Human-Relevant Sensitivity of iPSC-Derived Human Motor Neurons to BoNT/A1 and B1

The application of botulinum neurotoxins (BoNTs) for medical treatments necessitates a potency quantification of these lethal bacterial toxins, resulting in the use of a large number of test animals. Available alternative methods are limited in their relevance, as they are based on rodent cells or neuroblastoma cell lines or applicable for single toxin serotypes only. Here, human motor neurons (MNs), which are the physiological target of BoNTs, were generated from induced pluripotent stem cells (iPSCs) and compared to the neuroblastoma cell line SiMa, which is often used in cell-based assays for BoNT potency determination. In comparison with the mouse bioassay, human MNs exhibit a superior sensitivity to the BoNT serotypes A1 and B1 at levels that are reflective of human sensitivity. SiMa cells were able to detect BoNT/A1, but with much lower sensitivity than human MNs and appear unsuitable to detect any BoNT/B1 activity. The MNs used for these experiments were generated according to three differentiation protocols, which resulted in distinct sensitivity levels. Molecular parameters such as receptor protein concentration and electrical activity of the MNs were analyzed, but are not predictive for BoNT sensitivity. These results show that human MNs from several sources should be considered in BoNT testing and that human MNs are a physiologically relevant model, which could be used to optimize current BoNT potency testing.


Introduction
Botulinum neurotoxins (BoNTs), produced by bacteria of the genus Clostridium, are the most lethal naturally occurring toxins, and yet they are used for medical applications. They exhibit a high selectivity for motor neurons (MNs) at the neuromuscular junction, where they specifically inhibit the release of neurotransmitters, inducing a potentially lethal muscle paralysis. This selectivity is employed when BoNTs are used to treat neuromuscular or autonomic disorders, pain conditions or for cosmetic applications [1]. The potency of a batch needs to be quantified prior to its release either by an ethically questionable mouse bioassay, which is the gold standard, or by alternative methods. However, despite the fact that the use of several alternative methods was enabled by the European pharmacopoeia The aim of this study was to generate human MNs from induced pluripotent stem cells (iPSCs) with different protocols, to evaluate the sensitivity to BoNT/A1 and B1 and to compare it with that of the human neuroblastoma cell line SiMa, which has been used in several in vitro approaches [30][31][32][33]. By quantifying the cleavage of the respective substrates, we were able to show that human MNs can detect much lower activities of BoNT/A1 than the mouse bioassay and SiMa cells. In addition, human MNs could be used to quantify the potency of BoNT/B1, which SiMa cells were not capable of. The BoNT/B1 sensitivity of human MNs corresponded more with human sensitivity data than with the mouse bioassay, which indicates that these cells are a physiologically relevant model. Altogether, human MNs have a high potential to improve current alternative methods and to be used in universal methods that can detect more than one serotype at a time.

Sensitivity of iPSC-Differentiated MNs and SiMa Cells to BoNT/A1 and B1
The application of BoNTs requires prior potency determination, for which only partial replacement of the mouse lethality test has been achieved [2]. In this study, we compare the sensitivity of MNs differentiated according to three different protocols to the sensitivity of the tumor cell line SiMa, which is used in several (proprietary) alternative methods [34]. To this end, SiMa cells were partially differentiated into a more neuronal phenotype and MN populations were generated as described in detail in Schenke et al. [28]. Human iPSCs (IMR90-04) were used for direct differentiation to MNs with protocols based on Du et al. [35] and Maury et al. [36], while the differentiation based on the publication by Kroehne et al. [37] is separated into differentiation of iPSCs to neuronal progenitor cells (NPCs), which can be expanded and differentiated further into MNs. The differentiated cells were treated with several concentrations of BoNT/A1 and B1 for 48 h and then analyzed for the cleavage of the respective BoNT substrate. The fraction of remaining fulllength SNAP25 or VAMP2 after BoNT/A1 or B1 treatment, respectively, was analyzed via Western Blot, normalized to the total protein content per lane, and concentration-response data were modelled by nonlinear regression (Figure 1).
