Split luciferase-based assay to detect botulinum neurotoxins using human motor neurons derived from induced pluripotent stem cell.


 Botulinum neurotoxins (BoNTs) have been widely used clinically as a muscle relaxant. These toxins target motor neurons and cleave proteins essential for neurotransmitter release like Synaptosomal-associated protein of 25 kDa (SNAP-25). Most in vitro assays for BoNT testing use rodent cells or immortalized cell lines, which showed limitations in accuracy and physiological relevance. Here, we report a cell-based assay for detecting SNAP25-cleaving BoNTs by combining human induced Pluripotent Stem Cells (hiPSC)-derived motor neurons and a luminescent detection system based on split nanoluc luciferase. This assay is convenient, rapid, free-of-specialized antibodies, and can discriminate the potency of different BoNTs, with a detection sensitivity of femtomolar concentrations of toxin and can be used to study the different steps of BoNT intoxication. Abreviations: BoNT, Botulinum neurotoxin, SNAP-25, Synaptosomal-associated protein of 25 kDa, hiPSC, human induced Pluripotent Stem Cells, SNARE, soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptor, SV2, synaptic vesicle proteins, MLB, mouse lethality bioassay, LD50, toxin’s dose lethal for half of the animal injected, CB-assay, cell-based assays, FRET, Förster resonance energy transfer, Concanamycin A, EC50, Half maximal effective concentration, MNs, motor neurons.


Introduction
Botulinum neurotoxins (BoNTs) are a family of bacterial toxins produced by the bacteria Clostridium botulinum. Seven serotypes (named A to G) can be distinguished based on their immunological differences [1] and further divided in subtypes based on their amino acid sequence differences [2].
LC is translocated into the cytoplasm and the endopeptidase domain of the LC interacts and cleaves its substrate.
Because BoNTs have a biological origin and are highly toxic, the safe and effective dosing of pharmaceuticals BoNT requires accurate and reliable test. For years, the gold standard assay to test BoNT pharmaceuticals has been the mouse lethality bioassay (MLB) [12]. This assay measures the toxin's dose lethal for half of the animal injected (LD 50 ). This method causes severe suffering of laboratory animals and therefore doesn't follow the principle of the 3Rs (reduction, replacement and re nement of animal use in research). Therefore, the development of alternative methods is socially and ethically required.
Several in vitro cell-based assays (CB-assay) have been developed for BoNT potency. The main advantage of these CB-assays is that they can provide information on all essential steps of BoNT intoxication. Several methods were approved by the FDA and other regulatory agencies, to test the potency of pharmaceutical BoNT/A [13] [14] [15]. All these assays monitor the cleavage of BoNT substrate.
The rst replacement assay was developed by Fernandez et al. and relies on a monoclonal antibody that recognizes only the cleaved form of the synaptosomal-associated protein of 25 kDa (SNAP-25), a component of the neurotransmitter release apparatus. It was developed in human SiMa cell line in an ELISA-based format [13]. Another test developed by Dong  enabling FRET are linked together with a SNARE protein fragment. The proteolytic activity of BoNT abolishes energy transfer between the uorescent proteins [15].

