Active Traction Force Response to Long-Term Cyclic Stretch Is Dependent on Cell Pre-stress

Abstract

Mechanical stimulation is recognized as a potent modulator of cellular behaviors such as proliferation, differentiation, and extracellular matrix assembly. However, the study of how cell-generated traction force changes in response to stretch is generally limited to short-term stimulation. The goal of this work is to determine how cells actively alter their traction force in response to long-term physiological cyclic stretch as a function of cell pre-stress. We have developed, to our knowledge, a novel method to assess traction force after long-term (24 h) uniaxial or biaxial cyclic stretch under conditions of high cell pre-stress with culture on stiff (7.5 kPa) polyacrylamide gels (with or without transforming growth factor β1 (TGF-β1)) and low pre-stress by treating with blebbistatin or culture on soft gels (0.6 kPa). In response to equibiaxial stretch, valvular interstitial cells on stiff substrates decreased their traction force (from 300 nN to 100 nN) and spread area (from 3000 to 2100 μm²). With uniaxial stretch, the cells had similar decreases in traction force and area and reoriented perpendicular to the stretch. TGF-β1-treated valvular interstitial cells had higher pre-stress (1100 nN) and exhibited a larger drop in traction force with uniaxial stretch, but the percentage changes in force and area with stretch were similar to the non-TGF-β1-treated group. Cells with inhibited myosin II motors increased traction force (from 41 nN to 63 nN) and slightly reoriented toward the stretch direction. In contrast, cells cultured on soft gels increased their traction force significantly, from 15 nN to 45 nN, doubled their spread area, elongated from an initially rounded morphology, and reoriented perpendicular to the uniaxial stretch. Contractile-moment measurements provided results consistent with total traction force measurements. The combined results indicate that the change in traction force in response to external cyclic stretch is dependent upon the initial cell pre-stress. This finding is consistent with depolymerization of initially high-tension actin stress fibers, and reinforcement of an initially low-tension actin cytoskeleton.

Introduction

Cell traction forces are essential to numerous cell processes, such as migration (1, 2), adhesion (3), and extracellular matrix (ECM) assembly (4). Cells generate traction forces through the actin-myosin machinery of the cytoskeleton, and these forces are transmitted to the surrounding ECM via integrins. Cell traction forces produce stress and strain in the surrounding matrix, and the external stress and strain in the ECM modulate cellular functions such as ECM protein secretion (5) and differentiation (6). Many disease states, such as hypertension, fibrosis, and cancer, result in the dysregulation of a cell’s ability to generate traction forces (7, 8, 9); thus, traction force is an important functional readout of cell phenotype (10).

Traction force is also a useful measure of a cell’s integrated response to physical and biochemical cues. Cell traction forces have been shown to be influenced by substrate stiffness (11, 12), restriction of both cell shape (13) and spreading (11, 14), ECM composition (15), and the presence of soluble factors (12). Recently, rapid increases and decreases in traction force in response to stretch have been documented (16, 17); however, the ways in which the traction forces are altered upon remodeling of the cytoskeleton in response to long-term external stretch remains largely unknown. Understanding the ways in which cell-generated forces change with long-term cyclic stretch would provide insight into the regulation (or dysregulation) of cell tension during cell differentiation, proliferation, and apoptosis, biological events that occur over timescales not currently investigated in combination stretch/traction force studies.

Collectively, observations from a variety of studies indicate that many concomitant factors affect the ability of a cell to generate traction force in static culture. As substrate stiffness increases, the ability of cells to generate traction force increases (11, 12). On stiffer substrates, spread area (14), cell form factor (a function of cell area and perimeter) (13), and elongation (14) may also increase. These morphometric parameters affect the traction force that a cell is able to generate; it has been demonstrated that cells with larger spread area (14) and greater elongation (18) generate higher magnitudes of traction force compared to their respective controls. Relationships between stiffness, area, and cell shape become even more complicated with the addition of dynamic stimulation. We have previously demonstrated that with cyclic equibiaxal stretch for 6 h, cells that are small and rounded in static culture on soft (0.6 kPa) substrates increase in area and elongation similar to cells cultured statically on stiff substrates (19). Additionally, increases in traction force and cell proliferation have been reported with short-term stretch on soft micropillar substrates (20).

The cell’s ability to generate traction force changes rapidly with stretch and depends on the magnitude, direction, type, and duration of stretch, as well as the cell’s orientation with respect to the stretch direction. Fredberg and colleagues observed that on polyacrylamide (PA) gels, cell traction, as represented by the contractile moment, decreases precipitously in cells subjected to a rapid step and release of stretch (16, 21). The recovery after the immediate loss of traction depends upon the uniformity of stretch. With a homogeneous strain field, the traction recovers to baseline over ∼10 min, whereas with a nonhomogeneous strain field, cells increase their contractile moment (21). With multiple cycles of repetitive homogeneous stretch and release, the cells were not able to recover their baseline contractile moment within the 10-min dwell time. Subcellular measurements of traction force utilizing stretch of micropillar arrays indicate that the cell traction response can vary with location in the cell, with the largest changes in traction force occurring along the periphery (22). The duration of stretch also affects the traction force response; Fu and colleagues observed that cell-generated traction forces increase initially in response to a rapid step stretch of cells cultured on micropillar arrays, then return to or fall below initial baseline values over the next 60 min as the stretch is held (17). More recently, the same research group utilized microcontact-printed areas of various geometries and demonstrated that cell shape and orientation control the cell’s traction response to stretch (23); notably, elongated cells oriented perpendicular to the stretch exhibited little response to stretch, whereas those aligned with the stretch exhibited large changes in traction force. These pioneering studies indicate that the traction force changes rapidly with short-term stretch (<1 h); however, extrapolating the cytoskeletal response to long-term cyclic stretch, i.e., how stretch changes the traction force, is not possible in these studies due to active adaptations of the cytoskeleton relative to stretch, such as α-smooth muscle actin (α-SMA) incorporation into stress fibers and reorientation of the cell away from the direction of stretch.

