Apical expansion of calvarial osteoblasts and suture patency is dependent on fibronectin cues

The roof of the mammalian skull is comprised of a quartet of paired intramembranous bones, the frontal and parietal; in addition the mouse has the interparietal bone (Carter and Anslow, 2009; Ferguson et al., 2018; Ishii et al., 2015). Where these bones meet, fibrous joints, or sutures, are formed that provide areas for continued growth during postnatal stages (Zhao et al., 2015). Fate mapping in mice shows that the frontal bones originate from the cranial neural crest, whereas the parietal bones are derived from the paraxial mesoderm (Jiang et al., 2002). However, both the frontal and parietal bone primordia emerge as foci in the supraorbital arch mesenchyme (SOM) above the eye and commit to the osteoblast lineage (Ferguson et al., 2018; Jiang et al., 2002; Yoshida et al., 2008). Through development, the osteoblasts in the calvarial anlagen establish an osteogenic front and expand to the apex of the head to cover the brain (Fan et al., 2016). The mechanisms of how the calvarial osteogenic progenitors expand to their respective locations in a determinative manner and form without deformation are not understood.

Label based tracing ex utero as well as inducible genetic lineage mapping of the mouse SOM cells between embryonic day (E)10.5-E13.5 showed that SOM cells contribute to the frontal and parietal bones and adjacent dermal mesenchyme by E16.5 (Deckelbaum et al., 2006; Tran et al., 2010; Yoshida et al., 2008). The distribution pattern of labeled cells in these studies suggest that active cell movement may play a role in expansion of calvaria. Iseki et al. and Opperman et al. demonstrated that the population of osteoprogenitors located at the osteogenic fronts of the bone anlagen are highly proliferative (Iseki et al., 1999; Opperman et al., 2000). These findings are shadowed by the fact that inhibiting cell proliferation of the calvarial explants ex-vivo only diminished parietal bone growth by less than 20% (Lana-Elola et al., 2007). These data suggest that proliferation is not a prominent driver of apical expansion of the calvaria. There are small populations of ephrin-Eph-positive ectomesenchyme cells added directly at the osteogenic front in late development that provide complementary mechanisms for growth, but these are not sufficient to drive apical expansion (Ting et al., 2009). From these data, we hypothesize that the apical expansion of osteoprogenitors over a long distance results from directional cell migration.

Directional migration of calvarial osteoblast progenitors towards the skull apex would require guidance cues to establish coordinated cell activity. One would anticipate any such cues to be polarized or graded across the embryonic cranial mesenchyme (CM). RNA-seq analysis at E12.5 suggests that genes related to cell adhesion and extracellular matrix (ECM)-receptor interaction pathways are differentially expressed in mesenchyme populations located apically to the SOM (Dasgupta et al., 2019). The interstitial ECM comprises a set of fibrillar proteins such as collagen, fibronectin and laminin, as well as polysaccharide macromolecules. During embryonic tissue development, the ECM plays dual roles in providing structural support and regulation of signaling pathways important for cell movement and tissue growth (Walma and Yamada, 2020). Differential expression of ECM-receptor interactions suggests a role for matrix interactions and context in determining cues as well as substrates to guide directional cell movements (Shellard and Mayor, 2021a; Walma and Yamada, 2020; Yamada et al., 2019). Even with these insights, there remains little evidence for a fundamental role of cell matrix interaction in regulating calvarial growth and pattern, nor mechanisms of how these pathways may be dysfunctional in cases of defects. A fundamental understanding of early calvarial morphogenesis will provide new insights into the etiology of congenital defects of the skull such as craniosynostosis (CS) and fontanelles through shared biological processes that can be leveraged for therapies.

In this study, we identified fibronectin (FN1) as a key, differentially expressed ECM component during the process of apical expansion of calvaria. FN1 is an interstitial ECM protein identified as having a role in directional cell movement in vitro (Lo et al., 2000; Shellard and Mayor, 2020). We find that FN1 acts as a substrate for calvarial growth and is required to preserve suture patency of the coronal sutures. Filamentous-actin (F-actin) can serve as anchor points that mediate ECM interactions. We detail dependency of cell-matrix interactions of calvarial osteoblast progenitors by targeted deletion of Wiskott-Aldrich syndrome-like (Wasl; also known as N-Wasp), an F-actin nucleating factor in cranial neural crest cells that leads to the specific failure of calvarial osteoblasts to migrate apically (Caswell and Zech, 2018). These results are consistent with lamellipodia-mediated cell migration across an FN1 substrate leading to proper patterning of the skull roof; thereby, providing a new mechanistic model of calvarial morphogenesis.

To identify asymmetric cues present in the developing skull roof, we visualized the expression of interstitial ECM proteins during early stages of calvarial development (E12.5-E13.5) (Walma and Yamada, 2020). We initially visualized pan-collagen unfolded and denatured chains by staining with fluorescent Collagen Hybridizing Peptide, Collagen III antibody on E12.5 coronal sections (Fig. S1). Detection of these products was not regionally different in the mesenchyme. Using a broad assessment of mineralized collagen, sulfated proteoglycans and elastin with standard histological stains also did not show enrichment in the CM at E13.5 (Fig. S1). Laminin, which is a dominant ECM protein in basement membranes and a component of the fibrous networks of ECMs, was expressed in the developing meninges and blood vessels in the CM at E12.5 (Fig. S1) but not within the generalized mesenchyme. However, we found that FN1 protein was highly enriched towards the apex in the CM between E11.5 and E13.5, and ahead of the RUNX2⁺ calvarial osteoblast progenitors at E13.5 (Fig. 1A-C‴; Fig. S1). We detected a 2- to 2.5-fold difference in expression intensity of FN1 between basal and apical regions of the CM (Figs 1D-F and 2C,E). Notably, FN1 expression intensity on coronal sections quantified by line graph and 3D surface plot showed a continuous increase in FN1 expression intensity towards the apex (Fig. 1D,E). We independently validated the graded expression of FN1 in cranial mesenchyme by assessing the expression of fibulin 1 protein (Fig. S1), which is associated with assembled FN1 protein fibers in the ECM (Balbona et al., 1992). We found that fibulin 1 protein was enriched in the CM toward the apex, similar to the graded expression of FN1. Together, these results show that, among the core fibrous interstitial ECM components (Walma and Yamada, 2020), the expression of FN1 stands out because it is enriched and graded towards the apex in CM, in the direction of calvarial expansion (Fig. 1F), suggesting that it may provide external directional cues for calvarial osteoblasts.

