Microelectrode arrays (MEAs) have attracted strong interest due to their use in various applications, including cellular recording, biosensors, and drug screening.1,2,3,4,5,6,7,8,9,10,11 Perhaps one of the most promising application of MEA devices are biomedical implants in which the MEA serves as a vital tool for monitoring or restoring biological functionality.12,13,14 Historically, MEA devices, as developed by Wise and co-worker15 in the late 1960s, consisted of conductive metallic material covered by an insulating layer, except for a small electrode opening to establish a connection to the surrounding tissue. They were fabricated on stiff silicon substrates using photolithographic techniques and silicon etching technology. These and similar devices have enabled recording and stimulation of electrical activity and provided neuroscientists with a tool for studying cellular signaling processes in complex organs, such as the human brain.16 Furthermore, micro- and nanoelectrode arrays have provided possibilities for studying electrochemical signals in cell networks.17,18,19 Nevertheless, establishing a reliable communication between a biological cell and an electrode remains a challenging task partly due to the mechanical mismatch between the soft biological tissues and the rigid electronic chip.20,21,22 An important factor to consider when evaluating electronic interfaces for implants is the Young’s modulus of the biological tissue, which lies in the range of 100 Pa to 10 kPa for tissue of the central nervous system.23,24,25 In contrast, electronic implants exhibit very high elastic moduli in the range of GPa for rigid silicon-based chips.26 In vitro, the stiff substrate alters cell shape, organization, and function and therefore does not represent a model close to the natural cellular behavior with soft surrounding tissue.27,28,29,30 In vivo, the stiff synthetic substrate may trigger inflammatory response or loss of functionalities, indicating rejection of the electronic interface and thus precluding successful translation of these probes to clinical research.20,2331,32 Furthermore, rigid electronic implants can alter the physiological movement of organs such as the heart33,34,35 and neuronal tissue.36,37,38 Therefore, numerous studies have been conducted to optimize flexibility and geometrical structure of different substrates for future implants.39,40,41,42,43,44,45,46 However, adding electronic functionality to soft substrate materials remains difficult due to technical limitations arising from standard fabrication methods. Recently, bioactive coating of MEAs using hydrogels has been introduced in an effort to overcome the mechanical mismatch of the metal–biological interface.47,48,49,50,51 Adding a soft hydrogel layer to neuronal implants significantly decreases the local strain and modulates the immune response in the brain.52,53 Likewise, several methods have been investigated to fabricate bioelectronic interfaces on flexible substrates such as polyimide6,54 and parylene42,55, or to transfer a metallic pattern onto soft polydimethylsiloxane (PDMS) substrates.56,57 Very recently, an implanted neural MEA interface has been developed, which is capable of restoring voluntary control of locomotion after traumatic spinal cord injury.58 Moreover, an MEA with platinum–silicone electrodes has been patterned on a PDMS substrate and used for successful in vivo recordings from the spinal cord of a rat.44 One of the advantages of using flexible MEAs is to have a conformal contact between the living tissue and the electrode with minimal invasiveness.44 Nevertheless, current microfabrication strategies typically require expensive instrumentation as well as time-consuming research and development cycles for each substrate. In addition, most of the flexible substrates and common electrode materials that are compatible with classical fabrication approaches are still relatively stiff compared to the interfacing tissue (see Table 1). This calls for technologies that can add electronic functionality to truly soft substrates, such as hydrogels, in a rapid prototyping approach.

