Analgesic effect of Botulinum toxin in neuropathic pain is sodium channel independent

Botulinum neurotoxin type A BoNT/A is used off-label as a third line therapy for neuropathic pain. However, the mechanism of action remains unclear. In recent years, the role of voltage-gated sodium channels (Nav) in neuropathic pain became evident and it was suggested that block of sodium channels by BoNT/A would contribute to its analgesic effect. We assessed sodium channel function in the presence of BoNT/A in heterologously expressed Nav1.7, Nav1.3, and the neuronal cell line ND7/23 by high throughput automated and manual patch-clamp. We used both the full protein and the isolated catalytic light chain LC/A for acute or long-term extracellular or intracellular exposure. To assess the toxin ’ s effect in a human cellular system, we differentiated induced pluripotent stem cells (iPSC) into sensory neurons from a healthy control and a patient suffering from a hereditary neuropathic pain syndrome (inherited erythromelalgia) carrying the Nav1


Introduction
Chronic pain plays a major role in the social and economic burden of the world population and is unmet by the options in today's clinics (Cohen et al., 2021;Gaskin and Richard, 2012;Klint et al., 2015;Macchia, 2022).Neuropathic pain is considered as one of the most difficult pain syndromes to treat and outcomes are often unsatisfactory (van Hecke et al., 2014).Botulinum toxin type A (BoNT/A) has gained awareness for its use as analgesic treatment for neuropathic pain for which it is frequently used off-label as a third line therapy (Deutsche Gesellschaft für Neurologie, 2019).It has also shown its efficacy in studies for diseases such as trigeminal neuralgia (Wu et al., 2012), migraine (Jackson et al., 2012;Shaterian et al., 2022), post herpetic neuralgia (Xiao et al., 2010) and diabetic neuropathy (Lakhan et al., 2015), and is included in national and international therapy guidelines (Bendtsen et al., 2019;Cruccu and Truini, 2017;Deutsche Gesellschaft für Neurologie, 2019;Finnerup et al., 2015;Hange et al., 2022;Moulin et al., 2014).
BoNT/A's well understood mechanism of muscle relaxation is based on its selective cleavage of Synaptosomal-Associated Protein 25 (SNAP-25) (Blasi et al., 1993), inhibiting acetylcholine release from its presynaptic vesicles at the neuromuscular synapse, leading to paralysis (Montecucco and Schiavo, 1995).Although it is well-known that BoNT/A impairs function of the neuromuscular endplate, muscle relaxation is unlikely the reason for pain relief in the context of neuropathic pain, which is often non-muscular (Cui et al., 2004a;H. J. Park et al., 2006).Several mechanisms contributing to BoNT/A's effect to alleviate neuropathic pain were suggested (Matak et al., 2019;J. Park and Park, 2017).Reduced activity of SNAP-25 may decrease release of neuropeptides from the sensory nerve endings, thus inhibiting sensitization (Durham et al., 2004;Purkiss et al., 2000).Furthermore, the toxin was suggested to lessen local inflammation around nerve endings (Cui et al., 2004b;Lucioni et al., 2008), involvement of TRPA1 was discussed (Burstein et al., 2014;Paterson et al., 2014), and reduced membrane expression of TRPV1 was observed (Shimizu et al., 2012).Also, the inhibition of voltage-gated sodium channels (Navs) by BoNT/A was reported (Shin et al., 2012).
Navs (Nav1.1 to Nav1.9) are essential for the initiation and propagation of action potentials (Catterall et al., 2020;Hodgkin and Huxley, 1952).Gain-and loss-of-function mutations in Nav1.7 are linked to chronic pain syndromes, stressing its importance in the generation of human pain (Cox et al., 2006;Cummins et al., 2004;Faber et al., 2012;Fertleman et al., 2006;Körner and Lampert, 2020).Effects of Nav mutations and their pharmacology can be assessed by patch-clamp in heterologous expression systems (Le Cann et al., 2021).Ideally, the impact of Nav mutations on neuronal excitability is assessed in sensory neurons of the patients themselves, which is almost impossible for obvious reasons.Patient derived induced pluripotent stem cells (iPSC) offer an alternative as they can be differentiated into sensory neurons, recapitulating the Nav pathophysiology in the patient's neuronal background (Meents et al., 2019).This system is currently used for disease modelling and identification of personalized medicine (Cao et al., 2016;Namer et al., 2019;Neureiter et al., 2022).
BoNT/A was reported to block Navs in rat hippocampal and sensory neurons at femtomolar concentrations (Shin et al., 2012).Here we set out to investigate the effects of BoNT/A on the pain relevant Nav isoforms Nav1.7 and Nav1.3 using a high-throughput automated patch-clamp system and manual patch-clamp recordings.Furthermore, we assessed the toxin's effect on neuronal excitability of iPSC-derived sensory neurons on multielectrode-arrays of a patient carrying the Nav1.7 gain-of-function mutation p.Q875E (Skeik et al., 2012;Stadler et al., 2015) leading to inherited erythromelalgia (IEM, a hereditary neuropathic pain syndrome) and a healthy control (HC).Indeed, we did not observe current reduction or biophysical alterations in Nav gating, nor a relevant change in activity of iPSC-derived sensory neurons upon application with BoNT/A.

Botulinum toxin for electrophysiology
The full-length toxin BoNT/A and the light chain (LC/A) (kindly provided by Ipsen, Bioinnovation Ltd, Oxfordshire, United Kingdom) was diluted in sterile water into stock solutions of 670 nM and kept at − 80 • C according to the manufacturers' recommendations.For manual patch-clamp experiments, the stock solution was diluted with recording solution directly prior to the experiment and used for up to two days (mostly only on the same day).The toxin was applied to the selected cell via a perfusor pen (250 μm, AutoMate Scientific, Perfusion Pencil) positioned within about 800 μm from the cell.For automated patchclamp experiments, the toxin was directly pipetted into the chambers of a 384 well recording plate during the measurements with a Biomek i5 pipette robot.We used 10 pM as the common concentration for both the full-length toxin (BoNT/A) and the light chain (LC/A), if not stated otherwise.
For high throughput patch-clamp experiments, cells were seeded days to one week in advance to achieve confluence of 80-90 %.The medium was exchanged a day before the recordings.Cells were washed twice with cold PBS-EDTA (PBS: PAN-Biotech; EDTA: Sigma-Aldrich) and separated with Accutase + 0.5 mg/ml DNAse (Roche Diagnostics GmbH).After 8 min of incubation at 37 • C, the external solution (External NMDG 60) was added, and the cells were cooled for 10 min at 4 • C. Next, the cells were separated using a fire polished pipette and transferred to a Teflon plate shaking in the automated patch-clamp device for 60 min at 10 • C and 200 rpm.

Manual patch-clamp experiments
HEK293T cells were recorded using an EPC 10 USB patch-clamp amplifier (HEKA Electronics, Germany).Measurements took place at room temperature (22 • C ± 2 • C, airconditioned).Recording patch pipettes were manufactured with a DMZ puller (Zeitz Instrument GmbH, Germany) and had a tip resistance between 1.0 and 2.0 MΩ.The external bath solution contained the following (in mM): 140 NaCl, 3 KCl, MgCl 2 , 1 CaCl 2 , 10 HEPES, 20 glucose.The pH was adjusted to 7.4 using NaOH and osmolarity was 300-310 mOsm.The patch pipette internal solution contained (in mM): 10 NaCl, 140 CsF, 1 EGTA, 10 HEPES, sucrose; pH was adjusted to 7.3 with KOH and osmolarity to 300-310 mOsm.Capacitive transients were cancelled, and series resistance was compensated by at least 75 %.Leak currents were corrected with the P/4 procedure.For all protocols, the holding potential (Vhold) was set at − 120 mV.Liquid junction potential of +8.2 mV was corrected in all experiments.After establishing the whole-cell configuration, inward Na + currents were allowed to stabilize in a "monitor-protocol" for about min during repeated − 20 mV depolarization steps before starting the recordings.
Sampling rates were set to 100 kHz for "monitor" and activation protocol and a 10 kHz low-pass filter was used.Current-voltage relations of activation were obtained with depolarizing steps from the holding potential starting from − 90 mV onto +50 mV with 10 mV incremental steps with an interval of 5 s.The Nav conductance G Na was calculated using the following equation: G Na = I Na /(V m − E rev ), where I Na is the peak of the current at the voltage V m , and E rev is the reversal potential for sodium, which was determined for each cell individually.Activation curves were derived by plotting normalized G Na as a function of test potential and fitted with the Boltzmann equation: , where G Na,max is the maximum sodium conductance, V 1/2 is the membrane potential at half-maximal activation, V m is the membrane voltage and k is the slope factor.
Tonic block of Navs was assessed using a pulse protocol that depolarized to − 20 mV with 10 s (0.1 Hz) between each sweep for a duration of about 13 min (80 sweeps).The ratio of the current following BoNT/A treatment to that before was calculated as "I post /I pre ", describing the relative current reduction of the cell after substance administration during pulse stimulation.I post /I pre is calculated as the quotient of the mean value of the last two maximum peak currents before substance administration (I pre ) and the mean value of the last two maximum peak currents after substance administration (I post ).

