Cell fate coordinates mechano-osmotic forces in intestinal crypt formation

Abstract

Intestinal organoids derived from single cells undergo complex crypt-villus patterning and morphogenesis. However, the nature and coordination of the underlying forces remains poorly characterized. Here, using light-sheet microscopy and large-scale imaging quantification, we demonstrate that crypt formation coincides with a stark reduction in lumen volume. We develop a 3D biophysical model to computationally screen different mechanical scenarios of crypt morphogenesis. Combining this with live-imaging data and multiple mechanical perturbations, we show that actomyosin-driven crypt apical contraction and villus basal tension work synergistically with lumen volume reduction to drive crypt morphogenesis, and demonstrate the existence of a critical point in differential tensions above which crypt morphology becomes robust to volume changes. Finally, we identified a sodium/glucose cotransporter that is specific to differentiated enterocytes that modulates lumen volume reduction through cell swelling in the villus region. Together, our study uncovers the cellular basis of how cell fate modulates osmotic and actomyosin forces to coordinate robust morphogenesis.

Extended Data Fig. 1. Cell and tissue quantification

A, Example of selecting crypt and villus region for quantification of single-cell apical and basal size in bulged organoid. Upper panel, from left to right: Z-projection of ZO-1 and Lyz staining in bulged organoid, selected apical segmentation with Lyz signal in 3D and ZO-1 signal (white) projected in 3D apical side in villus from xy view, in villus from yz view, in crypt from xy view and in crypt from yz view. Lower panel, from left to right: selected basal segmentation with Lyz signal in 3D and ZO-1 signal (white) project in 3D basal side in villus from xy view, outline of villus basal segmentation from xy view, crypt basal segmentation from xy view, outline of crypt basal segmentation from xy view and yz view. Colour of heatmap indicates the size of single-cell apical or basal membranes. B-C’. Tissue compaction along crypt-villus axis in the development of intestinal organoid and in vivo tissue. B. Representative images of bulged and budded organoids with CD44 and DAPI staining. B’. Plot of cell density in crypt (red dashed double arrowhead line) and villus (blue dashed double arrowhead line) tissue as indicated in B. Two-tailed t-test for bulged crypt (n=22) and bulged villus (n=22), p < 10 ^-5; for budded crypt (n=19) and budded villus (n=19), P < 10 ^-11. C. Representative images of immunostaining on the section of mouse intestine at the age of P1 and P11. C’. Plot of cell density in crypt (red dashed double arrowhead line) and villus (blue dashed double arrowhead line) regions as indicated in C. Two-tailed t-test for P1 crypt (n=32) and P1 villus (n=31), p < 10 ^-10; for P11 crypt (n=40) and P11 villus (n=44), P < 10 ^-23. Experiments and imaging analysis in A, and experiments in B-C were repeated at least three times independently with similar results. Scale bars, 20 μm. Violin plot lines in B’ and C’ denote quartile for each group.

Extended Data Fig. 2. Sensitivity analysis for how model parameters affect crypt morphology

A-C. Schematic of the model and morphometric parameters used (see Supplementary Note). D. Phase diagrams of crypt morphologies with varying volumes v and spontaneous curvature of crypt γ_c, for different values of in-plane contraction α (left to right: 1, 0.6, and 1.5). E-H. Evolution of thickness ratio h _c/h _v and radius ratio R _c/R _v during the inflation of an organoid (crypt size φ=0.2, shape factor ${\tilde{\kappa }}_0=10$): increasing α (γ_c = 0.1) (panel E) and γ_c (α = 1, 1.5, and 0.6) (panels F-H). I. Influence of cell swelling on crypt morphology (degree of crypt opening) with varied crypt size φ (α = 1.5, γ_c = −0.02), α (γ_c = −0.1, φ = 0.5), and γ_c (α = 1, φ = 0.5). J. Morphological evolution during the inflation of an organoid (α = 1.5, γ_c = −0.02) with swollen villus cells (v _ev = 5). K. Influence of spontaneous curvature of villus γ_v on crypt morphology (φ=0.2) with varied γ_c (α = 1) and α (γ_c = −0.08). L. Influence of γ_v on the evolution of h _c/h _v and R _c/R _v during the inflation of an organoid with α is respectively 1 and 1.5 (γ_c = −0.1, φ = 0.2).