Table 1 Summary of the elastic modulus of flexible substrates, electrodes, and tissue

Ink-jet printing has recently emerged as a versatile alternative fabrication tool for patterning high-resolution microstructures with complex electrode geometries on the micrometer scale.59,60,61,62,63,64,65,66,67,68,69 A major advantage of the fabrication using ink-jet printing is the possibility of changing the structure design in flight. Ink-jet printing eliminates the need for pre-patterned lithographic masks and thus allows for the adaptation of different geometries in a cost- and time-efficient manner. Another advantage of this method is the ease of incorporating newly emerging ink materials such as carbon or PEDOT: PSS (poly(3,4-ethylenedioxythiophene) doped with poly(4-styrenesulfonate)),70 which could serve as a better electrode material compared to standard noble metals for cellular interfaces.71 In this work, we use a carbon nanoparticle ink to print the interfacing electrode material due to its electrochemical stability, wide electrochemical water window, and low impedance for electrical sensing and stimulation of cellular activity.72,73,74,75,76,77,78,79 We shed light on the process required for ink-jet printing high-resolution MEAs, feedlines, and passivation layers on a soft substrate. Functional mircoelectrode arrays are printed on PDMS, agarose, and even gelatin-based substrates including candies (gummy bears). Additionally, we introduce a printed hydrogel MEA for cell recordings, which is challenging to directly pattern using classical photolithographic methods. We demonstrate the functionality of the MEA by extracellular recording of action potentials from HL-1 cells.

Results and discussion

Printed soft MEA arrays

The overall design and fabrication process of the printed PDMS MEAs are shown in Fig. 1. The schematic presents the design of a line MEA (Fig. 1a–c) and the subsequent printing of inks on a PDMS substrate (Fig. 1d–f). Developing functional electronics on soft materials requires the printing of electrically continuous lines. Therefore, it is important that adjacent ink droplets, which are deposited on the substrate, are connected to a functional entity. However, this is typically difficult to achieve with a water-based ink containing the functional material and a hydrophobic substrate such as PDMS. Our pristine PDMS substrates exhibited a static contact angle of 109° ± 3 with a surface energy of 25 ± 4 mN/m, which is consistent with observations reported in the literature.80 The high contact angle consequently leads to a breakup of the deposited liquid structure due to dewetting, effectively causing discrete islands of printed liquid similar to condensed water droplets on a cold surface. One way to avoid dewetting of ink on PDMS surfaces is to implement a multilevel matrix deposition method, in which few drops separated by a defined distance in the X- and Y-direction are printed each time until a whole film is completed. This method has been adapted to print on a wide variety of substrates as reported previously.81,82 Another way to overcome the dewetting problem is to increase the substrate surface energy via oxygen plasma exposure.83 Although the commonly employed method of oxygen plasma treatment does improve the wettability of PDMS it causes spontaneous cracking of the PDMS surface. Upon drying of the printed film, this causes microscopic cracks and consequently failure of the feedline connection. To circumvent this problem, we investigated the alternative of modifying the PDMS surface using (3-mercaptopropyl) trimethoxysilane (MPTMS), which has recently evolved for improving the surface wettability as well as enhancing the adhesion of the deposited ink and preventing crack formation.82,84 MPTMS has two different functional groups at its terminals: a methoxy (–OCH3) group, which binds to the PDMS and a thiol (–SH) group, which changes the surface properties of the pristine PDMS.57,82,85 Here, we modulate the wetting degree of PDMS to pattern large (>1 mm) and small (<0.03 mm) structures required for developing MEA. In order to fine tune the PDMS surface energy to meet the requirements of our printing resolution, we controlled the degree of surface modification using MPTMS by changing the incubation time. Figure 2a shows the printing results of carbon ink on the PDMS substrate with different MPTMS incubation using the same line pattern. As seen in Fig. 2b, the water contact angle decreases with longer incubation time. Optimal conditions for continuously formed lines are observed after approximately 60 min corresponding to a contact angle of ~80°. Larger features can be patterned on PDMS by further increasing the wettability of the surface. Thus, a wide range of structures ranging from high-resolution individual lines of 30 μm up to a several millimeter patches can be printed by tuning the MPTMS incubation time.

