Lineage-specific mutation of Lmx1b provides new insights into distinct regulation of suture development in different areas of the calvaria

The calvaria (top part of the skull) is made of pieces of bone as well as multiple soft tissue joints called sutures. The latter is crucial to the growth and morphogenesis of the skull, and thus a loss of calvarial sutures can lead to severe congenital defects in humans. During embryogenesis, the calvaria develops from the cranial mesenchyme covering the brain, which contains cells originating from the neural crest and the mesoderm. While the mechanism that patterns the cranial mesenchyme into bone and sutures is not well understood, function of Lmx1b, a gene encoding a LIM-domain homeodomain transcription factor, plays a key role in this process. In the current study, we investigated a difference in the function of Lmx1b in different parts of the calvaria using neural crest-specific and mesoderm-specific Lmx1b mutants. We found that Lmx1b was obligatory for development of the interfrontal suture and the anterior fontanel along the dorsal midline of the skull, but not for the posterior fontanel over the midbrain. Also, Lmx1b mutation in the neural crest-derived mesenchyme, but not the mesoderm-derived mesenchyme, had a non-cell autonomous effect on coronal suture development. Furthermore, overexpression of Lmx1b in the neural crest lineage had different effects on the position of the coronal suture on the apical part and the basal part. Other unexpected phenotypes of Lmx1b mutants led to an additional finding that the coronal suture and the sagittal suture are of dual embryonic origin. Together, our data reveal a remarkable level of regional specificity in regulation of calvarial development.

The calvaria (top part of the skull) is made of pieces of bone as well as multiple soft tissue joints called sutures. The latter is crucial to the growth and morphogenesis of the skull, and thus a loss of calvarial sutures can lead to severe congenital defects in humans. During embryogenesis, the calvaria develops from the cranial mesenchyme covering the brain, which contains cells originating from the neural crest and the mesoderm. While the mechanism that patterns the cranial mesenchyme into bone and sutures is not well understood, function of Lmx1b, a gene encoding a LIM-domain homeodomain transcription factor, plays a key role in this process. In the current study, we investigated a difference in the function of Lmx1b in different parts of the calvaria using neural crest-specific and mesoderm-specific Lmx1b mutants. We found that Lmx1b was obligatory for development of the interfrontal suture and the anterior fontanel along the dorsal midline of the skull, but not for the posterior fontanel over the midbrain. Also, Lmx1b mutation in the neural crest-derived mesenchyme, but not the mesodermderived mesenchyme, had a non-cell autonomous effect on coronal suture development. Furthermore, overexpression of Lmx1b in the neural crest lineage had different effects on the position of the coronal suture on the apical part and the basal part. Other unexpected phenotypes of Lmx1b mutants led to an additional finding that the coronal suture and the sagittal suture are of dual embryonic origin. Together, our data reveal a remarkable level of regional specificity in regulation of calvarial development.

