Wild type mESCs (E14) (Hooper et al., 1987), a kind gift of Marta Roccio (Inner ear research lab, Department of Clinical Research, University of Bern, Switzerland) were maintained in a 37°C tissue culture incubator at 5% CO₂ and differentiated into mESN using previously described protocols with several adaptations (Wichterle et al., 2002; Wichterle and Peljto, 2008; Wu et al., 2011). Briefly, cells were maintained in tissue culture flasks (TPP, Trasadingen, Switzerland) that had been coated for 30 min at 37°C with 0.1% gelatine in ES medium consisting of DMEM with Glutamax, 15% fetal bovine serum, embryonic stem cell-qualified (GIBCO^®, Thermo Fisher Scientific, Waltham, MA, USA) supplemented with MEM non-essential amino acids to a final concentration of 1 mM for all amino acids (glycine, L-alanine, L-asparagine, L-aspartic acid, L-glutamic acid, L-proline, L-serine), penicillin-streptomycin-glutamine to a final concentration of 1000 Units/ml penicillin, 1000 μg/ml streptomycin, and 0.292 mg/ml L-glutamine, 2-mercaptoethanol (0.1 mM) as well as leukemia inhibitory factor to a final concentration of 1000 Units/ml (Merck Millipore, Darmstadt, Germany). Unless otherwise specified, all cell culture supplies for embryonic stem cell maintenance, mESN differentiation and mESN culture were obtained from Invitrogen (Paisley, Scotland, UK). To elicit embryoid body (EB) formation and induce differentiation of the mESCs, 1–2 × 10⁶ cells were plated on day 0 in a 10 cm tissue culture dish (TPP, Trasadingen, Switzerland) and the medium was changed to a differentiation medium termed DMNK⁺ consisting of a 1:1 mixture of DMEM-F12/Glutamax and Neurobasal medium, supplemented with penicillin-streptomycin-glutamine to a final concentration of 1000 Units/ml penicillin, 1000 μg/ml streptomycin, and 0.292 mg/ml L-glutamine, 2-mercaptoethanol (0.1 mM), and 15% KnockOut^TM Serum Replacement. After 1 day in culture the floating EBs were transferred into a 15 ml tube, centrifuged at low speed (3 min; 35 × g), resuspended in 10 ml of DMNK⁺ medium and replated in a new 10 cm dish. On day 2 of differentiation, EBs were supplemented with 1 μM retinoic acid to induce neuralization. The next day, the medium was supplemented with 1 μM of smoothened agonist (Merck Millipore, Darmstadt, Germany). On day 7, EBs were dissociated using 2 ml of Accumax (Merck Millipore, Darmstadt, Germany) for 10 min at room temperature, diluted with 2 ml of DMNK⁺ medium and carefully pipetted up and down 30 times using a 1000 μl pipette. Cells were filtered through a 100 μm cell strainer (Becton Dickinson, Bedford, MA, USA) to obtain a single cell suspension. This step was performed three times to enrich the yield of single cells. The suspension was then centrifuged (3 min; 35 × g) and resuspended in 2 ml of DMNK⁺ medium supplemented with recombinant rat glial cell-derived neurotrophic factor (10 ng/ml), recombinant human brain-derived neurotrophic factor (10 ng/ml), and recombinant rat ciliary neurotrophic factor (20 ng/ml) (all R&D Systems, Minneapolis, MN, USA). 2 × 10⁵ cells/ml in 100 μl of DMNK⁺ medium supplemented with factors were then plated on MEAs which were previously coated for 30 min at 37°C with Matrigel^TM (Corning, Corning, NY, USA) diluted in 1:10 DMNK⁺ medium. The cells were then cultured at 37°C, 5% CO₂ changing medium every 3 days. Exposure to toxins and electrophysiological recordings were conducted 21 days (±2 days) after plating.

Multi-electrode array slides (Qwane Biosciences, Lausanne, Switzerland) composed of a glass substrate (700 μm thick, 21 mm × 21 mm) holding 68 black platinum electrodes with a dimension of 40 μm × 40 μm, an inter-electrode distance of 200 μm holding an impedance of 10 kΩ at 1 kHz and a 5 μm thick SU-8 polymer layer for insulation were used. For culturing of the cells, each MEA was placed into a 35 mm culture dish (TPP, Trasadingen, Switzerland) holding 2 ml of DMNK⁺ medium.

