High-Aspect-Ratio Semiconducting Polymer Pillars for 3D Cell Cultures.

Hybrid interfaces between living cells and nano/microstructured scaffolds have huge application potential in biotechnology, spanning from regenerative medicine and stem cell therapies to localized drug delivery and from biosensing and tissue engineering to neural computing. However, 3D architectures based on semiconducting polymers, endowed with responsivity to visible light, have never been considered. Here, we apply for the first time a push-coating technique to realize high aspect ratio polymeric pillars, based on polythiophene, showing optimal biocompatibility and allowing for the realization of soft, 3D cell cultures of both primary neurons and cell line models. HEK-293 cells cultured on top of polymer pillars display a remarkable change in the cell morphology and a sizable enhancement of the membrane capacitance due to the cell membrane thinning in correspondence to the pillars' top surface, without negatively affecting cell proliferation. Electrophysiology properties and synapse number of primary neurons are also very well preserved. In perspective, high aspect ratio semiconducting polymer pillars may find interesting applications as soft, photoactive elements for cell activity sensing and modulation.

Introduction

In recent years, materials scientists, biotechnologists, and neuroscientists have invested joint, extensive efforts toward the realization of three-dimensional structures suitable for interfacing with living cells and tissues. The most recent developments in the realization of structured surfaces suitable for cell interfacing, with regularly or randomly arranged nano- and microstructures protruding from a flat surface, have followed multiple paths. Different geometries (nanowires, nanopillars, mushrooms, nanocavities),[1] structure densities, and materials and fabrication technologies, as well as diverse cellular models, have been investigated. The huge interest in nano- and microstructures as cell substrates or 3D scaffolds originates from a number of promising applications, spanning from regenerative medicine to neuroscience and from pharmacology and physiology to neural computing and tissue engineering.[2−7] Some notable examples include the realization of nanostructures as low-impedance electrodes for electrical recording of neural culture activity, biosensors with enhanced signal-to-noise ratios, devices for cell proliferation and motility control, cell culture controlled patterning, induction of stem cell differentiation, and highly localized delivery of various functional molecules.[8−12]

Reported nanostructures mainly rely on the use of inorganic metals and semiconductors, such as gold, silicon, and silicon oxides, which guarantee optimal repeatability through the use of standard lithography fabrication, as well as excellent electrical conductivity.[8,13,14] In particular, nanostructured conducting electrodes allow for sizable reduction of the cell/device interface impedance, combined with spatial resolution at the single-cell level and parallelization of the excitation and/or recording of the cell electrical activity from multiple sites.[4,15,16] Electrically inert polymer substrates (among many others, PDMS, SU-8, polycarbonate, PLA) have been largely employed as well, mainly for tissue engineering and regenerative medicine applications.[17,18] A variety of synthetic and bioderived polymers have been also developed ad hoc as biocompatible scaffolds for 3D cell cultures. Their distinct advantages over inorganic materials comprise easier and faster processing, increased design flexibility and versatility, softness, and outstanding biocompatibility. More recently, the use of electrically conducting polymers has started to attract considerable attention due to the opportunity to integrate their optimal mechanical properties, typical of polymer soft materials, with unprecedented ionic–electronic mixed conduction capabilities, peculiar of conducting polymers. A notable example is represented by nanostructures made of PEDOT:PSS, a workhorse material for the realization of highly performing electrochemical sensors and actuators.[19,20] Based on their well-known electroactive properties, PEDOT:PSS and other conducting polymers have been proposed for the realization of bioscaffolds,[21] able to electrically modulate the activity of excitable cells and to guide cell proliferation and stem cell differentiation.[22−28] To this goal, PEDOT:PSS has been processed in the form of hydrogel-like scaffolds, sponges, and foams.[29,30] In many cases, device transparency remains an issue, thus hampering the use of bioimaging analysis techniques. Very recently, an interesting work by Inal and colleagues reported the fabrication of a PEDOT:PSS-based macroporous device, integrating electrochemical sensing capabilities with in situ live cell monitoring.[31] Besides the use of ionic- and electronically conducting polymers like PEDOT:PSS, the use of 3D nanostructures based on semiconducting polymers, with distinctive optoeletronic properties and visible-light emission/detection capabilities, has been very rarely taken into account. The realization of ordered polymer-based nanostructures, characterized by a high aspect ratio (HAR), has been considered to a much lower extent than disordered hydrogel-like structures, especially for biotechnology applications.[32,33]

In this work, we report fabrication of HAR polymer pillars entirely made of polythiophene (rr-P3HT), a well-known semiconducting polymer with distinctive optoelectronic features and good biocompatibility.[34−36] rr-P3HT pillar fabrication is based on an original, highly repeatable push-coating technique.[37,38] Polymer pillar arrays display optimal biocompatibility properties, serving as ideal cell culturing substrates for both neurons (primary cortical neurons) and cell lines (human embryonic kidney cells, HEK-293). rr-P3HT structured substrates exhibit excellent elasticity and flexibility while maintaining robustness and stability in aqueous environments for several weeks. The pillared structure leads to the realization of three-dimensional cell cultures, with a remarkable change in the cell morphology. We observe a sizable enhancement of the cell membrane capacitance due to the cell membrane thinning in correspondence to the pillars’ top surface. Importantly, this neither negatively affects cell proliferation nor leads to alteration of the neuronal electrophysiology properties on a macroscopic scale. In addition, while neurons grown onto polymer pillars show a slight decrease in dendritic arborization, the number of both excitatory and inhibitory synapses is not altered. HAR rr-P3HT pillars may act as soft, electrically active elements in a variety of perspective applications in biotechnology, including locally confined recording and actuation of living cell activity and realization of polymer scaffolds for spatially controlled drug delivery. To the best of our knowledge, light-sensitive nanostructured devices recently reported in the literature are all based on inorganic semiconductors.[39−42] In this perspective, our work represents also the first necessary step toward the future realization of optically responsive polymer 3D structures for photomechanical, photochemical, and/or photoelectrical modulation of cell activity, proliferation, and differentiation.

