A multidisciplinary approach to study the functional properties of neuron-like cell models constituting a living bio-hybrid system: SH-SY5Y cells adhering to PANI substrate Open Access

One of the more challenging aspects in cognitive or in rehabilitation neurosciences is the design of functional hybrid systems able to mimic the brain functionality, to connect and to exchange information between biological materials, like brain or neurons, and man-made electronic devices.^1–4 A demanding task in designing neuromorphic circuits is to reproduce the adaptive/learning mechanism: the ability to learn from the external stimuli and, accordingly, to adapt themselves is a typical behaviour always present even in the simplest living organisms but scarcely reproduced in the artificial components. In future prosthetic devices, attractive candidates are the new class of electronic elements called memristors.^5–7 The properties of these devices mimic the synapse activity of a nervous system by modifying their conductivity depending on the previous electric history. This peculiar property represents a sort of memory for the device.

Strategic importance to obtain neuromorphic systems, where the memristors play a role of synapses, is covered by organic polymers such as PANI^8,9 or PEDOT.⁵ It is well documented that these polymers, if assembled with polyelectrolytes, show memristive properties.¹⁰ 

Thanks to their easy handling, also using their self-assembling properties, the memristive polymers are the best candidates to obtain memory switching devices able to interface with biological materials.¹¹ Implementation of the functional properties of living systems into electronic memristive devices has recently obtained with the use of the slime mold PhysarumPolycephalum, a multinuclear unicellular mass of protoplasm.^12,13

To optimize the use of memristive devices as bio-electronic interfaces, it is necessary to asses their biocompatibility measuring the functional viability of the cells adhering to them and moreover assuring their ability to collect the signal directly from the cells. In this study we use SH-SY5Y cells as suitable cell model to test the PANI bio-compatibility and to analyse the substrate effects on the cell functionality using multidisciplinary approaches that include neurobiological, electrophysiological and vibrational spectroscopy methodologies.

SH-SY5Y human neuroblastoma cell line was grown in Modified Eagle’s medium (DMEM) containing 10% (v/v) heat-inactivated fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin (SIGMA Aldrich, St. Louis) and maintained at 37°C in a 5 % CO₂ humidified atmosphere. Cells were grown in sterile conditions on different substrates: on silicon, on glassy cover slips and finally on a coverslips covered with multilayers of synthetic PANI. The PANI film was obtained by the deposition of Polyaniline (PANI) multilayers on glass substrates with the well known Langmuir–Schaefer (LS) technique (for details see Ref. 14). To induce a neuron-like phenotype, the cells were plated at a density of 10⁴ cells/cm² and grown for 24 hours in complete medium. 24 hours after the seeding, the culture medium was changed and replaced with DMEM + 15% FBS + 10nM all-trans Retinoic Acid (Sigma Aldrich). After 5 days, the medium was then substituted with DMEM + 50ng/ml BDNF. These conditions were maintained for several days.

Cells were fixed and immunostained as previously described.¹⁵ The anti-b Tubulin primary antibody (Santa Cruz) was used at 1:500. The Alexa Fluor 488-conjugated goat anti mouse IgGs secondary antibody (1:500) was purchased from Life Technologies. Microscopy analysis was performed using the Zeiss Observer Z.1 Microscope with a PlanApo oil immersion lens (x20, NA 1.4). Pictures were acquired using the AxioVision imaging software package (Zeiss) and assembled with Adobe Photoshop CS3. Images were not modified other than adjustments of levels, brightness and magnification.

Advanced electrophysiological technique was applied to human neuroblastoma SH-SY5Y cells according to material and methods published in Ref. 16 In brief, standard methods of splitting and manipulation were performed under isolated sterile conditions and proliferation at 37°C (60% humidity, 1% CO₂), using completed DMEM as culture medium (Glutamax, 10% FBS and 1% of penicillin/streptomycin). During splitting, a portion of cells was drop over 3 ml petri dishes (NUNC. Thermo Scientific, USA) with/without Ø15 mm coverslips covered with PANI (sterilized previously with ethanol). Cell were used for experiments 2-6 days after split.

Patch-clamp recordings were done in whole-cell configuration¹⁷ at non-sterile conditions and at room temperature (∼21°C). Patch-clamp recording micropipette were fabricated from GB150-8p (OD 1.5 mm, ID 0.86 mm) borosilicate glass capillaries (Science Product, Hofheim, Germany) using a PIP6 temperature controlled pipette puller (HEKA, Lambrecht/Pfalz, Germany). Once attached to membrane, whole-cell configuration was obtained accessing to cell’s interior after the rupture of the patch seal (electric resistance >1 GΩ). To test the voltage activation, the membrane potential was set to a holding level of -40mV, then starting from -100mV a stimulation protocol (1Hz frequency) were used consisting of 15 pulses of 10 mV steps with a long temporal durations of 100 ms. Bioelectric signals were picked up using a universal voltage/current clamp amplifier (ELC-03XS, npi, Tamm, Germany) connected to a PC computer via a breakout-box interface (INT-20X, npi, Tamm, Germany). High-precision positioning was provided by a piezo-drive micromanipulator (Sensapex, Oulu, Finland). Instant patch clamp parameters were monitored with the signal acquisition software (WinWCP Electrophysiology Software, ©John Dempster, University of Stratchlyde, Scotland). Current-Voltage (IV) where used to compare the ionic current behavior between cell types adhering to PANI and controls (no-PANI adhering). The mean and steady state values of the current component (considered as outward-rectifying K⁺ currents) were calculated at a given command potentials and used to compare between cell types and surface cases (PANI/no-PANI). Further details about stimulation and recording protocols, and solutions are available in Ref. 16.