Fig. 1
figure 1

Sketch of the device principle and printing procedure. a Step 1: outer feedlines are printed with a silver nanoparticle ink on a 12 × 12 mm² substrate. b Step 2: inner feedlines and MEAs are printed with carbon nanoparticle ink. c Step 3: A 9 × 9 mm² passivation layer is printed with polyimide ink (PI). df Microscopic images of the successive printing process of a carbon MEA on PDMS subsequently depositing d silver ink, e carbon ink, and f PI ink. Scale bars represent 200 μm. g Principle of the recording of action potentials from electrogenic cells using the printed soft MEA

Fig. 2
figure 2

Effect of MPTMS incubation on printed line formation. a Microscopic images of printed carbon lines with a fixed drop spacing of 20 µm versus the incubation time of MPTMS, scale bars represent 200 µm. b Schematic drawing of the ink spreading on a PDMS substrate. c Measured contact angles of a water drop as a function of the incubation time of MPTMS. d Optical microscopy images of printed PI ink on PDMS as a function of oxygen plasma exposure time. The design of the structure to be printed was a rectangular shape as shown in the third image of this sequence. All structures were printed with a fixed DPI (dots per inch) of 846. Scale bars represent 200 µm

Fundamentally, the microfabrication of electronic devices for applications in wet environments requires the patterning of a dielectric material, which serves as a passivation layer. In the case of MEA chips for electrophysiological or electrochemical recordings, the passivation layer must be compatible with the cells and provide insulation of the feedlines versus conducting electrolytes. Moreover, it needs to be chemically resistant against the medium supporting the cell culture. For this reason, we chose a polyimide insulation material, which has already demonstrated good passivation properties in clean-room fabricated MEA arrays.86 As we are aiming for a fully printed device, it is crucial to choose a suitable ink and to understand possible factors that could induce passivation failure. One common problem in printing continuous passivation films is dewetting, which is especially important for covering areas that are larger than two printed drops.87 Considerable work has been reported in the literature to avoid dewetting when printing films either by using algorithm-generated spacings or by radically changing the surface energy of the substrate.88,89 As we discussed earlier, one can tune the PDMS wetting degree for water-based inks to a desired state by controlling the incubation time with MPTMS. However, it is not possible to combine two different wetting states: one for printing fine structures while the other enables the printing of large-area films. Obviously, there is no universal wetting state that can meet all the requirements for printing small and large structures with different inks in a single pass. Thus, the development of printed multilayer arrays with carbon MEAs and passivation films requires a change in the surface energy of the PDMS between the individual printing steps. This change has to match the requirements for optimal wetting in order to form a continuous passivation film. We have recently shown that the MPTMS treatment acts as a protective layer and prevents spontaneous crack formation upon plasma treatment.84 Here, we took advantage of the chemical modification conducted prior to printing the MEAs and investigated the influence of the plasma dose on the wetting of the passivation ink on PDMS surfaces (see Fig. 2d). This way it was possible to print the passivation layer in a controlled way covering the feedlines without compromising the MEAs or the bond pads for connecting the headstage. The design and fabrication process of our PDMS MEA are shown in Fig. 1. The passivation was printed to expose only a small carbon MEA to the liquid as seen in Fig. 1f. The average width of a single MEA was 30 ± 1.5 μm (n = 30) with a center-to-center distance of 40 ± 1.5 μm (n = 30). Details on the characterization of the printed MEA, including electrical, electrochemical, and mechanical evaluation of the device can be found in the Supplementary Information (Figures S1 and S2).