Experimental Section

Materials

rr-P3HT (purity 99.995%, molecular weight 15000–45000), o-dichlorobenzene, Dulbecco’s modified Eagle’s medium (DMEM), trypsin–EDTA, penicillin, phosphate-buffered saline (PBS) tablets, streptomycin, fibronectin (from bovine plasma), glutaraldehyde, paraformaldehyde, sucrose, gelatin, Triton X-100, phalloidin-FITC, l-glutamate, glutamine, and 4′,6-diamidine-2′-phenylindole dihydrochloride (DAPI) were purchased from Sigma Aldrich. PDMS elastomer (Sylgard 184) was purchased from Dow Corning. Glass/ITO substrates (15 Ω/sq) were purchased from Xin Yan Technology. Fetal bovine serum (FBS) was purchased from Euroclone. HOECHST 33342 and NucGreen Dead 488 ReadyProbes reagent were purchased from Thermo Fisher. AntiMAP2 (lot number: GR143561) was purchased from Abcam. Anti-PSD95 (cat. n. 444-1LC-49) was purchased from NeuroMab. Anti-vGAT (cat. n. 131002), anti-v-GLUT (cat. n. 135303), and anti-Synapsin (cat. n. 106002) were purchased from Synaptic Systems. Secondary antibodies FITC-conjugated anti-mouse (cat. n. 715-095-150), FITC-conjugated anti-rabbit (cat. n. 711-095-152), and Cy5-conjugated anti-mouse (cat. n. 715-175-150) were purchased from Jackson ImmunoResearch. All chemicals were used without any further purification.

PDMS Mold Fabrication

PDMS precursor was mixed with the curing agent (10:1 volume ratio) and left in vacuum for 30 min in order to remove air bubbles formed during the mixing process. The degassed mixture was put inside a glass Petri dish and left in an oven for 4 h at 65 °C. The bottom part of the Petri dish was covered with a silicon wafer to obtain a highly planar PDMS surface. After the thermal curing, a 4 × 5 mm2 PDMS area was patterned with a microhole array of 2 μm diameter and 7 μm pitch (distance between the centers of two adjacent holes) using femtosecond pulse laser micromachining. The system is equipped with a regenerative amplified mode-locked femtosecond laser source based on an Yb:KGW active medium (Light Conversion, Pharos) whose pulses at a fundamental wavelength of 1030 nm are characterized by a duration of 240 fs, repetition rate up to 1 MHz, and pulse energy up to 0.2 mJ. In order to fabricate the desired array of holes, a combinatorial optimization process of the writing parameters was carried out. The ideal parameters set, obtained by using the second harmonic λ = 515 nm, were 50 pulses on the same position, 100 kHz repetition frequency, and 15 mW average power (see the Supporting Information for other parameter combinations). The laser beam is statically focused on the surface substrate through a microscope objective (20×, Mitutoyo, NA 0.40). The 2D structure is achieved by moving the sample, placed on a high-precision three-axis air-bearing translation stage (Aerotech, ABL 1000 series) with a resolution of up to 20 nm. Usually, direct laser ablation performed on a substrate (in our case, PDMS) creates a large amount of debris that strongly affects the quality of both the surface and the geometry of the obtained structure (microholes) in terms of morphology, aspect ratio, and homogeneity. In order to overcome this serious problem and to obtain a high degree of repeatability in the realization of the holes, the microperforation of the PDMS substrate was then performed in a controlled atmosphere using a vacuum chamber (10–1 mbar). The low pressure promotes the separation of the ablated material from the surface that is free to ″fly″ away from the unprocessed area due to the increase in its average free path and the low kinetic energy. In this way, the obtained structure geometry is more easily controllable and of better quality with a residual roughness of about 2 orders of magnitude lower than the conventional ablation in air. After the laser process, the mold was washed with EtOH for an additional surface cleaning.

rr-P3HT Sample Fabrication

Commercial glass/ITO slabs were cut into 18 × 18 mm2 slides, washed by subsequent rinsing in an ultrasonic bath by using distilled water, acetone, and isopropanol (10 min each), and then dried with a N2 flux. rr-P3HT was dissolved in o-dichlorobenzene (20 g L–1) and stirred for one night at 50 °C. A 1 μL drop of the rr-P3HT solution was pushed onto the cleaned glass/ITO surface using the micropatterned PDMS mold. After a thermal treatment at 90 °C for 2 min, the mold was gently removed ending up with a 4 × 5 mm2 rr-P3HT pillar array surrounded by a flat rr-P3HT region deposited on top of glass/ITO substrates.

