-CTGTCTCAGAGAGGAGAGTGAGGAGGAC-3′; reverse, 5′-TACCTTACACCCTGAGACGGCGTC-3′; 34 cycles of: 94°C-30 s, 62°C-45s, 72°C-60s; (2) for the Cre transgene: forward, 5′-TTCCCGCAGAACCTGAAGATG-3′; reverse, 5′-CCCCAGAAATGCCAGATTACG-3′; Twsg1 internal control forward, 5′-AACAACAATGGCACAACCTAAT-3′, Twsg1 reverse, 5′-ACTTTCTCCCCACCCGTCTA-3′; 35 cycles of: 94°C-15s, 60°C-30s, 72°C-90s.

Mice were imaged using a MILabs μCT scanner at the School of Dentistry at the University of Alberta. Heads were fixed in 4% paraformaldehyde (PFA) for 24 h, then scanned at the following parameters: voltage = 50 kV, current = 0.24 mA, exposure time = 75 ms, 1600 exposures/full rotation. Scans were reconstructed at a voxel size of 25–35 μm, and the volumes were exported as NifTI-1 files. NifTI-1 files were directly analyzed using Amira software (version 2019.2, Life Technologies). NifTI-1 files were batch-converted to MINC-2 format using a custom made bash script and the MINC toolkit (Vincent et al., 2016).

Embryos were dissected from uterus at embryonic days (E) 13.5, 14.5, and 15.5 and collected at birth at postnatal day (P) 0. Embryos were decapitated and heads were fixed overnight at 4°C in 4% PFA. P0 heads were decalcified in 0.5M EDTA overnight before processing. Heads were processed and sectioned as previously described (Baddam et al., 2020). Before further analysis, slides were warmed in a 60°C oven, deparaffinized with xylene, and rehydrated through graded ethanol washes.

Density of cells forming the palatal shelves was quantified at E13.5 using DAPI nuclear stain (not shown). Semi-automated cell counting was performed in ImageJ on spatially defined regions within the palatal shelves (oral or nasal domains).

The number of immunostained Ki67-positive cells were counted on a minimum of six separate images per age and genotype from a palatal shelf. Sections were from a minimum of five different embryos from at least three different litters. The region counted is indicated in Figure 5 with white dotted lines. Counting was automated in ImageJ using thresholding, watershed and particle analysis functions. Values presented in the graph are number of Ki67 positive cells divided over DAPI positive cells to normalize for hypoplasticity of shelves. Quantification of Runx2 and Sp7 domains at E12.5/13.5 was performed similarly, with values reported as percent of a defined field of view occupied by the stain and/or number of positive cells within the field of view. At least 3 (E14.5) or 6 (P0) sections were analyzed from 3 to 5 mice per genotype from at least three different litters. Statistical significance was determined via an unpaired two-tailed t-test.

Primary osteoblast cultures were established from calvaria for the gene expression time-course experiment. Calvaria were dissected and cut into small fragments, discarding cranial suture tissue. Calvarial bones were dissected into HBSS (Sigma H-9394) and used for osteoblast isolation using a modified protocol (Bakker and Klein-Nulend, 2012; Perpétuo et al., 2019). HBSS was replaced with 0.25% trypsin for 10-min digestion at 37°C, washed in αMEM (Gibco 12561-056), and digested twice with 0.2% Collagenase Type II solution (Worthington 4176) for 30 min each. Final digestion product was collected. Remaining bone pieces were washed with αMEM – solution was added to final digest and bone pieces were rinsed three times with cCM (complete culture medium: αMEM supplemented with Penicillin-Streptomycin [Gibco 15240-062], ascorbate [100 μg/mL, Sigma, A-8960], 10% Fetal Bovine Serum [Sigma F-1051]), then transferred to 75 cm² flasks containing cCM. Final digest was centrifuged for 5 min at 1500 g and the pellet was resuspended in 1 mL αMEM per calvarium. Cell suspension was combined with bone pieces in flask and incubated at 37°C and 5% CO₂ in air until confluent. cCM media changed every 3 days.