Introduction
The calvaria (top part of the skull) comprises bone plates and soft tissue joints, called sutures. During fetal and early postnatal stages, large areas of soft tissue appear where multiple sutures intersect, and they are called fontanels. There are five pieces of bone in the mammalian calvaria. In mice, they are a pair of the frontal bone, a pair of the parietal bone, and a single piece of the interparietal bone. The interfrontal suture (also known as the metopic suture) joins the two frontal bones, and the sagittal suture connects the two parietal bones. In addition, the coronal suture lies between the frontal bone and the parietal bone, whereas the lambdoidal suture connects the parietal bone and the interparietal bone. While the bone provides physical protection to the brain, the sutures contain progenitors and stem cells for osteoblasts, which allow continued growth of the skull to accommodate the rapidly expanding brain during childhood (Ishii et al., 2015;Twigg and Wilkie, 2015;White et al., 2021;Stanton et al., 2022). Craniosynostosis (premature fusion of one or more sutures) is a major type of birth defects in humans, found in 1 out of~2300 live births (Boulet et al., 2008;Levi et al., 2012;Stanton et al., 2022). It can profoundly affect overall morphology of the skull, and even impair development of the brain through an increase in intracranial pressure (Levi et al., 2012;Yu et al., 2021;Stanton et al., 2022). Therefore, investigating how the sutures are generated and maintained is highly relevant to human health.
During embryogenesis, the calvaria develops from the head mesenchyme layer covering the brain (cranial mesenchyme), which also gives rise to the meninges and the dermis of the scalp (Ishii et al., 2015;Ferguson and Atit, 2019). This mesenchyme contains cells from the neural crest and the mesoderm. Lineagetracing studies in mice have found that the frontal bone, the interfrontal suture, and the sagittal suture are made of the neural crest-derived cells, whereas the parietal bone and the coronal suture are made of the mesoderm-derived cells (Jiang et al., 2002;Yoshida et al., 2008). The interparietal bone has cells from both origins. A more recent report has shown that the frontal bone receives a minor contribution of mesoderm-derived cells also (Deckelbaum et al., 2012).
The calvarial bone develops through intramembranous ossification, and this process begins around embryonic day (E) 12 in mice, with expression of an osteogenic gene Runx2 in the condensed mesenchyme just above the eye. This region is called the supra-orbital mesenchyme (Deckelbaum et al., 2012;Ferguson and Atit, 2019;Galea et al., 2021), and it contains rudiments of the frontal bone and the parietal bone separated by presumptive coronal suture progenitors. Between E12 and birth [= E19 or postnatal day (P) 0], the frontal bone, the coronal suture, and the parietal bone expand apically side-by-side. The interfrontal suture and the sagittal suture are established when the osteogenic fronts from the left and the right approximate at the dorsal midline late in gestation (Yoshida et al., 2008;Deckelbaum et al., 2012;Ishii et al., 2015).
The calvarial bone and sutures are organized in a stereotypic pattern in each species, but the mechanism that determines the layout remains poorly understood (Teng et al., 2019;White et al., 2021). It is thought to involve both genetic regulations and mechanical forces (Galea et al., 2021). For example, a computational model that combined the mechanical strain from the brain growth with the Turing reaction-diffusion model was able to accurately predict positions of the bone and sutures in the normal calvaria (Lee et al., 2019). Furthermore, we have shown that a large number of genes are differentially expressed along the apical-basal axis of the cranial mesenchyme, and one such gene, Lmx1b (LIM homeobox transcription factor 1 beta), plays a crucial role in early calvarial patterning (Cesario et al., 2018;. LMX1B is a key transcriptional regulator for development of multiple body parts including the limb, the brain, the kidney, and the calvaria (Chen et al., 1998a;Chen et al., 1998b;Morello et al., 2001;Guo et al., 2007). In humans, LMX1B mutation underlies the Nail-Patella syndrome, affecting the nails, kneecaps, and the kidney, and it was also detected in one case of familial craniosynostosis (Chen et al., 1998a;Morello et al., 2001;Wilkie et al., 2010). In the cranial mesenchyme, Lmx1b is specifically expressed in the area apical to the supra-orbital mesenchyme (also known as early-migrating mesenchyme) and inhibits osteogenesis in this region. Deletion of Lmx1b using Prrx1-Cre (Logan et al., 2002), which is broadly active in the cranial mesenchyme, led to heterotopic ossification at the vertex and fusion of multiple sutures (Cesario et al., 2018).
In the current study, we have further dissected Lmx1b function in the developing calvaria through lineage-specific deletion and overexpression. We defined the extent of the requirement of Lmx1b in the calvaria. In addition, there was a difference in Lmx1b function from the neural crest-side and the mesodermside of the coronal suture. Lastly, by analyzing the coronal suture at multiple points along the apical-basal axis, we obtained results suggesting distinct regulation at different positions.
The sample size for each experiment is indicated in the figure legends and the text of the Results section, and the phenotypes were Frontiers in Physiology frontiersin.org 03 consistent in all samples unless indicated otherwise. We did not separate the samples by sex because we found comparable calvarial phenotypes in male and female Lmx1b mutants in a previous study using Prrx1-Cre (Cesario et al., 2018). All the animal work was performed following a protocol approved by Institutional Animal Care and Use Committee of New York University.
For Alizarin Red staining of the skulls, the heads were skinned and incubated in ethanol for 2 days and acetone for 2 days and stained with 0.01% Alizarin Red S in 1% KOH solution for 2 days. The samples were cleared in 1% KOH and stored in glycerol. For alkaline phosphatase staining, the surface ectoderm was removed, and the heads were stained with NBT/BCIP (Sigma B6404).

Preparation of frozen sections, immunofluorescence, and RNA in situ hybridization
Frozen sections were prepared at 12 μm thickness as previously described (Jeong et al., 2012). For transverse sections, a section's position along the apical-basal axis was calculated as (No. of sections between the top of the head and the section of interest) divided by (No. of sections between the top of the head and the top of the eye), rounded to one digit below the decimal point. We tallied all sections including those lost in the process. "Top of the head" in this context was defined as the area of the head that would appear on the first section when we cut through the head embedded upside down. Its precise location varied slightly from one sample to another because different heads were tilted slightly differently during embedding. Generally speaking, the top of the head was at or close to the dorsal midline in terms of the medio-lateral position, and around the posterior end of the cerebral hemispheres in terms of the anteroposterior position, because when this area was touching the bottom of the embedding mold, the inverted head tended to stand still.
RNA in situ hybridization for Lmx1b was performed on frozen sections using RNAscope Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics, 323100) and an RNAscope probe for Lmx1b (Advanced Cell Diagnostics, 412931) according to the manufacturer's protocol. It was followed by immunofluorescence for SP7, for which the secondary antibody was donkey anti-rabbit IgG, horseradish peroxidase-conjugated (Novex, A16023), with Cy3-tyramide (PerkinElmer, NEL74001KT) as a substrate.
Fluorescence pictures showing specific areas of the tissue were captured with Nikon Eclipse E600, except for Lmx1b fluorescence RNA in situ hybridization images, which were acquired with Leica SP8 confocal microscope. Whole-section pictures were captured with Nikon SMZ1500 in tiles and manually stitched with Adobe Photoshop.