Both SiMa cells and MN populations showed a dose-dependent response to addition of BoNT/A1. The cells differentiated with the protocol based on Kroehne et al. [37] showed substrate cleavage at subpicomolar concentrations, with IC 50 values of 0.0027 pM (77.6% confidence interval [CI]: 0.0004-0.0131 pM), followed by the protocol based on Maury et al. [36] with 0.0337 pM (77.6% CI: 0.0102-0.0818 pM) and the protocol based on Du et al. [35] 0.519 pM (77.6% CI: 0.014-1.979 pM). SNAP25 cleavage in SiMa cells occurred only at significantly higher concentrations, resulting in the IC 50 value of 5.509 pM (77.6 % CI: 3.443-7.790 pM). When BoNT/B1 was used for the intoxication, detection of a dose-response correlation was only possible for MNs. In SiMa cells, the concentration of full-length VAMP2 varied in the analyzed replicates and showed no decrease, even at concentrations as high as 1000 pM BoNT/B1 ( Figure 1A). When only untreated samples of SiMa cells were compared for full-length VAMP2, the concentration differed by a factor of four between the sample with the highest and lowest concentration, respectively ( Figure 1B). Furthermore, a pronounced increase in the concentration of VAMP2 could be found in MNs after treatment with lower concentrations of BoNT/B1. In a small sample of blots probed with the antibodies for both substrate proteins, an increase in VAMP2 could also be seen when MNs generated according to the protocols based on Kroehne et al. [37] and Du et al. [35] were treated with BoNT/A1 ( Figure A1, Appendix A). In MNs, higher concentrations of BoNT/B1 were necessary for the cleavage of VAMP2 compared with the cleavage of SNAP25 by BoNT/A1. IC 50 values for BoNT/B1 treatment of cells differentiated with the protocols based on Maury et al. [36] were at 0.687 pM (77.6% CI: 0.319-1.560 pM), at 1.039 pM (77.6% CI: 0.766-3.314 pM) for the protocol based on Du et al. [35], which were significantly lower than the IC 50 value of 14.84 pM (77.6% CI: 7.54-33.09 pM) for the protocol based on Kroehne et al. [37]. No IC 50 value could be determined for SiMa cells treated with BoNT/B1. cation by Kroehne et al. [37] is separated into differentiation of iPSCs to neuronal progenitor cells (NPCs), which can be expanded and differentiated further into MNs. The differentiated cells were treated with several concentrations of BoNT/A1 and B1 for 48 h and then analyzed for the cleavage of the respective BoNT substrate. The fraction of remaining full-length SNAP25 or VAMP2 after BoNT/A1 or B1 treatment, respectively, was analyzed via Western Blot, normalized to the total protein content per lane, and concentration-response data were modelled by nonlinear regression (Figure 1). From the IC 50 values in pM, the correspondent mouse LD 50 /mL were calculated using mouse toxicity data generated by the BoNT manufacturer ( Figure 2). As different batches of toxins were used, only the corresponding mouse LD 50 units can be used for direct comparison of the protocols. Interestingly, cells generated by the protocol based on Maury et al. [36] were the most sensitive cell type for both BoNT/A1 and B1, with a sensitivity equating 0.05 and 4.12 mouse LD 50 /per mL. The cells differentiated according to the protocol based on Kroehne et al. [37] were only highly sensitive to BoNT/A1 with 0.06 mouse LD 50 /mL, while cells differentiated with the protocol based on Du et al. [35] showed intermediate sensitivity to BoNT/A1 and B1 with 10.90 and 6.24 mouse LD 50 /mL, respectively. SiMa cells were between 10 to 2000 times less sensitive to BoNT/A1 than the analyzed MN populations.

Molecular and Functional Differences between the Analyzed Cell Types
Large sensitivity differences among the MN populations generated with different protocols were detected regarding the BoNT serotypes A1 and B1. A factor that contributes to a high toxin affinity to MNs is the expression of corresponding protein receptors in high affinity isoforms. SiMa and MN populations generated with the differentiation protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] were analyzed for their mRNA expression levels of proteins relevant for BoNT toxicity in Schenke et al. [28]. In this study, to account for the different protein turnover rates that might be encountered in mature neurons, receptors for BoNT/A1 and B1 were analyzed on the protein level by Western blot in MN populations, SiMa cells and compared with mouse brain synaptosomes (mSYSOs), which were used as a control ( Figure 3). . Data provided from the manufacturer were used to estimate mouse LD50/mL, with details regarding the batches used given in Section 5.6. Mean IC50 values or mouse LD50/ml are shown with their respective 77.6% confidence intervals. To obtain the significance level of 5% for hypothesis tests α = √0.05 = 0.224 was chosen, and 77.6% CIs considered. Non-overlapping CIs show significantly different IC50 values indicated by *.

Molecular and Functional Differences between the Analyzed Cell Types
Large sensitivity differences among the MN populations generated with different protocols were detected regarding the BoNT serotypes A1 and B1. A factor that contributes to a high toxin affinity to MNs is the expression of corresponding protein receptors in high affinity isoforms. SiMa and MN populations generated with the differentiation protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] were analyzed for their mRNA expression levels of proteins relevant for BoNT toxicity in Schenke et al. [28]. In this study, to account for the different protein turnover rates that might be encountered in mature neurons, receptors for BoNT/A1 and B1 were analyzed on the protein level by Western blot in MN populations, SiMa cells and compared with mouse brain synaptosomes (mSYSOs), which were used as a control ( Figure 3).  [37] and Maury et al. [36] as well as SiMa cells were detected via Western blot using isoform-specific antibodies for detection. Each panel depicts the   [37] and Maury et al. [36] as well as SiMa cells (all with n = 3) to BoNT/A1 and B1 in IC50 [pM] (A, C) and mouse LD50/mL (B, D). Data provided from the manufacturer were used to estimate mouse LD50/mL, with details regarding the batches used given in Section 5.6. Mean IC50 values or mouse LD50/ml are shown with their respective 77.6% confidence intervals. To obtain the significance level of 5% for hypothesis tests α = √0.05 = 0.224 was chosen, and 77.6% CIs considered. Non-overlapping CIs show significantly different IC50 values indicated by *.