Once confluent, cells were trypsinized and seeded in a 12-well plate at 5–10 × 10⁴ cells per well. Osteogenic medium (αMEM supplemented with ascorbate [50 μg/mL] and 2 mM β-glycerophosphate [Sigma G-5422]) was added when cells reached 80–90% confluency. Medium was changed every 3 days. Cells from three wells were harvested using Trizol (Invitrogen 15596026) at days 0, 3, 6, 9, 12, 15 for mRNA isolation and RT-qPCR quantification. Three independent experiments with two mice each were performed with three wells/time-point/experiment. For mRNA isolation, wells from within an experiment were combined. A representative result is shown.

RNA extraction, cDNA synthesis, and quantitative RT-qPCR were performed as previously described (Malik et al., 2020) using appropriate primer pairs normalized to 36B4 as a reference gene (see Supplementary Table 2). Fold change was calculated using 2^{–ΔΔ Ct  method} (Livak and Schmittgen, 2001). MIQE guidelines were followed (Bustin et al., 2009).

Three-dimensional facial images from four genetically confirmed KBG syndrome patients were combined and mapped against a reference face (Figure 1A) (Hallgrímsson et al., 2020). The resulting mesh heatmaps revealed a consensus hypoplastic mid-and lower faces, with an expansion of the upper third of the craniofacial complex. The Ankrd11 neural crest-specific haploinsufficient deletion was generated by crossing Ankrd11 floxed mice with Wnt1Cre2⁺ mice (Ankrd11^nchet) (Figure 1B). Analysis of μCT scans from mice Ankrd11^nchet mice revealed distinct changes to several craniofacial structures (Figure 1C): a persistent open fontanelle, a unique ossification defect in the posterior frontal suture, a slight change in pterygoid bone morphology evident from coronal ortho-slices anterior to the coronal suture (Figure 1C, lower panels). The angle of medial aspect of the pterygoid bone relative to ventrolateral was altered in Ankrd11^nchet mice. Comparative mesh heatmaps of Ankrd11^nchet and control scans revealed growth alterations in comparable regions as described for the human faces: reduced facial width, hypoplasia of midface, nasal region, and an expanded cranial vault (Figure 1D). Note: mandibles were omitted from analysis because of their variable position with respect to the skull.

While pups lacking both copies of Ankrd11 in the neural crest (Ankrd11^ncko) died at birth, precluding our analysis at the postnatal and adult stage, visual inspection of the neonatal head at postnatal day 0 (P0) revealed partially open eyelids, variable midfacial hypoplasia, reduction of calvarial growth (evident as reduction in calvarial microvasculature), loss of pigment on the nose, fully penetrant cleft palate, and a smaller tongue relative to control mice (Supplementary Figure 2).

Skeletal preparations of P0 neonates demonstrated an underdeveloped midface and severe micrognathia in Ankrd11^ncko mice when compared to littermate controls (Figure 2A). A cleft palate was evident, as was the reduced ossification of the palatine bones and the anterior cranial base (presphenoid and sphenoid) (Figure 2A). Micro-computed tomography (μCT) showed that all intramembranously formed orofacial bones were reduced in size, resulting in underdeveloped midface and mandible (Figure 2B). Ossification of the anterior cranial bones was similarly severely stunted, and bones failed to cover the majority of the cranium at birth when compared to control mice (Figure 2B). Ossification of the anterior cranial base (presphenoid and basisphenoid) was similarly reduced. Several of the primary ossification centers failed to expand and formation of the pterygoid wings was largely missing. In contrast, more posterior, mesoderm-derived components of the cranial base (basioccipital bone) appeared unaffected. Quantification of the volumes of isolated craniofacial bones confirmed 39% reduction of the mandible (p < 0.001), 83% reduction of the frontal bone (p < 0.001), and 76% reduction of parietal bone (p < 0.05), with interparietal, occipital and basioccipital bones being not significantly different (Figure 2C).

