Crawling from soft to stiff matrix polarizes the cytoskeleton and phosphoregulates myosin-II heavy chain

Abstract

On rigid surfaces, the cytoskeleton of migrating cells is polarized, but tissue matrix is normally soft. We show that nonmuscle MIIB (myosin-IIB) is unpolarized in cells on soft matrix in 2D and also within soft 3D collagen, with rearward polarization of MIIB emerging only as cells migrate from soft to stiff matrix. Durotaxis is the tendency of cells to crawl from soft to stiff matrix, and durotaxis of primary mesenchymal stem cells (MSCs) proved more sensitive to MIIB than to the more abundant and persistently unpolarized nonmuscle MIIA (myosin-IIA). However, MIIA has a key upstream role: in cells on soft matrix, MIIA appeared diffuse and mobile, whereas on stiff matrix, MIIA was strongly assembled in oriented stress fibers that MIIB then polarized. The difference was caused in part by elevated phospho-S1943-MIIA in MSCs on soft matrix, with site-specific mutants revealing the importance of phosphomoderated assembly of MIIA. Polarization is thus shown to be a highly regulated compass for mechanosensitive migration.

Figure 1.

MSCs migrate from soft matrix toward scarlike, stiff matrix with increasing polarization of MIIB. (A) Human MSCs durotax across a gradient in stiffness on collagen-I–coated polyacrylamide hydrogels, even with a 3D overlay gel of collagen-I. MSCs also migrate upward into the soft overlay but only from the soft gel side of the gradient as indicated by arrows. (inset plot) Elasticity, E (±SD), determined by AFM and the gradient in E (red curve). Error bars show SEMs. (B) MSCs were immunostained for nonmuscle MIIB and with rhodamine-phalloidin for F-actin and imaged by fluorescence microscopy. MIIB localizes to the rear of the cell body only on the stiff matrix in 3D and 2D (2D lacks the collagen-I overlay). On soft matrix, MIIB is diffuse in the main cell body but is still absent from the farthest extensions in lamellipodia. Arrows show migration direction. Bars, 50 µm.

Figure 2.

MIIB polarization is suppressed in cells on soft matrix, but both the total density and insoluble fraction of MIIB remain constant with matrix stiffness. (A) F-actin and MIIB (gray) were stained in MSCs on soft or stiff matrix. The presence of the lamellipodium was used to identify the cell front. (bottom left) The front half of the cell and the rear half were marked as shown in the image. Migrating MSCs on soft or stiff matrix show uniform F-actin, but MIIB localizes to the cell rear only on stiff matrix. (B) Line scans of 6–10 µm width from the cell rear to front (of cells in A) were used to analyze the distributions of MIIB and F-actin. Each graph consists of data for one representative cell. Such data taken from three experiments are averaged in C and D. (C) Ratios of intensity from the front half of the cell to the rear half show that F-actin is always uniform. Inset image is representative of MIIA, which is throughout the cell body, although it tends to be depleted from the front lamellipodium. MIIA distribution is independent of matrix elasticity. (D) On soft matrix, MIIB is nearly homogeneous (except for the lamellipodium), whereas on stiff matrix, MIIB is more concentrated in the cell rear per Fig. 1 B. (E) Total MIIB density (black, open circles) is independent of matrix elasticity, and increasing matrix stiffness minimally affects the fraction of insoluble MIIB determined by Triton X-100 extraction (green triangles). Shaded regions indicate range of means. Means ± SEM for n ≥ 20 cells among three experiments per data point. au, arbitrary unit; Coll., collagen. Bars, 50 µm.

Figure 3.

MSCs crawl from soft to stiff matrix, with durotaxis appearing more sensitive to MIIB KD than MIIA KD. (A) Trajectories of cells with final positions in red after 12 h of imaging. Speed is determined from the cell’s contour length along the trajectory divided by time. (B) Durotaxis index versus time averaged for all cells. We define a step as the net displacement of the cell every 15 min, and we quantify the bias from the time-dependent number of steps to the right (R, stiff) and to the left (L, soft) as durotaxis index = (R − L)/(R + L). If 100% of steps are toward the stiff side, the durotaxis index = 1, whereas if there is no bias to cell migration, the durotaxis index = 0. A positive durotaxis index (P < 0.01) found only on gradient gels indicates biased migration toward stiff matrix after the indicated time constant. On homogeneous soft and stiff matrices, durotaxis index is not statistically different from zero, indicating a random direction. In all plots, means ± SEM for n ≥ 12 cells per data point among three independent experiments are given. (C) Durotaxis index vanishes for both low and high knockdown (KD) of MIIB (red), but a 50% KD of MIIA (blue) still yields a significantly nonzero durotaxis index, whereas a higher KD causes the durotaxis index to vanish. The error bars in the KD level reflect the variation determined from quantitative immunofluorescence imaging. Means ± SEM for n ≥ 20 cells per data point. For shallow MIIB KD, P < 0.05 compared with WT. For shallow MIIA KD, P < 0.05 when compared with both WT and to durotaxis index = 0. (D) Western blots against MIIB and MIIA for graded KDs of MIIA and MIIB, with actin used as a loading control. Bar graph summarizes three independent experiments in showing that minor KD of MIIB has minimal impact on MIIA. Error bars show SEMs. (E) MSCs with MIIB KD have extended tails and increased migration speed compared with WT. MIIA KD cells crawl more slowly than WT. Cells do not durotax with sufficient KD of either isoform, as indicated by the flower plots for cells on the gradient. The cell front and rear were determined from migration paths in time-lapse imaging. The arrow points to the extended tail. au, arbitrary unit; ctl, control. Bars, 50 µm.

