Autonomous push button-controlled rapid insulin release from a piezoelectrically activated subcutaneous cell implant.

Traceless physical cues are desirable for remote control of the in situ production and real-time dosing of biopharmaceuticals in cell-based therapies. However, current optogenetic, magnetogenetic, or electrogenetic devices require sophisticated electronics, complex artificial intelligence-assisted software, and external energy supplies for power and control. Here, we describe a self-sufficient subcutaneous push button-controlled cellular implant powered simply by repeated gentle finger pressure exerted on the overlying skin. Pushing the button causes transient percutaneous deformation of the implant's embedded piezoelectric membrane, which produces sufficient low-voltage energy inside a semi-permeable platinum-coated cell chamber to mediate rapid release of a biopharmaceutical from engineered electro-sensitive human cells. Release is fine-tuned by varying the frequency and duration of finger-pressing stimulation. As proof of concept, we show that finger-pressure activation of the subcutaneous implant can restore normoglycemia in a mouse model of type 1 diabetes. Self-sufficient push-button devices may provide a new level of convenience for patients to control their cell-based therapies.

INTRODUCTION

Engineered cell-based therapies hold great promise for the treatment of a wide range of chronic diseases. Advances in the field of synthetic biology have made it possible to engineer cells with reliable sense-and-response therapeutic programs, which, upon implantation in mouse models, can treat chronic pain (), obesity (), gouty arthritis (), liver disease (), and diabetes (–), and can reverse muscle atrophy (). Early synthetic gene switches relied on soluble small molecules, e.g., antibiotics (, ) or food components (, , ), as external input signals to control cellular behavior. The development of genetic switches responsive to traceless physical cues, such as light (–), magnetic fields (), ultrasound (), and, more recently, electrical signals (), overcame the challenges associated with systemic delivery of chemical inducers and improved the spatiotemporal resolution of therapeutics delivery. Among currently available types of physically triggered gene switches, electrogenetic switches have the advantage of simplifying the process of energy conversion, enabling direct electrical control of cellular functions.

Electrical signals regulate a variety of processes in human physiology, such as contraction of cardiomyocytes (), secretion of hormones by endocrine cells, and release of neurotransmitters by neurons (–). These actions are mediated by many proteins in the plasma membrane, including ion channels that allow ions to flow across the membrane. Inspired by the action potentials in electrically active cells, we recently engineered nonexcitable cells to perceive externally applied electrical signals by overexpression of specific ion channels and to react by secreting insulin (). In the first proof-of-concept application of electrogenetics to control cell-based therapies, we showed that an implant containing these cells could successfully treat type 1 diabetic mice (). However, the cell encapsulation device required sophisticated electronics and relied on an external power supply to electrically stimulate the implanted therapeutic cells. Instead, encapsulation devices independent of an external power source would have great practical significance for disease therapy.

The ability of piezoelectric materials to generate voltage upon mechanical deformation has attracted considerable attention for biomedical applications, such as health monitoring (–) and tissue engineering (–). Advances in the nanofabrication of piezoelectric materials have improved various properties, including flexibility (), higher voltage supply (), and stretchability (). Biocompatible piezoelectric polyvinylidene difluoride (PVDF)–based films (, ) have been used as power sources in many implantable devices, serving to convert the mechanical forces from heart, diaphragm, or muscles (–) into electricity. We considered that combining piezoelectric materials with electrogenetics could enable self-powered and self-controlled, voltage-based systems for programming electro-inductive designer cells. However, the low amplitude and frequency of mechanical forces generated by organs restricts the voltage generated by piezoelectric materials to a level below the voltage required for triggering these cells by direct coupling stimulation (, ). In this work, we aimed to develop and test a cellular encapsulation device that can efficiently use the low voltage generated by gentle finger-pushing a PVDF film-based piezoelectric module to trigger encased electro-sensitive human designer cells to produce a therapeutic output. As a proof of concept, we show that finger push–triggered insulin secretion from the developed subcutaneously implanted button-like devices could successfully restore normoglycemia in type 1 diabetic mice.

RESULTS

Toward piezo-triggered electro-inducible vesicular secretion

First, we evaluated the suitability of piezoelectric PVDF films for electrical stimulation of electro-sensitive cells derived from the pancreatic beta cell line 1.1E7 (), which was engineered to express proinsulin, and for constitutive expression of CaV1.2 and Kir2.1 channels (). We found that the piezoelectrically generated electric charge stimulated the opening of the voltage-gated CaV1.2 channels, triggering the release of insulin stored in secretory vesicles (Fig. 1A). We initially tested the most common electrical stimulation method, which is direct coupling using electrodes placed in the culture medium (, ), connected to each side of the PVDF-based piezoelectric module to provide the positive and negative charges (Fig. 1B). The electrodes were arranged in two configurations in relation to the cell monolayer, generating an electric field above (Fig. 1, C and E) or across (Fig. 1, D and F) the monolayer. However, although KCl-induced membrane depolarization activated insulin secretion in electro-sensitive cells, they did not respond to the voltage generated by finger-pushing a 52-μm-thick piezoelectric module for 3 min at 1 Hz in either of the electrode arrangements, as similar insulin levels were secreted by stimulated and nonstimulated cells (Fig. 1, G and H).