Automated patch-clamp system
Automated patch-clamp experiments were performed with the Syn-chroPatch 384i (Nanion Technologies GmbH, Munich, Germany).For all measurements "1× S-Type" chips (2 μm holes) were used and data acquisition was enabled by the software PatchContol384 (Nanion Series resistance was compensated by 70 %.Leak currents were corrected automatically with the P/4 procedure.For all protocols, the holding potential (Vhold) was set to − 120 mV.Liquid junction potential of +12 mV was corrected in all experiments.After establishing the whole-cell configuration, a quality control protocol was applied: three depolarizing sweeps to 0 mV.Cells with a current below − 200 pA or over − 10 nA were excluded, as well as cells with a seal below 300 MΩ.Fast inactivation, use dependency (10 Hz) and activation protocols were performed in each cell before and after toxin application.
The channels' activation was accessed using 100 ms pulses to a range of test potentials (-85 mV to + 30 mV) in 5 mV steps with an interval of 5 s.Steady-state fast inactivation was measured using a series of 450 ms pre-pulses from − 130 mV to − 20 mV followed by a 50 ms test pulse to 0 mV that assessed the non-inactivated transient current.Use dependent Nav block was recorded at 10 Hz with a series of 10 depolarizing steps for 5 ms to 0 mV.During a ≥13 min long pulse protocol with depolarizing steps to − 10 mV and an interval of 4 s, the toxins were added either to the intracellular or extracellular recording solution.The protocol had at least a duration of 13 min, however due to limitations of the solution exchange unit it was prolonged to up to 19 min for some experiments.This was then followed by the same protocols that were measured before, namely fast inactivation, use dependency 10 Hz and an activation protocol to compare before and after.
Activation curves and current reduction I post /I pre were evaluated in the same way as for manual patch-clamp experiments.Activation curves were again derived by plotting normalized G Na as a function of test potential and fitted with the Boltzmann equation.Inactivation curves are displayed as normalized peak inward current amplitudes (I Na /I Na, max ) at each test pulse as a function of pre-pulse potential and fitted using the following (Boltzmann) equation: the potential of half-maximal inactivation) and the slope factor k.For use dependency analysis, I 10 /I 1 describes the current reduction during the 10 Hz protocol after substance administration.It is a quotient formed from the peak current of the 10th sweep (I 10 ) and the first peak current (I 1 ).A detailed overview of the values is provided in Supplementary Table 1.

IPSC-derived sensory neurons
Two different iPS cell lines with two independent differentiations were investigated.The cell lines were obtained from a healthy control (HC) and a patient suffering from IEM, a chronic neuropathic pain condition caused by a gain-of-function mutation in Nav1.7, in this case the Q875E mutation (Stadler et al., 2015).The cell samples were reprogrammed into iPSCs and afterwards differentiated into sensory neurons using a viral forced NGN-1 expression approach (Schrenk- Siemens et al., 2022).Starting from d+84 of differentiation, sensory neurons' spontaneous activity was measured with a multielectrode-array (MEA) whilst treated with BoNT/A 10 pM or vehicle (Fig. 1).

Reprogramming
IPSCs derived from the blood cells of an IEM patient were used for differentiation to sensory neurons.Reprogramming was performed with written consent from the patient and her parents (ethics committee of the medical faculty of the RWTH, reference number EK243/18).Briefly, peripheral blood mononuclear cells (PBMCs) were reprogramed into iPSCs using the CytoTune-iPS 2.0 Sendai Reprogramming Kit (Thermo Fisher Scientific, Waltham, MA, USA) containing Sendai virus vectors for OCT4, KLF4, SOX2 and c-MYC.Single clone from the IEM patient was used for differentiation to sensory neurons.Reprogramming of the IEM patient cells was not part of this study.Its original work by Anil Kumar Kalia is currently under revision.
For the HC, iPSCs were derived from fibroblasts of a healthy donor and reprogrammed by the Huntington' Disease iPSC Consortium (Los Angeles, USA) (HD iPSC Consortium, 2012).

Virus stock production
Lentiviruses for co-expression of NGN1, eGFP and a puromycin resistance or rtTA were produced in HEK293T cells by co-transfection of the three helper plasmids pRSV-REV, pMDLg/pRRE, pMD2.G and FUW-TetO-Ngn1-P2A-EGFP-T2A-Puro or FUW-rtTA (all obtained from Katrin Schrenk-Siemens, Institute of Pharmacology, University of Heidelberg), respectively, using calcium phosphate transfection.Lentivirus containing cell culture supernatant was collected 48 and 72 h post-transfection.Both lentiviruses for expression of NGN1 and rtTA were pooled for further concentration.For that, the lentivirus-containing medium was mixed with 80 μg/ml Polybrene and 80 μg/ml Chondroitinsulphate (Sigma Aldrich), incubated for 20 min at 37 • C, centrifuged for 20 min at 2000 rpm and resuspended in serum free medium (Gibco) supplemented with 20 mM HEPES.Viral titre was determined by transduction of 1 × 10 5 HEK293T cells with a defined volume of the virus solution.Percentage of transduced cells was determined by flow cytometry (BD FACSCanto) 24 h after induction of eGFP expression by adding 2 μg/ml doxycycline to the cell culture medium (DMEM (Gibco), 10 % FBS (Biochrom)).

Immunostainings
At d+74 of sensory neuron differentiation, cells were fixed with 4 % cold PFA for 10 min, permeabilized with either 0.1 % Triton X-100 in PBS, Saponin permeabilization buffer (0.1 % Saponin, 2 % BSA in PBS) or not permeabilized.Non-specific antibody binding was blocked with % normal goat serum diluted in PBS or saponin staining buffer.Primary antibody incubation was performed at 4 • C for several hours or overnight.After 3 times washing, the secondary antibody staining was performed at room temperature for 2 h.The staining procedure for each antibody is listed in Table 1.A secondary antibody control was performed for each condition.For Imaging, a LSM 700 confocal microscope (Carl Zeiss) was used (Supplementary Fig. 1).

Multielectrode-array recordings
For both the cell lines (HC or IEM), we treated one sample of sensory neurons with and one sample without the full toxin BoNT/A 10 pM (Fig. 1).The recordings took place every day for 14 days starting from d+84 of differentiation.We chose this time frame since most neuropathic patients treated with BoNT/A in clinics usually experience pain relief in the span of two weeks (Bohluli et al., 2011;Ranoux et al., 2008;Sotiriou et al., 2009;Yuan et al., 2009;Zhang et al., 2014;Zúñiga et al., 2008).
On the first recording day (rd+0), two measurements were performed with the first one representing a baseline scan before any treatment and the second one being performed 30 min after adding either BoNT/A or vehicle control, respectively.On all subsequent days, only one recording was performed for each chip which was carried out before medium exchange to avoid the influence of mechanical stimuli to the cells as a source of error.50 % of the medium was changed three times a week with a fresh preparation of BoNT/A 10 pM from the stock in the treatment group.
The recordings took place with the MaxTwo Multiwell high density CMOS MEA system from Maxwell Biosystems (Zurich, Switzerland).Each well of the 6-well plate contains a chip including 26,400 platinum electrodes in a sensing area of 3.85 × 2.10 mm 2 and 1024 recording channels with a sampling rate of 10 kHz.The wells were pretreated with 1 % Tergazyme (Merck) for 1 h and sterilized with 70 % EtOH for min.After preconditioning of the chip for 24 h with neurobasal medium (Gibco), the whole chip area was coated with Poly-Ornithine (Sigma Aldrich), Fibronectin (Gibco) and Laminin (Sigma Aldrich).At day +5 of differentiation, 50,000 cells were seeded on the whole area of one chip.Each of the two different cell lines was plated on two chips of which one chip was assigned to a BoNT/A treatment, while the other was assigned to a vehicle control.
MEA recordings were performed at 37 Software v.21.1.22was used for the recordings and analysis of data.Spontaneous activity of the neurons was measured using the manufacturers' activity scan assay where the whole chip area was scanned with a checker-board configuration (recording every second electrode, recording time for each electrode 45 s, total scan duration 11 min).The spike detection threshold was set to five times the root mean square noise level of the bandpass-filtered voltage trace (300 Hz) and a second cut-off frequency (3000 Hz).Active electrodes were identified based on their firing rate and spike amplitudes obtained from the activity scan.More precisely, an electrode was considered active if the firing rate detected by this electrode (electrode spike count divided by time the electrode was recorded) exceeds 0.1 Hz and the 90th percentile of the negative amplitudes of all spikes detected on that electrode exceeds 20 μV, respectively.Due to limited samples (one measuring well for each condition, healthy control-BoNT/A; IEM-BoNT/A, healthy controlvehicle; BoNT/A-vehicle), we looked at the spiking activity on electrode level and not the total mean firing rate per well.The high density of electrodes of the CMOS MEA system leads to the problem that an action potential from a single neuron is typically detected by more than one electrode (Jäckel et al., 2012).To avoid redundant information hampering the independence of observations required for common statistical tests, spike filtering was done using MATLAB 2020 (The Mathworks, Inc) before further statistical analysis.Therefore, if at any given sampling point, a spike was detected on more than one electrode, we first extracted the position of the electrode recording the highest spike amplitude, el max , which was considered to be the closest electrode to the spike-generating neuron.All other simultaneous spike detection events were removed from the spike count statistics of the individual electrodes, if their Euclidian distance to el max was below 250 μm suggesting that the events originated from the same neuron (Yger et al., 2018).Inspection of the spatial distribution of active electrodes revealed clusters of active electrodes closely resembling the neuronal clusters observed in the immunostaining of cells grown on cover slips (Supplementary Figs. 1 and 2).

