Microtubules regulate GEF-H1 in response to extracellular matrix stiffness

Abstract

Breast epithelial cells sense the stiffness of the extracellular matrix through Rho-mediated contractility. In turn, matrix stiffness regulates RhoA activity. However, the upstream signaling mechanisms are poorly defined. Here we demonstrate that the Rho exchange factor GEF-H1 mediates RhoA activation in response to extracellular matrix stiffness. We demonstrate the novel finding that microtubule stability is diminished by a stiff three-dimensional (3D) extracellular matrix, which leads to the activation of GEF-H1. Surprisingly, activation of the mitogen-activated protein kinase kinase/extracellular signal-regulated kinase pathway did not contribute to stiffness-induced GEF-H1 activation. Loss of GEF-H1 decreases cell contraction of and invasion through 3D matrices. These data support a model in which matrix stiffness regulates RhoA through microtubule destabilization and the subsequent release and activation of GEF-H1.

FIGURE 1:

Endogenous GEF-H1 activity is regulated by extracellular matrix stiffness. (A) T47D human breast carcinoma cells were cultured in compliant (1.3 mg/ml) and stiff (2.0 mg/ml) collagen gels, and (B) normal murine mammary gland (NMuMG) cells were cultured in compliant (2.0 mg/ml) and stiff (4.0 mg/ml) collagen gels for 2 h, and then cells in collagen gels were lysed. Activated GEF-H1 was isolated by pull-down using purified GST-Rho(G17A), and the amount of GEF-H1 that was pulled down and the total amount in the lysates were determined by Western blot with anti-GEF-H1 antibodies (top). Results from four experiments were quantified and normalized to total GEF-H1 (bar graphs). In T47D cells there was an ∼40% increase in GEF-H1 activation with increased matrix stiffness; n = 4 and **p < 0.001. Similarly, in NMuMG cells there was a nearly twofold increase in GEF-H1 activity in response to increased matrix stiffness; n = 4 and *p < 0.05.

FIGURE 2:

GEF-H1 is required for proper RhoA regulation and Rho-mediated actin–myosin contractility in normal murine mammary gland cells. (A) NMuMG cells were transiently infected with vector control (pGIPz) or mouse-specific GEF-H1 shRNA #1 or stably transfected with vector control (pLK0.1) or mouse-specific GEF-H1 shRNA #2. (B) Control and GEF-H1–knockdown cells were cultured in compliant and stiff collagen gels, and the amount of active RhoA (GTP bound) was measured after 2 h and normalized to total Rho. RhoA activity was increased with increased matrix stiffness; however, loss of GEF-H1 significantly disrupted proper RhoA regulation and its activation in response to increased matrix stiffness (shRNA#1, n = 3; and shRNA #2, n = 5; *p < 0.05. (C) Stable control and GEF-H1–knockdown (shRNA #2) cells were cultured in compliant collagen gels for 10 d, and the area of the collagen gel was measured daily. The resulting contraction curve demonstrates that GEF-H1–knockdown cells contracted the matrix twofold less compared with control cells over 10 d; n = 6. (D) Contractility measurements were normalized to total cellular protein on day 10 and demonstrate that the decrease in cell contraction is not due to fewer cells when GEF-H1 shRNA #2 was stably expressed; n = 6 and ***p < 0.001. (E) NMuMG cells were stably transfected with GFP vector control or GFP-GEF-H1. Note the additional, larger GEF-H1 immunoreactive band representing GFP-GEF-H1 in the transfectants. (F) GFP vector control and GFP-GEF-H1–expressing cells were cultured in compliant and stiff collagen gels, and the amount of active RhoA (GTP-bound) was measured after 2 h and normalized to total RhoA. RhoA activity was increased with increased matrix stiffness in GFP vector control cells. GFP-GEF-H1 overexpression increased RhoA activity compared with GFP vector control cells, and this activation was up-regulated further in response to matrix stiffness. n = 4 and *p < 0.05. (G) GFP vector control and GFP-GEF-H1 cells were cultured in compliant collagen gels with or without exoenzyme C3 for 7 d, and contractility was measured on days 0, 3, 5, and 7. The resulting contraction curve demonstrates that GFP-GEF-H1 overexpression increased cellular contractility compared with GFP control cells on day 7, and inhibition of RhoA blocks gel contraction in both control and GEF-H1xoverexpressing cells; n = 3; *p < 0.05 and **p < 0.0001.