Extended Data Fig. 3. Comparison between numerical solutions of the full model and analytical scaling laws

A-A” Three possible mechanical scenarios that could drive crypt morphogenesis. Stable organoid configuration is calculated by minimizing the energy F (α,γ_c,γ_v,φ,v,λ), which depends on a few key parameters (see SN for details): α, ratio of in-plane contractions in crypt and villus regions; γ_c, spontaneous curvature crypt region; γ_v, spontaneous curvature of villus region. φ, relative size of the crypt region; v, normalized lumen volume; λ, potential line tension along the crypt/villus boundary. A. crypt budding driven by smaller crypt in-plane contraction α, leading to thinner crypts (decreased epithelial thickness ratio h _c/h _v and decreased radius of curvature R _c/R _v). A’. crypt budding driven by spontaneous curvature α_c, leading to thicker crypts (increased h _c/h _v and decreased R _c/R _v). A”. organoid budding driven by line boundary tension, leading to constant thickness (constant h _c/h_vand decreased R _c/R _v). The width and location of the green lines indicate the strength and distribution of the driving forces in each model. B. Radius ratio (R _c/R _v) from six experimental samples with the corresponding model fit (see Supplementary Note). C-D. All six samples of bulged organoids can be collapsed via h _c/h _v vs. time (C) or R _c/R _v vs. time (D). E-F. Comparison of numerical (symbol) and analytic (line) results to verify that, thickness ratio h _c/h _v and radius ratio R _c/R _v respectively depends on in-plane contraction α (with crypt size φ=0.05) and coupled parameter φ⁻¹ γ_c (with α = 1.15) for a bulged organoid (normalized volume v=5) (panel A), h _c/h _v and R _c/R _v depend on parameter u for a budded organoid (φ=0.2) (panel B), and analytic results also fit well with numerical results for the inflation of a bulged organoid (α = 1.15, γ_c = −0.025, φ = 0.05) or a budded organoid (u=6, φ=0.2). G. Phase diagram of crypt morphologies of an organoid (φ=0.2, ${\tilde{\kappa }}_0=10$) under infinite volume expansion (v=10⁸) (see Supplementary Note), with varying spontaneous curvature γ_c and in-plane contraction α. H. Influence of normalized volume of a crypt cell v _ec (top) and that of a villus cell v _ev (bottom) on h _c/h _v and R _c/R _v of a budded organoid (u=6, φ=0.2).

