Use-dependent potentiation of voltage-gated calcium channels rescues neurotransmission in nerve terminals intoxicated by botulinum neurotoxin serotype A

Botulinum neurotoxins (BoNTs) are highly potent toxins that cleave neuronal SNARE proteins required for neurotransmission, causing flaccid paralysis and death by asphyxiation. Currently, there are no clinical treatments to delay or reverse BoNT-induced blockade of neuromuscular transmission. While aminopyridines have demonstrated varying efficacy in transiently reducing paralysis following BoNT poisoning, the precise mechanisms by which aminopyridines symptomatically treat botulism are not understood. Here we found that activity-dependent potentiation of presynaptic voltage-gated calcium channels (VGCCs) underlies 3,4-diaminopyridine (3,4-DAP)-mediated rescue of neurotransmission in central nervous system synapses and mouse diaphragm neuromuscular junctions fully intoxicated by BoNT serotype A. Combinatorial treatments with 3,4-DAP and VGCC agonists proved synergistic in restoring suprathreshold endplate potentials in mouse diaphragms fully intoxicated by BoNT/A. In contrast, synapses fully intoxicated by BoNT serotypes D or E were refractory to synaptic rescue by any treatment. We interpret these data to propose that increasing the duration or extent of VGCC activation prolongs the opportunity for low-efficiency fusion by fusogenic complexes incorporating BoNT/A-cleaved SNAP-25. The identification of VGCC agonists that rescue neurotransmission in BoNT/A-intoxicated synapses provides compelling evidence for potential therapeutic utility in some cases of human botulism.

ventilation and parenteral feeding, until neuromuscular function is restored. Some serotypes produce paralysis that persists for months, requiring sustained supportive care with increased risk of co-morbidities.
Although post-exposure prophylaxis with antitoxin can efficiently neutralize toxin prior to neuronal uptake 4,5 , antitoxin administration requires clinical evidence of progressive paralysis, and thus even patients that receive a timely administration of antitoxin are likely to suffer symptoms of botulism. Despite intensive research efforts over the past three decades, small molecule LC inhibitors with clinically suitable pharmacokinetics and pharmacodynamics have not been identified 6,7 . Consequently, there remains a critical need for treatments that reverse neuromuscular paralysis in botulism patients.
In vitro studies suggest that neuromuscular function of intoxicated tissues can be transiently improved by enhancing neuronal Ca 2+ influx, such as through increased extracellular Ca 2+ , treatment with aminopyridines or exposure to cationic ionophores [8][9][10] . Of these, the aminopyridines are particularly intriguing because derivatives such as 4-aminopyridine and 3,4-diaminopyridine (3, are currently in clinical use for neuromuscular indications 11 . Aminopyridines block the intracellular domain of voltage-gated K + channels (VGKC), extending the duration of action potential-induced presynaptic depolarization and increasing neurotransmitter release 12,13 . However, clinical evaluations of aminopyridines in patients with severe serotype A and B botulism have resulted in highly variable and conflicting reports of efficacy [14][15][16] . Because the precise mechanisms by which aminopyridines enhance muscle contraction in paralyzed tissues are not known, conditions under which they can mitigate botulism symptoms have not been established.
While augmenting presynaptic cytosolic Ca 2+ influx is likely to increase release probability at non-intoxicated release sites in partially intoxicated NMJs 17 , it is unknown whether other mechanisms can also contribute to symptomatic rescue. Here, we investigated the cellular and molecular mechanisms of 3,4-DAP-mediated rescue of neurotransmission in networked cultures of primary rat neurons and isolated mouse diaphragms intoxicated by BoNT serotypes A, D or E (the serotypes most commonly associated with human disease). These studies were conducted under conditions in which neurotransmission was thoroughly blocked by BoNT intoxication, thereby enabling the evaluation of treatments that restore the ability of intoxicated synapses to undergo synaptic release. These data provide compelling evidence that enhancing presynaptic Ca 2+ influx through agonism of voltage-gated Ca 2+ channels (VGCCs) is sufficient to rescue neurotransmission in synapses intoxicated by BoNT/A. Furthermore, they reveal that drugs that synergize with 3,4-DAP to enhance phasic Ca 2+ influx represent a potentially novel class of candidate botulism therapeutics with improved efficacy profiles.

3,4-DAP rescue of evoked neurotransmission is unique to BoNT/A-poisoned synapses and dependent on VGCC activation.