The expression of FN1 in the cranial mesenchyme along the baso-apical axis could function as a tissue level directional substrate for guiding the calvarial osteoprogenitors. Humans with Apert syndrome have calvarial growth defects and CS, and suture fibroblasts from these patients secrete higher FN1 in culture (Bodo et al., 1997). We visualized FN1 protein expression in CM of the Apert syndrome mouse model in coronal and transverse planes before suture fusion at E13.0-E13.5 (Fig. 2B; Fig. S2) (Chen et al., 2003; Holmes and Basilico, 2012; Wang et al., 2005). Expression along the baso-apical axis in the frontal bone primordia at E13.0 showed significantly elevated FN1 expression in all three regions of the CM in Apert mice compared with controls (Fig. 2A,B,D,E). However, FN1 expression in Apert mutants remained graded towards the apex (Fig. 2C-E). Similarly, FN1 expression appears to be elevated qualitatively in the coronal suture mesenchyme in Apert mutants (Fig. S2). To further investigate the relationship between FN1 expression in the CM and the apical expansion of calvarial bones, we quantified the apical length of the alkaline phosphatase (AP)-positive domain, an early osteogenic marker, across the frontal bone primordia at E13.0 (Ishii et al., 2015). Relative to litter-matched control embryos, we found the relative apical expansion of frontal bone primordia in Apert mutants was significantly increased (Fig. 2F-I). Together, these data suggest that elevated expression of FN1 in the CM correlates with increased expansion of the embryonic frontal bone primordia.

FN1 plays many crucial roles in tissue growth and morphogenesis during development and FN1-integrin interactions are required for differentiation of osteoblasts in vitro (Globus et al., 1995; Yamada et al., 2019). Whether and how FN1 functions in cellular behavior and differentiation of osteoblasts during calvarial development and patterning in vivo is unknown. To assess this regulation in early specification of the calvaria, we conditionally deleted Fn1 in the CM in mouse between E10 and E11.0 using tamoxifen-inducible Pdgfra-CreER-mediated recombination (CM-Fn^fl/fl) (Fig. 3A) (Ferguson et al., 2018). We validated the efficiency of Fn1 deletion in cranial mesenchyme at E13.5 and found a significant decrease of Fn1 mRNA by qPCR and loss of FN1 protein expression by immunofluorescence staining in the CM (Fig. 3B-D). To determine functional outcomes of FN1 deletion, we quantified the cell proliferation index of calvarial osteoblasts between E12.5 and E14.5 in the primordia of the frontal bone, as defined by morphology/OSX (SP7) expression, and found that it was not significantly different between controls and CM-Fn^fl/fl mutants (Fig. 3E-H). Similarly, cell survival in the CM-Fn^fl/fl cranial mesenchyme at E12.5 was not altered (Fig. S3). We next tested whether osteoblast lineage specification and commitment was affected in CM-Fn^fl/fl (Nakashima et al., 2002). Similar to prediction by cell proliferation and viability, the discrete number and percentage of RUNX2⁺ osteoblast progenitors in three non-overlapping regions in the frontal bone primordia was comparable in the control and CM-Fn^fl/fl embryos (Fig. 3I-L). The discrete number of OSX⁺ cells was elevated in the basal and intermediate regions, but not in the apical region at E14.5, though the percentage of OSX⁺ committed osteoblasts was comparable in control and CM-Fn^fl/fl mutants along the entire baso-apical axis of the frontal bone primordia (Fig. 3M-P). Consistently, the nuclei density in the frontal bone primordia was significantly increased in CM-Fn^fl/fl mutants relative to the controls, notably at the basal and intermediate region of interest (ROI) in the frontal bone primordia (Fig. S3). These results are consistent with a failure of cells to extend apically, pooling at the more basal domain of the primordia when FN1 expression is absent. We did observe that the spatial domain of OSX expression was markedly diminished in the absence of FN1 (Fig. 3M,N). We measured the normalized apical extension, the length-width ratio of the frontal bone primordia and the frontal bone area marked by AP expression at E14.5 in sections or wholemount staining (Fig. 3Q-S; Fig. S3). There was a significant decrease only in the normalized apical length and frontal bone area in CM-Fn^fl/fl at E14.5 (Fig. 3Q-S; Fig. S3).

To investigate the robustness of the dependency of calvarial expansion on FN1 protein, we titrated the tamoxifen down by 100-fold (0.25 μg/g bw) to create a knockdown model of FN1 that was initiated at E9.5 (CM-Fn^KD) (Fig. 3T). The relative quantity of Fn1 mRNA in E13.5 cranial mesenchyme was decreased but comparable in the CM-Fn^KD mutants to controls (Fig. 3T). Compared with FN1 protein expression in controls at E13.5, FN1 protein expression was either absent in patches as ‘potholes’ or diminished in expression overall in the CM-Fn^KD mutants (Fig. 3U-W; Fig. S3). The percent apical expansion of the frontal bone primordia in the E13.5 embryos was significantly decreased in all CM-Fn^KD mutants compared with the controls (Fig. 3X-Z). All together, these results show that FN1 expression in CM is required for efficient apical expansion of the frontal bone progenitors without impacting cell proliferation, cell survival or osteogenesis (Fig. 3AA).

