Substrate compliance versus ligand density in cell on gel responses

Abstract

Substrate stiffness is emerging as an important physical factor in the response of many cell types. In agreement with findings on other anchorage-dependent cell lineages, aortic smooth muscle cells are found to spread and organize their cytoskeleton and focal adhesions much more so on "rigid" glass or "stiff" gels than on "soft" gels. Whereas these cells generally show maximal spreading on intermediate collagen densities, the limited spreading on soft gels is surprisingly insensitive to adhesive ligand density. Bell-shaped cell spreading curves encompassing all substrates are modeled by simple functions that couple ligand density to substrate stiffness. Although smooth muscle cells spread minimally on soft gels regardless of collagen, GFP-actin gives a slight overexpression of total actin that can override the soft gel response and drive spreading; GFP and GFP-paxillin do not have the same effect. The GFP-actin cells invariably show an organized filamentous cytoskeleton and clearly indicate that the cytoskeleton is at least one structural node in a signaling network that can override spreading limits typically dictated by soft gels. Based on such results, we hypothesize a central structural role for the cytoskeleton in driving the membrane outward during spreading whereas adhesion reinforces the spreading.

FIGURE 1

Ligand density and substrate compliance are postulated to influence cellular responses (Cukierman et al., 2001; Geiger, 2001). We tested that here with smooth muscle cells studied on various collagen-coated gels and glass in terms of morphological and organizational features at short times.

FIGURE 2

Polyacrylamide gel elasticity versus bis-acrylamide cross-linker concentration (in w/w %). (A) The elastic modulus, E, was measured for multiple 5% acrylamide gels by both macroscopic tension test (n = 157) and nanoindentation with an AFM (n = 36) as detailed in Materials and Methods. The inset demonstrates the linearity of the elasticity for the highest and lowest % bis-acrylamide gels as measured by tension tests on macroscopic samples; E is the slope. Such measurements of E were made on ∼1-mm-thick gels by both tension tests and AFM, and by AFM alone on 70-μm-thick gels with and without a monolayer of collagen (coll). The best-fit (dashed) curve through the ∼1-mm gel data is E = 42.6 [bis]–48.1 [bis]² (R² = 0.99). For AFM, the Poisson ratio, ν—indicating how a sample shrinks laterally when extended—must be assumed, but macroscopic tension tests suggest a value near 0.4–0.45. (B) For ν = 0.45, the two methods plotted against each other are linearly correlated (R² = 0.98) with slope nearly 1. Values of ν = 0.3–0.5 shift the cross-correlation between the two measurements methods by <±10%.

FIGURE 3

Representative SMC spreading on substrates that range from soft PA gels to rigid glass and with an intermediate collagen density of ∼100 ng/cm². Quantitation of cell areas is given in Figs. 4 and 5. In addition to differences in a mean spread area, the average cell shape factor tends to decrease from S = 0.49 ± 0.06 on a soft substrate (1 kPa gel) to S = 0.25 ± 0.04 for a cell on glass; the latter decreases even more with increasing collagen (see Fig. 6). The schematic depicts model spreading of a constant volume vesicle on a surface. The projected area of the spherical vesicle is just the equatorial area (πr²); when flattened, the surface area of the original sphere (4πr²) flattens to a projected area of 2πr², which is twofold larger than the sphere (scale bar = 20 μm).

FIGURE 4

Substrate-dependent spreading of SMCs. Projected SMC areas at 4 h (A, B) and 24 h (C, D) after plating were measured by image analysis and averaged for various PA gels or glass substrates with near-constant collagen I levels (∼5 × 10² ng/cm²) as assessed by fluorescence. Collagen I gels were also used. On a linear scale for the substrate modulus E (A, C), the results increase asymptotically toward glass, defining the saturation point of the hyperbolic fit (see Eq. 1 in text). On log-log scales (B, D), the sharp dependence of cell spreading on low modulus substrates is expanded, and plots fit a weak power law which can be used to estimate the effective gel elastic modulus that cells see when spreading on glass.

FIGURE 5

Spread cell area as a function of ligand density on soft, stiff, and rigid substrates. (A) The projected cell area was determined 4 h after plating (n > 10 per datapoint), giving the indicated average (mean ± SE). The smooth curves are calculated from a model for two-phase spreading (see Appendix) expressed in terms of both E (or E_app) and collagen density. Note that cells respond strongly to increasing collagen density on glass and hardly at all on soft gels. (B) Curved surface in three dimensions that fits SMC spreading.

FIGURE 6

Cell shape-dependence on collagen density. The cell shape factor, S, for the cell periphery is high for circular shapes and low for more ramified cell boundaries. The amphiphilic, cell-viable fluorophore PKH 67 highlights the cell boundary (scale bar = 20 μm) and allows for high-resolution fluorescence imaging of cell shape. Only the results for collagen on glass at 4 h after plating are shown here, but Fig. 6 illustrates similar trends for cells on gels. The inset is an AFM tapping mode image (scale bar = 2 μm) of an SMC spread on a rigid substrate; note the filamentous cytoskeleton clearly extending up to the leading edge.

FIGURE 7

GFP-actin and GFP-paxillin expressing SMCs on various collagen-coated substrates. The cells had been transfected en masse at least 1 day before detachment and replating on the desired substrate. Fluorescent cells shown were imaged at 4 h after replating (scale bar = 50 μm). Cells were observed on rigid glass (A, E), stiff PA gels (B, F), soft PA gels (C, C′, and G), and cross-linked collagen I gels (D). The contrast in the images reveals the assembly within live cells in terms of a ratio of freely diffusible versus organized component (e.g., G-actin versus F-actin).

FIGURE 8

On soft PA gels, slight overexpression of actin with GFP-actin transfection amplifies and overrides the weak optimum seen in spread area (A) and F-actin mass (B) over a range of collagen densities. Control cells were either transfected with bare GFP or nontransfected. For spread area, the solid bell-shaped curve through the control cells is the same as in Fig. 5 A; for F-actin mass measured as the integrated intensity of rhodamine phalloidin, the same curve shape is used to fit the control cells. As explained in the text, the dotted horizontal lines define baselines and the dotted, bell-shaped curves are stretched (2.6× or 3×) forms of the respective solid curves. Nonexpressing cells are excluded from analyses, as are completely rounded cells (<30%). Scale bar is 50 μm. (C) A schematic of F-actin assembly versus total actin pool is overlaid on hypothesized signaling curves. On optimum collagen, there is an impetus to assemble F-actin, which is amplified greatly with GFP-actin expression; on high or low collagen, the signal is effectively absent or inhibitory.

FIGURE 9

Schematic of cytoskeleton-driven spreading coupled to membrane tension. F-actin is sketched as green lines which (i) drive with a force f against the membrane while being anchored by focal adhesions at the other end. Newton's law of action-reaction equates f near adhesion sites with cell tractions (Lo et al., 2000), while balancing f at the plasma membrane by the tension τ_o. Slight increases in this tension—driven by the cytoskeleton (see Fig. 6, inset)—will tend to reinforce spreading (ii) by driving into adhesive contact plasma membrane reservoirs such as vesicles widely known in platelets (Grouse et al., 1990). Such recruitment processes tend to relax any increases in tension back to τ_o.