To characterize the effects of 3,4-DAP on action potential-elicited neurotransmission at intoxicated synapses, whole-cell patch-clamp electrophysiology was used to measure excitatory postsynaptic currents (EPSCs) in response to extracellular stimulation of proximal axons in primary neuron cultures 18 . These studies were done in the presence of GABA and NMDA receptor antagonists, permitting aggregate analysis of synaptic transmission at afferent glutamategic synapses. Field stimulation of primary rat hippocampal and cortical neurons at 18-24 d after plating produced EPSCs with amplitudes of 512.2 ± 77.1 pA (Fig. 1A). Spontaneous miniature EPSCs (mEPSCs) measured in the same neuron population had mean amplitudes of 14.8 ± 1.1 pA, indicating that field stimulation elicited release of approximately 35 quanta per trial. The elimination of EPSCs by addition of either tetrodotoxin (TTX) or the AMPA/kainic acid receptor antagonist CNQX confirmed that EPSCs constituted post-synaptic responses to presynaptic release (Fig. 1A). 3,4-DAP treatment of naïve cultures increased EPSC amplitudes by 166.7 ± 26.4% (p = 0.027; Fig. 1A), without altering spontaneous miniature EPSCs (mEPSC) amplitudes or frequencies (Supplementary Figure S1A), consistent with the known pharmacological role of 3,4-DAP as a voltage-gated potassium channel blocker.
The effects of 3,4-DAP on neurotransmission were first tested in cultured neurons intoxicated with high concentrations of BoNT/A 19 . This approach allowed us to specifically test for treatments that restored synaptic function at fully intoxicated synapses. Western blot analysis of whole cell lysates revealed conversion of SNAP-25 to the BoNT/A cleavage product SNAP-25Δ9 in intoxicated cultures (Fig. 1B). BoNT/A intoxication reduced EPSC amplitudes to 4.0 ± 1.3% of naïve controls, demonstrating near-complete blockade of evoked release (Fig. 1C). 3,4-DAP treatment restored evoked release in 100% of evaluated neurons, increasing ESPC amplitudes to 45.1 ± 14.4% of naïve controls (p = 0.0012; Fig. 1C). As expected, 3,4-DAP had no effect on intoxicated neurons in the presence of TTX, indicating that synaptic rescue by 3,4-DAP was not mediated post-synaptically (Supplementary Figure S1B). We next evaluated the molecular mechanisms involved in 3,4-DAP rescue of neurotransmission. A secondary neuropharmacological effect of 3,4-DAP is prolongation of voltage-gated Ca 2+ channel (VGCC) activation, which increases presynaptic Ca 2+ influx and vesicle release probability 12 . This suggested that enhanced Ca 2+ influx through VGCCs in response to 3,4-DAP treatment contributed to the rescue effect. To test this hypothesis, BoNT/A-intoxicated cultures were treated with 3,4-DAP and L, N, R and P/Q-type VGCC antagonists (nimodipine, ω-agatoxin IVA and ω-conotoxin MVIIC). The presence of VGCC antagonists completely blocked rescue of neurotransmission by 3,4-DAP, demonstrating a requirement for VGCC activity (p = 0.029; Fig. 1C).

VGCC agonism is sufficient to rescue neurotransmission in synapses intoxicated by
BoNT/A. The above data demonstrated that VGCC activation was necessary for 3,4-DAP rescue of neurotransmission in BoNT/A-intoxicated synapses. To more specifically determine whether VGCC currents were necessary for rescue, we recorded spontaneous synaptic release (mEPSCs) in the presence of elevated Ca 2+ as well as VGCC agonists. In contrast to evoked release, spontaneous release is a probabilistic phenomenon, and consequently, mEPSC frequencies reflect the average release probability among the hundreds or thousands of afferent synapses on each neuron 20,21 . The ability to surveil the behavior of many synapses, with single-synapse resolution, facilitated a more specific evaluation of the effects of treatment conditions on release probabilities 22 .
While blockade of VGCCs proved to prevent rescue under multiple contexts, it remained possible that this was a reflection of the critical role that Ca 2+ plays in neurotransmission, and not directly related to the rescue Mean EPSC amplitudes are normalized to recordings from age-matched, nonintoxicated control cultures. Arrows represent stimulation times. Scale bars represent 400 ms (x-axis) and 100 pA (y-axis). All data presented as mean ± SEM. *Indicates p < 0.05, **Indicates p < 0.01, ns = not significant.

