Integrins are required for synchronous ommatidial rotation in the Drosophila eye linking planar cell polarity signalling to the extracellular matrix

Abstract

Integrins mediate the anchorage between cells and their environment, the extracellular matrix (ECM), and form transmembrane links between the ECM and the cytoskeleton, a conserved feature throughout development and morphogenesis of epithelial organs. Here, we demonstrate that integrins and components of the ECM are required during the planar cell polarity (PCP) signalling-regulated cell movement of ommatidial rotation in the Drosophila eye. The loss-of-function mutations of integrins or ECM components cause defects in rotation, with mutant clusters rotating asynchronously compared to wild-type clusters. Initially, mutant clusters tend to rotate faster, and at later stages they fail to be synchronous with their neighbours, leading to aberrant rotation angles and resulting in a disorganized ommatidial arrangement in adult eyes. We further demonstrate that integrin localization changes dynamically during the rotation process. Our data suggest that core Frizzled/PCP factors, acting through RhoA and Rho kinase, regulate the function/activity of integrins and that integrins thus contribute to the complex interaction network of PCP signalling, cell adhesion and cytoskeletal elements required for a precise and synchronous 90° rotation movement.

Keywords: Drosophila eye; extracellular matrix; integrins; ommatidial rotation; planar cell polarity.

Figure 1.

Loss of βPS/mys integrin causes defects in OR. (a) Schematic presentation of photoreceptor (R-cell) differentiation posterior to MF. Anterior is left and dorsal up in all panels. The first R-cell to be specified is R8 (shown in purple). With the induction and specification of R2/R5 and R3/R4 pairs, the five-cell precluster is established (mystery cells (m) are initially also part of the precluster but later excluded). Subsequently, the R1/R6 pair, R7 and cone cells (c) are added. Ommatidia rotate an initial 45°, usually completed by columns 9–10 and then continue rotation until they reach their final 90°. Cartoons of typical ommatidia from columns 4, 7, 11, 16 and 22 are shown in third instar eye imaginal discs. The respective photoreceptors, R cells, are numbered. (b–b″) Confocal microscopy image of the third instar eye imaginal disc stained with anti-Elav (red: marking all R cells), anti-Bar (green; staining R1/R6 cells) and anti-Boss (purple; central R8 cell). (b′) Enlargement of boxed area in b; white arrows indicate rotation angles. (b″) Single ommatidium (boxed in b′). Numbers indicate respective R cells. (c–c″) Confocal microscopy image of wild-type third instar eye imaginal disc stained with anti-Elav (red: marking all R cells), svp-lacZ (green; staining the R3/R4 and R1/R6 cell pairs) and anti-Boss (purple; central R8 cell). (c′) The same eye disc, but lacking the Elav staining to easier appreciate OR angles, and (c″) is a monochrome of svp-lacZ staining. Note that both, anti-Bar (b–b″) and svp-lacZ (c–c″) are excellent markers to measure the respective rotation angles of individual ommatidia. See also electronic supplementary material, figure S1 for additional marker combinations used. (d–d″) Higher magnification area of eye imaginal disc with mys/βPS mutant clones (mys¹ allele is a reported protein null). Mutant tissue is marked by lack of β-gal staining (blue in d,d′ and shaded in grey in schematic in d″). R cells are marked with anti-Elav (red in d), and anti-Bar (green in d,d′, staining the R1/R6 pair and reflecting the rotation angle). (d″) Schematic of the rotation angles in wild-type (grey shaded area) versus mutant (white area) with bars reflecting angle; orange bars indicate rotation angles of R1/R6 pairs in mys¹ mutant or mosaic clusters; black bars mark orientation in fully wt clusters. Note that whereas black bars (wild-type) are close to a 45° at this developmental stage, orange bars display an abnormally wide range of angles (often reflecting an over-rotation of an ommatidium). Also, note that the majority of ommatidia in mys/bPS integrin mutant tissue are out of synchrony when compared with the neighbouring clusters. While some reach the final 90° rotation angle earlier than ommatidia of equivalent developmental stages in adjacent wild-type tissue, others lag behind (see also figures 2 and 3 for quantifications).

Figure 2.

