Planar polarization of the atypical myosin Dachs orients cell divisions in Drosophila

Abstract

Tissues can grow in a particular direction by controlling the orientation of cell divisions. This phenomenon is evident in the developing Drosophila wing epithelium, where the tissue becomes elongated along the proximal-distal axis. We show that orientation of cell divisions in the wing requires planar polarization of an atypical myosin, Dachs. Our evidence suggests that Dachs constricts cell-cell junctions to alter the geometry of cell shapes at the apical surface, and that cell shape then determines the orientation of the mitotic spindle. Using a computational model of a growing epithelium, we show that polarized cell tension is sufficient to orient cell shapes, cell divisions, and tissue growth. Planar polarization of Dachs is ultimately oriented by long-range gradients emanating from compartment boundaries, and is therefore a mechanism linking these gradients with the control of tissue shape.

Figure 1.

Planar polarization of Dachs is required for orientation of clone and tissue growth. (A, top) The Dpp and Wg morphogens are expressed along the A–P and dorsal–ventral (D–V) compartment boundaries, respectively, in the wing imaginal disc. (Bottom) The corresponding positions in the pupal wing. (B, top) Dachsous is expressed in a gradient from proximal (P) to distal (D). (Bottom) The corresponding gradient in the pupal wing. (C, top) Clone shapes (gray) in the wing pouch (the future wing) of the wing disc are oriented along the P–D axis. (Bottom) Clone shapes (gray) in the adult wing are also oriented along the P–D axis. (D) The Dachsous (Ds) gradient orients the planar polarization of Dachs (red) along the P–D axis such that Dachs localizes to the distal side of each cell's apical surface. (Bottom) A clone in the wing disc expressing Dachs-V5. (E–H) The shape of clones randomly induced by heat shock and marked by the expression of the lacZ gene (anti-βgal) was examined in the wing pouch. (E) Clones are oriented along the P–D axis in wild-type discs. (F) Clones fail to orient in dachs mutants. (G) Clones fail to orient in fat mutants. (H) Quantification of E–G (see Supplemental Fig. S6), with overlays of clone shapes shown on the right. Mean clone elongation ratios are as follows: wild-type = 3.4±1.2 (n = 55), dachs mutant = 1.6±0.8 (n = 40, P < 0.001), fat mutant = 1.6±0.5 (n = 41, P < 0.001). (I–K) Spindle orientations determined by Tub and phospho-HistoneH3 staining were quantified relative to the P–D axis for wild-type (n = 199) (I), dachs mutant (n = 236) (J), and fat mutant wing discs (n = 235) (K). (L) Wild-type wing. (M) dachs mutant wing. (N) fat mutant wing.

Figure 2.

Reorientation of Dachs with a new Dachsous gradient is sufficient to reorient tissue growth. (A) Diagram of the graded expression of dpp.Gal4, which is highest at the A–P compartment boundary and decays into the apical compartment. The dotted line shows the location of the region of interest. (B, top) Portion of a wild-type wing pouch showing oriented clones (βgal) along the P–D axis (vertical). A diagram of the endogenous gradient of Dachsous (Ds) in this portion of the wing pouch is shown to the left. (Bottom) Expression of Dachs-V5 using dpp.Gal4 shows polarization in the P–D axis. (C, top) Portion of a wing pouch expressing Dachsous (Ds) and GFP with dpp.Gal4 driver. Clones (βgal) orient according to the new Ds gradient. (Bottom) Expression of Dachs-V5 under dpp.Gal4 UAS.Ds shows polarization of Dachs along the new Ds gradient. (D) Quantification of spindle orientation in B (n = 70) versus C (n = 91). (E) Wild-type wing. (F) Pattern of dpp.Gal4 in the adult wing. (F′) A dpp.Gal4 UAS.Ds wing grows outward, perpendicular to the P–D axis, in the anterior compartment. (F″) Overlay with a wild-type wing. (G) Pattern of omb.Gal4 in the adult wing. (G′) An omb.Gal4 UAS.Ds wing grows outward, perpendicular to the P–D axis, in the both compartments. (G″) Overlay with a wild-type wing.

