Cellular transitions during cranial suture establishment in zebrafish

Abstract

Cranial sutures separate neighboring skull bones and are sites of bone growth. A key question is how osteogenic activity is controlled to promote bone growth while preventing aberrant bone fusions during skull expansion. Using single-cell transcriptomics, lineage tracing, and mutant analysis in zebrafish, we uncover key developmental transitions regulating bone formation at sutures during skull expansion. In particular, we identify a subpopulation of mesenchyme cells in the mid-suture region that upregulate a suite of genes including BMP antagonists (e.g. grem1a) and pro-angiogenic factors. Lineage tracing with grem1a:nlsEOS reveals that this mid-suture subpopulation is largely non-osteogenic. Moreover, combinatorial mutation of BMP antagonists enriched in this mid-suture subpopulation results in increased BMP signaling in the suture, misregulated bone formation, and abnormal suture morphology. These data reveal establishment of a non-osteogenic mesenchyme population in the mid-suture region that restricts bone formation through local BMP antagonism, thus ensuring proper suture morphology.

Fig. 1. Single-cell RNA-sequencing analysis of juvenile and adult calvaria.

A Schematic of dissected calvaria from juvenile and adult zebrafish. B Uniform Manifold Approximation and Projection (UMAP) plot of integrated juvenile (16,245 cells) and adult (5066 cells) datasets. Connective and skeletogenic clusters outlined by dashed lines. C UMAP analysis of reclustered connective and skeletogenic subset, including juvenile (4324 cells) and adult (4595 cells) datasets. Cell identities grouped by circles. D UMAP plot of osteoblast and prrx1a+ mesenchyme subset (517 juvenile cells, 656 adult cells). E Dot plot showing markers enriched for each cluster within the osteogenic/mesenchymal subset. F RNA velocity (arrows) showing differentiation flow for cells within the osteogenic/mesenchymal subset. G, H Dot plots showing zebrafish paralogs for markers of osteogenic and meningeal clusters in the mouse coronal or frontal suture datasets mapped onto the zebrafish osteoblast/mesenchyme subset.

Fig. 2. Conserved features of juvenile bone front and adult suture mesenchyme.

A Confocal image of juvenile frontal bones (Alizarin Red+) from animals receiving GFP+ neural crest precursor transplants. Dotted line indicates region for adjacent orthogonal projection. Brackets denote mesenchyme ahead of the growing frontal (F) bone. N = 3. B Confocal image of juvenile frontal bones. Mosaic recombination within Sox10:Cre; actab2:loxP-BFP-STOP-loxP-dsRed fish labels a neural crest clone in magenta, and RUNX2:GFP labels nascent osteoblasts. Dotted line indicates region for adjacent orthogonal projection. Brackets denote mesenchyme ahead of the bone front. N = 5. C Dot plot of mesenchyme-enriched genes. D, E In situ hybridization for six2a in the juvenile frontal bone and adult coronal suture. Dotted lines outline the frontal (F) and parietal (P) bones. Arrow marks the tip of the growing frontal bone. Orange boxed inset shows expression of six2a at suture edges and red boxed inset shows six2a expression in mid-suture region. In situs were performed in biological triplicate. F, G Confocal images show fli1a:GFP+ mesenchyme relative to Calcein Blue+ calvarial bones at juvenile and adult stages. Arrow denotes fli1a:GFP+ cells ahead of the growing frontal bone, and arrowheads suture mesenchyme expression. Note extensive labeling of vasculature by fli1a:GFP. N = 5 each stage. H, I Orthogonal sections of confocal images show fli1a:GFP+ cells at the juvenile frontal bone tip (H, arrow) and adult suture relative to sp7:mCherry+ osteoblasts. Asterisks mark fli1a:GFP+ endothelial cells. N = 3 each stage. Scalebars: 50 μM (A, B), 100 μM (D–I).

Fig. 3. scRNA-seq resolves osteoblast trajectories in the skull.