The goal of this work is to determine how cells alter their traction force in response to long-term physiological cyclic stretch. We hypothesize that changes in traction force with stretch are dependent upon the cell pre-stress. As defined in prior reports (24), cell pre-stress is the overall level of cytoskeletal tension under static conditions (before stretch). Cell pre-stress can be quantified by total traction force, as done here, or by total contractile moment applied to the substrate, although it is understood that neither of these metrics have units of engineering stress. In this study, cell pre-stress was modulated by culturing valvular interstitial cells (VICs) and U2OS cells on tunable-modulus PA hydrogels (7.5 kPa and 0.6 kPa) and treating the cultures with transforming growth factor β1 (TGF-β1) or blebbistatin. The hydrogels were attached to custom-designed polydimethylsiloxane (PDMS) wells, and 10% biaxial or uniaxial stretch was applied with microcontrolled motors. After 24 h of stretch, wells were removed from the stretch device and imaged for traction force measurements and analysis of cell morphology.

Materials and Methods

Cell culture

VICs were chosen because they are fibroblastic-like cells capable of alternating their pre-stress level with TGF-β1 supplementation (25). VICs were isolated from porcine heart valve leaflets obtained from a local abattoir (Blood Farm, Groton, MA) within 3 h of tissue harvest, according to published protocols (19). Cells from passages 3–4 were used for all experiments. U2OS (human osteosarcoma) cells, generously donated by Dr. Roland Kaunas (Texas A&M University), were used as comparison cells to determine if there is a cell-type-specific traction force response to stretch.

For all experiments, cells were seeded at a density of 2,500 cells/cm² and cultured for 24 h at 37°C, 10% CO₂, in standard media (1X DMEM, 10% FBS, and 1% antibiotic-antimycotic). To predifferentiate the VICs to the myofibroblast phenotype and thus increase the cell pre-stress, a subset of VICS was pretreated with 5 ng/mL TGF-β1 (AcroBiosystems, Newark, DE) for 5 days before cell seeding. To inhibit myosin II activity and reduce the cell pre-stress, a separate subset of cells was treated with 10 μm blebbistatin (Sigma-Aldrich, St. Louis, MO), prepared according to the manufacturer’s directions, for 1 h before stretching. After addition to the wells, blebbistatin remained in the media for the duration of the experiment.

Stretch system and culture well development

A custom stretch system and compliant culture well system were developed with maximal optical clarity and uniformity in the strain field to apply long-term cyclic stretch to the cells (please see the Supporting Material for details). The stretch device consists of four microcontrolled stepper motors on a low-profile aluminum base that can be mounted on an inverted microscope or placed in a standard temperature- and CO₂-controlled incubator. The system is controlled by an open-source multiplatform integrated development environment program that allows for stretch patterns, such as pure uniaxial stretch and uniaxial stretch that alternates direction with every cycle which are not available on commercial cell-stretching systems (e.g., Strex).

Custom compliant culture wells were designed to circumvent limitations of commercially available systems for stretching cells that make traction force measurements challenging, namely, thick membranes and nonuniform strain fields in Flexcell and Strex wells. The inverse molds for the custom wells were created with a thin metal spacer to form the bottom membrane. This allowed for the wells to be created in one casting of Polydimethylsiloxane, enabling the bottom to form without prestretch (for more information on the design of the well, please see the Supporting Material). The molds for the wells were drawn in Solidworks (Daussalt Systems, Waltham, MA) (Fig. 1 a) and machined out of polycarbonate sheets (Tap Plastics, Stockton, CA) using a CNC machine. The wells were optimized for uniformity of strain field using finite-element analysis (Ansys, Canonsburg, PA). Wells were cast with a 16:1 base/cross-linker ratio of polydimethylsiloxane (PDMS; Sylgard 184 Silicone Elastomer Kit, Dow Corning, Midland, MI) and strain patterns under various stretch regimens were validated using high-density displacement mapping software (courtesy of Dr. Glenn Gaudette, Worcester Polytechnic Institute) (Fig. 1 c). The custom well has an integrated bottom membrane that is sufficiently thin (<150 μm) to allow for quality images to be obtained even with attachment of a 70-μm-thick PA gel (details in the next section).

Figure 1.

Overview of the method. (a) Wells were created in a custom polycarbonate mold out of 16:1 PDMS. (b) PA gels were attached to PDMS wells according to a previously published protocol. After chemical treatment that activated the silicone substrate, an unpolymerized droplet of PA was placed onto a well. A 22 mm coverslip was placed onto a droplet (left) and a uniform gel thickness was acquired by capillary action. After polymerization, the coverslip was removed (right). (c) Stretch wave form of the strain-field analysis. Colored dots indicate where representative images of the wells were taken, and they are outlined for clarity. (d) The analysis region of the uniaxially stretched gel for strain-field analysis is indicated by the transparent red square. Longitudinal (ε_xx) and lateral (ε_yy) strain fields are shown. To see this figure in color, go online.

For all experiments, wells were stretched 10% (equibiaxial or uniaxial) at 1 Hz with a saw-tooth waveform for 24 h in a temperature- and CO₂-controlled incubator. The well was then transferred to a microscope for traction force measurements in the zero-strain configuration. Measurements were completed ∼15 min after cessation of stretch to allow for equilibration of cell-generated traction forces in the undeformed configuration. Static control cells, cultured in identical wells that were not mounted on the stretch device, were analyzed in the same manner.

Tunable-stiffness traction force substrate preparation

For traction force measurements, PA hydrogels were chosen as the compliant substrate as they are optically clear, have well-defined mechanical properties, and have an easily tunable modulus. Our previously published method was utilized with minor modifications (19) to attach 70-μm-thick PA to the PDMS wells. Briefly, the PDMS surface was plasma treated, then activated with a series of organic solvents to facilitate PA bonding to the PDMS membrane. Then, 50 μL of PA solution was pipetted onto the surface and coverslipped under continuous nitrogen flow. Even with these aggressive treatments, when removing coverslips after polymerization, the low-modulus gels (<1 kPa) often stick to the glass coverslip and pull off the PDMS membranes. To facilitate more robust attachment, a thin (50 μm) PA gel with a modulus of 7.5 kPa was attached to the gel first, and then 50 μL of the low-modulus (0.6 kPa) solution was pipetted onto the high-modulus gel. Attachment was validated using different-colored fluorescent beads within the gels for verification of each layer.