Fig. 1.

Stimulation of electro-sensitive cells with piezoelectric pulses.

(A) Upon membrane depolarization of engineered electro-sensitive cells constitutively expressing the channels Cav1.2 and Kir2.1 and a synthetic Proinsulin construct, the intracellular calcium concentration is increased, which drives the release of vesicle-stored insulin. (B) Illustration of the prototypic piezoelectric module. The electrodes are connected to the top and bottom sides of a silver-coated PVDF film, providing positive and negative charges, respectively. (C and E) Scheme and image of the stimulation cell culture multiwell plate with electrodes aligned vertically in relation to the cell monolayer. (D and F) Scheme and image of stimulation cell culture inserts and multiwell plate with electrodes aligned horizontally in relation to the cell monolayer. Cells were cultured on a porous membrane inside inserts, with one electrode placed above and one below the cell monolayer to generate an electric field across it. For both vertical (G) and horizontal (H) electrode configurations, piezoelectrical pulses were produced for 3 min at 1 Hz by finger-pressing a piezoelectric module with a 52-μm PVDF film placed on human skin to allow sufficient deformation. Cells were also either nonstimulated or treated with 40 mM KCl (chemical depolarization) as negative and positive controls, respectively. Insulin expression level was measured in culture supernatant samples collected after 3 min of stimulation. Data are means ± SEM; n = 4. Statistical analysis was done with a two-tailed t test. ns, not significant; ***P < 0.001.

Design of a piezoelectrically powered implant

We hypothesized that a more conductive cell-culture chamber might enable the low voltage generated from the piezoelectric material to trigger electro-sensitive cells. Therefore, we modified commercially available cell culture inserts (Fig. 1D) by coating their entire surface with a thin platinum layer (Fig. 2A), including the semipermeable membrane, which retained an average pore diameter of 0.4 μm (Fig. 2B), and connected two electrodes on opposite sides of the cell chamber. To test the conductive chamber for low-voltage electrostimulation, we applied unipolar square pulses with alternate polarization to electro-sensitive cell cultures and measured the insulin responses to different voltages, stimulation frequencies, and stimulation periods. The conductive chamber significantly improved the efficiency of electrostimulation, with secretion of insulin peaking at 1 V (Fig. 2C), 1 Hz (Fig. 2D), and after 3 min of stimulation (Fig. 3E). The electro-sensitive cells were not activated by lower-frequency pushes, which may reflect occasional inadvertent contacts (fig. S2). The electro-sensitive cells had released all of their stored insulin after 10 min of stimulation, reaching levels similar to those obtained with KCl stimulation (Fig. 2, C and E). Stimulation for 10 to 15 min afforded the highest level of insulin production, without affecting the cell viability. Nevertheless, 3-min stimulation also provided significant insulin induction, generating levels only about 20% lower than the maximum achieved levels. Therefore, we selected 3 min for follow-up experiments because shorter stimulation times are preferable from the viewpoint of convenience. The cell viability was not affected at these voltages (Fig. 2F) and frequencies (Fig. 2G). Next, we tested whether the PVDF-based piezoelectric module (Fig. 1B) connected to the electrodes in the conductive cell chamber could stimulate the electro-sensitive cells. For this purpose, the piezoelectric module was placed above human skin and gently finger-pressed for 3 min, which showed maximum insulin secretion at frequencies over 1 Hz (Fig. 2H). The pressure generated by finger-pushing corresponds to 1 kPa (0.1 N/cm2), which is lower than typing on a keyboard (). The electrosensitive cells could also be stimulated by an electric toothbrush (E-toothbrush), which vibrates at 50 Hz and produces 1 V when placed on top of the piezoelectric module (fig. S1). The use of an E-toothbrush also activated insulin secretion within 3 min of stimulation, generating similar levels to those achieved by finger-pushing at 1 Hz (Fig. 2H). E-toothbrush control of insulin expression could be a very convenient and straightforward way to trigger insulin secretion after each meal with high compliance, simply by brushing for 3 min.

Fig. 2.

Design and in vitro characterization of the conductive chamber.