Western Blot
IPSC differentiated sensory neurons (d+40) from healthy control were either treated with three dilutions (3 nM, 10 pM, 1 pM) of BoNT/A and incubated for 48 h or left untreated as a control at 37 • C and 5 % CO 2 .HEK Nav1.5 cells were used as a negative control that do not express SNAP-25.Cells were washed three times with ice cold PBS, lysis buffer was added to each sample and incubated for 1-2 min on ice.The lysate was transferred to pre-chilled Eppendorf tubes and maintained under constant agitation for 30 min at 4 • C. Lysates were centrifuged and the supernatant transferred to be prepared with reducing NuPage LDS Sample Buffer 4× (Invitrogen; #NP0007) supplemented with dithiothreitol (DTT, Thermo Fischer).The electrophoresis chamber was assembled using NuPage 12 % Bis-Tris gels, 1 mm x 15 wells (Invitrogen; #NP0343BOX), NuPage MOPS SDS running buffer (20X) (Invitrogen; #NP000102) and NuPage antioxidant (Invitrogen; #NP0005).The gel was transferred to a 0.2 μM nitrocellulose membrane (BioRad; mini #1704158) and after blocking for 1 h with 5 % non-fat milk, incubated with the Anti-SNAP25 primary antibody (Sigma; #S9684) over night.The secondary antibody, HRP-conjugated Anti-Rabbit IgG (Invitrogen; #A16096), was incubated for 1 h.Proteins were visualized using Intas Chemostar PC ECL, Fluorescence Imager with Chemostar Touch (Intas-Science Imaging).

Data analysis and statistics
Manual patch-clamp data were analyzed and plotted using Fitmaster software (HEKA Elektronik), Igor Pro software (Wavemetrics), Graph Pad Prism 9 (GraphPad Software) and CorelDRAW 2017 (CorelDRAW Graphics Suite).Automated patch-clamp data were analyzed using DataControl384 Version 2.0 (Nanion Technologies GmbH) to depict and export results in a csv file for further database integration.An in-house Excel Add-Inn generated by Lennart Müller was used for analysis and Graph Pad Prism 9 (GraphPad Software) and CorelDRAW 2017 for graphing.
For sample size estimation, we performed an a priori analysis for experiments where two groups-assuming normal distribution-were compared (d = 0.8, α = 0.05, power = 0.7, N = 21 per group), for experiments with three groups (f = 0.4, α = 0.05, power = 0.7, total sample size = 54), and for experiments with five groups (f = 0.4, α = 0.05, power = 0.7, total sample size = 70) F-tests were performed.Sample size calculation was based on this analysis and other studies with similar investigations (Cox et al., 2006;Rühlmann et al., 2020).Studies were designed to generate groups of equal size.Inequality in group size may occur due to exclusion of individual cells during data analysis (e.g., high series resistance).
Statistical analysis was performed with Graph Pad Prism 9 (Graph-Pad Software).In case of one factorial tests and group number of two we used a t-test.In case of one factorial tests and number of groups bigger than two we used one-way ANOVA and Bonferroni's multiple comparison test for normal distributed data and otherwise Kruskal-Wallis test and Dunn's multiple comparison test.Two-way and three-way ANOVAs were used for multifactorial analyses without post hoc tests.Tests were only performed for studies where each group size (number of independent values) was at least n = 5.No outlier exclusion was performed.Statistical significance was defined as P ≤ 0.05 (*) and number of degrees of freedom (DF) are stated when data was analyzed with ANOVA as F (DFn, DFd).Statistical ANOVA results can be found in Supplementary Table 2. Exact value of n (number of cells) is outlined in the figure legends or in Supplementary Table 1.If not stated otherwise, data are presented as mean and error bars depict the standard deviation (SD).
Sample randomization was ensured in by measuring cells from more than one dish per group from varying passages if multiple time points of measurements were performed for one experiment.
For the analysis of the MEA experiments, individual observations (spike counts) were drawn from the electrode level.Due to the high number of electrodes of which a substantial proportion may not have contact to any neurons, the inclusion of all electrodes may lead to an artificial inflation of the degrees of freedom, thus increasing type I errors.Therefore, for our analysis, we only included electrodes which were considered as active electrodes in at least one of the 16 different recordings.
The complex experimental structure motivated the use of a mixed model considering spike count as the dependent variable and including the fixed effect between-factors "cell line" (IEM, healthy control) and "pharmacology" (BoNT/A, vehicle), the fixed effect within-factors "condition" (pre, post treatment) and "day of recording (rd0-rd14)" as well as a random (intercept) effect of the individual electrode.Moreover, we included the two-way interaction terms "cell line x pharmacology", "cell line x condition", "pharmacology x condition", "cell line x day of recording" and "pharmacology x day of recording", as well as the three-way interaction terms "cell line x pharmacology x condition" and "cell line x pharmacology x day of recording".Since it was our main focus to determine the effect of BoNT/A on neural activity (i.e.spike count) -both in general and depending on disease status (i.e.cell line), the interaction terms "pharmacology x condition" and "cell line x pharmacology x condition" were of primary interest (Supplementary Table 3).This statistical analysis was performed using GraphPad Prism 9 (GraphPad Software) and RStudio (RStudio Team (2020)).
Since spike counts or firing rates (i.e.spike counts/recording time) obtained from MEA recordings typically exhibit a non-normal right skewed distribution (Fig. 8C), statistical models requiring a normal distribution of the response variable are not well-suited for the analysis (Negri et al., 2020).From a theoretical perspective, for count data such as spike counts, models applying Poisson or negative binomial distributions should be ideal.The high percentage of zero counts motivated us to use a zero-inflated negative binomial generalized linear mixed model applying the R package "glmmTMB" Model (Brooks et al., 2017), Model fit was evaluated using the R package "DHARMa" (Hartig, 2023).
Blinding was not performed as high safety requirements in handling of Botulinum toxin required clear toxin traffic monitoring and disposal in special waste.This made blinding unsuitable.

Extracellular BoNT/A does not alter Nav1.7 biophysical characteristics in HEK cells
Published results suggested that Botulinum toxin can block sodium currents in sensory neurons (Shin et al., 2012).As Nav1.7 is essential for the sensation of pain in the human body (Körner and Lampert, 2020;Yu and Catterall, 2003), we assessed its potential inhibition by the toxin.We started our analysis with a concentration of 10pM BoNT/A, as this was reported to evoke robust inhibition of Navs in rat sensory neurons (Shin et al., 2012).
HEK cells stably expressing Nav1.7 were manually patch-clamped and peak inward current and voltage of half maximal activation was assessed (Fig. 2).10pM BoNT/A in the extracellular recording solution showed no significant alteration compared to time controls exposed to bath solution (Fig. 2A,B and C,D).To assess the effect of several concentrations we used an automated patch-clamp system to determine a dose-response curve ranging from 100 fM to 1 nM for tonic (0.1 Hz) and use-dependent (10 Hz) block of Nav1.7 (Fig. 3).In the automated setting as well, BoNT/A did not block Nav1.7 at any tested concentration, neither tonically (Fig. 3A,C) nor in a use dependent manner (Fig. 3B,D).In summary, the experiments show that Nav1.7 expressed in HEK cells was not affected by BoNT/A.We decided to maintain the concentration of 10 pM for the following experiments.

Intracellular BoNT/A and LC/A application does not alter Nav1.7 gating
The Botulinum toxin protein consists of a heavy and a light chain.The light chain is the catalytically active domain that acts as the protease.The heavy chain contains the translocation and the receptor binding domain with which the protein can bind to and is endocytosed into a synaptic vesicle (Binz and Rummel, 2009;Oguma et al., 1995).To exclude the possibility that the toxin may not have crossed the cell membrane to act on any potential intracellularly positioned part of the  C, relative current reduction I post /I pre before and after toxin application.Mean ± SD values were 0.73 ± 0.25 for Control (black circles, n = 13), 0.74 ± 0.18 for fM (blue squares, n = 19), 0.67 ± 0.24 for 1 pM (blue triangles up, n = 13), 0.73 ± 0.26 for 10 pM (blue triangles down, n = 22), 0.65 ± 0.23 for 100 pM (purple diamonds, n = 18), 0.82 ± 0.12 for 1 nM (purple circles, n = 6).No significant difference detected, Dunn's multiple comparison test.D, relative current reduction I 10 /I 1 obtained from I max values during the 10 Hz protocol (grey symbols = before toxin, colored symbols = after toxin).The data was analyzed with a two-way ANOVA that reviewed the between factor Toxin (Control; 100 fM BoNT/A; 1 pM BoNT/A; 10 pM BoNT/A; 100 pM BoNT/A; 1 nM BoNT/A) and the within factor Time (pre or post application).None of the different BoNT/A concentrations yielded a significant effect on I 10 /I 1 ratio as confirmed by the nonsignificant interaction term Toxin x Time with P = 0.5516.There was also no significant difference between the levels of the factor Toxin (P = 0.9762).Values can be found in Supplementary Nav, we decided to directly add the toxin into the intracellular recording solution.
In addition to adding the full neurotoxic protein as 10 pM BoNT/A, we decided to perform the experiments with only the catalytically active light-chain 10 pM LC/A as well, assessing Nav activation, fast inactivation, tonic and use-dependent inhibition.Intracellular solution was exchanged during the experiments, allowing for direct comparison of results pre and post application on the identical cell (Fig. 4).Similar to the extracellular toxin application, we could not observe any inhibitory tonic or use-dependent effect of the toxin applied intracellularly, neither in its full form nor with the light chain alone (Fig. 4A and B).There was no shift in activation or inactivation induced by the toxin (Fig. 4C-F).
For some toxins it is known that they have a boosting effect on local anesthetics at very low concentrations (Templin et al., 2015).As our experiments indicated no effect on Nav1.7 by BoNT/A, we also evaluated this effect by adding 100 μM of the local anesthetic mexiletine in a later, separate application step during the same measurements (Fig. 4G,  H).Cells showed the well-known significant inhibition of the sodium current when treated with 100 μM mexiletine in a tonic and use dependent manner, however there we observed no differences in the inhibitory effect between the groups pretreated with Botulinum toxin or untreated cells (Fig. 4G,H).In summary, BoNT/A full protein and the catalytic light chain had no inhibitory effects on Nav1.7 in HEK cells on tonic or use dependent block, activation and fast inactivation when applied extracellularly or intracellularly nor had it an impact on mexiletine's potency on this channel.