Extended Data Fig. 4. Actomyosin drives and maintains crypt budding

A-A’. Supplementary to Fig. 2D. A. Montage images across the regions of laser cutting and opening with surrounded LifeAct-GFP signal. Dashed lines outline the size of opening. A’. Plot for distance of openings after cutting. B. Staining of phosphorylated Myosin light chain (pMLC) in budded organoid with maximum z-projection (upper panel) and single section (lower panel). C. Day4 LifeAct-GFP organoids in Control (n=96), Blebbistatin (n=188) or Cytocalasin D (n=45) cultures. Left images, maximum z-projection of LifeAct-GFP; right images, merged maximum z-projection of LifeAct-GFP with Lyz and DAPI staining. Plots, quantification of organoid eccentricity and volume ratio (Lumen vs total). D. Representative time-lapse recording of organoid expressing Myh-9-GFP during crypt morphogenesis. Upper images, middle sections. Lower images, maximum z-projections. E. Membrane-targeted GFP (mG) in organoid before bulging (n=12), bulged (n=20) and budded (n=22) organoids in the middle section (images); quantification for ratios of mG intensity in cells (box plots). F. Representative time-lapse recording of budded LifeAct-GFP (white) organoid treated with DMSO control, Blebbistatin or Aphidicolin (left images), and corresponding plot for aspect ratio of the organoids (right plot, numbers of recordings in bracket). Solid lines represent average values, shadow regions standard deviations. G-G’, Merged images for Myh-9-GFP and ZO-1 staining in wild-type/Myh-9-GFP (G, left image) or in myh-9^+/-/Myh-9-GFP (G’, left image) mosaic organoid with zoom-in areas (G and G’, right image) from yellow dashed rectangles. White dashed line outlines the morphology of Myh-9-GFP cells. G”, quantification on percentage of basal constriction in Myh-9-GFP cells next to wild-type (n=62) or myh-9^+/- (n=79) cells Arrows: red (crypt), blue (villus), white (G and G’, basal domain of Myh-9-GFP cells). P-values are calculated from two-tailed t-test. Box-plot elements show 25% (Q1, upper bounds), 50% (median, black lines within the boxes) and 75% (Q3, lower bounds) quartiles, and whiskers denote 1.5× the interquartile range (maxima: Q3 + 1.5× (Q3-Q1); minima: Q1-1.5× (Q3-Q1)) with outliers (rhombuses). Scale bars: B (20μm), C-D, F, G-G’ left (50μm), G -G’ right zoom-in (10μm). Images in A-G are representative of three independent experiments. Data in G” are pulled from two independent experiments.

Extended Data Fig. 5. In vivo tissue-specific localization of Claudin2 and ZO-1

In vivo staining of Claudin2 (A) and ZO-1 (B) in mouse small intestine at P1, P2, P5, P7, P11, P12, P13, P14, P15, P16, P17 and 6-month adult stages. Stainings were repeated at least three times independently with similar results. Yellow dashed rectangles in B indicate zoom-in regions of villi randomly selected from the same field of views in the presented images. Scale bars: A (100 μm), B (100 μm, zoom-in, 20 μm)

Extended Data Fig. 6. Sensitivity analysis of model results as a function of crypt apical tension and osmotic changes

A. Phase diagram of crypt morphologies (${\tilde{\kappa }}_0=6$, φ=0.2) as a function of normalized crypt apical tension m and villus cell volume v _ev, showing the villus cell swelling favors crypt budding synergistically with crypt apical tension. B. Influence of crypt size φ and shape factor${\tilde{\kappa }}_0$ on critical values of m (with v _ev=1, i.e. no cell swelling) and v _ev (with m=2, i.e. constant crypt apical tension) for organoid budding. C-E. Influence of preferential proliferation of crypt cells, characterized as crypt size φ_g, on crypt morphology, with normalized organoid volume v=1 (panel C) or 1.2 (panel E). After rescaling volume v, crypt morphologies with preferential crypt growth (solid line) are close to those without preferential growth (dash line) (panel D, see SN for further details). F-H. Dependence of thickness ratio h _c/h _v and radius ratio R _c/R _v on normalized crypt apical tension m (v _ev=1), with varied ${\tilde{\kappa }}_0$ (φ=0.2) (panel F) and φ (${\tilde{\kappa }}_0=6$) (panel G), and on villus cell volume v _ev (panel H), with varied m (${\tilde{\kappa }}_0=6$, φ=0.2). I. Evolution of h _c/h _v and R _c/R _v during organoid inflation with varied m (${\tilde{\kappa }}_0=6$, φ=0.2, v _ev=1). J. Phase diagram of crypt morphologies upon infinite volume expansion (v=10⁸) (showing three possible phases: fully closed, partially opening, or fully closed with vanishing apical surface), with varying ${\tilde{\kappa }}_0$ and m (φ=0.2).