Figure 4.

MIIA is more assembled on stiff matrix than soft, and the phosphomimetic S1943D is more soluble and mobile than WT or S1943A. (A, i) MSCs transfected with GFP-MIIA or GFP-MIIB on stiff matrix were photobleached along actin bundles and allowed to recover. MIIA recovered faster and thus was more mobile than MIIB. (ii) MSCs expressing GFP-MIIA on either soft 1-kPa matrix or stiff 34-kPa matrix were similarly analyzed, and MIIA in cells on soft matrix recovered more quickly. (iii) Phosphomimetic GFP-MIIA S1943D mutant recovered faster than GFP-MIIA S1943A, which was less than or equal to the behavior of GFP-MIIA on stiff matrix. Data are means ± SEM for at least five cells. (B) Fraction of insoluble MIIA is quantified by immunofluorescence of cell ghosts that were derived from Triton X-100 extraction of cells on matrices for 4 min (±SEM; n = 3). Western blotting of the insoluble fraction of the cells on matrices was also performed. Representative blot for MIIA is shown, normalized to total protein. (C) MSCs transfected with either GFP only, GFP-MIIA (WT), S1943D, or S1943A mutants were Triton X-100 extracted, and the extracted amount of MIIA (soluble) was compared with the insoluble fraction. Immunoblots show anti-MIIA blotting with β-actin as a loading control. When the amount of MIIA is normalized to β-actin, the only significant difference in the insoluble/soluble MIIA ratio is between S1943D and S1943A (P ≤ 0.05; ±SEM; n = 3). Endog., endogenous.

Figure 5.

S1943 of MIIA is matrix elasticity E regulated and impacts structural integration into filaments. (A) MSC on stiff matrix stained with anti-MIIA and with anti-pS1943. pS1943 is more diffuse compared with MIIA and less integrated into stress fibers. The total intensity of pS1943 normalized to MIIA signal always appeared higher for MSCs on soft than stiff matrix (n ≥ 20 cells). (B) Representative Western blot for pS1943 from MSCs cultured on soft or stiff matrix with MIIA as a loading control shows relatively more pS1943 per MIIA (representative of Western blots from n = 3 independent experiments). (C) The orientation of the fluorescence signal of MIIA was compared with that of pS1943 on both soft and stiff matrix conditions and plotted as a normalized abundance as a function of the angle. The peak of the fitted curve marks the oriented signal, whereas the baseline represents the isotropic structure. (inset) MIIA was less structured on soft matrix compared with signal on stiff matrix. (D) Orientation of myosin structure with GFP-MIIA and MIIA mutants. GFP-MIIA S1943D was less structured than MIIA WT or S1943A. Endogenous pS1943 was immunostained and again showed less structure on soft matrix in all transfects but showed more fiber integration with S1943A. We numbered all conditions in a table and then calculated the p-value in comparing all possible datasets. Squares filled in with yellow indicate a P < 0.05. Data are means ± SEM with ≥35 cells. endog., endogenous. Bar, 50 µm.

Figure 6.

Durotaxis and MIIB polarization are disrupted by MIIA S1943D overexpression and are maximal for WT levels of MIIB expression and pS1943. (A) MSCs on gradient gels transfected with WT GFP-MIIA show a durotaxis index similar to nontransfected, WT cells (yellow band). The S1943A mutant shows a reduced durotaxis index, and the S1943D mutant shows no significant durotaxis above background. Data are means ± SEM for ≥12 cells. (B) The same transfected MSCs were immunostained for MIIB, which polarizes in GFP-MIIA cells on stiff matrix (34 kPa) but does not on soft matrix (1 kPa) similar to nontransfected, WT cells (Fig. 1). The S1943A mutant shows a modest increase in MIIB polarization even on soft matrix, where these cells tend to spread more than any other cell (Fig. S5), as is found on stiff matrix. The S1943D mutant shows no significant MIIB polarization on soft or stiff matrix, even though these cells tend to spread on stiff matrix more so than any other cell (Fig. S5). The blue band is the range of durotaxis index when there is no durotaxis per Fig. 3 C. (C) Summary of all durotaxis index and polarization data from WT, KD, and overexpression experiments shows that WT cells are optimized for durotaxis and polarization. For only the data point involving transfection with S1943D, the S1943D is considered equivalent to pS1943 and considered part of the percentage of pS1943 in the graph. The surface plot illustrates the sensitivity of durotaxis and MIIB polarization to both MIIB expression level and the percentage of MIIA that is phosphorylated at S1943 within the cell. Circles are red for the durotaxis index, whereas squares are black for rear/front fluorescence. Data are means ± SEM for ≥20 cells among three independent experiments. Bars, 50 µm.

Figure 7.

Schematic summary of matrix elasticity–dependent assembly of MIIA, polarization of MIIB, and persistence in migration of single cells.