Histological staining of sections
P21 heads were fixed in 4% paraformaldehyde solution, dehydrated through an ethanol series, and embedded in methyl methacrylate. The samples were sectioned into 400 µm-thick slices with a precision diamond saw (Isomet 2000, Buehler Ltd). The sections were glued to an acrylic plate with a photolabile acrylatebased adhesive (Technovit 7210 VLC adhesive, Heraeus Kulzer GMBH) before grinding and polishing to a final thickness of 200 µm. The sections were subsequently stained with Stevenel‫׳‬s Blue and Van Gieson‫׳‬s Picro Fuschin, and imaged with a slide scanner (Aperio Technologies).
For histology of P0 heads, frozen sections from Section 2.3 were stained with hematoxylin and eosin, and imaged with a slide scanner (Aperio Technologies).

Quantitative analysis of images and statistical tests
Volume measurements from microCT data were obtained by Amira (Supplementary Figure S2) or Avizo ( Figure 5) (Thermo Fisher Scientific). The length and area measurements were done by FIJI (Schindelin et al., 2012) from images acquired at consistent settings between controls and mutants. Allometry was not considered. A comparison between genotypes was made with samples at the same developmental stage. We did not employ additional standardization across samples. However, except for the bone volume, our data were represented as a ratio of two measurements from the same sample, which reduced the effect of variation in the overall size of the head. All samples were measured without their genotypes indicated. Two-tailed Student's t-test was used for a comparison between control and mutant samples, and p < 0.05 was considered significant. In the charts, values from individual samples are presented with an average (bar) and standard deviation as error bars. For incidental data, Fisher's exact test was used to calculate p values.
fontanel. We had noted that the basal coronal suture did not express Lmx1b, and thus we concluded that this region developed independently of Lmx1b function (Cesario et al., 2018). However, it was unclear why other areas remained unossified in Prrx1-Cre; Lmx1b fl/mutants. One possibility was that these regions were occupied by cells still expressing Lmx1b in the mutants, since Prrx1-Cre was inactive in some parts of the cranial mesenchyme (Logan et al., 2002;Cesario et al., 2018). To answer this question, we used Mesp1-Cre to delete Lmx1b in the mesoderm-derived cells, and Sox10-Cre to delete Lmx1b in the neural crest-derived cells, as the two lineages make up the whole calvaria (Saga et al., 1999;Jiang et al., 2002;Matsuoka et al., 2005;Yoshida et al., 2008). Furthermore, this approach allowed us to address another important question, i.e., whether Lmx1b has any non-cell autonomous function in inhibiting osteogenesis, by examining the phenotype at the lineage boundary. Lastly, although we have shown that LMX1B does not inhibit osteogenesis in the limbs (Cesario et al., 2018), which undergoes endochondral ossification, it was unknown whether LMX1B has an anti-osteogenic function in the facial mesenchyme, which undergoes intramembranous ossification just like the calvaria (Galea et al., 2021). Lmx1b expression was evident in the nose and the mandible from E13.5 (Schweizer et al., 2004) (Supplementary Figure S2). Prrx1-Cre has limited activity in these areas (Logan et al., 2002), but Sox10-Cre can be effective because the nose and the mandible contain mostly neural crest-derived mesenchyme.
We did not use Wnt1-Cre for neural crest-specific mutation because it is active in a significant part of the brain in addition to the neural crest (Danielian et al., 1998;Keuls and Parchem, 2021). The Wnt1-Cre domain in the brain extensively overlaps with Lmx1b expression along the dorsal midline and in the midbrain (Guo et al., 2007;Mishima et al., 2009). Because Lmx1b plays an important role in brain development (Guo et al., 2007), Wnt1-Cre-mediated mutation of Lmx1b can lead to calvarial defects that are secondary to the brain defects.
3.1 Mesoderm-specific deletion of Lmx1b results in partial fusion of the coronal suture and the lambdoidal suture but no ossification in the posterior fontanel We used microCT and Alizarin Red staining to examine the morphology of the calvaria at E18.5 and P0 (Figure 1). In E18.5 Mesp1-Cre;Lmx1b fl/fl mutants (n = 5), the apical end, but not the basal end, of the coronal suture was displaced anteriorly within the frontal bone-coronal suture-parietal bone unit (Figures 1A-D, AC; see Supplementary Figure S1 for definitions of different axes). The parietal bone was enlarged while the frontal bone was reduced ( Supplementary Figures S2A, B, G). The morphology of the coronal suture exhibited a complex phenotype. There was partial fusion of the coronal suture affecting all mutants examined at P0 (n = 6, p = 0.00058), either unilaterally (2 out of 6 mutants) or bilaterally (4 out of 6 mutants). The fusion was always only in the middle part along the apical-basal axis ( Figures 1E-H). Furthermore, the apical part and the basal part were completely disjointed, with the apical part located significantly anterior to the expected position based on the trajectory from the basal part ( Figure 1H).
The lambdoidal suture was also affected in all mutants examined (n = 6, p = 0.00058), with partial fusion either unilaterally (2 out of 6 mutants) or bilaterally (4 out 6 mutants), and it had irregular morphology even where unfused ( Figures 1I, J). However, the posterior fontanel remained unossified and became a large soft spot in the skull with the bulging brain, similar to what was described in Prrx1-Cre;Lmx1b fl/mutants ( Figure 1F) (Cesario et al., 2018). We could not examine Mesp1-Cre;Lmx1b fl/fl mutants at postnatal stages because the newborns were visibly in distress with brain hemorrhage and thus had to be euthanized.
3.2 Neural crest-specific deletion of Lmx1b results in ossification of the entire interfrontal suture and abnormal morphology of the coronal suture, but the sagittal suture remains patent In E18.5 and P0 Sox10-Cre;Lmx1b fl/fl mutants (n = 12), the entire interfrontal suture and the anterior fontanel were occupied by heterotopic bone (arrowheads in Figures 1N, P, R, T), while the sagittal suture appeared widened ( Figure 1O-P). Also, there was a suture-like gap off the midline between the heterotopic bone and the endogenous frontal bone (arrows in Figure 1N, P, T). The apical end, but not the basal end, of the coronal suture was displaced posteriorly within the frontal bone-coronal suture-parietal bone unit ( Figures  1K-N, AC). These changes were concomitant with an increase in frontal bone volume (Supplementary Figures S2C, D, G). Although the coronal suture was patent in Sox10-Cre;Lmx1b fl/fl mutants, the morphology was affected in all samples examined at P0 (n = 7, p = 0.00002). The characteristic overlap between the parietal bone and the frontal bone was lost, unilaterally (2 out of 7 mutants) or bilaterally (5 out of 7 mutants) ( Figures 1U, V). This phenotype was confirmed on sections ( Figures 1W, X). At P21 (n = 3), the coronal suture showed the same phenotype as P0, i.e., posterior displacement at the apical end and a loss of bone overlap (Figures 1Y, Z, AA, AB, AD), but the frontal bone had become a continuous piece without any gap ( Figure 1Y, Z). We also measured the volumes of the nasal bone and the dentary bone in Sox10-Cre;Lmx1b fl/fl mutants at P0, but did not find a significant difference from controls ( Supplementary Figures S2C-F 3.3 The sagittal suture and the coronal suture are of dual embryonic origin, whereas the lambdoidal suture develops from the mesoderm Because the sagittal suture had been known to be neural