Mechanical force application to the nucleus regulates nucleocytoplasmic transport

Abstract

Mechanical force controls fundamental cellular processes in health and disease, and increasing evidence shows that the nucleus both experiences and senses applied forces. Such forces can lead to the nuclear translocation of proteins, but whether force controls nucleocytoplasmic transport, and how, remains unknown. Here we show that nuclear forces differentially control passive and facilitated nucleocytoplasmic transport, setting the rules for the mechanosensitivity of shuttling proteins. We demonstrate that nuclear force increases permeability across nuclear pore complexes, with a dependence on molecular weight that is stronger for passive than for facilitated diffusion. Owing to this differential effect, force leads to the translocation of cargoes into or out of the nucleus within a given range of molecular weight and affinity for nuclear transport receptors. Further, we show that the mechanosensitivity of several transcriptional regulators can be both explained by this mechanism and engineered exogenously by introducing appropriate nuclear localization signals. Our work unveils a mechanism of mechanically induced signalling, probably operating in parallel with others, with potential applicability across signalling pathways.

Extended data figure 1

a,b) Examples of curves showing fluorescence intensity as a function of time in the nucleus and cytoplasm in FLIP experiments on two example cells transfected with the diffusive 41kDa construct and seeded on a) 30 kPa in control condition and b) 30kPa with DN-KASH overexpression. Data represent the mean fluorescence intensity of the compartments (nucleus/cytoplasm), normalized with the mean of the whole cell before the beginning of photobleaching, and corrected for background signal. Each curve depicts a representative experiment of one cell each. c,d) Cartoon and equations describing the model used for fitting curves as in A,B, and calculating influx and efflux rates. The model considers the molecules to freely diffuse inside the nuclear and cytoplasmic compartments (see methods). e) Mobile fraction of the L_NLS 41kDa construct in the nucleus (Nuc) and cytoplasm (Cyt) of cells seeded on 1.5/30 kPa gels. N=19 cells from 3 independent experiments, lines show mean ±SEM f) For cells seeded on 1.5 and 30 kPa gels, correlation between nuclear to cytosolic ratios of volume, and of areas as measured in confocal slices used for FLIP measurements; regression equation y = 0,6075 x + 0,05375. N=20 (1.5kPa) and N=14 (30kPa) cells from 2 independent experiments. Black line shows the linear regression. Source numerical data are available in source data.

Extended data figure 2

a,b) Influx and efflux rates of diffusive constructs for cells seeded on 30 kPa gels, with or without DN-KASH overexpression. In a, both MW (p<1e-15) and DN KASH (p=1e-6) effects tested significant. In b, both MW (p<1e-15) and DN KASH (p=0,0002) effects tested significant. c,d) Influx and efflux rates of constructs containing L_NLS for cells seeded on 30 kPa gels, with or without DN-KASH overexpression. In c, both MW (p=0,0025) and DN KASH (p<1e-15) effects tested significant. In d, both MW (p<1e-15) and DN KASH (p=3.4e-10) effects tested significant. In all panels, N= 30 cells from 3 independent experiments. Two-way ANOVA, Šídák’s multiple comparisons test was used to obtain p-values between conditions. Data are mean ±SEM. Source numerical data are available in source data.

Extended data figure 3

a-c) Average fluorescence intensities of nuclear and cytoplasmic areas of cells seeded on substrates of 1.5 or 30 kPa stiffness and immunostained for importin α3 (imp α3) importin α1 (imp α1), and importin β1 (imp β1). N= 90 cells from 3 independent experiments. The effect of substrate stiffness tested significant for importin α3 (p=7.2e-8) and importin α1 (p=1.7e-5), but not for importin β1 (p=0.4971). p-values from Two-way ANOVA d-e) Corresponding example images showing the nucleus (Hoechst) and the distribution of the different importins. f) Corresponding quantification of nuclear to cytoplasmic ratio of importin localization. N= 91,98, 91, 98, 90, 90 cells (from left to right) from 3 independent experiments. p-values from independent two-tailed Mann-Whitney tests. g) N/C ratios of L_NLS-41 kDa or BFP constructs in cells seeded on 1.5 kPa gels before, during, and after nuclear deformation with AFM. h) L_NLS-41 kDa ratios normalized by BFP ratios, from panel g) paired measures. i,j) from g, corresponding paired dot plots of the time points right before and after force application. k) from g, corresponding % change in N/C ratios right after force application for both constructs. In g,h,i,j,k N= 15 cells from 3 independent experiments, p-values were calculated with a two-tailed paired t-test. l) N/C ratios of H_NLS-27 kDa construct in cells seeded on 1.5 kPa gels before, during, and after nuclear deformation with AFM. m) from l, corresponding paired dot plots of the time points right before and after force application. In l, m, N= 15 cells from 3 independent experiments. p-values were calculated with a two-tailed paired t-test. n) Corresponding images of constructs before and during force application, dotted line marks nucleus outline. o) N/C ratios of the L_NLS-41 kDa construct in cells co-transfected with DN-KASH and seeded on 1.5 or 30 kPa gels before, during, and after nuclear deformation with AFM. Data are mean ±SEM. p,q)from o, corresponding paired dot plots of the time points right before and after force application. In o,p,q, N= 15 cells from 3 independent experiments. p-values were calculated with a two-tailed paired t-test, traces of all cells are shown in Extended Data Fig. 8. r) Corresponding images of constructs before and during force application, dotted line marks nucleus outline. Scale bars, 20 μm. Note: in AFM experiments, non-mechanosensitive constructs (BFP and H_NLS) still show a small increase with force, likely due to lensing effects caused by changes in cell shape during indentation. This increase (~6% for BFP, ~2% for H_NLS) is much smaller than that of the mechanosensitive construct (L_NLS 41 kDa, ~14%), see panel k. Panel h in fact shows the response of the L_NLS construct after factoring out the response of BFP. Data are mean ±SEM in all panels. Source numerical data are available in source data.