Lack of BoNT/A effect is not Nav subtype specific
As sensory neurons not only express Nav1.7,we decided to expand our experiments to another Nav subtype.Nav1.3 is discussed to be upregulated in injured sensory neurons and to account for higher firing rates in models of neuropathic pain (Cummins and Waxman, 1997;Kim et al., 2001;Lindia et al., 2005;Waxman et al., 1994).Thus, we set out to include this channel in our research as a possible BoNT/A target.
We performed automated patch-clamp experiments exposing Nav1.3 expressed in HEK cells to extra-or intracellular BoNT/A (Fig. 5).No difference between control and toxin exposed groups were identified when treated with the full toxin or only its catalytic active unit extra-or intracellularly (Fig. 5A).Also, a use-dependent block at 10 Hz was not visible with toxin application (Fig. 5B).There was no shift in the voltagedependence of activation or fast inactivation (Fig. 5C-F).In conclusion, BoNT/A does not alter the biophysical properties of Nav subtypes Nav1.7 nor Nav1.3 in heterologous expression systems.

Neuronal background of expression system does not unveil toxin's effect on Navs
Nav function also depends on associated proteins in its cellular expression context.To use a more physiological neuronal cellular background which may contain necessary accessory proteins that could be a possible requirement for BoNT/A's full effect, we analyzed the effect of BoNT/A on endogenous Navs in the neuronal cell line ND7/23.Endogenous sodium currents are mainly mediated by Nav1.6 and Nav1.7, but also Nav1.1, Nav1.2, Nav1.3 and Nav1.9 at low levels (Lee et al., 2019).We observed that neither intracellular nor extracellular application of BoNT/A, or LC/A resulted in a reduced current (Fig. 6A,B) or shifts in activation or fast inactivation (Fig. 6C-F).Mexiletine, on the other hand, showed a reliable tonic and use dependent block at 10 Hz (Fig. 6G,H).There was no change in mexiletine's potency upon toxin pre-incubation (Fig. 6G,H).Altogether, our data suggest that the endogenous sodium currents in this more neuron like model were not A, Tonic block: relative current I post /I pre , reduction obtained from I max values before and after toxin application with the same protocol from Fig. 2A.Mean ± SD values were 0.98 ± 0.14 for Control cells (black circles, n = 76), 0.97 ± 0.16 for cells with 10 pM BoNT/A (blue circles, n = 74), 0.97 ± 0.12 for cells with 10 pM LC/ A (green circles, n = 72).No significant difference detected, Dunn's multiple comparison test.B, use dependent block I 10 /I 1 : relative current reduction obtained from I max values during the 10 Hz use dependent protocol (grey circles = before toxin, colored circles = after toxin) with the same protocol from Fig. 2C.The data was analyzed with a two-way ANOVA that reviewed the factor Toxin (Group: Control; BoNT/A; LC/A) and Time (pre; post).There was no significant difference between the groups in relation to the time factor (Toxin x Time) with P = 0.2363.There was also no significant effect of the factor Toxin (P = 0.4654), but a significant effect of Time (P= <0.0001), indicating a small but consistent reduction of the I 10 /I 1 ratios across the different experimental conditions.Values can be found in Supplementary Table 1 and two 2. The used protocol for inactivation is depicted next to the graph.
G, Positive control: mexiletine (Mex 100 μM).The graph shows relative current reduction I post /I pre obtained from I max values before and after application of either substance.Mean ± SD values were 0.97 ± 0.12 for Control cells with vehicle (black circles, n = 39), 0.97 ± 0.09 for BoNT/A cells with vehicle (blue circles, n = 32), 0.96 ± 0.04 for LC/A cells with vehicle (green circles, n = 33), 0.51 ± 0.11 for Control cells with mexiletine (black circles on the right, n = 30), 0.45 ± 0.17 for BoNT/A cells with mexiletine (blue circles on the right, n = 36) and 0.53 ± 0.09 for LC/A cells with mexiletine (green circles on the right, n = 37).Cells treated with vehicle (extracellular solution) were not reduced in their current and did not differ in their tonic block properties in their subgroups (vehicle/BoNT/A and LC/A).Cells treated with mexiletine (Mex) were reduced in their current equally and did not differ in their subgroups (vehicle/BoNT/A and LC/A).Accordingly, the respective two-way ANOVA reviewed the factors: Toxin (e.g.Control; BoNT/A; LC/A) and Local anesthetic (vehicle; mexiletine) and revealed a significant effect of mexiletine as compared to the control condition (P= <0.0001), but no significant effect of the Toxins (P = 0.0872) as well as no significant interaction between the two factors (P = 0.0641).Two-way ANOVA results can be found in Supplementary Table 2. H, The right graph shows the relative current reduction obtained from I max values during the 10 Hz use-dependent protocol (grey circles = before mexiletine or vehicle, colored circles = after mexiletine or vehicle).The data was analyzed with a three-way ANOVA that included the factors: Toxin (e.g.Control; BoNT/A; LC/A), Local anesthetic (vehicle; mexiletine) and Time (pre; post).Mexiletine successfully blocks the Nav1.7 current in a tonic and use dependent manner (significant interaction of Local anesthetic x Time; P < 0.0001).There was no significant effect of the Toxin (P = 0.2610) and no significant interaction of Toxin x Local anesthetic (i.e. the effect of mexiletine did not change when one of the toxins was co-administered; P = 0.3053).Values can be found in Supplementary Table 1 and three-way ANOVA results in Supplementary Table 2.

A.B. Kesdogan et al. (caption on next page)
A.B. Kesdogan et al. affected by Botulinum toxin application.

Longer toxin exposure does not change the outcome
So far, we only assessed acute effects of toxin exposure.To assess potential effects of longer time exposures, we added the toxin either as its full protein or only the light chain to the culture medium of ND7/23 cells 24 h before recordings.Control cells were seeded and prepared at the same time.We carried out the same protocols and examined for an inhibitory effect by either tonic or use dependent block or any shifts in voltage dependency of activation or fast inactivation.We did not detect any clear toxin effects in these experiments.There was no significant change in the tonic block peak current before and after application compared with the Control (Fig. 7A).Mexiletine successfully blocked the sodium current in a use-dependent manner (Fig. 7B).Treatment with BoNT/A was associated with a slight reduction of the I 10 /I 1 ratio.This effect was very small in comparison to mexiletine's effect (see Supplementary Table 1; mex on average 33 % reduction, BoNT/A < 1.5 % reduction) and in contrast to the results of LC/A (no reduction, see Supplementary Table 1).There was no shift in voltage-dependence of activation when incubated with the toxin (Fig. 7C,D).We detected a small but significant shift of voltage-dependence of fast inactivation for Control vs. LC/A cells with P = 0.0346 and BoNT/A vs. LC/A with P = 0.0340.Nevertheless, the total mean differences were rather small (1.67 mV and 1.68 mV (Fig. 7E,F)).These results suggest that prolonged toxin exposure does not result in changes in Nav properties and are in line with our results using acute toxin exposure.

Hyperexcitability of pain patient derived iPSC-sensory neurons is unaltered by BoNT/A
So far, we were not able to see any BoNT/A effect on Navs in heterologous expression systems or in a neuronal cell line.As the clinical effects of BoNT/A in neuropathic pain are evident, we strived for a more relevant model for neuropathic pain.To this end, we differentiated iPSCs of the patients carrying the disease-causing gain-of-function mutation pQ875E of Nav1.7 (Skeik et al., 2012;Stadler et al., 2015) and a healthy control (HD iPSC Consortium, 2012) into sensory neurons on multielectrode-arrays.Successful differentiation was confirmed by immunostainings (Supplementary Fig. 1): Cells expressed Tuj1 (neuron specific marker), Peripherin (peripheral neuronal marker), NF200 (sensory neuronal marker), cGRP (peptidergic marker), ISL1 (transcription factor), Substance P (peptidergic marker) and TRPV1.
An exemplary overview of the active electrodes of the MEA chip harboring healthy iPSC-derived sensory neurons can be seen in Fig. 8A.There is no major difference in the activity level when comparing pre and post BoNT/A treatment for one day.For a better overview of the different wells see Supplement Fig. 2. Fig. 8B illustrates raw data showing band pass-filtered voltage traces obtained from a representative electrode from the chip containing healthy iPSC-derived sensory neurons with an overlay of the spike event waveforms that were used to assess the activity.
To assess a potential time course of toxin effects, we treated one MEA well with BoNT/A 10 pM (Control-BoNT/A; IEM-BoNT/A) and one sample with vehicle (Control-vehicle; BoNT/A-vehicle) for both control and patient cells.Due to limited Cells were investigated daily for 14 days.Analysis of detected spike counts i.e., firing rate over that period of time shows that the activity measured looks very similar between BoNT/ A treated or untreated cells (Fig. 8C).
As expected, the iPSC-derived neurons from the inherited erythromelalgia patient exhibited a higher spiking activity in total compared to the healthy control cell line (Fig. 8D), which was treatment independent.Similar to the healthy cells, spontaneous activity of the patientderived sensory neurons was unaffected by BoNT/A at any time point investigated.The difference between the pre and post treatment condition in total did not differ between the BoNT/A and the vehicle control condition (Fig. 8D), as confirmed by the non-significant interaction term "pharmacology x condition" of our statistical model (χ2(df = 1) = 0.2155; P = 0.6425) (Supplementary Table 3).This tendency did also not differ between the IEM patient and the healthy control as reflected by the non-significant three-way interaction term (χ2(df = 1) = 3.6856; P = 0.0549 (Supplementary Table 3).All in all, the iPSC-derived sensory neurons did not show any susceptibility to the toxin.