βPS/mys integrin mutant clusters tend to rotate too far at early stages. (a) Confocal microscopy image of wild-type third instar eye imaginal disc stained with anti-E-cad (red), anti-Fmi (blue) and CnoGFO (green), representing typical wild-type examples of our quantifications with rotation angles (white arrows) measured in each cluster in the respective columns (marked by red numbers within two regions of the dorsal eye area: mid dorsal and more central, respectively). Inset in top right corner delineates how angle is determined relative to the E-cad/CnoGFP staining, which strongly marks the adherens junctions of R2/R5. For quantitative analyses, see also electronic supplementary material, figure S1, which shows the same disc with the respective rotation angle values (electronic supplementary material, figure S1A), and rotation angle arrows and values for discs stained for ant-Elav co-stained with anti-Bar (R1/R6 marker) and CnoGFP (electronic supplementary material, figure S1B) and svp-lacZ, anti-Elav and anti-Boss (electronic supplementary material, figure S1C). Equivalently, marked discs containing mutant mys¹ clones were used to establish rotation angles in mys/βPS mutant tissue (examples shown in electronic supplementary material, figure S2). (b) Comparative quantification of rotation angles of preclusters in columns 5–10 in wt and mys¹ mutant tissue presented in rosette diagrams. The presented intervals are in 5° wedges; the average is shown by green line and standard deviation (as a measure of angle spread and variation) in orange. 0° is horizontal right (as in MF at onset of rotation) and 90° (as in adults after completion of OR) is down in all examples. Columns 6–9 of wt and mys¹ mutant clusters are shown. Note that clusters in mys¹ mutants displayed generally a wider spread of rotation angles and at these early stages had a tendency to rotate farther early on. Statistical analyses with Student's t-test, p-values were: *<0.011 and **<0.001. See also electronic supplementary material, figure S3 for additional samples of comparative rotation angle quantifications.

Figure 3.

βPS/mys integrin mutant clusters do not rotate synchronously. (a) Confocal microscopy image of the third instar eye imaginal disc containing mys¹ mutant clones (marked by the absence of β-Gal expression, blue) stained with anti-Elav (red: marking all R cells) and anti-Bar (green; staining R1/R6 cells). The disc represents a typical example of mutant tissue analyses with rotation angles measured in each cluster (white arrows, angle values indicated), the respective columns are numbered in yellow in ventral disc area. See also electronic supplementary material, figure S2 for additional examples of discs containing mys¹ mutant clones. (b) Comparative quantification of rotation angles, presented as rosette diagrams, of clusters in columns 10, 12, 14, 16 and 18, and adult eyes in wt and mys¹ mutant tissue (the adult phenotype was recorded in sev-mys flies, which are dominant negative for βPS/Mys function, as analysed in figure 4). The presented intervals are in 10° wedges (blue); the average is shown by green line and the standard deviation (measure of angle spread and variation) in orange. 0° is horizontal right (representing angle prior to OR, see also figure 2) and 90° (representing completion of OR) is down in all samples. Note the much wider rotation angle distribution in mys¹ tissue (or in sev-mys adult eyes) when compared with the equivalent wt stages. See electronic supplementary material, figure S3 for additional examples of comparative rotation angle quantifications. (c) Comparative quantification of the difference in rotation angles between a cluster and its nearest neighbours within the same column. Data are grouped into columns 6–9 (left) and 10–18 (right). Individual data points are plotted in dark blue and the mean with standard deviation are shown (n = 69 and 158 for wt and n = 52 and 74 for mys¹). Statistical studies were performed with the Kolmogorov–Smirnov analyses, with focus on angle variation distribution. p-values were: *<0.05 and ***<0.0001. Note that average rotation angles were largely not different between wt and mutant, but the angle distribution was significantly changed, and there was greater variation in angle compared to the neighbouring clusters.

Figure 4.

Overexpressed βPS/mys acts as a dominant-negative causing OR defects. (a–d) Tangential sections through equatorial regions of adult eyes with schematic presentations indicating ommatidial orientation below each section. Dorsal and ventral ommatidia are indicated with black and red arrows, respectively. Anterior is left. Ommatidia unscorable for orientation (due to photoreceptor loss) are marked with dots. βPS/Mys is expressed under sevenless-Gal4 control (sev-mys) at the indicated temperature. (a) sev > mys at 25°C; such eyes are almost indistinguishable from wt, with ommatidia rotating largely at 90°. (b) sev-mys (2 copies) display OR defects. (c) sev-mys at 29°C (Gal4-driven expression is stronger at 29°C) causing mild rotation defects. (d) sev-mys; if^B2/+ (at 29°C). Heterozygosity for Integrin α-subunit, αPS2 (if−/+) enhances sev > mys, consistent with sev > mys acting as a dominant negative, by interfering with the balance between the α and β-integrin subunits. Note wide spread/distribution of rotation angles in backgrounds with ‘defective’ βPS/mys function. See figure 3b for quantification of genotype shown in b here.

Figure 5.

