Cell tumbling enhances stem cell differentiation in hydrogels via nuclear mechanotransduction

Abstract

Cells can deform their local niche in three dimensions via whole-cell movements such as spreading, migration or volume expansion. These behaviours, occurring over hours to days, influence long-term cell fates including differentiation. Here we report a whole-cell movement that occurs in sliding hydrogels at the minutes timescale, termed cell tumbling, characterized by three-dimensional cell dynamics and hydrogel deformation elicited by heightened seconds-to-minutes-scale cytoskeletal and nuclear activity. Studies inhibiting or promoting the cell tumbling of mesenchymal stem cells show that this behaviour enhances differentiation into chondrocytes. Further, it is associated with a decrease in global chromatin accessibility, which is required for enhanced differentiation. Cell tumbling also occurs during differentiation into other lineages and its differentiation-enhancing effects are validated in various hydrogel platforms. Our results establish that cell tumbling is an additional regulator of stem cell differentiation, mediated by rapid niche deformation and nuclear mechanotransduction.

Extended Data Fig. 1 |. Quantification of cell tumbling and nuclear movements upon cytoskeleton inhibition and LINC complex inhibition.

(a) (Left) Centroid speed and (Right) Major axis angular speed upon cytoskeletal inhibition. N ≥ 15 across 3 gels per condition (b) Nuclear movement speed upon cytoskeleton inhibition. N ≥ 20 across 3 gels per condition. (c) (Left) Centroid speed and (Right) Major axis angular speed upon LINC complex inhibition. N ≥ 15 across 3 gels per condition. (d) Nuclear movement speed (5-minute intervals for a total period of 16 hours) upon LINC complex inhibition. N ≥ 20 across 3 gels per condition. ns, not significant; *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001. P value, one–way ANOVA with Tukey’s multiple comparisons test.

Extended Data Fig. 2 |. Characterization of focal adhesions and the role of adhesive ligands in cell tumbling.

(a) Representative immunofluorescence images for focal adhesion markers in SG (top) and CG (bottom) after 24 hours of chondrogenic induction and Fibronectin after 12 hours of chondrogenic induction as observed from 3 independent experiments. Marker of interest-green, F-Actin-red, and DAPI-blue. Scale- 10µm. (b) Centroid speed and (c) Major axis angular speed upon perturbation of adhesive ligands. N ≥ 15 across 3 gels per condition (d) Nuclear movement speed upon perturbation of adhesive ligands. N ≥ 20 across 3 gels per condition. ns, not significant; *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001. P value, one–way ANOVA with Tukey’s multiple comparisons test.

Extended Data Fig. 3 |. Characterization of Ezrin in cell tumbling.

Note- NSC is the Ezrin inhibitor NSC668394. (a) Representative western blotting images and (b) Quantifications of western blotting images after 24 hours of chondrogenic induction. Protein expression for each marker was normalized to GAPDH. N = 3 biological replicates per condition. Data reported represent mean value ± s.d. (c) Representative immunofluorescence images for Ezrin (green), F-Actin (red) and DAPI (blue) in SG (top), SG with ezrin inhibitor (middle), and CG (bottom) after 24 hours of chondrogenic induction. Scale- 10µm. (d) Quantification of interquartile range of Ezrin intensity at the cell cortex highlighting ezrin expression heterogeneity. N ≥ 15 across 3 gels per condition (e) Cortical Actin, (f) Centroid speed, (g) Major axis angular speed, and (h) nuclear movement speed upon Ezrin inhibition. N ≥ 15 across 3 gels per condition ns, not significant; *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001. P value, one–way ANOVA with Tukey’s multiple comparisons test.

Extended Data Fig. 4 |. Role of cytoskeleton in early-stage tumbling-enhanced chondrogenesis.

(a–c) Gene expression of chondrogenic markers after 3 days of chondrogenic induction with cytoskeletal inhibition. N = 3 gels per condition. (a) SOX9 (b) ACAN (c) COL2A1 (d) (Top) Treatment regimen of different cytoskeleton inhibitors for the first 3 days in the 21-day chondrogenic induction period. (Bottom) Representative Safranin-O histology on cryo-sectioned samples for sGAG deposition after 21 days of chondrogenic induction with different cytoskeleton inhibitors as observed from 3 independent experiments. Scale- 200 µm (shared between all images). ns, not significant; *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001. P value, one–way ANOVA with Tukey’s multiple comparisons test. Data are presented as mean ± SD for Extended Data Fig. 4a–c.