Glass/ITO/P3HT flat samples for electrochemical measurements were prepared by spin-coating (speed 1600 rpm, acceleration 1600 rpm s–1) a 20 g L–1 rr-P3HT solution in o-dichlorobenzene on top of 18 × 18 mm2 glass/ITO slides.

Electrochemical Characterization

Electrochemical impedance spectroscopy (EIS) was carried out in Krebs–Ringer Hepes (KRH) extracellular solution (composition [mM]: 135 NaCl, 5.4 KCl, 5 HEPES, 10 glucose, 1.8 CaCl2, 1 MgCl2) at room temperature using an Autolab potentiostat PGstat 302 (Metrohm). An electrochemical cell in a three-electrode configuration was employed, comprising the planar/micropillars rr-P3HT devices as the working electrode, a platinum wire as the counter electrode, and saturated-KCl Ag/AgCl as the reference electrode. The planar glass/ITO/P3HT part of the flat/pillar devices was removed in order to guarantee that only the impedance contribution from the pillar array is taken into account. For the comparison with the planar rr-P3HT, flat glass/ITO/P3HT devices were employed. Impedance spectra were recorded in the 0.01 Hz to 100 kHz frequency range with an ac amplitude of 0.02 V by applying a constant bias equal to the device open circuit potential (0.11 and 0.08 V for the rr-P3HT flat and pillar cases, respectively). Nova 1.8 software was used for data analysis.

Cell Culture Preparation

rr-P3HT flat and microstructured devices were sterilized in an oven at 120 °C for 2 h. In the case of HEK-293 cells, a layer of fibronectin (2 μg mL–1 in PBS buffer solution) was deposited on the sample surface and incubated for 1 h at 37 °C, in order to promote cellular adhesion. Excess fibronectin was then removed by rinsing with PBS prior to cell plating. HEK-293 cells were cultured in cell culture flasks containing Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS), 100 μg mL–1 penicillin, and 100 μg mL–1 streptomycin. Culture flasks were maintained in a humidified incubator (Forma Series II water jacketed CO2 incubator, Thermo Fisher) at 37 °C with 5% CO2. When at confluence, HEK-293 cells were enzymatically dispersed using trypsin–EDTA, then plated on the different samples at a concentration of 20000 cells cm–2, and maintained in the incubator at 37 °C with 5% CO2.

Primary cortical neuron cultures were prepared from 18 to 19 day old rat embryos (pregnant females were obtained from Charles River Laboratories) as described elsewhere.[43] Prior to neuron plating, the sterilized devices were treated with a poly-l-lysine solution (1 mg/100 mL of borate buffer pH 7.4) overnight at room temperature. After washing three times with sterile water, primary cortical neurons were plated on the substrates at a density of 60000–75000 cells cm–2 in a Neurobasal medium supplemented with 2% B27 prepared in the laboratory, l-glutamate (at final concentration of 10 mM), glutamine (at final concentration of 2 mM), and 1% penicillin and streptomycin. The cells were maintained in the incubator at 37 °C with 5% CO2. At days in vitro 4 (4 DIV), half of the medium was replaced with a fresh medium without l-glutamate, and neurons were maintained in this medium until they were fixed for the staining at 14 DIV.

Scanning Electron Microscopy (SEM)

To evaluate cell morphology and spreading, HEK-293 cells (1 DIV) and primary neurons (14 DIV) plated on flat/pillar rr-P3HT substrates were prepared for SEM with the following procedure: (1) fixation in glutaraldehyde 2.5% in PBS overnight at 6 °C; (2) immersion in increasing concentrations of ethanol (between 20 and 100%, in steps of +10%, 20 min for each concentration) followed by air drying; and (3) evaporation of a thin gold layer on top of the sample surface (thickness 6 nm, 1.5 nm Cr adhesion layer). rr-P3HT-pillar-based devices without cells instead did not require any treatment prior to SEM image acquisition. All SEM micrographs were acquired by using a TESCAN MIRA III scanning electron microscope (operating voltage 4 kV, working distance 18 mm, stage tilting angle 45°).

Cell Viability Assay

The evaluation of the viability of all cell types employed in this work (primary neurons after 14 DIV, HEK-293 cells after 1 DIV) was accomplished by the HOECHST 33342/NucGreen Dead 488 ReadyProbes assay. The flat/pillar rr-P3HT substrates were incubated in KRH extracellular solution containing the two dyes (HOECHST 33342 10 μg mL–1 and NucGreen Dead 488 ReadyProbes Reagent 2 drops mL–1) for 5 min protected from ambient light. The samples were then washed with extracellular solution, and multiple images were acquired with a Nikon Eclipse Ti-S epifluorescence inverted microscope. Standard DAPI and FITC filter sets were employed for HOECHST and NucGreen, respectively. The percentage of viable cells was estimated by counting the total number of cell nuclei (stained by HOECHST) and the total number of dead cell nuclei (stained by NucGreen). The results obtained on the flat and microstructured regions were compared (n = 450 cells for each substrate type).