Combination treatment with 3,4-DAP and GV-58 restores EPSC amplitudes in neurons intoxicated by BoNT/A, but not
BoNT/D or /E. We next evaluated whether combinatorial addition of 3,4-DAP and a VGCC agonist had additive effects on evoked release in BoNT/A-intoxicated cultures. GV-58 was chosen for these studies because it specifically targets VGCCs that are expressed at the synapse and associated with evoked release 23,24 . In BoNT/A-intoxicated neurons, combination treatment with 3,4-DAP and GV-58 restored EPSC amplitudes to levels that were indistinguishable from non-intoxicated cultures (Fig. 5A,B; 106.6 ± 21.0% versus 100 ± 15.1% respectively, p = 0.80), and 17-fold improved over intoxicated cultures treated with vehicle alone (6.0 ± 1.3%, p < 0.001). Combination treatment proved more effective than 3,4-DAP alone (p = 0.019), though there was no statistical improvement compared to GV-58 alone (p = 0.41).

3,4-DAP plus GV-58 restores suprathreshold release in mouse diaphragm endplates fully intoxicated by BoNT/A.
Although these data demonstrated that enhanced VGCC activity could restore neurotransmission to CNS synapses severely intoxicated by BoNT/A, the physiological target of botulism is peripheral neurons. Thus, we used intracellular endplate recordings to evaluate the effects of treatment with 3,4-DAP and/or GV-58 on phrenic nerve-elicited endplate potentials (EPPs) in mouse hemidiaphragms preparations intoxicated ex vivo by BoNT/A or BoNT/E. The hemidiaphragm assay closely mimics the physiological progression of respiratory paralysis, and has been used extensively to study toxin mechanisms of action 25,26 .
To reproduce the comprehensive synaptic blockade used in CNS cultures, diaphragms were intoxicated with high doses of BoNT/A or /E and EPPs were serially monitored in sequential endplates until phrenic nerve stimulation failed to elicit EPPs in five consecutive endplates. This represented the statistical point where the majority of endplates are fully intoxicated, and few or no release sites remained competent to undergo vesicle fusion. Addition of either 3,4-DAP (2 µM) or GV-58 (50 µM; doses established in pilot dose-ranging studies) to BoNT/A-intoxicated preparations rapidly restored sub-threshold release, producing EPPs that were 3.49 ± 0.44 mV and 0.89 ± 0.11 mV, respectively (p < 0.01; Fig. 6). Surprisingly, combinatorial addition of 3,4-DAP plus GV-58 had a supra-additive effect in BoNT/A-intoxicated endplates, producing EPPs with amplitudes of 34.7 ± 5.19 mV (Fig. 6). In contrast, 3,4-DAP plus GV-58 had no apparent effect on diaphragms intoxicated by BoNT/E (Fig. 6), consistent with previous data suggesting that VGCC-mediated rescue was specific to BoNT/A-intoxicated synapses.

Discussion
In this study, we investigated the cellular and molecular mechanisms responsible for 3,4-DAP-mediated rescue of neurotransmission with the goal of determining the conditions under which 3,4-DAP could be effective in restoring synaptic function. We found that 3,4-DAP rescue of synaptic release in intoxicated synapses was specifically mediated through VGCC activity. Furthermore, treatments that either increased Ca 2+ influx through VGCCs or prolonged VGCC opening times were sufficient to rescue evoked and spontaneous release in BoNT/A-intoxicated neuron cultures. Functional rescue of neurotransmission was unique to BoNT/A and did not occur with serotypes D or E under any circumstance. The physiological relevance of these findings were confirmed in diaphragm NMJs, where combination treatment with 3,4-DAP and GV-58 proved synergistic, restoring putatively suprathreshold release to hemidiaphragms comprehensively paralyzed by BoNT/A.
The finding that rescue was unique to BoNT/A-intoxicated synapses is consistent with biochemical data suggesting that SNAP-25Δ9 remains competent to engage in or facilitate vesicle fusion 27 . In in vitro vesicle fusion studies, SNAP-25Δ9 directly engages with SYB1/2 and STX1 to mediate low-efficiency vesicle fusion. In contrast, vesicle fusion is abolished by substitution with the BoNT/E-cleavage product SNAP-25Δ26, illustrating that SNAP-25Δ9 and SNAP-25Δ26 have functionally distinct abilities to engage in neurotransmission. Deletion . mEPSC frequencies are normalized to recordings from age-matched, non-intoxicated control cultures. Scale bars represent 5 s (x-axis) and 40 pA (y-axis). All data presented as mean ± SEM and n ≥ 12 neurons for all conditions. *Indicates p < 0.05, ***Indicates p < 0.001. studies conducted in BoNT/A-intoxicated PC12 adrenal chromaffin cells found that overexpression of SNAP-25Δ9 restores low-level catecholamine release 28 . Collectively, these data suggest that SNAP-25Δ9 can directly participate in vesicle fusion, albeit with reduced efficiency compared to intact SNAP-25.