(A) Schematic representation and picture of the conductive cell culture chamber containing a porous membrane for cell attachment and two electrodes connected on opposite sides of the chamber. The whole chamber, including the porous membrane, is coated with a thin platinum (Pt) layer. (B) Scanning electron microscopy image of the porous membrane, showing that the Pt deposition had no effect on the average pore size of 0.4 μm (shown in black). (C to E) Insulin secretion level upon stimulation of electro-sensitive cells by a pulse generator. (C) Voltage dependence. Electrical stimulation was performed for 3 min at 1 Hz, at the indicated voltages. (D) Frequency effect. Electrical stimulation was performed at the indicated frequencies for 3 min at 1 V. (E) Stimulation time dependence. Electrical stimulation was performed for the indicated time periods at 1 V and 1 Hz. (F and G) Cell viability after electrical stimulation for 30 min at (F) 1 Hz and the indicated voltages or (G) 1 V and the indicated frequencies. (H) Insulin secretion level upon stimulation of electro-sensitive cells with voltage generated by finger pressure or E-toothbrush on a 28-μm-thick piezoelectric module for 3 min at the indicated frequencies. (I) Reload kinetics of vesicular insulin. Cells were stimulated twice during 3 min with the indicated time intervals between stimulations. (J) Reversibility assay. Electro-sensitive cells were stimulated twice for 3 min with a 4-hour interval between the first and second electrostimulation on each day in period of 3 days. Data are means ± SEM. Statistical analysis was done with a two-tailed t test (n = 3). **P < 0.01, ***P < 0.001.

Fig. 3.

Design and in vitro characterization of the button-like cellular device.

(A) Schematic representation of the device. The cell chamber is delimited by two porous membranes, with the bottom one also coated with platinum (Pt). A reinforcement ring is placed between the piezoelectric module and the conductive cell chamber. The piezoelectric module is linked by Pt wires to the conductive cell chamber, which encases engineered electro-sensitive cells. The conductive chamber allows diffusion of nutrients and therapeutic protein in and out of the device, as well as protecting the cells from immune system attack. (B) Pictures of the (i) top and (ii) bottom sides of unfolded and folded devices. The cells are injected into the cell culture chamber via a port in the reinforcement ring, followed by heat sealing of the porous membrane. Piezoelectric voltage output generated by finger-pushing the piezoelectric module on different materials. (C) Output voltage generated by periodically finger-pressing PVDF film–based piezoelectric modules of 28 and 52 μm thickness on top of human skin (applied pressure, around 1 kPa). (D to F) Voltage generated from piezoelectric module placed above 0.5 mm (D), 1 mm (E), and 2 mm (F) reinforcement rings. (G) Insulin secretion by electro-sensitive cells upon electrostimulation by finger-pushing the button-like implant for 3 min at 1 Hz. Nonstimulated and KCl-stimulated cells were used as negative and positive controls, respectively. (H) Insulin secretion kinetics. Electro-sensitive cells were stimulated for 30 min by finger-pushing and insulin levels in the cell chamber were analyzed by ELISA. Data are means ± SEM. Statistical analysis was done with a two-tailed t test (n = 3). ***P < 0.001.

We also evaluated how long it would take for the cells to replenish their insulin pool in secretory vesicles, by performing a second stimulation 1 to 4 hours after the first stimulation. As expected, the amount of secreted insulin increased as the interval between stimulations became longer, with the insulin response to the second stimulation reaching that of the first stimulation when the interval was 4 hours (Fig. 2I). The cells responded repeatedly with reproducible insulin secretion kinetics to two consecutive stimulations within 4 hours (Fig. 2J). Together, these data demonstrate that the electricity generated by transient deformation of the piezoelectric module can stimulate the electro-sensitive cells in the conductive chamber, matching the stimulation achieved by an external power source (Fig. 2, C to E), which shows that the piezoelectric module can be used as a reliable, self-sufficient power source for electro-stimulation.

Optimization and validation of the push-button implant design in vitro

To make a compact, subcutaneously implantable piezoelectric-based device, we sandwiched a semipermeable platinum-coated cell chamber and a piezoelectric module with a reinforcement ring between them (Fig. 3A). The bottom membrane of the cell culture chamber allows exchange of oxygen, nutrients, and therapeutic proteins and protects the cells from the host’s immune system. This conductive chamber is linked by platinum wires to the piezoelectric module for voltage transfer. The reinforcement ring, three dimensionally (3D) printed in U.S. Food and Drug Administration–licensed polylactide, protects the top membrane from the mechanical force of pressing and provides space for deformation of the piezoelectric module. The resulting button-like implant device is smaller than a 1-cent Euro coin and smaller than a U.S. dime (Fig. 3B). The magnitude of the voltage generated on the surface of piezoelectric module when a mechanical force is applied depends on the thickness of the film. Finger-pushing a 28- or 52-μm-thick PVDF film–based piezoelectric module placed on human skin with pressures around 1 kPa provides voltages of 1 or 2 V, respectively (Fig. 3C and movie S1). We also characterized the voltages generated by the piezoelectric module on top of the cell culture chamber using reinforcement rings of different thicknesses (0.5, 1, and 2 mm) to determine the minimum thickness (for a more compact design) that would provide enough space for deformation of the module by finger-pushing. Rings of 1 or 2 mm were sufficient to generate 1 or 2 V from 28- or 52-μm PVDF films, respectively (Fig. 3, D to F). Thus, we chose 28 μm as the optimal thickness of the PVDF film to deliver 1 V upon deformation and 1 mm as the minimum thickness of the reinforcement ring to provide sufficient space for film deformation. To validate the button-like device in vitro, electro-sensitive cells were loaded inside the cell culture chamber and the piezoelectric module on top was finger-pressed for 3 min at 1 Hz. The stimulated cells secreted 20-fold more insulin than nonstimulated cells, and the amount was comparable to that released by cells depolarized with 40 mM KCl (Fig. 3G). When the stimulation time was extended to 30 min, the level of secreted insulin plateaued (Fig. 3H), suggesting that cells released all of their vesicle-stored insulin during this period.

