Suppression of epithelial folding at actomyosin-enriched compartment boundaries downstream of Wingless signalling in Drosophila

ABSTRACT Epithelial folding shapes embryos and tissues during development. Here, we investigate the coupling between epithelial folding and actomyosin-enriched compartmental boundaries. The mechanistic relationship between the two is unclear, because actomyosin-enriched boundaries are not necessarily associated with folds. Also, some cases of epithelial folding occur independently of actomyosin contractility. We investigated the shallow folds called parasegment grooves that form at boundaries between anterior and posterior compartments in the early Drosophila embryo. We demonstrate that formation of these folds requires the presence of an actomyosin enrichment along the boundary cell-cell contacts. These enrichments, which require Wingless signalling, increase interfacial tension not only at the level of the adherens junctions but also along the lateral surfaces. We find that epithelial folding is normally under inhibitory control because different genetic manipulations, including depletion of the Myosin II phosphatase Flapwing, increase the depth of folds at boundaries. Fold depth correlates with the levels of Bazooka (Baz), the Par-3 homologue, along the boundary cell-cell contacts. Moreover, Wingless and Hedgehog signalling have opposite effects on fold depth at the boundary that correlate with changes in Baz planar polarity.


Drosophila strains
the genotypes for each figure. Name Genotype Origin yw y 1 w 67c23 Bloomington #6599 wg CX4 wg CX4 Baker, 1987 enlacZ en-lacZ Busturia and Morata, 1988 rokGFP Ubi-Rok::GFP Gift from Vincent Mirouse flwYFP flwYFP CPTI-002264 Lowe et al,, 2014;Lye et al.,2014 bazGFP bazGFP CC01941 Buszczak et al., 2007 prdGal4 prd-Gal4 Bloomington #1947 (Brand & Perrimon 1993) UASdeGradFP UAS-deGradFP Caussinus et al., 2011 armGal4 arm-Gal4 Sanson et al., 1996 UASwg UAS-wg Lawrence et al., 1995 MTDGal4 otu-GAL4::VP16, w * ; GAL4-nos.NGT; GAL4::VP16-nos.UTR Bloomington #31777 (Petrella et al., 2007) UASbazGFP Bloomington #7251 (Seugnet et al 1997) Immunostainings Embryonic staging was as in (Hartenstein and Campos-Ortega, 1985). Embryos were collected in a basket from one-hour collections on plates containing apple or grape juice hardened with agar. They were dechorionated by immersion in commercial bleach diluted 1:2 in pure water, for 2 minutes, rinsed, blotted dry and then transferred into heptane. For most experiments, embryos were fixed for 5 minutes in the interface of a 1:1 solution of Heptane:Formaldehyde 37% followed by manual devitelinization in PBS with 0.1% Triton X-100 in (PTX). For staining against phospho-Moesin, 10% trichloroacetic acid in dH2O was used instead of the formaldehyde, and the embryos fixed on ice for 1 hour. Embryos were then blocked with PTX containing 1% bovine serum albumin (PTB) for 30 minutes, and incubated overnight at 4ºC with primary antibodies. They were washed three times for 15 minutes in PTX, then incubated for one hour with secondary antibodies in PTB. They were washed a further three times in PTX, and stored at -20ºC in Vectashield (Vector laboratories). When biotin-conjugated secondary antibodies were used an extra step was used. After the second antibody washes the embryos were incubated with streptavidin-conjugated Alexa-405 for 30 minutes before three further washes in PTX, and stored at -20ºC in Vectashield.

Confocal imaging
Embryos were mounted individually under a coverslip supported by a tape bridge on either side. This flattened the embryos sufficiently so that all cells were roughly in the same z-plane. Embryos were imaged on a Nikon Eclipse TE2000 microscope incorporating a C1 Plus confocal system (Nikon) and images captured using Nikon EZ-C1 software; or, a Leica TCS SP8 confocal microscope and images captured using LAS X software (Leica). Optical zstacks were acquired with a depth of 0.25 µm between successive optical z-slices. All embryos were imaged using a violet corrected 60x oil objective lens (NA of 1.4). The gain and offset were optimized for each embryo.