A Lineage analysis and (B) pseudotime analysis of reclustered osteogenic subset using Monocle 3. The black lines in (A) represent the lineage relationship between cells along the trajectory. In (B), cells at the beginning of the trajectory are darkest (purple) and more mature cells are progressively lighter (orange and yellow). C Expression plots of select genes across pseudotime. D Violin plot split by age for osteogenic markers from Monocle analysis. E, F In situ analysis of juvenile frontal (F) bone and adult coronal sutures for mmp9, spp1, and ifitm5. Dotted lines outline the frontal and parietal (P) bones. Arrows mark enriched signal at bone tip and suture edges (mmp9) or along bone surfaces (spp1, ifitm5). In situs were performed in biological triplicate. G Orthogonal section of adult coronal suture labeled by mmp9:GFP and sp7:mCherry. Arrows mark edges of the coronal suture. N = 2. Model of osteogenic cell types at bone fronts and cranial sutures. Scalebars: 100 μM.

Fig. 4. Spatially restricted osteogenesis at bone fronts and suture edges.

A Confocal image of juvenile skull (dorsal view) showing osteoblast:nlsEOS relative to the osteoblast marker sp7:mCherry in the frontal (F) and parietal (P) bones. N = 6. B Schematic of treatment scheme for juvenile calvaria. Red box indicates area imaged in the next panels. C, D Confocal images of osteoblast:nlsEOS+ frontal bone 24 h post-conversion (hpc) showing merged and individual channels, and extracted de novo osteoblasts (green only). N = 3. E Schematic of treatment scheme for adult calvaria. Red box indicates area imaged in the next panels. F, G Confocal images of osteoblast:nlsEOS+ sagittal suture 7 days post-conversion (dpc) showing merged and individual channels, and extracted de novo osteoblasts (green only). Scalebars: 100 μM (A), 50 μM (C, F).

Fig. 5. Establishment of distinct molecular signature for mid-suture mesenchyme.

A Dot plot shows shared and cell type-specific markers for subsets of mesenchyme. B Violin plot split by age for genes expressed in regulatory mesenchyme. C In situ analysis of juvenile frontal bone and adult coronal suture. Dotted lines outline the frontal (F) and parietal (P) bones. Arrows mark bone front and suture mesenchyme, arrowheads the periosteum. D, E Double in situ analysis of adult coronal suture. Orange boxed insets show expression of fli1a and hyal4, but not grem1a, at suture edges, and red boxed insets show grem1a but not fli1a expression in mid-suture region. In situs were performed in biological triplicate. Scalebars: 100 μM.

Fig. 6. Mid-suture mesenchyme upregulates grem1a and does not contribute to bone.

A Confocal images show Alizarin Red staining of frontal (F) and parietal (P) bones relative to induction of grem1a:nlsEOS in forming sutures (arrowheads). Asterisks denote brain expression. N = 5 each stage. B Schematic of conversion scheme for calvaria. grem1a+ cells were globally converted, imaged, and re-imaged at 28 days post conversion (dpc). C Confocal images of grem1a:nlsEOS+ calvaria at 0 and 28 dpc. N = 6. D Orthogonal section of adult coronal suture shows grem1a:nlsEOS in mid-suture cells non-overlapping with sp7:mCherry+ osteoblasts. N = 8. E Quantification of grem1a:nlsEOS single-positive and grem1a:nlsEOS; sp7:mCherry double-positive cells at cranial sutures. Scalebars: 200 μM (A, B), 50 μM (D). P-values were calculated using a two-tailed non-parametric Student’s t tests. Error bars represent S.E.M. Source data are provided as a Source Data file.

Fig. 7. Loss of mid-suture mesenchyme and misregulated osteogenesis in twist1b; tcf12 mutants.