Extended Data Fig. 7. The impact of lumen volume on organoid morphology

A-B”. Fitting for lumen inflation experiments with theoretical model. Images, microinjection of bulged (A) and budded (B) organoids for lumen inflation. Fittings (A’ and B’), fitting experimental measurements with its predictive model based on epithelial thickness ratio (h _c/h _v), organoid radius ratio (R _c/R _v) and lumen volume (v) change. Plots (A” and B”), measurements of h _c/h _v and R _c/R _v in bulged (A”, n=7 biologically independent organoids) and budded (B”, n=12 biologically independent organoids) organoids. P-values are calculated from two-tailed Paired Student’s t-test. C-C’. Evolution of morphometric parameters in bulged (C) or budded (C’) organoids during volume inflation, induced by PGE treatment (n=3 biologically independent organoids) and microinjection (n=3 biologically independent organoids), can be collapsed via R _c/R _v vs. v and h _c/h _v vs. v. Scale bars, 50 μm.

Extended Data Fig. 8. Regulation of lumen volume by enterocytes and the membrane transporters

A. Representative images of DMSO-treated control (n=96), CHIR-(n=145) and IWP2-(n=78) treated organoids (left panels), and corresponding quantification on eccentricity and lumen ratio (right panels). B. Segmentation (Seg.) of single-cell volume based on β-Catenin signal in bulged and budded organoids (left images), violin plot for the corresponding quantification of single-cell volume (left plot), swarm plot for the lumen ratio of each organoids (right plot). Two-tailed t-test for cells in crypt (n=385) vs. villus (n=218) in bulged organoid (p = 0.311), crypt (n= 551) vs. villus (n=769) in budded organoids (p < 10^-71), and bulged villus vs. budded villus (p < 10^-75); for lumen ratio bulged (n=8) vs. budded (n=6) (p < 10^-4). C. Immunostaining of β-Catenin in the section of mouse intestinal tissue at P1 and P11 stages, segmentation (seg.) of single-cell area, zoom-in crypt (red dashed rectangles) and villus (blue dashed rectangles) areas in β-Catenin staining image overlapping with single-cell segmentations (left images), violin plot for the quantification of single-cell areas (left plot), swarm plot for the quantification of distance between villi (right plot). Two-tailed t-test for cells in P1 crypts (n=54) vs. villi (n=156) (p < 10^-14), in P11 crypts (n=82) vs. villi (n=159) (p < 10^-14), and cells in P1 villi vs. P11 villi (p < 10^-15); for distance between villi, P1 (n=32) vs. P11 (n=30) (p < 10^-7). D. TSNE-based visualizations of single-cell RNA sequencing data indicate mRNA expressions of aquaporins, atp1a1 and atp1b1. E. Day4 organoid treated with DMSO control, Ouabain and CuSO4 (left images), quantification on the eccentricity and lumen ratio of control (n=155), Ouabain (n=146), CuSO4 (n=192) and Sotagliflozin (n=119) (right plots). P-values in box plots (A, E) are calculated from two-tailed t-test. Box plot elements show 25% (Q1, upper bounds), 50% (median, black lines within the boxes) and 75% (Q3, lower bounds) quartiles, and whiskers denote 1.5× the interquartile range (maxima: Q3 + 1.5× (Q3-Q1); minima: Q1-1.5× (Q3-Q1)) with outliers (rhombuses). Violin plot lines (B, C) denote quartile for each group. Scale bars: A-B (50 μm), C (left, 50 μm, zoom-in,10 μm), E (20 μm). Images in A, B, C and E are representative of > 3 independent experiments.

