Cellular responses to beating hydrogels to investigate mechanotransduction

Cells feel the forces exerted on them by the surrounding extracellular matrix (ECM) environment and respond to them. While many cell fate processes are dictated by these forces, which are highly synchronized in space and time, abnormal force transduction is implicated in the progression of many diseases (muscular dystrophy, cancer). However, material platforms that enable transient, cyclic forces in vitro to recreate an in vivo-like scenario remain a challenge. Here, we report a hydrogel system that rapidly beats (actuates) with spatio-temporal control using a near infra-red light trigger. Small, user-defined mechanical forces (~nN) are exerted on cells growing on the hydrogel surface at frequencies up to 10 Hz, revealing insights into the effect of actuation on cell migration and the kinetics of reversible nuclear translocation of the mechanosensor protein myocardin related transcription factor A, depending on the actuation amplitude, duration and frequency.

The VPTT of crosslinked 100 % NIPAM gels tethered to a glass coverslip (as used in the present study) is measured in water and RPMI cell culture medium to be ~ 28.8 °C and ~ 25.0 °C, respectively, demonstrating a significant decrease upon crosslinking and in the presence of media (LCST of non-crosslinked NIPAM in water is ~ 32 °C) (c) Diameter of free-standing gel discs, prepared with different copolymer ratios in a mould with a diameter of 30 μm and a height of 5 μm, at different temperatures in cell culture medium. (d) The swelling ratio of the discs obtained from the measurements in (c). (e) A linear dependence of the VPTT with the NEAM % is observed for the free-standing discs, where a 65/35 mole % NIPAM/NEAM shows a VPTT of ~ 36 °C. This is similar to the linear dependence obtained for tethered gels (used in the present study), where 60/40 mole % NIPAM/NEAM gels have a VPTT of ~ 36 °C, shown in Figure 1c  Quantification of the AuNRs using inductive coupled plasma-optical emission spectroscopy (ICP-OES) in the gel after preparation and after swelling, and in the swelling medium (water). No AuNRs are observed in the medium after 3 days, demonstrating that the AuNRs do not leach out from the gel. Error bars represent standard deviation, n = 3. Figure 6. Variation of the microstructure ridge width of 60/40 NIPAM/NEAM gels without AuNRs and with AuNRs (OD 100) in thermal equilibrium at different temperatures shows that the VPTT of the gels is not affected due to the addition of AuNRs. Error bars represent standard deviation, n = 3. Results of the simulation obtained for a time independent photothermal equilibrium that is achieved in response to a laser pulse of 340 mW, 100 ms laser ON time and 1 Hz frequency, depicting the variation of temperature at the surface of the gel. The laser spot is shown with the overlaid pink rectangle (d) A 3D distribution of the heat dissipation from the irradiated region of the gel. The asymmetric nature of the distribution in the z-direction stems from the difference in thermal conductivities of glass and medium (i.e. water). Figure 9. (a) A representative MTS proliferation assay for L929 cells grown on TCPS after 24 h at different temperatures, n = 3. (b,c) The signal intensity of (b) heat shock protein Hsp 70 (n = 2, N ≥ 93 cells) and (c) Elf (that stains against stress granules and P-bodies, n = 2, N ≥ 155 cells) is shown in the box plots for actuated and non-actuated cells. (d,e) Representative images of cells stained against Hsp 70 (d) when actuated and (e) non-actuated, (f,g) against Elf that stains stress granules and P bodies for (f) actuated cells and (g) non-actuated cells. The non-actuation images are acquired from cells grown on the non-irradiated portion of the gel. In the box plots, the interquartile range (IQR) between the first and the third quartiles is indicated by the box, while whiskers denote 1.5 IQR. The hollow square, the horizontal line, and the filled dots represent the average, the median, and the outliers, respectively. Error bars represent standard deviation. *, **, *** are determined using one way ANOVA or Welch test, depending on the homogeneity of variances, and represent statistical significance at p < 0.05, 0.01 and 0.001, respectively. Scale bar = 20 µm. Schematic of the sample chamber designed for actuation. The gel, attached to a small cover glass, is glued on the sides to another larger glass coverslip and a PDMS ring is put around it to contain the media. On the opposite side of the glass coverslip, a grid with markings is glued to keep track of the actuated area. (*illustrations not to scale). Error bars represent standard deviation, n = 3. Figure 11. (a) A confocal z-stack of a gel with fibronectin coating. Sulfo-SANPAH is used as a bi-functional crosslinker to covalently link fibronectin to the gel. Fibronectin is stained red, while the polymer is stained green. (b-c) The effect of the (b) fibronectin concentration and (c) temperature of fibronectin incubation on L929 cell proliferation. Based on these results, a fibronectin concentration of 265 nM, incubated at 34 °C, is further used for this study. (d) When gels with an AuNR concentration of 0.004 vol % are prepared, the MTS assay does not show a toxic effect as AuNRs do not leach out of the gels. (e) Cells seeded on the gels are exposed to a pulsed laser (340 mW, 1 Hz, 100 ms laser ON time) for a period of 22 h. No significant cell death is observed in both the actuated and non-actuated regions of the gel. Scale bar = 100 µm. Error bars represent standard deviation, n = 3. In the box plots, the interquartile range (IQR) between the first and the third quartiles is indicated by the box, while whiskers denote 1.5 IQR. The hollow square, the horizontal line, and the filled dots represent the average, the median, and the outliers, respectively. On the left of the box plot, all data points are shown, the normal distribution curve serves to guide the eye. *, **, *** are determined using one way ANOVA or Welch test, depending on the homogeneity of variances, and represent statistical significance at p < 0.05, 0.01 and 0.001, respectively.  Figure 14. Immunofluorescent staining for MRTFA in cells grown on (a,d) 60/40 NIPAM/NEAM gels with AuNRs which show photothermal heating with mechanical deformation (actuation), (b,e) 0/100 NIPAM/NEAM gels with AuNRs, which show photothermal heating without mechanical deformation, (c,f) and 60/40 NIPAM/NEAM gels without AuNRs, which do not display photothermal heating or mechanical deformation. The top row shows the region of the gels that is pulsed with NIR light for 12 h (340 mW power, 1 Hz, 100 ms laser ON time), while the bottom row shows a region of the gel that is not exposed to NIR light. The actuated region demonstrates translocation of MRTFA from the cytoplasm to the nucleus, while this is neither observed on the control gels nor in the regions of the gel that do not actuate. Scale bar = 50 µm. (g) The nuclear MRTFA (%) is measured and represented by box plots where the interquartile range (IQR) between the first and the third quartiles is indicated by the box, while whiskers denote 1.5 IQR. The hollow square, the horizontal line, and the filled dots represent the average, the median, and the outliers, respectively. On the left of the box plot, all data points are shown, the normal distribution curve serves to guide the eye. (n ≥ 2, N ≥ 18 cells). *, **, *** are determined using one way ANOVA or Welch test, depending on the homogeneity of variances, and represent statistical significance at p < 0.05, 0.01 and 0.001, respectively. The local modulus of the confined gels tethered to a glass coverslip is measured using Atomic Force Microscopy (AFM), n = 2. The interquartile range (IQR) between the first and the third quartiles is indicated by the box, while whiskers denote 1.5 IQR. The hollow square, the horizontal line, and the filled dots represent the average, the median, and the outliers, respectively. On the left of the box plot, all data points are shown, the normal distribution curve serves to guide the eye. Error bars represent standard deviation. *, **, *** are determined using one way ANOVA or Welch test, depending on the homogeneity of variances, and represent statistical significance at p < 0.05, 0.01 and 0.001, respectively.