Rodent cells and immortalized cell lines have limitations concerning accuracy and physiological
relevance to the responses observed in humans. Human induced Pluripotent Stem Cells (hiPSC)-derived neurons could serve as a useful cell model to improve these assays . Several studies have shown that these cells are a highly sensitive model to test BoNT, especially hiPSC derived motor neurons (MNs) that represent the most physiological system to study BoNT [16] [17] [18] [19].
In this work we describe the development of a new cell-based assay combining human iPSC-derived MNs with an antibody-free detection system uses the engineered split-Nanoluc luciferase reporter (NanoBiT), [20] [21] to detect toxin action. We developed a toxin sensor constituting of a single chain polypetides composed of two complementary fragments of Nanoluc luciferase linked through substrates of BoNT. In presence of toxin, the linker region is cleaved, and the two luciferase pieces are separated resulting in decrease of luminescence emission.
The assay presented here offers the possibility to test toxins after an acceptable timeframe of cell culture (9 days) in physiologically relevant cells and can be read directly after toxin treatment. Its sensitivity allows to detect functionally active toxin in the femtomolar range, to compare and rank potency of different BoNTs. In addition, we demonstrate that our assay can be used to evaluate not only the process of target cleavage, but also toxin binding and translocation.
For BoNT/A detection in neurons, the construct was designed using split Nanoluc, then re y as internal controls. The internal control is located in front of SNAP25 (Full length). The vector was based on dual hSyn promotor vector [22] and BamHI/NotI were used for cloning. This cellular sensor is named here "BoNT sensor4".
His tag sensor proteins puri cation Different constructs of sensor proteins were puri ed using IMAC. The His6 tag was cloned into pET28a vector at C terminal. BL21 containing correct plasmid of in vitro construct was inoculated overnight and cultivated at 37°C until O.D. around 0.6-0.8, then induced by 0.25 mM of IPTG ( nal concentration) for overnight at 20 °C. The cells are harvested by centrifuging at 4500g, 10 minutes. After that, cells pellets were dissolve in 50 mM HEPES, 150 mM NaCl, pH7.4 buffer and followed by sonication for 3 minutes.
The supernatant of cell lysate was saved and mixed with Ni2+ beads for 1 h with rotation in 4°C. Later, the mixture was loaded into column and the resin was washed with 10 fold of bed column of 20 mM Imidizole in 50 mM HEPES, 150 mM NaCl, pH7.4 buffer. Finally, the protein was eluted in 4 fold of bed column 500 mM Imidizole in 50 mM HEPES, 150 mM NaCl, pH7.4 buffer. The puri ed sensor protein was dialyzed into 50 mM HEPES, pH 7.1 buffer to perform the assay immediately.
In vitro detection assay of Botulinum toxin Sensor protein concentration was estimated by SDS-PAGE with standard BSA as reference. 30 nM of sensor protein was prepared in 50 mM HEPES, 20 μM ZnCl2, 2 mM DTT, pH 7.1 buffer. The Botulinum toxin A was diluted and added into sensor protein solution with a concentration series from 1 nM to 1 fM with dilution factor 10. After 4 h and 24 h incubation at 37°C, Nano-Glo susbrate (Promega) was added into each sample with equal volume. The luminescent signal was measured in plate reader. Assay was preformed duplicate. Negative controls were also designed and performed. BoNT/A was mixed with Nnano-VAMP-Cnano sensor protein or BoNT/B was mixed with Nnano-SV2C-p25-Cnano sensor protein.

Rodent detection assay of Botulinum toxin
Lentivirus particles expressing the polypeptide BoNT sensor4 was added into 7-days neuron cells and then Botulinum toxin A of a series concentration from 300 pM to 1 pM was added into 13-day neuron cells with triplicate. After 48h challenging to toxin, the medium was removed, and a luciferase assay was done as described below.
hiPSC-derived Motor Neurons culture

Image acquisition
Images were acquired at room temperature using the ImageXpress Micro Confocal High-Content Imaging System (Molecular Devices, San José, CA, USA) at 10X and 20X magni cation. Images were analyzed in MetaXpress software (Molecular Devices). A multi wavelength cell scoring analysis was used to count DAPI + and ISLET + nucleus. Finally, pictures were saved in TIFF format and compiled into gures in Illustrator (Adobe).
Recombinant research grade botulinum neurotoxins A and E were manufactured from E.Coli by Ipsen as previously described [23] [24] [25]. Inactive (endonegative) BoNT/A (BoNT/A(0)) was produced by point mutations E224Q, H227Y in BoNT/A [25]. All BoNTs were puri ed and activated to more than 91% nal purity. For assessing the effect of BoNT serotypes A and E, hiPSC-derived neurons were plated in 96-well plate and exposed to the indicated doses of BoNT/A or BoNT/E in 100 µl of culture medium. For each experiment, each dose of BoNT was tested in triplicate. A negative control (medium without toxin) was always included.
hiPSC-derived Motor Neurons detection assay of Botulinum toxin 48 h after plating, iCell MN were transduced with lentivirus particles expressing the polypeptide BoNT sensor4, at MOI of 16, in 40 µl of culture media. 4-6 h after, the media was removed, and 100 µl of fresh medium was added. After 8 days of culture, cells were treated with BoNTs for 24 h, as described above, after which time the medium containing the toxin was removed and a luciferase assay was done as described below.
For SNAP-25 cleavage assay, intensities of the total form and cleaved form of SNAP25 were measured with GeneTools (Philomath, OR, USA).