Figure 3.

Dachs promotes constriction of apical cell–cell junctions to control cell shape, which may orient the mitotic spindle. (A1–6) Frames (5 min apart) from a live-imaged cell division in the wing pouch. (A′) Cell junction diagram with geometric centers of neighboring cells marked. (A″) Ellipse drawn through vertices of dividing cells (elongation ratio = 1.45±0.31, n = 95) predicts the orientation of the mitotic spindle (light blue) and the plane of division. (B) Correlation between a dividing cell's elongation angle and orientation of cell division, both relative to the P–D axis. Pearson's correlation coefficient, r = 0.81 (n = 72). (C) dachs mutant clones (GFP-negative) dilate the apical surface (marked by E-cad). (D) Ectopic localization of Dachs around the circumference of apical junctions in a fat mutant clone (GFP-negative) constricts the apical surface. (E) In fat dachs double-mutant clones (GFP-negative) apical junctions are not constricted but are dilated like dachs clones. (F) Quantification of C–E (Supplemental Table S1; Supplemental Fig. S7). Error bars, standard deviation. (*) fat mutant cells are apically more constricted than wild-type (WT) (P < 0.001). (**) dachs and fat/dachs mutants are apically more dilated than wild type (P < 0.001) but indifferent from each other. (***) Overexpression of Dachs enhances constriction of fat mutant cells (P < 0.001). (G) A model of cell growth, shape, and divisions in an isotropic manner (top) and when planar-polarized by Dachs-mediated tension (bottom). (H) Oriented growth of dachs mutant clones is rescued when they are surrounded by wild-type (WT) cells. (I) Quantification of H versus wild-type clones. The elongation ratio of dachs clones surrounded by wild-type (WT) cells is 3.22±1.07 (n = 38), indifferent from wild-type clones (P = 0.3) (Fig. 1H).

Figure 4.

Computer simulations show that polarized tension is sufficient to orient cell divisions and tissue growth. See Supplemental Movies S2–S4. For detailed parameter explorations, see Supplemental Figures S6 and S7. (A) Isometric tension at junctions leads to isometric clone shapes. (B) Polarized tension at junctions orients most cell divisions in the P–D axis, producing clone shapes and P–D orientation similar to in vivo clones. (C) Isometric tension with cell divisions forcibly oriented in the P–D axis produces clone shapes that are more elongated and P–D-aligned than in vivo clones. (D) Quantification of simulated clone shapes compared with wild-type in vivo clones.

Figure 5.

Evidence against Dachs directly orienting the mitotic spindle by tethering it in the P–D axis. (A) The difference (Δ) between the angle of daughter cells (relative to P–D = 0) and the long axis of their mother cell (relative to P–D = 0) is near 0 if orientation of cell shape determines the orientation of division. The difference (Δ) is negative if the daughter cells are more P–D-aligned than their mother. According to the “cell shape orients the spindle” model, the difference (Δ) should be centered around 0. According to the “Dachs directly orients the spindle” model, the difference (Δ) should be negative, because divisions would be expected to be biased along the P–D axis regardless of the shape of the mother cell. (A′) Analysis of in vivo cell division data from live imaging. The spread of the difference (Δ) is symmetric around the P–D axis (0). For rounder cells (elongation ratio close to 1), the long axis is less defined, thus giving a wider spread of difference (Δ). (B) In simulation B (polarized tension and division in long axis), cell shape orients the mitotic spindle and the difference (Δ) is symmetric, with a wider spread of difference (Δ) for rounder cells, as in vivo. (C) In simulation C (isometric tension and polarized cell division), spindles are tethered in the P–D axis (by Dachs) and the difference (Δ) is no longer symmetric and is negative. There is also no correlation between the elongation ratio and spread of the difference (Δ). This remains true even when the spindle-tethering mechanism is made imprecise enough to produce the correct clone shape (not shown). These results indicate that, in vivo, Dachs does not directly control spindle orientation.