Biocompatibility and longevity of the push-button implant in vivo

The biocompatibility of the button-like implant was assessed at 30 days after subcutaneous implantation in the dorsal back of mice. We observed no systemic toxicity of the kidneys or liver of treated animals (table S1) and no local immune cell infiltration of the implant, as validated according to ISO 10993. Histological images of a middle section of the cell culture chamber (Fig. 4A) and a section at the electrode side (Fig. 4J) of explanted devices revealed no apparent difference between the well-vascularized fibrous capsule surrounding cell-free (Fig. 4, B, C, K, and L, and table S2) and cell-containing (Fig. 4, F and G) implants. Likewise, there was no apparent difference in immune cell infiltration in the fibrotic tissue around the cell-free (Fig. 4, D and E) and cell-containing (Fig. 4, H and I) implants. Last, a functionality assessment of the explanted devices showed that they could still produce about 1 V in response to finger-pushing the piezoelectric module at 30 days after implantation (Fig. 4M), and there was no decline in performance compared with before implantation (Fig. 3E).

Fig. 4.

Biocompatibility of the piezoelectrical cell implants.

(A) Images of the implant sections (red frames) analyzed for histopathology and immunohistochemistry. The button-like implants were placed subcutaneously on the dorsal side of mice for 30 days. The samples were stained with Paragon (toluidine blue and basic fuchsin) for assessment of fibrotic capsule formation. Overview of the cell culture chamber section (B) without and (F) with cells. Green arrow, piezoelectric module; yellow arrow, reinforcement ring; black arrow, platinum wires; red arrow, membrane of conductive chamber; orange arrow, fibrotic tissue. (C and G) Magnified view of the chamber section. Red arrow, blood vessel; orange arrow, fibrocytes; green arrow, piezoelectric module. (D and H) The adherent tissues around the implant were immunostained with anti-CD68 antibodies. CD68-positive cells are red-circled and mark activated macrophages and fibroblasts. (E and I) CD11b-positive cells are red-circled and indicate macrophages, neutrophils, and natural killer cells. Histopathology of electrode section. (J) Image of the electrode section (red frame) analyzed for histopathology. (K) Overview of the electrode section. Yellow arrow, fibrotic tissue; blue arrow, piezoelectric module; black arrow, platinum wires; red arrow, electrodes. (L) Magnification of the chamber section. Blue arrows, piezoelectric module; yellow arrow, fibrocytes. (M) Functionality of the explanted devices at 30 days after implantation. Representative output voltage generated by periodically finger-pressing an explanted piezoelectrical cell implant.

Push-button therapeutic control of experimental type 1 diabetes

The mechanoelectrical actuation of engineered electro-sensitive cells enabled by piezoelectric materials has several advantages for implementation of cell-based therapies, as it enables traceless control, can be easily activated by gentle finger-pushing without the need for external power sources to apply the electrical stimulation, and provides an extremely compact design not requiring any electronic parts or batteries (Fig. 5A). For preclinical proof of concept, the button-like devices loaded with electro-sensitive insulin-producing cells were subcutaneously transplanted into type 1 diabetic mice. Mice stimulated by finger-pushing the skin immediately above the implant showed a rapid decrease in fasting blood glucose levels, while nonstimulated mice showed continued hyperglycemia (Fig. 5B), confirming the effectiveness of finger push–mediated remote control of insulin release by the implanted cells. Peak blood insulin levels were reached 2 hours after initiating the stimulation (Fig. 5C). The button-like cellular device could also provide sufficient insulin to rapidly attenuate postprandial blood-glucose levels during glucose tolerance tests (Fig. 5D). In contrast, blood glucose levels remained high in mice without the implant and in implanted animals that were not stimulated with finger-pushing. Blood insulin levels were monitored during 1 week, and mice stimulated by finger-pushing consistently showed significantly higher insulin levels compared with those of nonstimulated mice (Fig. 5E). Furthermore, it should be noted that all of the mice (finger-push treated and controls) were moving freely inside the cages and occasionally bumped into each other or into the cage walls. For the real-time blood glucose–monitoring experiments and glucose tolerance tests (Fig. 5, B and D), the mice were anesthetized during the finger-pushing period but were in a state of normal activity during the 2-hour monitoring period. Therefore, we believe that our device does not have potential risk for inadvertent activation during normal activity. Together, these results indicate that simply pushing the skin above the subcutaneously implanted button-like cellular device can activate insulin secretion by encapsulated electro-sensitive designer cells to control type 1 diabetes.