Extracellular recordings using PDMS and hydrogel MEAs

In order to evaluate the functionality of the soft MEA, we performed extracellular recordings of cardiomyocyte-like HL-1 cells90 (Fig. 3). To this end, the bond pads of the printed PDMS MEAs (Fig. 3a) where electrically connected to a carrier board (Fig. 3b). Cells were cultured on the MEAs until a confluent cell layer developed and evaluated for viability using fluorescent live–dead staining (Fig. 3c). After a few days in culture, the HL-1 cells were spontaneously contracting, confirming the compatibility of the printed devices with the active cell layer. We used the PDMS MEAs to locally monitor action potential generation within the cell culture. Figure 3d shows an example of the electrical signals recorded on different electrodes on the same PDMS MEA. The HL-1 cells generated spontaneous action potentials. The maximum cell signal amplitude recorded was 906 μVpp at a background noise of about 62 µVpp, which is comparable to reported values of HL-1 recordings using gold MEAs on polyimide substrates,68 as well as advanced clean-room-fabricated cell interfaces such as nanocavity electrodes91 and nanopillar electrodes92 on ceramic substrates.

Fig. 3
figure 3

Final device and demonstration of the printed MEAs on PDMS. a Photograph of printed carbon MEAs on a PDMS substrate. b Final chip bonded to a printed circuit board and encapsulated for use in cell culture. c Fluorescence microscopy image of live/dead staining of HL-1 cells growing on a PDMS MEA (scale bar 100 µm). Live cells appear green and a single dead cell red. d Action potential recording of different HL-1 cells growing on the same PDMS MEA over a time span of 2 s. e Magnified display of a single recorded trace. f Overlay of two individual extracellular recordings. The temporal shift in the two signals indicates the signal propagation across the cell network

To show the versatility of our approach for developing MEA structures on soft materials, we investigated printed MEAs on hydrogel substrates such as agarose, gelatin, and edible gummy bears (see Fig. 4). The functionality of the hydrogel MEAs was evaluated by performing cellular recording of action potentials. Cells were plated on a gelatin-based gummy bear MEA and after a few days in culture, electrical recordings were performed. Figure 4e shows five exemplary traces from the same MEA. We observed a maximum amplitude of 442 μVpp at background noise of approximately 80 μVpp on the hydrogel MEA.

Fig. 4
figure 4

Printed MEA on soft hydrogel substrates for extracellular recording. a Photograph of printed carbon MEAs on a gummy bear substrate. b Final chip bonded to printed circuit board with HL-1 cells culture. c Exemplary photograph of a printed MEA on a gelatin substrate. d Action potential recording from HL-1 cells using printed carbon MEAs on a gummy bear substrate, traces are offset in y-direction for clarity of representation. e HL-1 cells stimulation with noradrenaline (NA). f Photograph of a printed MEA on an agarose substrate (scale bar 10 mm). g Microscopic image of printed MEA on an agarose substrate (scale bar: 100 µm)

To further examine the specificity of the recorded signals, the cells were chemically stimulated with noradrenaline (NA), a catecholamine that triggers a sympathetic response. As expected, the firing rate of the spontaneous action potential was increased from 1.3 Hz up to 1.8 Hz upon addition of 4 µL of a 10 mM NA solution (Fig. 4e). We have chosen gelatin as a substrate for bioelectronic interfaces for several reasons. First, it is a soft material with a Young’s modulus in the range of 100–102 kPa.93 Second, it has been used as a scaffold for tissue engineering and shown to be a promising material for repairing traumatic injuries to the brain as it improves the brain-tissue reconstruction.94 Third, it exhibits antibacterial and hemostatic effects,95,96 which is beneficial for recovery after mechanically inserting the electrodes to the desired tissue location.97,98 Thus, it can be expected that adding functionality to gelatin-based devices will have an impact on future implant technology. We believe that our approach of printing MEA structures on hydrogel materials such as agarose and gelatin will provide opportunities for developing soft as well as low-cost and disposable functional devices. Furthermore, the concept can be applied in future work for the development of dense multilayer MEA arrays to meet requirements for neuroscience applications in vitro and in vivo.