Extended data figure 4

Relationship between mean N/C ratio as reported in figures, and corresponding coefficient of variation (standard deviation divided by the mean). The different points show all different constructs and conditions reported in the manuscript. Black dots indicate values of overexpressed engineered constructs, red squares indicate values of stained endogenous proteins. Source numerical data are available in source data.

Extended data figure 5

(a-d) Model predictions for N/C ratios (a), mechanosensitivities (b), influx rates (c) and efflux rates (d) for 41kDa constructs as a function of NLS affinity (modelled by the binding rate k_on between the NLS and importin α). e-f) Experimental Influx and efflux rates of 41 kDa constructs containing NLS signals of different affinity for importin β. In both cases (e,f), NLS strength and substrate stiffness effects tested significant (respectively: e) p<1e-15, p<1e-15, f) p<1e-15, p=2.4e-10). N= 30 cells from 3 independent experiments. p-values from Two-way ANOVA. Data are mean ±SEM. Source numerical data are available in source data.

Extended data figure 6

For M_NES constructs, influx rates (mediated by passive transport) and efflux rates (mediated by facilitated transport) as a function of molecular weight. N= 30 cells from 3 independent experiments. Substrate stiffness effects tested significative in both cases (a) p=5.1e-13; b) p<1e-15); MW only tested significative for influx, a) p<1e-15; b) p=0.2138). Two-way ANOVA, Šídák’s multiple comparisons test was used to obtain p-values between conditions. Data presented as mean ±SEM. c-d) Model predictions of N/C ratios (c) and mechanosensitivities (d) for an NLS with a binding rate k_on of 54 ms^-1 as a function of MW. Data are shown for experimentally measured N/C volume ratios (0.29) and for inverted volume ratios (3.5). e-f) Same predictions as in c,d for an NLS with a binding rate kon of 205 ms^-1. Note that these predictions simply evaluate the role of N/C volumes on import, they do not explicitly model the export cycle (and hence mechanosensitivities are above and not below 1). Source numerical data are available in source data.

Extended data figure 7

a-c) For Snail stainings at different conditions, quantifications of N/C ratios on 1.5/30 kPa substrates (a, N= 100 cells from 3 independent repeats), corresponding mechanosensitivities for the 3 different repeats (b), and representative images (c). d-f) For SMAD3 stainings at different conditions, quantifications of N/C ratios on 1.5/30 kPa substrates (d, N= 100 cells from 3 different repeats), corresponding mechanosensitivities for the 3 different repeats (e), and representative images (f). g-i) For GATA2 stainings at different conditions, quantifications of N/C ratios on 1.5/30 kPa substrates (g, N= 90 cells from 3 independent repeats), Corresponding mechanosensitivities for the 3 different repeats (h), and representative images (i). j-l) For NF-κβ stainings at different conditions, quantifications of N/C ratios on 1.5/30 kPa substrates, (j, N= 90 cells from 3 independent repeats), corresponding mechanosensitivities for the 3 different repeats (k), and representative images (l). For a-l, data are presented as mean ±SEM, scale bars correspond to 20 μm, and p-values from corrected multiple two-tailed Mann-Whitney (a,d) and two-tailed Mann-Whitney (g,j) tests. m) Relative gene expression of different genes as assessed with qPCR. Conditions are cells seeded on 1.5 or 30 kPa substrates, overexpressing or not a WT twist1 construct (Ctrl V5-twist1). Gene expression is shown relative to the 1.5 kPa condition without overexpression. n=2 independent experimental repeats. Source numerical data are available in source data.