Molecular and Functional Differences between the Analyzed Cell Types
Large sensitivity differences among the MN populations generated with different protocols were detected regarding the BoNT serotypes A1 and B1. A factor that contributes to a high toxin affinity to MNs is the expression of corresponding protein receptors in high affinity isoforms. SiMa and MN populations generated with the differentiation protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] were analyzed for their mRNA expression levels of proteins relevant for BoNT toxicity in Schenke et al. [28]. In this study, to account for the different protein turnover rates that might be encountered in mature neurons, receptors for BoNT/A1 and B1 were analyzed on the protein level by Western blot in MN populations, SiMa cells and compared with mouse brain synaptosomes (mSYSOs), which were used as a control ( Figure 3).   [37] and Maury et al. [36] as well as SiMa cells were detected via Western blot using isoform-specific antibodies for detection. Each panel depicts the detection with one of the antibodies. A total of 10 µg of lysate pooled from three independent differentiations was analyzed and compared with mouse brain synaptosomes (mSYSOs). For SV2A and SV2B blots, the signal intensity from mSYSOs was far greater than for the other samples, and images with different exposure times were combined (Exposure time of cell lysates: SV2A 1 min, SV2B 10 min and for mSYSOs 2 s and 10 s, respectively).
All protein receptor isoforms except SV2B could be detected in MN populations and in SiMa cells, albeit at different concentrations. SV2A and SV2B were found in much higher concentrations in mSYSOs and were recorded with different exposure times. For SV2A, the highest concentration was detected in lysates from MNs generated based on the protocol by Du et al. [35], while the lowest concentration was found in MNs generated according to the protocol by Maury et al. [36]. The isoform SV2B could not be detected neither in MNs Toxins 2021, 13, 585 6 of 16 nor in SiMa cells. SV2C, however, which is found in high concentration in MNs, could be detected in higher concentration in the analyzed MN populations, but hardly in SiMa cells. The low-affinity receptor for BoNT/B, SYT1, was expressed in comparable concentrations in all analyzed lysates, while SYT2 was found only in MN and hardly in SiMa cells. For SYT1 and SYT2, the BoNT/B1 binding site was sequenced in IMR90, NPCs and SiMa cells and all sequences are consistent with the human genome (data not shown, alignment with RefSeq sequences NC_000012 [SYT1] and NG_041776 [SYT2]).
As the uptake of clostridial neurotoxins is dependent on the endocytosis, which mainly occurs during action potential-dependent exocytosis, we analyzed the electrical activity of the cultivated MNs on high-density microelectrode arrays (HD-MEAs). Progenitor cells from the protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] were seeded on HD-MEAs at the latest time point prior to neurite outgrowth, which was at the stage of immature neurons for the protocol based on Du et al. [35] and at the MN progenitor stage for the other two protocols, and then differentiated further. The differentiations of the cells generated with the protocol based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] were finalized after 10, 15 and 21 days (days in vitro, DIV) on the HD-MEAs, respectively. The neuronal activity was recorded weekly after 7, 14 and 21 days on HD-MEAs, for which the entire chip (26,400 electrodes) was scanned for spiking activity ( Figure 4).
All protein receptor isoforms except SV2B could be detected in MN populations and in SiMa cells, albeit at different concentrations. SV2A and SV2B were found in much higher concentrations in mSYSOs and were recorded with different exposure times. For SV2A, the highest concentration was detected in lysates from MNs generated based on the protocol by Du et al. [35], while the lowest concentration was found in MNs generated according to the protocol by Maury et al. [36]. The isoform SV2B could not be detected neither in MNs nor in SiMa cells. SV2C, however, which is found in high concentration in MNs, could be detected in higher concentration in the analyzed MN populations, but hardly in SiMa cells. The low-affinity receptor for BoNT/B, SYT1, was expressed in comparable concentrations in all analyzed lysates, while SYT2 was found only in MN and hardly in SiMa cells. For SYT1 and SYT2, the BoNT/B1 binding site was sequenced in IMR90, NPCs and SiMa cells and all sequences are consistent with the human genome (data not shown, alignment with RefSeq sequences NC_000012 [SYT1] and NG_041776 [SYT2]).