Next, to allow for traction force measurements, 0.2 μm fluorescent beads were embedded into the top surface of the gel. This was accomplished by evaporating 35 μL of a 0.5% solution of 0.2 μm fluorescent beads (Life Technologies, Grand Island, NY) in 100% ethanol (Sigma-Aldrich) onto 22 mm circular glass coverslips (VWR, Randor, PA). The beaded coverslip was placed on top of the unpolymerized top gel. The gels were allowed to sit undisturbed for 1 h before the coverslips were removed.

To facilitate cell attachment, PA gels were coated with a 0.5 mg/mL solution of sulpho-SANPAH (Thermo Fisher, Waltham, MA) and reacted under ultraviolet light as per (19). A monomeric pepsin-extracted collagen solution (PurCol, Fremont, CA) of 3 mg/mL was diluted to 200 μg/mL in 1 mL of 0.02 N acetic acid and was placed on the gels. The collagen solution remained on the gels overnight at 4°C. As the process is not sterile, to avoid contamination, gels were treated with an antibiotic solution and rinsed with phosphate-buffered saline before cell seeding.

The modulus of PA gels was measured by atomic force microscopy indentation at 1 μm/s (0.06 N/m cantilever, conical tip; Asylum Research, Santa Barbara, CA). Ten measurements were made over the gel, and two gels per stiffness were analyzed. A custom MATLAB (Mathworks, Natick, MA) script was then used to extract the Young’s modulus from each curve by fitting the first 200 nm of indentation data to the Hertz model for a conical indenter. The extracted values were averaged to determine the mean modulus for each gel.

Traction force measurements

Images were captured using a Zeiss inverted microscope and a charge-coupled-device camera (Carl Zeiss, Thornwood, NY). Using a motorized stage and Axiovision software (v. 4.8.2 SP1, Carl Zeiss), three images were acquired for each cell: a phase contrast image of the cell, a fluorescent image of beads within the substrate beneath the cell, and a fluorescent image of beads in relaxed substrate when the cell was removed via 0.25% trypsin with EDTA (Gibco, Grand Island, NY) (see Fig. 2 c). Six to twenty cells per treatment/stretch condition per group were imaged in the static control and stretched groups (see Table 1 for exact numbers). The cell traction force in Table 1 is presented as a population average ± SD, not as a reflection of individual cells before and after stretch. Due to the flexible nature of the PDMS membrane, the weight of the liquid media and trypsin caused slight image distortion. A custom well holder was created with a No. 1 25 mm square coverslip (VWR) to support the thin membrane during imaging (Fig. 2 b).

Figure 2.

Overview of traction force measurements. (a) Time course schematic of experiments. All stretch experiments were at 10% magnitude (either uniaxial or biaxial stretch). At 24 h, stretch was stopped and the well was removed from the stretch device and placed in the microscope viewing dish for image acquisition. Wells were in their original conformation during the image acquisition process. (b) Side view of the image acquisition setup. (c) Representative images of traction force calculation. Phase images of the cell were acquired, as were fluorescent images of beads embedded within the PA gel. Green and red colors show the stressed and relaxed (cells trypsinized) substrate, respectively. Using a particle-tracking algorithm, bead displacements were calculated. Finally, knowing the modulus of the substrate and the bead displacements, a stress map of the substrate was created. To see this figure in color, go online.

Table 1.

Population averages of static and stretched cells in each of the experimental groups for all of the metrics examined in this study.

Cell Type	Modulus (kPa) Treatment	Stretch Type	No. of Cells	Cell Area (μm2)	Traction Force (nN)	Contractile Moment (pNm)	Angle of Principal Stress (°)	Maximal Stress (Pa)	Shape Factor	Elongation Ratio	Angle of Orientation (°)
VICs	7.5	static	6	3007 ± 1180	304 ± 152	−14.1 ± 8.3	66.8 ± 72.1	311 ± 171	0.14 ± 0.07	0.64 ± 0.22	49.0 ± 23.6
	10% equibiaxial	17	2080 ± 907	99 ± 64∗	−3.32 ± 3.26∗	49.4 ± 60.0	141 ± 69∗	0.31 ± 0.12∗	0.63 ± 0.17	57.6 ± 21.7
VICs	7.5	static	20	2525 ± 1002	519 ± 386	−15.4 ± 15.7	88.5 ± 49.6	203 ± 142	0.17 ± 0.08	0.67 ± 0.14	47.5 ± 24.1
	10% uniaxial	19	1680 ± 877∗	239 ± 292∗	−5.3 ± 5.8∗	80.0 ± 35.3	129 ± 93∗	0.17 ± 0.07	0.71 ± 0.13	77.1 ± 12.06∗
VICs	7.5−5 ng/mL TGFβ	static	6	3898 ± 1007	1130 ± 1182	−57.0 ± 52.0	99.9 ± 55.2	844 ± 643	0.10 ± 0.05	0.78 ± 0.10	44.7 ± 25.3
	10% uniaxial	14	2411 ± 984∗	510 ± 340∗	−21 ± 20∗	98.4 ± 39.2	591 ± 353	0.18 ± 0.06∗	0.75 ± 0.06	62.1 ± 24.7∗
VICs	7.5–10 μM Bleb	static	21	1625 ± 744	41 ± 37	−1.16 ± 3.7	111 ± 51.7	79 ± 68	0.09 ± 0.03	0.80 ± 0.08	52.1 ± 25.2
	10% uniaxial	10	2341 ± 641∗	63 ± 41	2.66 ± 1.97∗	102 ± 46.6	98 ± 49	0.08 ± 0.06	0.75 ± 0.13	44.9 ± 26.4∗
VICs	0.6	static	11	533 ± 271	14 ± 10	−0.10 ± 0.14	94.9 ± 54.8	92 ±65	0.75 ± 0.12	0.11 ± 0.06	40.1 ± 26.1
	10% uniaxial	9	1214 ± 489∗	43 ± 37∗	0.03 ± 0.17∗	72.7 ± 56.5	127 ± 88	0.31 ± 0.13∗	0.72 ± 0.12∗	68.6 ± 23.1∗
U20S	7.5	static	9	1847 ± 647	108 ± 63	−2.45 ± 1.88	97.0 ± 49.2	63 ± 80	0.10 ± 0.03	0.73 ± 0.55	50.1 ± 22.0
	10% uniaxial	10	1685 ± 272∗	26.7 ± 18.7∗	−0.51 ± 0.61∗	56.3 ± 30.6∗	37.1 ± 22.2∗	0.13 ± 0.04	0.55 ± 0.24	81.7 ± 7.6∗