Histological analysis confirmed the notable size difference of the head and the cleft palate phenotype (Figure 3A). All major structures were present and developed to a recognizable stage, with exception of the tongue and palatal shelves that remained hypoplastic (Figure 3A). Note that due to retrognathia, mandibular structures on the sections appear more anterior. The developing palatal shelves were hypoplastic and dysmorphic already at embryonic day 12.5 (E12.5) (Figure 3B). Hypoplasia of the tip of the palatal shelf remained evident at later developmental stages. Elevation of palatal shelves appeared normal, but hypoplastic shelves did not meet to fuse, resulting in failure to separate oral and nasal cavities (Figure 3B). Analysis of cell density in the oral and nasal domains of the palatal shelves at E13.5 revealed a 15% increase in cell density in the nasal domain only (p < 0.05), with no differences observed in the oral domain (Figure 3C).

To gauge when and how Ankrd11 could cause the above-described phenotypes, we performed expression analysis for Ankrd11 at E13.5, E14.5, and P0 (Figure 4A). At E13.5, Ankrd11 expression was discrete in orofacial regions with clear expression in a mesenchymal condensation in the developing maxillary region (sphenoid wing). Some expression was noted in the developing palatal shelves and lining oral epithelium. Note the distinct expression in the developing forebrain at this stage. At E14.5, Ankrd11 expression was overall stronger, more widely expressed, and could be observed in several developing mesenchymal structures. Distinct expression was observed in ossification centers and at sites of bone development in the mandible and maxilla, as well as in pre-odontoblasts in the developing molars. Ankrd11 was also present in some cartilaginous structures (nasal septum, developing turbinates), parts of the tongue, and the midline epithelial seam of the fusing palate. In the eye, expression was prominent in the developing lens as well as in the anterior segment (Figure 4A). Extent and intensity of Ankrd11 expression waned by P0, with the majority of Ankrd11 at this point restricted to the oral epithelium. Expression in the mandibular and maxillary bones was strongly reduced or absent, as was expression in teeth. In the eye, expression was now restricted to the posterior segment (Figure 4A). This indicates dynamic, but discrete, expression of Ankrd11 throughout development of craniofacial structures.

This dynamic expression is nicely illustrated in the developing mandibular bone (Figure 4B), one of the earliest bones to mature in the craniofacial complex (Flaherty and Richtsmeier, 2018). While at E14.5, at the onset of ossification, Ankrd11 signal lined the forming mandibular bone, expression was negligible in the comparable region at P0 (Figure 4B). Similarly, in the cranial vault, Ankrd11 was localized to the forming ossification centers at E14.5, whereas at P0 weak expression was observed in cells lining the calvarial bone, but not in osteocytes within the mature bone (Figure 4B).

In comparison, Ankrd11 expression in the developing palatal shelves and the developing maxilla was more discrete. At E12.5, a mostly epithelial expression pattern was observed in palatal shelves (Figure 4C). A few Ankrd11-positive mesenchymal cells were evident in the buccal aspect toward its tip, as well as in its upper, dorsal aspects (Figure 4C). At E13.5 and E14.5, the epithelial and mesenchymal expression largely remained comparable to E12.5. Overall, at E14.5, Ankrd11 expression was strongest in epithelium of the newly forming nasal cavity, the oral epithelium, the developing maxillary bones, as well as in the fusing palatal shelf, restricted to the midline epithelial seam. With the exception of some epithelial structures, Ankrd11 expression at P0 was reduced in all of these places. Residual weak expression in presumptive bone-lining cells was noted (Figure 4C).

To assess the cause for the hypoplastic palatal shelves in Ankrd11^ncko embryos, we assessed apoptosis and proliferation. No significant apoptosis was observed at any of the time points between E12.5–E13.5 (Supplementary Figure 3). The overall ratio of Ki67-positive cells to DAPI-positive nuclei was comparable at E12.5 and E13.5 in Ankrd11^ncko palatal shelves when compared to Ankrd11^ctrl (Figures 5A,B). When shelves were separated into oral and nasal domains and a spatially restricted quantitative analysis was performed (Figure 5A, right panels), a 40% decrease in mesenchymal proliferation became obvious in the E13.5 oral domain (n ≥ 5 each, p < 0.05) (Figure 5B). To test for changes to cellular and extracellular organization, E13.5 palatal sections were stained with Alcian blue to identify distribution of glycosaminoglycans (GAGs). The overall organization and direction of cell orientation differs between Ankrd11^ncko and control shelves (Figure 5C). In addition, a disorganization of the mesenchymal cells lining the shelf epithelium was noticed (Figure 5C). These findings indicate that loss of Ankrd11 has multiple subtle effects on palatal shelf organization and maturation, which likely underlie the cleft palate phenotype.