During the process of tissue morphogenesis, cells often become polarized and change shape, which can be coordinated by cues from the ECM (Yamada et al., 2019). However, it is not known whether calvarial osteoblast progenitors share these characteristics during apical expansion and if such behavior is dependent of FN1. The asymmetric position of the condensed Golgi complex is a well-known indicator of cellular polarity and is a required aspect of directed cell movement in a variety of cells in culture (Drabek et al., 2006; Yadav et al., 2009). Migrating cell types can position the Golgi complex forward or rearward of the nucleus and this positioning is greatly impacted by geometric constraints in fibronectin substrate (Pouthas et al., 2008; Serrador et al., 1999). The Golgi-nuclei angle distributions of calvarial osteoblasts in the control frontal bone primordia have a bipolar bias in distribution along the axis of growth, suggesting they are under mechanical and geometric constraints and capable of bidirectional movement (Fig. S4) (Shih and Keller, 1992; Zallen and Wieschaus, 2004). In the CM-Fn1^fl/fl mutants, the calvarial osteoblasts maintain the bias, but the Golgi-complex position was rearward of the nucleus as seen by the significant difference in the angle distributions in the basal and intermediate regions (Fig. S4). In the apical region of the frontal bone primordia, the Golgi-nuclei angle distribution became more unipolar and was highly biased toward the apex in controls and CM-Fn^fl/fl mutants, suggesting that FN1-independent variables are in play (Fig. S4). Thus, the bipolar bias in the cellular polarity of calvarial osteoblasts occurs along the baso-apical axis of growth and the deletion of FN1 either leads to the loss of a directional cue or the forward positioning of the Golgi complex, which is dependent on the FN1 matrix.

Next, we examined whether the FN1 matrix is required for cellular elongation of calvarial osteoblasts during apical extension of the frontal bone primordia. Mesenchymal cells have uneven boundaries; however, nuclei shape tracks cell morphology with high fidelity (Chen et al., 2015). Thus, to quantify the change in morphology, we measured the length-width ratio of calvarial osteoblast nuclei in the OSX domain at E13.5 (Fig. 4A,B). In controls, nuclei of calvarial osteoblasts become progressively elongated along the baso-apical axis. The nuclei length-width aspect ratio of calvarial osteoblasts in the basal region was between 1:1-4:1 and between 1:1-7:1 apically (Fig. 4C). The calvarial osteoblasts in CM-Fn^fl/fl mutants had significantly lower length-width ratios in the basal and apical region, indicating that cellular elongation was diminished throughout the frontal bone primordia (Fig. 4C). In the CM-Fn^KD, we found a significant decrease in cellular elongation only in the apical region (Fig. 4D). Collectively, these findings indicate that apical expansion of the frontal bone at the tissue level occurs by cellular polarization and elongation in calvarial osteoblasts and these cellular behaviors are dependent on FN1 expression.

Actin filaments are part of the force generating complex used for cell migration (Walma and Yamada, 2020; Yamada and Sixt, 2019). During cell movement, mesenchymal cells can have enriched and aligned contractile stress fibers of actin cytoskeleton associating with the ECM through polarized actin-rich focal adhesions to promote movement (Walma and Yamada, 2020; Yamada and Sixt, 2019). To determine whether actin cytoskeleton expression and organization is dependent on extracellular FN1 expression, we visualized and analyzed the topography of F-actin in the CM by Phalloidin staining. In E13.5 controls, F-actin expression was highly enriched in the CM in the domain of the frontal bone primordia (Fig. 4E,G). In individual mesenchymal cells in the basal region of the frontal bone primordia, F-actin appears to be distributed at the cell boundary (Fig. 4E,E′). In the apical region, F-actin appears to be distributed throughout the cell (Fig. 4E,E″). Relative fluorescent intensity measurement of Phalloidin in ROIs shows a significant decrease in basal and apical regions of the frontal bone primordia in CM-Fn1^fl/fl mutants (Fig. 4E-G). To visualize the topography of actin independent of fluorescence signal intensity, we generated a vector map of actin fibers in ROIs using the Alignment by Fourier Transform (AFT) algorithm (Fig. S4) (Marcotti et al., 2021). The heat map and the median order parameter representing the vector angles show F-actin fibers are anisotropic and highly aligned to each other in controls (Fig. S4). In these analyses, the median order parameter approaches one in the apical region of controls to signify perfect alignment (Fig. S4). In the basal region of CM-Fn1^fl/fl mutants, the mean order parameter is significantly lower, showing a decrease in anisotropy in F-actin fiber organization. Thus, F-actin distribution and anisotropy in alignment in calvarial osteoblasts suggest that actin-dependent cell contractility is correlated with the apical movement of the calvarial osteoblasts and, by extension, frontal bone formation. Further, these coincident processes are dependent on the presence of graded extracellular FN1.

Next, to directly test whether lamellipodia-dependent cell migration is a mechanism underlying apical expansion of calvarial osteoblasts in vivo, we conditionally deleted Wasl, a gene encoding a known effector of CDC42, and ARP2/3, which are key regulators of actin nucleation in lamellipodia (Hüfner et al., 2002; Watson et al., 2017). ARP2/3 is specifically required to direct differential dynamics of lamellipodia for cells moving up along a fibronectin gradient in vitro by haptotaxis (Hüfner et al., 2002; King et al., 2016; Marcotti et al., 2021). Conditional deletion of Wasl using the Wnt1Cre2 line specifically removes WASL function in mouse cranial neural crest cells as they emerge from the neural tube before forming progenitors of the frontal bone and other facial structures. We found that conditional deletion of Wasl led to distinct craniofacial phenotypes including cleft palate and calvarial defects without CS (Fig. 5). MicroCT imaging of Wnt1Cre2; Wasl^fl/fl mutants at P0 show mineralized but dramatically reduced size and apical extent of the frontal bone (Fig. 5A); parietal bones were generally unaffected, consistent with the different embryonic origins of these bones. The leading edge of the Wasl mutant frontal bones lacked the irregular border of mineralized bone seen in controls, hinting at different growth dynamics.

