An ultra-fast mechanically active cell culture substrate

We present a mechanically active cell culture substrate that produces complex strain patterns and generates extremely high strain rates. The transparent miniaturized cell stretcher is compatible with live cell microscopy and provides a very compact and portable alternative to other systems. A cell monolayer is cultured on a dielectric elastomer actuator (DEA) made of a 30 μm thick silicone membrane sandwiched between stretchable electrodes. A potential difference of several kV’s is applied across the electrodes to generate electrostatic forces and induce mechanical deformation of the silicone membrane. The DEA cell stretcher we present here applies up to 38% tensile and 12% compressive strain, while allowing real-time live cell imaging. It reaches the set strain in well under 1 ms and generates strain rates as high as 870 s−1, or 87%/ms. With the unique capability to stretch and compress cells, our ultra-fast device can reproduce the rich mechanical environment experienced by cells in normal physiological conditions, as well as in extreme conditions such as blunt force trauma. This new tool will help solving lingering questions in the field of mechanobiology, including the strain-rate dependence of axonal injury and the role of mechanics in actin stress fiber kinetics.


Supplementary Note 1 -DEA cell stretcher design and control
The DEA cell stretcher design shown in Fig. S1(a)-(b) maximises the strain uniformity in the central transparent gap. The vertical electrodes generate tensile strain whereas the horizontal electrodes generate compressive strain. The high aspect ratio (e>g) of the gap separating the vertical electrodes minimizes the border effects (necking in the strain profile), while the narrow w feedlines minimize the actuation along the y-axis and provide a more uniaxial deformation (i.e. it maximizes ε x /ε y ). The horizontal electrodes are designed to expand along their length and compress the central transparent gap of dimension g x e. The width w of the horizontal electrodes has to be larger than the gap w>g to provide a uniform strain profile, but significantly smaller than the membrane width to maximise the actuation strain 1 . It is also important to keep transparent passive regions in the membrane to be used as control areas where cells are not exposed to mechanical strain.
The electrical contacts between the DEA cell stretcher and the voltage power supply are achieved using a custom made holder presented in Fig. S1(c). To connect the device, each metallic pad is coated with a drop of electrolyte gel and the device is assembled on top of the holder. The holder is then connected to a high-voltage power supply 2 as presented in Fig. S1(d). The device can be controlled with a single bipolar power supply and a bridge rectifier, connecting the positive and negative output of the rectifier to the tensile and compressive electrodes respectively. In this configuration, the actuation mode of the DEA cell stretcher is controlled by the polarity of the driving signal. The main fabrication steps of the DEA cell stretcher are presented in Fig. S2. In addition to the information provided here, the fabrication process of a DEA cell stretcher 3 , as well as protocols for DEA materials preparation and processing 4 have been reported in prior work.
(a) The cell stretcher fabrication starts with a silicone (Sylgard 186, Dow Corning) membrane which is stretched biaxially by λ L =1.2 in one axis, and λ H =2.7 in the perpendicular axis. The prestretched membrane is then fixed between two rigid Poly(methyl methacrylate) (PMMA) frames using silicone adhesive (ARclear, Adhesive Research) and silicone RTV (Silpuran 4200, Wacker). (f) The bottom well is covered with a glass coverslip, hence creating a microfluidic reservoir which is then filled with vegetable oil and sealed using silicone RTV. The oil immersion provides a barrier between the cell culture and the external environment, otherwise only separated by a 30 micron thick silicone membrane. The immersion also prevents optical degradation of the silicone membrane (i.e. formation of micro cracks at its surface) over periodic actuation.
Once the fabrication is completed, the culture well is filled with growth medium and cells are cultured directly on top of the DEA. The compact transparent device can be easily mounted on top of an inverted microscope for live cell imaging. Cells are separated from the glass coverslip by approximately 0.3 mm, which gives a lower limit for the working distance of microscope objectives compatible with the system. The device response was characterized for voltage steps of increasing amplitude and the results are presented in Fig. S3. The strain is normalized to the strain reached 1 s after the voltage step is applied.
The results show that the response time of the DEA cell stretcher is independent of the driving voltage and actuation strain. A voltage-independent transfer function can therefore be used to describe the device and calculate the exact driving signal required to precisely mimic any time-varying strain profile. Future work will investigate this approach and evaluate its reliability. The response of the DEA cell stretcher to an overdrive function is discussed in the main manuscript. The function we used is describes by equation (S1) where V is the driving voltage in volt and t is the time in millisecond.