crestderived (Jiang et al., 2002;Yoshida et al., 2008), it was unexpected that the sagittal suture was patent in Sox10-Cre;Lmx1b fl/fl mutants ( Figure 1). Therefore, we scrutinized the embryonic origin in detail using a yellow fluorescent protein (YFP) Cre reporter (Srinivas et al., 2001) and a preosteoblast marker SP7 (Supplementary Figure S3). We discovered that, in normal embryos, only the anterior half of the sagittal suture mesenchyme was neural crest-derived whereas the posterior half was mesoderm-derived (Supplementary Figures S3H,I,N,O). We also examined the embryonic origin of the sagittal Frontiers in Physiology frontiersin.org 05 suture in Lmx1b mutants. In Mesp1-Cre;Lmx1b fl/fl ;R26 R-YFP/+ mutants (n = 3), the composition of the sagittal suture mesenchyme resembled that of the controls (Supplementary Figures S3K, L). In other words, the mesodermal cells were able to contribute to the posterior sagittal suture mesenchyme even in the absence of Lmx1b. In contrast, in Sox10-Cre;Lmx1b fl/fl mutants (n = 3), the whole sagittal suture was made of cells from the mesoderm, in which Lmx1b was not deleted (Supplementary Figure S3R). A section through a position that would normally be the anterior sagittal suture showed that the neural crest-derived cells became bone (arrowhead in Supplementary Figure S3Q), while the mesoderm-derived cells formed the suture mesenchyme next to the parietal bone (open arrowheads in Supplementary Figure S3Q).
As to the coronal suture, it was fused along the entire length in Prrx1-Cre;Lmx1b fl/mutants except for the basal part, which indicated that Lmx1b was required in the apical and the middle parts (Cesario et al., 2018). Based on the notion that the coronal suture is mesoderm-derived, we were unable to explain why the apical part was not fused in Mesp1-Cre; Lmx1b fl/fl mutants (Figure 1). Similarly, it was unclear why the coronal suture morphology was affected in Sox10-Cre;Lmx1b fl/fl mutants. Therefore, we re-examined the embryonic origin of the coronal suture at multiple positions along the apical-basal axis (Figure 2). We found that the apical part of the coronal suture mesenchyme was normally neural crest-derived (Figures 2D, G, M) even though most of the coronal suture was mesodermderived as previously reported (Jiang et al., 2002;Yoshida et al., 2008). This explained the lack of fusion in the apical coronal suture in Mesp1-Cre;Lmx1b fl/fl mutants, as the suture mesenchyme here was made of cells still expressing Lmx1b ( Figure 2J; n = 3). In contrast, the basal part of the coronal suture was patent in Mesp1-Cre;Lmx1b fl/fl mutants despite being mesoderm-derived ( Figure 2L; n = 3). In Sox10-Cre;Lmx1b fl/fl mutants, the entire coronal suture, including the apical part, was made of mesoderm-derived cells, which also formed the osteogenic front of the frontal bone ( Figure 2O-P; n = 3). However, the contribution of the mesodermal cells was only at the posterior tip of the frontal bone (arrow in Supplementary  Figures S4B, D, F), and thus the overall composition of the frontal bone did not change significantly in Sox10-Cre;Lmx1b fl/fl mutants (Supplementary Figure S4G).
We examined the origin of the lambdoidal suture because it was not clearly determined in earlier studies. At P0, the lambdoidal suture mesenchyme was entirely from the mesoderm lineage, whether it abutted the medial part of the interparietal bone, which is neural crest-derived (Supplementary Figures S5D, G), or the lateral part of the interparietal bone, which is mesodermderived (Supplementary Figures S5E, F, H, I). A previous study reported presence of neural crest-derived cells in the lambdoidal suture at P7 (Behr et al., 2011), and thus the composition may change over time.
3.4 Inactivation of Lmx1b leads to widespread osteogenesis in the cranial mesenchyme from early stages of calvarial development, but not over the midbrain  Frontiers in Physiology frontiersin.org 06 arrowheads in Figures 3B, D, I). However, the aberrant alkaline phosphatase expression was weak at the dorsal midline in Mesp1-Cre;Lmx1b fl/fl mutants (arrowhead in Figure 3F). More importantly, it was absent from the mesenchyme over most of the midbrain, the position of the future posterior fontanel ( Figures 3A-D). This spatial restriction in osteogenesis was clear at E12.5, when the bulging of the midbrain was not yet evident in the mutants (Figures 3A, B). Besides, the brain is not where primary defects are expected from Mesp1-Cre-mediated mutation. Thus, the bulging of the brain at the posterior fontanel is likely to be a consequence, rather than the cause, of the mesenchyme over the midbrain being an 'island' of soft spot encircled by bone. We confirmed that Mesp1-Cre had been active in this region of the mesenchyme in 3.5 Lmx1b in the neural crest-derived mesenchyme, but not the mesodermderived mesenchyme, has a non-cell autonomous effect on coronal suture development Although we found neural crest-derived cells in the apical part of the coronal suture at P0 (Figure 2), this could have been due to the coronal suture merging with the neural crest-derived sutures at the dorsal midline late in gestation, rather than due to coronal suture progenitors arising from the neural crest-derived mesenchyme during early development. Therefore, we examined the coronal suture at E14.5 to clarify the contribution of the neural crest lineage (Figure 4). To identify matching sections across different samples consistently, we labeled apical-basal positions using a numerical system in which the top of the head was 0 and the top of the eye was 1 ( Figure 4A; see Materials and Methods Section 2.3 for details). We were able to discern the prospective coronal suture flanked by the frontal bone and the parietal bone rudiments up to 0.4 position ( Figures 4C, I). Here, we found that the presumptive Frontiers in Physiology frontiersin.org coronal suture progenitors were mostly neural crest-derived (YFP − in the box in Figure 4C and YFP + in the box in Figure 4I), whereas they were mesoderm-derived at more basal positions (YFP + in the box in Figure 4D and YFP − in the box in Figure 4J). In E14.5 Mesp1-Cre;Lmx1b fl/fl mutants (n = 3), SP7 was detected at an apical position (0.2) where it is not detected normally (arrow in Figure 4E), and also in what should be mesoderm-derived coronal suture progenitors at 0.6 (arrow in Figure 4G). However, there was no evidence of aberrant osteogenesis in the neural crest-derived cells in these mutants ( Figures 4B-G). In contrast, Sox10-Cre;Lmx1b fl/fl mutants showed ectopic or enhanced expression of SP7 not only in the neural crest-derived cells (arrows in Figures 4M-O, Q) but also  Figures 4O, Q). This is a non-cell autonomous phenotype because the Cre was not active the mesoderm. Still, the coronal suture was able to form in these mutants, just a few cells removed from the neural crestmesoderm boundary (Figures 4N, O). Consistent with the observation at P0 (Figure 2), the coronal suture progenitors at 0.4 were from the mesoderm, instead of the neural crest, in Sox10-Cre;Lmx1b fl/fl mutans ( Figure 4N), indicating that cells lacking Lmx1b expression were unable to contribute to the coronal suture.
Lmx1b expression in the coronal suture had not been described before, and thus we examined it at various positions along the apical-basal axis during normal development (Figures 4R-U). As previously reported on coronal sections (Cesario et al., 2018), Lmx1b was expressed predominantly around the top of the head (0.2) while absent from the supra-orbital mesenchyme (0.8) (Figures 4R, U). In between, low levels of Lmx1b mRNA were detected in the coronal suture progenitors at 0.4 and 0.6 ( Figures 4S, T).