A Dot plot shows enrichment of twist1b and tcf12 expression in the regulatory mesenchyme cluster. B Double in situ analysis of adult coronal suture for twist1b and angptl1b. Dotted lines outline the frontal (F) and parietal (P) bones. In situs were performed in biological triplicate. Arrowheads mark double positive cells within the suture mesenchyme. Asterisk indicates autofluorescence from blood vessel. C Confocal images show grem1a:nlsEOS suture mesenchyme expression relative to the Alizarin Red+ frontal (F) and parietal (P) bones of control (twist1b^+/−;tcf12^+/−) and twist1b^−/−;tcf12^−/− mutant fish. Asterisk marks fused region of coronal suture, and boxes denote magnified regions. D Quantification of width of grem1a:nlsEOS domain in controls and in mutants sides with patent or partially fused coronal sutures. N = 8 control coronal sutures, N = 3 mutant patent coronal sutures, N = 5 partially fused coronal sutures, N = 4 control sagittal sutures, N = 5 mutant sagittal sutures. E Serial confocal imaging of individual control (twist1b^+/−; tcf12^+/−) and twist1b^−/−; tcf12^−/− mutant fish. The same fish were imaged before conversion of osteoblast:nlsEOS at approximately 8 mm, converted at 9 mm and imaged at 11 mm, and then converted again at 13 mm and imaged at 15 mm. Arrowhead marks the aberrant bone front, and arrow marks the fused region of the coronal suture. F Quantification of osteoblast addition at bone fronts and coronal sutures. N = 12 calvarial bones per genotype imaged from 9–11 mm, N = 6 patent coronal sutures per genotype imaged from 13–15 mm. Error bars report standard error of the mean, and statistics were performed using unpaired t tests. G Confocal images show fli1a:GFP vessels relative to the Alizarin Red+ frontal (F) and parietal (P) bones of control (twist1b^+/−;tcf12^+/−) and twist1b^−/−;tcf12^−/− mutant fish. Asterisk marks fused region of coronal suture, and boxes denote magnified regions. N = 5 controls and N = 8 mutants. H Quantification of density of fli1a:GFP vasculature in controls and mutants at patent or fused coronal sutures. Scalebars: 100 μM. P-values were calculated using a two-tailed non-parametric Student’s t tests. Error bars represent S.E.M. Source data are provided as a Source Data file.

Fig. 8. Deletion of BMP antagonists alters bone growth and promotes ectopic bone formation.

A Schematic of mutant alleles generated using CRISPR/Cas9. Red arrow indicates site of mutation relative to protein-coding regions (green boxes). B Confocal imaging of Calcein-stained calvaria show impaired suture formation in grem1a^−/−; nog2^−/−; nog3^+/− zebrafish compared to controls. Dotted lines represent regions of the orthogonal sections at the coronal suture shown below. Frontal (F) and parietal (P) bone. C Quantification of non-overlapping regions in the calvaria. N = 3 for controls, N = 5 for mutants. D Representative imaging of individual control and mutant zebrafish after repeated staining and imaging using Calcein green and Alizarin red stains. Arrowheads indicate non-overlapping bones at the coronal suture. E, F Quantification of area of bone growth between 11–13 mm and 13–15 mm at frontal and parietal bones. N = 7 control frontal bones, N = 8 mutant frontal bones at 11–13 mm. N = 6 control frontal bones, N = 10 mutant frontal bones at 13–15 mm. N = 7 control parietal bones, N = 10 mutant parietal bones at 11–13 mm. N = 6 control parietal bones, N = 10 mutant parietal bones at 13–15 mm. G Confocal imaging of supernumerary bones (numbered) at the sagittal suture (ss) between parietal bones in control and mutant zebrafish. H Quantification of supernumerary bones at the sagittal suture. N = 8 for control and mutant fish. I Confocal imaging of control and mutant coronal sutures reveal ectopic bone islands at the mutant coronal suture. Arrowheads mark ectopic bone islands between frontal and parietal bones. J Quantification of ectopic bone islands at the coronal suture. N = 8 for control and mutant fish. K Immunofluorescence for pSmad1/5 at the coronal suture of control and mutant zebrafish. Dotted line outlines bones. White box indicates quantified area for the coronal suture and red box indicates quantified area for parietal bone. L Quantification of fraction of pSmad1/5+ nuclei at coronal sutures or along the parietal bone. N = 3 for control and mutant fish. Scalebars: 100 μM. P-values were calculated using a two-tailed non-parametric Student’s t tests. Error bars represent S.E.M. Source data are provided as a Source Data file.

Fig. 9. Model for cranial suture formation.

Model for cranial suture formation shows approximating bone fronts with dedicated mesenchyme subsets preceding bone edges, followed by the establishment of overlapping bones and the emergence of the mid-suture mesenchyme program. In a zebrafish model for Saethre-Chotzen syndrome, loss of twist1b and tcf12 fusion leads to a failure to establish the mid-suture mesenchyme program that provides important regulatory factors to restrict osteogenesis and promote angiogenesis at cranial sutures.