In this work, we demonstrated the development and application of printed MEA arrays on soft substrates including PDMS and hydrogels. To this end, we introduced a straightforward printing process, which exploits controlled wetting properties of carbon and polyimide inks on PDMS, overcoming major problems that typically arise in printing structures at different spatial scales. We presented a printed hydrogel MEA for bioelectronic applications. The soft MEAs were applied for localized recordings of action potentials from HL-1 cells, validating the suitability of the printed devices for electrophysiological measurements. This work represents an important step toward the design of soft hydrogel-based bioelectronic devices using ink-jet printing.

We believe that the approach presented in this paper will allow for rapid prototyping of disposable sensor array structures on a variety of soft substrates for in vitro as well as in vivo applications. Potentially, future devices could be directly developed on gelatin taken from an individual and transplanted onto the tissue in the same organism.


Ink-jet printing

MEA arrays were fabricated using an OmniJet 300 ink-jet printer (UniJet Co., Republic of Korea). Silver nanoparticle ink was used for the fabrication of feedlines on PDMS samples (Nano-silver ink DGP 40LT-15C Advanced Nano Products, Co., Ltd, South Korea). The gelatin (gummy bear) and the agarose-based MEAs were printed using carbon ink prepared in our laboratory. Prior to printing, all inks were sonicated for 5 min and filtered with 0.2 µm PVDF syringe filter. Printing was performed using 1 pL DMC cartridges (Fujifilm Dimatix Inc., USA) for printing silver and carbon inks. Ten picoliters cartridges were used for printing the polyimide ink (PIN-6400-001; Chisso Corp., Japan) filtered with a 0.2 µm PTFE syringe filter. Printing parameters used for printing on all soft substrates were fixed to a jetting frequency of 2 kHz, a drop spacing of 20 μm, a substrate holder and printhead temperature of 25 °C, a jetting period of 16 μs, and a jetting voltage of 40 V. The distance between the printhead and substrates was set to 250 μm for planar substrates. For non-planar substrates (maximum height 3 mm) as shown in Fig. 4a, the distance between the printhead and the substrate holder was set to 10 mm. The sintering and curing steps were conducted on a precision hot plate (CT10; Harry Gestigkeit GmbH, Germany) in case of PDMS. In case of gelatin (gummy bears) and agarose gel, samples were cured using a photonic sintering system (PulseForge 1200; NovaCentrix, USA) with six 500 µs pulses at 100 V. All MEA chips were cut into 12 mm × 12 mm pieces.

Carbon ink preparation

The carbon ink was prepared as reported previously.66,67,99 Briefly, 1 g of carbon black (Orion Carbons) in 5 g of a 50/50 wt% mixture of ethylene glycol and water was milled with 100 μm yttrium zirconium beads at 1100 rpm for 1 h in a Pulverisette 7 ball mill (Fritsch, Germany). The milled mass was diluted with further 10 g of ethylene glycol/water mixture to adjust the viscosity with vigorous stirring. Next, 0.2 wt% polyacrylic acid (Sigma Aldrich, 35 wt% solution) was added to improve the adhesion of the ink to the substrate. Finally, the solution was filtered through a 0.45 μm filter to obtain the final ink.

MEA design layout

As shown schematically in Fig. 1, the multi-electrode microchip consists of 64 individually addressable carbon electrodes with a center-to-center spacing of 40 ± 1.5 μm. The size of the final printed chip is 12 × 12 mm. Figure 1 shows a schematic illustration of the design, the chip, and the printing process on PDMS substrates. In a first step, silver nanoparticle ink is printed to form the feedlines and bond pads of the chip as shown in Fig. 1a. The contact pads are used later for connecting the MEA to the electronic amplifier system. Next, the carbon nanoparticle ink is printed to cover the silver feedline and form an electrode area for cellular measurements (Fig. 1b). This procedure ensures that only carbon is exposed to the electrolyte solution. Finally, the polyimide passivation ink (JNC Corporation, Japan) is printed to cover the chip except for the contact pads and small electrode opening in the center of the chip. For the other substrates, carbon ink was printed directly on the substrate followed by the passivation layer. The final electrode area was 1900 ± 300 µm² (n = 30). The average resistance of the carbon feedlines was below 1 kΩ.

