Embryonic requirements for Tcf12 in the development of the mouse coronal suture

Craniosynostosis, the pathological fusion of the flat bones forming the top of the skull, occurs in ∼1/2000 live births (Lajeunie et al., 1995). Fusion occurs at fibrous joints called sutures that allow the skull to flex during childbirth and mastication. Craniosynostosis occurs in both sporadic and syndromic forms, and often affects particular sutures (Wilkie et al., 2006; Johnson and Wilkie, 2011). The coronal suture forms between the frontal and parietal bones and is preferentially affected in Saethre-Chotzen syndrome. The molecular genetics of this disorder have been under investigation for over two decades (Wilkie, 1997; Jabs, 2008). It is inherited as an autosomal dominant and has an incidence of 1/50,000 live births. In addition to coronal craniosynostosis, affected individuals also exhibit, with varying penetrance, digital abnormalities, abnormal facial structure and intellectual disability (Twigg and Wilkie, 2015). Heterozygous loss-of-function mutations in either of the basic helix-loop-helix (bHLH) transcription factor genes TWIST1 and TCF12 account for most Saethre-Chotzen syndrome cases (Howard et al., 1997; El Ghouzzi et al., 1997; Sharma et al., 2013). Twist1 belongs to a subfamily of bHLH proteins, members of which are expressed in specific tissues or cell types (Massari and Murre, 2000; Wang and Baker, 2015). Tcf12 is a member of the E protein subfamily of bHLH proteins, members of which typically exhibit broad expression (Wang and Baker, 2015). In mice, heterozygous deletion of Twist1 phenocopies the selective coronal suture loss seen in human Saethre-Chotzen patients (El Ghouzzi et al., 1997). Whereas heterozygosity of Tcf12 alone does not result in suture defects in mice, it does enhance coronal synostosis in Twist1+/− mice (Sharma et al., 2013). In zebrafish, loss of tcf12 or twist1b alone does not cause craniosynostosis, yet combined homozygous loss of twist1b and tcf12 results in coronal-specific synostosis (Teng et al., 2018), despite the zebrafish coronal suture likely not being homologous with the murine coronal suture (Teng et al., 2019). Although these experiments show combined requirements of Twist1 and Tcf12 for early development of the coronal suture (Teng et al., 2018), the individual role of Tcf12 in skull development had remained unexplored in model organisms, leaving it unclear whether Twist1 and Tcf12 control similar or distinct processes during suture development.

In mice and zebrafish, combinatorial loss of Twist1 and Tcf12 alleles influences the dynamics of osteogenic precursors during embryonic calvarial bone development (Teng et al., 2018). In compound mutants, frontal and parietal bone growth was accelerated, with the degree of acceleration in zebrafish predictive of later coronal suture loss. In addition to effects on osteogenic precursor dynamics in compound Twist1; Tcf12 heterozygous mice, heterozygous loss of Twist1 also results in inappropriate cell mixing at the boundary between the neural crest-derived frontal and mesoderm-derived parietal bones (Merrill et al., 2006; Ting et al., 2009). Loss-of-function of the ephrin receptor, EphA4, also causes cell mixing at the neural crest-mesoderm boundary and results in synostosis of the frontal and parietal bones (Merrill et al., 2006). Twist1 function is required in both neural crest-derived and mesoderm-derived cells for suture formation, with only heterozygous loss of Twist1 in both tissues resulting in synostosis (Teng et al., 2018). However, as these studies used non-conditional Cre drivers, distinct requirements in embryonic versus postnatal sutures could not be distinguished. This is a significant caveat, as recent work has demonstrated that the postnatal suture is a continuous source of osteogenic cells for the growth and repair of the calvarial bones. For example, Gli1:CreER can be used to mark long-term stem cells in the postnatal mouse suture, with genetic ablation of these labeled cells resulting in loss of all sutures (Zhao et al., 2015).