GT1b assay
Six days after thawing, GT1b (30µM) was added to the media recommended by supplier of previously transduced iCell MNs. 48h later, this media was removed, and the cells were treated with rBoNT/A (1pM). After 24h the medium containing the toxin was removed and a luciferase assay was done as described below.

Concanamycin A (ConA) Assay
Eight days after thawing, transduced iCell MNs were treated with rBoNT/A (1nM) in medium recommended by supplier supplemented with 60 mM KCl for 10 min. Toxins were then washed off using PBS. Where appropriate, ConA (250nM, Sigma-Aldrich)was added to the culture. 4h later, a luciferase assay was done as described below.

Luciferase assay
For rodent neurons assay, cells were lysed by adding 200 μl passive lysis buffer (Promega) and incubated at room temperature for 20 minutes. 50 μl of cell lysate was mixed with 50μl re y luciferase substrate (ONE-Glo(TM) EX reagent) and measured. Subsequently, 50μl Nanoluc luciferase substrate (NanoDLR Stop & Glo(R) reagent) and measure it again.
For hiPSCs derived neurons assay, after cell culture medium removing, 80 ml of fresh DMEM w/o phenol red was added. The cells were incubated for 30 min at RT.
Luciferase activity (NANOLUC™and Fire y luciferase) was assayed using the Nano-Glo â Dual-luciferase Reporter kit accordingly to the manufacturer's protocols (Promega) and using the multimode plate reader VICTOR â Nivo TM (Perkin Elmer, Waltham, MA, USA). With this protocol, luminescence was measured in less than one hour following toxin treatment.
For each cell lysate sample, both Fire y and NANOLUC™ luciferase signals were measured and the "luminescence ratio" (NANOLUC™ /Fire y = R i ) was calculated.
Each sample's ratio R i was normalized with the mean ratio of sample without toxin exposure.
For in vitro and rodent experiment, data are expressed as remaining sensor activity, as follows: For the human motor neurons data, results are expressed as a percentage of toxin activity, as follows: % activity = 100*[1 -((R i -R w toxin MAX ) / (R w/o toxin -R w toxin MAX ))] where R w/o toxin is the mean ratio of sample without toxin exposure and R w toxin MAX is the mean ratio of sample with the highest dose of toxin.
Finally, for one speci c toxin concentration, BoNT sensor4 cleavage percentage is determine as follows: BoNT sensor4 cleavage (%) = 100*[1 -(R i / R w/o toxin )] Results shown are from four independent experiments in triplicates.

Statistics
All results are presented as mean ± SEM of n independent experiments. Dose-response curves were tted with a four-parameter logistic equation, and the pEC50 was calculated. All data processing and statistical tests were done using GraphPad Prism version 8 (GraphPad Software Inc., La Jolla, CA, USA). Statistical analysis indicates one or two-tailed t-test or ordinary one-way ANOVA multiple comparisons P values, as indicated in gure legends. Signi cance is indicated by asterisks or NS, as follows: *P<0.05, **P<0.01, ***P<0.001; ****P<0.0001; NS: not signi cant P>0.05.