BoNT/A activity confirmation
To confirm that the toxin used in the presented experiments was intact and functional at the same concentrations applied to iPSC-derived neurons, we performed Western blot experiments with the identical stock used for the experiments after their completion.The well-known target at low concentrations of Botulinum toxin is SNAP-25 (Blasi et al., 1993).We incubated iPSC-derived sensory neurons at d+40 for Fig. 5. BoNT/A lacks an effect on Nav1.3 Nav1.3 cells were exposed intracellularly (yellow boxes) and extracellularly to 10 pM BoNT/A and LC/A during automated patch-clamp recordings.A, Relative current reduction I post /I pre obtained from I max values before and after intracellular toxin application in yellow boxes.Mean ± SD values were 1.02 ± 0.10 for Control cells (black circles, n = 48) and 1.00 ± 0.19 for cells treated with LC/A (green circles, n = 42).Next to it, relative current reduction I post /I pre after extracellular toxin application.Mean ± SD values were 0.91 ± 0.32 for Control cells (black circles, n = 37), 0.95 ± 0.19 for cells treated with BoNT/A (blue circles, n = 29), 0.96 ± 0.08 for cells treated with LC/A (green circles, n = 32).No significant difference detected, Dunn's multiple comparison test.B, the graph shows the relative current reduction I 10 /I 1 during the 10 Hz protocol (grey symbols = before toxin, colored symbols = after toxin, yellow box = intracellular solution exchange).There were no differences in between the subgroups of toxin treated (either intra-or extracellularly) or Control cells.The data was analyzed with a two-way ANOVA applying the factor Toxin (Control ICS; LC/A ICS; Control ECS; BoNT/A ECS; LC/A ECS) and Time (pre; post).None of the different toxins yielded a significant effect on I 10 /I 1 ratio as confirmed by the non-significant interaction term (Toxin x Time, P = 0.5648).There was also no significant effect of the factor Toxin itself (P = 0.1900).Values can be found in Supplementary Table 1 and two-way ANOVA results in Supplementary Table 2. C, voltage dependence of channel activation.The protocol for activation is the same as in Fig. 3C.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the activation curve in comparison to the Control.The curves are shifted to more hyperpolarized potentials in a time dependent manner (time induced shift toxin independent).D, values for half-maximal voltage dependence of activation (V 1/2 ) obtained from Boltzmann fits for individual traces.Mean ± SD values were − 40.02 ± 8.78 mV for Control Cells ICS (black triangles, n = 23), − 42.35 ± 8.23 mV for LC/A ICS cells (green triangles, n = 30), − 44.77 ± 9.37 mV for Control Cells ECS (black circles, n = 21), − 38.44 ± 6.13 mV for BoNT/A ECS cells (blue circles, n = 23), − 39.21 ± 7.36 mV for LC/A ECS cells (green circles, n = 22).No significant difference detected, Dunn's multiple comparison test.E, steady-state fast inactivation of the channel.The applied protocol is the same as in Fig. 3E.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the inactivation curve in comparison to the Control.The curves are shifted to more hyperpolarized potentials in a time dependent manner (time induced shift).F, half-maximal voltage-dependent fast inactivation values obtained from Boltzmann fits of individual traces.Mean ± SD values were − 74.41 ± 6.38 mV for Control Cells ICS (black triangles, n = 30), − 76.41 ± 6.41 mV for LC/A ICS cells (green triangles, n = 35), − 80.83 ± 5.61 mV for Control Cells ECS (black circles, n = 28), − 78.41 ± 5.17 mV for cells treated with BoNT/A ECS (blue circles, n = 27), − 76.92 mV ± 6.49 mV for cells treated with LC/A ECS (green circles, n = 30).No significant difference detected, Dunn's multiple comparison test.48 h with three different concentrations: 3 nM, 10 pM and 1 pM of whole protein Botulinum toxin type A. The results show that the iPSC-derived sensory neurons express the protein SNAP-25 whereas HEK cells used as controls do not (Fig. 9).The toxin clearly cleaves the protein already at concentrations as low as 1 pM (Fig. 9), leaving no doubt about the activity of the used toxin on its natural functionality.In summary, our results indicate toxin integrity and proper cellular uptake in our model system.