To better understand the reduction in intramembranous ossification observed in Figure 2, protein expression of two master regulators of ossification, Runx2 and Sp7 (Hojo et al., 2016; Komori, 2017) was assessed within the field of view indicated in Figure 6A. At E14.5, expression of Runx2 and Sp7 quite homogenously outlined the growing maxillary bones in control embryos (Figure 6A, top row). In contrast, in Ankrd11^ncko mutant embryos, Runx2 and Sp7 expression was only strong in the center, becoming more diffuse toward the periphery. Sp7 expression was more contained in comparison to Runx2 (Figure 6A). Quantification of the respective expression domains indicated that expression of both Runx2 and Sp7 was more contained, however the differences in the area occupied did not reach statistical significance (Figure 6B).

Differences in expression became even more striking at P0, a time-point of significant maxillary bone growth and remodeling. Expression of Runx2 and Sp7 could be observed in osteoblasts lining the newly formed trabeculae in control neonates (Figure 6C, top row). In Ankrd11^ncko, this trabecular pattern appeared disturbed. Expression of both Runx2 and Sp7 appeared patchy and locally more restricted to the small, sparse trabeculae in maxillary bone (Figure 6C, bottom row). Tracing the outlines of cells expressing Runx2 and Sp7 revealed the presence of extended, Runx2-positive cellular assemblies reminiscent of developing trabecula (Figure 6D, red dotted line), while the cellular structures in the mutant were more self-contained (Figure 6D, yellow arrowheads). Quantification indicated an 85% reduction of the Runx2-expression domain (p < 0.001) as well as 76% reduction in the number of Runx2-positive cells (p < 0.001). In contrast, despite a trend toward lower values, neither the number nor area of Sp7-positive cells were significantly different (Figure 6E). To test whether Ankrd11 is indeed expressed and changes during osteoblast differentiation, we performed an in vitro time course of calvarial osteoblast differentiation. RT-qPCR analysis revealed that expression of Ankrd11 changes dynamically over the course of osteogenic differentiation (Figure 7A). For reference, expression of several markers of osteoblast differentiation is shown: osteocalcin (Ocn), integrin binding sialoprotein (Ibsp), Osteopontin (Opn), alkaline phosphatase 1 (Alp1). This co-expression was mirrored in vivo, where Ankrd11, albeit more discrete, largely aligned with expression of Sp7 at E14.5, but not at P0 (Figure 7B).

The altered appearance of maxillary bones in Figure 7 prompted us to analyze maxilla, mandible, and calvaria trabeculae in more detail. In control mice, discrete trabeculae with widely interspaced osteocytes of mostly elongated appearance were readily identified (Figure 8A, top row). In contrast, trabeculae appeared less refined with a higher number of poorly aligned osteocytes evident in all bones from Ankrd11^ncko mice (Figure 8A, bottom row). Staining for the Wnt antagonist Sclerostin (Sost), a marker for mature osteocytes and an important regulator of bone remodeling (Winkler et al., 2003; Husain and Jeffries, 2017), revealed a strong reduction in Ankrd11^ncko mice (Figure 8A, right column), suggesting that osteocyte maturation is delayed and bone remodeling might be compromised. To better illustrate this, we stained coronal sections of the craniofacial complex with Picrosirius red to reveal collagen fiber networks and osteoclast-specific tartrate resistant alkaline phosphatase (TRAP). An overall reduction in trabeculation was seen in the maxilla, calvaria (Figure 8B), and mandible (Figure 8C). This was mirrored by a 99% reduction in TRAP-positive regions in the maxilla, 85% reduction in the mandible, and 99.5% reduction in the calvaria (Figure 8D bottom graph; p < 0.05) despite presence of comparable amounts of bone within the region of interest (Figure 8D top graph; n.s.). In the calvaria (Figure 8B, bottom panels), a single-layered bone was observed in Ankrd11^ncko mice indicative of reduced or defective bone remodeling. TRAP staining revealed an almost complete lack of bone resorption and, by extension, remodeling. Birefringent imaging under polarized light of picrosirius red-stained sections confirmed changes in trabeculation, interconnectivity, and crosslinking of collagen fibrils (Figure 8E). Thus, Ankrd11 appears to be comparatively more important in bone growth and remodeling than the initial induction of intramembranous bone, despite its prominent expression in ossification centers.