Although low-efficiency engagement of SNAP-25Δ9 in vesicle release provides a plausible mechanistic basis for rescue of neurotransmission in BoNT/A-intoxicated cultures, we cannot exclude the possibility that a compensatory release mechanism is activated in response to VGCC agonism. However, this putative mechanism must still rely on SNAP-25Δ9, otherwise rescue would have been observed in synapses intoxicated with BoNT/D or /E. Examples of such a mechanism could include direct stabilization of SNAP-25Δ9 in the fusogenic complex by coordination with Ca 2+ ions or recruitment of an accessory protein. However, synthetic vesicles incorporating SYB, STX1 and SNAP-25Δ9 are capable of supporting vesicle fusion in the absence of Ca 2+ or external protein factors, suggesting that neither Ca 2+ nor Ca 2+ -regulated accessory proteins are required for fusion activity 29 . Alternatively, we propose that increased Ca 2+ flux through activated VGCCs in the presence of 3,4-DAP, VGCC agonists or elevated extracellular Ca 2+ promotes rescue of neurotransmission by prolonging the kinetics of release. Synaptic release is initiated by local increases in Ca 2+ , which cause the Ca 2+ sensor synaptotagmin (SYT) to undergo a conformation change that permits assembly of SYB, STX1 and SNAP-25 into the fusogenic trans-SNARE configuration 30,31 . In the absence of Ca 2+ , SYT is a negative regulator of vesicle fusion, thus the duration of SYT permissiveness is determined by the magnitude and duration of local Ca 2+ transients 17 . This leads us to hypothesize that 3,4-DAP and/or VGCC agonists effectively extend the temporal window during which SNARE-mediated fusion is molecularly permissible by increasing presynaptic Ca 2+ transients. This would prolong the window during which less efficient fusion events can occur, such as those involving SNAP-25Δ9.
The ability of use-dependent VGCC agonists to rescue neurotransmission has therapeutic implications for stages of botulism other than complete paralysis. For example, most CNS neurons have 1-3 release sites per synaptic bouton 32 , whereas typical motor nerve terminals contain 800-1200 distinct release sites 33 . Continuous single-endplate recordings reveal the progressive depression of evoked release in response to BoNT intoxication 34 , illustrating that partially intoxicated endplates contain a mixture of intact and intoxicated release sites. Since 3,4-DAP treatment increases EPSC amplitudes in both non-intoxicated and BoNT/A-intoxicated synapses, we hypothesize that enhanced Ca 2+ influx will increase neurotransmission from partially intoxicated motor nerve terminals through two distinct mechanisms: increased release probability at non-intoxicated release sites, which would occur in a serotype-independent fashion; and rescue of fusion at release sites associated with SNAP-25Δ9.   If correct, Ca 2+ -mediated rescue of neurotransmission will then be influenced by factors such as the serotype involved and the extent of paralysis. These factors may contribute to the highly variable and conflicting results reported in the small number of clinical studies using aminopyridines as post-symptomatic treatments for serotype A and B botulism [14][15][16]35 .
Adult NMJs and CNS synapses are believed to predominantly rely on N-type (Ca v 2.2) and P/Q-type (Ca v 2.1) VGCCs for synaptic transmission 23,24 . These two VGCC subtypes incorporate synaptic protein interaction domains that interact with Stx1 and SNAP-25, placing the channels in close proximity to docked synaptic vesicles, and providing a structural association between voltage-gated Ca 2+ influx and synaptic vesicle fusion 36,37 . Although N-and P/Q-type VGCCs are often co-located within the presynaptic compartment, they appear to have functional distinctions. For example, in detonator synapses N-type VGCCs contribute to multivesicular release at specific release sites, whereas P/Q-type VGCCs drive widespread synaptic release 38 . In contrast, neuronal L-type VGCCs (Ca v 1.2 through Ca v 1.4) have been proposed to gate Ca 2+ -release from internal Ca 2+ stores, activating signaling processes that indirectly modulate presynaptic function 39,40 . L-type VGCCs are primarily believed to mediate somatic Ca 2+ transients 41,42 , although several studies suggest that they also directly contribute to neurotransmission 43,44 . While our data indicate that L-or N/P/Q-type VGCC agonists can promote release from naïve as well as BoNT/A-intoxicated CNS synapses, we did not determine whether release is exclusively mediated by direct modulation of Ca 2+ transients at release sites, or whether indirect Ca 2+ signaling mechanisms are also involved. Moreover, since central synapses, motor endplates and autonomic synapses are likely to exhibit morphological and functional differences in VGCC activity, VGCC agonists may have synapse type-specific effects on rescue of neurotransmission.
