Mesenchymal Wnts are required for morphogenetic movements of calvarial osteoblasts during apical expansion Free

The calvaria are a collection of intramembranous bones that form the roof of the skull and protect the brain and other sensory organs. Between each calvarial bone is a mesenchymal suture, which allows for growth and expansion of the skull after birth. Changes in the growth dynamics of calvaria can impact the positions of the sutures and may contribute to congenital calvarial defects such as craniosynostosis and calvarial dysplasias (Flaherty and Richtsmeier, 2018; Miraoui and Marie, 2010b; Ornitz and Marie, 2002; Rawlins and Opperman, 2008; Teng et al., 2018; Wilkie and Wall, 1996). These defects are detrimental to brain and sensory organ development (Morriss-Kay and Wilkie, 2005; Wilkie, 1997). Despite the clinical importance of the size and shape of calvaria, we lack a basic understanding of morphogenetic processes by which they are formed. Murine calvarial bones are akin to those of humans; hence, the mouse embryo is an ideal system for identifying the cellular mechanisms and cues that govern cell movement and apical expansion of mammalian calvarial osteoblast progenitors.

Mammalian calvarial osteoblast progenitors originate from cranial mesenchymal cells derived from cranial neural crest cells (CNCC) and paraxial head mesoderm (PM). The CNCC migrate from the early neural tube to the supraorbital arch (SOA) beginning by mouse embryonic day (E) 8.5. A subset of this SOA mesenchyme is fated to form the skeletogenic mesenchyme, which includes frontal and parietal bone primordia between E10.5 and E12.5 (Deckelbaum et al., 2012; Tran et al., 2010; Yoshida et al., 2008). At E11.5-E12.5, calvarial osteoblast progenitors of frontal and parietal bone primordia begin to condense in foci above the eye and express bone fate transcription factors such as Runx2. Between E12.5 and E15.5, calvarial osteoblasts commit and express osterix (Osx/Sp7) while expanding apically (cranially) to cover the brain. Although the transcriptional program of mammalian calvarial osteoblast differentiation has been extensively studied (Ang et al., 2022; Fan et al., 2021; Ferguson and Atit, 2018; Ishii et al., 2015), less is understood about the signals and cellular processes that control the morphogenesis and apical expansion of calvarial osteoblast progenitors from the SOA.

Apical expansion and growth of calvaria appears to be a temporally regulated process that can be perturbed by spatio-temporal pharmacological blockade or genetic deletion of various regulatory factors (Ferguson and Atit, 2018; Ferguson et al., 2018; Jiang et al., 2019). Agenesis at the apex of the skull and frontal bone dysplasia are common phenotypes observed after alterations in signaling and transcription factors within the cranial mesenchyme during early calvarial morphogenesis (Ferguson et al., 2018; Goodnough et al., 2016; Jiang et al., 2019; Matsushita et al., 2009; Roybal et al., 2010). For example, the bone transcription program is tightly regulated by several key signaling pathways, including Wnt, BMP, SHH and FGF signaling, which, when dysregulated, often result in defects in calvarial growth (Ang et al., 2022; Fan et al., 2021; Ferguson and Atit, 2018; Ishii et al., 2015). Temporal conditional genetic deletion of transcription factors between E8.5 and E10.5, such as Twist1, Msx1/2 and epigenetic regulators such as Ezh2 in cranial mesenchyme result in loss of mineralized bone at the apex of the skull (Dudakovic et al., 2015; Ferguson et al., 2018; Goodnough et al., 2016; Roybal et al., 2010). As these factors have been associated with apical expansion and growth of calvaria, reduction in expansion may underlie insufficient apical bone formation. However, it is not clear whether calvarial morphogenesis results from various mechanisms regulated by multiple pathways or whether a common downstream cellular mechanism regulates apical expansion.

There are a few putative cellular mechanisms that may contribute to rapid apical expansion of the calvaria (Ishii et al., 2015). One possibility is ‘end addition’, in which cells from ectocranial mesenchyme contribute to the growing osteogenic front under the regulation of ephrins (Ishii et al., 2015; Merrill et al., 2006). However, lineage tracing shows only few cells potentially participate in such a mechanism (Merrill et al., 2006), suggesting additional processes are involved in apical expansion. A second possibility is that spatio-temporally biased proliferation may contribute to apical expansion of calvarial osteoblasts, as in other examples of vertebrate morphogenesis (Heisenberg and Bellaïche, 2013; Stooke-Vaughan and Campàs, 2018; Stuckey et al., 2011). In a calvaria ex vivo culture model, inhibition of proliferation leads to a 44% decrease in cell division but only a 19% decrease in growth, suggesting that proliferation alone is insufficient to account for the apical expansion of maturing calvaria (Lana-Elola et al., 2007). A third potential mechanism is that of morphogenetic cell movement, such as radial cell intercalation (vertical); growth can occur in both axes of the plane, as seen during epiboly in frog and fish (Walck-Shannon and Hardin, 2014). This can be combined with convergent-extension through medio-lateral cell intercalations, as in closure of the neural tube in the frog (Huebner and Wallingford, 2018; Keller et al., 2008; Wallingford and Harland, 2002). Radial intercalation and convergent-extension would be consistent with the spreading of a sheet-shaped structure and narrowing of the osteoblast progenitor pool as they progress apically from the SOA, respectively. A fourth possibility is that directional movement of individual cells, either at different velocities along the apical axis of extension or in a bipolar apicobasal fashion, elongates the calvarial primordium.