Extended Data Fig. 9. Piezo channels do not strongly regulate crypt budding

A. TSNE-based visualizations of single-cell RNA sequencing data indicate the maker gene expression for crypt (left panel), villus (middle panel) and piezo1 (right panel). B. Piezo1 and DAPI staining in budded organoid. Piezo1 is detected in few single cells. Staining was repeated at least three times independently with similar results. C-E. Inhibition of Piezo1 (GdCl3 and GsMtx4) did not cause any defect in crypt morphogenesis, while activation of Piezo1 (Yoda-1) leads to slight increased lumen volume and reduced eccentricity. Reduced enterocytes (indicated by AdlB signal) are detected in Yoda-1-treated organoids. C. Representative images of Day 4 organoids treated in DMSO-control, GdCl3, GsMtx4 or Yoda-1 condition with AdlB, CD44 and DAPI staining in maximum z-projection. D-E. Corresponding box-plot quantifications of eccentricity (D) and lumen ratio (E) for organoids from C. Experiment (C-E) was repeated three times independently with similar results. Sample numbers: DMSO-control (n = 94 organoids), GdCl3 (n = 143 organoids), GsMtx4 (n = 68 organoids) and Yoda-1 (n = 71 organoids). P-values are calculated from two-tailed t-test. Box-plot elements show 25% (Q1, upper bounds), 50% (median, black lines within the boxes) and 75% (Q3, lower bounds) quartiles, and whiskers denote 1.5× the interquartile range (maxima: Q3 + 1.5× (Q3-Q1); minima: Q1-1.5× (Q3-Q1)) with outliers (rhombuses). Scale bars, 25 μm.

Extended Data Fig. 10. ECM remodelling is nonessential for crypt budding

A. Representative images of Day 4 organoids treated by DMSO as control, or broad-spectrum inhibitors of matrix metalloproteinases (GM6001 and Marimastat) with Laminin staining for Matrigel and DAPI staining for cell nuclei in maximum z-projection. Experiment was repeated > 3 times independently with similar results. B. Corresponding box-plot quantifications of eccentricity for organoids from A. Experiments were repeated three times independently with similar results. Sample numbers: DMSO-control (n = 265 organoids), GM6001 (n = 102 organoids), and Marimastat (n = 94 organoids). P-values are calculated from two-tailed t-test. Box-plot elements show 25% (Q1, upper bounds), 50% (median, black lines within the boxes) and 75% (Q3, lower bounds) quartiles, and whiskers denote 1.5× the interquartile range (maxima: Q3 + 1.5× (Q3-Q1); minima: Q1-1.5× (Q3-Q1)). Scale bars, 50 μm.

Figure 1. Crypt morphogenesis during intestinal organoid development.

A. Cartoon representation of crypt morphogenesis and workflow of image features extraction. B. Segmentation of single-cell apical and basal domains. Representative time-course imaging of ZO-1 staining in Day 3 organoid before bulging (left), Day 3.5 bulged and Day 4 budded organoids in top panels. Segmentation of single-cell apical (middle panels) and basal (bottom panels) domains corresponding to the top panels, Day 3.5 bulged organoid has xy and yz views for apical segmentation. Colour of heatmap indicates size of the domains. C. Violin plot of the size of single-cell apical and basal domains in B (sample number for apical domains, Day 3: 331 cells, Day3.5 crypt: 307 cells, Day 3.5 villus: 739 cells, Day 4 crypt: 550 cells, Day 4 villus: 1005 cells; sample number for basal domains, Day 3: 531 cells, Day3.5 crypt: cells, Day 3.5 villus: 1249 cells, Day 4 crypt: cells, Day4 villus: 1500 cells), and dot plot of the average of the domain size in organoids at Day 3 (n=9 organoids), Day 3.5 (n=7 organoids) and Day 4 (n=6 organoids). P-values in violin plot are calculated from two-tailed t-test, in dot plot are calculated from two-tailed Paired Student’s t-test (see Statistics and reproducibility). Plot lines denote quartile for each group. D. Representative time-lapse imaging of crypt bulging and budding. E. Plot for eccentricity, lumen volume, tissue volume and total volume of organoids in recording in D (n=4 independent recordings). Solid lines represent average values, shadow regions represent standard deviations. F. Quantification of the lumen ratio (Lumen/total), tissue ratio (Tissue/total) and eccentricity before bulging (00:00 hr) and after budding (16:30 hr). P-values are calculated from two-tailed Paired Student’s t-test (n=12 organoids). Red arrows in B and D indicate crypt regions. Scale bars in B and D, 20 μm. Images in B and D are representative of three independent experiments.