As the uptake of clostridial neurotoxins is dependent on the endocytosis, which mainly occurs during action potential-dependent exocytosis, we analyzed the electrical activity of the cultivated MNs on high-density microelectrode arrays (HD-MEAs). Progenitor cells from the protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] were seeded on HD-MEAs at the latest time point prior to neurite outgrowth, which was at the stage of immature neurons for the protocol based on Du et al. [35] and at the MN progenitor stage for the other two protocols, and then differentiated further. The differentiations of the cells generated with the protocol based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] were finalized after 10, 15 and 21 days (days in vitro, DIV) on the HD-MEAs, respectively. The neuronal activity was recorded weekly after 7, 14 and 21 days on HD-MEAs, for which the entire chip (26,400 electrodes) was scanned for spiking activity ( Figure 4). The share of active electrodes was low (~1%) but mirrored the differentiation of the MN populations on the HD-MEAs, with the highest activity for each of the protocols being recorded close to the finalization of the differentiation. The mean spike amplitudes did not increase during the cultivation time and were below 40 µV, indicating immature neurons. Due to the low number of active electrodes, effects of BoNT intoxication could not be studied. The share of active electrodes was low (~1%) but mirrored the differentiation of the MN populations on the HD-MEAs, with the highest activity for each of the protocols being recorded close to the finalization of the differentiation. The mean spike amplitudes did not increase during the cultivation time and were below 40 µV, indicating immature neurons. Due to the low number of active electrodes, effects of BoNT intoxication could not be studied.

Sensitivity of MNs to BoNT/A1 and B1 Is Superior to SiMa Cells
For potency determination of BoNT-based medicines only partial replacement of animal tests has been achieved [31,32,38,39]. Among potential test systems, human motor neurons (MNs) are the most physiologically relevant, being the cell type specifically targeted by BoNTs, and are able to recapitulate the major steps of toxin action [40]. In this study, MNs were generated from human iPSCs on the basis of three previously published differentiation protocols by Du et al. [35], Kroehne et al. [37] and Maury et al. [36]. MN populations generated with these protocols have been examined for their cellular composition, the presence of relevant ganglioside and protein receptors as well as of the SNARE substrates of all human relevant BoNT serotypes [28]. Here, the differentiated MNs as well as SiMa cells were used for treatment with BoNT/A1 and B1 and activity assessment via  (Figure 1). For both serotypes, a dose-dependent decrease in full-length substrates could be seen in all MN populations. Initially, full-length VAMP2 increased in relation to the total protein loaded, after treatment with low concentrations of BoNT/B1. This is likely part of an adaptive mechanism, which might be explained by a generally reduced protein turnover, induced by the blockage of signal transmission or a specific upregulation of VAMP2, which has been described under certain circumstances, e.g., electric shock or chronic stress [41][42][43]. In SiMa cells, no dose-dependent decrease in VAMP2 was found even at high BoNT/B1 concentrations, but instead, the VAMP2 concentration showed high variation even in untreated samples. In other studies using anti-VAMP2 antibodies from other manufacturers, no VAMP2 was detected in SiMa cells [33,44]. Therefore, SiMa cells might be unsuited to assess VAMP2 cleavage by BoNT/B. From the dose-response data of MNs and SiMa cells treated with BoNT/A1 and MNs treated with BoNT/B1, different models of non-linear regression were used to calculate the IC 50 values. As different batches of toxins were used in this study, these concentrations were translated into mouse LD 50 /mL by using mouse lethality data from the manufacturer. A reference volume of 1 mL was chosen, as a volume of 0.5 mL is usually injected during the mouse lethality assay, which results in a sensitivity of 2 mouse LD 50 /mL [25,45]. The sensitivity to BoNT/A1 was not only significantly higher (up to 2000 times) in the MN populations than in SiMa cells, but it was also at least 30 times higher than the mouse sensitivity. For MNs from commercial sources, a BoNT/A sensitivity with an IC 50 of 0.39 pM was reported by Duchesne De Lamotte et al. [27] and of 0.12 mouse LD 50 /mL by Pellett et al. [26]. The MNs generated here according to the protocols by Kroehne et al. [37] and Maury et al. [36], which had yielded 14% and 16% Islet1-positive MNs in a previous study, showed sensitivities to BoNT/A1 of 0.0027 and 0.0337 pM, which correlates to 0.06 and 0.05 mouse LD 50 /mL. One of the major differences is the larger proportion of MNs in the commercial cells used by Duchesne De Lamotte et al. [27] (approximately 82%) and by Pellett et al. [26] (approximately 87%), but it seems worth investigating different neuronal populations, which might be more sensitive than pure MN populations.
For MNs generated based on the protocol by Du et al. [35], we have reported a sensitivity of 0.96 mouse LD 50 /mL from an initial experiment, which is lower than the results in this study by a factor of ten [28]. The same batch of toxin was used for both studies, which was stored at −80 • C as single use aliquots. However, a slightly adapted method for protein normalization was used based on whole protein concentration instead of using only actin for normalization, in addition to the use of a different software for quantification as well as a different model of nonlinear regression, which reduces the comparability. In our previous study cells were differentiated starting from the same progenitor batch, while here we started from iPSCs wanting to investigate the reproducibility after independent differentiations, which resulted in a larger confidence interval [28]. When establishing an assay based on MNs, it would be favorable to scale up batch production of progenitors for cryopreservation to maximize reproducibility.