There is evidence for long-range displacement of calvarial osteoblasts. When Engrailed 1 (En1)-expressing cells were genetically labeled in the SOA at E10.5, they were subsequently found throughout the frontal and parietal bones, among adjacent ectocranial cells, and in the overlying dermis at E16.5 (Deckelbaum et al., 2012; Tran et al., 2010). Consistent with published studies of DiI-labeled mouse calvarial bone primordia at E13.5, the descendants of En1-marked cells in the SOA were visible long distances towards the apex (Ting et al., 2009; Yoshida et al., 2008). By both techniques, labeled cells were found much further apically than could readily be accounted for by proliferation alone, strongly supporting the possibility of morphogenetic cell movement during apical expansion of calvarial bones (Tran et al., 2010). However, cellular behaviors and potential tissue-scale cues that might orient efficient expansion of calvarial osteoblast progenitors in the SOA are unknown.

In this study, we found that multiple non-canonical Wnt ligands are expressed in cranial mesenchyme and are required for cellular behaviors engaged in morphogenetic movements and directional migration towards the apex during calvarial morphogenesis. Wnt5a contributes to cranial expansion primarily by permitting basal cell rearrangements within the SOA and directional cell movements further apically. This working model provides new insights into the basis of calvarial defects.

Calvarial osteoblast progenitors that form the murine frontal and parietal bones expand apically from the SOA region between E10.5 and E16.5 (Fig. 1A) (Deckelbaum et al., 2012; Tran et al., 2010; Yoshida et al., 2008). Between E13.0 and E14.5, we observed that length-width ratios of frontal bone primordia in coronal sections increased from 5:1 to 10:1 (Fig. 1B,C). During this period, a lack of apoptotic cells has been reported within the primordial tissue (Fig. S1) (DiNuoscio and Atit, 2019; Goodnough et al., 2016). To examine the potential role of cell proliferation in this expansion, we tested the proportion of mesenchymal cells in S phase. Uptake of 5-ethynyl-2′-deoxyuridine (EdU) was similar in basal, intermediate and apical regions of the frontal bone primordium at E13.5 and E14.5 (Fig. S1A,B), an observation that is consistent with data from later stages (Lana-Elola et al., 2007). Therefore, the spatial distribution of proliferating cells does not explain the directional and apically narrowing/radial thinning nature of calvarial expansion. Interestingly, cell shape analysis of E13.5 Osx⁺ calvarial osteoblasts in coronal plane, using the membrane marker concanavalin, identified a progressive increase in cellular elongation that corresponded to progressive apical narrowing/radial thinning of the frontal bone primordium (Fig. 1D). Spinning disk confocal images of intact embryos show cranial mesenchyme cells in the basal position have protrusive activity that decorate the cell surface (Fig. 1E). In the apical position, cranial mesenchyme cells become elongated and polarized, with protrusions present on the apical side of the cells (Fig. 1F). We postulated that dynamic cellular behaviors remodel the SOA mesenchyme as new cells are added.

To examine dynamic tissue and cell behaviors, we studied an earlier stage of SOA expansion at E10.5, when calvarial osteoblasts precursors are present (Kuroda et al., 2023). Using wild-type transgenic embryos that harbor the reporter mTmG (Muzumdar et al., 2007) without Cre recombinase (Wnt5a^+/−;R26-mTmG), we visualized cell membranes in red under live conditions (Tao et al., 2019) and applied an updated strain (deformation) analysis algorithm that we have described previously (Tao et al., 2019; https://github.com/HopyanLab/Strain_Tool). Briefly, strain was mapped in time-lapse movies by tracking numerous points frame-to-frame and correlating images using a sub-pixel registration algorithm (Guizar-Sicairos et al., 2008). Delauney tessellation was then applied to generate a triangular mesh of the tissue of interest upon which area-weighted averages in strain along orthogonal axes were color coded onto the first frame. In the basal, relatively broad, region immediately cranial to the eye, mesenchyme compressed along the mediolateral axis (E_xx) in a medial region and expanded along that axis in the lateral region, while extending apically along the apicobasal axis (E_yy).

In contrast, the narrower, most apical mesenchyme predominantly extended along the apicobasal axis, whereas an intermediate region contracted apicobasally. The tissue rotated (reflected by shear: E_xy) in the basolateral region where it wrapped partially around the orbit and, notably, within the mid-part of the SOA mesenchyme between narrowing baso-medial and extending apical regions in the coronal plane (Fig. 2A, Movie 1). These complex tissue-scale deformations suggest that morphogenetic movements remodel the SOA.