Cell Morphology

HEK-293 cells grown on fibronectin-coated rr-P3HT flat/pillar substrates for 2 DIV were washed twice with PBS and fixed for 15 min in 4% paraformaldehyde and 4% sucrose in 0.12 M sodium phosphate buffer pH 7.4, at RT. Labeling with phalloidin-FITC was applied in GDB buffer (30 mM phosphate buffer, pH 7.4, containing 0.2% gelatin, 0.5% Triton X-100, and 0.8 M NaCl) for 2 h at RT. Nuclei were marked with DAPI (5 min incubation in PBS). Fluorescence images were acquired with the same microscope employed for the viability assay, using standard FITC and DAPI filters set for recording the fluorescence emission of the phalloidin-FITC- and DAPI-stained actin and nuclei. Cell top-view surface area and shape parameters were quantified using ImageJ software. Cells shape was evaluated in terms of circularity c (4π × [cell area]/[cell perimeter]2, c = 1 indicates a circle, c = ∼0 indicates a highly elongated shape). The cell projection extension was evaluated by measuring the cell perimeter and by normalizing it to the cell top-view surface area. Mean values have been obtained by averaging over a statistical ensemble of n = 100 cells for each substrate type. Rat cortical neurons were fixed at 14 DIV in 4% paraformaldehyde plus 4% sucrose at room temperature. AntiMAP2 (1:200) was applied in GDB buffer (30 mM phosphate buffer pH 7.4, containing 0.2% gelatin, 0.5% Triton-X-100, and 0.8 M NaCl). Morphological analysis of dendrites was performed on the signal obtained by MAP2 staining, acquired using a confocal microscope (Zeiss LSM800) with a 40× objective and sequential acquisition setting at a resolution of 1024 × 1024 pixels. Sholl analysis was performed using NeuronStudio (Computational Neurobiology and Imaging Center, Mount Sinai School of Medicine, New York, NY) to evaluate the dendritic arborization and to measure the number of branching points. Labeled neurons were chosen randomly for quantification from three to six coverslips from two to three independent experiments. The number of neurons used for quantification is indicated in the figure legends. Statistical significance was determined by the one-way ANOVA Bonferroni post hoc test.

Electrophysiology

Electrophysiology was performed using a patch clamp set up based on an inverted fluorescence microscope (Nikon Eclipse Ti-S). Intracellular recordings of primary cortical neurons were carried out after 14 DIV with an Axopatch 200B (Axon Instruments) in a whole-cell configuration, using borosilicate glass pipettes (3–6 MΩ). Recordings were performed in KRH extracellular solution and in a current clamp configuration, with and without applying a current ramp (20 pA current steps, ranging from 0 up to 200 pA) for evaluating the neuron firing threshold. The patch pipette was filled with the following solution [mM]: 126 K-gluconate, 4 NaCl, 2 MgSO4, 0.2 CaCl2, 0.08 Bapta, 9.45 glucose, 5 Hepes, 3 ATP, and 0.1 GTP. Responses were amplified and stored with pCLAMP 10 (Axon Instruments), and resting membrane potentials were corrected for a −15 mV junction potential offset, evaluated using the pClamp10 junction potential calculation tool. All data were elaborated with Origin 9.0 software. The cell membrane capacitance (Cm) was measured by applying continuous square wave voltage pulses to the cells and by fitting the resulting current transients, using the Membrane Test tool of the pCLAMP 10 software (Axon instruments). The neuron top-view surface area was measured from bright field images using ImageJ software.

Primary Cortical Neuron Synapse Imaging

Rat cortical neurons were fixed at 14 DIV in 4% paraformaldehyde plus 4% sucrose at room temperature. Primary antibodies, namely, anti-PSD95 (1:200), anti-vGAT (1:100), anti-v-GLUT (1:200), and anti-Synapsin (1:400), and secondary antibodies, namely, FITC-conjugated anti-mouse (1:100), FITC-conjugated anti-rabbit (1:100), and Cy5-conjugated anti-mouse (1:100), were applied in GDB buffer (30 mM phosphate buffer pH 7.4, containing 0.2% gelatin, 0.5% Triton-X-100, and 0.8 M NaCl).

Confocal images were obtained using the same confocal microscope, objective, and image acquisition parameters employed for the neuron’s morphological analysis. Labeled neurons were chosen randomly for quantification from three to six coverslips from two to three independent experiments. The number of neurons used for quantification is indicated in the figure legends. Morphometric measurements were performed using MetaMorph image analysis software (Universal Imaging Corporation). Three single dendrites from each neuron in different conditions were manually traced. The number, area, and average intensity of synapses were automatically measured by a computer using MetaMorph image analysis software and logged into Microsoft Excel. The synapse density was calculated as the number of synapses per length of dendrites.

Statistical significance was determined by the one-way ANOVA Bonferroni post hoc test.