For BoNT/B1, we detected sensitivities in MNs that were lower by a factor of 2 (Maury et al. [36] protocol) to 90 (Kroehne et al. [37] protocol) in comparison with the mouse lethality data, which has an LD 50 of 2 units/mL. Considering the well-described interspecies sensitivity differences humans and mice towards BoNT/B by a factor of 30-100 due to the mutation of SYT2 in humans and chimpanzees, the lower sensitivity of human MNs reflects human sensitivity [12,15,16]. For the MN populations based on the protocols by Maury et al. [36] and Du et al. [35], we found a sensitivity of 4 and 6 mouse LD 50 /mL, which correlates well with 2 mouse LD 50 /mL reported by Pellett et al. [26]. However, according to the Western blot data, the sensitivity for single BoNT serotypes differs between the MN populations used in this study with the cells differentiated with the protocol based on Du et al. [35] being less sensitive to BoNT/A1, and with the cells differentiated with the protocol based on Kroehne et al. [37] showing low sensitivity to BoNT/B1. In Schenke et al. [28], we have reported a MN yield (Islet-positive cells) of 51%, 16% and 14% for the protocols based on Du et al. [35], Maury et al. [36] and Kroehne et al. [37], respectively.
Only the cells differentiated according to Maury et al. [36] detected low concentrations of both BoNT/A1 and B1 (0.05 and 4.12 mouse LD 50 /mL), even though the percentage of MNs was lower than in the cells differentiated with the protocol based on Du et al. [35], indicating that MN percentage is not the only factor determining the sensitivity.

Factors Contributing to MN Sensitivity
Among the factors that determine the BoNT sensitivity of a cell, which might be synergistic, are the expression and distribution of isoforms of substrates, protein and ganglioside receptors, glycosylation of receptors, cell maturity and synaptic activity [18,26,46,47]. Factors such as the glycosylation of receptors have been discussed as a reason for individual differences in patients treated with BoNTs [1,18]. The gene expression of protein receptors and substrates were previously assessed in Schenke et al. [28] and could be detected in MN populations generated with each of the protocols. As the protein turnover in mature neurons might be decreased, the concentration of the BoNT receptors was assessed via Western blot (Figure 3). However, the protein receptor concentrations largely mirror the gene expression data in Schenke et al. [28] with deviations with respect to SV2B, which is 10-fold less expressed than SV2A in all MNs and SiMA cells, but was not detected in any MNs or SiMa cells by Western blot. Additionally, SV2C and SYT2 are hardly detectable on the protein level in SiMa cells, whereas RT-qPCR yielded comparable expression level between MNs and SiMa cells for SYT2 and only slightly smaller SV2C expression in SiMa cells compared with MNs. In comparison with mSYSOs, it appears that the concentration of SYT2, which is the dominant isoform in MNs, was lower than the concentration of SYT1 in the analyzed MNs [48]. Since the share of MNs in these populations is only 14-51% on average, other cell types with other isoform expression patterns can have affected the average protein levels.
Both the gene expression data and the receptor protein concentration cannot explain the sensitivity differences alone, as, e.g., the cells differentiated according to the protocol based on Kroehne et al. [37] exhibit a relatively high concentration of both SYT1 and SYT2, but low BoNT/B1 sensitivity ( Figure 2). Additionally, the cells used for differentiation (IMR90-04 and NPCs) as well as SiMa cells showed full accordance in the DNA sequence encoding the BoNT/B binding site in SYT1 and SYT2 of the human genome. Since BoNT/B has stronger autonomic side effects than BoNT/A, it is expected to enter other subsets of neurons, which might be a reason for serotype differences seen in this study [16].
As another parameter to affect BoNT sensitivity, the neuronal maturation and synaptic activity of MN populations were studied with MEAs. For this purpose, MN progenitor cells, not yet showing neurite outgrowth, from the different protocols were seeded and differentiated further on HD-MEA chips. Therefore, the cultivation time on the MEAs is different from the differentiation day. Cells from the protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] reached maturation after 10, 15 and 21 days on the MEAs, respectively. In the data recorded here, neuronal activity was low and did not increase after finalization of the differentiation on the day set by the authors of the original differentiation protocols. In addition, the low spike amplitudes resemble rather immature neurons [49]. Due to this, the activity of MNs needs to be improved prior to further experiments such as BoNT potency evaluation utilizing MEAs. The low neuronal activity, especially in the MN populations generated with the protocols by Kroehne et al. [37] and Du et al. [35], might be due to experimental parameters such as evaporation, which was substantial in MEA wells. The activity might be optimized by adapting the cultivation parameters by, e.g., using different cell culture media or co-culture of MNs with astrocytes or Schwann cells. Co-culture with astrocytes has been shown to shorten the time needed until functional maturity is reached and to improve the neuronal activity [50]. A slightly greater functional maturity from the cells generated with the protocol based on Maury et al. [36] compared with the other protocols is apparent from the recorded neuronal activity and may have contributed to the superior sensitivity of these cells. Optimizing the neuronal Toxins 2021, 13, 585 9 of 16 activity might further increase the sensitivity of MNs to different BoNTs and while studying BoNT intoxication of MNs on MEAs is challenging, it remains a promising system.