Fig. 5.

In vivo performance of the button-like cellular implant.

(A) Electric pulses are generated by finger-pushing the skin above a subcutaneously implanted cellular device incorporating a piezoelectric module. (B and C) Wild-type or type 1 diabetic mice with subcutaneously implanted devices were fasted for 4 hours before finger-push stimulation at around 1-Hz frequency. Blood samples were collected at indicated time points. Nonstimulated mice were used as controls. Time courses of blood (B) glucose and (C) insulin levels. (D) Glucose tolerance test. Wild-type or type 1 diabetic mice were treated by finger-pressing the skin above the cellular implant and then injected intraperitoneally with 1.25 g of glucose per kilogram of body weight. (E) The button-like devices were subcutaneously implanted into type 1 diabetic mice and insulin levels were recorded every 2 days after of finger-push stimulation, over a period of 7 days. The statistical significance of the difference between the finger-push and no-push groups was calculated. Data are means ± SEM. Statistical analysis was done with a two-tailed t test (n = 5). *P < 0.05, **P < 0.01, ***P < 0.001.

DISCUSSION

Engineered cell-based therapies are changing the treatment landscape of life-threatening diseases, constituting an important pillar in modern medicine (, ). The first translational successes were recently achieved with chimeric antigen receptor T cell therapies approved for blood cancers () and an increasing number in clinical development. In these cell immunotherapies, patient’s own T cells are modified with antitumor synthetic receptors to trigger their killing activity upon recognition of tumor cells displaying the target antigens. However, cytokine release syndrome and off-tumor toxicity are common safety concerns, which are being addressed by applying multiple synthetic biology approaches (, ). Beyond cancer, other diseases requiring chronic administration of precisely dosed therapeutic proteins could benefit greatly from cell therapy approaches relying on synthetic gene switches to enable the required control. Engineered cell therapies based on allogeneic cell sources are at early stages of preclinical development, targeting numerous diseases, such as metabolic diseases (), infections (), and autoimmunity (), among others. They have been tested in animal models by implantation subcutaneously or intraperitoneally, using some form of encapsulation to protect the cells from the host’s immune system. While chemically induced gene switches were initially used, recent developments have focused mainly on regulation by traceless physical stimuli such as light (optogenetics) (–), magnetic fields (magnetogenetics) (), ultrasound (sonogenetics) (), and electrical stimulation (electrogenetics) (). Physical triggers are particularly attractive for chronic disease management as the stimulus can be controlled both temporally and spatially, enabling precise control over the cellular therapeutic response. In the present work, we have developed a new traceless way to activate transplanted therapeutic cells, which is as simple as pushing a button. Our piezoelectrically based button-like cellular implant acts as a voltage generator when the button is pushed and can trigger encased electro-sensitive cells to secrete an effector protein, bypassing the need for an external power source for activation and electronics for control. We designed the cell culture chamber to meet several key criteria, namely, (i) to have a highly conductive surface for electro-stimulation of cells with a relatively low resistance and energy input, (ii) to be permeable to nutrients and therapeutic protein but shield the implanted cells from the host immune system, (iii) to have good mechanical properties and thus be resilient to repeated finger-pushing, and (iv) to show good biocompatibility. For in vitro and in vivo testing, the device was loaded with engineered cells that can sense electrical signals and respond in a calcium-dependent signaling pathway by releasing vesicle-stored proteins. The piezoelectrically generated electrical charge produced by pressing the button is efficiently transmitted to the cells via the platinum-coated semipermeable cell chamber. As piezoelectric material, we selected PVDF films, on the basis of their high flexibility, easy processability, semipermeability, and biocompatibility. Such films have been widely explored for applications in biomedicine and tissue engineering (–). The voltage generated is dependent on the thickness of the PVDF film used, and we found that 28-μm-thick films deliver sufficient voltage (1 V) to maximally stimulate the electro-sensitive cells used here. The push-button device could be activated either by gentle finger pushing at 1 Hz or by using an electric toothbrush vibrating at 50 Hz, which represents a particularly attractive medical intervention to release insulin after each meal simply by brushing one’s teeth.