Twist1 can form homodimers as well as heterodimers with a variety of bHLH factors, including Tcf12 (Fan et al., 2020). Forced dimer approaches in embryonic stem cells (Fan et al., 2020) and in mouse sutures (Connerney et al., 2006) have provided evidence that Twist1 homodimers and Twist1-E heterodimers drive different cellular processes. In differentiating mouse embryonic stem cells, heterodimers promote expression of ectodermal and mesodermal markers and homodimers maintain cells in a pluripotent state. In developing mouse sutures, a higher Twist1/Twist1 to Twist1/E ratio was associated with an expansion of the osteogenic fronts and suture closure (Connerney et al., 2006). It thus appears to be clear that a change in the ratio of Twist1 homodimers versus Twist1-E heterodimers can produce different biological outcomes, although the exact role of Twist1 homodimers and heterodimers in suture development and craniosynostosis remains unclear.

Here, we investigate the extent to which Tcf12, a binding partner for Twist1 (Fan et al., 2020), participates in developmental processes of suture formation similar to those described for Twist1. In contrast to the expectation from forced dimer approaches, we show that homozygous loss of Tcf12 recapitulates many of the calvarial phenotypes of heterozygous Twist1 and compound heterozygous Twist1; Tcf12 mice, including accelerated frontal and parietal bone growth, and disruption of the boundary between sutural and osteogenic cells. We also find that Tcf12 function is required in both neural crest-derived and mesoderm-derived cells for coronal suture formation. The suture mesenchyme between the apposing frontal and parietal bones has been shown to house skeletal stem cells (Zhao et al., 2015). In the appendicular skeleton, Grem1 marks mesenchymal progenitors responsible for bone formation (Worthley et al., 2015). Interestingly, we find that Tcf12 also controls the asymmetry of Grem1+ mesenchyme cells at the parietal and frontal bone fronts, which may help explain the predictable overlapping nature of these bones at the normal coronal suture. In addition, we show that conditional deletion of Tcf12 in the postnatal Gli1+ sutural stem cell domain does not affect maintenance of sutures, further pointing to a largely embryonic role of Tcf12 in suture regulation.

To investigate the role of the murine Tcf12 gene in the development of craniosynostosis, we generated a Tcf12 null allele by crossing Tcf12^flox/flox mice with EIIa-Cre, a general deleter active in the germline (Lakso et al., 1996). It had previously been reported that homozygous Tcf12^−/− mice are born at the expected frequency from a Tcf12^+/− incross, with the vast majority dying by 2 weeks after birth (Zhuang et al., 1996). We similarly observed that most homozygous Tcf12^−/− mice died within the first 2 weeks. In order to examine the morphology of the coronal suture, we carried out Alizarin Red staining on the skulls from homozygous Tcf12 mutants. We found partial fusion of the coronal suture in 5 of 10 Tcf12^−/− mice (50%) at embryonic stages [embryonic day (E)17-E18] and 8 of 22 Tcf12^−/− mice (36%) surviving into postnatal stages [postnatal day (P)0-P21]; no coronal suture defects were observed in 35 heterozygous Tcf12^+/− and 26 wild-type sibling mice (Fig. 1; Movies 1, 2; Table 1,). The mutant parietal bones displayed an abnormal overlap at the sagittal suture at E18, P0 and P2 stages. We also observed developmental delay and rare exencephaly in Tcf12^−/− mice, as reported by Zhuang, et al. (1996), with the frequency of exencephaly increasing from 3% to 33% when one allele of Twist1 was also deleted (Table 2). In addition, Tcf12^−/− mice displayed kinks and curls of the tail, a phenotype known to be caused by a moderate delay in posterior neuropore closure (van Straaten and Copp, 2001). Tcf12^−/−; Twist1^+/− mice also infrequently displayed an open ventral body wall not observed in Tcf12^−/− mice (Table 2). The suture phenotypes and low-penetrance exencephaly and associated defects suggest that Tcf12 functions in both calvarial suture development and neurulation.