The factors not analyzed in this study, such as the intra-and intercellular distribution of the relevant proteins, their glycosylation and the presence of gangliosides, might have played a role in the apparent sensitivity of the MN populations. As neuronal differentiations in vitro can yield heterogeneous cell compositions, these factors may differ depending on the exact composition. For instance, the binding affinity of BoNT/B to SYT2 is largely enhanced by the presence of gangliosides [40,[51][52][53]. Therefore, cells with, e.g., a high expression of SYT1/2, which lack gangliosides, will have a low sensitivity to BoNT/B, DC and the G. A method suited for analyzing single cells for several parameters, such as the expression of multiple proteins, which is flow cytometry. This method is rarely used for neurons, and with the cells generated in this study, the strong interconnectivity of MNs with their particularly long axons made the generation of a viable single cell suspension virtually impossible. Ultimately, however, it will be necessary to look at the endpoint of BoNT toxicity, the inhibition of neurotransmitter release. The multitude of factors that contribute to the sensitivity of a given cell, e.g., the distribution of BoNT substrates within a cell, could also affect substrate cleavage and the inhibition of exocytosis [47]. Therefore, to have a BoNT in vitro assay, that can faithfully represent human sensitivity, either in already available or new pharmacological preparations from different BoNT serotypes or subtypes, it should be based on quantification of neurotransmission inhibition.
The data shown here aim to emphasize that human MNs are suitable to be used in further optimization of alternative methods for BoNT potency determination, likely in combination with MEAs or reporters of neurotransmitter release [54,55]. Since the sensitivity of a cell population could not be predicted with single parameters alone, it appears necessary to further investigate the potential of (motor) neurons from different sources. Nevertheless, the MNs generated in this study, depending on the protocol used, were able to quantify pico-to subpicomolar concentrations of both BoNT/A1 and B1, were significantly more sensitive than the tumor cell line SiMa for BoNT/A1 and could reflect human sensitivity, all of which makes human MNs a potential foundation for serotype-independent BoNT potency determination.

Differentiation Based on Du et al.
For the MN differentiation based on the protocol by Du et al. [35], IMR90 were seeded on Matrigel-coated plates at 5 × 10 4 cells per mL in StemMACS™ iPS-Brew XF medium with 10 µM Y-27632. The following day, neural medium was used, which was then exchanged every other day. For neuronal induction, 3 µM CHIR, 2 µM DMH1 (Bertin Pharma, Montigny le Bretonneux, France) and 2 µM SB were used from day 1 to day 6. Cells were passaged and cultivated in neural medium with 0.1 µM RA, 0.5 µM PMA, 1 µM CHIR, 2 µM DMH1 and 2 µM SB for six more days. Motor neuron progenitors were detached with Collagenase and transferred in neural medium with 0.5 µM RA and 0.1 µM PMA to low attachment plates (Corning), where they formed neurospheres. After 6 days, the spheres were dissociated, seeded on Matrigel coated plates and neurons maturated until day 28 in neural medium containing 0.5 µM RA, 0.1 µM PMA, 0.1 µM compound E (CE, Bertin Pharma), 2 ng/mL of GDNF, BDNF and CNTF (Peprotech).

Differentiation Based on Maury et al.
The MN differentiation based on the publication by Maury et al. [36] was conducted as follows: IMR90 were seeded as small aggregates with a density of 1 × 10 5 cells/mL in low-attachment plates in neural medium containing 0.5 µM AA, 3 µM CHIR, 2 µM DMH1, 2 µM SB and 5 µM Y-27632 and cultivated in suspension. Totals of 0.1 µM RA and 0.5 µM SAG were added and Y-27632 was withdrawn on day 2. On day 4, only AA, RA and SAG were added until motor neuron progenitors were obtained on day 9, where neurospheres were dissociated mechanically and enzymatically and seeded on Matrigel-coated plates. From day 9 to 14, 10 µM DAPT and from day 11 to 32, GDNF, BDNF and CNTF (5 ng/mL) and 1 µg/mL dbcAMP were added.