Extended Data Fig. 5 |. Role of LINC complex in early-stage tumbling-enhanced chondrogenesis.

(a–c) Gene expression of chondrogenic markers after 3 days of chondrogenic induction with LINC complex inhibition. N = 3 gels per condition. (a) SOX9 (b) ACAN (c) COL2A1 (d) (Top) Treatment regimen of LINC complex inhibition for the first 3 days in the 21-day chondrogenic induction period. (Bottom) Representative Safranin-O histology on cryo-sectioned samples for sGAG deposition after 21 days of chondrogenic induction with LINC complex inhibition as observed from 2 independent experiments. Scale- 200 µm (shared between all images). ns, not significant; *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001. P value, one–way ANOVA with Tukey’s multiple comparisons test. Data are presented as mean ± SD for Extended Data Fig. 5a–c.

Extended Data Fig. 6 |. Role of adhesive ligands in early-stage tumbling-enhanced chondrogenesis.

(a–c) Gene expression of chondrogenic markers after 3 days of chondrogenic induction with adhesive ligand perturbations. N = 3 gels per condition. (a) SOX9 (b) ACAN (c) COL2A1 (d) (Top) Treatment regimen of different adhesive ligand perturbations for the first 3 days in the 21-day chondrogenic induction period. (Bottom) Representative Safranin-O histology on cryo-sectioned samples for sGAG deposition after 21 days of chondrogenic induction with different adhesive ligand perturbations as observed from 3 independent experiments. Scale- 200 µm (shared between all images). ns, not significant; *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001. P value, one–way ANOVA with Tukey’s multiple comparisons test. Data are presented as mean ± SD for Extended Data Fig. 6a–c.

Fig. 1 |. Enhanced MSC chondrogenesis is associated with minutes-scale cell tumbling in three dimensions.

a, Schematic illustrating SG versus CG, with mobile or static crosslinks and ligands. Blue, PEG backbone; purple, αCD rings; green, RGD. b, Representative Safranin O staining for sGAG deposition by MSCs after 21 days of chondrogenic induction in SG (top) and CG (bottom). Scale bar, 200 µm. c, Representative storage and loss moduli of SG and CG as obtained by dynamic-light-scattering-based microrheology (left; N = 3 gels per condition) and bulk parallel-plate shear rheology (right). d–h, Characterization of cell tumbling. d, Time-lapse image sequence of a single live cell in SG (top) and CG (bottom). Scale bar, 10 µm. e, Time-lapse image sequence of the labelled hydrogel for the corresponding cells in d for SG (top) and CG (bottom). Scale bar, 10 µm. f, Superimposed colour-coded outlines of a single cell over time (5-min intervals for a total period of 16 h) for SG (top) and CG (bottom). g, Correlation coefficient of cell shape over time comparing the cell shape at time t > 0 with shape at time t = 0. N = 10 cells across three gels per condition. h, Centroid speed (left) and angular speed (right) of the major axis of cells. N ≥ 15 across three gels per condition. i, Force–strain curves obtained from the AFM point indentation on SG, CG and SG50:CG50 (left). The strain with the same amount of force in SG, CG and SG50:CG50 derived from the force–strain curve (right). N = 4 gels per condition. j, Cell tumbling in SG50:CG50 (left) centroid speed and major-axis angular speed (right). N = 15 across three gels per condition. *P < 0.05, **P < 0.005, ***P < 0.001, ****P < 0.0001. The P value is obtained using an unpaired two-tailed t-test in h, two-way analysis of variance (ANOVA) in i and one-way ANOVA in j. Data are presented as mean ± s.d. for c, g and i. For i, the boxes represent the interquartile range, and the centre of the boxes represents the median. The whiskers represent the maximum and minimum values.

Fig. 2 |. Cell tumbling is associated with enhanced seconds-to-minutes-scale cytoskeletal and nuclear dynamics in MSCs.

