Oscillatory cortical forces promote three dimensional cell intercalations that shape the murine mandibular arch

Abstract

Multiple vertebrate embryonic structures such as organ primordia are composed of confluent cells. Although mechanisms that shape tissue sheets are increasingly understood, those which shape a volume of cells remain obscure. Here we show that 3D mesenchymal cell intercalations are essential to shape the mandibular arch of the mouse embryo. Using a genetically encoded vinculin tension sensor that we knock-in to the mouse genome, we show that cortical force oscillations promote these intercalations. Genetic loss- and gain-of-function approaches show that Wnt5a functions as a spatial cue to coordinate cell polarity and cytoskeletal oscillation. These processes diminish tissue rigidity and help cells to overcome the energy barrier to intercalation. YAP/TAZ and PIEZO1 serve as downstream effectors of Wnt5a-mediated actomyosin polarity and cytosolic calcium transients that orient and drive mesenchymal cell intercalations. These findings advance our understanding of how developmental pathways regulate biophysical properties and forces to shape a solid organ primordium.

Fig. 1

Cell cycle times and tissue properties are insufficient to explain mandibular arch shape. a Sagittal OPT renderings of the right mandibular arch in the mouse embryo at different stages. b Cell cycle time was measured in 24 adjacent regions (4 epithelial and 4 mesenchymal regions each within proximal, middle and distal arch) of the 19 somite stage mandibular arch. c, d Spatial variation of epithelial (c) and mesenchymal (d) cell cycle times in the mandibular arch. Cell division was more rapid in the proximal region for both epithelial and mesenchymal layers; n = 3 embryos at 20 somite stage, 15–35 cells examined for each of 12 epithelial regions per embryo, 50–75 cells examined for each of 12 mesenchymal regions per embryo; asterisks denote p < 0.05, Student’s t-test, error bars denote standard error of the mean (s.e.m.). e Tissue indentation by AFM was employed to measure properties of intact, live mouse embryos. f Elastic (Young’s) modulus (stiffness) of epithelium and mesenchyme. g Viscosity of whole tissue in proximal, middle and distal regions of the mandibular arch at 19 and 21 somite stages. For f and g, 15 separate sites in each proximal, middle and distal region were indented in triplicate (45 measurements per region) per embryo; n = 2 embryos per condition, asterisks denote statistical significance, p value range: 10⁻⁶ to 10⁻¹⁹, two-tailed t-test, error bars denote standard deviation. h Finite element simulation of 4 h of growth beginning from the actual 19 somite stage mandibular arch shape to predict 21 somite stage shape. The model incorporated experimentally measured spatial variation of cell cycle time, elasticity and viscosity. Simulated growth (in blue) results in an arch that is shorter and broader than the actual 21 somite stage arch. Source data are provided as a Source Data file

Fig. 2

Two distinct patterns of growth characterise the mandibular arch. a 3D renderings of cell membranes labelled by mTmG based on live light sheet microscopy. Whole arch (left) and local cell neighbour relations (middle and right with each colour representing one cell) are shown. Scale bar: 40 μm. b Distribution of numbers of cell neighbours in middle (red curve) and distal (blue curves) mandibular arch (n = 2 embryos, 202 middle cells and 244 distal cells examined, p < 0.05, chi-squared; scale bar: 40 μm. c Voronoi tesselation of mesenchymal nuclei to estimate cell shapes. d, e Colour-coded spatial distribution of cell shape index (S/V^2/3) that correlates positively with liquid-like tissue phase (greatest in the middle region). f 4D tracks of a subset of mandibular arch cells are shown in two orthogonal views. Relatively directional tracks that were oriented distalward characterised the waist region, whereas short and tightly curved tracks characterise the bulbous region (representative of 3 embryos at 20 somite stage). g 3D dandelion plot of spatially colour-coded trajectories of cells at the start and end of a movie. Waist cells (red/orange) move predominantly outward whereas bulbous cells move in a relatively radial fashion. h Strain illustrated as deformation of a sagittal plane grid during a 150 min. movie. Nodes remain fixed to the same positions throughout the movie. Coloured lines to the left of each grid correspond to rostrocaudal rows of squares that correspond to dashed lines in i. Corresponds to Supplementary Movie 11. (Representative of 3 embryos at 20–21 somite stage.) i The proximodistal axis elongated (reflected by the positive ε_xx slope), while the midportion of the arch converged along the rostrocaudal axis (reflected by negative values of the ε_yy curve). Convergence of midportion tissue combined with expansion of distal tissue results in clockwise (positive ε_xy values) and counter-clockwise (negative ε_xy values) rotational deformation of adjacent regions, respectively (as reflected by the downward ε_xy slope). Dashed, coloured lines correspond to rostroacaudal rows of grid components of the strain tensor as depicted in h. The solid black line is the average of rostrocaudal strain tensors as it varied along the proximodistal axis. Source data are provided as a Source Data file