Further, MEAs from MultiChannel Systems (MultiChannel Systems GmbH, Reutlingen, Germany) composed of 60 Titanium nitride electrodes with a diameter of 30 μm (59 recording electrodes, 1 internal reference electrode), an inter-electrode distance of 200 μm holding an impedance of <100 kΩ and a silicon nitride isolation were used. Further they hold a 6 mm high glass ring around the electrodes allowing the addition of 1 ml of DMNK⁺ medium.

Spontaneous neuronal activity was measured by detecting extracellular voltage transients, induced by current flow through membranes of neurons. MEAs were incorporated into a Plexiglas chamber and mounted on an inverted microscope. To maintain the pH at 7.4 the cultures were superfused with an extracellular solution consisting of 145 mM NaCl, 4 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 5 mM Hepes, 2 mM Na-pyruvate and 5 mM glucose, (all Sigma-Aldrich, St. Louis, MO, USA). The 68 electrodes were AC-coupled to an individual custom-made amplifier and data were digitized at a rate of 6 kHz with 12 bit resolution and stored on a hard disk for offline analysis as described previously (Tscherter et al., 2001). To control the A/D card (NI-DAQ-card, AT-MIO-64E-3, National Instruments, Ennetbaden, Switzerland) a custom-made LabVIEW software (National Instruments, Ennetbaden, Switzerland) was used. The contribution of inhibitory neurons in our cultures was analyzed under the treatment of a GABA A (10 μM gabazine) and glycine (1 μM strychnine, both Sigma-Aldrich, St. Louis, MO, USA) receptor antagonist and signals were recorded for 10 min (n = 7, from three independent experiments) (Figure 3C). To discriminate intrinsic neuronal activity and synaptic activity, pharmacological identification of postsynaptic responses was performed by application of a mixture of GABA A (10 μM gabazine), glycine (1 μM strychnine), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid/kainate (10 μM, CNQX) and N-methyl-D-aspartate (50 μM APV) receptor antagonists (all from Sigma–Aldrich, St. Louis, MO, USA) thus inhibiting all synaptic transmission (Figures 3A,B). This step was conducted at the end of every recording with each culture that showed a remaining burst activity. Further, to distinguish background noise from neuronal activity, every culture was treated as a final step with 1 μM TTX (Alomone Labs, Jerusalem, Israel), a potent voltage-gated sodium channel blocker, and recorded for 10 min. The resulting activity was then set as a zero reference.

For long-term time-dependent analysis measurements a commercially available MEA system (MultiChannel Systems GmbH, Reutlingen, Germany) was used to measure the activity 21 days (±2 days) after plating the cultures. This setup enabled to conduct measurements at different time points without the risk of contamination. The signals from the MEAs were amplified with a MEA2100 headstage and data were digitized with a 60-channel A/D converter at a rate of 25 kHz with 16 bit resolution. The measurement was controlled via MC_Rack software (MultiChannel Systems GmbH, Reutlingen, Germany) and the data were stored on a hard disk for further analysis. The temperature was controlled with an external heater unit (TC02, MultiChannel Systems GmbH, Reutlingen, Germany) set to 37.6°C. To eliminate evaporation and contamination during the experiment the MEAs were sealed with a semi-permeable membrane (ALA MEA-MEM, ALA Scientific Instruments, Westbury, NY, USA), (Potter and DeMarse, 2001). All recordings were conducted in DMNK⁺ medium. To discriminate intrinsic neuronal activity and synaptic activity, pharmacological identification of postsynaptic responses was performed as described above.

Event detection and further analysis was performed offline using IGOR (WaveMetrics, Inc., Lake Oswego, OR, USA) as described previously (Tscherter et al., 2001). The detected signals are fast voltage transients (<4 ms) corresponding to single action potentials or spikes in neuronal somata or axons (single unit activity) and are represented by single time markers called events. These are shown in event raster plots for each electrode (Figure 2A). Usually, they appeared in clusters (multi-unit activity) originating from closely timed action potentials of several neurons seen by one or multiple electrodes. These multi-unit activities were defined as burst (Figure 2C). Detected bursts were subsequently defined offline in IGOR by reaching a certain threshold set individually for each culture with the beginning being defined as the first event in a time window of at least 5 ms and the burst end being the last event in a time window of at least 10 ms. Network activity plots (Figure 2B) show the total activity of active electrodes within a sliding window of 10 ms shifted by a 1 ms step. For calculation of the spike rate (spikes/s per active electrode), spontaneously active electrodes were defined as electrodes showing a minimum of 0.1 detected events per second. Electrodes which detected fewer or 0 signals were not taken into account.