a, Left: representative heat map of matrix deformation over 15 min as determined by the tracking of embedded bead displacements in SG (top) and CG (bottom). The cell locations before (dashed line) and after (solid line) are marked in black for SG and white for CG. Right: quantification of the maximum matrix displacement over 15 min. N ≥ 15 cells across three gels per condition. b, Left: representative image of F-actin in cells in SG (top) and CG (bottom). Scale bar, 10 µm. Middle: superimposed colour-coded cortical F-actin indicating the presence of F-actin at a given location in the cortex during 5 min of live imaging. Right: quantification of live-cell cortical actin (5-s intervals for a total period of 5 min). N ≥ 10 cells across three gels per condition. c, Representative change in F-actin protrusions highlighting dynamics over 30 s in SG (top) and CG (bottom) as observed from three independent experiments. Scale bar, 5 µm. d, Schematic depicting the calculation of F-actin dynamic area (area gained + area lost) between two consecutive time frames. e, Representative example of magnitude and fluctuations in the dynamic area for one cell in SG and one in CG (5-s intervals for a total period of 5 min). f, Quantification of F-actin dynamics as the average dynamic area (5-s intervals for a total period of 5 min). N ≥ 10 cells across three gels per condition. g, Time-lapse image sequence of a single nucleus in SG (top) and CG (bottom). Scale bar, 7 µm. h, Representative movement tracks for nuclei in SG and CG. N = 15 cells across three gels per condition. Grid size, 2 µm. i, Quantifications of nuclear movement speed (5-min intervals for a total period of 16 h) (left) and total distance moved (over 16 h) (right). N ≥ 50 cells across three gels per condition. j, Representative example of cell and nucleus movement (left) speed and (right) angle of a single cell during cell tumbling over time. The angle is calculated with respect to the horizontal axis. k, Spearman correlation values of cell and nucleus speed and angle (direction) correlations in cell tumbling. N ≥ 20 cells across three gels per condition. *P < 0.05, **P < 0.005, ***P < 0.001, ****P < 0.0001. The P value is obtained using an unpaired two-tailed t-test for a, b, f and i.

Fig. 3 |. Early stage cell tumbling is essential for enhanced MSC chondrogenesis in three dimensions.

a, Centroid speed (left) and angular speed (right) of the major axis of cells on different days of chondrogenic induction. N ≥ 15 across three gels per condition. b, Representative heat map of matrix deformation over 15 min as determined by the tracking of embedded bead displacements on different days of chondrogenic induction. The cell locations before (dashed line) and after (solid line) are marked in black for SG and white for CG. N ≥ 10 across three gels per condition. c, Quantification of the maximum matrix displacement over 15 min on different days of chondrogenic induction. N ≥ 15 cells across three gels per condition. d, Quantification of live-cell F-actin dynamics (5-s intervals for a total period of 5 min) on different days of chondrogenic induction. N ≥ 10 cells across three gels per condition. e, Quantification of nuclear speed on different days of chondrogenic induction. N ≥ 50 cells across three gels per condition. f, Blebbistatin treatment regimen highlighting treatments on different days of the 21-day chondrogenic induction period. g, Representative Safranin O histology on cryo-sectioned samples for sGAG deposition after 21 days of chondrogenic induction with different durations of blebbistatin blocking in SG as observed from three independent experiments. Scale bar, 200 µm (for all images). ns, not significant, *P < 0.05, **P < 0.005, ***P < 0.001, ****P < 0.0001. The P value is obtained using one-way ANOVA with Tukey’s multiple comparisons test.

Fig. 4 |. Cell tumbling is associated with a repressed global chromatin state that is required for enhanced MSC chondrogenesis in three dimensions.

a,b, Representative western blotting images (a) and quantifications of western blotting images (b) of markers associated with nuclear mechanical loading and chromatin accessibility after 24 h of chondrogenic induction. Protein expression for each marker was normalized to GAPDH. N = 3 biological replicates per condition. c, Representative immunofluorescence images for different markers in SG (top) and CG (bottom) after 24 h of chondrogenic induction. Scale bar, 10 µm (for all images). d, Quantification of nuclear localization of signal intensity along distance from the nuclear boundary based on the immunofluorescence imaging of different markers. The x axes represent the signal intensity and the y axes represent the distance from the nuclear boundary for all the plots. The number of cells tested per condition are indicated in the plots and was across three gels per condition. e–g, ATAC-seq analysis after 16 h of chondrogenic induction. e, Principal component analysis of top 10,000 accessible chromatin regions with the highest variance. f, Volcano plot indicating the accessibility patterns of genomic regions after differential analysis between SG versus CG (adjusted P value < 0.05 and log₂[fold change] > 0.5). g, Heat map of the top 50 genomic regions that are statistically significantly different between SG and CG (adjusted P value < 0.01 and log[fold change] > 0.5). h, Top: treatment regimen of different chromatin inhibitors for the first 3 days in the 21-day chondrogenic induction period. Bottom: representative Safranin O histology on cryo-sectioned samples for sGAG deposition after 21 days of chondrogenic induction with different chromatin inhibitors as observed from three independent experiments. Scale bar, 100 µm (for all images). ns, not significant; *P < 0.05, **P < 0.005, ***P < 0.001, ****P < 0.0001. The P value is obtained from an unpaired two-tailed t-test for b. Data are presented as mean ± s.d. for b and d.

