The nature of cell division forces in epithelial monolayers

Abstract

Epithelial cells undergo striking morphological changes during division to ensure proper segregation of genetic and cytoplasmic materials. These morphological changes occur despite dividing cells being mechanically restricted by neighboring cells, indicating the need for extracellular force generation. Beyond driving cell division itself, forces associated with division have been implicated in tissue-scale processes, including development, tissue growth, migration, and epidermal stratification. While forces generated by mitotic rounding are well understood, forces generated after rounding remain unknown. Here, we identify two distinct stages of division force generation that follow rounding: (1) Protrusive forces along the division axis that drive division elongation, and (2) outward forces that facilitate postdivision spreading. Cytokinetic ring contraction of the dividing cell, but not activity of neighboring cells, generates extracellular forces that propel division elongation and contribute to chromosome segregation. Forces from division elongation are observed in epithelia across many model organisms. Thus, division elongation forces represent a universal mechanism that powers cell division in confining epithelia.

Graphical abstract

Figure 1.

Cell division within epithelial monolayers is associated with three distinct stages of cell–matrix and cell–cell stress: Mitotic rounding, division elongation, and postdivision spreading. (a) Schematic of a generic epithelium in vivo (side view). (b) Timeline of mitotic rounding, division elongation, and postdivision spreading with respect to cell cycle and division progression. (c) Net change in length of parent and daughter cell pair from mitosis start to postdivision spreading completion along the major and minor axes. n = 20; mean ± SD; paired two-sample t test. (d) Experimental setup (side view) of MDCK epithelial monolayer grown on polyacrylamide gel substrate with embedded fluorescent beads. (e) Elastic moduli measurements of polyacrylamide gel. (f) Image of cell division event within MDCK epithelial monolayer (top view). (g) Schematic of division axis (x) and perpendicular axis (y) orientations for the traction force microscopy and monolayer stress microscopy heat maps. (h) Net average change in cell–matrix stress from start of mitosis to postdivision spreading completion. n = 300. (i) Side-view images of the three morphological processes occurring during cell division: Mitotic rounding, division elongation, and postdivision spreading. (j–l) Average change in cell–matrix stress (j) and their distribution (k and l) during the three stages associated with division. Outward (pointing away from the origin) traction stresses were defined to be negative, and inward (pointing toward the origin) traction stresses were defined to be positive. Vertical red lines indicate means; blue line at 0 Pa; edge bins contain exceeding values as well; n = 300; mean ± SD; Tukey’s multiple comparison test. Scale bars, 10 µm. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Figure S1.

MDCK epithelial cell division is accompanied by distinct stages of cell–cell stress. (a–i) Change in cell–cell stress along division axis (a–c), perpendicular axis (d–f), and shear directions (g–i) during division, represented as averaged heat maps (a, d, and g), distributions (b, e, and h), and average-value comparisons (c, f, and i) for entire division process (mitosis start to postdivision spreading completion), mitotic rounding, division elongation, and postdivision spreading. For heat maps, dividing cell is centered on the heat map, with the division axis oriented along the horizontal (x) axis. Mitotic rounding consists of inward and outward force generation stages. (j and k) Average change in cell–matrix stress during the course of mitotic rounding (from mitosis start to metaphase) from mitosis start to 6 min before metaphase, and from 6 min before metaphase to metaphase (j) and corresponding quantification (k). n = 300; ± SD for comparison bar graphs; for histograms, blue line indicates 0 Pa stress value, red line indicates mean; for comparisons, Tukey’s multiple comparison test. Scale bars, 10 µm. *, P < 0.05; ****, P < 0.0001.

Figure S2.

MCF10A epithelial cell division is accompanied by distinct stages of cell–matrix and cell–cell stress. (a–l) Change in cell–matrix stress (a–c) and cell–cell stress during division along division axis (d–f), perpendicular axis (g–i), and shear directions (j–l) represented as averaged heat maps (a, d, g, and j), distributions (b, e, h, and k), and average-value comparisons (c, f, i, and l) for entire division process (mitosis start to postdivision spreading completion), mitotic rounding, division elongation, and postdivision spreading. For heat maps, dividing cell is centered on the heat map, with the division axis oriented along the horizontal (x) axis. Outward (pointing away from the origin) traction stresses were defined to be negative, and inward (pointing toward the origin) traction stresses were defined to be positive. n = 300; ±SD for comparison bar graphs; for histograms, blue line indicates 0 Pa stress value, red line indicates mean; for comparisons, Tukey’s multiple comparison test. Scale bars, 10 µm. ***, P < 0.001; ****, P < 0.0001.