Results
Development of an assay based on a split form of Nanoluc luciferase to detect BoNT activity.
In order to establish a novel cell-based approach to detect the activity of BoNT, we rst developed an antibody-free detection system using recently developed split-Nanoluc luciferase [21]. More precisely, we expressed a single polypeptides chain composed of Nanoluc luciferase split in two parts and maintained close to each other by a linker containing BoNT substrates sequence. Once BoNT cleaves its target, the two parts of the Nanoluc luciferase are no longer in proximity, resulting in a decrease in its activity ( Figure 1A). The remaining luciferase activity can be inversely correlated with the toxin activity.
Several single chain polypeptides containing different BoNT substrates were designed ( Figure 1B). For instance, a sensor contains a linker composed of a SNAP-25 fragment that can be cleaved by BoNT/A, E and C and another sensor contains an additional fragment of SV2C, functioning as toxin receptor, in addition to the SNAP-25 fragment. The third version uses a VAMP 1 fragment as the linker, which can be cleaved by BoNT/B, D, F and G. These sensors proteins were puri ed as recombinant proteins.
Then, to evaluate the functionality of these sensor proteins and their respective sensitivity to BoNT, the three version of sensors were incubated with a gradient of concentration of BoNT/A or BoNT/B for 24 hours and the remaining luciferase activity was measured.
Both version of sensors containing a SNAP-25 fragment showed a concentration-dependant decrease of their luciferase activity after BoNT/A incubation ( Figure 1C). Moreover, the sensor that contains SV2 fragment has an estimated EC50 at 7.86 pM. This is approximatively 5-fold more sensitive than the sensor without SV2C (EC50=35.87pM), indicating that including fragment of toxin receptor can improve the sensitivity of the sensor ( Figure 1C). BoNT/B did not affect the luciferase activity of these two sensors, demonstrating the speci city of these two sensors for BoNT/A ( Figure 1C).
The third sensor showed a concentration-dependant decrease of their luciferase activity after BoNT/B incubation ( Figure 1D). Incubation of this sensor with BoNT/A did not affect the luciferase activity, demonstrating its speci city for BoNT B ( Figure 1D).
All these data show that the split-Nanoluc luciferase-based sensors can be used to detect BoNT activity.

NANOLUC TM sensor can detect BoNT/A in cells and human motor neurons
We next sought to develop cell-based assays using split-Nanoluc luciferase-based sensors. As the majority of BoNT products are BoNT/A, we focused on establishing a SNAP25-cleaving cell-based sensor.
For this purpose, a construct expressing a split-Nanoluc luciferase with a linker containing a re y luciferase, which serves as an internal control, and the full length of SNAP-25 was developed. This recombinant reporter was named BoNT sensor4 and it is illustrated in Supplementary Figure 1. Then, this sensor protein was expressed in neurons via lentiviral particles, as developed by Gascon and coworkers [22] and adapted by us for co-expressing BoNT sensor4 and GFP. The lentiviral expression vector is illustrated in Figure 2A.
Primary rat cortical neurons transduced with these lentiviral particles were exposed to various concentrations of BoNT/A and cell lysates were harvested 48 hours later. The luciferase activities of both re y and NANOLUC™ luciferase were measured. Incubating neurons with toxin reduced the signal of NANOLUC™ luciferase, showing BoNT sensor4 ability to sense toxin activity in cells ( Figure 2B).