The cells directly adhering on silicon and PANI substrates were prepared for the Raman measurements.

The chosen silicon substrate is covered with TEOS layer, which spontaneously converts silicon to vitreous silica upon the addition of water. This substrate was chosen in order to combine a low Raman signal substrate with the good bio-compatibility of the silica glass.

The cells are immersed in phosphate-buffer saline (PBS) solution during the Raman measurements after the culture medium washing.

Raman spectra are collected using a micro-Raman setup equipped with a solid state laser with emission wavelength at 532 nm, and power reduced to less than 10 mW to avoid the cells photo-damage. An edge filter with optical density higher than 8 prevents the acquisition of spectra below 100 cm⁻¹ from the laser line. The scattered light is dispersed by a 600 grooves/mm grating and the CCD detector (1024 × 256 Pixels) allows the simultaneous acquisition of 3000 cm^-1 Raman shift in a single spectrum at a resolution of about 10 cm⁻¹. The acquisition time for the whole spectrum is 100 s. A water immersion objective UPLSAPO 60XW from Olympus is chosen in order to reach high quality images and spectra thanks to its high numerical aperture (N.A. = 1.2) and its high transmission and chromatic aberration correction. In the chosen experimental configuration, the exciting source is focused onto a single SH-SY5Y cell through the objective directly immersed into the PBS solution.

In Figure 1 we report representative images of differentiated SH-SY5Y cells adhering on the PANI substrate. The immunofluorescence measurements has been done using an anti beta tubulin antibody to easily visualize the structure of the neurites (Figure 1 B). 3 days of incubation with BDNF was enough to differentiate cells on the substrates and to avoid further stress given by serum starvation. As shown, the cells developed neurites attached to the PANI surfaces. The high growth rate as well as the neurite formation confirms the good bio-compatibility of the substrate.

We tested the effect of PANI adhesion measuring the functional cell properties in terms of membrane ionic currents using whole-cell patch clamp¹⁷ in voltage- and current-clamp mode. Results here concentrate on neuroblastic and differentiated cells growing up over surfaces with PANI presence or absence (Nunclon^®  as deep dish). Information from experiments was compared computing mainly the steady state current value.

SH-SY5Y exhibited a different evolution of the ionic interchange depending on the temporal variations of the applied stimulus protocols. Outward rectifying currents were elicited when the voltage application past from a controlled condition value (holding -40 mV) to more than 0 mV of the voltage pulse. This rectifying current was observed using different time stimulus protocols showing a more marked variation with long stimulation (Figure 2A). The Current-Voltage (I-V) curve on cells in different preparations support also in this case a reduction (about 30%) of the whole-currents in PANI-adhering cells for long stimulation protocols (Figure 2B). This effect can be attributed to a reduction of flowing outward currents, presumably attributed to K+ ions (Figure 2B). Inward currents result was not seriously affected by the adhesion to PANI. Nevertheless the shape of the I/V curve slope is maintained in both PANI-adhering and no-PANI adhering conditions: these data indicate that the kinetics are almost preserved.

The viable conditions of the cells adhering to PANI have been demonstrated also with recordings in current-clamp mode of the spontaneous activity of the cells. Cells showed a robust activity of depolarizing potentials and action potentials, the latter up to 20 mV of amplitude (Figure 3A). In stimulated conditions, the cell excitability is sensitive to different temporal pattern of stimulations as shown also in voltage-clamp experiments: in this case, long stimulations seem to be more effective in the eliciting of largest responses while short stimulation elicits smallest potentials (Fig 3B). However in both cases the correspondence between current-stimulation and membrane potentials is well retained. Thus, also in these conditions the adhesion to PANI does not interfere with the overall excitability of the cells.

As a whole, our data demonstrate that a better bio-functionalization could restore the observed loss of ionic currents, improving the adhesion process, which may cause a relative deficit of ion channel availability. However, PANI shows no toxic effect on the cell physiology ensuring a full viability and a good excitability of the cells adhering to it.

The results obtained with the electrophysiology investigation can be compared with the spectroscopic ones in order to have a more complete characterization of the cells status. The ability of Raman spectroscopy to investigate the molecular vibrational modes is well known, and it was applied in different fields in condensed matter physics.^18–21 Recently reaching the micrometric spatial resolution, thanks to the coupling with microscopes, and due to its non-contact and label free nature, Raman micro-spectroscopy has been applied to analyse the status of biological cells^22–26 and tissues.^27,28 In order to investigate the status of SH-SY5Y cells by Raman spectroscopy, we compare the spectra of cells grown on the PANI polymeric films with those of cells grown on silicon covered with silica layers, a control substrate which assures the good biocompatibility.