Neural crest-specific deletion reveals the broad involvement of Ankrd11 during craniofacial development. In line with this, Ankrd11 shows restricted and dynamic expression in many craniofacial tissues. At E13.5 and E14.5, Ankrd11 is localized to discrete parts of the eye, bones, teeth, cartilage, and several muscular components. At a later developmental stage, expression in these regions is comparatively low and is primarily restricted to oral epithelium. Some expression in muscular compartments and olfactory bulb was noted in line with Gallagher et al. (2015). Relating expression to phenotypic manifestations allows strong predictions on the involvement of Ankrd11 in the development of these different structures. For instance, Ankrd11 expression is noted from E13.5 onward in mesenchymal condensations and at E14.5 in ossification centers. This implies that the bony defects observed in Ankrd11^ncko mice might precipitate from differences during bone induction, although some of the defects such as osteocyte maturation and remodeling manifest only at much later stages. A similar pattern is seen in the maxillary and calvarial bones, where Ankrd11 is strongly expressed early in development, yet the impact on bone maturation and remodeling is noted only in much later stages when Ankrd11 expression is markedly reduced. It was previously suggested that development of bone involves epigenetic mechanisms, which might in part be controlled by lineage-specific transcription factors such as Sp7/Osterix (Park-Min, 2017). Whereas induction of ossification centers appears not to be compromised, significant differences are observed with respect to growth and maturation of these ossification centers. Expression domains of Runx2 and Sp7 are differentially affected. Thus, our study illustrates how expression of these two master genes are variably affected by loss of Ankrd11. At present, it is unclear if the weak expression of Ankrd11 seen in more mature bone indicates a continuing requirement for Ankrd11 to maintain established epigenetic signatures, or whether it indicates ongoing, not-yet-understood de novo genomic fine-tuning.

Both 3D morphological as well as phenotypic analysis indicate that Ankrd11 is particularly important for intramembranous ossification. Establishment of ossification centers appears not to be compromised in the Ankrd11^ncko embryos; however, they universally fail to fully expand. This was not due to changes in proliferation or apoptosis (Supplementary Figure 3) but rather appeared to be the consequence of other, not fully understood mechanisms affecting specification, maturation, cell orientation, or migration of the induced bone-forming cells. In this light, Yoda mice or mice with Ankrd11 knockdown in cortical progenitors display aberrant positioning of neurons in the developing cortex linked to dysregulated epigenetic mechanisms (Gallagher et al., 2015). Together with our craniofacial data, this could underscore that control of cell migration may be a common mechanism.

Intramembranous calvarial bone development generally occurs in two stages: (1) development of bone primordia from mesenchymal condensations, e.g. anterior-superior of eye for the frontal bone (beginning at E12.5), and (2) expansion, e.g. of the frontal bone front to cover the top of the skull (Yoshida et al., 2008). Defects in either or both of these stages result in insufficient bone formation. Our results suggest that loss of Ankrd11 predominantly affects extension of the bone via growth at the osteogenic front and bone remodeling. The frontal bone insufficiency seen in Ankrd11^ncko mice mirrors the phenotype observed in the Beetlejuice mouse, which carries a mutation in Prickle1 (Prickle1^bj/bj) (Wan et al., 2018). Prickle1^bj/bj mice also have hypoplastic frontal bones, an expanded anterior fontanelle, and frequently cleft palate. The Prickle1 mutation is thought to affect a Wnt/Planar cell polarity (Wnt/PCP) signaling pathway. Apoptosis and proliferation are not affected, but unlike the Ankrd11 mutant, a notable decrease in Sp7 alongside minimal changes to Runx2 expression is observed (Wan et al., 2018). Mutations in core components of the Wnt/PCP pathway cause Robinow syndrome which, like KBG syndrome, results in a wider midface and shorter stature. Notwithstanding these phenotypic similarities, changes in Ankrd11^ncko mice are likely caused by inappropriate epigenetic regulation in differentiating osteoblasts. Histones associated with several gene loci for key osteogenic factors including Runx2, Sp7, and Alp undergo dynamic acetylation changes throughout the ossification process (Zhang et al., 2015). Of interest, Runx2 interacts directly with HDAC3, a histone deacetylase previously shown to directly interact with Ankrd11 (Schroeder et al., 2004; Zhang et al., 2004; Gallagher et al., 2015). It is thus possible that Ankrd11 and Runx2 participate in common gene regulatory networks to set up osteoblast differentiation during craniofacial bone development. Potential changes in epigenetic signatures were not investigated as part of this study, because craniofacial structures are too heterogeneous following different developmental kinetics, posing significant challenges for the analysis of subtle epigenetic changes.