Results and Discussion

rr-P3HT Microstructure Fabrication and Morphology

The fabrication of the rr-P3HT microstructured substrates is carried out by taking advantage of the push-coating technique. Push-coating is an adaptable and eco-friendly process for manufacturing thin polymer films, originally developed for organic field effect transistors[38] and recently applied also to light-emitting diodes and solar cells.[37] In push-coating, a small volume of polymer solution is sandwiched between the substrate and a millimeter-thick PDMS layer. The solution spreading between the substrate and the PDMS through capillary forces facilitates the solvent diffusion into PDMS to form uniform thin films. Because the surface roughness of the PDMS layer determines the morphology of the push-coated film, this technique has been very recently proposed also to create nanostructured light-emitting polymer layers.[44] However, push-coating has never been employed to fabricate high aspect ratio pillared structures.

rr-P3HT nano- and microstructures are typically obtained by self-assembly or by imprinting processes.[45−47] If the former approach is not suitable for high aspect ratio structures, the latter requires to heat the solid polymer film well above its glass transition and crystallization temperatures before imprinting (around 170 °C for rr-P3HT) and to apply a controlled pressure on a lithographed hard stamp. By contrast, in our approach, the pillared film is formed from solution, in a few seconds, at room temperature, without any other applied pressure than the weight of the soft PDMS layer, and only a mild thermal treatment is required to facilitate the final stamp detachment. Typically, a very small amount of semiconducting polymer solution is employed, approximately 20 times smaller than in standard spin-coating deposition for the same coated surface and thickness.[37]

The first step to manufacture the rr-P3HT pillars by push-coating is the fabrication of the PDMS mold. The process, schematically depicted in Figure a, starts with the fabrication of a PDMS layer by thermal cross-linking of a commercial silicone oil precursor. In this work, a 5 mm thick stamp is employed since a PDMS thickness in the millimeter scale guarantees a good retention of the solvent inside the PDMS layer and optimal mechanical properties, necessary for the success of the push-coating procedure.[37] The second step consists in the micropatterning of the PDMS mold by femto-laser micromachining, in order to obtain a negative stamp of the pillars’ structure. The PDMS stamp is then employed in the push-coating process, ending up with an array of rr-P3HT pillars surrounded by a planar rr-P3HT region (thickness 180 nm, evaluated by profilometry measurements) deposited on top of standard glass/ITO substrates (Figure b). Since the process does not damage the PDMS mold, the same stamp can be used for many subsequent depositions in a highly reproducible way.

Figure 1

Device fabrication and morphology. (a) rr-P3HT micropillar fabrication process. (b) Photograph of the rr-P3HT-pillar-based device, taken at the end of the fabrication process. SEM images depicting the single rr-P3HT pillar structure (c) (scale bar, 2 μm) and an overview of the pillar array at decreasing magnification (d,e) (scale bars, 10 and 100 μm, respectively).

Our technique offers a number of advantages. Besides the excellent repeatability and speed of the overall process, the use of femtosecond micromachining to fabricate the mold allows finely tuning of the geometrical parameters of the polymer pillars (three-dimensional shape, size, aspect ratio, pitch). The combination of all these characteristics is known to be the key in the interaction with the living cells since it strongly influences cells’ viability, adhesion, and proliferation.[9,48−51] In fact, HAR micro- and nanopillars have been recently successfully employed to probe cellular tractions for enhancing stem cell differentiation and to gain intracellular access for drug delivery.[7,9,50,52−54] Other interesting applications are focusing on neurodegenerative diseases.[17] In all these cases, the fabrication technique should ideally be highly repeatable, fast, and simple while providing high versatility and capability to rapidly adapt parameters to the specific cellular model. The size and aspect ratio of pillars can be easily finely tuned by properly changing the writing parameters and conditions employed during the PDMS mold mask-less fabrication, namely, the laser source pulse power, number of pulses, repetition rate, and pressure of the vacuum chamber. All these parameters have been systematically changed for engineering the pillars’ characteristics (Figure S1), and an optimal set has been sorted out (see the Experimental Section). Scanning electron microscopy (SEM) images show the HAR conical shape of the individual pillars (Figure c) and provide an idea of the repeatability of the overall fabrication procedure (Figure d,e). Average pillars’ height, base diameter, and half-height width are 6.4 ± 0.3, about 2.3 ± 0.1, and about 1.2 ± 0.2 μm, respectively. The mean distance between two adjacent pillars (from center to center) is 7.2 ± 0.2 μm. In Figure c, it is possible to appreciate the increased surface area due to the nanometer-scale roughness of the organic semiconducting pillars, usually achieved in HAR inorganic structures through expensive and time-consuming methods.[55,56] It was demonstrated that the presence of nanogrooves on micropillar side walls is an essential parameter for the formation of 3D neuronal networks in vitro since they enhance the adhesion of neuronal processes to the pillar body.[55] Importantly, the conical shape of the HAR rr-P3HT structures allows combination of the advantages of micro- and nanoscale topographies. In particular, the micrometer-sized base confers to the soft rr-P3HT pillars’ good mechanical stability, while the submicrometer rounded tip is expected to establish a tight interface with the living cell membrane. The distance between HAR pillars is another key parameter. In fact, high-density pillar arrays (adjacent pillar pitch lower than a critical value of about 2 μm) usually lead to limited cellular adhesion and proliferation. This effect has been unanimously attributed to a reduced contact area with the underlying flat substrate.[9,50] Conversely, when the interpillar distance is much higher (pillar density < 30 pillars 100 μm–2), the adhesion and proliferation of cells are enhanced. Interestingly, it was observed that Si and InAs pillar arrays, with a density in the range of the one employed in our case (2 pillars 100 μm–2), promote cells to spread out, with a larger area, without seriously affecting their viability.[57,58]

Aiming at interfacing the realized polymer microstructured substrates with a biological environment, it is important to preliminarily characterize their electrochemical behavior in contact with an aqueous saline medium. To this goal, we carry out electrochemical impedance spectroscopy (EIS) measurements at the electrochemical equilibrium, that is, at potential values corresponding to the device open circuit potential, by employing KRH extracellular solution as the electrolyte. This allows the study of the system in conditions similar to the ones employed in electrophysiological experiments.