Cell-ECM interactions depend on WASL activity to form F-actin-mediated cell contacts (Hüfner et al., 2002; Watson et al., 2017). Consistent with our data from conditional deletion of FN1 substrate, calvarial osteoblast lineage differentiation indicated by AP and OSX expression was visible in frontal bone primordia, but the apical expansion of the frontal bone primordia was markedly diminished in Wasl mutants compared with controls at E14.5 (Fig. 5B-D; Fig. S5). Consistently, the length-width ratio of nuclei in the basal and apical regions of the frontal bone primordia was also significantly decreased (Fig. 5E,F).

The cell proliferation index of the osteoblasts was comparable at E14.5, but the cell density was significantly higher in the frontal bone primordia of Wnt1Cre2; Wasl^fl/fl embryos (Fig. S5). At earlier embryonic stages (E12.5), the cell proliferation index of RUNX2⁺ calvarial osteoblast progenitors was elevated in the basal region and not perturbed in the apical region of the frontal bone primordia in Wnt1Cre2; Wasl^fl/fl mutants compared with stage-matched controls (Fig. 5G,H). The percent of RUNX2⁺/ROI is also comparable between control and Wnt1Cre2; Wasl^fl/fl mutants, suggesting that the process of cell specification of osteoblasts progenitors is intact (Fig. 5G,I). Further, cell survival at E12.5, assayed by cleaved caspase 3 expression, was comparable in cranial mesenchyme of controls or mutants (data not shown). To assess the impact of Wasl deletion on cell behavior, we visualized F-actin distribution and nuclei shape of calvarial osteoblasts in E12.5 frontal bone primordia. Compared with controls, F-actin expression was markedly diminished in RUNX2⁺ calvarial osteoblasts and the adjacent Wnt1Cre2-derived mesenchyme of conditional Wasl mutant embryos (Fig. 5J,K). Thus, similar to Fn1 conditional mutants, Wasl-dependent actin nucleation within lamellipodia is required for apical expansion of the frontal bone primordia without perturbing overall osteogenesis, survival and proliferation. Collectively, these analyses expose a new role for FN1 matrix substrate and Wasl dependent-lamellipodia in calvarial osteoblasts during apical expansion of the skull roof.

Calvarial bone fibroblasts from children with CS have dysregulated FN1 expression in vitro (Baroni et al., 2002; Bodo et al., 1997). This raises the question of whether maintenance of suture patency is associated with FN1 expression, as a consequence or as an addition to calvarial bone growth. FN1 is expressed in the coronal suture mesenchyme at E13.5 (Fig. S2), so we tested whether FN1 substrate is also required for the maintenance of sutures. In control E16.5 mouse embryos, the coronal suture remained patent (Fig. 6A). In contrast, in CM-Fn^fl/fl mutants that survive till E16.5, we observed continuous Alizarin Red staining across the coronal suture space indicating premature fusion of the coronal suture unilaterally as well as bilaterally (Fig. 6B). The penetrance of the CS phenotype in FN1-deficient CM cells was high approaching 80% (Fig. 6C). In transverse sections of CM-Fn^fl/fl mutants, we confirmed mineralization of the coronal suture space (Fig. 6D-F). Correlated with this inappropriate fusion, we found the length of head and proportion of frontal bone area were decreased, normalized lateral length of the frontal bone was increased in CM-Fn^fl/fl mutants, suggesting growth dynamics defects are linked to CS phenotype (Fig. 6G-J). In addition, the anterior metopic (interfrontal) suture distance between the paired frontal bones and posterior sagittal suture distance between the paired parietal bones was significantly increased in CM-Fn^fl/fl mutants (Fig. 6K-O). We initiated recombination at E13.5 and E14.5 and collected embryos at E18.5. We did not find evidence for apical extension defects or suture fusion in the coronal suture at E18.5, suggesting that these two phenotypes are related at earlier stages during the active apical expansion stage (Fig. S6). Together, these findings indicate that FN1 functions to promote apical expansion of calvaria and dysregulation of FN1 expression leads to coincident suture fusion when bones become juxta-positioned (Figs 2 and 6).

Overall, our findings in this study allow us to build a new model for development of the skull roof in which calvarial osteoblasts use WASL-dependent lamellipodia to migrate possibly through a baso-apical net movement along a gradient of stiffness or adhesion of FN1 substrate (Fig. 6P). Further, this mechanism is required for patterning of the calvarial primordia, and consequently retention of suture patency between these bones.

The mechanisms by which the skull roof forms and is patterned in a determinative fashion is not understood. As these developmental mechanisms underlie calvaria formation, it is attractive to ask if they also contribute to a common etiology of craniofacial disorders. Using a combination of genetic tools in the mouse, we developed a framework detailing the process of calvarial expansion and its impacts on suture patency. Central to these findings is the discovery that FN1 expression is graded towards the apex in the CM, and this ECM framework is required for efficient apical expansion of the calvarial osteoblasts. Deletion of FN1 in the CM leads to altered cellular behavior without directly perturbing differentiation or proliferation of calvarial osteoblasts. Analysis of Wasl-deficient mutants shows that calvarial osteoblasts require Wasl function for cellular integration of this graded substrate signal and apical expansion, presumably through lamellipodia-actin nucleation. Collectively, our data not only uncover the role FN1 plays in both growth of the calvaria and cranial suture formation/maintenance, but also illuminate the relationship between calvarial growth and presentation of common calvarial defects such as persistent fontanelles and synostosis.

One hypothesis for directional growth of calvarial osteoblasts from the initial pool in the basal region of the CM is the interpretation of graded environmental signals toward the apex that would provide directional cues. In support of this hypothesis, we observed graded FN1 expression between E11.5 and E13.5 along the baso-apical axis of the cranial mesenchyme in and around the calvarial osteoblasts (Fig. 1). We show that mice with either elevated (Apert) or loss (CM-Fn^fl/fl) of FN1 in cranial mesenchyme have accelerated or diminished apical expansion of frontal bones, respectively. These findings suggest that graded FN1 expression may provide directional cues to guide the apical expansion of either individual or a collective group of calvaria osteoblasts in the frontal bone primordia. The absolute limits or steepness of that FN1 gradient that is required for efficient apical expansion as well as visualizing the formation of FN1 graded expression needs further investigation.