Figure 2. Region-specific spontaneous curvature drives crypt morphogenesis.

A. Schematic of the morphometric parameters for 3D organoids. R _c, radius of curvature in crypt; R_v,radius of curvature in villus; h, height of cell along apical-basal axis represents tissue thickness, h _c, crypt thickness, h_v, villus thickness (see SN for details). B. Three possible mechanical scenarios that could drive crypt morphogenesis as a function of time (indicated by arrow). Green lines indicate the driving forces. C-C”. Experimental data validate model ii). C, time-lapse recording of representative organoid during bulging. Experiment was repeated more than 3 times independently with similar results. C’, thickness ratio (h _c/h _v) from six experimental samples with the model fit. C”, different samples collapse on the same theoretical master curve independent of time (R _c/R _v vs. a function of h _c/h _v). D. Test of apical tensions in bulged and budded organoids. Left panel, images of organoids visualized by LifeAct-GFP (white) in single sections before laser cutting, zoom-in areas from the coloured regions before (0.0s) and after (1.6s) laser cutting. Yellow dashed lines indicate the position of cutting on apical domains. Right panel, quantification of the initial recoil velocities. Sample numbers: bulged crypt (n=10 independent cuttings), budded crypt (n=17 independent cuttings), bulged crypt (n=6 independent cuttings), budded villus (n=8 independent cuttings). E. Test of basal tension in bulged and budded organoids. Left panel, imaging of micropipette aspiration on the basal side with zoom-in regions indicating in the white dashed rectangles. Right panel, quantification of basal tensions. Sample numbers: crypt (n=17 organoids), villus (n=8 organoids). P-values (D and E) are calculated from two-tailed t-test. Box-plot elements show 25% (Q1, upper bounds), 50% (median, black lines within the boxes) and 75% (Q3, lower bounds) quartiles, and whiskers denote 1.5× the interquartile range (maxima: Q3 + 1.5× (Q3-Q1); minima: Q1-1.5× (Q3-Q1)) with outliers (rhombuses). Scale bars: C (25 μm), D (organoid, 25 μm; zoom-in, 5 μm), E, (organoid, 50 μm; zoom-in, 10 μm). Images in D and E are representative of three independent experiments.

Figure 3. Myosin patterns in stem cells and enterocytes determine region-specific spontaneous curvature.

A. Myh-9-GFP in organoid before bulging, bulged, and budded organoids in the middle section, maximum z-projection in budded organoid, and zoom-in areas from dashed boxes. B. Plots for ratios of Myh-9-GFP intensity in cells. Abbreviations: ca, crypt apical; cb, crypt basal; va, villus apical; vb, villus basal. Arrows: villus (blue), crypt (red). C-F. Stem cell fate is responsible for apically enriched Myh-9 in crypt. C-D. Representative images for organoids treated by CHIR and VPA and enriched with stem cells (SC-enriched), treated by CHIR and DAPT and enriched with Paneth cells (PC-enriched) or enriched with enterocyte (enterocyst). C. Maximum z-projection of Lgr-5-DTR-GFP organoids with Lyz and DAPI staining (upper panels) and wild-type organoids with CD44, AdlB and DAPI staining (lower panels). D. Maximum z-projection (upper panels) and single-layer (lower panels) images of Myh-9-GFP organoids with Lyz and DAPI staining. E. Quantification of eccentricity from Maximum z-projection images in D. F. Quantification of Myh-9-GFP intensity from single-layer images in D. For box plots B, E, F n numbers are stated in brackets and represent organoids selected for multiple single-cell (B) or whole organoid (D) measurements. P-values (B, E and F) are determined by two-tailed t-test. Box-plot elements show 25% (Q1, upper bounds), 50% (median, black lines within the boxes) and 75% (Q3, lower bounds) quartiles, and whiskers denote 1.5× the interquartile range (maxima: Q3 + 1.5× (Q3-Q1); minima: Q1-1.5× (Q3-Q1)) with outliers (rhombuses). Experiments in A and C-E were repeated >3 times independently with similar results. Scale bars: A (organoid, 50μm; zoom-in, 5 μm), C and D (50 um).