Quantification of enrichment at PSBs
Two stages were used for quantification: stage 10 embryos in all genotypes except for arm>wg, where late stage 9 embryos were analyzed to avoid too much folding at ectopic boundaries. Quantifications were done in maximum intensity projections, which were made from the minimum number of z-slices needed to contain all the adherens junction signal. The adherens junctions were labelled by staining for either E-Cadherin or phospho-Tyrosin. Cortical signal of different proteins was quantified on line traces that went over cell interfaces. The position of the PSB was identified by co-staining with anti-En or anti-Wg, except for wg CX4 embryos in which these markers are gradually lost; in this case, an enlacZ transgene was used and staining with anti-ßGal showed the PSB location (ßGal protein has longer perdurance than En protein in embryos). The lines were manually traced by using the FIJI plugin Simple Neurite Tracer (Longair et al., 2011) or the ImageJ plugin NeuronJ (Meijering et al., 2004) based on membrane marker stainings and avoiding dividing cells. Average fluorescence intensity was quantified for 3-pixel wide line traces using ImageJ or FIJI (Schneider et al, 2012;Schindelin et al, 2012). We used the image-wide modal pixel intensity as an approximation of the average background fluorescence. The modal pixel intensity was then subtracted from all pixels to remove background fluorescence from the signal. PSB and Ectopic Boundary interface fluorescence intensity was then normalised to En interface fluorescence intensity for each PSB quantification (Example in Figure 1 D", G"), with the exception of Baz, for which it was normalized to DV tracks outside the Engrailed domain (Example in Figure 1E", H"). This is because contrarily to the other proteins we looked at, Baz shows a very weak remaining planar polarity at stage 10, in particular in the Engrailed domain. Statistics were performed in Prism (GraphPad). Pilot experiments were used to establish that n ≈ 20 PSBs was appropriate for the detection of enrichments or depletions. Data from all quantifications are reported as mean ± 95% confidence intervals. Results were considered significant when p < 0.05 (* when p < 0.05, ** when p < 0.01, *** when p < 0.001, **** when p < 0.0001).

3D image segmentation, quantification of cell areas, AJ position and index of straightness
Wildtype and arm>wg tage 10 embryos were stained with Engrailed and E-Cadherin antibodies as well as CF594-Phalloidin to mark PSBs, adherens junctions and actin respectively. Then, embryos were mounted under a coverslip suspended by a two-layer thick tape bridge on either side. The samples were imaged on a Leica TCS SP8 confocal microscope (CAIC, University of Cambridge). Optical z-stacks were acquired with a depth of 0.33 µm between successive optical z-slices, which is the optimal z interval thickness of the 63X objective used. The gain and offset were optimized for each embryo. Fluorescence images were denoised (Boulanger et al., 2010) and segmented using Real-time Accurate Cellshape Extractor (Stegmaier et al, 2016). Cell top was detected by the apical medial actin enrichment while cortical actin decorated cell contour. Segmented images were used in ImageJ to manually select cells of different populations (Control, PSB and ectopic PSBs: ECT) in wildtype and arm>wg embryos. Selected cells were saved as region of interests and used to quantify cell area per stack and 3D render. Custom written MATLAB scripts computed cell areas for the chosen cells in each plane of the stack.
For the adherens junctions apico-basal position, analysis contours were generated as described above for the quantification of protein enrichments at PSBs and saved as 2D binary masks. The cell walls corresponding to the regions of interest were determined by propagating these contours as open snakes on the cortical Phalloidin channel intensities (Shemesh and Ben-Shahar, 2001). These cell walls were then used to quantify the distance between the adherens junctions (E-cadherin) and the top of the cell, detected by medial actin (Phalloidin). The positions of the adherens junctions were given by the maxima of E-cadherin channel values in z direction along the wall. An estimate of the top of the cell was obtained by segmenting the Phalloidin channel stack in 2D (xz direction) via robust statistics based thresholding of the wavelet coefficients of the image. 2D projections of intensities in the Ecadherin and Phalloidin channel (across the width of the bounding box for each input contour) were saved as a mean of quality control by visual inspection. The distance between adherens junctions and the closest point of the cell top was computed taking into account voxel anisotropy. Finally, as a post-processing step of removing outliers, the highest 10% of distances were discarded for each region.
The index of straightness (IS, (Monier et al., 2010) was computed for each propagated contour in each plane of the 3D stack over a depth of 5 microns (starting from 0.6 microns above the adherens junctions). It is calculated as representing the percentage of curve length exceeding the length of the straight line joining the curve's endpoints: IS = (length of curve / distance between the two endpoints of the curve -1) * 100 Scanning Electron microscopy Embryos were fixed for 5 minutes in Heptane:Formaldehyde 37% (1:1) and devitellinised with Heptane:Methanol (1:1). Then, they were re-fixed immediately in 2% Glutaraldehyde, 2% Formaldehyde, 0.05M Sodium Cacodylate pH 7.4 and 2mmol/L Calcium Chloride overnight. Once rinsed twice in deionised water, embryos were treated with 1% osmium ferricyanide for 3 days. After that they were rinsed four times in deionised water, dehydrated to 100% ethanol and dried by either critical point drying, or drying from hexamethyldisilazane (HMDS). Where HMDS was used, embryos were transferred into 1:1 HMDS:ethanol for 10 minute, then HMDS for 10 minutes twice, and left to dry. Dry embryos were mounted on carbon tabs on 12.5 mm Cambridge stubs and sputter coated with 50nm of gold. Images were taken in a FEI XL30 FEG scanning electron microscope operated at 5 kV.