Extended data figure 8

Plots showing the evolution with time of N/C ratios before, during and after force application to the cell nucleus for all cells measured. a-b) AFM experiments reported in Figure 3, c) Figure 5, and d-h) Extended Data Figure 3. Source numerical data are available in source data.

Figure 1. Nucleocytoplasmic transport is mechanosensitive.

a) Cartoon of light-activated nucleocytoplasmic shuttling construct. Mild NLS is always active, NES is activated only upon light excitation. b) Time sequences of construct fluorescence before, during, and after excitation for cells seeded on 1.5/30 kPa substrates, with or without DN KASH overexpression. Scale bars, 20 μm. c-e) Corresponding quantifications of N/C ratios, and coefficients of exit and subsequent re-entry of constructs into the nucleus (in units of s^-1, obtained by fitting an exponential to the curves, see methods). (N=20, 22, 21, 21 cells per condition (1.5 kPa, 30 kPa, 1.5 kPa DN KASH, and 30 kPa DN KASH, respectively) from 3 independent experiments, data are presented as mean values +/- SEM.In c) the bar indicates the statistical significance between the last timepoint of 1.5kPa and 30kPa values. In d-e, p-values calculated with 2-way ANOVA Šídák’s multiple comparisons test. Source numerical data are available in source data.

Figure 2. Passive diffusion through NPCs is mechanosensitive for small MWs.

a) Cartoon of constructs with EGFP and different amount of repeats of PrA domains. b) Images showing fluorescence of indicated constructs on 1.5/30 kPa substrates. c) N/C ratios of constructs on 1.5/30 kPa substrates as a function of MW. N=120 cells from 3 independent experiments. Significant effects of stiffness and MW were observed (p <1e-15 and p <1e-15; computed via 2-way ANOVA). d) Example of a FLIP experiment: a laser photobleaches a region of the cell cytoplasm, and fluorescence intensities are recorded over time in nucleus and cytoplasm. Resulting curves are fitted to a kinetic model to obtain influx and efflux rates (see methods). e,f) Influx and efflux rates on 1.5 and 30 kPa substrates as a function of MW of the constructs. N=30 cells from 3 independent experiments. The effects of both substrate stiffness and MW were significant in both e,f). p-values e) 2.9e-8, <1e-15, f) 4.0e-8, <1e-15, computed via 2-way ANOVA. Scale bars, 20 μm. Data are mean ±SEM. Source numerical data are available in source data.

Figure 3. Differential mechanosensitivity of facilitated import versus passive diffusion explains force-induced nuclear translocation.

a) Example importin β-GFP images for cells on 1.5/30 kPa substrates. b-d) Corresponding importin β-GFP influx rates (b), efflux rates (c), and resulting N/C ratios (d). N=30, 30, and 60 cells from 3 independent experiments. p-values calculated with two-tailed Mann-Whitney test. e) Cartoon of constructs with EGFP, different number of repeats of PrA domains, and NLS of different affinities to importin α. f) Example images of L_NLS-41 kDa construct for cells on 1.5 and 30 kPa substrates. g-i) Corresponding Influx rates (g), efflux rates (h), and resulting N/C ratios (i) of L_NLS-41 kDa construct. N=30, N=30, N=120 cells from 3 independent experiments respectively each. p-values calculated with two-tailed Mann-Whitney test. j) N/C ratios of L_NLS-41 kDa or diffusive 41 kDa constructs in cells seeded on 1.5 kPa gels before, during, and after nuclear deformation with AFM. Graphs on the left show paired dot plots of the time points right before and after force application. p-values were calculated with two-tailed paired t-test. k) Corresponding % change in N/C ratios right after force application for both constructs. p-value was calculated with a two-tailed unpaired t-test with Welch’s correction. In j,k, N= 16 cells from 3 independent experiments, traces of all cells are shown in Extended Data Fig. 8. l) Corresponding images of constructs before and during force application, dotted line marks nucleus outline. Scale bars 20μm. m) Cartoon summarizing the effects of nuclear force and MW on active and passive transport. Passive transport decreases with MW, and depends on force only for low MW molecules. Active transport does not depend on MW, and depends on force regardless of MW. Note that active transport arrows also show a small arrow in the export direction, as discussed in the text. n) Influx rates (mediated by facilitated transport) of L_NLS constructs with different molecular weights. The effect of substrate stiffness and MW tested p<1e-15 and p=0.0004. o) Efflux rates of L_NLS constructs (mediated by passive transport) with different molecular weights. The effect of substrate stiffness and MW tested p=3,5e-11 and p<1e-15. In n), o), N= 30 cells from 3 independent experiments. Two-way ANOVA, Šídák’s multiple comparisons test was used to obtain p-values between conditions. Data are mean ±SEM in all panels. Source numerical data are available in source data.