To investigate where Tcf12 is required for suture development and skull bone growth, we deleted both copies of Tcf12 in neural crest (Tcf12^flox/flox; Wnt1-Cre), mesoderm (Tcf12^flox/flox; Mesp1-Cre) or both (Tcf12^flox/flox; Wnt1-Cre; Mesp1-Cre). Whereas homozygous loss of Tcf12 in either neural crest or mesoderm alone did not affect coronal suture patency, loss in both tissues caused coronal suture fusion (Fig. 2A-H). The severity of coronal suture defects appeared to be higher in combined conditional mutants (coronal synostosis index 1.7±1.8; mean±s.e.m.) than in conventional homozygous Tcf12 mice (coronal synostosis index 0.5±0.2), although the severity of synostosis was highly variable in combined conditional deletion mutants. Given a similar requirement for Twist1 in both mesoderm- and neural crest-derived tissues (Teng et al., 2018), these results support Tcf12 and Twist1 acting together in the same tissues for coronal suture formation.

When Tcf12 was deleted in the mesoderm, we observed an abnormal overlap of the parietal bones of the sagittal suture (Fig. 2C,D), consistent with accelerated growth of the mesoderm-derived parietal bones (Fig. 2C) and similar to what was seen in Tcf12 nulls (Fig. 1G,J,M). Upon mesodermal deletion, we also observed an expansion of the size of the parietal bone relative to the neural crest-derived frontal bone at P21, effectively moving the coronal suture anteriorly (Fig. 2G,I). Reciprocally, the size of the frontal bone expanded relative to the parietal bone upon neural crest deletion of Tcf12, shifting the coronal suture posteriorly (Fig. 2F,I). This relative size change was associated with increased apical growth of the frontal bone compared with the parietal bone rudiments in Tcf12^flox/flox; Wnt1-Cre mice, and a trend toward increased parietal versus frontal bone rudiment growth in Tcf12^flox/flox; Mesp1-Cre mutants, as revealed by alkaline phosphatase (ALP) staining at E13.5 and E14.5 (Fig. 2J-P).

At E18.5 and P1, the apical expansion of the frontal bones in Tcf12^flox/flox; Wnt1-Cre mice was also associated with the appearance of Alizarin Red+ heterotopic bones in the normally unossified area when compared with controls (Fig. 3A-G). Sectioning of neural crest-derived mesenchyme (Tcf12^flox/flox; Wnt1-Cre; R26RLacZ) at E15.5 showed that the ectopic bone is entirely of neural crest origin (Fig. 3H-O). Thus, the expansion of the frontal relative to the parietal bone upon neural crest-specific Tcf12 loss is due, at least in part, to accelerated osteoblast differentiation within neural crest-derived cells.

We showed previously that closer apposition of frontal and parietal bones in Twist1^+/−; Tcf12^+/− mutants was associated with increased proliferation yet reduced number of putative progenitors at the growing bone fronts and the developing coronal suture (Teng et al., 2018). Given the accelerated frontal and parietal bone growth we observed upon Tcf12 deletion, we assessed whether osteoblast and progenitor dynamics were also altered in Tcf12 null mice. At E14.5, we observed an ∼22% increase in the number of Sp7+ osteolineage cells along the frontal and parietal bone fronts of Tcf12 mutants, yet the percentage of Sp7+ osteolineage cells incorporating bromodeoxyuridine (BrdU; a marker of cell proliferation) and the total number of proliferative Sp7− cells were unchanged (Fig. S1). These findings indicate that Tcf12 and Twist1 function in common to limit osteogenic commitment.