In Figure 4, as an example, we report representative spectra a single SH-SY5Y cell adhering on the PANI substrate. The spectrum of the single cell can be analysed removing the buffer and the substrate contributions, once normalized to the intensity of their characteristic modes. In particular, the intense and wide band centred at about 3200 cm^-1, assigned to the OH- stretching vibrational modes of water has been chosen for the normalization. The so obtained spectra are reported in Figure 5, where the data acquired on cell adhering on the PANI and on the silicon substrate are compared (upper panel).

Both spectra present all the characteristic features of the supramolecular components of living cells.²⁹ The substrate on which the cells are deposited, however, affects the intensity of some of these bands as evidenced in the lower panel of Figure 5, where the difference between the two spectra is reported. Out of the experimental error, differences appear both in the low and in high frequency region. In the so called “fingerprint region”, at frequency lower than 1800 cm⁻¹, the main peaks of the nucleic acids, lipids and proteins are present. Many scientific works indicate in this region also the presence of spectroscopic markers of the cellular stress.^30–35 The initial degradation of DNA and/or RNA, and the alteration of lipid bodies which starts to occur in stressed cells, has been associated to the spectral changes in the 1000-1600 cm^-1 range. The intensity reduction of the peak centred at around 1600 cm^-1 has been recently found in living S. pombe cells investigated at different cell cycle stages,³⁰ in the yeast cells 26, in the HeLa cells during the induction of apoptosis³¹ and also in time lapse Raman spectra of SH-SY5Y cells stored under unfavourable conditions 24. In these studies, a clear correlation between the peaks intensity and the cell viability has been found.

Our data show that, the intensity of peaks around 1125 cm^-1, 1310 cm^-1, and 1580 cm^-1 is smaller in the sample grown on the PANI substrate than on the control as evidenced in the difference spectrum. These peaks can be assigned to proteins, lipids and the breathing of the Guanine and Adenine rings respectively²⁹ but at about the same frequencies, characteristic Raman bands of Cytochrome C can be also recognized.³¹ Due to the resonant scattering of Cytochrome C with the incident 532 nm excitation, here used, the intensity of its Raman bands probably prevails on the others. The intensity decreasing of the characteristic peaks of the Cytochrome C has be found during the cell suffering.³¹ Our data seems to indicate that the cells grown on the PANI substrate show an increase in the spectroscopic markers related to the cellular suffering probably induced by the substrate interaction.

Also the high frequency signals, reported in Figure 5, support this hypothesis. In this spectral region the contributions from C-H₂ and C-H₃ stretching modes of lipids and proteins are present. The spectral components of this band are respectively assigned to CH₂ stretching of lipids, centred at 2850 cm^-1 and 2880 cm^-1,^25,36 and the CH₃ stretching of proteins located at about 2930 cm^-1.³⁷ The comparison reported in the lower panel of Figure 5. shows a clear reduction of the low frequency side of the CH stretching profile.

The decrease of the low frequency components was indicated as a marker of cell viability²³ in the time-resolved experiment. In fact it can be associated to the degradation of bio-membranes and/or to the relative decrease of lipids with respect to proteins.^21,36 Therefore, the Raman data confirms that cells grown on PANI substrates present the spectroscopic signatures of living cells, though all the spectroscopic markers consistently indicate the existence of a state of cellular stress.

This paper reports the first joint study performed using immunocytochemistry, electrophysiology and Raman spectroscopy on living cells adhering on PANI substrates in bio-hybrid systems. By the immunocytochemistry study we found that the PANI films can support the cell adhesion and can also promote cell differentiation. Raman spectroscopy and electrophysiology are able to characterize the status of cells by monitoring the dynamic supramolecular changes occurring in living cells and by measuring their bioelectrical activity respectively. The reported data indicate that the bioelectric level is retained, with some modifications of the quantity of the whole cell membrane outward currents. In particular we found a reduction in the cells excitability upon positive stimulations that can be ascribed to membrane geometric factors related to the adhesion process as well as to a relative failure of ion channel activity. Both these hypotheses can be explain by a substrate effect on the cell properties. The effect of the substrate is also revealed by the spectroscopic investigation. In fact, if in the hybrid system the general biochemical properties of cells are not strongly modified, intensity modifications on the spectroscopic markers of cells stress have been found. The joint analysis consistently suggests the need of increasing the bio-compatibility of the PANI substrates.

In our opinion, the achievement of this study is the development of an experimental multidisciplinary approach able to characterize in real time the status of the living cells. We applied the analysis on a fully working hybrid bio-inspired device but in a prospective view, this work provide the proof of concept of an integrated study able to analyse the status of living cells in a large variety of applications merging nanosciences, neurosciences and bioelectronics.

This work is supported by PAT (Autonomous Province of Trento) “Grandi Progetti 2012” Project “MaDEleNA”; in this context, the author thanks Dr. Silvia Battistoni (CNR-IMEM, Parma, Italy) for the preparation and the supply of the PANI substrates.