Tissue fluidity mediated by adherens junction dynamics promotes planar cell polarity-driven ommatidial rotation

Abstract

The phenomenon of tissue fluidity-cells' ability to rearrange relative to each other in confluent tissues-has been linked to several morphogenetic processes and diseases, yet few molecular regulators of tissue fluidity are known. Ommatidial rotation (OR), directed by planar cell polarity signaling, occurs during Drosophila eye morphogenesis and shares many features with polarized cellular migration in vertebrates. We utilize in vivo live imaging analysis tools to quantify dynamic cellular morphologies during OR, revealing that OR is driven autonomously by ommatidial cell clusters rotating in successive pulses within a permissive substrate. Through analysis of a rotation-specific nemo mutant, we demonstrate that precise regulation of junctional E-cadherin levels is critical for modulating the mechanical properties of the tissue to allow rotation to progress. Our study defines Nemo as a molecular tool to induce a transition from solid-like tissues to more viscoelastic tissues broadening our molecular understanding of tissue fluidity.

Fig. 1. Dynamics of ommatidial cluster rotation reveal a pulsatile, continuous process.

a Representation of ommatidial clusters posterior to the morphogenetic furrow (MF, indicated by green arrowhead) in the region of the equator (yellow dashed line); blue dots denote centroids of photoreceptors R2 and R5, highlighting the angle of clusters relative to the MF and thus rotation of clusters from their initial position as they mature (away from the MF) (Supplementary Fig. 1a–c for more details). b Snapshots at the indicated timepoints from a movie of E-cad::GFP depicting ommatidial pre-cluster maturation and rotation. Image segmentation is shown in the white overlay. See Supplementary Fig. 1 for more details. Centroids of R2 and R5 cells were used to measure the degree of rotation over time as shown in the right panel. Cells highlighted in red are the eight photoreceptors of the cluster. c Rotation (blue) and apical surface area (red) of ommatidial clusters as a function of time. Measurements from different ommatidia were temporally aligned with the time when R7-cell joined the cluster (highlighted in this and subsequent graphs by dark gray dashed line, light gray dashed line highlights time point when R1/R6 join the cluster). Line shows mean, and shaded region indicates standard deviation. d Normalized ommatidial rotation (Rt(90), blue) and normalized apical surface constriction (1 − A(t)/A₀, red) of individual clusters are shown as a function of time. Line shows mean, and shaded region indicates standard deviation. Note the close association between rotation and constriction. e Sigmoid fit (dashed line) to ommatidial rotation, R(t) (line shows mean, and shaded region indicates standard deviation). The angular velocity during rotation is calculated from the fit parameters as $\mathrm{\Omega }=60\times 90/4\tau ~11\left[{}^{\circ }/\mathrm{h}\right]$. f Sigmoid fit (dashed line) to the ommatidial area, A(t) (line shows mean, and shaded region indicates standard deviation). The rate of ommatidial constriction is calculated from the fit parameters as $C=60\times A_1/4\tau ~2.5\left[\mathrm{\mu }{\mathrm{m}}^2/\mathrm{h}\right]$. g Change in degree of rotation over time of an individual cluster showing $\mathrm{d}R\left(t\right)/\mathrm{d}t$, in blue (left axis), and R(t), orange (right axis). Pulsatile behavior is marked by orange shaded boxes denoting periods, or pulses, of positive rotation with anti-rotation behavior in between such periods (white areas). h Cross-correlation between standardized ommatidial rotation and constriction as a function of time lag for individual ommatidia, denoted by different colored lines. Note that the maximum cross-correlation for all ommatidia occurs at zero time lag, implying a simultaneous occurrence of these two processes and a close association between them. Measurements in c–f and h are averaged over 11 clusters from two pupae in >300 min movies. OM; ommatidia. All scale bars: 3 µm.

Fig. 2. Interommatidial cells display dynamic behaviors during ommatidial rotation.

a Snapshots of the same cluster shown in Fig. 1 highlighting interommatidial cells (ICs) that divide (green) and delaminate (yellow; these cells can also divide, but at least one of the daughters delaminates subsequently). The ICs analyzed were first level neighbors of the cluster at the R7 recruitment stage. b Average number of dividing (green) and delaminating (orange) ICs that were the first-level neighbors of the photoreceptor cluster, as a function of time. Data were binned into 90 min time periods and were averaged over all ommatidia. The mean and standard error of the mean is shown. Gray dashed lines indicate the recruitment of the R1/R6 pair and R7, respectively (also in d). c Selected ICs, highlighted with random colors, illustrating junction remodeling and neighbor exchange. d Average number of ICs that were the first-level neighbors of the photoreceptor cluster and underwent neighbor exchange in between frames plotted as a function of time. Line shows mean, and shaded region indicates standard error of the mean. e Snapshots with trajectories of IC centroids, with dots representing the current centroid position in the respective image. f Average cell speed and displacement of ICs during ommatidial rotation. The trajectories of ICs from all ommatidia were pooled, and the average displacement (white arrows) and cell speed (heat map) were measured. A 5-cell pre-cluster schematic with gray cells is indicated in the middle of the plot with a curved red arrow representing the direction of rotation. Measurements in b, d, and f are averaged over 11 clusters from two pupae in >300 min movies. All scale bars: 3 µm.