We next examined the number of putative osteoprogenitors at the developing coronal suture in Tcf12 mutants. We had previously shown that Grem1+ cells, which function as skeletal stem cells in long bones (Worthley et al., 2015), were reduced in the coronal suture mesenchyme of Tcf12^+/−; Twist1^+/− mice (Teng et al., 2018). Interestingly, we observed that Grem1+ cells were asymmetrically distributed around the bones at the forming coronal suture, with more Grem1+ cells above (i.e. closer to skin) than below (i.e. closer to brain) relative to the frontal bone, and conversely more Grem1+ cells below than above the parietal bone in wild types at E14.5 (Fig. 4A-G). This asymmetric distribution of Grem1+ cells in the ectocranial versus endocranial layers correlates with the parietal bone reproducibly overlapping above the frontal bone at the mature coronal suture. In Tcf12 null mice, we observed selective reductions of Grem1+ cells above the frontal bone (Fig. 4A-G), resulting in a more symmetric arrangement of Grem1+ cells. Grem1+ cells above and below the parietal bone were also reduced in mutants. Consistently, we observed that Tcf12 and, in particular, Twist1 expression were also asymmetrically enriched in mesenchyme above versus below the frontal bone and below versus above the parietal bone tip at E15.5 (Fig. 4H-J, arrows). We also attempted to examine whether Grem1+ cell asymmetry was established earlier at E13.5, a stage before the parietal and frontal bones meet, but did not observe significant Grem1 expression at this stage. Instead, we examined the expression of Six2, which we had recently shown broadly labels coronal suture mesenchyme (Farmer et al., 2021). At E12.5 and E13.5, we observed more Six2 expression above the frontal bone than above the parietal bone (Six2 expression in the dura below the bones prevented us from assessing relative expression below the frontal and parietal bone) (Fig. 4K,L). These results indicate selective enrichment of Grem1+/Six2+ mesenchyme above the frontal bone compared with the parietal bone before these bones contact to form the coronal suture. The loss of asymmetry in Tcf12 mutants further correlates with the frontal and parietal bones meeting end-on-end and often fusing.

Disruption of the boundary between the osteogenic neural crest cells of the frontal bone and the mesoderm-derived coronal suture mesenchyme was observed in Twist1 heterozygous mice and was accompanied by a reduction in Eph-ephrin signaling in the ectocranial layer above the coronal suture (Merrill et al., 2006). We used Wnt1-Cre; R26R mice to trace the fate of neural crest-derived cells at the coronal sutures of E13.5 embryos (Jiang et al., 2002). In wild types, lacZ+ neural crest-derived cells contributed to the frontal bone and were excluded from the mesoderm-derived coronal suture mesenchyme and parietal bone. In contrast, neural crest-derived cells invaded the coronal suture mesenchyme in Tcf12^−/− mutants, consistent with a defect in the neural crest-mesoderm boundary (Fig. 5A-F). We next examined ephrinA2-EphA signaling in Tcf12 null mice, as we had previously shown that this pathway is required in the ectocranial layer above the forming suture to regulate neural crest-derived cell migration and boundary formation (Merrill et al., 2006; Ting et al., 2009; Yen et al., 2010). In Tcf12^−/− mutants, ectocranial expression of Efna2 was completely abolished (Fig. 5G-J). We conclude that Tcf12 and Twist1 act similarly to regulate Efna2 ectocranial expression and the boundary between osteogenic and non-osteogenic cells at the nascent coronal suture.

We next investigated potential roles for Tcf12 in regulating suture patency at postnatal stages. Using RNAscope in situ hybridization of 14-week-old mouse skulls, we observed postnatal expression of Tcf12 throughout the mesenchyme of the coronal suture, as well as in periosteum above and dura mesenchyme below the suture (Fig. 6C). As the suture mesenchyme expression of Tcf12 overlaps with the reported labeling by Gli1-CreERT2 at 4 weeks after birth (Zhao et al., 2015), we treated Gli1-CreERT2; Tcf12^flox/flox mice at this stage with tamoxifen to delete Tcf12 in the postnatal suture mesenchyme. Four weeks after induction, micro-CT and histological analysis revealed no defects in the coronal suture when compared with untreated wild-type controls (Fig. 6A,B,D,E). At this same stage, in situ hybridization for Tcf12 mRNA revealed an ∼60% reduction in expression within the suture region, which likely includes non-skeletogenic cells not derived from Gli1+ skeletal stem cells (Fig. 6F-H). This finding argues against a major postnatal role of Tcf12 in Gli1+ suture mesenchyme for maintaining suture patency, although we cannot rule out postnatal roles for Tcf12 in cells outside of the Gli1+ domain.