After activating relatively bright green fluorescence of mTmG-labelled cell membranes using Wnt1:Cre (Wnt5a^+/−;Wnt1:Cre;R26-mTmG) in the SOA mesenchyme, we rendered cell membranes in 3D using Imaris. Mesenchymal cell intercalations within the baso-medial region were identified where convergent extension takes place. Cells exchanged neighbors such that, in the most basic situation, an individual cell entered or exited a surrounding group of about four to six other cells (Fig. 2B, Movie 2). That process is akin to a 3D T1 exchange that was described for foams (Tao et al., 2019b; Weaire et al., 2012) and has been recognized in mandibular arch mesenchyme (Tao et al., 2019b). It is not currently possible to track and reliably quantify the relative proportion, complexity or orientation of these types of 3D intercalations because multiple cells in each identifiable group participate to variable degrees in rearrangements within neighboring groups. Instead, we propose that Golgi positions relative to nuclei offer reasonable proxies for quantification of the spatial distributions of cell orientation (as shown below). We infer that rearrangements among basal cells tend to liquify and permit remodeling of basal mesenchyme.

In contrast to the basal region, cells within the apical region were elongated along the apicobasal axis. To visualize filipodia that are inserted between neighboring cell membranes, we sparsely converted mTmG-labeled cell membranes from red to green to identify them using a unique Wnt1:Cre strain (Wnt5a^+/−;Wnt1:Cre;R26-mTmG) that has spontaneously become mosaically expressed in our colony. Cell protrusions oriented towards the apex or base of the SOA preceded movement of the cell body in the same direction (Fig. 2C, Movies 3 and 4). Based on their elongated shapes (Fig. 1D), these bipolar cell movements should tend to elongate relatively apical mesenchyme along the apicobasal axis. At early stages, we propose that a transition of cellular behaviors from basal rearrangements to apical bipolar shear-type movements contributes to remodeling of the SOA.

Non-canonical Wnt pathway and Wnt5a/11 ligands are key regulators of cell polarity, convergent-extension and directional cell movement (Butler and Wallingford, 2017; Gray et al., 2011; Zallen, 2007). After modifications by Wntless, Wnt ligands are secreted from the cell and bind to cell-surface Frizzled receptors and to the co-receptors ROR2 and VANGL2 to initiate a signaling cascade (Andre et al., 2015; Davey and Moens, 2017). We observed robust expression of the key signaling components Wnt5a and Wnt11 (Goodnough et al., 2014), the co-receptor ROR2 and phosphorylated Jun, a downstream effector of non-canonical Wnt signaling, throughout the cranial mesenchyme between E11.5 and E13.5 (Fig. S2A-I). Interestingly at E11.5, VANGL2, a regulator of cytoskeleton-mediated changes in cell polarity, was readily identifiable by immunostaining in basal, but not apical, mesenchyme, suggesting that pathway function, like cell behaviors, is different in those two regions (Fig. S2F,G). Overall, these data suggest that non-canonical Wnt/PCP signaling is active within cranial mesenchyme during the remodeling of the SOA.

To test for Wnt/PCP pathway function, we conditionally deleted mesenchymal Wntless to prevent secretion of all Wnt ligands and circumvent functional redundancy. We used tamoxifen-inducible Pdgfrα:Cre-ER to delete Wntless in cranial mesenchyme (CM-Wls^fl/fl or Pdgfrα:Cre-ER; Wls^fl/fl) (Fig. 3A). As Cre-recombination occurs within 24 h of oral gavage initiated at E8.5, we obtained efficient deletion of Wls mRNA by E12.5 (DiNuoscio and Atit, 2019; Ibarra et al., 2021). Using this method, the cranial mesenchyme loses its ability to secrete Wnt ligands; however, the ectodermal Wnts are sufficient to maintain canonical Wnt pathway activation through E12.5 (DiNuoscio and Atit, 2019; Ibarra et al., 2021) (Fig. 3A). Fate mapping of PdgfraCre-ER-expressing cells between E9.0 and E10.5 showed β-galactosidase⁺ cells were present broadly in the cranial mesenchyme, including the frontal bone primordia and adjacent dermis in controls and in CM-Wls^fl/fl mutants at E14.5 (Fig. 3B,C).

At E14.5 and E16.5, CM-Wls^fl/fl mutants exhibited diminished expansion of the forelimb and craniofacial structures, and a curly tail, phenocopying mutants of other core-PCP components (Gao et al., 2011) (Fig. S4A-D). CM-Wls^fl/fl mutants expressed the early calvarial osteoblast markers Runx2, Osx and alkaline phosphatase (AP) (Fig. 3B-I), indicating that mesenchymal Wnts are not required for cell fate specification and early commitment to the osteoblast lineage. The proportion of Runx2- and Osx-expressing cells per region of interest was comparable in the basal and apical regions of the frontal bone primordia (Fig. 3J,K). In contrast, ectodermal Wnt-dependent activation of cranial mesenchymal canonical Wnt/β-catenin signaling is required for expression of Runx2 and Osx for osteoblast differentiation and inhibition of Sox9 expression and cartilage fate by E12.5 (Goodnough et al., 2012, 2014). Thus, we infer that the CM-Wls mutant phenotype here is primarily the result of a lack of noncanonical Wnt signaling.