Supplementary Method 2. Dispersion and quantification of AuNRs
Since AuNRs show a surface plasmon resonance, this property is used to quantify the amount of AuNRs in the solution. The optical density of the gel precursor solution, as well as of the hydrogels, is measured on a JASCO UV-Vis spectrophotometer from 1100 to 400 nm at 1000 nm/min. Quartz cuvettes with a path length of 1 mm are used to measure the absorbance.
A gel precursor solution, which does not have AuNRs, is used as a reference solution.
Hydrogels without AuNRs are used as reference for measuring the spectra of the gels. The number density of the AuNRs in the solution is calculated according to the following equation:

Supplementary Method 4. Functionalization of the gel surface
The gel surface is covalently functionalized with fibronectin to promote cell adhesion using Sulfo-SANPAH mediated succinimide crosslinking. This protocol is adapted from 5,6 . Persistence (L): The distance that the cell covers without changing its direction. To calculate this, the net displacement of the cell from the initial to the final position is divided by the sum of the total distance travelled to reach the final position. Hence, the persistence is the end-toend distance divided by the contour length.

= Supplementary Equation 6
Projected contour length ( ′ ): The actuation is directionally controlled using a micro-patterned substrate. Therefore, the topography of the gel can also influence cell motility. To calculate the movement of cells in the direction of substrate micro-patterns, the projected contour length is calculated as the distance travelled by the cells in a direction parallel to the topography.

