Previously we reported that 0.81 pM BoNT/A treatment for 24 h results in cleavage of 50% of SNAP-25 in ESNs [28]. Reasoning that application of higher doses for a shorter period might accelerate toxin internalization, we exposed ESNs to 0.67–670 pM BoNT/A for 3 h or 6 h and evaluated SNAP-25 cleavage after 24 h (Figure 1A). In comparing the percent cleaved SNAP-25 at 24 h following either 3 or 24 h intoxication, we found that roughly three-quarters of toxin internalization occurs within the first few hours. Since both 6.7 and 67 pM produced roughly 50% cleaved SNAP-25 after 3 h exposure, we longitudinally evaluated BoNT/A activity from 3–96 h (Figure 1B). At each concentration the rate of SNAP-25 cleavage per hour slowed dramatically after 24 h, suggesting a balance between rates of SNAP-25 synthesis and cleavage (Figure 1C). Based on these data, we designed a screening assay in which ESNs were exposed to 6.7 pM BoNT/A for 3 h, then washed and incubated for 21 h to allow full toxin activation. Candidate therapeutics were applied at 24 h, and the recovery of full-length SNAP-25 was evaluated 48 h after therapeutic treatment (summarized in Figure 1D). The duration of incubation following treatment was selected based on reports that approximately 2% of cellular SNAP-25 is recycled per hour [30].

To evaluate whether LTX treatment altered light chain (LC)/A activity in ESNs, BoNT/A-intoxicated ESNs were exposed to 400 pM LTX for 6.5 or 13 min, and SNAP-25 integrity was evaluated after 48 h. As a control, ESNs were also treated with 60 mM K⁺ (KEB) for 1.5 min, which evokes Ca²⁺-dependent glutamate release [28]. LTX treatment resulted in rescue of 92 ± 4.6% and 98 ± 1.7% (6.5 and 13 min, respectively) of full-length SNAP-25 within 48 h, whereas KEB treatment showed no difference from untreated neurons (Figure 2). The restoration of uncleaved SNAP-25 indicates that some aspect of LTX treatment results in the inactivation or clearance of LC/A from synaptic termini. Furthermore, these data suggest that latrotoxin may be the active moiety in experiments demonstrating that the administration of crude homogenate from black widow spider venom glands to BoNT/A-paralyzed neuromuscular junctions dramatically accelerates recovery [25].

LTX treatment incurs fulminant neurotransmitter release from primary neurons, partly in response to profound levels of Ca²⁺ internalization [31]. Since LTX treatment resulted in recovery of full-length SNAP-25 in ESNs, whereas KEB did not, we used the fluorescent intracellular calcium sensor Fluo-4 to compare the amplitude and duration of Ca²⁺ internalization evoked by these two treatments. Time-lapse confocal microscopy analysis of DIV 21 ESNs treated with KEB demonstrated an immediate rise in Fluo-4 fluorescence (Figure 3A). Fluorescence intensity increased within 15 s of KEB treatment, remained high for about 90 s and subsided prior to washout at 2.5 min (Figure 3B). Conversely, while ESNs treated with 400 pM LTX also showed a strong rise in Fluo-4 fluorescence, there were several key differences in the kinetics of the Ca²⁺ response. First, there was a 1.5 min delay between LTX addition and the onset of Fluo-4 fluorescence, suggesting that spider toxin requires time to bind synaptic receptors and form pores within the membrane at this concentration (Figure 3B). Unlike the self-limiting response from KEB treatment, LTX induced a steady increase in Fluo-4 signal throughout the experiment, even following rinses to remove residual toxin, demonstrating that LTX results in loss of cellular ionic homeostasis and that the KEB response was not limited by reduced levels of extracellular Ca²⁺. No apparent differences in onset, amplitude or duration were noted within the 20 min experimental window between the 6.5- and 13 min LTX treatment, nor did a 24 h intoxication with 67 pM BoNT/A prior to LTX or KEB treatment affect Ca²⁺ internalization (not shown).