Figure 4. Balance between affinity to importins and MW defines the mechanosensitivity of nuclear localization.

a,b) Qualitative prediction from conceptual model of how MW and affinity to importins should affect N/C ratios (a) on soft substrates and their mechanosensitivity (b) (see methods). Mechanosensitivity is defined as (N/C)_stiff/(N/C)_soft. c-e) Representative examples of construct distribution in cells seeded in substrates of 1.5kPa or 30kPa, for L_NLS constructs at different MW, M_NLS constructs at different MW, and 41kDa constructs at different NLS strengths. f-h) N/C ratios corresponding to the same conditions as C-E. i-k) Mechanosensitivity corresponding to the same conditions as C-E. l-m) Kinetic model predictions of N/C ratios (l) and mechanosensitivities (m) for NLS of different affinities for importin α (modelled through the binding rates k_on between the NLS and importin α, with values of 54 and 205 ms^–1) as a function of MW. n-o) Model predictions of N/C ratios (n) and mechanosensitivities (o) for 41kDa constructs, as a function of increasing NLS strength. Statistics: f) N= 120 cells from 3 independent experiments. Both MW (p<1e-15) and Stiffness (p<1e-15) effects tested significant. g) N= 120 cells from 3 independent experiments. Both MW (p<1e-15) and Stiffness (p=0,0015) effects tested significant. h) N= 120 cells from 3 independent experiments. Both NLS strength (p<1e-15) and Stiffness (p=0,0012) effects tested significant. Two-way ANOVA, Šídák’s multiple comparisons test was used to obtain p-values between conditions. Scale bars: 20 μm. Data are mean ±SEM. Source numerical data are available in source data.

Figure 5. Balance between affinity to Exportin1 and MW defines the mechanosensitivity of nuclear localization in constructs containing NES signals.

a-c) Representative examples of construct distribution in cells seeded in substrates of 1.5kPa or 30kPa, for H_NES constructs at different MW, M_NES constructs at different MW, and L_NES constructs at different MW. d-f) N/C ratios corresponding to the same conditions as A-C. d) N= 90 cells from 3 independent experiments. Both MW (p<1e-15) and Stiffness (p=0,0162) effects tested significant. e) N= 120 cells from 3 independent experiments. Only MW effects tested significant (p<1e-15). f) N= 90 cells from 3 independent experiments. Both MW (p<1e-15) and Stiffness (p=0,0001) effects tested significant. Two-way ANOVA, Šídák’s multiple comparisons test was used to obtain p-values between conditions. g-i) Mechanosensitivity corresponding to the same conditions as A-C. Mechanosensitivity is defined as (N/C)_stiff/(N/C)_soft (n=3 experiments). j) N/C ratios of H_NES 41 kDa construct in cells seeded on 1.5 kPa gels before, during, and after nuclear deformation with AFM. k) From data in j, paired dot plots of the time points right before and after force application. In j and k, N= 15 cells from 3 independent experiments. p-values were calculated with a two-tailed paired t-test, traces of all cells are shown in Extended Data Fig. 8. l) Corresponding images of constructs before and during force application, dotted line marks nucleus outline. Scale bars 20μm. Data are mean ±SEM. Source numerical data are available in source data.

Figure 6. The mechanosensitivity of twist1 can be re-engineered with exogenous NLS sequences.

a) N/C ratios of endogenous twist1 for cells on 1.5/30 kPa substrates, and under indicated treatments. N= 100 cells from 3 independent experiments. p-values from two-tailed Mann-Whitney tests, corrected for multiple tests in the intracondition comparisons with the two-stage step-up method of Benjamini, Krieger and Yekuteili. b) Corresponding images of twist1 distribution. c) Scheme of different twist1 mutants. Mutations inactivating both NLS sequences and the G3BP2 binding motif are indicated in red. d) N/C ratios of transfected twist1 mutants for cells on 1.5/30 kPa substrates. N= 90 cells from 3 independent experiments. p-values from two-tailed Mann-Whitney tests, corrected for multiple tests with the two-stage step-up method of Benjamini, Krieger and Yekuteili. e) Corresponding construct mechanosensitivities, defined as (N/C)_stiff/(N/C)_soft (N= 3 experiments). f) Corresponding images showing the distribution of the different mutants. Scale bars, 20 μm, data are mean ±SEM. Source numerical data are available in source data.