Supplementary Equation 7
To interpret the trajectory plots of cells (erratic or directed), the classical method of analysis  Table 1). By measuring their diameter at varying temperatures in media, a VPTT of ~ 38 °C is observed for free-standing (unconfined) 60/40 % NIPAM/NEAM hydrogel discs, which is slightly higher than the hydrogels tethered to a glass slide. This may be the result of confinement, as previously reported in literature 13 In the collapsed state, the ridges (surface topography) of the hydrogel are rectangular (~ 25 µm width) and resemble the silicon wafer pattern, while in the swollen state, the ridges appear concave with ~ 32 µm width and the grooves acquire a reduced width of ~ 18 µm (Figure 1 d).
The thickness of the dry gel is 3.2 ± 0.7 μm (measured at the ridge). When performing volume phase transitions by successive heating and cooling cycles, there is no shift in the VPTT and thus no hysteresis is observed (data not shown), in contrast to pNIPAM in solution 14 .
Monotonic compression of free-standing hydrogel discs at different temperatures demonstrates a rise in elastic modulus from 120 kPa at 33 °C to 320 kPa at 42 °C

Supplementary Note 2. Gel surface topography
When preparing and confining flat gel films of a similar composition and thickness to a rigid substrate, creases are observed. However, in the case of the microstructured gels used here, the surface density and location of creases depend on the filling fraction of ridges relative to the microstructure period. In addition, a larger fractional change is measured in ridge width compared to the ridge height (Δh/Δw ~ 0.3). This is in contrast to a confined flat film, for which Δh/Δw ≥ 1 is expected.
Together with the absence of creases when the ridge filling fraction is sufficiently high, this may indicate weak confinement of the ridges relative to the bulk of the gel below the topography.
Furthermore, the swelling ratio of the ridges is compared with the swelling ratio of freestanding discs, prepared from the same 60/40 NIPAM/NEAM composition and a diameter of 30 μm. This reveals a lateral swelling ratio of ~ 1.8 in case of the ridges, which is only marginally larger than the change in diameter of the free swelling discs (~ 1.6) (Supplementary Figure 3). This shows that although the ridges are connected to the surface of a tethered hydrogel film, their swelling and shrinking ability in response to temperature changes is not compromised.

Supplementary Note 3. Photothermal actuation
Considering the length scales involved in this photothermal hydrogel system, the AuNRs are ~ 140 times smaller in dimension than a typical cell (assumed to have a size of ~ 10 µm).
Moreover, the generated heat can be controlled using a pulsed laser, which enables tunable gel actuation, depending on the kinetics of the gel's phase transition. In response to the incident laser, the AuNRs heat up the surrounding medium and gel within nanoseconds to reach a steady  Thermal confinement is undesirable as it can lead to very high temperatures in the exposed area of the gel and impair actuation 17 . Due to the large heat sink, the temperature in the medium or in the non-exposed portion of the gel does not increase, leading to local actuation only in the target area. The laser pulses are selected to avoid overheating of the sample and keep the mean temperature in the actuated area always below 38.5 °C. However, since the laser ON (20 or 100 ms) and OFF time (80-900 ms) are shorter than the characteristic time required for shrinking and swelling of the gel (τcharac, 23 s), respectively, the gel is modulated around a steady state during actuation but does never fully shrinks or swells (Movie 5).

Supplementary Note 4. Estimation of gel temperature
To enable visualization of the initial contraction when the laser is turned on and the final swelling when the pulsing is stopped, the laser is pulsed between 60 and 120 s (100 ms pulses, 340 mW). Figure 2 shows that there is a spatial distribution of temperature changes, confined to the laser spot. The local total increase in temperature, ∆ , can be subdivided in to two parts: the overall increase in the steady state temperature compared to the initial state when the laser is OFF ( ) and the additional fluctuations in temperature when the laser is ON From the mean temperature measurements, the estimated values are ~ 0.6 °C and ~ 0.3 °C leading to a net ∆ of ~ 0.9 °C. Besides the average temperature, we also measure the maximum temperature during actuation (Supplementary Figure 7). This is the maximum temperature recorded in the gel and does not reflect the temperature of the entire gel.
The maximum total temperature increase ∆Tmax in the irradiated portion of the gel is approximately 3.0 °C at equilibrium, with ~ 1.7 °C. The maximum temperature (due to positive oscillations) around the new equilibrium (dTmax) isaround ~ 1.5 °C, keeping the maximum temperature experienced by the cells below 39.0 °C and thus below the temperature of heat shock for cells 18 . Therefore, no portion of the gel is over-heated beyond the hyperthermia temperature, even after prolonged actuation. While the temperature increases very quickly leading to a rapid collapse, the recovery is slower during one actuation cycle (Movie 5).
In addition to the temperature measurements using an IR camera, the changing ridge width of the gel during actuation is measured, demonstrating a ridge width of ~ 32 µm before light exposure and a change in ridge width from ~ 31.7 to 27.4 µm and vice-versa during actuation.
Based on the calibration curve correlating the ridge width with temperature, obtained when the gels are heated without light in medium and at thermal equilibrium (Figure 1 e) an adverse effect of temperature on cells is prevented.