We next decided to transfer this assay into a well characterized, commercially available hiPSc model, the iCell Motor neurons. This hiPSC model has been shown to be highly sensitive to BoNT serotypes A and E [26], and to express, after 14 days of culture, markers of authentic and functional motor neurons [19]. First, we checked the biosensor expression 14 days after thawing, using lentiviral particles. We transduced iCell MNs and after two weeks of culture we observed that the majority of the MNs were GFPpositive (Supplementary Figure 2), indicating that lentiviral particles are e cient in transducing these human cells. After cell lysis we checked the expression of the engineered polypeptide by Western blot using an antibody against Fire y luciferase. This blot revealed a single band at the expected molecular weight suggesting that the SNAP-25 sensor is well expressed (Supplementary Figure 2).
Because our previous work strongly suggested that 7 days of maturation were enough for iCell MNs to express motoneuronal markers and proteins involved in BoNT intoxication [19], we compared toxin sensitivity after 1 or 2 weeks of culture. At these two different stages of maturation, we treated cells with BoNT/A for 24 hours and we tested SNAP-25 cleavage by western blot. At the two time points tested, we observed a concentration-dependent increase in SNAP-25 cleavage with a complete cleavage of the target at higher concentrations. The two curves were similar with a comparable pEC50 (-11.89±0,03 at day 7 and -11.89±0.04 at day 14-15) indicating a similar sensitivity to the toxin ( Figure 2C). This result indicates that the extra week in culture did not change toxin-sensitivity of MNs.
Finally, we checked, at an early maturation time point, if transduction disturbed cell identity of MNs. After 8 days of culture, we xed transduced cells and quanti ed MNs using the nuclear motoneuronal marker ISLET1. Figure 2D, 2E and supplementary gure 3 show that, at this stage, the number of ISLET + cells per well in transduced conditions were similar to the control. Therefore, no negative impact on the motoneuronal identity of the cell culture was detected, showing that transduction does not impair motor neurons differentiation.
Altogether these results con rm that iCell MN overexpress BoNT sensor4 and already after 7 days in culture can be used for a cell-based assay without losing toxin sensitivity.
Femtomolar concentrations of BoNT/A or /E can be measured in human MNs expressing NANOLUC TM sensor less than one hour after toxin treatment end.
Based on these results we established a new CB-assay protocol that can be run in 9 days, as summarized in Figure 3A. Brie y, iCell MNs are transduced 48 hours after plating, followed by 6 days of further cell culture. Then, a 24 h toxin treatment 8 days after rst plating is done, after which luminescence signals can be directly measured.
To evaluate the sensitivity of this new CB-assay, we established concentration-response curves for two different recombinant BoNT serotypes, BoNT/A and BoNT/E. Both toxins showed a concentrationdependant increase of toxin activity ( Figure 3B). Maximal activity resulted in complete cleavage of SNAP- 25 and was observed at the highest concentration for both toxins. As shown by us previously using a western blot SNAP-25 cleavage assay [19], BoNT/E (pEC 50=13.57±0.05) was signi catively more potent than BoNT/A (pEC50=12.57±0.05) in this new assay ( Figure 3C). As a control, a mutant form of BoNT/A with inactivated LC did not lead to any changes in luminescence.