Discussion
Over recent years, Botulinum toxins' beneficial effect in neuropathic pain has been demonstrated.Based on earlier work suggesting a role of Navs in BoNT/A effects (Shin et al., 2012), we aimed to pharmacologically characterize the specific mode of action on Nav subtypes.However, with various methodological approaches, substance derivates and model systems, we were not able to show any significant effect of the toxin on Nav gating.
From all characteristics investigated, two biophysical readouts reached statistical significance namely a left-shift of voltage dependence of fast inactivation in pre-incubated ND7 cells treated with LC/A (Fig. 7F) in comparison to cells treated with a control or BoNT/A.However, the respective p-values (P = 0.0346 and P = 0.0340) fell just slightly below the conventional threshold of p < 0.05 and the corresponding mean differences were small (1.68 mV and 1.67 mV), questioning the physiological relevance of this finding.Finally, it remains unclear why this effect is only observed for LC/A, but not the original toxin BoNT/A.Additionally, we detected a small statistically significant effect in the use-dependent block analysis of the same experiment.The treatment with BoNT/A showed a slight reduction of the I 10 /I 1 ratio (Toxin x Time, P = 0.0296) (Fig. 7B).This effect was very small in comparison to mexiletine's effect (see Supplementary Table 1), and in contrast to the results of LC/A.
We used a high toxin concentration of 10 pM for our experiments in comparison to the usually administered dose in neuropathic pain treatment ranging from 50 to 200 Units (Deutsche Gesellschaft für Neurologie, 2019) which corresponds to a concentration of about 3 pM-12 pM (Field et al., 2018).The local effective concentration is expected to be lower (Richter and Jacobsen, 2014).Thus, also small effects should have been measurable.Shin et al., (2012) used hippocampal rat neurons and rat dorsal root ganglions (DRG) to illustrate BoNT/A's activity at very low concentrations.They report an inhibitory effect of BoNT/A's which seemed to be similar in TTX sensitive and TTX resistant Nav subtypes.The results of our experiments contradict the findings of Shin et al. (2012), as we did not observe any Nav block in the numerous experimental settings used.Sodium currents in heterologous expression systems and primary cultivated neurons show a variation in current density, especially in the first minutes after establishing the whole cell configuration (Coste et al., 2004;Todt et al., 1999).This effect is associated with Cesium fluoride (CsF) in the intracellular solution (Qu et al., 2000;Saab et al., 2003), as also used by the Shin et al. group.For this reason, it is common practice to apply pulses for the first 3-5 min ("monitor protocol") to ensure the current amplitude reached a steady state and is not overlaid by peak current densities due to adjustment of the experimental setting (Meents and Lampert, 2016).It is possible that Shin et al. did not use such a protocol.Not only the current density, but also the voltage dependence of activation and fast inactivation changes time dependently in whole cell patch-clamp recordings (Coste et al., 2004).Thus, a further explanation for the divergent results may arise from the lack of time-controlled experiments in Shin et al., which would have ruled out a potential bias due to shifts in the voltage dependency in whole cell configuration, as we have observed it in our recordings (Fig. 4C, E, 5C, E, 6C, E).The study of Shin et al. focused on BoNT/A's effects in rodent cells, i.e. rat DRG neurons and rat hippocampal neurons.We used cells of human origine, such as HEK cells and human iPSC-derived sensory neurons.The ND7/23 cells, however, are a fusion cell line of embryonic rat DRG and mouse neuroblastoma, and thus Fig. 6.BoNT/A has no effect on endogenous sodium currents in ND7/23 cells ND7/23 cells were exposed to intracellular or extracellular 10 pM BoNT/A and LC/A or vehicle (either intracellular or extracellular solution) in automated patchclamp recordings.A, Relative current reduction I post /I pre obtained from I max values before and after intracellular toxin application.Mean ± SD values were 0.95 ± 0.21 for Control cells (black circles, n = 37), 1.07 ± 0.34 for 10 pM BoNT/A cells (blue circles, n = 44), 0.99 ± 0.17 for 10 pM LC/A cells (green circles, n = 36).B, shows relative current reduction I post /I pre obtained from I max values before and after extracellular toxin application.Mean ± SD values were 1.26 ± 0.37 for Control cells (black circles, n = 38), 1.17 ± 0.37 for 10 pM BoNT/A (blue circles, n = 31), 1.11 ± 0.22 for 10 pM LC/A cells (green circles, n = 42).No significant difference detected, Dunn's multiple comparison test.C, voltage dependence of channel activation is depicted.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the activation curve in comparison to the Control.D, values for half-maximal voltage dependence of activation (V 1/2 ) obtained from Boltzmann fits for individual traces.Mean ± SD values were − 39.70 ± 8.03 mV for Control cells (black circles, n = 36), − 36.15 ± 7.94 mV for cells treated with 10 pM BoNT/A (blue circles, n = 41), − 38.22 ± 8.06 mV for cells treated with 10 pM LC/A (green circles, n = 35).No significant difference detected, Bonferroni's multiple comparison test, ANOVA results in Supplementary Table 2. E, steady-state fast inactivation of the channel is depicted.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the inactivation curve in comparison to the Control.There is a time induced shift in the inactivation curves between the cells before and after substance administration.G, tonic block with extracellular application of mexiletine or vehicle (extracellular solution).Relative current reduction I post /I pre obtained from I max values before and after application of either substance.Mean ± SD values were 0.94 ± 0.09 for Control cells with vehicle (black left circles, n = 17), 0.98 ± 0.16 for BoNT/A cells with vehicle (blue left circles, n = 20), 0.96 ± 0.03 for LC/A cells with vehicle (green left circles, n = 17), 0.62 ± 0.10 for Control cells with mexiletine (black right circles, n = 18), 0.66 ± 0.05 for BoNT/A cells with mexiletine (blue right circles, n = 17), 0.66 ± 0.10 for LC/A cells with mexiletine (green right circles, n = 18).The data was analyzed with a two-way ANOVA that included the factor Toxin (Control; BoNT/A; LC/A) and Local Anesthetic (vehicle; mexiletine).There was no significant effect of the Toxin (P = 0.2033).Mexiletine treatment consistently induced a reduction of I post /I pre ratios across all the different toxin conditions as confirmed by the significant effect of the factor Local Anesthetic (P= <0.0001) and the non-significant interaction term (Toxin x Local Anesthetic, P = 0.8775).Two-way ANOVA results can be found in Supplementary Table 2. H, relative current reduction I 10 /I 1 obtained from I max values during the 10 Hz use-dependent protocol (grey circles = before mexiletine or vehicle, colored circles = after mexiletine or vehicle).Mexiletine successfully blocks the sodium current in a tonic and use-dependent manner.There we no differences in between the subgroups mexiletine or vehicle treated cells.The data was analyzed with a three-way ANOVA that reviewed the factors Toxin (e.g.Control; BoNT/A; LC/A), Local Anesthetic (vehicle; mexiletine) and Time (pre; post).Application of the different toxins did not lead to significant change of I 10 /I 1 ratios (Toxin x Time, P = 0.2066).As expected, mexiletine treatment consistently induced a reduction of I 10 /I 1 ratios across all the different toxin conditions as confirmed by the significant interaction term Local Anesthetic x Time (P= <0.0001) and non-significant interaction term Toxin x Local anesthetic x Time (P = 0.2170).Values can be found in Supplementary possess more "rodent-like" components (Lee et al., 2019;Wood et al., 1990).Nevertheless, we were unable to reproduce the previously published findings in these cells.Still, the divergent findings could also result from species differences.
While heterologous expression systems are very helpful to study the biophysics of ion channel gating, rodent models aid to investigate excitability on a cellular and systemic level.However, the latter have been shown to exhibit significant differences in their nociceptive system compared to the human nociceptive system.This discrepancy makes it challenging to translate findings based on animal models, potentially limiting their relevance in translational research (Rostock et al., 2018).
IPSC-derived neurons offer the unique possibility to study neuropathic pain mechanisms in cells carrying the patient's genetic background.These "disease-in-a-dish" models for neuropathic pain have been shown to replicate important parts of clinical phenotype (Cao et al., 2016;Meents et al., 2019;Mis et al., 2019) and have led to identification of personalized medical treatments for neuropathic pain (Namer et al., 2019).We made use of this model to study the potential inhibitory effects of BoNT/A on control and patient derived sensory neurons.The patient derived sensory neurons carrying the Nav1.7/Q875Emutation showed enhanced unevoked activity, mimicking findings observed in peripheral nerve fibers of neuropathic pain patients (Kleggetveit et al., 2012).BoNT/A did not alter this activity suggesting that its pain-relieving effect is not mediated by direct interference with neural activity, neither via Navs nor via other ion channels or signaling mechanisms represented in the iPSC-derived sensory neurons.As its well-known mechanism of action, inhibition of vesicle release, is preserved and functioning also in this experiment, these results also show, that spontaneous activity in erythromelalgia iPSC-derived sensory neurons is most likely intrinsic and not dependent on synaptic stimulation or release of neuropeptides from other neurons in the dish.
BoNT/A was shown to hinder trafficking of the TRPV1 capsaicin receptor (Morenilla-Palao et al., 2004;Shimizu et al., 2012), and a similar mechanism was suggested for Navs (Matak et al., 2019).It has been shown that Nav trafficking involves vesicle dependent anterograde and retrograde transport (Liu et al., 2005;Mercier et al., 2018).Yet in contrast to the TRPV1 channel, there is no data of SNAP-25 dependent vesicle transport in Nav trafficking.However, there is data on the half-life of neuronal Nav subtypes in the membrane.It has been estimated that it ranges from about 17 to 50 h, depending on cell type and culture conditions (Mercier et al., 2018;Monjaraz et al., 2000).If Botulinum toxin type A does interfere with the trafficking of voltage-gated Navs, we should have seen an effect during the 14-day time span of our MEA experiments.
BoNT/A's C-terminal binding domain region is homologous to those of the extracellularly secreted fibroblast growth factors FGF1, FGF2 and FGF9, and is thus a potential ligand for Fibroblast Growth Factor Receptors (FGFRs) (Jacky et al., 2013).FGFs are signaling molecules for the development, maintenance and repair of the brain (Guillemot and Zimmer, 2011;Klimaschewski and Claus, 2021).Mainly the non-secreted intracellular FGFs (FGF homologous factors, FHFs) contain an interaction site with Navs and alter their gating, current density and localization (Körner and Lampert, 2020;Pablo et al., 2016).As the FHFs are the factors FGF11 to FGF14, it is unlikely that BoNT/A directly interferes with Navs via the FHF binding site.If such a regulation had occurred nevertheless, we would have observed it in our MEA experiments.
The here presented results, while negative, show thorough evidence, that voltage-gated sodium channels are no target of BoNT/A, contradicting previous findings (Shin et al., 2012) that have been discussed subsequently (Matak et al., 2019;J. Park and Park, 2017).Thus, our study helps to enhance a scientific system that produces reproducible, reliable results (Baker, 2016).Since Botulinum toxin's analgesic effect in clinics is undisputed (Bendtsen et al., 2019;Cruccu and Truini, 2017;Deutsche Gesellschaft für Neurologie, 2019;Finnerup et al., 2015;Hange et al., 2022;Moulin et al., 2014), the study also demonstrates that peripheral nociception can be relieved independently of voltage gated sodium channels.
Our study strongly suggests that voltage gated Navs are not involved in the analgesic potential of BoNT/A.Possible alternative pathways of the toxin include the inhibition of pain mediators, reduced expression of TRPV1 and reduced local inflammation leading to less sensitization of peripheral nerve fibers (Matak et al., 2019;J. Park and Park, 2017).Future studies which further investigate the exact mechanism of action of the toxin will help to optimize its use in pain treatment.Fig. 7. 24 h pre-incubation with BoNT/A in ND7/23 cells did not alter Nav currents ND7/23 cells were exposed to 10 pM BoNT/A or 10 pM LC/A for 24 h.Subsequently, mexiletine or vehicle (extracellular solution) was applied and endogenous Nav currents were recorded with automated patch-clamp.A, relative current reduction I post /I pre obtained from I max values before and after extracellular toxin application.Mean ± SD values were 1.03 ± 0.20 for Control cells with vehicle (black circles on the left, n = 37), 1.00 ± 0.21 for BoNT/A cells with vehicle (blue circles on the left, n = 36), 1.03 ± 0.13 for LC/A cells with vehicle (green circles on the left, n = 42), 0.68 ± 0.08 for Control cells with mexiletine (black circles on the right, n = 26), 0.68 ± 0.11 for BoNT/A cells with mexiletine (blue circles on the right, n = 29), 0.74 ± 0.18 for LC/A cells with mexiletine (green circles on the right, n = 22).The data was analyzed with a two-way ANOVA that reviewed the factor Toxin (Control; BoNT/A; LC/A) and Local anesthetic (vehicle; mexiletine).There was no significant effect of the Toxin (P = 0.3805).Again, mexiletine treatment consistently induced a reduction of I post /I pre ratios across all the different toxin conditions as confirmed by the significant effect of the factor Local anesthetic (P= <0.0001) and non-significant interaction term (Toxin x Local anesthetic, P = 0.5923).Values can be found in Supplementary Table 1 and twoway ANOVA results in Supplementary Table 2. B, relative current reduction I 10 /I 1 obtained from I max values during the 10 Hz use-dependent protocol (grey circles = before vehicle or mexiletine, colored circles = after vehicle or mexiletine).The data was analyzed with a three-way ANOVA that reviewed the factors: Toxin (e.g.Control; BoNT/A; LC/A), Local Anesthetic (vehicle; mexiletine) and Time (pre; post).Mexiletine successfully blocked the sodium current in a use-dependent manner (significant interaction of Local anesthetic x Time factor; P= <0.0001).Treatment with BoNT/A was associated with a slight reduction of I 10 /I 1 ratio, too (significant interaction of Toxin x Time, P = 0.0296), but this effect was very small in comparison to mexiletine's effect (see Supplementary Table 1).Values of three-way ANOVA results can be found in Supplementary Table 2. C, voltage dependence of channel activation.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the activation curve in comparison to the Control.D, values for half-maximal voltage dependence of activation (V 1/2 ) obtained from Boltzmann fits for individual traces.Mean ± SD values were − 25.94 ± 9.52 mV for Control cells (black circles, n = 72), − 26.60 ± 6.81 mV for cells with 10 pM BoNT/A (blue circles, n = 75), − 27.57± 6.87 mV for cells with 10 pM LC/A (green circles, n = 74).No significant difference detected, Dunn's multiple comparison test.E, steady-state fast inactivation of the channel is depicted.Continuous lines represent conductance voltage curves.F, half-maximal voltage-dependent fast inactivation values obtained from Boltzmann fits of individual traces.Mean ± SD values were − 71.03 ± 4.12 mV for Control cells (black circles, n = 72), − 71.04 ± 4.00 mV for cells incubated and treated with 10 pM BoNT/A (blue circles, n = 75), − 72.71 ± 3.81 mV.Bonferroni's multiple comparison test detected a significance between Control vs. LC/A cells with P = 0.0346 and a total mean difference of 1.68 mV and BoNT/A vs. LC/A with P = 0.0340 and a total mean difference of 1.67 mV, ANOVA results in Supplementary Table 2.  A, Overview of the active electrodes of the MEA chip harbouring iPSC-derived sensory neurons obtained from a healthy control before and after treatment with BoNT/A, respectively.Note that the distribution of active electrodes (yellow color code) closely resembles the neuronal clusters present in the culture as revealed by immunostaining (Suppl.Fig. 1).B, Band pass-filtered voltage traces obtained from a representative electrode from the chip containing iPSC-derived sensory neurons from a healthy control before treatment with BoNT/A.Red dots indicate spikes.On the right zoomed-in image, an overlay of the typical spike waveforms can be seen.The waveforms refer exactly to the spikes that can be seen in the zoomed-out image.C, Temporal trajectory of detected spike counts from iPSC derived sensory neurons obtained from a healthy control (upper panels) or a patient with inherited erythromelalgia (IEM, lower panels) before and after treatment with BoNT/A or a vehicle control, respectively.Inspection of the box plots reveals heavily right skewed distributions which is a common finding in MEA experiments and motivated us to use a negative binomial mixed model.D, 95 % confidence intervals of estimated marginal means (means which are predicted by the model for the different level combinations of the factors "cell line" (IEM, healthy control), "pharmacology" (BoNT/A, vehicle) and "condition" (pre, post treatment) while controlling for the day of recording) given on the model (i.e.logarithmic) scale.As our main finding, the difference between the pre and post treatment condition did not differ between the BoNT/A and the vehicle control group.IPSC-derived neurons from the IEM patient (right panels) exhibited a higher spiking activity in total compared to the healthy control (left panels).