Neural crest-specific overexpression of
Lmx1b uncouples the coronal suture from the lineage boundary in the basal part, but not in the apical part To confirm the non-cell autonomous function of LMX1B, we overexpressed it in the neural crest-derived cells using Sox10-Cre and a ROSA26 knockin allele with a floxed stop cassette followed by Lmx1b coding sequence (R26 Lmx1b/+ ) (Li et al., 2010).
At E18.5, the frontal bone volume was reduced by more than 50% in Sox10-Cre;R26 Lmx1b/+ mutants compared with controls ( Figures 5A-E; n = 8). The parietal bone was also reduced bỹ 25% on average, as expected from a non-cell autonomous effect of Lmx1b overexpression. We also examined the relative position of the coronal suture within the frontal bone-coronal suture-parietal bone unit in Sox10-Cre;R26 Lmx1b/+ mutants ( Figures 5A-D, F). At the apical end, the mutant coronal suture shifted in the anterior direction compared with controls, consistent with the frontal bone being more severely reduced than the parietal bone ( Figures  5C, D, F). Surprisingly, the basal end of the coronal suture shifted posteriorly in these mutants ( Figures 5A, B, F). Examination of the frontal bone and the parietal bone separately showed that they were not uniformly affected along the apical-basal axis in the mutants. The frontal bone was preferentially reduced in the apical part, whereas the parietal bone was preferentially reduced in the basal part ( Figure 5G).
To better understand the coronal suture phenotype of Sox10-Cre; R26 Lmx1b/+ mutants, we examined the transverse sections of the heads in which YFP labeled the neural crest-derived cells (Figure 6). The measurements from perinatal samples confirmed that the apical part of the coronal suture, but not the basal part, shifted to the anterior direction in the mutants (Figures 6A-D, I; n = 3). The mutant frontal bone contained a larger fraction of mesoderm-derived cells (YFP − ) than controls (Supplementary Figure S6; Figures 6A-H, J). In the apical part (0.4), these cells were broadly scattered, and the coronal suture more or less coincided with the neural crest-mesoderm boundary (Figures 6E, F). In contrast, in the basal part (0.8), the posterior half of the mutant frontal bone was almost entirely made of the mesoderm cells, and the coronal suture was located inside the mesoderm domain ( Figures 6G, H, J).
We examined how the initial osteogenesis pattern was affected in Sox10-Cre;R26 Lmx1b/+ mutants at E13.5. Consistent with the phenotype at E18.5, both the frontal bone and the parietal bone rudiments were reduced in the mutants (n = 3). However, throughout the apical-basal axis, the frontal bone rudiment was Frontiers in Physiology frontiersin.org more severely affected than the parietal bone rudiment, and thus the prospective coronal suture showed an anterior shift (Supplementary Figure S7; Figures 6K-M). Therefore, the early phenotype was similar to that of the apical part at E18.5, which suggested that the basal part had undergone a patterning change in Sox10-Cre; R26 Lmx1b/+ mutants later in development. We also noted that Lmx1b overexpression affected the parietal bone formation preferentially in the mesenchyme overlying the forebrain meninges, such that this region (# in Figures 6K, L) contained most of the parietal bone rudiment in control embryos, but less than half in the mutant embryos ( Figure 6N).