Ink characterization

The particle size of the carbon ink was determined using a scanning electron microscope (SEM-Nova Nano FEI, USA) with an accelerating voltage of 3 kV. We observed nanoparticle diameters in the range of 40–200 nm. The 3-D profile for the printed MEAs was measured using a laser confocal microscope (Keyence VK- X130K) at a wavelength of 658 nm. The height mapping in the z-direction revealed a homogeneous topography and a thickness of 500 and 600 nm for the printed silver and carbon layers, respectively.

The electrical resistance of the printed carbon lines was measured after exposure to a temperature of 120 °C. We performed a four-terminal resistivity measurements yielding a sheet resistance of 47 ± 12 Ω/sq (n = 10); further details can be found in the Supplementary Information.

PDMS substrate preparation

Step-1 silicon wafer coating

A silicon wafer (5-inch) was first silanized in order to generate a repellent surface. The coating was achieved by covalently linking perfluorooctyltrichlorosilane (FOTCS) 97% (from Alfa Aesar) via vapor deposition. First, the wafer was activated in oxygen plasma (0.8 mbar, 3 min, 80 W) and silanized with FOTCS at 45 mbar for 1.5 h in an argon atmosphere.

Step-2 PDMS casting

The silicone elastomer PDMS substrates (Sylgard 184 from Dow Corning) were prepared by manually mixing the curing agent and base material at a ratio of 1:10 by mass. This ratio results in a substrate with Young’s modulus of ~2.5 MPa as described previously.100 In order to cast the elastomer substrate, 10 mL of the mixture was poured onto a FOTCS-coated silicon wafer and subsequently degassed in a vacuum chamber at room temperature. The PDMS was cured at 60 °C overnight. Afterwards, the PDMS substrate was easily peeled from the wafer.

Step-3 PDMS surface modification

The surface modification of PDMS substrate was carried out as mentioned previously.84 Briefly, PDMS samples were coated by (3-mercaptopropyl) trimethoxysilane (MPTMS 95%—Sigma Aldrich). To this end, the PDMS substrate was immersed in 1:200 solution of MPTMS in ethanol for 1 h. Afterwards, the sample was rinsed with deionized water (Millipore Milli-Q System, 18 MΩ/cm). Finally, the PDMS samples were immersed in 1 mM HCl solution for 1 h and washed again with deionized water. The PDMS substrates were kept in the refrigerator and were used within 2 days.

Oxygen plasma

An oxygen plasma chamber (100-E plasma system; Technics Plasma GmbH) was used for tuning the hydrophilicity of the PDMS surface. The typical dose as used prior to passivation printing was applied exposing the PDMS surface for 150 s at 40 W and 0.2 mbar pressure, unless stated otherwise.

Gelatin-based substrate preparation

For the MEAs printed on gummy bear substrates, commercial gummy bears from Haribo® (Haribo GmbH & Co. KG, Bonn, Germany) were melted and casted on a silicon wafer. The casted substrate was cleaned by ethanol and washed with deionized water. Finally, the substrate was immersed in deionized water for 8 h before printing. For the gelatin substrates, 20% w/v gelatin solution was prepared by soaking gelatin powder (Sigma-Aldrich®, from porcine skin) in 100 mL deionized water for 2 h. Next, the solution was heated to approximately 70 °C until a homogeneous solution was formed. Finally, the warm solution was poured into a Petri dish and allowed to from a gel at room temperature.

Agarose substrate preparation

Gels were prepared by dissolving 3 g of agarose (Sigma-Aldrich®) dissolved in 100 mL tris/acetate buffer. The gel was poured in Petri dish and kept in the refrigerator. The thickness of the agarose gel was 3.5–4.0 mm.