The normalized apical extension and area of AP-positive frontal bone primordia was significantly diminished among CM-Wls^fl/fl mutants relative to controls at E13.5 (Fig. 3L, Fig. S3C-E). The decrease in apical expansion of frontal bone primordia in CM-Wls^fl/fl mutant embryos was not attributable to cell proliferation index, cell density or cell survival, which were comparable with the controls (n=4-7, two or three litters) (Fig. S3F-J). We also measured the individual Osx⁺ cell nuclei length in micrometers in the coronal plane at E13.5, and estimated the amount of apical shortening that can be accounted for by change in cellular level length. The control nuclei were 19-21 μm and mutants were 16.5-18.5 μm, the differences in which were statistically significant (unpaired Student's t-test, P=0.015) (n=4 embryos/genotype, two litters). We found that the shortened mutant nuclei account for half of the normalized 25-40% difference in the apical extension between controls and CM-Wls^fl/fl mutant (Fig. 3L), suggesting other variables are contributing to the extension. Consistent with apical shortening, we found the widest region of the frontal bone in the coronal plane was 102-104 μm in the controls and 121-155 μm in the CM-Wls^fl/fl mutants (n=3 embryos/genotype, two litters).

The versatility of the conditional inducible Pdgfrα:Cre-ER allowed us to diminish the extent of Cre-recombination within cranial mesenchyme. By significantly decreasing the dose of tamoxifen (by ∼25-fold) at E8.5, we limited the number of cells in which Cre was activated, effectively generating a mosaic population of mutants. We used R26R β-gal to monitor Cre-ER activity within the CM, starting at E12.5 (Fig. 3M,N). Mosaic loss of Wls in a small number of mesenchymal cells was sufficient to attenuate apical expansion of frontal bone primordia, as measured by AP staining (Fig. 3O-Q). Next, we used the PDGFRα:Cre-ER;R26 tdTomato reporter to monitor the distribution of Cre-ER⁺ descendants. Similar to control embryos, tdTomato⁺ cells were distributed throughout the Osx⁺ frontal bone primordium, in the adjacent dermal mesenchyme and cranial mesenchyme that is further apical to the calvarial osteoblast domain at E13.5 (Fig. S4A-E). Together, these data indicate that mesenchyme Wls is not required for osteoblast commitment to the bone lineage but for apical expansion of the frontal bone primordium in a non-cell autonomous manner.

The preceding data suggest that Wls regulates paracrine cues that affect cellular parameters to remodel SOA mesenchyme. To study the cellular shapes of calvarial osteoblasts in mutants, we examined the nuclear length-width ratio of calvarial osteoblasts because it is feasible and consistently tracks cell shape (Chen et al., 2015; Green, 2022) (Fig. 1D). Runx2⁺ (at E12.5) and Osx⁺ (at E13.5) nuclei shape elongation was reduced and circularity was increased in the basal and apical regions of mutant embryos (Fig. 4A,B, Fig. S3K). The enrichment of filamentous actin in the cranial mesenchyme apically, as assessed by phalloidin staining intensity, was also diminished (Fig. 4C-E), indicating CM-Wnts contribute to cellular elongation and cytoskeletal organization.

As the positions of Golgi with respect to nuclei correspond to axes of cell polarity (Ede and Wilby, 1981; Yadav et al., 2009), we measured GM130⁺ Golgi-nuclei angles among calvarial osteoblasts within the Osx⁺ domain of E13.5 frontal bone primordia. With respect to the midline of the brain as a reference axis, Golgi-nuclei angles were distributed more broadly in the basal region than in the apical region (Fig. 4F,G). In the intermediate and apical regions, cells were increasingly bi-polar along the apical-basal axis, consistent with our quantitative observation that protrusive activity seems bipolar. Comparatively, more mutant cells exhibited a basal-ward bias in all regions and the distribution of Golgi-nuclei angles was significantly different in CM-Wls mutants in each region (Fig. 4F,G). These data indicate that cell polarity is progressively biased along the apical-basal axis of growth and the repositioning of the Golgi complex to the front of the nuclei is dependent upon CM-Wls.

A likely paracrine effector in the SOA mesenchyme is WNT5A (Fig. S3). Using Wnt5a^−/− embryos labeled with the transgenic membrane reporter mTmG (Wnt5a^−/−;R26-mTmG), convergent-extension-type remodeling of the basal tissue and extension of apical tissue were diminished by time-lapse light sheet microscopy and strain analysis (Fig. 5A, Movie 5). At the basal aspect of the mutant SOA, mean strain was similar to that of wild-type embryos (Fig. 5B). However, the variance from the mean (root mean square) of strain among individual foci was diminished in mutants (E_xx root mean square Wnt5^+/−=0.18, Wnt5a^−/−=0.12, unpaired Student's t-test P=0.02, n=3 embryos; E_yy root mean square Wnt5^+/−=0.18, Wnt5a^−/−=0.13, unpaired Student's t-test P=0.03, n=3 embryos), indicating that tissue remodeling was relatively incomplete. For quantitative assessment at cellular resolution, we again depended upon the altered Golgi positions among mutant cells (Fig. 4G). The qualitative correlate of those data based on time-lapse imaging of three separate Wnt5a^−/−;Wnt1:Cre;R26-mTmG embryos revealed a paucity of cell rearrangements in 3D. Although cell shapes fluctuate, outright neighbor exchange is not observed (Fig. 5C, Fig. S5A,B, Movies 6-8) in marked contrast to wild-type embryos in which neighbor exchange was readily observed (Fig. 2B). In apical mesenchyme in the coronal plane, mediolateral narrowing (E_xx) and apicobasal extension (E_xy) were diminished among Wnt5a^−/− embryos. Consistent with this finding and with altered cell orientation based on Golgi positions (Fig. 4G), protrusive activity among apically located mutant cells was qualitatively less abundant and multidirectional, rather than bipolar (Fig. 5D, Movie 9). We conclude that Wnt5a is required to remodel the SOA mesenchyme by facilitating morphogenetic cell movements that restrain mediolateral expansion of the base while promoting elongation of the apical aspect of the SOA mesenchyme.

