Figure 1.
Comparison of strains imposed upon cells in different designs of equibiaxial stretching devices.
(A) and (B) illustrate two representative equibiaxial cell stretching systems currently available, while (C) depicts the design that we have developed. The left and right columns correspond to the states before and after stretching, respectively. (A) When the circular clamp holding the membrane is moved downwards along the concentrically placed indenter, homogeneous strains are generated on the membrane inside the indenter. However the periphery of the membrane is stretched with different magnitudes. (B) Strain heterogeneity is also seen in the devices that adopt a vacuum to stretch the membrane. The vacuum creates a low-lying well outside the indenter, which can impart compressive forces to cells [32]. (C) When cells are confined to the central region of the membrane lying within the boundary of the indenter, they are expected to be subject to homogeneous strains.
Figure 2.
Finite element simulations to select membrane dimensions.
(A) Cross sectional sketch of proposed membrane with wall (not to scale). Red arrows indicate the region near the junction of the wall and the membrane, where strain heterogeneity can arise. To reduce this heterogeneity, values for the wall thickness (i) and the wall height (ii) were selected by varying them in finite element simulations. (B) Deformation of the PDMS membrane for Design 1 (top) and Design 2 (bottom) computed by asymmetrical finite element modeling. Colors indicate magnitude of total displacement. Wall thicknesses for Design 1 and 2 are 0.5 mm and 1 mm, respectively. For both Designs 1 and 2, the thickness of the base membrane is 0.5 mm, and the wall height is 10 mm. (C) Normalized radial strain (er/er0) profiles for both designs near the wall-membrane junction. Design 1: er0 = 8.24%, Design 2: er0 = 7.76%. Note the smaller variation near the wall interface in Design 2 as compared to that in Design 1.
Figure 3.
Fabrication of PDMS membranes and design of cell stretching device.
(A) An aluminum mold consisting of a plunger, a middle piece and a base, was used to cast PDMS. Mold surfaces that form the base of the membrane (indicated by cross-hatching) are polished to a mirror-like finish to produce a clear surface for imaging. (B) Cross-section of an assembled mold. The base (3) and middle (2) components are assembled to form a cavity into which PDMS elastomer is poured. The top plunger (1) compresses the elastomer into the desired shape. The blowout details how mold assembly leads to the formation of a 1 mm thick wall and a 0.5 mm thick base. (C) The cured PDMS membrane has an outer diameter of 95 mm and a central well of inner diameter 35 mm. The depth of the well is 10 mm. (D) Cross-section of the membrane holder and indenter assembly. The membrane is affixed to its holder using a clamping ring made of aluminum which holds it in a circular groove at the base. Vertical motion of the holding plate leads to stretching of the membrane over the concentrically placed indenter beneath. (E) Schematic of the cell stretching device. The membrane is affixed in a membrane holder which is engaged with a holding plate. A linear slide operated by a stepper motor controls movement of the plate along a vertical axis. Bearings moving along guidance shafts constrain movement to this axis. Downward displacement of the holding plate causes membrane stretching upon a hollow indenter. The indenter, shafts, and linear slide are mounted upon an aluminum base. The motion of the holding plate is controlled by a computer.
Figure 4.
Variation of strain on the membrane surface against downward displacement.
(A) A snapshot of the regular grid of displacement vectors computed using the particle image velocimetry (PIV) algorithm. This set of displacement vectors tracks the motion of the marker positions as the holding plate is displaced downwards by 3.5 mm. The white circle and the red polygonal line denote the boundary of the culture surface and the boundary for strain computation presented in (E), respectively. (B – D) Mean and standard deviation of the radial (Err), circumferential (Ecc), and shear (Erc) components of the strain field over the entire culture surface, respectively, plotted against corresponding downward holding plate displacement during one loading-unloading cycle. At each holding plate displacement, data points from the four membranes are horizontally staggered to allow results from individual membranes to be seen. The data points in red indicate the loading phase, during which the holding plate is being displaced downward from its initial position (0 mm) to a maximum of 11.5 mm; the data points in blue indicate the unloading phase, during which the holding plate is displaced back to its initial position. The shear strain is consistently zero regardless of the displacement of the holding plate. The curves for the radial and circumferential strains show strong overlap. Some hysteresis is seen when the membrane is unloaded; however, upon returning to the zero position, Err and Ecc are both consistently zero. (E) Color intensity maps of the radial and circumferential components of strain over the culture surface of membrane 4 in (B), (C), and (D) during loading phase and displaced downward by 3.5 mm (top panel), 5.5 mm (middle panel), and 9.5 mm (bottom panel). The coefficient of variation (CV), defined as the ratio of the standard deviation to the mean, is given for each intensity map.