Supplementary Note 5. Simulation of heat dissipation from the hydrogel film in response to a NIR laser
Local actuation of the gel is driven by local temperature changes. The temperature changes are kept at a minimum to prevent overheating of the gel, which can lead to adverse effects on cells.
Infrared imaging is used to measure the temperature changes that occur in the irradiated zone on the gel. However, it is well known that time dependent measurements of the local temperature resolved in three dimensions in such small volumes are prone to error. To gain more insight into the heat dissipation during photothermal heating of this system, heat transfer is modeled using finite element modeling.
In this basic model, all time dependent variations of the system, such as volume changes and the increase in the AuNR density that occur due to the collapse of the gel are ignored. Based on the results from the infrared experiments, it is assumed that a steady state is rapidly reached during irradiation (t < laser ON time). Therefore, heat diffusion and the temperature distribution across the gel surface is calculated for the situation after photothermal equilibrium is achieved.
It is assumed that heat conduction dominates the scene (thus convection and radiation are neglected) and that the media and the side border of the gel are in thermal equilibrium with the environmental chamber (temperature constant at 36 o C). Another assumption is that the power absorbed by the AuNRs is uniformly distributed over the volume of the irradiated gel, resulting in an isotropic heat flux from the irradiated volume at any time.
The actuation chamber consists of the gel that is covalently attached to a small glass coverslip (diameter = 9 mm, thickness = 300 μm), which is in turn fixed on a larger glass coverslip giving the following equation to solve:

Supplementary Equation 13
where v is the variation function in the mesh, vanishing at the boundary, v(A) = 0, T is the temperature, V is the volume, q is the heat flux incident on the sample, which is 0 everywhere outside the area of irradiation, A is the area exposed to the laser i.e. the heat source. The outer boundary of the sample is at 309.15 K (36 o C).
The material parameters used in the simulation are listed in Table 3. The gel is irradiated using laser pulses of 100 ms at 1 Hz, with a laser power of 340 mW. The absorbance value of a gel prepared from a precursor solution with OD 100 is used to calculate the power reaching the surface of the gel, P abs as determined by the Lambert-Beer law (Supplementary Method 3) Using the parameters mentioned above, the heat intensity for a duty cycle of 10 % is given by The simulations show that the average temperature of the irradiated area is ~ 1.8 °C higher than the surrounding temperature. This is in very close agreement with the IR temperature measurements revealing a ΔTmean ~ 1 °C. The heat dissipation over the hydrogel in the z direction ( Supplementary Figure 8 d) is asymmetric, suggesting that a major part of the heat is dissipated through glass. This finding is also significant for cells, as it conveys that cells are not exposed to elevated temperatures.

Supplementary Note 6. Surface functionalization of the gel to support cell growth
Physisorption of fibronectin to the gel does not result in a uniform coating and leads to inhomogeneous cell growth (data not shown). Therefore, Sulfo-SANPAH is used as a bi- The effect of persistence is reflected in the contour length (distance covered by the cells) and the end-to-end distance of the trajectories (the net displacement of the cell from time t = 0 h to t = 12 h, Supplementary Figure 12 a-d).
The trajectories of single cells are analyzed by calculating the mean-squared displacement (MSD) as a function of lag time 21 . The MSD reflects the average end-to-end distance that is traversed by the cells during the lag time interval. In this study, the displacement (r) is calculated from time lapse frames acquired at time intervals (∆t) of 1 hour (13 frames acquired over 12 h, including the initial time t = 0). The squared displacement (SD) for a defined lag time equals the square of the end-to-end distance between t = 0 and t = n∆t, with n ranging between 0 and 12. The MSD < r 2 > at a given lag time n∆t is calculated as the mean of the

Supplementary Equation 15
Where P t (x, y) is the position vector of the cell at time t, described by the Cartesian coordinates (x,y). Comparing the motion to particle diffusion 22 , the mean squared displacement shows the characteristics of anomalous subdiffusion. However, due to the high standard deviation of the data, no fitting is possible. The trajectories also indicate persistence related to the microstructure of the surfaces and an increase of this persistence in correlation to the mechanical actuation of the gels.