Fig. 3. Nmo kinase affects ommatidial rotation and interommatidial cell motility.

a Snapshots of control, nmo^mut (loss-of-function), and nmo^GOF (gain-of-function) clusters during ommatidial rotation, respectively. The eight R-cells are highlighted in red, the R2/R5 pair is identified by blue centroid dots. b Ommatidial rotation as a function of time in nmo^mut (red) and nmo^GOF (blue), compared to the wt control (green). Line shows mean, and shaded region indicates standard deviation (in this panel and other equivalent panels). c Apical area of ommatidial clusters as a function of time for the three genotypes, as indicated. d Comparison of area constriction as a function of time for representative ommatidial clusters for the three genotypes. Note that all three constrict their apical area (orange line and scale on the right), but the constriction dynamics, measured by constriction rate (blue line and scale on the left), vary between the genotypes. In wt control, constriction happens in pulses followed by expansion (indicated by orange shaded boxes). In nmo^mut, the expansion periods were fewer as compared to control, while in nmo^GOF, expansion periods were longer and more frequent relative to control. e Violin plot of cluster constriction for control, nmo^mut, and nmo^GOF as indicated, measured as the total area during constriction periods. f Violin plot of cluster expansion for control, nmo^mut, and nmo^GOF genotypes measured as the total area during expansion periods. (***p = 0.005 and 0.004, respectively). g Snapshots of randomly selected ICs colored to illustrate neighbor exchange in control, nmo^mut, and nmo^GOF, respectively. h Fraction of ICs that were the first-level neighbors of the photoreceptor clusters and underwent neighbor exchange in between frames plotted as a function of time. i Snapshots with IC centroid trajectories for the three genotypes, each differently colored, in a 3 h window. j Violin plot of average cell displacement for ICs (***p = 5.76 × 10⁻³²). k Average cell speed and displacement of ICs during ommatidial rotation in control, nmo^mut, and nmo^GOF, respectively. For each genotype, the trajectories of ICs from all ommatidia were pooled, and average displacement (black arrows) and cell speed (heat map) were measured. All scale bars: 3 µm. Measurements in b, c, e, f, h, j, and k are averaged over 17 nmo^mut ommatidia from two pupae and 18 nmo^GOF ommatidia from two pupae from movies >300 min. To calculate the p-values in (f and j), two-tailed Student’s t-test was used.

Fig. 4. Nmo regulates cell shape in eye and wing tissues.

a Snapshots of representative ommatidia (at stage of R7 joining clusters) in wt control, nmo^mut, and nmo^GOF, where interommatidial cells are colored based on their shape index ($p_0=\mathrm{cell}\mathrm{perimeter}/\sqrt{\mathrm{cell}\mathrm{area}}$). The color map is shown. Scale bar: 3 μm. b Violin plots of the perimeter length of interommatidial cells for each respective genotype, black line shows the mean here and all subsequent panels (***p = 2.06 × 10⁻⁸² and 3.13 × 10⁻²⁷, respectively). c Violin plot of the area of interommatidial cells for the respective genotype (***p = 1.16 × 10⁻⁶³ and 1.20 × 10⁻²¹, respectively). d Violin plot of the shape index of interommatidial cells for the respective genotype (***p =  2.44 × 10⁻²⁵ and 4.06 × 10⁻⁹, respectively). e Standard deviation of shape index as a function of average shape index for all the first and second level neighbors of an ommatidium. Each ommatidium is represented as a single dot, and color-coded by respective genotype as indicated. The shaded ellipse covers the smallest area that is enclosed by the individual points (note the linear relation between these two quantities). f Mosaic wing tissue (fixed) at 22 h after puparium formation (APF) with cells colored based on their shape index; a nmo^mut clone (nmo^DB null allele) and nmo^GOF cells are juxtaposed to wt control cells, as indicated on left. Blue line marks the boundary between experimental (nmo^mut or nmo^GOF) cells and control cells. The color map is shown. Scale bar: 5 µm. g Violin plot of the cell perimeter in control and nmo^mut, and in control and nmo^GOF 22 h APF pupal wing tissues (***p = 7.64 × 10⁻⁶² and 4.49 × 10⁻¹⁴¹, respectively). h Violin plot of the cell area in control and nmo^mut, and in control and nmo^GOF wing tissues (***p = 1.12 × 10⁻⁵⁵ and 4.05 × 10⁻¹⁰⁵, respectively). i Violin plot of the shape index in control and nmo^mut, and in control and nmo^GOF wing tissues, respectively (***p = 2.37 × 10⁻¹⁴ and 1.35 × 10⁻⁸⁰ respectively). Note that the change in cell perimeter, area, and shape index between different genotypes are consistent in the eye and wing tissues. To calculate the p-values in (b–d) and (g–i), two-tailed Student’s t-test was used. In a–e, the n values were as follows: 290 cells/11 ommatidia (control); 430 cells/17 ommatidia (nmo^mut), and 450 cells/18 ommatidia (nmo^GOF), each from two pupae. In f–i, the n values were as follows: >700 cells from three mosaic wings (control and nmo^mut); and four mosaic wings (nmo^GOF).