The ability of heterozygous mutations in TWIST1 or TCF12 to cause coronal synostosis in Saethre-Chotzen syndrome, as well as genetic interactions of Twist1 and Tcf12 homologs in zebrafish and mice, had suggested that Twist1 and Tcf12 may similarly control several key processes in calvarial development. However, Twist1 can also form homodimers, as well as heterodimers with other partners besides Tcf12, suggesting that there could be differences between Twist1 and Tcf12 roles during coronal suture development. Here, we show that Twist1^+/− and Tcf12^−/− mutants share many, but not all, defects related to coronal suture formation, suggesting a primary role for Twist1-Tcf12 heterodimers in this process.

We find that Tcf12^−/− mutants display the same partially penetrant coronal synostosis as Twist1^+/− mutants, which is similarly preceded by accelerated frontal and parietal bone growth and disruption of the boundary between sutural and osteogenic cells (Merrill et al., 2006; Yen et al., 2010; Ting et al., 2009; Teng et al., 2018). The observation of coronal suture fusion by Alizarin Red staining as early as E17.5, as well as the similar penetrance and severity of suture defects at late embryonic and postnatal stages, points to defects in the formation as opposed to the maintenance of the coronal suture in Tcf12^−/− mice. At a molecular level, this is reflected by an increase in Sp7+ osteolineage cells and a concomitant decrease in Grem1+ mesenchyme cells at embryonic bone fronts, in addition to loss of ectocranial Efna2 expression. As with Twist1, Tcf12 must be mutated in both the neural crest and mesoderm lineages to disrupt the coronal suture. These similarities suggest that Tcf12 and Twist1 may act together to control common gene targets and cellular processes that are important for proper coronal suture development. We did not, however, observe the same increases in osteoblast and/or progenitor proliferation in Tcf12 mutants as reported for Twist1^+/− mutants (Teng et al., 2018). This difference could be explained by the milder suture fusion seen in Tcf12^−/− mutants, and/or a selective requirement of Twist1 homodimers for restricting the proliferative expansion of bone-forming cells.

Our results suggest several potential mechanisms that may lead to synostosis upon loss of Tcf12 and/or Twist1. Mutation of either gene results in excess osteoblast production and acceleration of frontal and parietal bone growth. This is clearly seen upon inactivation of Tcf12 in the neural crest. In these animals, small disorganized islands of bone form in the frontal fontanelle, the space in which the posterior portion of the parietal bones and anterior portion of the frontal bones come together. These bony islands arise from neural crest-derived cells of the frontal foramen, reminiscent of our previous findings that inactivation of Msx1 and Msx2 in neural crest causes inappropriate conversion of non-osteogenic neural crest-derived cells to osteoblasts (Roybal et al., 2010). Why only inactivation of Tcf12 or Twist1 in both the neural crest and mesoderm causes synostosis is unclear. In one model, stem cell depletion in one lineage can be compensated for by stem cells from the other lineage. Hence, depletion in both lineages would be required to bring stem cell numbers below the threshold necessary to maintain separate bones. Consistent with this model, we observe depletion of Grem1+ mesenchyme cells associated with the bone fronts in both Twist1^+/− and Tcf12^−/− mutants, though future lineage tracing will be required to determine whether Grem1+ cells act as osteogenic stem cells as described in the limbs (Worthley et al., 2015). In addition to acceleration of bone growth, disruption of either Tcf12 or Twist1 also results in a disruption of the boundary between the frontal bone and the suture mesenchyme that is associated with loss of overlying Efna2+ ectocranial expression. Whether this boundary defect is a secondary consequence of the accelerated bone growth, or is an independent phenotype that further contributes to synostosis, requires further investigation.