Fig. 3

Epithelial and mesenchymal cell rearrangements converge and extend the midportion of the mandibular arch. a Orientation of epithelial tetrads during T1 transitions at formation to resolution stages (separated by an arrow); n = 3 embryos at 21 somite stage. b Schematic representation of predominant orientation of epithelial T1 transitions in middle and distal regions. c Time series (taken from 120 min. time lapse moves) of volumes of mesenchymal cells of dual H2B-GFP;mTmG transgenic embryos visualised by light sheet microscopy at intermediate and high magnification. Select nuclei are coloured to show tissue and cell convergence at intermediate and small scales occurs in the middle, but not distal, region. (Representative of 5 embryos at 19–21 somite stage). d Schematic representation of oriented mesenchymal cell intercalations transverse to the axis of elongation in the middle region. e In the mid-portion of the arch, F-actin and phosphomyosin light chain (pMLC) were biased along proximal and distal epithelial and mesenchymal cell interfaces which is parallel to the rostrocaudal axis and to the direction of cell intercalations. The angular distribution of immunostain fluorescence intensity for epithelial (n = 4 embryos) and mesenchymal (n = 4 embryos) F-actin and epithelial phosphomyosin light chain (pMLC) (n = 5 embryos) relative to the arch long axis that was designated as 0^o was quantified in the middle region using SIESTA. Scale bar: 20 μm, asterisks denote p < 0.05, Student’s t-test, error bars denote s.e.m. Source data are provided as a Source Data file

Fig. 4

Vinculin force oscillations distinguish middle and distal regions of the mandibular arch. a Conditional Rosa26 knock-in mouse strains: full length vinculin tension sensor (VinTS), TFP (FRET donor) only control (VinTFP), vinculin tailless control (VinTL). b Tension sensor expression among epithelial cells in the mandibular arch with one cell cortex highlighted as region of interest. Colour scale shows range of lifetime (in nanonseconds, ns) and corresponding force values (in picoNewtons, pN). c Individual cell fluorescence lifetime values in middle (mid) and distal (dist) epithelium and mesenchyme of the mandibular arch. Boxplots show mean (x), median (---), central quartiles (coloured box), and range (transverse end bars); n = 15 cells per region in each of 3 embryos. d Representative vinculin force curves of individual cells in middle and distal regions, and donor only control. Lifetime readings were taken at two minute intervals, error bars denote s.e.m. e Multiple vinculin force curves. The sample variance of lifetime values, a measure of amplitude, was greater among middle (0.0201 ns) versus distal (0.0132 ns) mesenchymal cells (n = 5 embryos, 15 cells per embryo per condition, p = 0.03, t-test). Mean lifetime variance was lower among middle mesenchymal cells of control VinTL (0.0028 ns, p = 0.02, ANOVA) and VinTFP (0.0006 ns, p = 0.01, n = 3 embryos, 15 cells per embryo per condition) strains. f Correlations of VinTS fluorescence lifetime and calcium reporter X-rhod-1 fluctuation (defined as the percentage of X-rhod-1-positive area for each cell at each time point) in middle and distal arch regions over time (18 cells per region). Source data are provided as a Source Data file

Fig. 5

Deficient cell rearrangements in Wnt5a^−/− mutants. a The Wnt5a^−/− mutant mandibular arch is comparatively short and broad by OPT. b A finite element model incorporating the spatial variation of Wnt5a^−/− mutant cell cycle time, Young’s modulus and viscosity was employed to simulate 4 h of growth starting from the actual 19 somite stage Wnt5a^−/− mutant mandibular arch shape compared to the actual 21 somite stage Wnt5a^−/− mutant arch. c Ratio of rostrocaudal to proximodistal axis length (RC/PD) was comparatively better for finite element simulation of Wnt5a^−/− mutant growth than for WT. d, e Cells of H2B-GFP transgenic embryos were tracked by 4D light sheet microscopy and, according to a random walk model, persistence of cell movements (persistent time (d)) and direction (angle from the mean (e)) were diminished in Wnt5a^−/− mutants. CDF, cumulative distribution function. f Epithelial T1 transitions in the Wnt5a^−/− mutant mandibular arch did not tend to converge and extend the middle region as in WT (compare with Fig. 3b). g Angular change in long axis among resolving tetrads was diminished in Wnt5a^−/− mutant epithelium (n = lateral arch epithelium in each of 2–5 embryos per condition, asterisks indicate p < 0.05, Student’s t-test). h Distribution of mesenchymal cell face numbers was shifted rightward to higher values for Wnt5a^−/− mutant mesenchyme (n = 2 Wnt5a^−/− embryos, 92 middle cells, 146 distal cells, error bars denote s.e.m.). i Mesenchymal cells in the mutant middle region lacked the centripetal intercalary movements and longitudinal tissue flow as observed in the WT middle region (compare with Fig. 2f, g). Source data are provided as a Source Data file