Discussion
In this study, we have clarified a spatial specificity of Lmx1b function by analyzing lineage-specific deletion and overexpression mutants. In the calvaria, Lmx1b was essential to prevent osteogenesis in all sutures except for the basal part of the coronal suture, the posterior part of the sagittal suture, and the posterior fontanel ( Figure 7). Furthermore, Lmx1b in the neural crest-lineage, but not the mesoderm lineage, had a non-cell autonomous effect at the coronal suture. In addition, although Lmx1b was expressed in the facial mesenchyme, it did not play the same crucial role as in the calvaria. Together, our findings on Lmx1b highlight the regional difference in genetic regulation of osteogenesis during development, even between physically and/or ontogenically very close areas of the mesenchyme. More importantly, a couple of unexpected phenotypes of the Lmx1b mutants gave us new insights into calvarial development.

Embryonic origin of the calvarial sutures
From what had been known about the embryonic origin of different components for the calvaria, we expected that the interfrontal suture and the sagittal suture would be lost in Sox10-Cre;Lmx1b fl/fl mutants. However, the sagittal suture was patent in these mutants, which led us to discover that it normally comprised both neural crest-derived and mesoderm-derived cells. Also, the complex coronal suture phenotype in Mesp1-Cre;Lmx1b fl/fl mutants revealed that the coronal suture has three distinct segments along the apical-basal axis with respect to the embryonic origin and regulation by Lmx1b: neural crest-derived and Lmx1b-dependent, mesoderm-derived and Lmx1b-dependent, mesoderm-derived and Lmx1b-independent (Figure 7). A previous study reported neural crest contribution to the coronal suture, but it was only mentioned as a part of the periosteum and the dura mater (Doro et al., 2019), not the suture mesenchyme proper as we found. More importantly, all the earlier reports of lineage analyses showed whole mount samples, where it is difficult to distinguish overlapping layers of tissue, and/or sections from only one position for each suture (Jiang et al., 2002;Yoshida et al., 2008;Behr et al., 2011;Doro et al., 2019). The intra-suture heterogeneity described here indicates that a careful attention to the precise position is necessary when examining suture phenotypes at a high resolution.
According to our results, Lmx1b was essential in all the neural crestderived suture areas, but it was dispensable in some of the mesodermderived suture areas (Figure 7). It is possible that the embryonic origin of the cells is important for the role of Lmx1b in suture development. Alternatively, different positions of the suture cells within the calvaria may determine their dependence on Lmx1b, as discussed below.  Figure S3). Similarly, we believe that the neural crest-derived cells supposed to make the apical coronal suture mesenchyme were incorporated into the frontal bone, contributing to its posterior extension (reflected in the posterior displacement of the apical coronal suture). However, this is difficult to prove because the frontal bone is normally made of the neural crest-derived cells, those migrating from the supra-orbital mesenchyme (Jiang et al., 2002;Yoshida et al., 2008).

