Epithelial cells adapt to curvature induction via transient active osmotic swelling

Abstract

Generation of tissue curvature is essential to morphogenesis. However, how cells adapt to changing curvature is still unknown because tools to dynamically control curvature in vitro are lacking. Here, we developed self-rolling substrates to study how flat epithelial cell monolayers adapt to a rapid anisotropic change of curvature. We show that the primary response is an active and transient osmotic swelling of cells. This cell volume increase is not observed on inducible wrinkled substrates, where concave and convex regions alternate each other over short distances; and this finding identifies swelling as a collective response to changes of curvature with a persistent sign over large distances. It is triggered by a drop in membrane tension and actin depolymerization, which is perceived by cells as a hypertonic shock. Osmotic swelling restores tension while actin reorganizes, probably to comply with curvature. Thus, epithelia are unique materials that transiently and actively swell while adapting to large curvature induction.

Keywords: cell volume; curvature; epithelial mechanics; mTORC2; membrane tension.

Graphical abstract

Figure 1

Self-rolling substrates induce cylindrical deformation of epithelial cell monolayers (A) Schematic of concave curvature generation on a (Ai) multi-layer substrate composed of: glass slide (light blue); PDMS sheet (dark blue); central gelatin patch (brown); bottom PDMS leaflet and toluene solution (yellow); top leaflet of PDMS, toluene, and silicon oil solution (pink); (Aii) oil extraction by isopropanol bath and formation of the strain gradient; (Aiii) surface activation by plasma treatment on the central region through a PDMS mask; (Aiv) fibronectin solution (green) incubation on the central region; (Av) cell solution (light pink) incubation; (Avi) substrate cutting after cell monolayer formation (inset: orthogonal view of a polarized cell layer, ezrin [red], cell membrane [Myr-Palm-GFP, green], and nuclei [H2B-mCherry, magenta]. Scale bars, 20 μm); (Avii) self-rolling of the substrate upon cutting. (B) Top (top) and side (bottom) views of a PDMS tube without cells, dextran solution (green). Scale bars, 50 μm. (C) 3D view (left, scale bars, 100 μm), side (middle) and top (right) views (scale bars, 50 μm) of a PDMS tube with cells, cell membrane (deep-red CellMask, green), and nuclei (H2B-mCherry, magenta). (D) Distribution of inner radii of tubes with cells. The red line stands for the mean value, n = 278 images, N of independent replicates = 3.

Figure 2

Cell volume increase and recovery are an active response (A) Cell volume measurements. Top, from left to right: 3D view of a cell monolayer on a flat PDMS substrate (scale bars, 50 μm); centroid identification of the nuclei (yellow circles) by segmentation (scale bars, 20 μm); 3D cell membrane segmentation; 3D reconstruction of cell shapes (zoom and large views, random colors represent different cells). Bottom, from left to right: 3D view of a cell monolayer on the top region of a PDMS tube; 3 representative examples of 3D reconstructions of cell volume on PDMS tubes. Scale bars, 50 μm. (B) Distribution of cell height on flat regions (black squares) over time and on tubes (red circles) over time after cutting, n ≥ 81 cells/time point, N ≥ 3. (C) Distribution of cell width on flat regions (black squares) and on tubes (red squares) over time after cutting, n ≥ 81 cells/time point, N ≥ 3. (D) Weighted means and standard error of the weighted mean (SEWM, with variance weights) of cell volume on flat regions (black squares) and on tubes (red circles) over time after cutting, n ≥ 13 images/time points, N ≥ 4. (E) Distribution of cell volume on tubes (red circles) over time after cutting, n ≥ 2,729 cells/time point, N ≥ 4. (F) Distribution of cell volume on flat regions (black circles), n ≥ 3,224 cells/time point, N ≥ 4. (G) Distribution of cell volume on tubes formed after cell fixation in PFA (blue circles) over time after cutting, n ≥ 279 cells/time point, N = 1. Insets in (E)–(G) representative orthogonal views of cells on tubes or flat regions (see diagrams) at different time points after cutting. Scale bars, 20 μm. (H) Pictures of the stretchers in the initial position before compression (top) and stretching (bottom) of about 15% of the elastomer film. 1 small mark in the ruler, 1 mm. (I) Weighted means and SEWM (with variance weights) of cell volume on an elastomer film before compression (black square) or stretching (black rhombus) and after compression (magenta squares) or stretching (green rhombuses) over time, n ≥ 14 images/time point, N ≥ 3. Scale bars, 20 μm. In all images, green represents cell membrane, Myr-Palm-GFP, and magenta represents nuclei, H2B-mCherry. N is the number of independent replicates; the horizontal lines stand for the mean values.