Furthermore, we confirmed that the cellular response in terms of secreted protein levels can be tuned by varying the duration of the stimulation period to deliver the energy required for the conductive cell chamber to be active for long enough. The activation of the implant through pulsatile mechanoelectrical signals allows for more compact implants, as there is no need for electronic switchboards or batteries, which would increase the size, complexity, and susceptibility to errors of the implant.

As an in vivo proof of principle, we implanted the device subcutaneously in type 1 diabetic mice. Repeated gentle finger-pressing of the skin above the implantation site, which is reminiscent of the pulsatile nature of hormone secretion (), was effective to program fast insulin release by electro-sensitive cells in the button-like cellular implant, resulting in the restoration of normoglycemia. Notably, the implant was also able to decrease short-term postprandial high blood-glucose levels with fast kinetics. Although the 0.4-μm pore size facilitates transfer of nutrients and insulin to keep the therapeutic cells viable and functional, it would permit the escape of extracellular vesicles that might be produced by the therapeutic cells, including exosomes (size up to 0.1 μm) (). Various types of cells have been shown to release exosomes packed with nucleic acids, lipids, and proteins, that in a cell transplantation scenario might be transferred to host cells. However, the potential of exosomes released by donor cells to trigger a host immune response in diabetic patients is still unclear (). It should be noted that exosomes have been extensively explored as promising therapeutic delivery vehicles due to their low toxicity, good safety profile, and ability to diffuse easily through plasma membranes (, ). The finger-pressing was gentle, like typing on a keyboard, and the treated animals showed no signs of skin rashes or hepatoma. Furthermore, the need for repeated pushing prevented inadvertent biopharmaceutical release by accidental touches.

For clinical translation, it will be important to confirm the longevity of the piezoelectric device. Implanted PVDF membranes have been validated for up to 6 months and over 7000 bending cycles (), while cells have been shown to remain responsive inside implanted semi-permeable microcontainers for several months (, , ) or even years (). All of our piezoelectric implants showed a high degree of biocompatibility and remained functional in providing a reliable power source for electrically triggered biopharmaceutical release over the entire experimental period of 30 days. The encased engineered pancreatic cells are protected by a permeable membrane, which allows free diffusion of nutrients for cell survival and functionality. Nevertheless, the decrease in insulin production as compared with the in vitro studies might reflect some limitations of oxygen or nutrient supply because of insufficient vascularization of the implantation site. Some strategies have been proposed to overcome this issue. These include implanting empty (cell-free) push-button devices to allow for prevascularization before introducing the therapeutic cells () or incorporating oxygen delivery systems to reduce exposure of the therapeutic cells to hypoxia ().

As regards scaling up the push-button device to treat human diabetes, we could take as a size reference the Viacyte patch (27 cm2) encasing stem cell–derived pancreatic progenitors, which was recently used in phase 1/2 clinical trials (, ). Our patch was fabricated using 3D-printing technology, which is easily scalable to construct larger push-button devices, and, if combined with porous scaffolds (), should provide sufficient conductive surface area to stimulate the therapeutic cells. For human use, the immortalized pancreatic cells used in this study would need to be replaced by more clinically relevant cells, such as primary mesenchymal stem cells (), engineered with the necessary components to produce insulin (or another therapeutic protein) in response to electrical stimulation ().

In a clinical context, such a device could not only greatly simplify the treatment regime of type 1 diabetic patients who are dependent on frequent insulin injections and could be expected to increase compliance, it may also serve as a blueprint for the treatment of other medical conditions requiring sequential secretion of therapeutic peptides such as severe osteoporosis where daily injections of the parathyroid hormone is used to stimulate bone formation (). We believe that this simple, compact device also has great potential for the implementation of electrogenetics in other next-generation cell-based therapies targeting many currently intractable diseases.

MATERIALS AND METHODS

Cell culture

Electro-sensitive cells were previously established in our lab by engineering the human pancreatic cell line 1.1E7 (catalog no. 10070101-1VL, Sigma-Aldrich, Saint Louis, MO, USA) to stably coexpress the Cav1.2 and Kir2.1 channels, as well as proinsulin containing nano luciferase in C peptide portion (). These cells were routinely cultured in RPMI 1640 medium (catalog no. 72400-021, Thermo Fisher Scientific) supplemented with 10% (v/v) fetal bovine serum (FBS; catalog no. 022M3395, Sigma-Aldrich) and 1% (v/v) penicillin/streptomycin solution (PenStrep; Biowest, Nuaillé, France) at 37°C in a humidified atmosphere containing 5% CO2. For passaging, cells were detached by incubation in 0.05% trypsin-EDTA (catalog no. 25300-054, Life Technologies, Carlsbad, CA, USA) for 5 min at 37°C, resuspended in cell culture medium, and centrifuged for 1 min at 200g. Then, the supernatant was discarded, and the cells were resuspended in fresh medium and seeded at a density of 1.5 × 105 cells/ml. Cell number was quantified using an electric field multichannel cell counting device (Casy Cell Counter and Analyzer Model TT, Roche Diagnostics GmbH, Rotkreuz, Switzerland). For electrostimulation, electro-sensitive cells were cultured in RPMI 1640 medium supplemented with 10% FBS, 1% PenStrep, and 2.0 mM CaCl2.