Due to the high toxicity of BoNTs, the handling requires appropriate safety measures. Solubilized neurotoxins were handled in a level 2 biosafety cabined equipped with high-efficiency particulate air filters. All consumables (e.g., pipette tips, falcon tubes) and medium that were exposed to the toxin, as well as the neurotoxin itself, were treated with 2 M NaOH (Sigma-Aldrich, St. Louis, MO, USA) and incubated for at least 1 h prior to discarding them in a separate container. These were autoclaved and disposed in the biological waste containers within the laboratory.

To assess the biologic activity of BoNT, 21 day (±2 days) old cultures were exposed to either 1.66, 16.6, or 166 pM of purified BoNT/A (A1 Hall Strain, Metabiologics, Madison, WI, USA) diluted in PBS and incubated at 37°C, 5% CO₂ in 2 ml of DMNK⁺ medium. Cultures serving as control received the equivalent volume of PBS containing no toxin. After 6 h, cultures (treated and untreated) were washed twice with DMNK⁺ medium to remove all unbound toxin and neuronal activity was measured during 10 min as described beforehand in the electrophysiological recordings section. Serving as a negative control, cultures were treated with 166 pM heat inactivated BoNT/A (95°C, 5 min) in PBS and incubated for 6 h at 37°C, 5% CO₂.

For long-term measurements 21 day (±2 days) old cultures were exposed to either 1.66 pM of purified BoNT/A (A1 Hall Strain, Metabiologics, Madison, WI, USA) or 33 Units/ml of onabotulinumtoxin A (Botox^®, Allergan Inc., Dublin, Ireland) diluted in PBS and incubated at 37°C, 5% CO₂ in 1 ml of DMNK⁺ medium. Neuronal activity measured in all cultures at the beginning of the experiment (0 h) served as baseline activity. Upon treatment with the toxin, the cultures were transferred after 6, 12, and 24 h from the incubator to the MultiChannel Systems electrical recording system and neuronal activity was measured. Before each recording the cultures were allowed to settle for 5 min before spontaneous neuronal activity was measured for 2 min. After completion of the recordings the cultures were transferred back to the incubator.

On day 7 of differentiation, EBs were dissociated as described in Section “Mouse Embryonic Stem Cell Culture” and 2 × 10⁵ cells/ml in 500 μl DMNK⁺ medium supplemented with growth factors were plated on round cover glasses (Ø 12 mm, HUBERLAB, Aesch, Switzerland) which were previously coated for 30 min at 37°C with Matrigel^TM (Corning, Corning, NY, USA) diluted 1:10 in DMNK⁺ medium. The cells were then cultured at 37°C, 5% CO₂ for up to 21 days, changing medium every 3 days. Following, cells were fixed in 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO, USA) in PBS for 10 min, washed three times with PBS and permeabilized with 0.1% Triton X-100 for 5 min (Sigma–Aldrich, St. Louis, MO, USA). Prior to staining, cells were blocked for 2 h with 2% bovine serum albumin and 0.01% Triton X-100 (both Sigma–Aldrich, St. Louis, MO, USA). Primary antibodies against β-III Tubulin (monoclonal mouse anti β-III Tubulin, 1:500, R&D Systems, Minneapolis, MN, USA; polyclonal rabbit anti β-III Tubulin, 1:500, Abcam, Cambridge, UK), synaptic vesicle glycoprotein isoform A/B/C (monoclonal mouse anti SV2, 1:100), GD1a ganglioside (monoclonal mouse anti GD1a, 1:10), GT1b/2b ganglioside (monoclonal mouse anti GT1b/2b, 1:10) (SV2 antibody developed by Buckley, K.M., GD1a, and GT1b/2b antibodies developed by Schnaar, R.L. obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology, Iowa City, IA, USA), synaptic vesicle glycoprotein isoform C (polyclonal rabbit anti SV2C, 1:500), SNAP-25 (monoclonal mouse anti SNAP-25, 1:500), synapsin 1a/b (monoclonal mouse anti Syn1, 1:500), VGLUT 2 (monoclonal mouse anti VGLUT2, 1:500), glutamate decarboxylase 1 (polyclonal rabbit anti GAD1, 1:500) (all Synaptic Systems, Goettingen, Germany), and GFAP (polyclonal rabbit anti GFAP, 1:500) (Chemicon, Merck Millipore, Darmstadt, Germany) were incubated over night at 4°C in PBS with 2% bovine serum albumin and 0.01% Triton X-100. The samples were washed three times with PBS and secondary antibodies conjugated with Alexa 488, Alexa 555, and Alexa 647 (all Invitrogen, Paisley, Scotland, UK) were incubated the following day for 2 h at room temperature diluted 1:500 in 2% bovine serum albumin and 0.01% Triton X-100. Cells were washed three times in PBS and mounted in Fluoroshield^TM with DAPI (Sigma–Aldrich, St. Louis, MO, USA) for nuclear staining. Images were acquired with a Zeiss laser scanning confocal microscope 710 using a Zeiss Plan-Apochromat 60x 1.4 NA objective and a Nikon Eclipse Ti microscope using Nikon-Plan Fluor 40x 1.0 NA and 63x 1.4 NA objectives.