Figure 3

Both signs of curvature produce transient cell shape changes, but this effect is compensated on wavy substrates (A) Schematic of convex curvature generation: PDMS and toluene with (pink) and without (yellow) silicon oil leaflets are inversed compared with Figure 1A. Fibronectin coating (green) and cell monolayer (light pink). (B) Distribution of outer radii of tubes with cells, n = 36 images, N = 2. (C) Distribution of cell volume on tubes with inverted sign of curvature (green circles) over time after cutting and before rolling (black circles), n ≥ 1,179 cells/time point, N = 3. (D) Representative orthogonal views of cells on tubes with inverted sign of curvature at different time points after cutting. Scale bars, 20 μm. (E) Schematic of wavy substrate formation. From the top: infrared laser exposure of the pre-stretched elastomer film; cell growth on the flat surface of the pre-stretched elastomer film; wavy morphologies formation at the surface from the stress relaxation of the substrate. (F) Distribution of radii of the elastomer film in the concave and convex regions, n = 48 (concave) and 43 (convex) images, N = 1. (G) 3D view of a representative cell monolayer on a wavy substrate; 3D reconstruction of segmented nuclei on the concave (blue) and convex (orange) regions. Scale bars, 50 μm. (H) Distribution of cell volume on flat regions before substrate relaxation (black circles) and concave regions of wavy substrates (blue circles) over time after relaxation, n ≥ 602 cells/time point, N = 1. (I) Distribution of cell volume on flat regions before substrate relaxation (black circles) and convex regions of wavy substrates (orange circles) over time after relaxation, n ≥ 1,240 cells/time point, N = 1. (J) Representative orthogonal views of cells on flat and wavy regions at different time points after relaxation. Scale bars, 20 μm. In all images, green represents cell membrane, Myr-Palm-GFP, and magenta represents nuclei, H2B-mCherry. N is the number of independent replicates, the horizontal lines stand for the mean values.

Figure 4

Ion channels regulate the osmotic balance perturbed by the change of curvature (A–F) Weighted means and SEWM (with variance weights) of cell volume and representative orthogonal views on flat regions or tubes at 5 and 30 min after cutting of (A and B) DCPIB-treated cells on flat regions (light pink squares) and on tubes (dark pink circles) over time after cutting, n ≥ 6 images/time point, N = 3; of (C and D) EIPA-treated cells on flat regions (dark yellow squares) and on tubes (light orange circles) over time after cutting, n ≥ 6 images/time point, N = 3; and of (E and F) Bumetanide-treated cells on flat regions (dark blue squares) and on tubes (blue circles) over time after cutting, n ≥ 9 images/time point, N = 3. (G) Distribution of cell volume on flat substrates before (black circles) and over time (minutes) after switching to a hypotonic solution (∼−155 mOsm, brown circles), n ≥ 596 cells/time point, N = 1, the horizontal lines stand for the mean values. (H) Representative images of cells on flat regions before and after hypoosmotic shock (∼−155 mOsm), top (top) and side (bottom) views. (I) Representative images of cells on flat regions before and after hyperosmotic shock (∼+510 mOsm), top (top) and side (bottom) views. (J) Representative images of cells on a tube after hyperosmotic shock (∼+600 mOsm), maximum-intensity z-projection (top) and side (bottom) views. In all images, green represents cell membrane, Myr-Palm-GFP, and magenta represents nuclei, H2B-mCherry. Scale bars, 20 μm. N is the number of independent replicates.