Fig. 1 .
Fig. 1.Pictogram of the experimental set-up for multielectrode-array (MEA) recordings Cell samples derived from a chronic pain patient with IEM and from a healthy control were reprogrammed to iPSCs.Reprogramming of the cells was not part of this study.Differentiation of iPSC into sensory neurons was accomplished.Cells were recorded on multielectrode-array and treated for 14 days with BoNT/A or vehicle.IEM, inherited erythromelalgia; iPSCs, induced pluripotent stem cells; BoNT/A, Botulinum toxin type A.

Fig. 2 .
Fig. 2. Absence of tonic block of Nav1.7 by 10 pM BoNT/A in manual patch-clamp HEK293T cells, stably expressing Nav1.7, were exposed to either 10 pM BoNT/A or normal extracellular solution.A, Representative sodium currents before and after substance administration either with vehicle (upper traces) or with Botulinum toxin 10pM (lower traces).B, Relative peak current reduction during low frequency stimulation (0.1 Hz) induced by toxin application.For protocol see inset and methods.Mean ± SD values were 0.65 ± 0.23 for cells with BoNT/A (blue circles, n = 14) and 0.82 ± 0.26 for Control cells (black circles, n = 12).There was no significant deviation (P = 0.0917).Unpaired t-test, two-tailed, normal distribution.C, 10 pM BoNT/A application does not affect voltage dependence of channel activation.Values are given as mean value ± SD.Continuous lines represent conductance voltage curves.The activation curves are shifted to more hyperpolarized potentials in a time dependent manner (time induced shift independent of toxin).Inset: voltage protocol.D, values for half-maximal voltage dependence of activation (V 1/2 ) obtained from Boltzmann fits for individual traces after BoNT/A or vehicle.Mean ± SD values were − 55.89 ± 7.67 mV for cells with BoNT/A (blue circles, n = 12) and − 53.92 ± 10.37 mV for Control cells (black circles, n = 10).P = 0.6154, unpaired t-test, two-tailed, normal distribution.

Fig. 3 .
Fig. 3. Dose response of tonic and use dependent BoNT/A block of Nav1.7 Nav1.7 cells were exposed to BoNT/A concentrations ranging from 100 fM to 1 nM in automated patch-clamp experiments.A, Tonic block: exemplary traces obtained from the 10 th and the 96th sweep during the pulse protocol pre and post treatment of different Botulinum toxin concentrations or vehicle.The pulse protocol is displayed on top.B, Use-dependent block: exemplary traces measured after vehicle or toxin application during the 10 Hz use-dependent protocol.The pulse protocol is displayed on top with 10 pulses applied during 1 s.The first (I 1 ) on the left and the 10th current trace (I 10 ) are magnified for every condition (Scale bar y-axis: 500 pA and x-axis: ms).C, relative current reduction I post /I pre before and after toxin application.Mean ± SD values were 0.73 ± 0.25 for Control (black circles, n = 13), 0.74 ± 0.18 for fM (blue squares, n = 19), 0.67 ± 0.24 for 1 pM (blue triangles up, n = 13), 0.73 ± 0.26 for 10 pM (blue triangles down, n = 22), 0.65 ± 0.23 for 100 pM (purple diamonds, n = 18), 0.82 ± 0.12 for 1 nM (purple circles, n = 6).No significant difference detected, Dunn's multiple comparison test.D, relative current reduction I 10 /I 1 obtained from I max values during the 10 Hz protocol (grey symbols = before toxin, colored symbols = after toxin).The data was analyzed with a two-way ANOVA that reviewed the between factor Toxin (Control; 100 fM BoNT/A; 1 pM BoNT/A; 10 pM BoNT/A; 100 pM BoNT/A; 1 nM BoNT/A) and the within factor Time (pre or post application).None of the different BoNT/A concentrations yielded a significant effect on I 10 /I 1 ratio as confirmed by the nonsignificant interaction term Toxin x Time with P = 0.5516.There was also no significant difference between the levels of the factor Toxin (P = 0.9762).Values can be found in Supplementary Table1 and two-way ANOVA results in Supplementary Table 2.