Collectively, our work reveals that calvarial osteoblasts exhibit two different modes of cell movement and complex cellular behavior that are regionalized along the baso-apical axis during calvarial morphogenesis in vivo. Mechanistically, we found these behaviors, such as cell intercalation, cellular elongation, protrusive activity and filamentous actin enrichment are achieved via non canonical cranial mesenchyme Wnts and, in particular, Wnt5a (Fig. S6). Thus, our data improve upon our understanding of the physiological and pathological processes of directional expansion in mammalian morphogenesis, and its involvement in calvarial defects.

Despite the even distribution of mitotic cells within the calvarial primordia, remodeling during growth is achieved by cell movements and cell shape changes. Other examples of this type of growth control in mouse embryonic mesenchyme include the limb bud (Boehm et al., 2010; Gros et al., 2010; Wyngaarden et al., 2010) and mandibular arch (Tao et al., 2019), and epithelial examples include the anterior visceral endoderm (Trichas et al., 2012) and kidney tubules (Lienkamp et al., 2012). The transition of basal cell rearrangements to directional apical cell movements underscores how, as morphogenetic mechanisms alter tissue shape, different intercellular strategies are employed. Examination of calvarial expansion ∼3 days later at E13.75 has identified an intriguing mechanism of anisotropic calvarial growth dominated by ECM expansion rather than cytoskeletal activity (Dang et al., 2023 preprint; Feng et al., 2024). We infer that mechanisms of expansion evolve temporally as well as spatially during calvarial expansion.

A requirement for the noncanonical Wnt pathway in a paracrine fashion for two distinct modes of cell movement in vivo in the adjacent basal and apical region of SOA mesenchyme is interesting. It confirms the recognized role of the non-canonical Wnt pathway in facilitating cytoskeletal organization and contractions (Davey and Moens, 2017; Shindo, 2018; Shindo et al., 2019). However, our data imply the pathway does not necessarily control the type or orientation of contraction. Cell shape oscillations that drive junctional rearrangements (Gorfinkiel, 2016; Levayer and Lecuit, 2013; Tao et al., 2019) and protrusive activity that drives directional migration (Carmona-Fontaine et al., 2008) represent overlapping yet distinct cytoskeletal activities (Huebner and Wallingford, 2018). Although the SOA and frontal bone primordia does not elongate normally in CM-Wls mutants, our evidence suggests a non-cueing role for mesenchyme Wnts. Consistent with the broad distribution of the Wnt5a and Wnt11 ligands in the cranial mesenchyme, neither basal cells nor apical cells were unidirectionally oriented in control embryos. Furthermore, loss of Wnt5a diminished but did not abolish both the basal junctional rearrangements and the bipolar nature of the more apically located cells and protrusive activity. Those observations raise the possibility that the noncanonical Wnt pathway is permissive for morphogenetic cell movements and that other factors, such as tissue-scale geometric constraint or forces (Lau et al., 2015; Yamada and Sixt, 2019), physical (Aigouy et al., 2010; Shellard and Mayor, 2021; Zhu et al., 2020) and/or other extracellular cues (Feng et al., 2024; Koca et al., 2022; Yamada et al., 2022), potentially orient tissue growth.

A recent study identified differentially expressed genes in Wls-deficient mid-nasal process (MNP) mesenchyme using single cell-RNA-sequencing approach (Gu et al., 2022). They identified several well-known targets of canonical Wnt signaling, such as Msx1, Msx2, Lef1, Twist1, Pax3 and Fgf8, that also have important roles in midline facial development (Goodnough et al., 2012; Gu et al., 2022; Ishii et al., 2003; Song et al., 2009). Msx 1, Msx2 and Twist1 transcription factors have major roles in calvarial growth and early bone differentiation (Chai and Maxson, 2006; Goodnough et al., 2016; Ishii et al., 2005; Roybal et al., 2010). Gu et al. also found that mRNA for a non-canonical Wnt ligand, Wnt5a, was downregulated in Wls deficient MNP mesenchyme (Gu et al., 2022). Msx2 null mutants have diminished apical expansion of the frontal bone primordia that was a result of a specification defect to the Runx2⁺ osteogenic lineage at E12.5 and a proliferation defect visible at E14.5 in the primordia (Ishii et al., 2003). Msx1/2 compound null mutants lack frontal and parietal bone primordia (Han et al., 2007). Twist1^+/−; Msx1^−/− compound mutants have more severe apical expansion defects of the frontal bone and thought to work in parallel biological pathways (Ishii et al., 2003). Our studies on CM-Wls mutants reveal a role for non-canonical Wnt signaling in calvarial morphogenesis and apical expansion. The CM-Wls mutants have specific facial deformations that are seen in Wnt5a-null embryos, such as flattened snout and domed head (Yamaguchi et al., 1999). These craniofacial defects are distinct from those produced by the conditional cranial mesenchyme deletion of canonical Wnt/β-catenin signaling in Twist1, ectoderm-Wls, Msx2^−/− and Twist1^+/−; Msx1^−/− mutants (Goodnough et al., 2012, 2014, 2016; Ishii et al., 2003; Tran et al., 2010). The CM-Wls mutants do not often survive beyond E15.5 and we did not find proliferation defects in the Runx2- and Osx-expressing osteoblasts between E12.5 and E14.5. We also did not find significant differences in the number of Runx2⁺ and Osx⁺ cells in fixed ROIs in the basal and apical regions of the frontal bone primordia. We cannot fully rule out specification and differentiation defects in terms of total calvarial osteoblast cell numbers from analysis of our sections in coronal view. Thus, we infer that the craniofacial defects of the CM-Wls mutants are result of non-canonical Wnt/PCP signaling and Wnt/calcium signaling, which are required for directional tissue morphogenesis, but have yet to be fully addressed in calvarial morphogenesis.