Fig. 6

Actomyosin bias and vinculin tension were diminished in the Wnt5a^−/− mutant mandibular arch. a Spatial distributions of F-actin and pMLC were not biased in 21 somite stage Wnt5a^−/− mutants. The angular distribution of immunostain fluorescence intensity for epithelial (n = 4 embryos) and mesenchymal (n = 4 embryos) F-actin, and epithelial pMLC (n = 5 embryos) was quantified relative to the arch long axis that was designated 0^o using SIESTA. Scale bars: whole arch panel 100 μm, merge field 20 μm, asterisks denote p < 0.05, Student’s t-test, error bars denote s.e.m. b Average fluorescence lifetime values in middle (mid) and distal (dist) epithelium and mesenchyme of mandibular arch. Boxplots show mean (x), median (---), central quartiles (coloured box), and range (end bars); n = 15 cells per region x 2 (mutant) or 3 (WT) embryos. p values for pairwise WT/Wnt5a^−/− mutant comparisons are given in the graph. Of note, middle mesenchymal lifetime range was diminished in Wnt5a^−/− mutants and more closely resembled WT distal mesenchymal lifetime values. c Representative vinculin force curves of an individual Wnt5a^−/− mutant cell in the middle region that lacks the amplitude observed for WT cells in same region. d Multiple vinculin force curves; the variance (amplitude) of lifetime values for Wnt5a^−/− mutant middle mesenchyme (0.0122 ns, n = 3 embryos, 15 cells per embryo) was diminished relative to WT middle mesenchyme (0.0201 ns, n = 5 embryos, 15 cells per embryo, p = 0.04), but similar to WT distal mesenchyme (0.0132 ns, p = 0.38, ANOVA). Lifetime variance was similar between WT (0.0132 ns) and Wnt5a^−/− mutant (0.0096) distal mesenchyme (p = 0.13). e Variation of cytosolic calcium concentration using Fluo8 applied to embryos was quantified at 5 positions per cell over time. Corresponds to Supplementary Fig. 6D; scale bar 10 μm. Source data are provided as a Source Data file

Fig. 7

Wnt5a is a spatial cue and Yap/Taz and Piezo1 act downstream of Wnt5a. a Sox:Cre;Z/Wnt5a embryos expressed Wnt5a beyond the normal expression domains by in situ hybridisation. b OPT showing a short and wide mandibular arch in a 21 somite Sox:Cre;Z/Wnt5a embryo. c F-actin and pMLC biases were diminished in 20–21 somite Sox:Cre;Z/Wnt5a embryos. d Proportion of cells exhibiting nuclear YAP in the Wnt5a^−/− mandibular arch. Scale bars: 20 μm. e OPT showing phenotype of T:Cre;Yap^f/+;Taz^f/f mandibular arch. f Orientation of mesenchymal cell intercalations in the sagittal plane of mandibular arch mid-regions of WT and T:Cre;Yap^f/+;Taz^f/f embryos. Intercalation angles were binned into one of 4 arcs of 90^o each: anterior, posterior, proximal and distal (n for WT = 48, n for mutants = 42, where n = No. of groups of 5–7 cells). g Multiple vinculin force curves showed no difference in the variance (amplitude) of fluorescence lifetime values of Yap^f/+;Taz^f/f and Tcre;Yap^f/+;Taz^f/ middle mesenchymal cells (p = 0.185, ANOVA, n = 16 cells in each of two 20–21 somite embryos per condition). h PIEZO1 immunostain intensity was diminished in the Wnt5a^−/− mandibular arch. Scale bars: 20 μm. i The 21 somite Piezo1^−/− mutant mandibular arch partially phenocopied that of Wnt5a^−/− with a short and broad mid-portion. j YAP/TAZ and PIEZO1 partially mediate actomyosin bias and cortical oscillation amplitude, respectively, downstream of WNT5A to orient and promote mesenchymal cell intercalations. Source data are provided as a Source Data file