Signaling from the ectoderm and meninges are important for calvarial development and expansion, respectively (DiNuoscio and Atit, 2019; Goodnough et al., 2014; Vivatbutsiri et al., 2008). Fibronectin is expressed broadly in CM and meninges (Liesi, 1984) (Fig. 1; Fig. S1), raising the possibility that its action may impact either or both tissues. As lineage restricted Cre-deletor lines become available in osteoblast precursors and other lineages in the E9.5-E10.5 cranial mesenchyme and underlying meninges, future analysis can demonstrate which cell population secretes FN1 that is necessary for apical expansion of frontal bone.

Directional cell movement along a graded matrix substrate can be guided by a combination of adhesion/haptotaxis, mechanical/durotaxis or chemotaxis as shown in vitro and in vivo (Shellard and Mayor, 2021a,b). Fibroblasts and vascular smooth muscle cells migrate directionally in vitro in response to a mechanical stiffness gradient generated by fibronectin-coated polyacrylamide hydrogels but not by laminin (Hartman et al., 2016; Lo et al., 2000; Phuong Le et al., 2022). Interestingly, ARP2/3 is specifically required to direct differential dynamics of lamellipodia of cells moving along a fibronectin gradient in vitro by haptotaxis (King et al., 2016). This functional dependency on ARP2/3 in interpretation of fibronectin structure synergizes with our finding that loss of Wasl function, which regulates ARP2/3, leads to inability of calvarial osteoblasts to expand toward the apex. Our results are consistent with previous studies showing the involvement of upstream regulators of ARP2/3 and WASL such as Cdc42 and Rac1 in apical expansion of calvaria (Aizawa et al., 2019; Liu et al., 2013). Future studies using new mouse models will focus on which movement strategies are in play for apical expansion of calvaria and the underlying mechanism of how the graded expression of FN1 is established in cranial mesenchyme.

Altogether, our data support a model in which the skull roof forms from apical expansion of calvarial osteoblasts by directional cell migration. The variety of complex mesenchymal cell behaviors during migration within the rich 3D interstitial matrix in vivo are just emerging and still rely heavily on our understanding of in vitro studies. There is no single model to explain the complex movement and regulation of directional migration in a proliferating primordium, as multiple factors can be espoused as components of directional migration (Petrie et al., 2009). In order to build a conceptual framework for apical expansion in vivo based on the mouse conditional genetic mutants described here, we can begin to draw connections from common themes that emerge from in vitro studies. First, there is consensus that mesenchymal cells migrate by using ARP2/3-branched actin nucleated by WASL at the leading edge to generate lamellipodia and adhere to a matrix substrate (Caswell and Zech, 2018; Doyle et al., 2009; Yamada and Sixt, 2019). Supporting this logic, we find that filamentous actin is enriched and anisotropic in organization within the primordia of the frontal bone, and Wasl is preferentially required in a cell autonomous manner for apical expansion of frontal bone osteoblasts. Second, others have shown there are features of the ECM matrix, such as graded expression, adhesion, stiffness and alignment, that shed light on how ECM proteins regulate directional cellular movement (Shellard and Mayor, 2021b; Yamada et al., 2019). We demonstrate that the FN1 ECM is graded in the cranial mesenchyme towards the apex and is required for the efficient apical expansion of frontal bones in the calvaria. Future studies will need to focus on the how cells use FN1 in an in vivo context. Third, migrating mesenchymal cells and nuclei are elongated, with their actin cytoskeleton aligned to the ECM in vitro (Charras and Sahai, 2014; Faurobert et al., 2015; Friedl and Mayor, 2017; Petrie and Yamada, 2016; te Boekhorst et al., 2016; van Helvert et al., 2018). We found that the shapes of calvarial osteoblast nuclei become progressively more elongated with anisotropy in F-actin alignment along the baso-apical direction in the controls showing directional bias for this phenotype. Cellular elongation is significantly reduced in the Fn1 and Wasl mutants, as are the levels of F-actin in the frontal bone primordia. Fourth, many migrating cell types in vitro and in early development are polarized along the axis of growth, which can be visualized by the asymmetric position of the Golgi complex at the motile end in front of the nuclei or, in some cases, in the rearward position (Carney and Couve, 1989; Magdalena et al., 2003; Nemere et al., 1985; Serrador et al., 1999). We found that the Golgi complex position became more biased forward of the nuclei in the apical region along the axis of growth. Previous studies demonstrate that geometrical constraints of the FN1 matrix are key determinants of the Golgi complex position, cell polarity and cell shape (Serrador et al., 1999). Altogether, our data from in vivo studies support a model where calvarial osteoblasts orient and elongate towards the apex and migrate using WASL-dependent lamellipodia on a graded FN1 substrate leading to apical expansion calvaria during development (Fig. 6P).

There are many genes in which mutation can lead to CS in humans (Twigg and Wilkie, 2015). This disparate set of factors suggests that diverse mechanisms contribute to the pathogenesis of CS. Conversely, the broad genetic heterogeneity might suggest a shared mechanism that integrates multiple classes of signals.

We show that deletion of Fn1 conditionally in CM leads to CS. The coronal suture is preferentially lost in the FN1 mutant, similar to many well-studied CS models, such as Twist1^+/− and Apert mutants. How the suture phenotype arises and its direct dependency on FN1 is unclear. One argument is that the suture phenotype arises as a property of the altered growth of the calvaria. Teng et al. found that the frontal bones and parietal bones show increased diagonal growth in Twist1^+/− zebrafish, such that the coronal suture is vulnerable to be affected in these conditions (Teng et al., 2018). Alternatively, Holmes and Basilico suggest that the suture mesenchyme retains its identity and ectopically undergoes osteogenesis in Apert Fgfr2^+/S252W mice (Holmes and Basilico, 2012). Through the analysis of various CS models, the known cellular mechanisms underlying CS are different, including loss of boundary identity (Merrill et al., 2006) and altered timing of differentiation of suture mesenchyme (Holmes et al., 2009; Holmes and Basilico, 2012). It remains unclear whether the effect of FN1 on retaining suture patency is direct or an indirect effect of differential growth of the individual primordia.