Graph Pad Prism software (version 6.0) was used for statistical analyses. Differences between unpaired groups were evaluated either using the two-tailed unpaired Mann–Whitney test or the unpaired t-test. For comparison of paired groups a paired t-test was conducted. A p-value < 0.05 (two-tailed) was considered as statistically significant.

Three days after plating (data not shown) and throughout the next 21 days, mESN expressed the post-mitotic neuronal marker β-III Tubulin (Figures 1A–F, 2D). Immunocytochemistry performed 21 days after plating the cells revealed, that a large proportion of post-mitotic neurons expressed the VGLUT 2, a known marker for glutamatergic neurons (Figure 1D). However, only a minority of these cells were immunoreactive for GAD1, a marker for GABAergic neurons (Figure 1E). Co-staining with the universal pre-synaptic neuronal marker Syn1 proved the presence of synaptic vesicles in axons of β-III Tubulin positive cells (Figures 1D,E). In addition, neurons expressed the polysialo gangliosides GD1a and GT1b/2b, both known to mediate the initial anchorage of BoNT/A toward the presynaptic membrane (Figure 1C). Furthermore, the expression of all three pre-synaptic membrane receptor isoforms SV2 C/A/B (order expresses decreasing neurotoxin affinity) (Mahrhold et al., 2006; Rummel, 2013) (Figures 1A,B), necessary for docking and cytosolic uptake of BoNT/A as well as its target protein SNAP-25 (Figure 1A) further confirmed that these cells expressed the necessary components used by BoNT to inhibit synaptic activity. After 21 days in culture, a high amount of GFAP were observed, a proteins expressed by numerous cell types of the central nervous system including astrocytes (Figures 1F, 2D).

First single spikes, recorded from individual electrodes were measured beginning 3 days after plating of the cells. At 6 days after plating the cultures showed first synchronous spontaneous activity over several electrodes simultaneously and reaching a plateau at 21 days after plating, represented in the formation of oscillating bursts (Figure 2A). The size of the bursts was random and did not have any obvious periodicity and their occurrence could not be predicted.

Next, we analyzed the contribution of the different neurotransmitters to spontaneous postsynaptic events by pharmacological interference. Inhibition of GABA A, glycine, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid/kainate and N-methyl-D-aspartate receptors completely abolished the remaining occurrence of bursts in every analyzed culture (n = 60) (Figure 3A). However, asynchronous intrinsic activity was still detectable, indicating that the neurons were still able to fire single action potentials (Figure 3B). Inhibition of the GABA A and glycine receptor by gabazine and strychnine, thus blocking the effects of inhibitory neurons, led to a slight increase of the burst rate (22.8 ± 5.11 vs. 23.4 ± 6.1 bursts/min; n = 7; from three independent experiments), as well as mean spike rate (9.4 ± 3.9 vs. 11.5 ± 4.3 spikes/s per active electrode; p < 0.01; n = 7; from three independent experiments) (Figures 3C,D). This demonstrates the presence of inhibitory synaptic connections involving GABA A and glycine receptors, in accordance with our immunocytochemistry findings on the presence of a small amount of GABAergic neurons (Figure 1D). Finally, all spontaneous activity was eliminated by applying 1 μM TTX, indicating that the detected voltage transients originated from neuronal activity (Figure 3B, bottom raw data trace).

Collectively, these data demonstrate the functional development of glutamatergic and GABAergic synapses within our mESC-derived neuronal network.