FIGURE 3:

Microtubule stability is responsive to matrix stiffness and regulates GEF-H1 activity. (A) NMuMG cells cultured in compliant and stiff collagen gels were pretreated with vehicle control, 10 μM nocodazole, or 10 μM paclitaxel for 30 min, and then gels were floated in the presence of each treatment for an additional 2 h. Collagen gels were fixed in paraformaldehyde and stained for acetylated α-tubulin as a marker for stable microtubules (green) and total actin (red). Nocodazole and paclitaxel treatment resulted in clear differences in the structure of acetylated (stable) microtubules; however, there was no discernible change in actin. In contrast, there was a dramatic change in the structure of acetylated (stable) microtubules in stiff compared with compliant matrix conditions with vehicle (DMSO) treatment. (B) Acetylated (stable) microtubule structural morphology was quantified using CurveAlign software as a measure of anisotropy (increased organization of the microtubules) from 47 cells stained as in A. Each dot represents a data point, and the average value is represented by a horizontal line; n = 47 and ***p < 0.005. See Materials and Methods for details. (C) Cell lysates from NMuMG cells cultured in compliant and stiff collagen gels treated as described were probed for acetylated α-tubulin as a marker for microtubule stability and normalized to total α-tubulin by Western blot. The change in stability measured by acetylation was quantified for DMSO-treated cells (bar graph) and demonstrate that there was a significant decrease in acetylated α-tubulin, indicating a decrease in stable microtubules in stiff compared with compliant matrices; n = 3 and *p < 0.05. (D) NMuMG cells cultured in compliant and stiff collagen gels were pretreated with vehicle control (DMSO), 10 μM nocodazole, or 10 μM paclitaxel for 30 min, and then the gels were floated in the presence of DMSO or drug for an additional 2 h. The amount of activated GEF-H1 was determined by GST-Rho(G17A) pull-down and was normalized to total GEF-H1 by Western blot. There was a significant increase in GEF-H1 activity with increased stiffness; however, changes in microtubule stability via treatment with nocodazole and paclitaxel disrupted GEF-H1 regulation by matrix stiffness. Nocodazole treatment appeared to maximally activate GEF-H1 in both compliant and stiff matrix conditions, whereas treatment with paclitaxel prevented GEF-H1 activation in response to increased matrix stiffness compared with controls; n = 5 and *p < 0.05.

FIGURE 4:

ERK1/2 signaling is not required for GEF-H1 regulation and downstream signaling in response to matrix stiffness. (A) Untransfected NMuMG cells cultured in compliant and stiff collagen gels were pretreated with vehicle (DMSO) or 10 μM UO126 for 30 min and then floated in the presence of DMSO or drug for an additional 2 h. The amount of activated GEF-H1 was determined by Rho(G17A) pull-down and was normalized to total GEF-H1 by Western blot. Treatment with UO126 did not inhibit the activation of GEF-H1 by matrix stiffness; n = 4 and *p < 0.05 (bar graph). Western blots were also probed for pERK1/2 and total ERK1/2 to verify that the U0126 inhibited phosphorylation of ERK as expected. (B) NMuMG cells were cultured as described, and the amount of active (GTP-bound) RhoA was measured after 2 h and normalized to total Rho. RhoA activity was significantly increased with increased matrix stiffness in vehicle control and UO126 treatment, which suggests that its regulation did not occur downstream of MEK and ERK1/2; n = 6 and *p < 0.05.

FIGURE 5:

GEF-H1 expression is required for cell invasion across 3D matrices. (A) NMuMG and (B) MDA-MB-231 cells that were stably transfected with control shRNA or species-specific shRNA #5 targeting GEF-H1 were cultured within compliant and stiff collagen gels layered on top of a permeable membrane support and allowed to invade across the gel for 24 h (NMuMG) or 48 h (MDA-MB-231). Cell migration and invasion were quantified by Calcein AM live-cell staining of cells that transversed the membrane and normalized to the initial cell number. In both NMuMG and MDA-MB-231 cells, cell invasion through compliant matrices was significantly disrupted upon loss of GEF-H1 protein expression. n = 3 for each cell type and *p < 0.05. Cell invasion through stiff matrices was diminished compared with compliant matrices and may reflect either difficulties in traversing the denser matrix or a loss of persistent cell invasion toward the membrane.