Bone remodeling begins soon after bone formation. In Ankrd11^ncko mice, bone remodeling was severely compromised. The only notable remodeling occurred in the mandible, but even there it was dramatically reduced. Osteocytes appeared to retain an immature phenotype, evidenced by their increased numbers, plump morphology, apparent failure to align along stress/force lines, and lack of Sclerostin (Sost) expression, a characteristic of immature osteocytes. This could be the consequence of a delay in bone formation or an intrinsic defect in osteocyte differentiation. As Ankrd11 is not expressed in osteocytes (Figure 4B), this phenotype is likely caused by abnormal osteoblast differentiation, possibly involving inappropriate epigenetic osteoblast programming or specification as discussed earlier.

Immature osteocytes lack the exquisite mechanosensory properties of mature osteocytes. The primary cilium has been suggested to play an important role in bone mechanotransduction (Temiyasathit and Jacobs, 2010). Inability to sense or respond to mechanical cues would manifest as failure to align along stress lines, increased cell number, and failure to induce bone remodeling (exemplified by loss of TRAP staining). On the other hand, osteocyte maturation is dependent on mineralization, as the transition to mature osteocytes is triggered by differences in mechanosensing of mineralized and unmineralized matrix (Irie et al., 2008). Changes to ordered mineralization in Ankrd11^ncko bones could equally explain the failure in osteocyte maturation and associated bone remodeling. Osteoclasts are related to monocytes and are of hematopoietic origin. Neural crest-specific Ankrd11 deletion would not directly affect these cells. The defect in remodeling is therefore most likely due to defective mechanosensing or defective communication with osteoclasts, underlining interconnection of these systems.

While lack of Sost expression could reflect a failure of osteocyte maturation, it must be noted that Sost itself is under epigenetic control. HDAC5 deficiency results in increased Sost expression, impaired osteogenesis, and low bone density (Wein et al., 2015). Sost activity-blocking antibodies are an approved treatment modality for osteoporosis aiming to reduce bone resorption. A more detailed analysis will be required to carefully dissect the cause and consequence of these bone-related phenotypes.

In contrast to bone primordia, the pattern of Ankrd11 expression in the palate is relatively discrete. It also differs from the often broad expression domains described for many transcription factors and signaling molecules in the developing palatal shelves (He et al., 2011; Danescu et al., 2015). This restricted Ankrd11 expression was seen at all stages from E12.5-E14.5 and could be attributed to discrete cells organizing palate development, or a tight window of expression governing reprogramming. Thus, Ankrd11 might be involved in the spatial organization of the palatal shelf, possibly regulating the expression of a growth factor or its receptors, facilitating growth of the palatal shelf secondarily. Indeed, changes in cellular distribution were observed.

Furthermore, an asymmetric pattern of proliferation was observed at E13.5, with a decrease in the buccal half of the shelf. The affected region roughly corresponds to the gene expression domain of the Hedgehog signaling receptor Patched (Ptch) (Xiong et al., 2009), raising the possibility that Ankrd11 somehow intersects with Hh signaling and cilia. Hedgehog and Wnt signaling are indeed both critical for cilia maturation, and functional cilia are required for normal palatal development (Brugmann et al., 2010; Nakaniwa et al., 2019). Recently, Ankrd11 was directly associated with ciliopathies (Breslow et al., 2018). The interplay between Hedgehog and Wnt signaling and chromatin remodeling by Ankrd11 may provide important clues for understanding both palate and bone phenotypes and will be investigated in future studies.