The mean average stress for a group can be approximated within a few % error by dividing the mean traction force by the mean cell area. Data are presented as the mean ± SD and represent population averages rather than calculations from individual tracked cells. The asterisk indicates significance compared to the static control, using Student’s t-test (p < 0.05). TGFβ, TGF-β1; Bleb, blebbistatin.

A custom MATLAB program (Mathworks, Natick, MA) was created for image processing. Drift from the stage was removed by averaging the displacement of beads within small selected areas of the image corners. The images were then cropped around the cell of interest and cell boundaries were manually selected. Bead displacements were calculated using mass particle image velocimetry (code available from MATLAB Central). Substrate material properties, substrate dimensions, bead displacements, and the cell boundary were imported into ANSYS 14.0. Using finite-element analysis, stresses at the nodes under the cell were calculated. A custom MATLAB program was used to calculate total cell traction force magnitude, F, from the surface shear stresses, S, in the substrate and unit mesh area, da, using the equations:

$$S_i=\sqrt{S_{i\_xz}^2+S_{i\_yz}^2}$$ (1)

$$F_{\text{total}}=\sum _{i=1}^n\left(S_i\times da\right).$$ (2)

Only stresses from nodes that lie within the cell boundary were integrated for the traction force calculation in Eq. 2.

As a second measure of overall cell contractility, the contractile moment was computed as described previously (26).

$$\mu =M_{xx}+M_{yy},$$ (3)

where M_xx is the combination of the traction forces in the x-direction weighted by their coordinates in the x-direction and M_yy is the combination of the traction forces in the y-direction weighted by their coordinates in the y-direction. As defined, the net contractile moment is negative for a cell that is pulling inward and positive for a cell that is pushing outward. The angle of principle stress was also computed as described in (26).

Orientation and area measurements

Phase images were acquired at 24 h, and cell morphological measurements were made using ImageJ (27). To determine the alignment angle relative to the stretch axis, a line was fit through the long axis of the cell nucleus. For angles over 90°, the supplementary angle is reported due to symmetry. A minimum of 90 cells from each group were measured. Cell area, perimeter, and major and minor axis dimensions were measured. Form factor (19, 28), an indication of the number of cellular extensions, and elongation (29) were calculated using the formulas:

$$\text{form}\text{factor}=\frac{4\pi \times \text{area}}{{\text{perimeter}}^2}$$ (4)

$$\text{elongation}=\frac{\left(A_{\text{maj}}-A_{\text{min}}\right)}{\left(A_{\text{maj}}+A_{\text{min}}\right)},$$ (5)

where A_maj and A_min are the major and minor axes, respectively, of a fitted ellipse.

Statistics

All values are reported as the mean ± SD. Differences in metrics between static and stretch were compared for each treatment using two-tailed Student’s t-tests assuming equal variance. Due to dissimilarity of the baseline levels between treatment groups (e.g., TGF-β1, blebbistatin, soft, stiff), these treatment groups were not compared statistically. A p-value of <0.05 is considered to be statistically significant. Each stretched group was compared to its respective control with a Student’s t-test using Sigma-Plot version 11.0 (Systat Software, San Jose, CA). The angle was measured with respect to the stretch direction, which was defined as 0–180°.

Results

To assess the effect of long-term stretch on traction force, we first cyclically stretched the cells equibiaxially on stiff substrates to reduce effects of stretch avoidance by cell reorientation. The maximal substrate stress decreased significantly with stretch, as did the overall cell traction force and contractile moment. As our current experimental setup does not allow for tracking individual cells over time, traction force measurements reported herein are population averages of stretched cells compared to a separate, paired control group of statically cultured cells. Representative stress plots for cells from each culture condition are shown in Fig. 3. Fig. 4 shows the relative (%) changes in cell traction force, cell area, and contractile moment for each experimental group as compared to their respective controls. Fig. 5 shows cell elongation and average form factor changes with long-term cyclic stretch, and Table 1 lists the population averages as numerical values for all the aforementioned variables. After cyclic equibiaxial stretch, cells appeared to have fewer cellular extensions than respective controls and the average cell area was smaller (n.s.). The average form factor was significantly larger in stretched cells, indicating that cyclic equibiaxial stretch induced cells to adopt a more rounded phenotype (Fig. 5). As expected, there was no preferential angle of orientation for VICs under cyclic equibiaxial stretch; both static culture and stretched cells exhibited a random orientation (Fig. 6).

Figure 3.

Overview of experiments. VICs were cultured on 7.5 kPa substrates unless otherwise indicated. Pretreatments are indicated on the representative static stress maps. Arrows indicate whether uniaxial (straight arrows) or biaxial stretch (bisecting arrows) was used for a given experiment. For cells cultured under high pre-stress conditions, mean traction force, mean contractile moment, mean cell area, and mean maximal substrate stress all decreased with stretch. The mean form factor increased for both equibiaxially stretched cells and TGF-β1-pretreated cells when uniaxially stretched indicating a decrease in extent of cell extensions. The opposite was true for cells under low pre-stress conditions: mean traction force, mean maximal substrate stress, and mean cell area all increased when cells were stretched compared to static control cells. Scale bar, 50 μm. To see this figure in color, go online.

Figure 4.