Figure S3.

Increasing epithelial monolayer densities are correlated with reduced measured division forces and increasing heights, and observed cell–matrix stresses are specific for dividing cells undergoing mitotic rounding, division elongation, and postdivision spreading.(a–f) Change in cell–matrix stress during division at varying MDCK monolayer densities represented as averaged heat maps (a, c, and e) and average-value comparisons (b, d, and f), respectively, for mitotic rounding (a and b), division elongation (c and d), and postdivision spreading (e and f). For heat maps, dividing cell is centered on the heat map, with the division axis oriented along the horizontal (x) axis. Outward (pointing away from the origin) traction stresses were defined to be negative, and inward (pointing toward the origin) traction stresses were defined to be positive. n = 140 (low density), n = 300 (medium density), n = 168 (high density); ±SD; one-way ANOVA post-test for trend. ****, P < 0.0001. (g and h) Representative side-view images of MDCK cells stably expressing LifeAct:RFP at varying densities (g) and their height quantification (h). n = 8; Spearman’s rank. **, P < 0.01. The increased cell heights and reduced cross-sectional areas at greater monolayer densities make the plane stress assumption used in MSM less valid, which could explain why measured cell division stresses decrease at increasing monolayer densities (Tambe et al., 2011). (i and j) Average change in cell–matrix stress before mitotic rounding (24 min before mitosis start to mitosis start, n = 277; i) and after postdivision spreading (postdivision spreading completion to 24 min later, n = 300; j). (k–p) Neighboring cells along the division axis or perpendicular axis do not exhibit distinct cell–matrix stresses while the dividing cell undergoes mitotic rounding, division elongation, or postdivision spreading. Average change in cell–matrix stress for neighboring cells along division axis (k–m) and perpendicular axis (n–p), while the dividing cell undergoes mitotic rounding (k and n), division elongation (l and o), and postdivision spreading (m and p). Neighboring cells are centered on the heat maps. n = 78. Scale bars, 10 µm.

Figure 2.

Forces generated during division elongation do not originate from adjacent cells pulling on the dividing cell along the division axis. (a) Side-view schematics displaying potential mechanisms that drive division elongation include adjacent cells pulling, adjacent cells migrating, or the dividing cell pushing outward. (b and c) Schematic showing top view of cells dividing against cell edges (single cells), cell vertices (cell pairs), or a cell edge and a vertex (b), and corresponding frequency distribution (c). n = 212. (d) Fluorescent image of nuclei neighboring a dividing cell, with yellow lines indicating angle between the dividing cell and neighboring cell’s nuclei. (e) Distribution of angles between dividing and neighboring cell nuclei. n = 602 nuclei (from 152 dividing cells); χ² test against difference from uniform distribution. (f) Cartoon depicting structure of full-length and truncated E-cadherin protein (Hart et al., 2017). (g and h) MDCK cells with dominant expression of truncated E-cadherin exhibit reduced levels of extracellular E-cadherin. Representative images of fluorescent stain of extracellular E-cadherin for MDCK WT cells (full-length) and MDCK cells stably expressing truncated E-cadherin (g) and the fluorescent intensity quantification (h). n = 80 (full-length E-cadherin), n = 67 (truncated E-cadherin); mean ± SD; unpaired two-sample t test. (i) Image of mosaic monolayer with MDCK cell expressing E-cadherin with a truncated extracellular domain surrounded by WT MDCK cells. (j and k) Average change in cell–matrix stress during division elongation generated with WT and truncated E-cadherin cells. n = 116; mean ± SD; unpaired two-sample t test. Scale bars, 10 µm. *, P < 0.05; ****, P < 0.0001.

Figure 3.

Forces generated during division elongation do not originate from the movement of adjacent cells. (a) Representative movement of nuclei adjacent to a dividing cell. (b) Average displacement of nuclei neighboring dividing cells. Dividing cell is centered on the heat map, with the division axis oriented along the horizontal (x) axis. n = 152. (c) Quantification of adjacent nuclei movement along the division axis, with positive numbers indicating movement away and negative numbers indicating movement toward the dividing cell. n = 261 adjacent nuclei (from 152 dividing cells); one-sample t test. (d) Reverse-contrast images of vinculin::GFP in MDCK cell at metaphase and cytokinesis completion. Image shows area (50 × 30 µm) used for detection of adhesions in one of the experiments. Outline of dividing cell is shown. (e) The number of detected adhesions at metaphase and cytokinesis. n = 12; mean ± SD; paired two-sample t test. Scale bars, 10 µm. **, P < 0.01.