Mesh analysis of four patients with KBG syndrome revealed hypoplastic mid- and lower face and enlarged upper face. This is largely mimicked in Ankrd11^nchet and Ankrd11^ncko mice (Figures 1, 2). Given the stunted intramembranous ossification in the Ankrd11^ncko mouse, we propose that the expansion of the upper 1/3 of the face is a consequence of underdeveloped midfacial bones as well as intramembranously formed extensions of the cranial base – the pterygoid wings. The resultant narrower cranial base and midface would necessitate increased cranial bone growth to accommodate the expanding brain, leading to a triangular face. The calvaria would not be able to meet the increased demand in bone growth (as they are compromised in bone formation themselves) resulting in a persistent anterior fontanelle.

Patients with KBG syndrome show various accompanying craniofacial anomalies. Macrodontia, dental crowding, and hypo/oligodontia are often observed, necessitating invasive jaw surgery (Ockeloen et al., 2015; Low et al., 2016; Morel Swols and Tekin, 2018). Our results and expression analysis suggest that these malformations may be due to abnormal tooth and jaw development. Moreover, speech and feeding difficulties are frequently observed in KBG syndrome patients (Ockeloen et al., 2015; Low et al., 2016; Morel Swols and Tekin, 2018). The anomalies in palate development, more severe in Ankrd11^ncko mice but evident also in Ankrd11^nchet mice, can provide a structural explanation for these observations. High-arched palate or submucosal cleft, considered milder variants of defective palate development, lead to similar problems. Thus, our results will precipitate further investigations into the underpinnings of the various phenotypic craniofacial anomalies associated with ANKRD11.

ANKRD11 is one of the most disrupted genes in monogenic neurodevelopmental disorders with de novo mutations (Wilfert et al., 2017; Satterstrom et al., 2020). Yet, most patients with KBG syndrome are not diagnosed until 20–30 years of age, if at all (personal communication with Drs. Charlotte Ockeloen, Tjitske Kleefstra, and Peter Kannu). The gold standard for diagnosis of KBG syndrome is genetic testing to detect ANKRD11 variants or 16q24.3 deletion involving the gene. Specific craniofacial features such as persistent anterior fontanelle, submucosal or high-arched palate, macrodontia, and reduced bone mineral density are currently only suggestive of KBG syndrome and clinical phenotypic criteria for proper diagnosis remain poorly defined (Morel Swols and Tekin, 2018). This study confirms the power of 3D morphometric analysis to assist with syndrome diagnosis (Hallgrímsson et al., 2020) and illustrates the power of disease modeling in the mouse to unearth the underlying cellular basis for the malformation. Indeed, many craniofacial features observed in KBG syndrome (Ockeloen et al., 2015; Low et al., 2016; Morel Swols and Tekin, 2018) are recapitulated in the Ankrd11^ncko mouse. This study has begun to unravel some of the unknowns of Ankrd11-mediated regulation of craniofacial development and has allowed assessment of craniofacial malformations as a result of Ankrd11 deletion or loss-of-function. Systematic assessment of cataloged craniofacial features in patients with KBG syndrome can potentially drive further understanding of clinical penetrance and future establishment of definitive phenotypic diagnostic criteria to achieve earlier differential diagnosis, as well as clarify the impact of ANKRD11 variants of unknown significance (VUS). Similarly, studies in other organ systems often involved in KBG syndrome will further clarify the role of Ankrd11 in epigenetic and genomic control of tissue and organ development.

Our phenotypic characterization of conditional ablation of Ankrd11 in the murine neural crest revealed novel roles of Ankrd11 in craniofacial development, specifically in intramembranous bone formation and palate development. Our results will help improve clinical assessment of patients with KBG syndrome based on craniofacial phenotypic diagnosis, particularly early in life. Ultimately, early diagnosis and better understanding of ANKRD11 function will assist with genetic counseling for patients with KBG syndrome and their affected families.