Cultivation and Differentiation of SiMa Cells
The neuroblastoma cell line SiMa (ACC 164), which was kindly provided by G. Püschel, was cultivated in RPMI1640 (Biochrom, Cambridge, UK) with 10% inactivated FCS (Biochrom), 1% P/S and 1% L-glutamine. SiMa cells were differentiated into a more neuronal phenotype by replacing the serum with 2% MACS ® NeuroBrew ® -21, 1% N-2 Supplement and 1 mM non-essential amino acids. SIMA were differentiated on poly-L-lysine (Sigma)-coated plates for two days before BoNTs were added.

Analysis of SNAP25 and VAMP2 Cleavage
Cells differentiated and maturated according to the protocols based on Kroehne et al. [37], Du et al. [35] and Maury et al. [36] were treated with purified BoNT/A1 or B1 (Miprolab, Göttingen, Germany; #3101 and #3201) diluted in neural medium. BoNT/A1 from one batch (2.8 × 10 8 minimal lethal doses/mg) was used for treatment of all cells except for the cells differentiated according to Maury et al. [36], for which another batch (2 × 10 7 minimal lethal doses/mg) was used. BoNT/B1 from one batch (1.6 × 10 8 minimal lethal doses/mg) was used for treatment of all cells except for the cells differentiated according to Du et al. [35] and Maury et al. [36], for which another batch (8 × 10 7 minimal lethal doses/mg) was used. For the treatment of SiMa cells, toxins were diluted in SiMa differentiation medium, while neural medium without supplements was used for treatment of MNs. After 48 h, cells were washed with PBS, harvested in RIPA buffer (50 mM Tris, 150 mM NaCl, 1 mM EDTA, 1% Triton X100, 1% sodium deoxycholate, 0.1% SDS, 1% Protease Inhibitor Cocktail Set III) and lysed with ultrasonication. The cleared lysate (25 µg SiMa Lysate, 10 µg MN lysate) was separated with Any kD™ Mini-PROTEAN ® TGX™ gels (Bio-Rad, Feldkirchen, Germany) and transferred onto nitrocellulose membranes (GE Healthcare, Freiburg, Germany). Membranes were blocked with 5% milk powder in TBST for 1 h. Anti-SNAP25 and anti VAMP2 primary antibodies (Table A1) were diluted 1:1000 in 5% milk powder in TBST and incubated over night at 4 • C. Horseradish peroxidase (HRP) coupled secondary antibody (Sigma , Table A2) was diluted 1:20,000 and incubated for 1 h at room temperature. The antibodies directed against SNAP25 and VAMP2 recognize only the respective full-length proteins as they bind to the carboxy terminus, which is cleaved off by BoNT/A1 and B1. Bands were visualized with Immobilon Western HRP Substrate (Merck/Millipore, Burlington, MA, USA) and quantified with the ChemiDoc imaging system (Bio-Rad). Afterwards, the membrane was stained with India Ink (Pelikan #4001) with 2% acetic acid for 30 min, washed with H 2 O, quantified with colorimetric detection and used to normalize the signal intensities to the total protein loaded per band. Toxin-treated samples were normalized to the respective untreated control. Concentrationresponse data were modelled by nonlinear regression as described in Seeger et al. [57] in RStudio (RStudio Team 2021, PBC, Boston, MA, USA, http://www.rstudio.com/, accessed on 19 August 2021). The used regression models were chosen according to the Akaike Information Criterion (AIC) from probit, logit, Weibull, Aranda-Ordaz, generalized logit I and II. Confidence intervals (CIs) were estimated via 1000 bootstrap simulations. IC 50 values were computed as the point where the corresponding curve crosses the midpoint of range (codamain) of the corresponding approximation function. Since standard hypothesis tests cannot be applied directly to check whether the difference in IC 50 values might be explainable by random effects, CIs were used. The probability for (1-α) CIs to overlap under the null hypothesis that the data originate from the same distribution is α 2 . To obtain the usual significance level of 5% for hypothesis tests, α = √ 0.05 = 0.224 needs to be chosen, i.e., 77.6% CIs were considered. Non-overlapping CIs indicate significantly different IC 50 values.

Analysis of BoNT Receptors
Mouse brain synaptosomes were generated as detailed by Rummel et al. [58]. Briefly, functional synaptosomes were recovered from a Percoll gradient after homogenisation of cerebrum obtained from 6-10 mouse brains and various centrifugation steps, and finally diluted in physiological buffer (140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 20 mM HEPES, 10 mM glucose, 0.5% BSA, pH 7.4) with the final synaptosomal protein adjusted to a concentration of 10 mg/mL. Lysates of MNs and SiMa cells not treated with BoNTs were pooled from three independent differentiations. Totals of 10 µg protein each were additionally separated on a 12.5% gel via SDS-PAGE and transferred onto PVDF membranes (Roti ® -PVDF, pore size 0.45 µm, ROTH, Karlsruhe, Germany). Membranes were blocked with 5% milk powder in PBST (PBS, 0.05% Tween) for 1 h. Anti-SV2A, anti-SV2B, anti-SV2C, anti-SYT1 and anti-SYT2 primary antibodies (Table A1) were diluted in 5% milk powder in PBST and incubated over night at 4 • C. HRP-coupled secondary antibodies (both from Rockland, Limerick, PA, USA, Table A2) were diluted 1:20,000 and incubated for 4 h at room temperature. Bands were visualized with SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific) with the ChemoStar ECL Imager (Intas, Göttingen, Germany).