Localization of Mys/βPS–integrin during eye disc patterning. All panels show confocal images of third instar eye discs. Anterior is left and dorsal is up. Integrin βPS subunit, visualized by a specific antibody (red in a–c, and single channel in a′, b′ and c″), is highly enriched at membranes that are at the periphery of the developing ommatidial clusters, basal to adherens junctions (marked by DE-cadherin, green in a, b and d), and on the basal sides of each cell. (a–c) mys¹ clones (mutant for βPS–integrin subunit) are marked by the absence of GFP (blue in a and c, single channel in c′). (b) Enlargement of boxed area in (a). Approximately 9–11 columns posterior to the morphogenetic furrow, βPS/Mys is also detected around the cells of the R1/R6 pair (red arrow in b′). Numbers 1–10 mark the respective columns posterior to MF, the respective R cells are labelled with small white numbers (1–6) in most posterior cluster shown. See electronic supplementary material, figure S4 for more on R1/R6 expression. (c,c′) The loss of βPS/Mys does not cause apical–basal polarity defects, nor does it affect precluster integrity in third instar eye discs, as apical markers (DE-cadherin, green in a) and markers for the basolateral plasma membrane (FasIII, green in c; single channel in c′) are unaffected. (d) Three-dimensional rendering of confocal stacks showing the βPS/Mys localization (red) baso-laterally, having a ‘basket’-like appearance around each cluster. Photoreceptor nuclei are marked with anti-Elav (blue) and DE-cadherin (green) marks the apical adherens junctions.

Figure 6.

ECM components are required for OR. (a) Schematic presentation of selected components and factors associated with integrin-mediated adhesion and signalling. Integrin heterodimers (α and β chain shown in black) and components of the ECM are shown in the top half of the panel, components mediating integrin signalling in the cytoplasm are displayed in the lower half of the panel (ILK, integrin-linked kinase; FAK, focal adhesion kinase; vinc, vinculin). (b) Perlecan is required for correct OR. Tangential sections through adult eyes of the hypomorphic trol¹³ allele reveal OR defects. Ommatidial orientation is indicated with arrows (schematic to the right); arrows are as in figure 4. (c–c′) Confocal image of the third instar eye imaginal disc showing clones mutant for a laminin-α chain (wb^4Y18). Mutant tissue is marked by lack of β-gal staining (blue); photoreceptors are marked by anti-Elav (red) and anti-Bar (green), highlighting the R1/R6 pair reflecting ommatidial orientation. Ommatidia within mutant and wt tissue are misrotated indicating the non-autonomous function of secreted Laminin. (c″) Schematic of ommatidial orientation from c–c′: white areas represent wb/laminin-α mutant tissue; black arrows indicate rotation angles (as deduced from R1/R6 cell arrangement relative to equator).

Figure 7.

A dominant-negative integrin-signalling chimera results in rotation defects. (a–d) Tangential sections through equatorial regions of adult eyes, with schematics indicating ommatidial orientation below each section. Arrows are as in figure 4. Unscorable ommatidia (due to loss of R cells) are marked with dots. The chimeric torso^DβPS_cyt protein is expressed under sevenless-Gal4 control (sev > torso^DβPScyt) at 29°C. (a) sev > torso^DβPS_cyt: note approximately 10% misrotated ommatidia, (b) sev > torso^DβPS_cyt; wb^EY18/+. (c) sev > torso^DβPS_cyt; ILK¹/+. (d) sev > torso^DβPS_cyt; fmi^E54/+. Note that in addition to rotation defects, reducing the levels of these genes also caused an increase in R-cell number defects (loss of R cells—dots—and R7 to R1–6 transformation—asterisk in d—in schematic). (e) Quantification of sev > torso^DβPS_cytinteractions: * and ** indicates p-values < 0.05 and <0.005, respectively (with Student's t-test). Whereas no effect is seen in a wb/laminin heterozygous background as expected (b), reduced dosage of ILK (c) and fmi (d) enhances the sev > torso^DβPS_cyt phenotype.

Figure 8.

Activated Rho kinase causes accumulation of βPS/integrin levels. Confocal images of the third instar eye imaginal disc with activated dRok expression clones (generated via the Flip-out technique) marked by simultaneous GFP expression (green). R-cell neurons are labelled by anti-Elav (blue), βPS/Mys integrin staining is in red (single channel in monochrome panels a′ and b′). (a,a′) Both posterior and anterior to the furrow, βPS/Mys levels are upregulated in dRok-expressing clones (clone anterior to furrow is ‘overlayed’ by a peripodial membrane clone: thus, clone boundary and integrin staining do not appear exactly coincident, since dRok in peripodial membrane cells does not cause βPS/Mys upregulation). (b,b′) Enlargement of boxed area posterior to the furrow in (a); note that dRok-dependent βPS/Mys accumulation is mainly detected at the outside membranes of preclusters (where ‘rotation takes place’) as in wild-type (see figure 5 for wild-type details for comparison).