Figure 5

With ion channels, actin cytoskeleton and membrane tension control short and long-time scale response to curvature generation (A) Mean values and SD of fluorescence lifetime (difference values of each time point [T_i] with the final time point [T_F]) on flat regions (0 min) and on tubes over time after cutting, n ≥ 4 images/time point, N = 4. (B) Representative top views along the tubes (see diagram) at different time points after cutting. The color bar corresponds to lifetime in nanoseconds (ns). (C) Mean values and SD of apical to basolateral actin intensity ratio on flat regions (0 min) and on tubes over time after cutting, n ≥ 31 images/time point, N ≥ 3. (D) Representative orthogonal views along the tubes (see diagram) at different time points after cutting. Actin (SiR-Actin, gray). (E–H) Weighted means and SEWM (with variance weights) of cell volume and representative orthogonal views on flat regions or tubes at 5 and 30 min after cutting of (E and F) jasplakinolide-treated cells on flat regions (dark violet squares) and on tubes (light violet circles) over time after cutting, n ≥ 14 images/time point, N = 3; and of (G and H) blebbistatin treatment on flat regions (dark green squares) and on tubes (green circles) over time after cutting, n ≥ 16 images/time point, N = 3. In (F) and (H), green represents cell membrane, Myr-Palm-GFP. Scale bars, 20 μm. N is the number of independent replicates.

Figure 6

Short-time response involves mTORC2 activity and requires membrane tension drop (A–D) Weighted means and SEWM (with variance weights) of cell volume and representative orthogonal views on flat regions or tubes at 5 and 30 min after cutting of (A and B) Torin1-treated cells on flat regions (gray squares) and on tubes (dark gray circles) over time after cutting, n ≥ 10 images/time point, N = 3; and of (C and D) rapamycin-treated cells on flat regions (dark blue squares) over time and on tubes (blue circles) over time after cutting, n ≥ 16 images/time point, N = 3. (E) Representative images of cells fixed and stained for phosphorylated Akt (red) and nuclei (cyan) on flat regions close to the tubes over time and in tubes over time after cutting. (F) Mean values and SD of the cytosolic/nuclear (C/N) fluorescent ratio of the phospho-Akt antibody at different time points after cutting on flat and tubes. n ≥ 8 images/time point, N = 1. (G and H) Mean values and SD of fluorescence lifetime (difference values of each time point [T_i] with the final time point [T_F]) on flat regions (0 min) and on tubes over time after cutting of (G) jasplakinolide-treated cells, n ≥ 3 images/time point, N = 2; and of (H) Torin1-treated cells, n ≥ 3 images/time point, N = 2. (I) Mean values and SD of apical to basolateral actin intensity ratio on flat regions (0 min) and on tubes of Torin1- and rapamycin-treated cells over time after cutting. n ≥ 15 images/time point (Torin1), n ≥ 8 images/time point (rapamycin), N = 2. (J) Weighted means and SEWM (with variance weights) of cell volume after PalmC treatment on flat regions (light violet squares) and on tubes (dark violet circles) over time after cutting. n ≥ 5 images/time point (flat), n ≥ 24 images/time point (tubes), N = 3. (K) Mean values and SD of fluorescence lifetime (difference values of each time point [T_i] with the final time point [T_F]) of PalmC-treated cells on flat regions (0 min) and on tubes over time after cutting. n ≥ 3 images/time point, N = 2. (L) Mean values and SD of apical to basolateral actin intensity ratio on flat regions (0 min) and on tubes of PalmC-treated cells over time after cutting, n ≥ 15 images/time point, N = 2. N is the number of independent replicates. Scale bars, 20 μm.