Force-induced cell polarisation is linked to RhoA-driven microtubule-independent focal-adhesion sliding

Abstract

Mechanical forces play a crucial role in controlling the integrity and functionality of cells and tissues. External forces are sensed by cells and translated into signals that induce various responses. To increase the detailed understanding of these processes, we investigated cell migration and dynamic cellular reorganisation of focal adhesions and cytoskeleton upon application of cyclic stretching forces. Of particular interest was the role of microtubules and GTPase activation in the course of mechanotransduction. We showed that focal adhesions and the actin cytoskeleton undergo dramatic reorganisation perpendicular to the direction of stretching forces even without microtubules. Rather, we found that microtubule orientation is controlled by the actin cytoskeleton. Using biochemical assays and fluorescence resonance energy transfer (FRET) measurements, we revealed that Rac1 and Cdc42 activities did not change upon stretching, whereas overall RhoA activity increased dramatically, but independently of intact microtubules. In conclusion, we demonstrated that key players in force-induced cellular reorganisation are focal-adhesion sliding, RhoA activation and the actomyosin machinery. In contrast to the importance of microtubules in migration, the force-induced cellular reorganisation, including focal-adhesion sliding, is independent of a dynamic microtubule network. Consequently, the elementary molecular mechanism of cellular reorganisation during migration is different to the one in force-induced cell reorganisation.

Fig. 1.

Stretching forces induces cell repolarisation in a microtubule-independent manner. (A) Cell reorganisation was analysed by fitting an ellipse to each cell outline and measuring the orientation angle, ϕ, between the long axis of the cell and the stretch direction. Three still images of a time series of an 8-hour phase-contrast movie of stretched (+) non-treated (upper row), nocodazole-treated (middle row) and taxol-treated (lower row) NIH3T3 cells illustrate the cell reorganisation. The direction of cyclic stretch is indicated by the double-headed arrows. Scale bars: 100 μm. (B) Time course of the reorientation of NIH3T3 cells treated with the indicated drugs upon uniaxial cyclic stretching of 8% at 1 Hz. The mean values for the order parameter cos2ϕ were calculated from the orientation angle ϕ (see panel A). A value =1 indicates a perfectly parallel orientation, –1 a perfectly perpendicular orientation and 0 a random cell orientation with respect to the stretch direction. (C) Quantification of the maximum cell reorientation (_MAX). The value <cos2ϕ>_MAX resembles the cellular mean orientation of the last 4 hours of cyclic stretch under the indicated conditions. `Stretch (–)' indicates non-stretched control conditions and `stretch (+)' the application of cyclic stretch. The disruption of MTs enhanced the cellular reorientation in comparison with non-treated cells (^*P<0.001). (D) Quantification of F-actin, FA and MT orientation under the indicated conditions after 3 hours of cyclic stretch. Cellular structures were analysed by using fast Fourier transformation (FFT) analysis of cell subareas and background subtraction combined with threshold application to yield <cos2ϕ> values. The orientation of each analysed structure was significantly higher under non-treated, nocodazole-treated and taxol-treated stretching conditions (+) compared with non-treated, non-stretched conditions (–) (^*P<0.05). Cells treated with cytochalasin D revealed no difference in alignment of MTs under stretch compared to the non-treated, non-stretched control (^‡P>0.2). (E) MTs, FAs and actin filaments after 3 hours of cyclic stretch. MTs were visualised by an anti-β-tubulin antibody, FAs were stained with an anti-paxillin antibody and actin was marked using phalloidin. `Stretch (–)' indicates non-stretched control conditions and `stretch (+)' the application of cyclic stretch. Actin stress fibres and FAs oriented perpendicular to the stretch direction (double-headed arrow) under non-treated conditions and despite nocodazole or taxol treatment. MT reorientation was dependent on the orientation of the actin cytoskeleton and did not occur in cytochalasin-D-treated cells. Scale bars: 10 μm.

Fig. 2.

Stretch-induced, oriented migration relies on intact microtubules. (A) Quantification of NIH3T3 cell-migration distance was performed by tracking the cell nucleus every 10 minutes over 8 hours under indicated conditions. Cyclic stretching at 1 Hz and 8% [stretch (+)] did not significantly change the overall distance of migration compared to non-stretched cells [stretch (–)]. Cell migration was basically blocked by stabilisation (taxol treatment) or disruption (nocodazole treatment) of MTs. ^‡No statistical difference, P>0.05. ^*Significance, P<0.0001. (B) Oriented fibroblast migration was determined by analysing the linear displacement of non-treated cells from their starting point to their ending point. The direction of migration was perpendicular (<cos2ϕ>=–0.45±0.1) during stretching (+) and random (<cos2ϕ>=0.1±0.13) under non-stretching conditions (–) (^*P<0.01).

Fig. 3.