Recording of voltage from PVDF film–based piezoelectric module

PVDF films of 28 μm (catalog no. 1-1002608-0, TE Connectivity, USA) and 52 μm (catalog no. 2-1002608-0-M, TE Connectivity, USA) were placed on top of human skin or reinforcement rings and finger-pushed. The output voltage was recorded on a digital oscilloscope (Rigol DS1052E, Rigol Tehnologies Inc., PRC). The pressure generated by finger-pushing was measured with a digital force gauge (catalog no. 123946-AW, Sauter FK 50, Sauter, Switzerland).

Preparation of the conductive cell culture chamber

The conductive chamber was customized from Corning Transwell cell culture inserts (catalog no. 353095, VWR, Switzerland), which were cut to have a height of around 2.5 mm, followed by a sputter process to deposit 200 nm of platinum (Ionfab300, Oxford Instruments, Abingdon, UK) on a Polyester (PET) semipermeable membrane with a pore size of 0.4 μm. The resulting conductive layer was imaged using a scanning electron microscope (JSM-7100F, JOEL, Japan).

Electrostimulation through the culture medium

(i) Electrodes placed vertically in relation to the cell monolayer: A C-dish (Ionoptix, Dublin, Ireland) containing a pair of vertical carbon electrodes that fit into the wells of standard 6-well culture plate was used for transferring the voltage provided by PVDF films. The electro-sensitive cells were seeded at a density of 35,000 cells/cm2 in 1.4 ml of medium. (ii) Electrodes placed horizontally in relation to the cell monolayer: A customized 24-well plate containing a pair of 0.5-mm Pt electrodes (catalog no. HXA 050, Cooksongold Ltd., Birmingham, UK) placed above and below the cell monolayer was used. Cells were cultured in an insert with a semipermeable membrane (catalog no. 353095, Falcon) at a seeding density of 35,000 cells/cm2 in 1.7 ml of medium. (iii) Electrical stimulation applied through the conductive cell chamber: A customized cover contains a pair of 0.5-mm Pt electrodes that fit in individual wells of a standard 24-well plate. The electrodes were connected to the Pt conductive chamber on opposite sides. Cell culture chambers were pretreated with poly-l-lysine to promote cell adhesion (catalog no. P4832, Sigma-Aldrich) before seeding 35,000 cells/cm2 in 0.5 ml of medium. The cells were stimulated by an HP3245A Universal Source function generator or by pressing the piezoelectric module with a finger or using an electric toothbrush (ProtectiveClean 4300, Philips Sonicare).

Insulin quantification

Insulin quantification was measured by mouse enzyme-linked immunosorbent assay (ELISA) kits or correlation with nano luciferase NLuc. Mouse insulin ELISA kits were used to quantify recombinant mouse insulin levels in culture supernatants (catalog no. 10-1247-01, Mercodia, Uppsala, Sweden) and mouse serum (catalog no. 10-1249-01, Mercodia), according to the manufacturer’s instructions. Optical density was measured at 450 nm on a Tecan Infinite M1000 Pro plate reader and the corresponding concentrations were calculated on the basis of the measured absorbances of manufacturer-provided standard solutions. The concentration of NLuc in cell culture supernatants was measured by using the Nano-Glo Luciferase Assay System (catalog no. N1110, Promega, Madison, WI, USA). In brief, 10 μl of each supernatant sample was mixed with 10 μl of Nano-Glo substrate-containing buffer (in a ratio of 1:50) in black 384-well plates (catalog no. 781900, Greiner, Germany) and incubated at room temperature for 10 min. Total luminescence was measured with a Tecan Infinite M1000 Pro plate reader (Tecan Group AG, Maennedorf, Switzerland).

Resazurin-based cell viability analysis

To assess the percentage of viable cells with active metabolism, cells were incubated with resazurin (60 μg/ml; catalog no. R7017, Sigma-Aldrich, USA) for 2 hours. Fluorescence was then measured with a Tecan Infinite M1000 Pro plate reader at excitation and emission wavelengths of 560/9 nm and 590/20 nm, respectively. To calculate the relative cell viability, the fluorescence of non–electrically stimulated cells was set to 100%.