Y-27632 Rho kinase inhibitor injections
Early stage 9 arm>wg embryos were injected through the posterior into the yolk at room temperature with 1 mM Y27632 in dH2O, and dH2O in control experiments. Embryos were aged for 45 minutes at 25°C, then fixed in 8% formaldehyde over heptane for 20 minutes. They were rinsed with PBS, manually devitellinised by nicking with a needle, and then fixed for SEM as above.

Live Imaging
Dechorionated embryos were transferred into halocarbon oil (Voltalef PCTFE, Arkema), mounted ventral side up on stretched oxygen-permeable membrane, and covered with a coverslip supported by a bridge of a single coverslip on either side. Timelapse imaging was carried out on a Nikon Eclipse E1000 equipped with a spinning disk unit (Yokogawa CSU10), laser module with 491nm and 561nm excitation (Spectral Applied Research LMM2), and a C9100-13 EM-CCD camera (Hamamatsu). Z-stacks were acquired with an interval of 0.7µm or 1µm. Images were captured using Volocity software (PerkinElmer). Where multiple images were stitched together (Fig S4C), the FIJI plugin Grid/collection Stitching was used (Preibisch et al., 2009).
Laser ablations and analysis of recoil velocities Laser ablation experiments were carried out on a TriM Scope II Upright 2-photon Scanning Fluorescence Microscope controlled by Imspector Pro software (LaVision Biotec) using a tuneable near-infrared (NIR) laser source delivering 120 femtosecond pulses with a repetition rate of 80 MHz (Insight DeepSee, Spectra-Physics). The laser was tuned to 927nm, with power ranging between 1.40-1.70 W. The maximum laser power allowed to reach the sample was set to 220 mW and an Electro-Optical Modulator (EOM) was used to allow microsecond switching between imaging and treatment laser powers. The laser light was focused by a 25x, 1.05 Numerical Aperture (NA) water immersion objective lens with a 2mm working distance (XLPLN25XWMP2, Olympus). Images were collected every 0.731 ms for 5 frames before the ablation and 60 frames after the ablation.
Ablations were performed during image acquisition (with a dwell time of 9.27 µsec per pixel), with the laser power switching between treatment and imaging powers as the laser was raster scanned across the sample. Targeted line ablations of about 2 µm length were performed at the centre of junctions on the PS boundary or on control, non boundary dorsoventral (DV) oriented or antero-posterior (AP) oriented junctions, using a treatment power of 220 mW. 20-25 ablations per condition per genotype were carried out, 2-4 ablations per embryo.
To analyse recoil velocities, a kymograph spanning the ablated region was extracted using the dynamic reslice function in Fiji, and the distance between the two ends of the cut was measured up to 30 seconds after ablation. Linear regression was performed on the first 5 timepoints after ablation and the slope of the regressed line was used to measure the recoil velocity of the cut ends.

Panel
Parental genotype (         Movie 1: Myosin-II-like localisation of Flw-YFP during mesoderm invagination and early germband extension. Flw-YFP can be seen in the apices of presumptive mesoderm cells, and is present in the medial pulsatile flows at the apical cortex of cells in the extending germband, as well as being planar polarised at their junctions.