Fig. 5. Nmo controls E-cad recycling and affects junctional E-cad steady state levels.

a, b Mosaic 22 h after puparium formation pupal wings (fixed tissue) with either nmo^mut clones (a, nmo^DB null allele) or nmo^GOF (b, overexpressed under ptcGal4 driver control) demarcated by the yellow line showing in (a): E-cad::GFP (green and monochrome), Discs large (Dlg, blue) and Histone2A::RFP (His::RFP, red, marking wt control cells); and antibody staining of E-cad (green, monochrome) in (b), Patj (blue), and Ptc (red, marking nmo GOF overexpression region). Note a visual increase in junctional E-cad level in nmo^mut cells (a) and a decrease in nmo^GOF cells (b) relative to wt controls (see Ext Data Fig. 7h for monochrome presentation of control Dlg and Patj staining). Scale bar: 5 μm. c Boxplot of normalized intensity of junctional E-cad levels averaged over 300 adherens junctions from 3 mosaic wing tissues in control, nmo^mut, and nmo^GOF cells (mean and standard deviation; ***p = 4.74 × 10⁻¹¹⁹ and 1.19 × 10⁻⁵³, respectively). d–f Snapshots of eye tissue Fluorescence Recovery After Photobleaching (FRAP) experiments of E-cad::GFP in control, nmo^mut, and nmo^GOF. Bleaching occurred over a 10 µm junction along two adjacent cell boundaries (shown by orange arrowheads). Scale bar: 3 µm. g Measured recovery fraction after photobleaching as a function of time for control, nmo^mut, and nmo^GOF. Line shows mean, and shaded region indicates standard deviation. h Boxplot of initial recovery rate after photobleaching for the respective genotype measured as the slope of the fitted exponential curve at t = 0 (see Supplementary Information; *p = 0.03; ***p = 2.77 × 10⁻⁶). i Boxplot of half recovery time after photobleaching for the respective genotype measured from fitted exponential curve (see Supplementary Information; *p = 0.01). j Tangential sections of adult eyes (showing a region around the equator) of the indicated genotypes with corresponding schematics below. Chiral ommatidia are depicted as black (dorsal) or red (ventral) flagged arrows. Note that the sev > Nmo GOF phenotype (left panels) is suppressed by heterozygosity for dynamin (shi^FL54/+, sev > Nmo; right panels). k Quantification of eye sections from genotypes shown in (j) presented as violin plots of ommatidial angles in adult eyes (***p = 1.71 × 10⁻⁷). Red line indicates mean value. Note that shi−/+ heterozygosity causes a reduction of widespread angle distribution and loss of additional smaller peak around 270°, as compared to sev > Nmo alone. In c, three mosaic nmo^mut wings from two pupae and four nmo^GOF mosaic wings from two pupae were analyzed. Box plots in panels c, h, and i: black line indicates the median, the edges of the box are 25 and 75% percentiles, and the whiskers extend to the most extreme datapoints. Panels g–i are averaged over 10 control, 9 nmo^mut, and 5 nmo^GOF ommatidia. Panel k is averaged over 408 ommatidia from 3 eyes for sev > Nmo and 330 ommatidia from 3 eyes in shi−/+, sev > Nmo. To calculate the p-values in c, h, and i, two-tailed Student’s t-test was used. To calculate p-values in k, two-sample Kolmogorov−Smirnov test was used.

Fig. 6. Nmo regulates tissue mechanics and tension at adherens junctions.

a, c, e Snapshots of laser ablation experiments performed at the level of a single junction in wt control, nmo^mut (LOF), and nmo^GOF conditions. Orange arrowheads indicate ablated junctions. Note that in nmo^mut (LOF), the excessive pre-existing stress in the tissue causes ruptures in non-ablated junctions (red arrowhead in the right panel in c) Scale bar: 3 µm. b, d, f, Kymographs of ablated junctions for the respective genotypes (each pixel = 2 s). g Average boundary recoil after ablation in wt control (green, n = 5), nmo^mut (red, n = 7), and nmo^GOF (blue, n = 8) tissues. Line shows mean, and shaded region indicates standard deviation. Exponential fitted curves for wt control and nmo^mut are also shown. For nmo^GOF, an initial expansion was immediately followed by reformation and shrinkage of the ablated boundary, and an exponential function does not correctly fit this data. h Table showing the initial recoil speed and recoil amplitude for wt control and nmo^mut measured from the exponential fitted curves shown in (g). i Schematic summary of the effects of Nmo during rotation and the associated tissue fluidity requirements for rotation. Ommatidial cluster cells are colored in red and ICs in gray in all panels. The thickness of black lines between cells, representing AJs, correlates with levels of E-cad. Blue double arrows indicate tension levels, and red circular arrows the level of E-cad recycling. As E-cad recycling and tissue fluidity increase from nmo^mut, through wt (control) to nmo^GOF, tension decreases, facilitating cluster rotation in wt and ultimately allowing over-rotation in nmo^GOF.