The common phenotypic presentation of suture fusion from otherwise genetically heterogeneous cases suggests that the presentation of disease pathology may have shared developmental mechanisms. Stamper et al. (2011) suggested that perturbations in gene expression related to ECM-mediated focal adhesion occur in cases of non-syndromic single-suture CS. ECM plays an underappreciated role as the foundation for morphogenesis of the skeleton. During tissue development, ECM can directly guide cell migration as a ‘road’, but ECM can also ‘stop’ movement or restrict growth from particular destinations (Walma and Yamada, 2020). By acting to direct migration, ECM has been also shown to serve as a framework to unify the concentrations of cytokine or growth factors during chemotaxis or haptotaxis (Colak-Champollion et al., 2019; Malhotra et al., 2018; Sieg et al., 2000), restrict cells as physical barriers (Renkawitz et al., 2019; Zanotelli et al., 2019) or modify cellular migration machinery through altering adhesion, protrusion or traction force generation (Richier et al., 2018; Sekine et al., 2012; Sieg et al., 2000; Walma and Yamada, 2020; Yamada and Sixt, 2019). Taken together, our results demonstrate that FN1 is essential for calvarial osteoblast cell migration and formation of the skull roof. In its absence, altered calvarial growth occurs and leads to inappropriate suture fusion. Thus, the biological process governed by FN1 may serve as a convergence point downstream of other known pathways associated with CS disorders, thereby arriving at a similar phenotypic outcome. Our studies lend support to the concept that targeting ECM formation or remodeling may be a previously unexplored avenue for treatment strategies for calvarial growth, fracture healing and calvarial disorders.

Pdgfra-CreER (The Jackson Laboratory, #018280) (Rivers et al., 2008), Fn1^flox/flox (Fn^fl/fl) (The Jackson Laboratory, #029624) (Sakai et al., 2001), Wasl^fl/fl (MGI:3760279) (Cotta-de-Almeida et al., 2007; Hawkins et al., 2021), Wnt1Cre2 (The Jackson Laboratory, #022137) (Danielian et al., 1998), EIIaCre (Lakso et al., 1996), FGFR2^NeoS252W (Wang et al., 2005) were used to study the role of fibronectin in calvaria and suture development. For timed matings, Pdgfra-CreER/+; Fn^fl/fl males were crossed with Fn^fl/fl females and Wnt1Cre2/+; Wasl^fl/fl males were crossed with Wasl^fl/fl females overnight. All mice were genotyped as described previously and maintained on a mixed genetic background, except the Apert line which is inbred and maintained on C57BL/6J (Hawkins et al., 2021; Sakai et al., 2001; Wang et al., 2005). To yield the desired crosses, the mice were time mated overnight and checked for vaginal plugs in the morning. The vaginal plug day was assigned as E0.5. The pregnant dams were given 25 μg tamoxifen/g body weight at E9.5 and 12.5 μg tamoxifen/g body weight at E10.5 by oral gavage to induce the CreER recombination. For the knockdown model, 0.25 μg tamoxifen/g body weight was given at E9.5 and E10.5 to induce the CreER recombination. For late stage gavage, 25 μg tamoxifen/g body weight was given at E13.5 and 12.5 μg tamoxifen/g body weight was given at E14.5 to induce the CreER recombination. Tamoxifen-free base (Sigma-Aldrich, T5648) was dissolved in corn oil at 37°C in a rotating chamber and used for oral gavage within 14 days. Embryos were collected and processed as previously described (Atit et al., 2006). For each experiment, a minimum of four mutants with litter-matched controls from at least three litters were studied unless otherwise noted. Case Western Reserve University Institutional Animal Care and Use Committee approved all animal procedures in accordance with American Veterinary Medical Association guidelines (Protocol 2013-0156, Animal Welfare Assurance No. A3145-01).

E12.5-E15.5 tissue was collected and fixed in 4% paraformaldehyde (PFA) at 4°C for 35-50 min, respectively, sucrose dehydrated, embedded in O.C.T Compound (Tissue-Tek, Sakura) and cryopreserved at −80°C. The embryos were cryosectioned at 14 μm in the coronal or transverse plane using a Leica CM203. Immunofluorescence (IF) on cryosections was performed by drying slides at room temperature, washing in 1× PBS and blocking in 5-10% donkey or goat serum. Sections were incubated with primary antibodies overnight at 4°C, washed in 1× PBS, incubated with species-specific secondary antibodies for 1 h at room temperature, washed with DAPI (0.5 μg/ml) and then mounted with Fluoroshield (Sigma-Aldrich, F6182). The following primary antibodies were used: Collagen hybridizing peptide (CHP) (Hwang et al., 2017; Jussila et al., 2022) (1:250; 3Helix, BIO60), rabbit anti-Fibronectin (1:250; Abcam, ab2413; RRID:AB_2262874), rabbit anti-Fibulin 1 (1:250; Abcam, ab230994), rabbit anti-Collagen I (1:250; Abcam, ab270993; RRID:AB_2927551), Collagen III (Rockland Labs, 1:100), rabbit anti-Laminin (1:250, Millipore, L9393), 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-RUNX2 (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 secondary antibodies included Alexa 488 (1:500, Invitrogen, A32790; RRID: AB_2762833) and Alexa 594 (1:800, Invitrogen, A11012; RRID: AB_2534079). For IF staining against GM130, the Vector Laboratories M.O.M. immunodetection kit (BMK2202) was used before incubation with primary antibody.