Figure 4. Region-specific localization of cell junctions.

A-B. Single-cell RNA sequencing analysis of budded organoid. TSNE-based visualization of single-cell degree of expression of marker genes for cell types (stem cells/Crypt columnar cells, Paneth cells and enterocytes) (A), and cell junctions (Claudin2, Occludin and E-Cadherin) (B). C. ZO-1 staining in budded organoid in maximum z-projection (left panel), and heatmap visualization of basal signal (right panel). D. ZO-1 staining in 6-month mouse intestine in section along crypt/villus axis, and perpendicular sections of villus and crypt regions along the positions indicated by black dashed rectangles. E. Middle single-layer section and maximum z-projection of the Claudin2, Occludin and N-Cadherin staining with DAPI staining for cell nuclei (blue). F. Co-localization of ZO-1 and Myh-9-GFP in cell basolateral domains in villus region. From left to right: Z-projection (upper panels) and zoom-in region (lower panels) of ZO-1, Myh-9-GFP, and merged ZO-1 and Myh-9-GFP. G. Single-layer section and maximus z-projection of ZO-1 (merged with Lyz) and Claudin2 (merged with DAPI) co-staining in organoid before bulging, bugled and budded organoids. Arrows: red arrows indicate crypt regions; blue arrows indicate the villus regions. Experiments in C-G were repeated more than 3 times independently with similar results. Scale bars, 50 μm.

Figure 5. Lumen volume reduction promotes crypt budding.

A. Theoretical prediction (${\tilde{\kappa }}_0=6$, φ=0.2) for the relative contribution of the crypt apical tension (m) vs. lumen volume (v) for organoid shape (left panel), and schematic representing the experimental designs for phase diagram validation (right panel). The horizontal dashed line at m = 4.2 indicates the critical value, above which the budded shapes cannot be reversed by increasing lumen volume. B-C”. Fitting for the lumen inflation experiments with the theoretical model. Images show organoid morphology before and after lumen inflation by PGE treatment of bulged (B) and budded (C) organoids with tissue and lumen visualized by LifeAct-GFP (white). Scale bars, 20 μm. Experimental measurements (black dots, data) from the representative samples (epithelial thickness ratio h _c/h _v and radius ratio R _c/R _{v ,} B’ for B, bulged organoid, C’ for C, budded organoid) as a function of lumen volume (v) change were fitted from the model (black lines, see SN for details and best-fit values). Budded crypts systematically remained near-unchanged by inflation (decreased radius ratio R _c/R _v) (B’), while bulged crypts opened up (increased radius ratio R _c/R _v) (C’). Plots, measurements of epithelial thickness ratio and radius ratio in bulged (B”, n=9 biologically independent samples) and budded (C”, n=10 biologically independent samples) organoids. P-values are calculated from two-tailed Paired Student’s t-test. D-E. All samples from PGE and pipette inflation (bulged: D, budded: E) can be collapsed via fitting of two parameters (see Supplementary Note) on the same theoretical master curve (R _c/R_v vs. a function of h _c/h _v) independent of volume as predicted by the model (black line), showing opposite trends in bulged (D) vs. budded (E). Images in B and C are representative of three independent experiments.

Figure 6. Coordination of lumen volume reduction and region-specific spontaneous curvature promotes crypt morphogenesis.