Mechanistic question can be addressed by the new CB-assay
To determine if the new CB-assay could also be used to study the mode of action of BoNT, we evaluated the effect of two well-known modulators of BoNT binding and LC translocation.
It was previously shown that addition of ganglioside GT1b, a BoNT coreceptor, could improve toxin sensitivity of hiPSC derived neurons [14] by facilitating BoNT binding. To evaluate if our new CB-assay would be sensitive to gangliosides, we added 1 pM of GT1b to the cell culture, 48 hours before treatment with BoNT/A. Figure 4A shows that addition of GT1b increased the cleavage of the engineered target.
Moreover, using ConcanamycinA (ConA), a speci c vacuolar H + -ATPase inhibitor, which blocks endosomal acidi cation and therefore antagonized the action of BoNT [27], we tested if our assay is sensitive to this effect. Figure 4B shows that treating transduced MNs with 1 nM BoNT/A in high-K + concentration conditions (to stimulate synaptic vesicles), for 10 min, resulted in approximately 30% substrate cleavage during a subsequent 4 hours. This effect was completely blocked when 250 nM of ConA was co-administrated with the toxin to cells.
Together, our data show that the new cell-based assay described here reports on all intoxication steps by botulinum toxins and therefore can also be used for mechanistic studies.

Discussion
In this paper, we describe a new cell-based assay that provides a reproducible and highly sensitive model for SNAP-25 cleaving toxins and it is independent from using any specialized antibodies. Moreover, the assay combines human iPSC derived MNs, which represent a physiological cellular model for BoNT activity, with a straightforward, simple detection of the enzymatic action of the toxin activity by directly quantifying luminescence after toxin treatment.
BoNT testing by the new assay is a further development of previous work. Several studies already demonstrated the potential of hiPSC-derived MNs for BoNT testing [17] [18] [19]. More precisely, it has been shown that these cells are a sensitive neuron type for BoNTs [17] and express marker proteins of mature motor neurons [19].
Previous studies also suggested that the time in culture of iCell MNs, before BoNT testing, could be shortened [16] [19]. If 14 days are required to obtain highly mature and functional MNs, we con rmed here that we could test the toxin from day 7 without losing sensitivity.
Moreover, the "direct" luminescent read-out used here, also enabled to measure BoNT activity less than one hour after toxin treatment. Compared to a cell-based assay with an ELISA read out [13], the assay described here does not need several steps of incubation and wash due to the use of antibodies. Furthermore, it doesn't require any special antibody which can be costly and di cult to produce. Overall, our assay is cheaper, easier and faster than an ELISA-based readout.
While alternative assays have been developed with uorescent probes in an effort to avoid using antibodies [15], the major drawback of these assays required complicated equipment such as uorescent microscope. Alternative readouts based on natural luminescence [28] or on electrochemical multielectrode [29] arrays have been developed. For example, Pathe-Neuschafer-Rube and co-workers developed a pH-stable re y-luciferase targeted to secretory vesicles in the human SiMa cell line and showed that BoNT/A could inhibit luciferase secretion [28]. Such approaches do circumvent a limitation of our assay which is to work theoretically for all BoNT serotypes. Indeed, we designed a speci c SNAP25-cleaving BoNT assay, but our system is very versatile and BoNT target sequence can be easily changed. The production of in vitro sensor targeting VAMP1 illustrate that point.
To be a good in vitro alternative to the mouse lethality bioassay (MLB), assay sensitivity is a crucial parameter. It has been reported that the MLB has a LD50 range of 5-10 pg/ml for botulinum toxin [30] and here, we reached an EC50 of 40 fg/ml (=272 fM) for BoNT/A. While the cell-based assay developed here has thus the necessary sensitivity to serve a batch-release cell-based assay, further validation will be required to use it for such purposes.
The sensitivities measured here for BoNT/A and E are similar as those obtained by western blot cleavage assay in the same cellular model after 14 days of culture [19]. Interestingly, in both studies, BoNT/E is more potent compare to A. This does not t with previous experiments done on primary rodent cells [31] that indicated that BoNT/A was 30 times more potent than BoNT/E. But our ndings match with the observations from the rst clinical trial in human (phase one study) in which it was shown that BoNT/E potency is at least equivalent or even greater compare to toxin A [24]. These observations underline the translational value of a human cellular model.
Finally, to prove we can use our system to test different steps of toxin intoxication, we tried to modulate toxin binding by a pre-exposure of our cells with gangliosides. As it was previously shown for iCell GABA neurons [14], we found that GT1b addition also improves BoNT/A potency in MNs (Figure4A). This observation differs from earlier ndings which indicated that pre-exposure with gangliosides on iCell MNs did not impact their sensitivity to BoNT/A [17]. This discrepancy could be explained by the difference in nature of BoNT, protocols for gangliosides and toxin treatment. In the previous study, iCell MNs were pretreated with 45µM GT1b for 24 hours prior to natural BoNT/A (produced by Clostridium) exposure while in our study, cells were pre-treated with 30 µM for 48h prior to recombinant BoNT/A exposure. Apart from these small differences, another important parameter to differentiate these two experiments is the readout which is employed to quantify the action of BoNTs. This suggests that luminescent tool is more sensitive to small changes in BoNT cleavage compare to Western blot.
We also showed that our assay can be used to study toxin translocation. The blockage of this cellular process in our cells, completely inhibited rBoNT/A activity.
Nevertheless, the CB assay presented here can be improved. One way to do it is to generate a stable hiPSC line expressing the SNAP25 luminescent probe. From these stem cells, it would be possible to generate specialized neurons (MNs, glutamatergic neurons, sensory neurons…) bankable as precursors.
This approach would overcome the usage of lentivirus, limit number of manipulations and reduce time of experiment. All these improvements will facilitate the use of these engineered cells in a QC environment and/or for high-throughput screening.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryFigures.pdf