Preparation of the button-like implant device

The device contains a conductive cell culture chamber, a 28-μm PVDF film–based piezoelectric module (catalog no. 11028362-00, TE Connectivity), and a 3D-printed polylactide reinforcement ring. The PVDF film was glued to a PET film (catalog no. GF98750021-2EA, Sigma-Aldrich) on the back with light-curing silicone adhesive (catalog no. 1212167, Loctite 5055, Henkel, USA) and cured under ultraviolet (UV) light (365-nm wavelength) for 30 s to increase the mechanical strength and then trimmed to match the size of the reinforcement ring. Platinum (Pt) wires were used to link the conductive chamber to the electrodes of the PVDF film. They were connected to the bottom of the conductive chamber by applying electrical conductive gel (catalog no. SKU-0018, Bare Conductive), which was then covered with light-curing silicone adhesive and cured for 30 s. Light-curing silicone adhesive was also used to fix the wires to the electrodes in the PVDF film, using the same procedure mentioned above. After connection, the implants were washed in 500 ml of ddH2O for 2 days and sterilized by immersion in 70% ethanol. The bottom conductive membrane of the culture chamber was treated with 0.01% poly-l-lysine solution (catalog no. 25988-63-0, Sigma-Aldrich, Switzerland) for 1 hour and dried at room temperature. Then, porous polycarbonate (PC) membrane (catalog no. 110637, Whatman, USA) with a pore size of 0.4 μm was sealed onto the conductive chamber, sparing a region surrounding the port for cell injection. The reinforcement ring was bonded to the cell culture chamber using Epo-Tek 301-2 (catalog no. 301-2, Epoxy Technology Inc., Billerica, MA, USA), followed by washing in 500 ml of ddH2O for 2 days and UV disinfection. For seeding, a cell suspension of 2 × 106 cells in 200 μl of medium was injected into the conductive chamber via the entry port and afterward the PC membrane was fully sealed by heating.

Animal experiments

Type 1 diabetic mouse model. Wild-type 8-week-old male C57BL/6J mice were fasted for 4 hours and injected with daily doses of freshly diluted streptozocin (STZ; catalog no. S0130, Sigma-Aldrich, Switzerland; 50 mg/kg of body weight in ice-cold 0.1 M citrate buffer) for five consecutive days. Blood glucose levels were measured with a commercial glucometer (Contour Next; Bayer HealthCare, Levekusen, Germany). Five days after STZ treatment, the animals were anesthetized with inhaled isoflurane and the button-like devices were subcutaneously implanted in their backs. For the glucose tolerance test, treated mice were fasted for 8 hours and then finger-pressing was done above the implantation site, followed by injection of d-glucose (1.25 g/kg) and collection of blood samples from the tail vein for glucose measurement. Fasting blood glucose and insulin levels were measured after 4 hours of food restriction at the indicated time points of finger-pressing stimulation. All experiments involving animals were performed according to the directive of the European Community Council (2010/63/EU) and carried out by S.X. and M.-D.H. (license number, 2997/30779) at ETH Zurich in Basel, Switzerland.

Histology

For hematoxylin and eosin (H&E) staining, liver and kidneys from mice implanted with cell-free and cell-containing devices were collected at 30 days after device implantation. Tissue around the implants was excised, embedded in paraffin, and cut into 2- to 4-μm slices, followed by staining with H&E. Explanted devices and surrounding tissue were processed by methyl methacrylate resin embedding. The chamber of each device was diamond-sawed (EXAKT 300 CP System, EXAKT Technologies Inc., Oklahoma City, OK, USA) at its central position in the transverse direction at a thickness of approximately 400 μm. The sections were ground to a thickness of approximately 40 to 60 μm (EXAKT Technologies Inc., Oklahoma City, OK, USA) and stained with Paragon (toluidine blue and basic fuchsin).

Immunohistochemistry

Immunohistochemistry was performed using the rabbit monoclonal antibodies anti-CD11b (Ab133357; lot no. GR3209213-2, Abcam, Cambridge, UK) and anti-CD68 (ab125212, lot no. GR300628-28, Abcam, Cambridge, UK) for all samples. The optimal dilutions for the anti-CD11b and anti-CD68 were established to be 1:7500 and 1:1500, respectively. Epitope retrieval was performed with Bond Epitope Retrieval Solution 1 (citrate-based buffer, pH 5.9 to 6.1, Leica Biosystems, Wetzlar, Germany) for 20 min at 100°C, before the staining protocol was started.

Histopathology

Histopathology evaluation was performed by AnaPath GmbH using a scoring system according to ISO 10993-6:2016(E). The images were taken by an Olympus UC30 camera.

Image analysis

Quantitative analyses of immunohistochemistry sections were performed using the image analysis software QuPath (https://qupath.github.io).

Statistical analysis

All in vitro data represent means ± SEM of three independent experiments (n = 3 to 4). For mouse experiments, each treatment group was composed of five mice (n = 5), and the values are expressed as means ± SEM. Comparisons between groups were made using Student’s t test. Differences were considered statistically significant at P < 0.05. Prism 6 software (version 8.0.0, GraphPad Software Inc.) was used for statistical analysis.