Differential interference contrast images captured 15 h after treatment indicate that cultures treated with LTX for 6.5 min have large numbers of distributed varicosities and disrupted processes, whereas cultures treated with KEB for 13 min do not (Figure 4A). LTX treatment of ESNs decreased the total protein yield following cell lysis by 44 ± 16% and 52 ± 19% (6.5 and 13 min, respectively) compared to controls at 48 h after treatment (p < 0.01, n = 6; Figure 4B). We used calcein green staining to compare morphological changes in ESNs during the emergent and long-term response to LTX versus KEB. These observations are in agreement with results from LTX-treated spinal cord motoneurons and cerebellar granule neurons [31]. The development of axodendritic varicosities from existing processes was apparent within 30 min of LTX addition (Figure 5). Although neurons remained viable 24 h and 48 h after treatment, there was an overall decrease in viability and persistent evidence of varicosities. By 48 h some of these varicosities were resolved, suggesting that neuronal regeneration was underway.

Although we previously demonstrated that neurons derived from embryonic stem cells have the potential of acting as a biologically relevant platform for botulinum research, we did not develop screening protocols to maximize the likelihood of rapidly identifying therapeutically useful candidates [28]. In this work we started by developing a cell-based assay using the kinetics of BoNT/A internalization and activation in ESNs to standardize the evaluation of therapeutic candidates.

Although it is a widespread practice to use toxin concentrations that are several orders of magnitude above the EC50 followed immediately by therapeutic administrations to expedite therapeutic screening assays, for several reasons this approach significantly increases the risk of failing to identify a valid therapeutic [32,33,34,35]. First, internalization of BoNT at high concentrations may result in broad cellular distribution as opposed to synaptic localization, altering the ability to evaluate therapeutic efficacy. Second, only a fraction of internalized toxin is capable of cleaving the majority of SNAP-25 within 6 h (see Figure 1). Thus, evaluating therapeutic candidates before all the internalized toxin is activated will artificially increase the ratio between therapeutic molecules and toxin, thereby increasing the apparent therapeutic activity. Finally, a therapeutic candidate applied too early may disrupt toxin processing rather than interfere with fully functional toxin. This latter point is particularly relevant in high-throughput screening approaches for post-exposure therapeutics in which libraries are selected without a mechanistic bias.

In consideration of these concerns, we designed an assay around three principles intended to mitigate risks and ensure a high level of sensitivity for therapeutic efficacy. First, an incubation step was included between the removal of holotoxin and prior to therapeutic administration to allow internalized light chain to become fully activated. This step was designed to be sufficiently long to establish a steady state between cleavage of existing SNAP-25 and synthesis of new SNAP-25. Second, a dose was selected that would result in approximately 50% cleavage of SNAP-25 within the given time frame, yet be as low as possible to encourage toxin internalization and trafficking by synaptic endocytosis rather than non-specific endocytosis. Third, treatment efficacy timescales were selected around presumptive mechanisms of action. Thus, using dose response curves at multiple time points and exposure durations, we identified a 3 h exposure to 6.7 pM, with therapeutic application at 24 h and SNAP-25 evaluation at 72 h as the preferred model.

The finding that LTX treatment of BoNT/A-intoxicated ESNs rescued full-length SNAP-25 expression is the first instance of a successful therapeutic application in derived neurons. This finding resulted from the inactivation of LC/A by an unknown mechanism, allowing newly synthesized SNAP-25 to accumulate without being proteolytically cleaved by LC/A. It should be noted that we have not yet evaluated the functional rescue of synaptic signaling or the regeneration of normal synaptic morphologies with appropriate localization of synaptic proteins. The significant decrease in total protein recovered argues that some degree of neurotoxicity has occurred, and this is corroborated by morphological changes. It may be that most of the existing synapses have been disrupted in such a fashion to inactivate LC/A, or that the LC/A-intoxicated synapses are more sensitive to LTX-treatment. The data does suggest that LTX treatment results in calcium ‘overload’ and the abnormal flux of other ions through non-selective, cationic pores in the synaptic membrane, any of which may prove injurious to neurons [31,36]. The profound and sustained increase in cytosolic Ca²⁺ observed in ESNs followed by acute evidence of axodendritic varicosities is characteristic of excitotoxicity and presents a possible mechanism for clearance or inactivation of LC/A from synaptic termini via synaptic degeneration. Whether such degeneration results in the physical loss of LC/A from the synaptic compartment or in accelerated degradation of synapse-localized proteins via cellular proteosomal clearance mechanisms is unknown, but clearly of interest. Another possibility is that the significant increase in intracellular Ca²⁺ more directly inhibits the BoNT/A light chain cannot be discounted based on these data; e.g., through induction of autocatalysis [37].