Cell traction force and area change with long-term cyclic stretch. (a) The percentage change in cell traction force was normalized to respective controls for each treatment group. Error bars reflect raw stretch traction force values divided by the mean of the controls then multiplied by 100%. Cell traction force decreases with biaxial stretch, uniaxial stretch, and uniaxial stretch with TGF-β1 pretreatment (TGFβ) on a 7.5 kPa substrate. Cell traction force increases with stretch slightly when cells are pretreated with 10 μM blebbistatin (Blebb) before stretch and with cells that are stretched on a soft substrate. (b) The percentage change in cell area was normalized to respective controls for each treatment group. Error bars reflect the raw stretch cell area values divided by the mean of the controls. Cell area decreases with biaxial stretch, uniaxial stretch, and uniaxial stretch with TGF-β1 pretreatment on a 7.5 kPa substrate. Cell area increases with stretch when cells are pretreated with 10 μM blebbistatin and subsequently stretched and with cells that are stretched on a soft substrate. (c) The percentage change in contractile moment followed the same trends as that of traction force, with the average contractile moment decreasing significantly for cells under high pre-stress. For cells under low pre-stress, there was an increase in the average contractile moment with stretch. The asterisk indicates significance when compared via Student’s t-test to respective static controls (p < 0.05). To see this figure in color, go online.

Figure 5.

Changes in cell shape and elongation with long-term cyclic stretch. (a) Elongation ratio was unaffected by stretch under most treatment conditions, except in the case of culture on a soft (0.6 kPa) gel. (b) Form factor increased with biaxial stretch and uniaxial stretch for TGF-β1-pretreated cells (TGFβ) as the cell perimeter decreased with depolymerization of many cellular extensions in the direction of stretch. Form factor decreased with stretch on a soft substrate as cells elongated from a circular morphology. ^∗p < 0.05. To see this figure in color, go online.

Figure 6.

Angle of orientation for static and stretched cells represented as 0–90° histograms. The angle of orientation was measured with respect to the stretch direction (0–180°). For cells that had an orientation angle above 90°, the supplementary angle was reported. Cells that oriented perpendicular to the stretch had an angle of orientation of 90°. (a) biaxially stretched cells compared to static controls (b) uniaxially stretched cells compared to static controls (c) TGF-β1 pre-treated cells stretched uniaxially compared to TGF-β1 pre-treated static controls (d) Blebbistatin pre-treated cells stretched uniaxially compared to Blebbistatin static controls (e) Cells cultured on 0.6 kPa substrates stretched uniaxially compared to static control cells cultured on 0.6 kPa. For all panels, black bars are static cells, gray bars are stretched cells.

We then investigated whether VICs would maintain their level of traction force when allowed to reorient away from the direction of stretch (the membrane is essentially static perpendicular to the stretch direction). Despite orientation to the direction of zero stretch, cyclic uniaxial stretch caused a significant decrease in maximal substrate stress and traction force generated by the cells. The decrease in traction relative to matched static controls was somewhat lower for uniaxial stretch than for equibiaxial stretch (43% and 67% decrease, respectively). The average contractile moment also decreased significantly with stretch, though it was slightly less than for equibiaxial stretch (65% and 76%, respectively). The average cell area decreased significantly with stretch, but no significant changes in average elongation ratio or form factor were measured.

In the presence of TGF-β1, VICs cultured on stiff substrates differentiate into myofibroblasts with α-SMA-positive actin stress fibers (25) that can withstand high forces; thus, a 5-day pretreatment with TGF-β1 was used to increase the cell pre-stress. VICs cultured under static conditions with TGF-β1 pretreatment generated two and a half times the maximal substrate stress and traction force of the untreated static controls (see Table 1 for numerical values and Figs. S7 and S8 for the distributions). These cells also had a greater average cell area and appeared to have more cell extensions under static conditions when compared to untreated VICs under static conditions. Despite the ability of the actin cytoskeleton to generate higher tension with TGF-β1 treatment, these cells exhibited a significant decrease in the cell traction force after 24 h of uniaxial stretch, similar to the untreated cells (Fig. 4). Areas of locally high stress concentrated under the cellular extensions in statically cultured cells were not apparent with stretched cells, and there was a decrease in maximal substrate stress after stretch (n.s.). The average contractile moment decreased significantly with stretch. The average cell area also decreased significantly with stretch, likely due to the decreased number of cellular extensions, as indicated by the significant increase in form factor (Fig. 5). The average elongation ratio of the cells was not significantly altered by stretch. Like nontreated VICs, TGF-β1 pretreated cells were oriented perpendicular to the stretch direction, and there was no statistical difference in orientation between untreated VICs and VICs pretreated with TGF-β1 after 24 h of cyclic uniaxial stretch.

After our observation that the cells reduced their traction force in response to long-term stretch, we sought to determine whether stretch could increase cytoskeletal tension for cells below their optimal cytoskeletal tension level. Cell pre-stress was decreased in two ways. First, blebbistatin, a myosin II inhibitor, was used as a chemical means of decreasing the cytoskeletal tension before stretch. VICs treated with blebbistatin generated ∼25% of the traction force of static, untreated VICs. Although not significant, the maximal substrate stress and cell traction force increased with cyclic uniaxial stretch. The average contractile moment increased in magnitude significantly with stretch (Fig. 4). Additionally, the contractile moment for some control blebbistatin-treated VICs and all stretched blebbistatin-treated VICs was positive. The blebbistatin-treated VICs had outward displacement vectors near the cell periphery. This nonintuitive finding may indicate outward forces due to polymerization of actin stress fibers and/or nonrecoverable stretch in the stress fibers that push outward upon cessation of cyclic stretch. In these cells, the myosin II motors were inhibited and unable to provide the necessary inward contractile force (see representative images in the Supporting Material). VICs treated with blebbistatin had a smaller spread area, though they remained elongated, exhibiting a very thin spindle shape. The average cell area increased significantly with stretch (Fig. 4), whereas the average form factor and average elongation ratio (Fig. 5) were not statistically different for static and stretched blebbistatin-treated VICs. Blebbistatin-treated cells had a slight, yet significant, orientation parallel to the stretch direction.