Establishing the sutures along the dorsal midline of the calvaria
It is incompletely understood how positions of the bone and the sutures are determined during calvarial development. Regulatory factors for calvarial patterning likely include prepatterning of the cranial mesenchyme by region-specific gene expression, environmental cues via secreted signaling molecules from the neighboring tissue, and localized accumulation of mechanical strain produced by the growing brain. Once osteogenesis is initiated at certain positions, the pattern is enforced by a reaction-diffusion mechanism involving activators and inhibitors from the nascent osteogenic centers (Lee et al., 2019;Galea et al., 2021). There may be a hierarchy among these factors, and potentially a difference in their relative importance depending on the exact location.
A recent study suggested that a difference in the volumetric strain from the brain was the reason why the osteogenic centers appeared in the baso-lateral region of the head but not at the apex during normal development (Lee et al., 2019). However, ossification at the interfrontal suture, the anterior fontanel, and the anterior sagittal suture in Sox10-Cre;Lmx1b fl/fl mutants indicates that a superseding factor at this location is intrinsic patterning of the cranial mesenchyme by way of differential expression of Lmx1b. Furthermore, a suture-like gap appeared between the heterotopic midline bone and the endogenous frontal bone, most likely due to mutual inhibition via the reaction-diffusion mechanism (Galea et al., 2021), but this was also abrogated later in the absence of Lmx1b (Figure 1).
In contrast, there seems to be an Lmx1b-independent mechanism to prevent osteogenesis at the posterior sagittal suture and the posterior fontanel, based on the results from Mesp1-Cre;Lmx1b fl/fl and Prrx1-Cre;Lmx1b fl/mutants (Figure 1; Cesario et al., 2018). Notably, Prrx1-Cre-mediated deletion of Gnas, encoding a component of G protein-coupled receptor signaling, led to heterotopic ossification at the dorsal calvaria preferentially affecting the posterior fontanel (Xu et al., 2018). Gnas is ubiquitously expressed during development (Magdaleno et al., 2006), and thus it is not clear why the anterior and the posterior regions of the calvaria have opposite patterns of dependence on Lmx1b and Gnas to prevent osteogenesis. We speculate that a difference in interactions between the suture and different regions of the brain (forebrain vs. midbrain) might be involved.

Determining the position of the coronal suture
Because the frontal bone, the coronal suture, and the parietal bone arise from one continuous and seemingly homogenous block of mesenchyme, it is particularly intriguing how the suture position is determined within this unit. An interesting phenotype of Mesp1-Cre;Lmx1b fl/fl mutants was that the apical part and the basal part of the coronal suture were completely disjointed. This result suggested that the positions of the two segments could be determined independently of each other, although it does not necessarily mean that this is the case during normal development.
How we analyzed the relative position of the coronal suture in Lmx1b mutants is very similar to what others did for lineage-specific mutants of two craniosynostosis genes, Twist1 and Tcf12 (measuring the ratio between the lengths of the sagittal suture and the interfrontal suture) (Teng et al., 2018;Ting et al., 2022), except that we analyzed the apical end and the basal end separately. A common finding from all three studies is that the apical coronal suture position at perinatal or postnatal stages closely reflects the pattern of osteogenesis at early stages (E13.5 or earlier) determined by genetic regulation intrinsic to the cranial mesenchyme. When a mutation enhanced osteogenesis in the neural crest-derived mesenchyme, the coronal suture was displaced posteriorly. When a mutation enhanced osteogenesis in the mesoderm-derived mesenchyme, the coronal suture shifted anteriorly. Tcf12 and Twist1 were shown to regulate cell proliferation, and the displacement of the coronal suture in these mutants was attributed to differential growth of the frontal bone vs. the parietal bone (Teng et al., 2018). However, our earlier study did not find a significant effect of Lmx1b mutation on cell proliferation (Cesario et al., 2018). Therefore, the phenotype of Lmx1b mutants likely reflects mis-patterning of the cranial mesenchyme at the beginning of the calvarial development.
As to the position of the basal end of the coronal suture, we did not find a significant change in Sox10-Cre;Lmx1b fl/fl or Mesp1-Cre; Lmx1b fl/fl mutants. This was consistent with the fact that Lmx1b is normally not expressed in the supra-orbital mesenchyme, which contains the base of the coronal suture. However, in Sox10-Cre; R26 Lmx1b/+ mutants, Lmx1b was overexpressed even in the supraorbital mesenchyme, and thus we expected an anterior shift of the coronal suture associated with the frontal bone deficiency at the basal end as well as at the apical end. Instead, we found that the mesoderm-derived cells partially compensated for the frontal bone deficiency between E13.5 and E18.5, apparently at the expense of the parietal bone growth in the basal part (Figures 5, 6). As a result, the anterior shift of the coronal suture at E13.5 was "rectified" by E18.5 (in fact, over-corrected to the posterior shift), and in the process, the coronal suture became separated from the neural crest-mesoderm boundary. Although we do not know the molecular and cellular changes underlying this phenotype, it suggests that an unknown factor overrode the early pattern of osteogenesis in the cranial mesenchyme during the course of development. In addition, the phenotype revealed that being at the lineage boundary was not essential for the basal coronal suture to be maintained, although it might be important for initial specification.
It is unclear why the positions of the apical end and the basal end of the coronal suture might be regulated differently. One potential Frontiers in Physiology frontiersin.org factor is that the basal part of the calvaria, but not the apical part, develops in close association with the chondrocranium, which may influence development of the calvarial bone and sutures (Motch Perrine et al., 2022). While we found aberrant contribution of the mesoderm-derived cells to the frontal bone in both Sox10-Cre;Lmx1b fl/fl and Sox10-Cre; R26 Lmx1b/+ mutants, the mechanisms involved are most likely different between the two. In Sox10-Cre;Lmx1b fl/fl mutants, SP7 expression was activated in the mesodermal cells next to the neural crest-derived domain from an early stage (E13.5) (Figure 4), and this phenotype remained restricted to the lineage boundary at P0 (Supplementary Figure S4). Therefore, we believe that there was an error in mesodermal cell fate specification regulated by Lmx1b via short-range secreted signaling molecules (see Section 4.4 for details). In contrast, in Sox10-Cre;R26 Lmx1b/+ mutants, there was a large scale change in the frontal bone composition accompanied by a significant modification in the calvarial bone pattern between E13.5 and E18.5 ( Figure 6). Thus, we speculate that the incorporation of the mesodermal cells to the frontal bone is due to an adjustment secondary to the early defect, as discussed above, rather than due to regulation of mesoderm osteogenesis by Lmx1b.
We did not pursue Mesp1-Cre-mediated overexpression of Lmx1b because Mesp1-Cre is active in the cardiovascular system including the capillary endothelial cells throughout the body (Saga et al., 1999). Consequently, overexpressing Lmx1b with Mesp1-Cre is likely to cause a global disruption in the embryo, which in turn would complicate analysis of the phenotype. Hypothetically, overexpression of Lmx1b in the mesoderm-derived calvarial osteoprogenitors is expected to impair parietal bone formation, possibly with partial compensation by neural crest-derived cells. However, we are cautious with making predictions considering that we have encountered many surprising phenotypes in the current study.