By modeling the system with a simple frequency-dependent RC circuit and considering it at frequencies below 10 Hz, it is possible to obtain information about the device/electrolyte interface. In particular, within this low frequency range, the capacitor well approximates the Helmholtz double layer established at the interface.[59−63]

The equivalent capacitance C(ω) and impedance modulus |Z| versus frequency (f) plots in the 0.01 Hz to 100 kHz frequency range, extrapolated from the flat rr-P3HT EIS data, are displayed in Figure S2, and they show the typical trend reported in the literature.[60,61] Here, we focus our attention on the C(ω) and |Z| trends below 10 Hz (Figure ). In the case of the flat rr-P3HT surface, an approximately constant C(ω) value is observed at frequencies <1 Hz, ascribed to the Helmholtz double layer capacitance (Figure a). Conversely, in the case of the rr-P3HT pillar/electrolyte interface, C(ω) is characterized by a distinct behavior, being approximately constant between 1 and 0.2 Hz, steadily increasing at frequency values of <0.2 Hz and finally reaching a value of 53 μF cm–2 at 0.01 Hz. The latter regime can be ascribed to the establishment of a volumetric capacitance due to an enhanced percolation of ionic species through the organic semiconductor that expands the electrochemical active surface area.[61,64] A detailed understanding of this phenomenon, however, requires deeper investigations and falls out of the scope of the present work. We consider only the 0.2–1 Hz frequency range where the C(ω) is constant and the volumetric capacitance contribution is negligible, thus making it possible to evaluate the increase in the surface area exclusively depending on the pillars’ morphology. We observe that C(ω) at 0.2 Hz is equal to 12.2 μF cm–2 (normalized on the geometrical device area), almost 3 times higher than the corresponding value in the flat case (4.4 μF cm–2), in line with the existing literature.[59,60,65] The active surface area increment, due both to ion percolation and surface topography, leads to the noticeable decrease of the device impedance modulus by about 15 times (Figure b), passing from the flat rr-P3HT morphology to the microstructured one. This result is promising in view of the implementation of low impedance polymeric electrodes for electrically assisted cell proliferation and differentiation applications, as well as for electrical stimulation and recording of living cell activity, since it is related to a higher signal-to-noise ratio and higher charge injection limit.[56,66]

Figure 2

Electrochemical impedance spectroscopy. Equivalent capacitance C(ω) (a) and impedance modulus (b), extracted from EIS data, normalized to the device geometrical surface area.

Living Cell Cultures

Although the good viability of several cell models on top of rr-P3HT thin films was extensively verified,[67,68] cell growth and proliferation on top of rr-P3HT microstructures may be strongly affected by the underlying topography, and viability needs to be carefully assessed. Thus, viability of cells cultured on top of HAR rr-P3HT pillars was directly evaluated by employing both a cell line (human embryonic kidney cells, HEK-293) and neurons (primary cortical neurons). HOECHST/NucGreen staining was used to evaluate the relative percentage of healthy cells on the two substrates (Figure ), obtaining >90% viable cells both for HEK-293 and neuron cell cultures after 1 and 14 days in vitro (DIV), respectively. No significant differences were detected between rr-P3HT planar and microstructured regions.

Figure 3

HEK-293 and cortical neurons viability. Bright field and fluorescence images of HEK-293 cells and cortical neurons cultured on flat rr-P3HT (a,b,e,f) and rr-P3HT pillars (c,d,g,h), respectively. All cell nuclei are stained by HOECHST (blue); dead cell nuclei are stained by NucGreen (green). Scale bar, 50 μm. Histograms showing the percentage of viable HEK-293 cells (i) and cortical neurons (j) on the different device morphologies. n = 450 cells for each substrate type. Error bars represent the standard error of the mean (s.e.m.).

The morphologies of HEK-293 and cortical neurons grown on top of polymer flat and microstructured substrates are qualitatively assessed by SEM. Figure clearly shows a significant difference in the morphology of the cells plated on the two different substrates. Both HEK-293 and primary neuronal cells cultured on flat rr-P3HT present a planar, two-dimensional shape. Conversely, when cultured on top of polymer microstructures, HEK-293 cells and neuronal soma remain largely suspended on top of the pillars, rarely reaching the underlying substrate. It can be also appreciated how the selected array geometry leads to a more elongated morphology of the cell body, especially in the case of HEK-293 cells. Interestingly, the cell membrane thinning in the proximity of the pillar tips points to the attainment of a tight cell/material interface.

Figure 4

SEM micrographs of cells cultured on rr-P3HT flat and microstructured samples. Top-view SEM images of HEK-293 cells (a,b) and cortical neurons (c,d). Scale bars, 20 and 10 μm for panels (a,c) and (b,d), respectively.