Mutations in several major signaling pathway genes and transcription factors that perturb calvarial growth apically contribute to calvarial defects such as dysplasias and enlarged fontanelles, and are implicated in loss of sutures (Miraoui and Marie, 2010a; Ornitz and Marie, 2002; Rawlins and Opperman, 2008; Wilkie and Wall, 1996). In humans and mice, heterozygosity at TWIST1 (Ghouzzi et al., 1999) and loss of its downstream effector, EPHA4, lead to diminished calvarial growth and are accompanied by craniosynostosis (Merrill et al., 2006). Twist1, Epha4 and Twist1-EphA4 mouse mutants show loss of segregation of osteogenic and non-osteogenic cells at the suture boundary, presumably owing to mismigration of osteogenic cells in the coronal suture and, consequently, ectopic differentiation of suture mesenchyme cells to osteoblast fate (Ting et al., 2009). Activating mutations of FGFR2 in mouse and human cause craniosynostosis with patent metopic/frontal bone fontanelle (Chen et al., 2003; Flaherty et al., 2016; Holmes and Basilico, 2012). The role of cell movement and directional cell behaviors in congenital calvarial bone defects, such as craniosynostosis and dysplasias, and in craniofacial bone diseases, such as craniometaphysial dysplasia and craniodiaphyseal dysplasia, remain to be analyzed.

We show that morphogenetic cell movements in vivo and associated cellular behaviors are new players to calvarial bone morphogenesis and apical expansion. The role of cell movement and multiple modes of cell movement in calvarial defects remains to be investigated and could advance our morphogenetic understanding of congenital defects.

Analysis was carried out using the following mouse strains from Jackson Laboratory: mTmG Gt(ROSA)26Sortm4 (007676), (ACTB-tdTomato-EGFP)Luo/J (007576), 129S4.Cg-Tg(Wnt1-cre)2 Sor/J (022137), CAG::myr-Venus (16604528), Wnt5a^+/− (10021340), PdfgrαCreER (018280) (Rivers et al., 2008), R26R-β-gal (022616), R26TdTomato (007905) and Wntless^fl/fl mice (Carpenter et al., 2010) (a gift from Richard Lang, Cincinnati Children's Hospital Medical Center, OH, USA) were used in this project. All mouse lines were outbred to CD1 and genotyped as previously described for PDFRαCreER/+;Wntless^fl/fl embryos; PDFRαCreER/+; Wntless^fl/fl males and R26R/R26R; Wntless^fl/fl females were time mated at night. The day of vaginal plug detection was counted as embryonic day (E) 0.5. Cre alleles were transmitted to experimental embryos via male breeders. To induce CreER recombination, tamoxifen at either 25μg or 0.125μg tamoxifen/g body weight dose (Sigma T5648) was given by oral gavage twice to pregnant dams carrying E8.5 and E9.5 embryos. Tamoxifen was dissolved in corn oil and gavaged in the late afternoon of the designated day. For each experiment, a minimum of three controls and three mutant mice from at least two different litters were used. Case Western Reserve University Institutional Animal Care and Use Committee and Animal Care Committee of Hospital for Sick Children approved all animal procedures in accordance with AVMA guidelines (Protocol 2013-0156, Animal Welfare Assurance No. A3145-01).

Three-dimensional (3D) time-lapse microscopy was performed on a Zeiss Lightsheet Z.1 microscope. Embryos were suspended in a solution of DMEM without Phenol Red containing 10% rat serum and 1% low-melt agarose (Invitrogen) in a glass capillary tube. Once the agarose had solidified, the capillary was submerged into an imaging chamber containing DMEM without Phenol Red, and the agarose plug was partially extruded from the capillary until the region containing the embryo was completely outside the capillary. The temperature of the imaging chamber was maintained at 37°C with 5% CO₂. Images were acquired using a 20×/1.0 objective with single-side illumination, and a z-interval of 0.479 µm that was automatically calculated based on the numerical aperture (1.0) of the objective. Images were acquired for 2-3 h with 3 min intervals. Fluorescent beads (Fluorospheres 1 µm, Thermo Fisher, 1:10⁶) were used as fiducial markers for 3D reconstruction and to aid in drift-correction for cell tracking. Multi-view processing was performed with Zen 2014 SP1 software to merge the three separate views and generate a single three-dimensional image. Further analysis and cell tracking were performed using Vision4D (Arivis) software. Spinning disk confocal microscopy in whole embryos was performed to visualize SOA cranial mesenchyme cells using the Quorum WaveFX-X1 (Quorum Technologies), as described previously (Tao et al., 2019).