Further arguing for a largely embryonic role of Tcf12 in suture patency, deletion of Tcf12 in Gli1-expressing cells of the 1 month postnatal coronal suture had no effect on suture maintenance one month later. A similar lack of postnatal suture phenotype was recently reported for mice in which Twist1 was conditionally deleted by Gli1-CreERT2 at similar stages (Yu et al., 2021). However, expression of Tcf12 in the coronal suture and surrounding tissues was not completely eliminated in Gli1-CreERT2; Tcf12^flox/flox mice following postnatal tamoxifen administration, potentially due to incomplete activity of CreERT2 and/or expression of Tcf12 in some Gli1-CreERT2− lineage cells. Given the continued expression of Tcf12 in coronal suture mesenchyme until at least 14 weeks after birth, future experiments using different tools to completely eliminate postnatal Tcf12 function will be needed to better assess requirements for Tcf12 after birth.

Our results also provide a potential explanation for the overlap of the frontal and parietal bones at the coronal suture. Unlike the medially placed interfrontal and sagittal sutures, which are formed by the direct apposition of osteogenic fronts on a level plane, the transversely situated coronal suture is reproducibly formed by the parietal bone overlapping above the frontal bone. In wild types, we find a preferential enrichment of Grem1+/Six2+ mesenchyme cells above the frontal bone and below the parietal bone, similar to Tcf12 and Twist1 expression and our recent reports of Erg expression (Farmer et al., 2021). We find that enrichment of Six2 above the frontal but not above the parietal bone begins as early as E12.5, when the bones are still distant from one another. One possibility is that these asymmetries skew the growth of the two bones off their central axes such that the parietal reproducibly overlaps above the frontal bone. In mice deficient for Tcf12 or Twist1, the osteogenic fronts of the frontal and parietal bones fail to overlap, and this correlates with a more symmetrical distribution of Grem1+ cells (this study) and Six2+/Erg+ cells (Farmer et al., 2021) above and below the growing frontal bone. In contrast, distribution around the parietal bone was less affected. One explanation for the reduction in frontal but not parietal bone mesenchymal asymmetry is loss of the Efna2+ ectocranial layer in mutants. The ectocranial Efna2+ layer overlays the enriched Grem1+/Six2+ cells of the frontal bone but does not significantly extend into the parietal side of the coronal suture. It is therefore possible that the ectocranial layer serves to enrich Grem1+/Six2+ mesenchyme above the frontal bone in normal animals, with loss of this ectocranial layer in mutants disrupting asymmetry and leading to increased apposition of bone fronts that could further contribute to synostosis.

In summary, our study points to similar roles for Twist1 and Tcf12 in controlling embryonic processes essential for coronal suture formation, thus explaining why Saethre-Chotzen syndrome can be caused by heterozygous inactivating mutations in either TWIST1 or TCF12. The similarities of defects in heterozygous Twist1 and homozygous Tcf12 mutant mice further support the model that Twist1 and Tcf12 function as heterodimers to control unique sets of genes important for regulation of osteoblast progenitors. Future experiments will be required to identify genomic regions specifically bound by such heterodimers, as well as the molecular and cellular mechanisms by which these transcription factors regulate the timing and rate of calvarial bone addition to ensure proper coronal suture formation.

The Institutional Animal Care and Use Committee of the University of Southern California approved all animal experiments (Protocol 20552). We did not separate mice for experiments by gender, and wild-type and mutant mice were maintained on a C57BL/6 background. The Tcf12 (Wojciechowski et al., 2007), Twist1 (Chen and Behringer, 1995), R26R (Soriano, 1999), Wnt1-Cre (Danielian et al., 1998), Mesp1-Cre (Saga et al., 1999) and Gli1-CreERT2 (Zhao et al., 2015) alleles were genotyped by PCR as described. Gli1-CreERT2; Tcf12-fl/fl mice were injected intraperitoneally with 20 mg/ml tamoxifen in corn oil (Sigma-Aldrich, C8267) for three consecutive days starting at 1 month of age as previously described (Yu et al., 2021).

