Self-propagating wave drives morphogenesis of skull bones in vivo

Abstract

Cellular motion is a key feature of tissue morphogenesis and is often driven by migration. However, migration need not explain cell motion in contexts where there is little free space or no obvious substrate, such as those found during organogenesis of mesenchymal organs including the embryonic skull. Through ex vivo imaging, biophysical modeling, and perturbation experiments, we find that mechanical feedback between cell fate and stiffness drives bone expansion and controls bone size in vivo. This mechanical feedback system is sufficient to propagate a wave of differentiation that establishes a collagen gradient which we find sufficient to describe patterns of osteoblast motion. Our work provides a mechanism for coordinated motion that may not rely upon cell migration but on emergent properties of the mesenchymal collective. Identification of such alternative mechanisms of mechanochemical coupling between differentiation and morphogenesis will help in understanding how directed cellular motility arises in complex environments with inhomogeneous material properties.

Fig. 1. Inhomogeneous material properties and cell behaviors during skull morphogenesis.

A Illustration depicting anisotropic expansion of frontal and parietal bones (gray) toward the top of the head between E13.5, 15.5, and 17.5. B Diagram depicting a coronal section of the developing skull with the frontal bones colored in grey and labeled in cyan. Dotted box indicates an inset showing the locations of the bone center (BC), bone front (BF), and undifferentiated mesenchyme (UM). C Atomic force microscopy (AFM) measurements of the stiffness (apparent Young’s modulus) at E15.5 at the three locations indicated in (B). Horizontal bars indicate mean (N = 8, bone center n = 105, bone front n = 109, undifferentiated mesenchyme n = 25, Welch’s two-tailed t-tests, **p = 0.0017 for BC vs BF, ***p < 0.00001 for BF vs UM). D Representative image of nuclear 540 nm autofluorescence generated by 940 nm excitation of the osteogenic front of E14.5 flat-mounted Osx1:GFP-Cre skull cap. D’ Osx1::GFP-Cre fluorescence. D” Second Harmonic Generation signal with 940 nm excitation. Scale bar = 50 µm, (N = 3) E Schematic showing excision of Osx1-GFP::Cre labeled skull caps. Scale bar = 100 µm. F Diagram depicting ex vivo imaging setup. G Max projection of Osx1-GFP::Cre labeled frontal bone at 0 h at E13.75. Dotted box indicates insets at 0 h and 14 h shown below. Scale bar = 100 µm. H Graph showing example lateral (n = 139), intermediate (n = 117), and front (n = 199) cell tracks from live imaging experiments combined (N = 4), color-coded by time. I Graph showing mean squared displacement for osteoblasts in different mediolateral positions, averaged over N = 4 images. Shading indicates SEM. J Max projection of Osx1-GFP::Cre labeled E13.75 live explant showing two late anaphase nuclei with pink lines indicating angle of divisions, which are quantified in (K, L). Scale bar = 10 µm. K Angle of osteoblast division in E13.75 Osx1-GFP::Cre labeled live explants, (Mean angle, 88°, two-tailed Rayleigh z-test = 497.4, p < 0.00001, n = 862, N = 8). L Angle of division orientation in fixed Osx1::GFP-Crelabeled E13.5 flat-mounted skull caps (Mean angle, 84°, two-tailed Rayleigh z-test = 90, 73697178, p < 0.00001, n = 149, N = 6). M Graph showing the percentage of osteoblasts that maintain (Maint., gray) or do not maintain (Not maint., black) their mediolateral neighbor relationship (n = 60, N = 4). N Representative stills from Osx1-GFP::Cre labeled cultured explants. Osteoblast nuclei have been false colored to demonstrate neighbor relationships reflecting data in (M), sister nuclei are overlaid with pink (N = 4). Scale bar = 10 µm. O Representative time series of an osteoblast division from Osx1-GFP::Cre labeled explant showing how nuclear displacement is measured (N = 8). Scale bar = 10 µm. P Graph showing significant difference in the displacement of medial and lateral sister nuclei after a division (two-tailed two-sample t-test, ***p < 0.0001, n = 499, N = 8).

Fig. 2. Ex vivo live imaging of bone expansion reveals progressive differentiation of osteoblasts.