Second, as a mechanical means of reducing pre-stress, VICs were cultured on low-modulus (0.6 kPa) PA gels (termed “soft” herein in accordance with mechanobiology literature). Cells statically cultured on these soft substrates had a circular morphology with low traction force and low maximal stress values (Fig. 3). With uniaxial stretch, the traction force and cell area increased significantly (Fig. 4) and the maximal stress increased slightly (n.s.). The mean contractile moment was positive for cells cultured on soft substrates with stretch compared to static controls, which had a negative contractile moment (Fig. 4). The positive contractile moment for the stretched cells on soft gels may indicate that the cells are spreading outward, as evidenced by the increase in mean cell area, due to dynamic stimulation. VICs became elongated with a significant increase in the average elongation ratio. The average form factor significantly decreased, and average elongation significantly increased (Fig. 5), indicating a transition to a more polarized cell phenotype. Similar to VICs cultured on stiff substrates, VICs on soft substrates had a statistically significant orientation away from the direction of stretch. There was a uniform distribution of angles for static cells, whereas the average angle for stretched cells was 70° relative to the axis of stretch (Fig. 6). As orientation away from stretch on a soft substrate contradicted previous findings by Kaunas and colleagues with U2OS cells (human osteosarcoma) cultured on soft collagen substrates (30), an additional study with U2OS cells on soft substrates was done to compare orientation responses on soft (0.6 kPa) and very stiff (1 MPa) substrates. Similar to VICs, U2OS cells reoriented away from stretch when cultured on both the soft and stiff substrates (Fig. S4). After cyclic uniaxial stretch, U2OS cells on soft substrates were equivalent in size to stretched U2OS cells on 1 MPa substrates. An additional experiment was conducted on U2OS cells to ascertain how cells integrate stretch signals over time in the PDMS culture wells. U2OS cells were subjected to three patterns of stretch: 10% uniaxial stretch, 10% equibiaxial stretch, and alternating uniaxial stretch (10% uniaxial stretch in the x-direction followed by 10% uniaxial stretch in the y-direction). The cytoskeletons of groups with alternating uniaxial stretch appeared diamond-like in shape, with peaks at ∼45° from the direction of stretch (see Fig. S5).

Discussion

In this study, we tested the hypothesis that changes in cell-generated traction forces in response to long-term cyclic stretch are dependent upon the pre-stress in the cell. We developed and implemented a system capable of measuring traction force generated by cells cultured on substrates with a tunable elastic modulus. After cyclic stretch for 24 h, decreases in traction force in cells with high pre-stress (for this particular cell phenotype) and increases in traction force for cells with low pre-stress were observed. This behavior, along with reorienting away from uniaxial stretch, is consistent with maintaining homeostatic cytoskeletal tension during dynamic stretch, i.e., if the cytoskeletal tension is at a maximal level in static culture, it is reduced to accommodate dynamic stretch by retraction of cellular extensions and/or reorientation, whereas if it is below the maximal level (e.g., on a soft substrate), it is increased toward its optimal level. The method and data presented herein can be used to validate future mathematical and phenomenological models of cell reorientation and cell shape changes with stretch based on cytoskeletal tension.

Effects of stretch on cytoskeletal structure and mechanics

Rapid cytoskeletal remodeling with stretch has been observed with metrics other than traction force, including immunofluorescence, storage modulus, and cell spread area. Early work by Pender and McCulloch showed rapid changes in the cytoskeleton with stretch using immunofluorescence; F-actin in gingival fibroblasts was reduced by 50% at 10 s after stretch, but was increased by >100% at 50 s after stretch (31). Further, Costa et al. observed that release of stretch caused human aortic endothelial cells cultured on prestretched silicone membrane to depolymerize stress fibers in the direction of stretch release when release rates were >5% stretch per second, and rapid remodeling was observed within 60 s of the release of stretch (32). In response to transient stretch-unstretch, human bladder smooth muscle cells rapidly disassemble F-actin (16). In parallel to changes in the actin cytoskeleton, cell mechanical properties also change with stretch. Fredberg and colleagues demonstrated that human airway smooth muscle cells abruptly soften (as measured by G′ decrease) with transient equibiaxial stretch, with cell stiffness recovery over ∼10 min (21). This behavior was interpreted as “fluidization” and “resolidification” of the cytoskeleton (33). Throm et al. demonstrated cytoskeletal remodeling (via an increase in cell spread area) of VICs when they were cultured on 0.6 kPa substrates and stretched equibiaxially at 1 Hz for 6 h (19). Although traction force was not measured in the studies cited here, these works support rapid cytoskeletal remodeling with stretch, which would likely lead to changes in cellular forces.

Effect of long- and short-term stretch on cell traction force

Before this study, changes in cell traction force have been measured with stretch on short timescales (<1 h) using stretch and release as well as stretch and hold experimental designs. Navajas and colleagues (34) rapidly stretched epithelial cells cultured on collagen gels. They observed an initial passive increase in traction force with stretch (with 5.5% and 11% stretch); with release of stretch, there was a significant decrease in traction with recovery over 10 min. Fredberg and colleagues applied a transient homogeneous strain field and, in contrast to Gavara, found that the contractile moment decreased rapidly after stretch and recovered over a period of ∼10 min (21). Fu and colleagues applied an instantaneous step stretch (without release) to smooth muscle cells cultured on a micropillar device and observed the cell-generated forces for 1 h (17). Cells increased their traction force within the first half-hour of observation and decreased to below baseline after 60 min. Although the time course of these studies is not long enough to observe cytoskeletal adaptation to stretch involving protein incorporation (such as αSMA, which has been reported to take 72 h (35)), these studies shed light on the passive transfer of load within the cytoskeleton and the ways in which the rapid reorganization of the existing cytoskeletal proteins translates into the evolution of localized active cell-generated traction.