Next, we evaluated the effect of BoNT/A exposure on the average network activity, defined by changes in burst and spike rates, in a dose- and time-dependent fashion. Exposure to BoNT/A for 6 h resulted in a significant decrease of the burst rate for toxin concentrations of 166 pM (43.9 ± 24.9%; p < 0.0005; n = 6) and 16.6 pM (66.7 ± 42.5%; p < 0.005; n = 16) compared to untreated cultures (100 ± 19.7%; n = 27) (Figure 4A). A significant decrease in the mean spike rate was observed for toxin concentrations of 166 pM (28.7 ± 20.0%; p < 0.005; n = 6) and 16.6 pM (78.0 ± 70.9%; p < 0.05; n = 16) compared to untreated cultures (100 ± 31.9%; n = 27) (Figure 4B). However, no significant decrease in burst rate (108.7 ± 25.2%; p = 0.25; n = 9) (Figure 4A) and mean spike rate (91.5 ± 67.3%; p = 0.38; n = 9) (Figure 4B) was observed upon treatment with 1.66 pM BoNT/A for 6 h.

To exclude that the observed effect of the toxin on neurons may be due to protein properties unrelated to its biological activity, cultures (n = 5) were treated with 166 pM of heat inactivated (5 min, 95°C) BoNT/A for 6 h. It has previously been shown, that heating BoNT/A for 15 s at 72°C was sufficient to reduce its toxic activity by more than 99.99% (Weingart et al., 2010). Compared to a control (untreated group, n = 3), we could not detect any significant difference in the burst or spike rate after 6 h (Figure 4C) indicating that the decrease of activity was indeed caused by the specific biological activity of BoNT/A.

It has been shown, that BoNT/A leads to a time-dependent increase of SNAP-25 cleavage as well as decrease in the network activity (Pellett et al., 2011; Beske et al., 2015). We therefore examined whether longer exposure periods with 1.66 pM BoNT/A would lead to a decrease of the network activity. In control cultures exposed to PBS only, there was no significant changes over time in either the burst or mean spike rate (Figures 5A,C). After 12 h of incubation with 1.66 pM BoNT/A the overall burst rate of the neuronal cultures was reduced by 62.9% (37.1 ± 48.0%; n = 4) and the mean spike rate was reduced by 36.6% (63.4 ± 36.9%; n = 4) comparing to the baseline activity (highlighted by the dotted line) determined prior to treatment (Figure 5C). At 24 h post treatment with 1.66 pM BoNT/A a complete loss of synaptical driven activity was observed resulting in the complete absence of burst formation when compared to the baseline activity (Figure 5B, right panel and Figure 5C, left panel). Further, the average spike rate was reduced by 80.5% (19.5 ± 17.4%; n = 4) when compared to the baseline activity. Solely the asynchronous intrinsic activity was detectable (Figure 5B, right panel and Figure 5C, right panel). By comparing the untreated control group receiving only PBS toward BoNT/A treated cultures 24 h post treatment, significant decreases of both the burst rate (105.0 ± 10.1%; n = 3) vs. (0.0 ± 0.0%; p < 0.0001; n = 4) (Figure 5C, left panel) and the mean spike rate (103.6 ± 9.5%; n = 3) vs. (19.5 ± 17.4%; p < 0.001; n = 4) (Figure 5C, right panel) were observed.

Pharmaceutical preparations with onabotulinumtoxin A (Botox^®, Botox Cosmetics^®, Vistabel^®, Vistabex^®) contain the 150 kDa holotoxin BoNT/A and several non-toxic clostridial accessory proteins which are covalently bonded to the 150 kDa holotoxin forming a BoNT/A complex. We therefore assessed, if BoNT/A complex interferes with the sensitivity or kinetics of the described assay. Upon treatment with 33 Units/ml of Botox^®, corresponding to approximately 1.66 pM of BoNT/A based on its reported specific activity, and incubation for 24 h a complete loss of synaptical driven activity was observed resulting in the complete absence of burst formation when compared to the baseline activity (Figure 5C, left panel). Comparing the average spike rate 24 h post treatment with an untreated control group, a significant decrease in activity was observed (103.6 ± 9.5%; n = 3) vs. (19.2 ± 3.2%; p < 0.001; n = 3) (Figure 5C, right panel).

Overall these data indicate, that mESN grown on MEAs can detect the biological activity of BoNT/A holotoxin in a time- and dose-dependent manner resulting in a decrease of synaptically mediated neuronal activity. Further, no impairment regarding kinetics or sensitivity was observed using BoNT/A complex that is contained in the pharmaceutical preparation Botox^®.