Non-cell autonomous function of Lmx1b
We found that Lmx1b deletion in the neural crest-derived cells, but not in the mesoderm-derived cells, led to non-cell autonomous induction of osteogenesis at the coronal suture ( Figure 4). In Sox10-Cre;Lmx1b fl/fl mutants, osteogenesis was enhanced in the neural crest-derived cells, judging from SP7 expression. More importantly, SP7 was also induced in the adjacent mesoderm-derived cells even though Lmx1b was not deleted in them. In the coronal suture region, the main differences between the contributions from the neural crest and the mesoderm are that only the neural crest-derived cells make the fibroblasts of the forebrain meninges underneath the suture, while only the mesoderm-derived cells make the capillary endothelial cells (Le Douarin and Kalcheim, 1999). Therefore, we hypothesize that the non-cell autonomous effect of Lmx1b involves its expression in the meninges . This is consistent with the phenotype in Sox10-Cre;R26 Lmx1b/+ mutants, in which the parietal bone development was preferentially inhibited in the mesenchyme over the forebrain meninges.
It has been known for decades that the meninges regulate the sutures, although most studies focused on suture maintenance at late fetal or postnatal stages (Grova et al., 2012;. The meninges express diverse secreted signaling molecules including members of transforming growth factor β (TGFβ), fibroblast growth factor (FGF), bone morphogenetic protein (BMP), Chemokine (C-X-C motif) ligand, and retinoic acid , and one or more of them may be regulated by Lmx1b.
However, the non-cell autonomous mode is clearly not the major mechanism for the function of Lmx1b. The coronal suture was able to form in Sox10-Cre;Lmx1b fl/fl mutants, and the parietal bone eventually grew over the forebrain meninges in Sox10-Cre;R26 Lmx1b/+ mutants. Therefore, we conclude that Lmx1b regulates the fate of cranial mesenchyme cells mainly via a cell-autonomous mechanism, which is common for a transcription factor. Understanding the exact mechanism of LMX1B function awaits identification of its transcriptional targets. Nevertheless, our finding suggests a potential novel role for Lmx1b in regulating development of the meninges.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Materials, further inquiries can be directed to the corresponding author.

Ethics statement
The animal study was reviewed and approved by Institutional animal care and use committee of New York University.

Author contributions
JJ, YC, LW, and KD contributed to the conception and the design of the study. AC, KD, T-VH, MP-V, JK, NK, YL, JM, JY, JJ, and LW contributed to the acquisition, analysis, and interpretation of data. JJ prepared the first draft of the manuscript. All authors contributed to the article and approved the submitted version.

Funding
This work was supported by a grant from National Institute of Health (R01 DE026798). MicroCT core at New York University College of Dentistry was supported by a Shared Instrument Grant from National Institute of Health (NIH) (S10OD010751).