Craniosynostosis is most easily assessed at postnatal stages (e.g. Yen et al., 2010). Because Tcf12 homozygous mutants often die perinatally (before P6), we typically used P0 animals. Some mice survived into the first postnatal week, and very rarely until P21. When we obtained such mice, we scored them for craniosynostosis and made them a part of our analysis. We also used embryonic stages to assess early events of suture development. At E14-E15, the frontal and parietal bone rudiments are coming into apposition and osteogenic cells and putative osteogenic stem cells are identifiable by marker gene expression (Ting et al., 2009; Yen et al., 2010; Teng et al., 2018; this study).

Skulls of newborn (P0) and postnatal mice were stained for bone with 2% Alizarin Red S in 1% KOH for 3-5 days. The specimens were then cleared and stored in 100% glycerol. Embryonic skulls were stained with 2% Alizarin Red S in phosphate buffered saline (PBS) for 1 h, destained for 1 h and imaged using a Leica MZ12.5 dissecting microscope. MicroCT imaging of postnatal skulls was performed using identical conditions as previously described (Yu et al., 2021).

E13.5 and E14.5 heads were fixed in 4% paraformaldehyde (PFA) in PBS, and these were then bisected midsagitally after fixation. Presumptive calvarial bones were stained with NBT and BCIP (Roche, 11681451001). To detect lacZ, skulls of late-stage embryos were fixed in cold 4% PFA for 30 min followed by PBS washes. Calvarial bones were then stained with β-galactosidase staining solution overnight at 37°C.

BrdU labeling to assess cellular proliferation at E14.5 was carried out as previously described (Teng et al., 2018). BrdU (Sigma-Aldrich, B9285) was injected into pregnant females (200 μg/g body weight) 2 h before dissection. Heads of embryos were embedded in Optimal Cutting Temperature (OCT) medium (VMR, 25608-930) before sectioning. Frozen sections were cut at 10 µm. Immunohistochemistry was performed using rat anti-BrdU (Bio-Rad Laboratories, MCA2060, 1:1200), rabbit anti-Sp7/Osx (Santa Cruz Biotechnology, sc-22536-r, 1:500), rabbit anti-ephrin-A2 (Santa Cruz Biotechnology, sc-912, 1:200) and goat anti-Grem1 (Thermo Fisher Scientific, PA5-47973, 1:40) diluted in 1% bovine serum albumin (BSA)/PBS and incubated overnight at 4°C. Detection of primary antibody of anti-Brdu, anti-Osx, anti-ephrinA2 and anti-Grem1 was performed by incubating goat anti-rat FITC (Santa Cruz Biotechnology, sc-2011, 1:150), goat anti-rabbit Alexa Fluor 568 (Abcam, ab1175471, 1:500), donkey anti-rabbit Alexa Fluor 568 (Abcam, ab1175470, 1:200) and donkey anti-goat IgG-Alexa Fluor 488 (Abcam, ab150129, 1:1200), respectively, for 1 h at room temperature followed by 4,6-diamidino-2-phenylindole (DAPI) counterstaining and examination by epifluorescence microscopy. Analysis of β-galactosidase activity of Wnt1-Cre/R26R reporter gene expression was carried out as previously described (Merrill et al., 2006). Postnatal calvaria were dissected and fixed in 10% neutralized buffered formalin (NBF, Sigma-Aldrich, HT501128) overnight at room temperature, then decalcified with 10% EDTA in 1× PBS for 1-14 days based on mouse age. Decalcified calvaria were dehydrated and embedded in paraffin. Tissue blocks were sectioned at 5 μm using a microtome (Leica) and mounted on SuperFrost Plus slides (Fisher, 12-550-15). Hematoxylin and Eosin (H&E) staining was completed following standard protocols.