Contact angle measurement

The wetting behavior of the PDMS substrate was investigated by contact angle measurement. The sessile drop technique using an OCA H200 instrument (DataPhysics Instruments GmbH) was performed at room temperature. A 2 μL drop of deionized water was dispensed on top of the PDMS surface. The acquired image of the water on the sample was taken using an integrated camera. The drop profile of a liquid–vapor interface was extracted and fitted by the Young–Laplace function provided by the OCA H200 software. The contact angle at the liquid–solid interface was assigned according to the fitted profile. Measurements were repeated five times for each substrate.

Electrical characterization

The resistance of the printed test structures was measured using a multimeter (Voltcraft Plus VC 960, Conrad, Germany). The sheet resistance was measured using a four point probe (Jandel CYL-HM21, Bridge Technology, USA).

Electrochemical characterization

Prior to electrochemical measurements, all chips were cleaned by incubation with ethanol and deionized water for 5 min, each. A glass ring with a height of 10 mm and a diameter of 7 mm was glued to the MEAs using PDMS in order to create a reservoir. Electrochemical experiments were performed using a Biological potentiostat (VSP-300 potentiostat from BioLogic Science Instruments). All experiments were carried out in a supporting electrolyte of phosphate-buffered saline (PBS 1×, pH 7.4) using an equimolar mixture of potassium ferricyanide (1 mM) and potassium ferrocyanide (1 mM) (Sigma Aldrich) as a redox tracer dissolved in PBS. The signals were recorded against an Ag/AgCl reference electrode (Super Dri-ref SDR 2; World Precision Instruments, USA).

Cellular recording

Three PDMS MEAs and one gummy-bear-based MEA were sterilized by incubation with 70% ethanol for 10 min, followed by rinsing with sterile distilled water thrice. They were coated with 2.5 µg/cm2 fibronectin from bovine plasma (Sigma Aldrich, Schnelldorf, Germany) in calcium and magnesium free PBS (Life Technologies GmbH, Darmstadt, Germany) at 37 °C for 1 h. The chips were rinsed once with supplemented Claycomb medium just before cell seeding. Cardiomyocyte-like HL-1 cells were maintained in Claycomb medium (Sigma Aldrich, Steinheim, Germany) supplemented with 10 v% fetal bovine serum (Life Technologies GmbH, Darmstadt, Germany), 100 μg/ml penicillin–streptomycin (Life Technologies GmbH, Darmstadt, Germany), 0.1 mM (±)-Norepinephrine (+)-bitartrate salt (Noradrenaline, Sigma Aldrich, Steinheim, Germany) and 2 mM l-glutamine (Life Technologies GmbH, Darmstadt, Germany) in a humidified incubator at 37 °C and 5% CO2. The medium was changed daily. Once confluency was reached, the contracting cell layer was first washed and then detached by incubation with 0.05% trypsin-EDTA (Life Technologies GmbH, Darmstadt, Germany) at 37 °C. Trypsin digestion was then inhibited by addition of supplemented Claycomb medium and the cells were sedimented by centrifugation at 200 rcf for 5 min. The cells were resuspended in pre-warmed, supplemented Claycomb medium and 100 µL were added to the center of each chip. The cells were left to adhere in a humidified incubator at 37 °C and 5% CO2 for 30 min. Afterwards, 500 µL of medium were added to each chip. The medium was exchanged daily until confluency was reached. Once the confluent cell layer was beating (after ~2 days) action potentials were recorded employing a 64 channel MEA amplifier system developed in-house. The system consists of a headstage connected to a main amplifier, which is connected to the controlling PC via a 16-bit A/D converter (USB-6255; National Instruments, Austin, Texas, USA). Data acquisition is controlled through an in-house developed software, which allows the definition of the recording parameters such as gain and filter settings. We limited the effective bandwidth with a bandpass filter from 1 to 3 kHz for all measurements reported.

Data Availability Statement

All experimental data generated or analyzed during this work are included in the article and the Supplementary Information Files.