Timelapse 3D datasets of cell membranes were processed with the ImageJ macro Tissue Cell Segment Movie (kindly provided by Dr Sébastien Tosi from the Advanced Digital Microscopy Core Facility of the IRB Barcelona) to generate membrane segments of embryos expressing the cell membrane reporter ROSA26;mTmG before analysis with Imaris 9.0 software (Bitplane). Surface objects were created using Imaris and, using the ‘Add New Cells’ function, cell boundaries from a region of interest were detected using the local contrast filter type and fluorescence intensity/quality thresholds. Cell tracking was performed using the autoregressive motion algorithm. Analysis was performed on three separate areas in basal and apical regions of the supraorbital arch over two independent experiments for each condition.

Tracking of tissue deformation was performed over 16 time steps in the rostrocaudal-mediolateral plane of time-lapse light sheet movies of the supraorbital arch. For each sample, tracking was performed for ∼50 points by image correlation of a 24×24 pixel box centered on each point in subsequent images. Image correlation was carried out using a subpixel registration algorithm explained by Guizar-Sicairos et al. (2008). Once points were tracked, we applied Delauney tessellation to generate a triangular mesh from the initial positions representing the initial tissue shape. We then calculated strain in the horizontal and vertical directions for each triangular element, and assigned to each point the area weighted average of elements, including that point. The custom tool used for deformation tracking is available at https://github.com/HopyanLab/Strain_Tool. Box and whisker plots show median and inter-quartile range, and were obtained with default settings for matplotlib.pyplot.baxplot. An unpaired Student's t-test was performed using python library scipy.stats.t_test.

Heads of E12.5-E14.5 embryos were drop-fixed in 4% paraformaldehyde (PFA) for 30-45 min, respectively, at 4°C and cryopreserved as previously described (Atit et al., 2006). Embryos were cryosectioned at 14 μm in the coronal plane of the frontal bone primordia. For immunofluorescence, cryosections were dried at room temperature for 10 min, washed in 1×PBS and blocked in 10% species-specific serum with 0.01% Triton for 1 h. Primary antibodies were diluted in blocking buffer and slides were incubated overnight at 4°C. The next day, the cryosections were washed in 1×PBS, incubated with species-specific fluorescent secondary antibodies (below) for 1 h at room temperature, washed with DAPI 0.5 μg/ml and mounted with Fluoroshield (Sigma F6057). The following primary antibodies were used for immunofluorescence experiments: rabbit anti-Osx (1:2000 or 1:4000; Abcam ab209484, RRID:AB_2892207), rabbit anti-caspase 3 (1:250; Abcam ab13847; RRID:AB_443014), goat anti-Runx 2 (1:100; R&D Systems AF2006-SP; RRID:AB_2184528) and mouse anti-GM130 (1:100; Thermo Fisher Scientific BDB610822, RRID:AB_398141). Appropriate species-specific Alexa-Fluor secondary antibodies were used (Alexa 488, 1:500, Thermo Fisher A32790; RRID: AB_2762833 and Alexa 594, 1:800 Thermo-Fisher A11012; RRID: AB_2534079). For immunohistochemistry staining against GM130 (BD Dickinson), Vector M.O.M. kit (BMK2202) was used for blocking before incubating with primary antibody, Diva-Decloaker (Biocare) was used for heat-mediated antigen retrieval for 15 min and Vector ABC elite kit was used for tertiary amplification followed by staining with DAB. For concanavalin A membrane staining, sections were dried at room temperature, rinsed in Hanks buffered salt solution (HBSS), stained in 10 μg/ml concanavalin-A (Thermo Fisher, C11253) diluted in Hanks balanced buffered solution for 30 min at room temperature. Sections were rinsed in HBSS, counterstained with DAPI (1:2000) and mounted with Fluoroshield. Phalloidin-488 conjugate (1:1000) was used to visualize filamentous actin staining by incubating samples for 30 min at room temperature before rinsing cells in 1×PBS, counterstaining with DAPI (1:2000) and mounting with Fluoroshield. Alkaline phosphatase(AP) staining on whole-mount embryos and sections was performed as previously described (Ferguson et al., 2018). AP stained sections were imaged on the Hamamatsu Nanozoomer S60 Digital Whole Slide Scanner and embryos were imaged at the same magnification on a Leica MZ16F stereoscope.