Fig. 5 |. Cell tumbling is associated with enhanced nuclear stretch and PLA2 signalling, which can be modulated to enhance MSC chondrogenesis.

a, Schematic depicting the PLA2–ARA nuclear mechanosensing pathway by which nuclei sense and transduce confinement and activate actomyosin contractility. b, Seconds-scale time-lapse image sequence of a portion of a nucleus labelled with H2B-eGFP highlighting nuclear deformations in SG (top) and CG (bottom). Scale bar, 2 µm (for all images). c, Representative example of nuclear area fluctuations over 60 s (3-s intervals) for one nucleus in SG and one in CG. d, Average nuclear area fluctuations over 60 s in SG and CG. N = 20 nuclei across three gels per condition. e, Representative western blotting images (left) and quantifications of western blotting images (right) of markers related to PLA2 signalling after 24 h of chondrogenic induction. The protein expression for each marker was normalized to GAPDH. N = 3 biological replicates per condition. Data reported represent mean ± s.d. f, Centroid speed (left) and angular speed (right) of the major axis of cells in response to the inhibition of PLA2 and supplementation of ARA in SG and CG. N ≥ 15 across three gels per condition. g, Nuclear speed in response to the inhibition of PLA2 and supplementation of ARA in SG and CG. N ≥ 15 across three gels per condition. h, Representative Safranin O histology on cryo-sectioned samples for sGAG deposition after 14 days of chondrogenic induction with PLA2 inhibition and ARA supplementation in SG as observed from three independent experiments. Scale bar, 200 µm (for all images). ns, not significant; *P < 0.05, **P < 0.005, ***P < 0.001, ****P < 0.0001. The P value is obtained using one-way ANOVA with Tukey’s multiple comparisons test. Panel a created with BioRender.com.

Fig. 6 |. Cell tumbling is associated with enhanced differentiation towards other lineages in three dimensions and occurs in various hydrogel platforms.

a, Representative Alizarin Red S histology on cryo-sectioned samples for mineral deposition after 28 days of osteogenic induction in SG (top) and CG (bottom). Scale bar, 200 µm. b, Centroid speed (left) and angular speed (right) of the major axis of cells on day 0 on osteogenic induction. N ≥ 15 across three gels per condition. c, Representative Oil Red O staining on fixed whole-gel samples for fat droplet formation after 28 days of adipogenic induction in SG (top) and CG (bottom). Scale bar, 200 µm. d, Centroid speed (left) and angular speed (right) of the major axis of cells on day 0 on adipogenic induction. N ≥ 15 across three gels per condition. e, Representative Safranin O histology on cryo-sectioned samples for sGAG deposition after 14 days of chondrogenic induction in degradable CG (top) and CG (bottom). Scale bar, 200 µm. f, Centroid speed (left) and angular speed (right) of the major axis of cells on day 0 on chondrogenic induction. N ≥ 15 across three gels per condition. g, Representative Safranin O histology on cryo-sectioned samples for sGAG deposition after 14 days of chondrogenic induction in viscoelastic CG (top) and CG (bottom). Scale bar, 200 µm. h, Centroid speed (left) and angular speed (right) of the major axis of cells on day 0 on chondrogenic induction. N ≥ 15 across three gels per condition. i, Schematic illustrating minutes-scale cell tumbling, and the associated timescales and molecular mechanisms. Left: cell tumbling involves whole-cell movements in a 3D matrix, subject to physical confinement and allowing local matrix reorganizability. Right: subcellular features of cell tumbling, including dynamics of matrix reorganization, F-actin and the nucleus occurring at the seconds-to-minutes timescale. The mechanisms identified that are associated with cell-tumbling-enhanced MSC chondrogenesis include enhanced actomyosin contractility, ezrin activity, LINC complex, PLA2–ARA-mediated nuclear mechanosensing and chromatin condensation. ns, not significant; *P < 0.05, **P < 0.005, ***P < 0.001, ****P < 0.0001. The P value is obtained using an unpaired two-tailed t-test for b, d, f and h. Panel i created with BioRender.com.