RhoA activity increases, whereas the activity of Rac1 and Cdc42 remains constant, in response to stretching. (A) For ELISA measurements for active RhoA, Rac1 and Cdc42 proteins, NIH3T3 cells were investigated under non-treated, non-stretched [stretch (–)] and non-treated, cyclic stretched conditions [1 Hz, 8%; stretch (+)] at indicated time intervals. The data set was normalised to stretch (–) measurements. ELISA data show increased RhoA activity levels upon cyclic stretching (^*P<0.01); Rac1 and Cdc42 activity stays constant. (B) NIH3T3 cells were transfected with either pRaichu-RhoA or pRaichu-Rac1. FRET was determined for non-treated, non-stretched [stretch (–)], and non-treated, cyclic stretched conditions [stretch (+)] at indicated time intervals. FRET images were normalised to the acceptor fluorescence intensity and were displayed using a spectral colour look-up table indicating FRET levels. FRET measurements show that the RhoA activity level increased upon cyclic stretching in a non-polarised fashion (stretch direction is indicated by a doubled-headed arrow). Rac1 activity was high in protruding cell areas (arrowhead) and low in retractions (arrow). Black squares indicate areas of analysis for Rac1 activity gradient (supplementary material Fig. S3). Local Cdc42 activity was high in cell protrusions and did not change upon stretching (data not shown). Scale bars: 10 μm.

Fig. 4.

Inhibition of Rac and Rho activity influences cell and cytoskeleton polarisation under stretching conditions. (A) MTs, F-actin and FAs in NIH3T3 cells subjected to cyclic stretch for 3 hours. MTs were stained by an anti-β-tubulin antibody, FAs were visualised with an anti-paxillin antibody and actin was marked using phalloidin. The direction of cyclic stretch is indicated by a double-headed arrow. Cells either expressed dominant-negative Rac (RacN17) or were treated with C3 transferase for Rho inhibition. RacN17-expressing cells were identified by YFP co-transfection (supplementary material Fig. S6). Scale bars: 10 μm. (B) MT alignment occurred perpendicular to the stretch axis and correlates with F-actin orientation in RacN17-expressing cells (=–0.26). MTs were randomly oriented in C3-toxin-treated stretched cells (≈0) (^*P>0.05). `Stretch (–)' indicates non-stretched control conditions and `stretch (+)' the application of cyclic stretch.

Fig. 5.

Microtubules have limited control over (localised) RhoA and Rac1 and Cdc42 GTPase activity. (A) FRET measurements for RhoA and Rac1. NIH3T3 cells were transfected with either pRaichu-RhoA or pRaichu-Rac1 and treated with taxol or nocodazole prior to the FRET measurements. FRET was determined for non-stretched [stretch (–)] and cyclic stretching [1 Hz, 8%; stretch (+)] conditions at indicated time intervals. FRET images were normalised to the acceptor fluorescence intensity and are displayed using a spectral colour look-up table indicating FRET levels. RhoA activity increased upon cyclic stretching independent of MT stabilisation with taxol (stretch direction is indicated by a double-headed arrow). RhoA activity was high upon disruption of MTs (nocodazole) and did not further increase upon stretching. Under all conditions (taxol or nocodazole treatment), Rac1 activity levels were high in protruding cell areas (arrowhead) and low in retractions (arrow). Black squares indicate areas of analysis for Rac1 activity gradient (supplementary material Fig. S3). Local Cdc42 activity did not change upon application of cyclic stretch (data not shown; refer to the Results section). Scale bars: 10 μm. (B) For ELISA measurements for active RhoA, Rac1 and Cdc42 proteins, NIH3T3 cells were treated with either taxol or nocodazole and investigated under non-stretched [stretch (–)] and cyclic stretching conditions [stretch (+)] at the indicated time intervals. The data set was normalised to stretch (–). ELISA data show an increase in RhoA activity upon stretch in presence of taxol (^*P<0.01). Cyclic stretching did not further increase RhoA activity in nocodazole-treated cells (compare with A). Rac1 and Cdc42 activity in taxol- and nocodazole-treated cells did not change upon stretching. (C) Kymograph analysis of directional NIH3T3 protrusion activity over a time course of 3 hours. The direction of stretch is indicated by a double-headed arrow. To illustrate the analysis, a nocodazole-treated cell is displayed. A line was drawn along the cell edge of a composite phase-contrast image and the peaks were counted to yield a frequency of membrane protrusions per hour. Cell protrusions occurring perpendicular to the stretch axis were determined as `end'; cell protrusions parallel to the stretch direction were called `side'. Application of stretching doubled the protrusion activity at the ends but decreased activity by about half at the sides of cells (^*P<0.01). Independent of MT stabilisation (taxol) or disruption (nocodazole), the rate of protrusions remained higher at ends compared with those measured at the sides of the cells (^*P<0.01).

Fig. 6.

FA reorganisation occurs through a MT-independent sliding mechanism. FA dynamics was investigated in cells subjected to cyclic stretch of 8% at 1 Hz under indicated conditions (a, non-treated; b, nocodazole-treated; c, taxol-treated). NIH3T3 cells were transfected with pGFP-vinculin and time-lapse fluorescent movies were recorded (see supplementary material Movies 5-10). Grey-scale images on the left of each condition show single FA tracks over a time period of 115 minutes for non-treated and 180 minutes for nocodazole- and taxol-treated cells. Enlarged areas are indicated by white boxes. Selected FAs are colour-coded: green for FAs before stretch application (0 minutes), red for FAs after stretch (for non-treated after 100 minutes, for nocodazole and taxol after 160 minutes). Scale bars: 10 μm.