Morphological Analysis

In order to corroborate and quantify the observed changes in the morphology of HEK-293 cells cultured on top of rr-P3HT microstructured substrates, we carry out immunofluorescence imaging experiments. Figure depicts representative fluorescence images of the nuclei and cytoskeleton, marked with DAPI (blue) and phalloidin-FITC (green), respectively, of the cells grown on top of planar (Figure a) and pillar-modified (Figure b) regions. The cells plated on rr-P3HT pillars show a more elongated shape (Figure c) than the ones grown on the flat controls, in line with the obtained results of HEK-293 plated on similar densities of HAR microstructures[57] and also with other cell types.[69] The pillar microstructures induce a sizable shrinking of the cell membrane in the x–y plane and at the same time a spreading in the z direction, qualitatively observed in the SEM images (Figure ) and quantitatively confirmed also by the measurement of the average top-view surface area of cells (Figure S3). Interestingly, the quantitative evaluation of the average cell perimeter (normalized to the cell area, Figure d) shows that the cells adhered on the rr-P3HT pillars develop more cellular projections, similar to what was already observed in the case of cells cultured on HAR silicon pillars with an interdistance of >2 μm[58] as well as on silicon nanowires.[6,70]

Figure 5

HEK-293 cell morphological analysis. Immunofluorescence images of HEK-293 cells cultured on top of rr-P3HT flat (a) and rr-P3HT pillars (b). Cell nuclei and cytoskeleton are stained with DAPI (blue) and phalloidin-FITC (green), respectively. Scale bar, 50 μm. Quantification and comparison of cell morphological parameters in terms of circularity (c) and perimeter normalized to the cell area (d). Mean values are averaged over a statistical ensemble of n = 100 cells for each substrate type. Error bars represent the s.e.m. ***, p < 0.001 (Student t test).

While it is widely accepted that mechanical cues and subsequent changes on cell morphology strongly influence fundamental cellular processes, including cell–cell communication, metabolism, regulation of cellular proliferation, and genetic reprogramming of the cell fate, a detailed knowledge of the cellular mechanosensory and mechanotransduction networks is still lacking.[71−73] The observed substantial remodeling of the cytoskeleton here observed in HEK-293 model cell lines driven by polymer pillars thus deserves specific attention, and it will be the object of future investigations in different cell models. In this work, we opted for extending our morphological analysis to the case of cortical neurons as one of the most relevant examples of excitable cells (Figure ). We compared the dendritic morphology of cortical neurons grown in the absence of the polymer (Figure a), on top of rr-P3HT pillars (Figure b), and on planar rr-P3HT (Figure c). MAP2 staining showed that the neurons cultured in the presence of the rr-P3HT polymer (flat or pillar) have a slight decrease in the number of primary and secondary dendrites (Figure d,e), as well as in the number of branching points (Figure f), as compared to neurons cultured on glass coverslips. Interestingly, we did not find any significant difference between rr-P3HT pillars and flat conditions. In this regard, several reports in the literature have addressed the effect of micro- and nanostructured substrates, including continuous geometries, like grooved substrates or electrospun fibers,[74−76] and discontinuous geometries like pillar arrays,[75,77] on neuronal morphology. To this purpose, silicon micropillars were mainly employed, while different cell models were used, including dorsal root ganglion neurons,[78] spiral ganglion neuronal cells,[79] and hippocampal neurons.[55,80−83] In general, it has been reported that the neuron outgrowth on micropillars is strongly affected by the pillar density,[77,79−81] and in particular, it was found that pillar spacing lower than 4.5 μm heavily affects the neuron outgrowth. Conversely, at higher interpillar distances, the neuron morphology is comparable to the one of the flat control substrates.[55,81] These findings are in line with our results, even though a direct comparison is not strictly possible due to the differences in substrate materials composition and in cell models and maturation stage.[78]

Figure 6

Cortical neuron morphological analysis. Rat cortical neurons cultured on top of glass (a), rr-P3HT pillars (b), and rr-P3HT flat (c) stained for MAP2. Histograms showing the number of primary dendrites (d), secondary dendrites (e), and branching points (f) of neurons on different devices. n = 30 cells for each condition. Error bars represent the s.e.m. Scale bars, 50 μm. One-way ANOVA followed by Bonferroni correction.

Importantly, the fact that structured substrates do not significantly alter the neuron morphology suggests that their functional properties should be preserved as well, given the intimate relationship between the dendritic expression and the information transduction capability. To this goal, we evaluated the expression of synaptic markers and carried out electrophysiology assays.

Neuronal Functionality

To investigate whether the microstructured polymer structure affects the number and the size of synapses, we stained excitatory synapses with the presynaptic marker vGLUT and the post synaptic marker PSD-95 and inhibitory synapses with the presynaptic marker vGAT (Figure ). We did not find any difference both in the density and size of vGLUT positive puncta in neurons cultured in the presence of the rr-P3HT polymer (flat or pillar) and in neurons cultured on glass coverslips (Figure a–c). The number of PSD-95 positive puncta was unchanged among the three conditions, while the size of PSD-95 clusters was slightly increased in neurons cultured in the rr-P3HT flat condition compared with neurons cultured on glass coverslips or on rr-P3HT pillars (Figure d–f). Moreover, when we labeled inhibitory synapses, we did not detect any difference both in the density and size of vGAT positive puncta in neurons cultured in the presence of the rr-P3HT polymer (flat or pillar) and in neurons cultured on glass coverslips (Figure g–i). Altogether these data demonstrate that synapse formation is not altered on neurons grown on top of rr-P3HT pillars compared with neurons grown on the flat polymer or glass coverslips.