To quantify the cell proliferation index, mice were administered 250 μg EdU in PBS/10 g of body weight by intraperitoneal injection 30 min before sacrifice. Embryos were then collected, cryopreserved and sectioned as described above. EdU was detected using Click-iTEdU Alexa Fluor 594 Imaging kit as per the manufacturer's directions (Invitrogen, C10639). The slides were then stained with 1:2000 DAPI for nuclei and mounted with Fluoroshield (Sigma-Aldrich, F6182). The percentage of proliferation cells in the frontal bone primordia (domain of condensed DAPI nuclei by morphology and expression of OSX) were quantified using ImageJ/Fiji (see below).

Alizarin Red staining on cryosections was performed on sections after post-fixing in 4% PFA, washing in 1× PBS for 5 min, quickly rinsing with ddH₂O before incubating with 2% Alizarin-S staining solution (Sigma-Aldrich, TMS-008-C) for 15 s. The slides were washed with ddH₂O and mounted with Vectashield mounting medium (Vector Laboratories, H-1400-10). RGB Trichrome staining was performed on paraffin sections that were stained as described previously (Electron Microscopy Sciences, 26357-02; Thermo Fisher Scientific, F99-10; Sigma-Aldrich, A3157) (Gaytan et al., 2020; Jussila et al., 2022). The paraffin sections were stained with Masson's Trichrome for mature Collagen I or with Van-Gieson's staining for elastin according to the manufacturer's instructions. AP staining was performed on cryosections after rinsing in PBS and NTMT (100 mM Tris, pH 9.4, 100 mM NaCl, and 50 mM MgCl₂) for 5 min. Embryos were stained with 20 μl/ml NBT/BCIP substrate (Roche, 11681451001) in the dark for 5-10 min at room temperature. Then the slides were washed with 1× PBS and mounted with aqueous mounting medium (Vector Laboratories, H5501). Slides were scanned using Hamamatsu Nanozoomer S60 by 20× brightfield scanning.

Phalloidin staining of E13.5 coronal sections were performed by incubating 1× Phalloidin working solution (1× PBS, 10% bovine serum albumin and Phalloidin-iFluor™ 488) 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). The images were captured by the Leica STELLARIS 5 confocal microscope with 63× oil immersion objective (Leica; HC PL APO 63×/1.40 OIL CS2) using Application suite X.

The brightfield images for AP staining, Alizarin Red, Masson's Trichrome staining, RGB Trichrome staining and Van-Gieson's staining as well as immunofluorescent staining images for CHP, collagen III, laminin, fibronectin, fibulin 1, OSX, caspase 3 and RUNX2, and EdU staining were captured at 10× and 20× using an Olympus BX60 microscope (UPIanFI 4/0.13) with CellSens Entry Software. Confocal images were captured using a 63× oil immersion objective on Leica STELLARIS 5 (HC PL APO 63×/1.40 OIL CS2) using Application suite X software. Images were processed in Adobe Photoshop and Fiji/ImageJ.

For whole-mount AP staining, E14.5 embryos were drop fixed in 4% PFA at 4°C overnight, rinsed in PBS, bisected in the sagittal plane, and fixed in 70% ethanol at 4°C overnight on a moving platform. Sequentially, the embryos were washed in PBST, Tris-buffered saline with 0.1% Tween-20 (TBST), NTMT buffer (described above) for 10 min, respectively. Embryos were stained with 20 μl/ml NBT/BCIP substrate (Roche, 11681451001) in the dark for 10 min on ice. Then the embryos were washed with 1 mM EDTA and fixed in 4% PFA for 10 min. The embryos were stored in 1 mM EDTA at 4°C. Images of controls and mutants were obtained at the same magnification on a Leica MZ16F stereoscope.

For Alizarin Red whole-mount staining, embryos at E16.5 were fixed in 95% ethanol and acetone at 4°C overnight before staining as previously described (Ferguson et al., 2018). Briefly, the embryos were pre-cleared in 1% KOH (Thermo Fisher Scientific, 1310-58-3) solution for 1 h and stained in 0.005% Alizarin Red (Sigma-Aldrich, A5533) dissolved in 1% KOH overnight at 4°C. Then the embryos were placed in 50% glycerol (Sigma-Aldrich, F6428-40ML): 1% KOH until clear. The embryos were kept in 100% glycerol for long-term storage. Images of controls and mutants were obtained at the same magnification on a Leica MZ16F stereoscope.

MicroCT images were taken from fixed specimens maintained in 70% ethanol using a Bruker Skyscan 1173, 240-degree scan with 0.2 rotational step, X-ray source voltage 70 kV and current sett 80 mA. Exposure time was 1500 ms. Resolution of scan was 7 voxels. Images were processed using Amira software package, version 6.0 (FEI).

The brightfield images for histological stains were captured using the Olympus BX60 microscope with a digital camera (Olympus, DP70) and CellSens Entry Software (Olympus Corporation 2011; Version 1.5) with a 10× or 20× objective (Olympus UPIanFI 4/0.13). White balance was performed before the images were captured. The immunofluorescence images were captured on the Olympus BX60 wide-field microscope with Olympus DP70 camera or Leica STELLARIS 5 confocal microscope with a 63× oil immersion objective (Leica; HC PL APO 63×/1.40 OIL CS2) using Leica Application Suite X software (4.1.1.232273). The exposure was held constant between controls and the experimental groups. Wholemount skeletal preparations were imaged with a Leica MZ16F stereoscope and a Leica DFC490 camera using Leica software. Images were processed using Adobe Photoshop and Fiji/ImageJ and page set in Adobe InDesign.

The line profile analysis was performed by Fiji/ImageJ software (Fiji/ImageJ; 2.14.0/1.54f). A Gaussian blur filter with a proper sigma (radius) value of 50 (Fig. 1D) or 20 (Fig. 2C) was applied to the images. Four ROI lines were drawn from the basal to apical region of the CM on every image. Then the resulting CSV files were imported into a custom MATLAB script in order to obtain the intensity line profiles of the FN1 expression. The custom codes are available at https://github.com/atit-lab/FN1-line-graph-analysis. The standard deviations were also incorporated into the plots as colored shaded area around the averaged intensity profiles.