A Max projection of Osx1-GFP::Cre labeled frontal bone at 0 h (E13.75) and 6 h, together with the osteogenic front (magenta) and individually tracked cells (cyan). Scale bar = 100 µm. B The relative displacement for the osteogenic front (magenta) and tracked cells (cyan) over the course of 6 h starting from E13.75, defined as the displacement normalized by the total displacement of the osteogenic front at 6 h. Shaded areas show SEMs (N = 4, n = 199 tracked cells). C Box plot with all individual data points of rate of expansion at the osteogenic front compared to that of tracked cells, computed from the same data as in (B). The center represents the mean, and the bounds represent the standard error. D Average GFP intensity profiles along the medial-lateral axis for a labeled frontal bone in an ex vivo imaging experiment starting at E13.75. Each graph arises from one time frame, with the color indicated in the legend. E GFP labeling scheme in Osx1-GFP::Cre; R26RmT/mG explants. F Representative fixed tissue image of Osx1-GFP::Cre; R26RmT/mG (N = 6). The arrows with solid outline indicate cells that have both membrane and nuclear stain, and the arrow with dashed outline indicates a cell with only nuclear stain. G Max projections showing GFP localization in E13.75 live movies of Osx1-GFP::Cre; R26RmT/mG explants at 5 h time intervals. Scale bars = 50 µm. H Schematic of a coronal section with the bone labeled in cyan. The inset shows the imaged areas of I with the approximate domains of Sp7+ osteoblasts (cyan) and Runx2+ precursor cells (magenta). I DAPI, Runx2, and SP7 immunoreactivity at the osteogenic front in E14.5 coronal sections (N = 8).

Fig. 3. Biophysical model with mechanical feedback between stiffness and cell fate recapitulates imaging results.

A Schematic of the model. B Dynamics of the osteogenic front and the tracked cells for the theoretical model, the ex vivo live imaging experiments (N = 4), and in vivo fixed tissue images obtained by comparing bone sizes between E13.5 (n = 7) and E14.0 (n = 7). Shaded areas indicate SEM for the ex vivo data. C Cell velocity for the theoretical model compared to the velocities obtained from the ex vivo tracked cells over 6 h (Fig. 1H). Error bars show standard deviations in both horizontal and vertical directions. D The simulated fraction of osteoblasts shows a sigmoidal profile across space that travels as a wave. The horizontal axis shows the position along the medial-lateral axis, where the origin represents the initial position of the osteogenic front, here defined to be the location where ϕ = 1/2.

Fig. 4. Perturbing collagen crosslinking leads to a larger difference in stiffness across the tissue, resulting in larger bones.

A Schematic showing embryological removal of the bone center from skull explant for live imaging. B Stills showing Osx1-GFP::Cre fluorescence of live-imaged skull caps where the bone center has been excised. 0 min (yellow) and 290 min (blue) are overlaid (N = 5). Scale bar indicates 100 µm. C Schematic of the BAPN feeding protocol. D Schematic showing collagen (black) fibrils crosslinked by Lysyl Oxidase (magenta). E Schematic showing collagen without crosslinkers in BAPN-treated embryos. F TEM image of transected collagen fibrils in differentiating mesenchyme in E15.5 control embryos (N = 2, n = 100). Scale bars indicate 25 nm. G TEM image of transected collagen fibrils in differentiating mesenchyme from E15.5 BAPN-treated embryos (N = 22, n = 103 (Welch’s two-tailed t-test, p < 0.0002)). Scale bars indicate 25 nm. H Violin plot with median (solid line) and quartiles (dashed lines) showing cross-sectional area of collagen fibers in control (n = 100) and BAPN-treated embryos (n = 103) (Welch’s two-tailed t-test, p < 0.0002). I AFM measurements of the stiffness (apparent Young’s Modulus) at E15.5 at two different locations of the tissue, for control samples (cyan) and BAPN-treated samples (magenta) (N = 2, n = 409), with horizontal lines indicating the median (Kruskal–Wallis test, *p < 0.02, ****p < 0.0001). J Representative image of a control Osx1-GFP::Cre labeled frontal bone at E15.5 (N = 2, n = 6). K Representative image of a BAPN-treated Osx1-GFP::Cre labeled frontal bone at E15.5 (N = 3, n = 10). Scale bars indicate 200 µm. L Box plot with all individual data points showing measured frontal bone areas for the two conditions at E13.5. The center represents the mean, and the bounds represent the standard error. (WT, N = 2, n = 4, BAPN, N = 3, n = 13), E14.5 (WT, N = 2 n = 4, BAPN, N = 2, n = 4), and E15.5 (WT, N = 2, n = 6, BAPN, N = 3, N = 10) (Mann–Whitney test, p < 0.0001). M Line plot showing the normalized and aligned GFP intensity profiles at the osteogenic front in fixed control (green) and BAPN-treated (magenta) skull caps at E13.75. The shaded area represents values within one standard deviation of the mean. N Box plot showing the fitted slope of GFP intensities between control and BAPN-treated embryos represented in (M). The box extends from the first quartile (Q1) to the third quartile (Q3) of the data, with a line at the median. The whiskers extend from the box to the farthest data point lying within 1.5× the inter-quartile range (IQR) from the box. (** represents p < 0.01, two-sided Mann–Whitney U-test, p = 0.00786). O 2D schematic showing a proposed model of anisotropic frontal bone expansion at the level of individual cells. P 1D schematic showing proposed model for a self-propagating wave of differentiation, stiffness, and cell motion.