= � 0 , − 500 ≤ < 0 500 • � − 100 ⁄ + −100 2 + −10 3 � + 2500, 0 ≤ ≤ 1500 (S1) The shape of the overdrive function is presented in Fig. S4. At 0 ms the voltage instantly ramps from 0 V to 4 kV (due to the finite resolution of the waveform generator the voltage step takes 0.1 ms) after which it exponentially decreases to 2.5 kV over the following 1,000 ms. This empirical function was obtained by trial and error and could be further improved. The frequency response of the DEA cell stretcher was characterized in tensile mode. The actuation strain was measured at different frequencies using a 3 kV sinusoidal signal and the results are presented in Fig. S5(a). As the frequency increases, the device doesn't have time to fully expand or relax between cycles and the actuation strain decreases. As discussed in the main manuscript, the system is limited by its mechanical time constant and not by its electrical time constant. We calculated the strain amplitude ∆ε=ε max -ε min as the difference between the maximum ε max and minimum ε min strain values, and normalized it to the strain amplitude measured at 0.1 Hz. The normalized strain amplitude is reported in Fig. S5(b) as a function of frequency for different driving voltages. The results show that the cut-off frequency varies with the driving voltage and actuation strain, and is equal to 800 Hz at 3 kV. The lifetime and stability of the DEA cell stretcher was characterized in culture conditions. The culture chamber was filled with growth medium and the device was actuated using a 1 Hz square wave signal with a 50 % duty cycle. Every 10'000 cycles, the strain was measured at rest (0 V) and in the actuated state (4 kV), waiting 1 s between applying the voltage and measuring the strain. The results obtained over 80,000 actuations cycles are reported in Fig. S6 as a function of time. The experiment was stopped after 80,000 cycles due to a failure of the DEA by dielectric breakdown. The strains at rest and in the actuated state show a small drift over the first 10,000 cycles, after which the actuation strain remains very stable over the next 70,000 cycles. The strain drift can be due to a complex combination of stress relaxation, viscoelasticity, and plastic deformation of the membrane and the electrodes. 3 Figure S6: Actuation strain of the DEA cell stretcher at rest (0V) and in the actuated state (4 kV) over periodic actuation. The device was actuated using a 1 Hz square wave signal with a 50 % duty cycle. The experiment was stopped after more than 20 hours of operation and 80'000 actuation cycles due to the DEA failure by dielectric breakdown. The device exhibit a small 1-2 % strain drift over the first 10'000 cycles, after which the actuation strain remains very stable.
The optical transparency and compact design of the DEA cell stretcher enables live cell imaging. Inverted microscope objectives can be brought 300 µm from the cultured cells, limited by the thickness of the DEA membrane and the oil backing. Standard objectives can be used at up to 40x magnification as seen in Fig. S7(a) which shows human lung carcinoma A549 cells imaged at 40x magnification (DNA and mitochondria are seen in blue and green respectively). Higher magnification can be achieved using long working distance objectives as seen in Fig. S7(b) which shows human lung carcinoma A549 cells imaged at 60x magnification (DNA and mitochondria are seen in blue and red respectively).
Between the rest and actuated (10 % strain) states, the cell monolayer moves by approximately 30 µm out-of-plane. It is therefore necessary to refocus the microscope when working at high magnification. The amplitude of the displacement increases with the driving voltage and the actuation strain. For a 10 % actuation strain on a 30 µm -thick membrane, the thickness compression contributes to the out-of-plane motion by only 1.5 µm. The motion is induced mainly by electrostatic forces generated between the high-voltage electrodes and the microscope objective, effectively pulling the membrane towards the objective when the DEA is actuated. This effect can be suppressed by changing our two electrodes configuration to a three electrodes configuration where a central high-voltage electrode is bounded on both sides by ground electrodes.

Supplementary Videos legends
Supplementary Video S1 Light microscopy videos demonstrating the device's ultra-fast response time, its capability to reproduce complex strain profiles, and compatiblity with high magnification microscope objectives. The first clip shows the device's response to a 3 kV voltage step. The video was recorded using a high-speed camera and it is slowed down by 1,000x in order to see the sub-millisecond response time of the system. The second clip shows the device generating a complex strain profile that reproduces the mechanical environment of the myocard. The last clip shows a real time video of cells being stretch and imaged at 40x magnification.

Supplementary Video S2
Animations showing fluorescence images of cells on the DEA cell stretcher at the rest and actuated states. Human lung carcinoma A549 cells were stained to see DNA in blue (Hechst, Invitrogen) and mitochondria in green (MitoTracker Green FM, Invitrogen), and imaged using a Nikon A1r confocal microscope. The first animation displays images acquired at 10x magnification, showing the displacement of a small cell population's nuclei upon actuation. The second and third animations display images acquired at 40x magnification, showing the deformation of single cells and their intracellular content upon actuation.