Figure 5.
Cell culture on collagen-coated PDMS membranes.
(A) Phase-contrast images (20 X) of RPTP-α+/+ MEFs (left) and HEK 293A cells (right), respectively. The upper row shows images of cells cultured on the PDMS membrane, while the lower row depicts cells on standard tissue-culture plastics (polystyrene). Scale bars: 50 µm. (B) Images of an RPTP-α+/+ MEF cell before (left) and after (right) application of 10% equibiaxial strain. The diagonal lines indicate the cell length before stretching the membrane. Note the increase of the cell dimension upon membrane stretching. Scale bars: 20 µm. The images were acquired using a manually operated device (Figure S3).
Figure 6.
Immunoblot analysis of cell signaling induced by cyclic stretching.
(A) RPTP-α+/+ fibroblasts cultured on collagen-coated membranes were subjected to no mechanical stimulus (lane 1), periodic vertical motion with no strain (lane 2), and cyclic strains of 6% (lane 3) and 13% (lane 4), at a frequency of 0.5 Hz for 30 minutes. While no change was observed in the levels of ERK expression, its phosphorylation level increased in lanes 3 and 4 as compared to the first two. At least three independent experiments were conducted to quantify stretch-dependent ERK phosphorylation. Paired Student’s t-tests showed that cell stretching produced a significant increase in ERK phosphorylation (**p<0.01, n = 3). (B) Anti-phospho-tyrosine immunoblot analysis showed different strain responses among individual tyrosine-phosphorylated proteins. Arrows indicate bands at molecular weights of 200 kDa, 125 kDa, 33 kDa and 17 kDa, respectively. The signal intensity from these bands was normalized against the corresponding α-tubulin intensity, and the mean values of the fold change in tyrosine phosphorylation from three independent experiments were plotted. At 13% strain, the 200 kDa band showed a significantly decreased intensity, while the intensity of the other three bands (125 kDa, 33 kDa and 17 kDa) increased significantly as compared to the unstrained control (* <0.05, **p<0.01, paired Student’s t-tests, n = 3). (C) MEFs subjected to cyclic strain ranging from 1% to 6% (0.5 Hz, 30 min) exhibited a stepwise increase in the level of ERK phosphorylation in response to strain. All the stretched samples showed a statistically significant increase in ERK phosphorylation relative to the unstrained control (*p<0.05, **p<0.01, ***p<0.005, paired Student’s t-tests, n = 3). (D) At small magnitudes (≤6%) of cyclic strain, tyrosine phosphorylation did not show significant changes from the unstrained control. Note: All error bars in this figure indicate standard deviations.
Figure 7.
Correlation between strain magnitudes and phosphorylation level.
(A) Scatterplot of ERK phosphorylation levels against applied strains, obtained from three independent experiments (Figure 6C). Pearson’s correlation coefficient (R) was calculated and found to be statistically significant (R = 0.80, n = 15, p<0.01). (B) Scatterplot of tyrosine phosphorylation levels against applied strains, obtained from three independent experiments (Figure 6D). Pearson’s correlation coefficients between phosphorylation level and strain magnitude were calculated for four bands (200 kDa, 125 kDA, 33 kDa, 17 kDa) in the phosphotyrosine blot (Figure 6D). The 125 kDa and 17 kDa bands showed moderate positive correlation with applied strain magnitude (R = 0.48 and R = 0.36, respectively). However, none of the coefficients were statistically significant (n = 15, p>0.05 for all R).