Figure 4.

Forces generated during division elongation originate from the dividing cell pushing outward against adjacent cells along the division axis. (a–d) Images of adjacent cells (a) and nuclei (c) being deformed during division elongation, with corresponding averaged area strain heat maps (b and d). Dividing cell is centered on the heat map, with the division axis oriented along the horizontal (x) axis. n = 200 dividing cells (b), n = 152 dividing cells (d). (e and f) Quantification of cell (e) and nuclei (f) area strain between adjacent, perpendicular, and neighboring cells. n = 12,371 (far), n = 369 (adjacent), n = 369 (perpendicular; from 200 dividing cells; e); and n = 8,621 (far), n = 258 (adjacent), n = 273 (perpendicular; from 152 dividing cells; f); mean ± SD; Tukey’s multiple comparison test. (g and h) Schematic displaying average change in curvature (κ) of an adjacent cell (g) and quantification (h) of curvatures at metaphase and cytokinesis completion. n = 708 adjacent cells (from 200 dividing cells); mean ± SD; paired two-sample t test. Scale bars, 10 µm. ****, P < 0.0001.

Figure 5.

Stress generated by dividing cell during elongation transmitted to substrate through adhesions of adjacent cells. (a) Mosaic imaging of LifeAct displaying dividing cell in blue and adjacent cell in red at the center and basal planes. Yellow segment indicates distance from center of the dividing cell to dividing-adjacent cell boundary. (b) Average substrate deformations generated during division elongation. n = 300. Dividing cell is centered on the heat map, with the division axis oriented along the horizontal (x) axis. (c) Comparison between distance to dividing cell edge at the basal plane and distance to maximum substrate deformation from averaged substrate deformation heat map. n = 10 dividing cells (dividing cell edge), n = 6 maximum substrate deformation values obtained from averaged substrate deformation heat maps from 300 dividing cells; mean ± SD; scale bars = 10 µm. (d and e) Average change in cell–matrix stress (d) and their distribution (e) during the three stages associated with division, for single cells. Outward (pointing away from the origin) traction stresses were defined to be negative, and inward (pointing toward the origin) traction stresses were defined to be positive. n = 7; mean ± SD; Tukey’s multiple comparison test; scale bars = 10 µm. (f) 3D rendering of adjacent cell surface at metaphase and cytokinesis completion, with red arrow indicating direction of dividing cell. Scale bar, 5 µm. (g) Top view and side view of LifeAct showing adjacent cell deformation during division elongation. Scale bar = 10 µm. (h and i) Deformation heat map of adjacent cell with 0 nN and 500 nN applied force (h), and normalized (against 0 nN) SSIM between experimental and simulated adjacent cell deformation for various force values (i). Scale bar, 5 µm. n = 32 z-slices (from three dividing cells); mean ± SD; Dunnett's multiple comparison test. (j) Schematic displaying outward substrate deformations due to pushing forces exerted by the dividing cell. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Figure 6.

Perturbation of interpolar spindle elongation impacts division elongation in only some cases. (a) Side-view schematics illustrating that forces for division elongation can, in principle, be generated from interpolar spindle elongation and cytokinetic ring contraction. (b and c) Cell elongation (n = 94; b) and change in cell–cell stress along the division axis (Δ σ_xx; n = 300; c) during the early (metaphase to late anaphase) and late (late anaphase to cytokinesis completion) stages of division elongation. Mean ± SD; paired two-sample t test. (d) Quantification of interpolar spindle elongation between metaphase and late anaphase. Scale bar, 10 µm. (e) Lack of correlation between interpolar spindle and cell elongation. n = 25; Pearson’s r. (f) Schematic of inhibition experiments in which inhibitor is added to cells at metaphase. (g–i) Interpolar spindle elongation (n = 53 control, n = 54 experimental; g), cell elongation (n = 61 control, n = 62 experimental; h), and adjacent cell area strain (n = 45 control, n = 44 experimental; i) quantification with and without kinesin-5 inhibition with BRD9876 treatment. Mean ± SD; unpaired two-sample t test. (j and k) Interpolar spindle elongation (n = 46 control, n = 45 experimental; j) and cell elongation (n = 61 control, n = 62 experimental; k) with and without kinesin-5 inhibition with S-trityl-L-cysteine (STLC) treatment. Mean ± SD; unpaired two-sample t test. (l) Laser ablation reveals the interpolar spindle does not consistently bear compression. Representative image of dividing MDCK cell before and after ablation of the interpolar spindle, with magenta line indicating chromosome separation and cyan line indicating cell length. Yellow box indicates ablation region. Scale bar, 5 µm. (m) Quantification of change in distance of chromosome separation (n = 14) and cell length (n = 13) before and after ablation. Negative values indicate chromosome or cell retraction. Mean ± SD; one-sample t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Figure 7.