Recovery from botulinum requires the re-innervation of the inactive muscle fiber, whether by regeneration of the original endplate or by establishment of a new endplate via axonal sprouting. In vivo, it appears that the endplate disassembles and the presynaptic compartment regresses in response to BoNT intoxication [38]. Shortly afterward, new axonal sprouts develop and attempt to re-innervate the target muscle fiber. The original endplate is eventually regenerated once the light chain has been cleared from the presynaptic terminal and appears to regain control as the primary efferent to the muscle fiber. The coordination of these events demonstrates the importance of synaptic remodeling in recovery of normal synaptic function after intoxication. It may be possible that by disrupting the integrity of the original synaptic terminal, BoNT light chains no longer remain associated with intoxicated synapses, thereby enabling the immediate re-innervation of endplates by re-activated nerve terminals. Similarly, in vivo studies have demonstrated dramatically accelerated recovery from paralysis following BoNT intoxication following nerve crush injury [26,27]. If this general approach proves to be efficacious in facilitating recovery from intoxication in vivo, then induced regeneration of paralyzed synapses could offer a novel, serotype-independent therapy for intoxication by persistent BoNTs.

The formation of varicosities in response to LTX treatment is suggestive of the morphological changes described following application of excitotoxic stimuli to cultured neurons. For example, glutamate treatment leads to varicosity formation in glutamatergic dendrites, whereas veratridine inactivation of voltage-gated sodium channels causes varicosity formation in both axons and dendrites. Under conditions of a sublethal excitotoxic stimulus, these varicosities have proven to be reversible via volume recovery pathways, although it is not known if axons and dendrites utilize similar mechanisms to resolve neuronal swelling [39,40,41]. While the nature of the potential relationship between excitotoxicity resolution and loss of BoNT persistence is still unclear, these results indicate that ESNs provide a stable cell-based platform to further interrogate the effect of LTX-induced synaptic regeneration on BoNT-persistence in cultured neurons.

There are several aspects of LTX activity that are not well understood [10]. For example, in addition to the formation of cation-selective pores permeable for Ca²⁺, LTX mutants incapable of forming pores can still induce membrane depolarization through inhibition of voltage-gated potassium channels and activation of L-type Ca²⁺ channels [42,43]. Whereas synaptic exocytosis induced by this mode of LTX action is dependent on the presence of the classical SNARE machinery, there is evidence that substantial levels of neurotransmitter release induced by pore-mediated Ca²⁺ internalization can still occur despite the absence of SNAP-25, VAMP-2 or Munc-13 [16]. Whether this involves a novel complex that can effect exocytosis at high Ca²⁺ concentrations or has a more mundane explanation such as enhanced reversal of membrane-associated amino acid transporters in response to changes in cytoplasmic ion concentrations or ATP-depletion is not known [44,45]. However, here we have demonstrated that ESNs offer an LTX-sensitive, neuron-based platform that has the potential to replace the use of primary neurons and neurogenic cells in studying the molecular, cellular and functional consequences of intoxication by LTX. As we have previously described for BoNTs, the additional capabilities of genetic tractability, neuronal subtype homogeneity and lot-to-lot consistency mean that the ESN model may also be a transformative tool for LTX research [46].