Mice were administered 250 μg EdU in PBS/10 g mouse weight by intraperitoneal injection 1 h prior to sacrifice. Embryos were then collected and prepared for cryopreservation as previously stated and EdU staining was conducted using the Click-iT Plus EdU kit (Thermo Fisher Scientific, C10639) by following the manufacturer's instructions. The percentage EdU⁺ positive calvarial osteoblasts/total DAPI⁺ nuclei was calculated in the coronal sections of frontal bone primordia that was identified by morphology of condensed nuclei. Three to four different sections per embryo were used for analysis using ImageJ/Fiji, Adobe Photoshop and Cell Profiler. First, regions of interest (ROIs) in the apical, intermediate and basal regions of the frontal bone primordia were generated in Fiji. Images were then cropped in Adobe Photoshop and analyzed in Cell Profiler using the following pipeline: (1) split each image into two channels: blue and red; (2) smooth using a Gaussian filter; (3) calculate uneven background illumination using a block size of 22 for both channels; and (4) employ a global two class threshold strategy using the Otsu method for objects with a diameter of 10-30 (EdU positive cells) and 10-15 (DAPI positive cells). The number of EdU positive and DAPI positive cells were exported as a Microsoft Excel spreadsheet.

At least three sections per embryo in the coronal plane were used for all histomorphetry. The length and width of the E13.5 frontal bone primordia was determined from the Osx⁺ domain from coronal plane sections in ImageJ/Fiji. The length-width ratio of individual calvarial osteoblasts nuclei was obtained regionally from the basal, intermediate or apical region of interest in the frontal bone primordia after immunostaining for Runx2 or Osx. The length-width ratio of calvarial osteoblasts cells was obtained from Osx⁺ cells co-stained with concavalin A lectin for membranes at E13.5 (n=3 wild-type embryos). The length-width ratio and circularity of Osx⁺ nuclei was obtained at E13.5 (n=3 embryos).

For quantifying the normalized apical extension of the frontal bone, frozen sections of embryonic mouse heads in the coronal plane were stained for alkaline phosphatase as described above. Images from the Hamamatsu Nanozoomer were opened on the NDPI viewer software. Using a fixed zoom, the baso-apical length of the frontal bone primordia was measured using the freehand line tool and converted to micrometers. Similarly, the cranial length was measured from the eyelid to the midline of the brain. The percentage apical extension was calculated as the ratio of length of the frontal bone to the cranial length. Frontal and parietal bone primordia area was obtained from whole-mount alkaline phosphatase-stained embryos. Individual bone primordia area measurements were normalized to cranial area in Fiji.

Phalloidin staining of E12.5 coronal sections were performed by incubating 1×phalloidin working solution (1×PBS, 10% bovine serum albumin and Phalloidin-iFluor 488) and 1000×conjugate in DMSO (1:1000; Cayman Chemical Company, 20549) for 45 min. After washing in 1×PBS twice, the sections were stained with DAPI and mounted with fluoroshield (Sigma Aldrich, F6182). Images were captured on Olympus BX60 and corrective fluorescence was quantified in Fiji/ImageJ. For fluorescence intensity, images were transformed to 8-bit and the average of mean gray values of three circles on non-fluorescence regions was calculated as the background fluorescence. The corrected fluorescence per ROI was calculated using the following formula: corrected fluorescence+integrated density – (area of ROI×average of background fluorescence).

The GM130⁺ Golgi complex to nuclei angle was calculated in the Osx⁺ expression domain of the frontal bone primordia in the coronal plane at E13.5. The image was oriented so that the eyes were along the x-axis (0 degree) and the midline of the brain/apex was along the y-axis (90 degrees). The Golgi complex-nuclei angle was measured relative to the center of the nuclei manually using ImageJ/Fiji software. Approximately 25-30 cells in the apical, 35-45 in the intermediate and 50 cells in the basal regions of the frontal bone primordia were randomly selected from controls and CM-Wls mutants in MS Excel (n=4 embryos, three or four sections/embryo).

All wide-field images were captured using the Olympus BX60 microscope, Olympus DP70 digital camera using DC controller software. Whole-mount AP-stained embryos were imaged with a Leica MZ16F stereoscopes and Leica DFC 490 camera using Leica software. Images were processed in Adobe Photoshop 2022, ImageJ/Fiji (Schneider et al., 2012) and assembled on Adobe Indesign or Illustrator 2022.

All graphs were created using GraphPad Prism version 9.0 for Mac (https://www.graphpad.com/scientific-software/prism/). Data are presented as mean±s.d. in all graphs in Figs 3 and 4. Pairwise sample comparisons on Prism 9.0, using an unpaired two tailed t-test with Welsh's correction due to unequal variance and sample numbers where indicated. For comparison across more than two groups, one-way ANOVA was used with Bonferroni correction (Fig. 1). Circular distribution of Golgi-complex nuclei angles was analyzed using a Kolmogrov-Smirnov test developed for intracellular organization (Apte and Marshall, 2013). Data that are not significant are indicated by ‘n.s.’ or not marked with significance. The P-values for statistical tests in all figures are stated or represented as *P<0.05, **P<0.005 and ***P<0.0005.

We thank all the present and past members of the Atit Lab and Hopyan Lab for contributing ideas or effort to this project. Special thanks to Nyoka Lovelace and Mia Carr for help with animal husbandry and immunostaining, James Ferguson and L. Henry Goodnough for expression analysis of Wnt/PCP components in Fig. S2, and Yifan Zhai for frontal bone area analysis in Fig. S3. We thank the Case Western Reserve University-bio[box] shared instrumentation facility for microscopes and the chemidoc gel reader, and the Imaging Facility at the Hospital for Sick Children for access to lightsheet and scanning disk confocal microscopes.