Forces for division elongation are generated by cytokinetic ring contraction. (a) Change in cell volume before and after cytokinesis. n = 22; mean ± SD; paired two-sample t test. (b and c) Simulation of cytokinetic ring contraction progression (b) and corresponding computational model (c). Scale bar, 5 µm. (d) Predicted cell elongation due to cytokinetic ring contraction for a 5% cell volume reduction. (e) Predicted cell elongation based on varying levels of cell volume reduction during division. (f–h) Cell elongation (n = 49 control, n = 59 experimental; f), adjacent cell area strain (n = 44 control, n = 47 experimental; g), and chromosome separation (n = 47 control, n = 59 experimental; h) with blebbistatin treatment. Mean ± SD; unpaired two-sample t tests. (i) Cell elongation with BTO-1 treatment. n = 20 control, n = 20 experimental; mean ± SD; unpaired two-sample t test. (j) Western blot indicating knockdown of anillin. Anillin and GAPDH were imaged at different intensities. n = 1. (k) Division time course displaying cell treated with siANLN fails to complete cytokinesis. Scale bar, 10 µm. (l) Cell elongation with siANLN or control (siCTRL) treatment. n = 40 control, n = 40 experimental; mean ± SD; unpaired two-sample t test. (m) Representative images of actin for siANLN cells at maximum cytokinetic ring contraction. Scale bar, 10 µm. (n) Correlation between maximum contractile ring contraction and cell elongation for siANLN cells. n = 40; Pearson’s r. Scale bars, 10 µm. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

Figure S4.

The interpolar spindle does not bear compressive forces during division elongation, while varying levels of volume and area conservation during cytokinesis lead to cell elongation. (a) Array of side-view images of MDCK cells before and after ablation, showing that 3D ablation of the interpolar spindle does not result in consistent retraction of the dividing cell. Scale bars, 5 µm. (b) A snapshot depicting coarse-grained triangular mesh consisting of pentagons and hexagons. (c and d) Volume change (c) and area change (d) during the progression of cytokinetic ring contraction corresponding to the case shown in Fig. 7 (b–d). (e and f) Final cell elongation vs final volume change (e), and final area change vs. final volume change (f) for varying degrees of area conservation.

Figure 8.

The forces of division elongation are universal across epithelia in vivo. (a andb) Deformation of adjacent cells in early Drosophila embryos during cell division (a), with quantification of adjacent and perpendicular cell area before and after division (b). Mean ± SD; n = 41 adjacent cells, n = 30 perpendicular cells (from 26 dividing cells); paired two-sample t test. Image in a is of utrophin::GFP, which marks the cell cortex. (c and d) Schematic displaying average change in curvature of an adjacent cell during division (c) and quantification of curvatures before and after division (d). Mean ± SD; n = 41 adjacent cells (from 26 dividing cells); paired two-sample t tests. Scale bars, 10 µm. (e and f) Deformation of adjacent cells quantified with area strain (e) and curvature change (f) for C. elegans embryo E5 (C. el E5; n = 1), embryonic (E15.5) mouse epidermis (Ms E15.5; n = 2), Drosophila embryo segmentation (Dr seg; n = 2), C. elegans embryo E8-16 (C. el E8-16; n = 1), Drosophila embryo after the blastoderm stage (Dr blast; n = 41), adult mouse intestinal organoid (Ms adult; n = 2), embryonic (E3) mouse (Ms E3; n = 3), Xenopus embryo (Xe; n = 2), and zebrafish larva epidermis (Ze; n = 4). (g and h) Change in distance of daughter cell nuclei after division completion within the adult Drosophila intestine (g) and their quantification (h). Mean ± SD; n = 17; one-sample t test. Scale bar, 2.5 µm. ***, P < 0.001; ****, P < 0.0001.

Figure S5.

The forces of division elongation are universal across epithelia in vivo. (a–h) Images of adjacent cell deformation during division within mouse blastocyst (E3; a), adult mouse intestinal organoid (b), embryonic (E15.5) mouse epidermis (c), Drosophila embryo segmentation (d), zebrafish larva epidermis (e), C. elegans embryo E8-16 (f), Xenopus embryo (g), and C. elegans embryo E5 (h). Scale bars, 15 µm (c), 10 µm (a, b, and d–h).