At least three sections per embryo were used for histomorphometry. The percentage of cells expressing markers such as OSX, RUNX2 and EdU was measured as the ratio of marker-expressing number of cells to total DAPI-stained nuclei number in the ROI boxes in the frontal bone primordia. Corrected fluorescence intensity was calculated from immunofluorescence images using Fiji/ImageJ. For fluorescence intensity (such as FN1, Phalloidin), 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 using standard methods (McCloy et al., 2014; Shihan et al., 2021). The corrected fluorescence per ROI was calculated as integrated density – (area of ROI × average of background fluorescence).

The apical length of AP-expressing frontal bone primordia was measured using NDP view2 (Hamamatsu, U12388-01) with free-hand line and then normalized to the length of cranial length from eyelid to apex.

For GM130-stained Golgi-complex to nuclei angle distribution, the orientation of the images was adjusted to the midline of the brain (90°). A vector from GM130-stained Golgi complex to the center of the nucleus was drawn and the angle of the vector and the x-axis (0°) was measured using Fiji/ImageJ. In order to avoid oversampling, the total numbers of the measurements from each embryo and from the same regions were kept similar by using the random selection function in Excel and plotted to wind-rose graphs in Excel.

For cellular nuclei shape analysis, the nuclei were stained with DAPI (1:2000) and the images were captured with a 63× oil immersion objective using a confocal microscope (Leica STELLARIS 5). The measurements related to cell shape analysis of the nuclei in apical, intermediate and basal ROIs in the OSX⁺ frontal bone primordia were determined by CellProfiler (3.1.8) from four controls and four mutants, respectively. Maximum and minimum Feret diameters were obtained by CellProfiler. In addition to the measurements automatically assessed by the CellProfiler pipeline, the length-width aspect ratio was obtained by ⁠. Excel randomly selected 25 nuclei per embryo for each measurement to avoid oversampling. Graphs and statistical tests were made by Prism 9.

For quantification of skeletal histomorphometries, the length and width of the E16.5 skulls were photographed in the dorsal view as normalization variables (labeled with red dashed lines in Fig. 6). The anterior and posterior metopic suture distances as well as the proportion of frontal bone area (colored in gray) to the frontal region (colored in pink) were measured. The graphs and statistics were generated by Prism 9.

For all images, the boundary of frontal bone primordia was defined on the expression of OSX, RUNX2 or AP shown in the cranial mesenchyme region, or by morphology of condensed nuclei stained with DAPI that also label with OSX in the SOM.

To quantify the alignment of the Phalloidin-stained F-actin filaments, a single confocal microscope image of 0.5 μm thickness E12.5 frontal bone primordia was used. Three non-overlapping ROIs of fixed size in both the apical and basal regions of the frontal bone primordia in two histological sections of three biological replicates were analyzed. The ROIs were converted to 16-bit grayscale using FIJI/ImageJ and input into ‘Alignment by Fourier Transform’ (AFT) (Marcotti et al., 2021) using MATLAB (Mathworks, v2022a). The following parameters were used for calculating an image order parameter: window size, 30 pixels; window overlap, 50%; neighborhood radius, 2× vectors. Local masking and filtering were not applied to the images. Vector angles distribution from an ROI can be visualized as a heat map and quantified collectively as an order parameter, which normally ranges between 0 for random alignment and 1 for perfectly aligned. The vector angle distributions from one ROI from each of the three embryos were combined and reported in the heat map. The median value of order parameter from three ROIs/section were averaged in the apical and basal region of two sections per embryo and graphed.

The supraorbital cranial mesenchyme of E13.5 embryos was manually enriched and RNA was isolated as previously described (Ferguson et al., 2018; Hamburg-Shields et al., 2015). Relative mRNA expression was quantified using 4 ng of cDNA on a StepOne Plus Real-Time PCR System (Life Technologies) and the ΔΔCT method (Schmittgen and Livak, 2008). Commercially available TaqMan probes (Life Technologies) specific to each gene were used: Fn1 (Mm01256744_m1), Hprt (Mm03024075_m1) and ActB (Mm00607939_s1). The CT values were normalized to the ActB or Hprt CT value. ΔΔCT values were obtained by normalizing the ΔCT values to the average ΔCT values of the controls. Relative mRNA fold change was determined using the ΔΔCT values.

All graphs and statistical analysis were generated using Microsoft^® Excel version 16.16.27 for Mac and GraphPad Prism version 9.0.0 for Mac. Data are presented as mean±s.d. in all graphs. Outliers were excluded using the outlier calculator in Microsoft^® Excel. The pairwise sample comparisons for cell proliferation, cell density and the quantification of FN1 and Phalloidin corrective fluorescence expression were performed using unpaired two-tailed Welch's t-test due to unequal variance and sample numbers. The results for GM130 analysis were compared using a Kolmogorov–Smirnov test developed for intracellular organization (Apte and Marshall, 2013). The P-values for statistical tests in all figures are represented as: *P<0.05, **P<0.005, ***P<0.0005 and ****P<0.00005.

We thank all the members of the Atit Lab past and present for helpful comments and suggestions in guiding this project. We thank Nyoka Lovelace, Julie Denker and Mia Carr for support with animal husbandry and genotyping. We thank Alex Flores and Michael Mohn for help on apical extension analysis in Figs 2, 3 and Fig. S3. We thank Yifan Zhai for taking images in Fig. 3 and Fig. S3. We thank Dr Samuel Senyo and Chao Liu for generating the custom MATLAB codes used in Figs 1 and 2. Thank you to Dr Nicole Crown for the generous use of the Leica confocal and to the Case Western Reserve University bio[box] facility for use of microscopy and shared instrumentation facilities in the Biology Department.