Several theories have been developed that predict that cell tension is either maintained or increased with stretch (36); however, no combination studies of simultaneous traction force and stretch have reached long enough timescales to allow for changes in protein expression and differentiation, which many groups report with stretch (37). Many research groups have observed reorientation with stretch; however, substrates outside of the physiologically relevant range of stiffness are generally utilized (38, 39, 40). To date, the only study of simultaneous cell reorientation and traction force has been done with very slow tidal stretches (26). Human embryonic vascular endothelial cells were subjected to a trapezoidal waveform (∼1 s loading, 3 s hold, and ∼1 s unloading) every 49 s for 2 h. There was an initial drop in cell tension ∼50 min into the experiment. Once cells reoriented perpendicular to the stretch, the authors report that cells recovered their full contractile moment, in contrast to the depressed traction exerted by the cells on our stiff substrates after 24 h of stretch at 1 Hz. The difference in frequency of stretch or duration of dynamic culture may be responsible for the differences: repeated cyclic stretching for an extended duration may result in changes in protein levels so that VICs become less contractile, which has been shown with pulmonary fibroblasts (37) as well as vascular smooth muscle cells (M. Rolle, personal communication).

Changes in traction force with stretch are heterogeneous within and between cells. Fu and colleagues observed the largest changes in force along the cell periphery (17, 22). We also found that the stresses under cellular extensions at the periphery of the cells cultured on stiff substrates showed the most marked decrease after stretch. In contrast, on very soft substrates, cells were rounded in static culture and elongated and created areas of high substrate stress under the cellular extensions. The ways in which the peripheral tension fluctuations due to external loading are translated into a global cell response, such as cell reorientation and cell shape change, are still unknown. In terms of cell-to-cell differences in the response to stretch, Gavara et al. reported highly variable changes in traction between cells upon release of equibiaxial stretch (34). The authors attributed these differences to biological variations in the initial traction force (i.e., pre-stress). Epithelial cells with low initial traction force had greater recovery of traction force and recovered to baseline levels, but cells with high initial traction force did not recover their ability to generate traction on the collagen substrate over the 10 min time span measured. Matsumoto and colleagues found biological variability in how smooth muscle cells respond to two cycles of slow stretch (22). Cells either increased their traction force after the first cycle of stretch (termed an active myogenic response) or decreased their traction force after the first cycle of stretch (a passive response). The baseline traction force (i.e., pre-stress) was similar between groups, and the authors did not discuss a mechanism that would account for the differences in a cell responding actively or passively to stretch. We also saw variability in traction force with stretch, with the greatest differences on the soft (0.6 kPa) substrates. Although the average traction force was greater for the stretched cells, several of the stretched cells on soft substrates had traction forces lower than the average for static control cells. These cells were as elongated as those that generated higher traction.

Changes in traction force and spreading with stretch depend upon initial pre-stress

Our combined results indicate that the traction force response to long-term cyclic stretch is dependent upon the cell pre-stress level. In the presence of TGF-β1, the cell incorporates α-SMA into the actin stress fibers and increases contractile activity. The cell is thus able to generate more force (>1000 nN); however, when cyclically stretched beyond the homeostatic tension level for long duration, the cytoskeleton is remodeled by retraction of cellular extensions and/or reorientation away from stretch, resulting in a decrease in traction force. The actin cytoskeleton of U2OS cells appears to have a lower homeostatic tension level compared to VIC cells cultured on substrates of the same stiffness (7.5 kPa) (Fig. S3). However, U2OS cells reorient similarly to the VICs and decrease traction force in response to long-term cyclic stretch, indicating that stretch causes the cytoskelton to exceed it's homeostatic tension despite starting at a relatively low pre-stress (∼100 nN) compared to VICs (∼500 nN). In contrast to cells on stiff substrates, VICs cultured on soft substrates and those with inhibited myosin II motor activity generate submaximal cytoskeletal tension when cultured statically. The traction force increased in these cells with long-term cyclic stretch, yet not to the level seen in cell culture on stiff substrates (either static or stretched). Sheetz and colleagues observed similar spreading of fibroblasts cultured on soft micropillars that were cyclically stretched equibiaxially (20).

In contrast, Krishnan et al. found that changes in cell traction in response to homogeneous stretch are not dependent upon substrate modulus (and thus initial cell pre-stress level) (21). Their reported results indicate that in response to a transient equibiaxial stretch, the contractile moment is decreased to 20% of its initial value, then recovers to initial levels within 3 min regardless of the modulus of the substrate (1, 4, or 6.2 kPa). The soft substrate utilized in our experiments has a somewhat lower modulus (0.6 kPa), resulting in a rounded cell morphology; this distinct morphological change was not reported by Krishnan et al.

Reorientation of initially spread cells and the effect of pre-stress

Both VICs (Fig. 6) and U2OS cells (see Fig. S5) cultured on stiff substrates orientated away from the direction of uniaxial stretch, in agreement with the literature for a variety of cell types (41, 42, 43, 44, 45). With equibiaxial stretch, the decrease in spread area was significant for both VICs and U2OS cells and, as expected, neither cell type reoriented. Alternating uniaxial stretch resulted in diamond-shaped cells with stress fibers at ∼±45° from the alternating directions of stretch. These data provide additional support for the theory that stress fibers exposed to high-magnitude stretch depolymerize and that those with low tension are reinforced with stretch (Fig. S5). To our knowledge, this is the first time that an alternating uniaxial stretch regimen has been utilized.

We speculated that the rate of reorientation with cyclic stretch is related to the amount of cell pre-stress. Specifically, we predicted that cells with lower traction force and associated weaker substrate adhesion would reorient more readily and those with higher traction forces would do so to a lesser extent. Differences in reorientation were found between the two cell types we tested. U2OS cells, which on average generate less than one-third of the overall traction force of VICs (Table 1), were shown to reorient to a greater degree than VICs over 24 h of uniaxial stretch (Figs. 6 and S3). Like VICs, U2OS cells had statistically lower traction force and statistically smaller area with stretch (Table 1). Differences in reorientation rate between cell types have been reported in the literature, but the relationship with traction force has not been measured simultaneously (45). In contrast, TGF-β1-treated VICs (myofibroblasts), which generate more than double the traction force of nontreated VICs and have mature focal complexes (46), reoriented to an extent similar to that for the untreated VICs and had an equivalent relative decrease in cell area with cyclic stretch. Thus, it appears that degree of reorientation is more strongly associated with cell type than pre-stress. One possible explanation as to why TGF-β1-pretreated VICs were still able to reorient is that stretch may also act to modulate phenotype in mechanosensitive cells, causing them to revert to a less contractile phenotype (37) and allowing them to reorient; however, more experimentation is needed to test this hypothesis.