Microelectrode Arrays
For recording of electrical activity of MNs, MN progenitors from the protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] were each seeded on six MaxOne high-density (HD) MEAs (MaxWell Biosystems, Zurich, Switzerland). MaxOne HD-MEAs consist of a single well with 26,400 electrodes, of which 1024 can be recorded simultaneously on a sensor area of 3.85 × 2.10 mm 2 , with a center-to-center electrode distance of 17.5 µm, which facilitates subcellular resolution. To enhance the attachment of the neurons to the surface, HD-MEAs were prepared as recommended by the manufacturer: At first, HD-MEAs were pretreated with 1% Terg-a-zyme ® enzyme detergent (Sigma) over night at room temperature. On the next day, HD-MEAs were thoroughly rinsed with H 2 O, sterilized with 70% Ethanol in H 2 O, dried, placed in 100 mm Petri dishes and coated with 50 µL of 0.1 mg/mL poly-D-lysine (Thermo Fisher Scientific) for 1 h at 37 • C. HD-MEAs were rinsed three times with H 2 O, allowed to dry for 30 min and then coated with 50 µL of Geltrex™ hESC-Qualified, Ready-To-Use, Reduced Growth Factor Basement Membrane Matrix (Thermo Fisher Scientific). 30 mm dishes filled with H 2 O were placed next to the HD-MEAs to reduce evaporation. After incubation for 1 h at 37 • C, Geltrex was removed, and MN progenitors or immature neurons were seeded on the HD-MEAs in the respective differentiation medium. As the differentiation protocols differed with regard to the timing of subcultivation and choice of suspension or adherent culture, the latest time point where cells were subcultivated prior to neurite formation was chosen for plating of HD-MEAs. Additionally, the number of cells seeded was adjusted to ensure similar cell density on the MEAs and to compensate for cell death after thawing. For the protocol based on Du et al. [35], 150,000 immature neurons frozen on differentiation day 18, were seeded onto the sensor and surrounding area in a volume of 50 µL. For the protocols based on Kroehne et al. [37] and Maury et al. [36], 150,000 and 450,000 MN progenitors frozen on differentiation day 6 and day 9, respectively, were seeded onto the sensor and surrounding area in a volume of 50 µL. After cells were allowed to attach for 1 h in the incubator, 600 µL of respective differentiation medium was added. In parallel to the MEAs, cells were seeded on similarly coated cover slips at the same density in relation to the surface area, in a volume of 500 µL. A total of 50% of the cultivation medium was exchanged as required by the respective differentiation protocol until the differentiation was finalized, after which at least 50% of the medium was exchanged twice a week. For the differentiation based on Maury et al. [36], 10 µM DAPT was supplemented during the whole cultivation time to prevent overgrowth of neuronal progenitors, which was only added until differentiation day 14 in the other experiments with this protocol. Electrical activity and network properties of neurons on HD-MEAs were assessed weekly with a MaxOne recording unit (MaxWell Biosystems), which was placed inside the incubator, starting after seven days of cultivation on the chips. To distinguish from the differentiation day, days in vitro (DIV) is used to describe the cultivation time on the MEAs. Spiking activity on the whole sensor area was assessed with the "activity scan" module of the MaxLab Live software (MaxWell Biosystems) in sequential configurations, each of which was recorded for 30 s. Recorded metrics are the spike amplitude (90th percentile of the negative amplitude of detected spikes in µV) and the active area (electrodes with firing rate > 0.1 Hz and spike amplitude > 20 µV). Funding: This research was funded by grants from the German Federal Ministry of Education and Research, 031L0132B to B.S. and 031L0111B to A.R. This publication was supported by Deutsche Forschungsgemeinschaft and University of Veterinary Medicine Hannover, Foundation within the funding programme Open Access Publishing. Figure A1. Concentration of SNAP25 and VAMP2 after treatment with BoNT/A1. Western Blots of motor neurons (MNs) generated with the protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] treated with different concentrations of BoNT/A1 with detection of full-length substrate protein (n = 1), normalized to total protein loaded. Each panel represents one differentiation protocol. Mouse S12920, BD Transduction Laboratories #612717 1:1000 Figure A1. Concentration of SNAP25 and VAMP2 after treatment with BoNT/A1. Western Blots of motor neurons (MNs) generated with the protocols based on Du et al. [35], Kroehne et al. [37] and Maury et al. [36] treated with different concentrations of BoNT/A1 with detection of full-length substrate protein (n = 1), normalized to total protein loaded. Each panel represents one differentiation protocol.