Fig. 4 .
Fig.4.Intracellular application of Botulinum does not alter Nav1.7 block Nav1.7 cells were exposed to intracellular 10 pM BoNT/A and its catalytic light chain LC/A during automated patch-clamp experiments.A, Tonic block: relative current I post /I pre , reduction obtained from I max values before and after toxin application with the same protocol from Fig.2A.Mean ± SD values were 0.98 ± 0.14 for Control cells (black circles, n = 76), 0.97 ± 0.16 for cells with 10 pM BoNT/A (blue circles, n = 74), 0.97 ± 0.12 for cells with 10 pM LC/ A (green circles, n = 72).No significant difference detected, Dunn's multiple comparison test.B, use dependent block I 10 /I 1 : relative current reduction obtained from I max values during the 10 Hz use dependent protocol (grey circles = before toxin, colored circles = after toxin) with the same protocol from Fig.2C.The data was analyzed with a two-way ANOVA that reviewed the factor Toxin (Group: Control; BoNT/A; LC/A) and Time (pre; post).There was no significant difference between the groups in relation to the time factor (Toxin x Time) with P = 0.2363.There was also no significant effect of the factor Toxin (P = 0.4654), but a significant effect of Time (P= <0.0001), indicating a small but consistent reduction of the I 10 /I 1 ratios across the different experimental conditions.Values can be found in Supplementary Table1 and two-way ANOVA results in Supplementary Table 2. C, voltage dependence of channel activation.Continuous lines represent conductance voltage curves.Toxin application does not affect voltage dependence of channel activation.The activation curves are shifted to more hyperpolarized potential in a time dependent manner (time induced shift).Inset: voltage protocol for activation.D, values for half-maximal voltage dependence of activation (V 1/2 ) obtained from Boltzmann fits for individual traces.Mean ± SD values were − 44.03 ± 9.34 mV for Control cells (black circles, n = 71), − 45.23 ± 9.86 mV for cells treated with BoNT/A 10 pM (blue circles, n = 65), − 44.51 ± 10.37 mV for cells treated with LC/A 10 pM (green circles, n = 70).No significant difference detected, Dunn's multiple comparison test.The used protocol for activation is depicted on the right.E, steady-state fast inactivation of the channel.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the inactivation curve in comparison to the control.F, half-maximal voltage-dependent fast inactivation values obtained from Boltzmann fits of individual traces.Mean ± SD values were − 84.38 ± 5.33 mV for control cells (black circles, n = 71), − 86.65 ± 6.27 mV for cells treated with 10 pM BoNT/A (blue circles, n = 65) and − 86.35 ± 4.94 mV for cells treated with 10 pM LC/A (green circles, n = 71).No significant difference detected, Bonferroni's multiple comparison test, ANOVA results in Supplementary Table2.The used protocol for inactivation is depicted next to the graph.
Fig.4.Intracellular application of Botulinum does not alter Nav1.7 block Nav1.7 cells were exposed to intracellular 10 pM BoNT/A and its catalytic light chain LC/A during automated patch-clamp experiments.A, Tonic block: relative current I post /I pre , reduction obtained from I max values before and after toxin application with the same protocol from Fig.2A.Mean ± SD values were 0.98 ± 0.14 for Control cells (black circles, n = 76), 0.97 ± 0.16 for cells with 10 pM BoNT/A (blue circles, n = 74), 0.97 ± 0.12 for cells with 10 pM LC/ A (green circles, n = 72).No significant difference detected, Dunn's multiple comparison test.B, use dependent block I 10 /I 1 : relative current reduction obtained from I max values during the 10 Hz use dependent protocol (grey circles = before toxin, colored circles = after toxin) with the same protocol from Fig.2C.The data was analyzed with a two-way ANOVA that reviewed the factor Toxin (Group: Control; BoNT/A; LC/A) and Time (pre; post).There was no significant difference between the groups in relation to the time factor (Toxin x Time) with P = 0.2363.There was also no significant effect of the factor Toxin (P = 0.4654), but a significant effect of Time (P= <0.0001), indicating a small but consistent reduction of the I 10 /I 1 ratios across the different experimental conditions.Values can be found in Supplementary Table1 and two-way ANOVA results in Supplementary Table 2. C, voltage dependence of channel activation.Continuous lines represent conductance voltage curves.Toxin application does not affect voltage dependence of channel activation.The activation curves are shifted to more hyperpolarized potential in a time dependent manner (time induced shift).Inset: voltage protocol for activation.D, values for half-maximal voltage dependence of activation (V 1/2 ) obtained from Boltzmann fits for individual traces.Mean ± SD values were − 44.03 ± 9.34 mV for Control cells (black circles, n = 71), − 45.23 ± 9.86 mV for cells treated with BoNT/A 10 pM (blue circles, n = 65), − 44.51 ± 10.37 mV for cells treated with LC/A 10 pM (green circles, n = 70).No significant difference detected, Dunn's multiple comparison test.The used protocol for activation is depicted on the right.E, steady-state fast inactivation of the channel.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the inactivation curve in comparison to the control.F, half-maximal voltage-dependent fast inactivation values obtained from Boltzmann fits of individual traces.Mean ± SD values were − 84.38 ± 5.33 mV for control cells (black circles, n = 71), − 86.65 ± 6.27 mV for cells treated with 10 pM BoNT/A (blue circles, n = 65) and − 86.35 ± 4.94 mV for cells treated with 10 pM LC/A (green circles, n = 71).No significant difference detected, Bonferroni's multiple comparison test, ANOVA results in Supplementary Table2.The used protocol for inactivation is depicted next to the graph.
A.B. Kesdogan et al. (caption on next page)A.B.Kesdogan et al.
Fig. 6.BoNT/A has no effect on endogenous sodium currents in ND7/23 cells ND7/23 cells were exposed to intracellular or extracellular 10 pM BoNT/A and LC/A or vehicle (either intracellular or extracellular solution) in automated patchclamp recordings.A, Relative current reduction I post /I pre obtained from I max values before and after intracellular toxin application.Mean ± SD values were 0.95 ± 0.21 for Control cells (black circles, n = 37), 1.07 ± 0.34 for 10 pM BoNT/A cells (blue circles, n = 44), 0.99 ± 0.17 for 10 pM LC/A cells (green circles, n = 36).B, shows relative current reduction I post /I pre obtained from I max values before and after extracellular toxin application.Mean ± SD values were 1.26 ± 0.37 for Control cells (black circles, n = 38), 1.17 ± 0.37 for 10 pM BoNT/A (blue circles, n = 31), 1.11 ± 0.22 for 10 pM LC/A cells (green circles, n = 42).No significant difference detected, Dunn's multiple comparison test.C, voltage dependence of channel activation is depicted.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the activation curve in comparison to the Control.D, values for half-maximal voltage dependence of activation (V 1/2 ) obtained from Boltzmann fits for individual traces.Mean ± SD values were − 39.70 ± 8.03 mV for Control cells (black circles, n = 36), − 36.15 ± 7.94 mV for cells treated with 10 pM BoNT/A (blue circles, n = 41), − 38.22 ± 8.06 mV for cells treated with 10 pM LC/A (green circles, n = 35).No significant difference detected, Bonferroni's multiple comparison test, ANOVA results in Supplementary Table2.E, steady-state fast inactivation of the channel is depicted.Continuous lines represent conductance voltage curves.Toxin application leads to no shift in the inactivation curve in comparison to the Control.There is a time induced shift in the inactivation curves between the cells before and after substance administration.F, half-maximal voltage-dependent fast inactivation values obtained from Boltzmann fits of individual traces are shown.Mean ± SD values were − 80.52 ± 5.78 mV for Control cells (black circles, n = 36), − 79.25 ± 4.85 mV for cells treated with 10 pM BoNT/A (blue circles, n = 40), − 78.98 ± 5.23 mV for cells treated with 10 pM LC/A.No significant difference detected, Dunn's multiple comparison test.G, tonic block with extracellular application of mexiletine or vehicle (extracellular solution).Relative current reduction I post /I pre obtained from I max values before and after application of either substance.Mean ± SD values were 0.94 ± 0.09 for Control cells with vehicle (black left circles, n = 17), 0.98 ± 0.16 for BoNT/A cells with vehicle (blue left circles, n = 20), 0.96 ± 0.03 for LC/A cells with vehicle (green left circles, n = 17), 0.62 ± 0.10 for Control cells with mexiletine (black right circles, n = 18), 0.66 ± 0.05 for BoNT/A cells with mexiletine (blue right circles, n = 17), 0.66 ± 0.10 for LC/A cells with mexiletine (green right circles, n = 18).The data was analyzed with a two-way ANOVA that included the factor Toxin (Control; BoNT/A; LC/A) and Local Anesthetic (vehicle; mexiletine).There was no significant effect of the Toxin (P = 0.2033).Mexiletine treatment consistently induced a reduction of I post /I pre ratios across all the different toxin conditions as confirmed by the significant effect of the factor Local Anesthetic (P= <0.0001) and the non-significant interaction term (Toxin x Local Anesthetic, P = 0.8775).Two-way ANOVA results can be found in Supplementary Table2.H, relative current reduction I 10 /I 1 obtained from I max values during the 10 Hz use-dependent protocol (grey circles = before mexiletine or vehicle, colored circles = after mexiletine or vehicle).Mexiletine successfully blocks the sodium current in a tonic and use-dependent manner.There we no differences in between the subgroups mexiletine or vehicle treated cells.The data was analyzed with a three-way ANOVA that reviewed the factors Toxin (e.g.Control; BoNT/A; LC/A), Local Anesthetic (vehicle; mexiletine) and Time (pre; post).Application of the different toxins did not lead to significant change of I 10 /I 1 ratios (Toxin x Time, P = 0.2066).As expected, mexiletine treatment consistently induced a reduction of I 10 /I 1 ratios across all the different toxin conditions as confirmed by the significant interaction term Local Anesthetic x Time (P= <0.0001) and non-significant interaction term Toxin x Local anesthetic x Time (P = 0.2170).Values can be found in Supplementary Table1and three-way ANOVA results in Supplementary Table2.
A.B. Kesdogan et al. (caption on next page)A.B.Kesdogan et al.

Fig. 8 .
Fig.8.BoNT/A does not affect iPSC-derived sensory neurons' spontaneous activity Multielectrode-array (MEA) recordings of healthy control and pain patient iPSCderived sensory neurons' spontaneous activity with and without BoNT/A 10 pM.A, Overview of the active electrodes of the MEA chip harbouring iPSC-derived sensory neurons obtained from a healthy control before and after treatment with BoNT/A, respectively.Note that the distribution of active electrodes (yellow color code) closely resembles the neuronal clusters present in the culture as revealed by immunostaining (Suppl.Fig.1).B, Band pass-filtered voltage traces obtained from a representative electrode from the chip containing iPSC-derived sensory neurons from a healthy control before treatment with BoNT/A.Red dots indicate spikes.On the right zoomed-in image, an overlay of the typical spike waveforms can be seen.The waveforms refer exactly to the spikes that can be seen in the zoomed-out image.C, Temporal trajectory of detected spike counts from iPSC derived sensory neurons obtained from a healthy control (upper panels) or a patient with inherited erythromelalgia (IEM, lower panels) before and after treatment with BoNT/A or a vehicle control, respectively.Inspection of the box plots reveals heavily right skewed distributions which is a common finding in MEA experiments and motivated us to use a negative binomial mixed model.D, 95 % confidence intervals of estimated marginal means (means which are predicted by the model for the different level combinations of the factors "cell line" (IEM, healthy control), "pharmacology" (BoNT/A, vehicle) and "condition" (pre, post treatment) while controlling for the day of recording) given on the model (i.e.logarithmic) scale.As our main finding, the difference between the pre and post treatment condition did not differ between the BoNT/A and the vehicle control group.IPSC-derived neurons from the IEM patient (right panels) exhibited a higher spiking activity in total compared to the healthy control (left panels).

Fig. 9 .
Fig. 9. Confirmation of the biological activity, cleavage of SNAP-25 Image of Western Blot with iPSC-derived sensory neurons from healthy control treated with SNAP-25 antibody.IPSC-derived sensory neurons (DIV 40) incubated for 48 h with different concentrations of BoNT/A (3 nM; 10 pM; 1 pM).M = Marker at a molecular weight of 28 kDa depicted; negative control: HEK Nav1.5 cells that do not express SNAP-25 and were not treated (− ) with BoNT/ A; iPSC-derived sensory neurons express SNAP-25 and that BoNT/A cleaves the protein at concentrations as low as 1 pM.
Primary and secondary antibodies used for immunostaining of the iPSC-derived sensory neurons.
Table 1 and three-way ANOVA results in Supplementary Table 2.