Figure 7

Analysis and quantification of excitatory and inhibitory synapses of cortical neurons. (a) vGLUT staining (green) of rat cortical neurons cultured on top of glass, rr-P3HT pillars, and rr-P3HT flat (from left to right, respectively). Histograms show vGLUT cluster density (b) and area (c). (d) PSD-95 staining (green) on glass, microstructured, and flat polymer samples (from left to right, respectively). Histograms show PSD-95 cluster density (e) and area (f). (g) vGAT staining (green) on glass, microstructured, and flat polymer samples (from left to right, respectively). Histograms show vGAT cluster density (h) and area (i). Cell nuclei are stained with DAPI. n = 8/16 neurons for each substrate. Panels (a,d,g), scale bar 20 μm. Insets of panels (a,d,g) show a magnification of the stained puncta (scale bar, 3 μm). Error bars represent the s.e.m. (one-way ANOVA followed by Bonferroni correction).

Finally, in order to directly verify whether the membrane passive properties and the electrophysiological activity of primary cortical neurons cultured on rr-P3HT HAR pillars are also successfully preserved, we carry out patch clamp recordings in a whole-cell configuration (Figure ). Figure a,b compares the equilibrium parameters of the cells plated both on flat and on microstructured rr-P3HT areas. The average cell membrane resting potentials (Vm = −68 ± 1 mV and Vm = −66 ± 1 mV on pillars and flat regions, respectively, displayed as mean ± s.e.m.) do not show any significant dependence on the substrate type, showing, in both cases, the typical value recorded in vitro (Figure a). Conversely, the average cell membrane capacitance Cm is significantly higher in the case of rr-P3HT pillars, possibly due to the increase in the cell membrane surface area (Figure b). Since the Cm value reported in Figure b is normalized over the cell body top-view surface area, it is possible to relate the surface area increment only to the cell membrane elongation in the z direction exerted by the pillars, observed in the SEM images (Figure ). Figure c,d shows whole-cell recordings in the current clamp configuration carried out on neurons cultured on both the microstructured and the planar polymer substrates, respectively. The intracellular current injection in subsequent steps of 20 pA amplitude and 20 ms time duration and the simultaneous recording of the membrane potential allow evaluation of the threshold current value (Ith), above which action potential firing is observed. Ith is very similar in the two cases, with no statistically significant difference (Figure e). The action potential characteristics are also not affected by the substrate morphology, as displayed by the similar time to peak versus intensity trends of the neurons grown on pillars and flat regions (Figure f). Overall, synaptic expression assays and whole-cell patch clamp experiments demonstrate that the main functional properties of primary cortical neurons cultured on top of polymer pillars are very well preserved.

Figure 8

Cortical neuron electrophysiology. Average cell membrane resting potential Vm (a) and cell membrane capacitance Cm (b) of neurons grown on planar and microstructured substrates. Cm values are normalized to the cell surface area in the x–y plane. *, p < 0.05 (Student t test). Representative action potential traces elicited in neurons plated on rr-P3HT pillars (c) and rr-P3HT flat (d) by 20 ms current steps of 20 pA amplitude. (e) Average threshold current intensity, as normalized to the cell membrane capacitance. (f) Neuron action potential time to peak versus amplitude. Error bars represent s.e.m.

Conclusions

This paper reports the successful fabrication of semiconducting polymer microstructures of conical shape, based on rr-P3HT, by coupling the femtosecond micromachining and the push-coating technique. The fabrication process allows for the realization of biocompatible, three-dimensional devices in a highly versatile, fast, straightforward, and repeatable way. Realized semiconducting polymer pillars are characterized by a HAR and usefully combine the advantages of micro- and nanoscale topographies, without making recourse to complex fabrication techniques, as it is required instead in the case of inorganic, stiff electrodes. In fact, the micrometer-sized base confers the soft rr-P3HT pillars good mechanical stability, while the submicrometer rounded tip establishes a tight interface with the living cell membrane. We have demonstrated that polymer pillars are highly biocompatible substrates suitable for both primary cortical neurons and cell line HEK-293. The versatility of the fabrication technique, however, allows for prompt and straightforward variation of geometrical parameters, adaptable to specific requirements of other cellular models. Interestingly, we observe a sizable change in the morphology of HEK-293 cells cultured on polymer pillars, showing a pronounced tendency to develop in the three-dimensional space. Moreover, the adopted shape and density of polymer pillars lead to establishing a close contact with the neuronal cell membrane, without however negatively affecting cell viability and electrophysiological properties. Importantly, rr-P3HT structured devices are compatible with visible-light modulation of the cell activity. Organic semiconducting thin-film substrates were recently proposed by our group and other groups for localized optical control of the activity of living cells.[59,67,68,84−86] Underlying physical transduction mechanisms include electrical, electrochemical, and thermal mechanisms.[87−90] Based on these reports, we conclude that the use of rr-P3HT polymer pillars represents a useful implementation of the existent devices for living cell optical excitation, thus first opening the way to the combined use of three-dimensional functional polymer structures and optical stimulation.