A. Actomyosin maintains crypt morphology in inflated budded organoid. Images, time-lapse recording of budded LifeAct-GFP organoid treated with PGE to inflate lumen (upper panel) or PGE (whole time) and Blebbistatin at 3:20 hrs (lower panel). Plot, corresponding plot for radius ratio (R _c/R _v) of the organoids. Lines in the middle represent the average values, shadow regions represent the standard deviations. B. Lgr5-DTR-GFP in inflated organoid with or without spontaneous curvature. Images, representative Lgr5-DTR-GFP organoids cultured in DMSO control, PGE or PGE + Blebbistatin (from 3:20 hr) condition with DAPI staining. Box plot, quantification of Lgr-5-DTR-GFP intensity in three groups from left images. C. Osmotic deflation in organoids with different spontaneous curvatures. Cartoon and images from upper to lower panels: Day2.5 DMSO-treated organoid, Day3 organoid before bulging, Day3.5 bulged organoid, and Day3.5 CHIR-treated organoid without bulging. Colours in the cartoon: red (Paneth cell), yellow (stem cell), blue (cell in villus tissue). Organoid images, fluorescent images of organoid expressing Lgr-5-DTR-GFP and H2B-iRFP before and after osmotic shock. Plot, quantification on eccentricity of organoids before and after osmotic shock. Red arrows in D and F indicate crypt regions. For plots (A-C), n numbers are stated in brackets and represent the number of independent recordings (A) or independent organoids (B and C) selected for measurements. P-values are calculated from two-tailed t-test (B) or two-tailed Paired Student’s t-test (C). Box-plot elements show 25% (Q1, upper bounds), 50% (median, black lines within the boxes) and 75% (Q3, lower bounds) quartiles, and whiskers denote 1.5× the interquartile range (maxima: Q3 + 1.5× (Q3-Q1); minima: Q1-1.5× (Q3-Q1)) with outliers (rhombuses). Experiments in A-C were repeated at least 3 times independently with similar results. Scale bars, 50 μm.

Figure 7. Enterocytes control lumen volume reduction through SGLT-1.

A. Schematic representation of the experimental design applied. B. Light-sheet time-lapse recording of enterocyst, the organoid that is composed of only enterocyte, from Day 3 with tissue and lumen visualized by LifeAct-GFP (white). C. Light-sheet time-lapse recording of organoid treated with 3 μM CHIR, the organoid that is highly enriched with stem cells and Paneth cells, from Day 3 with tissue and lumen visualized by LifeAct-GFP (white). D. Corresponding plots of B on eccentricity, lumen volume, tissue volume and total volume (lumen + tissue) (n=5 independent recordings). E. Corresponding plots of C on eccentricity, lumen volume, tissue volume and total volume (n=3 independent recordings). In D and E, solid lines represent average values, shadow regions represent the standard deviations. F. Model prediction of how increased volume of different regions promotes budding, showing that fluid uptake specifically by villus cells is best for budding. G. Model prediction of increased cell volume in villus region with and without corresponding lumen volume changes, showing that cell volume changes alone can still promote budding (full line). H. TSNE-based visualizations of single-cell RNA sequencing data indicate the expression of sglt. I. Immunostaining of SGLT-1 (white) in budded organoid with crypt regions visualized by Lgr-5-DTR-GFP (green) indicates the enrichment of SGLT-1 in the villus apical domain. J-J’. SGLT-1 is required for lumen volume reduction. J. Representative images of Day 4 organoid from DMSO-treated control and Sotagliflozin-treated sample for SGLT-1 inhibition (from 3 independent experiments). J’, quantification of the eccentricity (left plot) and lumen ratio (lumen volume vs. total volume, right plot) of control (n=155 organoids) and Sotagliflozin treated organoids (n=119 organoids). P-values are calculated from two-tailed t-test. Box-plot elements show 25% (Q1, upper bounds), 50% (median, black lines within the boxes) and 75% (Q3, lower bounds) quartiles, and whiskers denote 1.5× the interquartile range (maxima: Q3 + 1.5× (Q3-Q1); minima: Q1-1.5× (Q3-Q1)) with outliers (rhombuses). Time-lapse recordings in B-C and experiments in I-J were repeated at least 3 times independently with similar results. Scale bars: B, C and J (20 μm), I (50 μm).