Culture in the presence of blebbistatin, a myosin II motor inhibitor, resulted in cells with lower pre-stress. VICs that had been treated with blebbistatin reoriented parallel to stretch. This observation is in agreement with comparable studies in the literature with blebbistatin and stretch on PDMS substrates (29). It has been proposed that myosin II cross-bridge cycling helps maintain stress fiber tension, and stress fibers depolymerize when they are not at their homeostatic level of tension (47). However, when myosin II is inhibited, the stress fibers are below their ideal tension level. With uniaxial stretch, stretch fiber tension is passively increased by the dynamic stimulus in the direction of stretch. Active tension generation by the cell is decreased with myosin II inhibited, but stress fibers are able to form, likely by using the passive tension generation of the stretch. This would explain why the blebbistatin-treated cells orient parallel, and not perpendicular, to stretch. As blebbistatin can affect some myosinII-independent processes (48), future work to determine the contribution of myosin to stress fiber tension and cell reorientation should also include myosin light chain inhibitors ML-7 and ML-9. Interestingly, approximately half of the blebbistatin-treated control cells and all of the blebbistatin-treated stretched cells had a positive contractile moment, indicating outward (away from the cell body) forces. Outward pushing forces have been reported by Sheetz and colleagues for cells that had been stretched when the membrane returned to its original conformation (20). We believe that is the first reported finding of outward pushing forces in blebbistatin-treated cells and in any nonstretched cells. A possible explanation for the outward displacements by these cells is the outward polymerization of actin stress fibers while the myosin motors are inhibited and unable to provide an inward (contractile) force. These contractile-moment values are above the level of noise in the measurement, and U2OS cells had similar measured cell traction force magnitudes but negative contractile moments (as indicated by the inward displacements toward the cell body).

Reorientation and spreading of initially rounded cells with long-term stretch

Both VICs and U2OS cells reoriented away from the direction of stretch when cultured on soft substrates. Results of similar experiments reported by Kaunas and colleagues showed that cells reorient parallel to the direction of stretch when cultured on soft collagen gels (10% uniaxial at 1 Hz for 3 h) (30). This finding raises questions about the time course of cell spreading and reorientation on soft substrates, whether the cells first spread out and then reorient, and what mechanism allows cells to maintain their area on soft substrates once they have reoriented (as in theory, the angle of zero strain should have no mechanical stimulation). An additional study of cell reorientation on soft substrates showed no reorientation with 16 h of stretch; however, that study was done at mHz frequencies and low strain amplitude (3%). Cell reorientation has been shown to be dependent on frequency (30, 49).

Many models have been developed to examine the mechanisms of cell reorientation and stress fiber dynamics with cyclic stretch (49, 50, 51). Few models can predict the reorientation behavior of cells cultured on soft (0.6 kPa) substrates. The Wei model (51), a polymerization-based model, predicts cell reorientation if the cell is not elongated (as is the case with the soft substrate) but does not include a term for matrix stiffness and is not predictive of an already elongated cell (as is the case for cells cultured on the 7.5 kPa gel). Stamenovic et al. developed a model incorporating the phenomena of cytoskeletal fluidization and resolidification in response to cyclic stretching (36). This model accurately predicts the increase in cell traction force and cell lengthening we observed with stretch on soft 0.6 kPa substrates. To our knowledge, there is no model that can simultaneously predict the pre-stress-dependent behaviors of traction force and elongation with reorientation in the cells cultured on soft and stiff substrates.

Limitations and future studies

As traction force, reorientation, and cell morphology were only assessed after 24 h of stretch, we cannot comment on the time course of changes in these metrics during stretch or after cessation of stretch. In the future, we wish to track cells over time on both the soft and stiff substrates presented here. These studies would help us better address how traction force changes as a measure of pre-stress, cell orientation with respect to the stretch direction, and total strain over time. Additionally, they will allow us to study the timecourse of cell elongation and reorientation on the very soft substrates. We also did not measure changes in protein expression (e.g., α-SMA), which may explain changes in traction force with long-term stretch. Although it was not studied in VICs, reversal of the myofibroblastic phenotype in the presence of stretch has been shown with lung myofibroblasts (52). We believe that traction force can be used as a powerful functional indicator of cell phenotype that can supplement biochemical metrics (protein synthesis and gene expression).

Conclusions

Here we report, to our knowledge, the first traction force measurements after 24 h of cyclic stretch. We observe that the traction force response to stretch is dependent upon initial cell pre-stress. The cell traction force decreased with long-term cyclic stretch when the cell pre-stress levels were high, whereas traction force increased when cell pre-stress levels were low. The decrease in traction force with stretch was larger in the case of equibiaxial stretch than uniaxial stretch, as cells were unable to undergo reorientation. Cell pre-stress levels were further increased with culture in TGF-β1. However, TGF-β1-treated cells still decreased traction force after long-term stretch. Decreasing pre-stress levels either chemically, with blebbistatin, or by culture on soft substrates resulted in increasing traction force with stretch. We conclude that when cell pre-stress is low, mechanical stimulation may serve as a means to increase cytoskeletal tension. The methods and data from this study will make it possible to account for cell pre-stress in the development and validation of models of cell reorientation and cell spreading with stretch. To our knowledge, this study is the first to measure cell reorientation and traction force with stretch at frequencies and stiffness levels relevant to the cardiovascular system.

Author Contributions

H.C. designed experiments, performed research, and analyzed data from the traction force experiments presented in the article; she also wrote the article and the Supporting Material section. M.M. designed research, performed research, and analyzed the data for Fig. 5 and contributed to programming of the stretch device. N.D. developed prototypes for the well mold, designed research, performed research, and analyzed the data for Fig. S4. J.F. contributed analysis techniques for strain-field verification and wrote a section of the Supporting Material. Q.W. developed analysis techniques, helped with experimental design, and oversaw data analysis for the manuscript. K.B. oversaw development of the stretch device, well, and validation, designed traction force experiments, oversaw data analysis, and wrote the manuscript.

Editor: Christopher Yip.