For postnatal calvaria, tissues were fixed in 10% NBF overnight at room temperature, decalcified with diethyl pyro carbonate (DEPC)-treated 10% EDTA in 1× PBS at 4°C, dehydrated sequentially with 15% sucrose in 1× PBS (4°C, overnight) and 30% sucrose in 1× PBS/OCT (4°C, overnight), and embedded in OCT on liquid nitrogen immediately. Frozen tissue blocks were sectioned at 8 μm on a cryostat (Leica, CM3050S) and mounted on SuperFrost Plus slides. Probes were purchased from Advanced Cell Diagnostics: Tcf12 (504861); Twist1 (414701-C2); Six2 (500011); Sp7 (403401-C3). Colorimetric in situ RNA detection was performed using RNAscope^® 2.5 HD Detection Reagent (Advanced Cell Diagnostics, 322360) according to instructions provided by the manufacturer. Fluorescent RNAscope in situ hybridization for embryonic and adult time points was performed with the RNAscope Multiplex Fluorescent Kit v2 (Advanced Cell Diagnostics, 323100) according to the manufacturer's protocol for the fixed-frozen sections. Heat antigen retrieval was omitted for embryonic tissue to maximize nuclei quality, and TSA^® Plus (PerkinElmer Fluorescein, NEL741001KT; Cy3, NEL744001KT; Cy5, NEL745001KT) reagents were used at 1:750 dilution. As the exon 18 deleted transcripts of the Tcf12-flox allele still retain some sequence recognized by the Tcf12 RNAscope probe, the 60% reduction of Tcf12 transcripts in tamoxifen-treated Gli1-CreERT2; Tcf12^flox/flox animals may be an underestimate of deletion efficiency in the postnatal suture mesenchyme.

The severity of coronal synostosis was quantified using a scoring method adapted from Oram and Gridley's craniosynostosis index (Oram and Gridley, 2005; Yen et al., 2010) applied specifically to the coronal suture (Teng et al., 2018). For each skull, the two coronal sutures were scored. The extent of fusion was assessed microscopically (0: unfused; 1: <50% fused; 2: ≥50% fused; 3: 100% fused). Scores for left and right sutures were added. Thus, for example, the maximum score given to two coronal sutures is 6. Phenotypic scoring was performed blinded to genotype. Comparisons among genotypes were analyzed by an unpaired two-tailed Student's t-test. For the measurement of coronal suture position, and hence the relative proportions of the parietal and frontal bones, we took the ratio of the sagittal suture to the metopic suture. The sagittal suture was defined as the distance between the juncture of the sagittal and lambdoid suture and the juncture of the sagittal and coronal suture. The metopic suture was defined as the distance between the juncture of the metopic and coronal suture and the anterior end of the frontal bones. For the measurement of frontal and parietal bone size at embryonic stages, the area of developing bones were quantified by outlining bones using ImageJ software and normalizing the size to that of the neighboring eye (Table S1). For assessment of BrdU+, Sp7+ and Grem1+ cells in mouse coronal suture area, stained-positive cells in a defined area along the bone fronts were manually counted (Tables S2 and S3). Four to five sections from the medial to lateral aspects of the coronal suture were quantified per animal and averaged. Comparisons were analyzed by an unpaired two-tailed Student's t-test, and the P-values were corrected for multiple testing with Bonferroni correction. For quantification of RNAscope signal, sutures were defined as all cells between neighboring frontal and parietal bones. Individual foci were defined as single transcripts, and each transcript was assigned to the nearest DAPI+ nucleus by manually counting in ImageJ (Table S4). Comparison of transcript count per cell was analyzed by an unpaired two-tailed Student's t-test. GraphPad Prism v8 was used for statistical analyses.

We thank members of the Maxson, Crump and Chai labs for helpful discussions.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.199575.