Although aminopyridines have been studied as BoNT countermeasures for decades, their clinical potential has been compromised by adverse CNS effects at presumptive therapeutic doses 45,46 . Therefore, there is interest in reducing the central access of 3,4-DAP, or alternatively in identifying co-treatments that allow reduction in 3,4-DAP concentrations to tolerable levels while retaining or improving therapeutic efficacy 47,48 . While the combination of VGCC agonists and 3,4-DAP appeared very robust in rescuring synaptic release, direct modulation of Ca 2+ influx carries the risk of neurotoxic responses. For example, excessive cytosolic Ca 2+ levels can activate compensatory changes in presynaptic function that reduce Ca 2+ sensitivity 49 or cause presynaptic Ca 2+ toxicity 50 . However, most VGCC agonists are use-dependent, enhancing Ca 2+ influx predominantly during action potential firing 51,52 . This results in the temporally constrained influx of Ca 2+ at specific subcellular compartments, as opposed to tonic Ca 2+ currents. For similar reasons, combinatorial use of 3,4-DAP and GV-58 is under preclinical evaluation as a treatment for myasthenic diseases 53 . However, it should be noted that although both FPL and GV-58 produced encouraging in vitro results as BoNT/A treatments, in vivo use of both drugs is limited by their poor aqueous solubility and there is a need for derivatives with improved pharmacokinetic and/or pharmacodynamics profiles.
In conclusion, we demonstrate that enhanced Ca 2+ influx through VGCCs restores neurotransmission to BoNT/A-intoxicated synapses. Our findings suggest that co-administration of use-dependent VGCC agonists with 3,4-DAP may increase the safety and efficacy of treatments to prolong survival during the acute phase of botulism, or alternatively, to accelerate recovery of muscle function during the chronic phase of botulism.

Experimental procedures
Animal use statement. The  Neuronal culture and BoNT intoxication. Fresh E18 rat hippocampal, cortical and ventricular zone tissues were obtained from BrainBits (Springfield, IL, USA), dissociated according to the manufacturer's instructions, and plated at a density of 125,000 cells/cm 2 on polyethylenimine/laminin-coated glass coverslips (Sigma-Aldrich). Neuronal cultures were maintained at 5% CO 2 , 37 °C, and 95% humidity in NbActiv4 medium (BrainBits). Experiments were performed 16 to 24 days after plating. For neuronal intoxication, BoNT/A, BoNT/D or BoNT/E were prepared at 100x final concentration in fresh NbActiv4 medium and added to neuron cultures at 100 pM. Neurons were analyzed 20-24 hours after addition of BoNT.
Endplate electrophysiology. For ex vivo diaphragm studies, Male C57BL/6 mice (6-10 wk; Jackson Labs, Bar Harbor, ME) were group-housed and provided a standard diet with regular enrichment and water ad libitum. Mice were thoroughly anesthetized using 5% isoflurane and euthanized by decapitation. Diaphragm muscles and corresponding phrenic nerves were isolated from euthanized mice by dissection at 22-24 °C in Tyrode's solution (137 mM NaCl, 5 mM KCl, 1.8 mM CaCl 2 , 1 mM MgSO 4 , 24 mM NaHCO 3 , 1 mM NaH 2 PO 4 and 11 mM D-glucose, pH 7.4). After dissection, muscles were attached with dissection pins to a 10-mm tissue culture dish containing Sylgard (Dow Corning; Midland, MI, USA) and the preparation was perfused with oxygenated Tyrode's. Muscle viability was verified using a 0.2 ms squarewave of direct current at suprathreshold amplitudes to the phrenic nerve using a bipolar stimulating electrode driven by a stimulation isolation unit (Digitimer North America; Ft. Lauderdale, FL, USA).
Muscle potentials were recorded on a HEKA Elektronik EPC10 patch clamp amplifier using sharp glass electrodes (10-20 MΩs) pulled with a Sutter Instrument P1000. Muscles fibers were impaled close to endplate junctions and recordings with a starting resting membrane potential (RMP) > −60 mV were rejected. Muscle contraction was selectively blocked with 1 μM μ-conotoxin GIIIB, which preferentially blocks muscle-specific voltage-activated Na + channels (Alamone Labs). For all experimental conditions, recordings were performed at 22-24 C in perfused oxygenated Tyrode's. EPPs were elicited by suprathreshold phrenic nerve stimulation with a square wave of constant current for 0.2 ms at 0.05 Hz. EPP amplitudes were averaged from 6 consecutive stimulations per muscle fiber and corrected for non-linear summation using the following conversion: EPP /1-(0.8 * EPP /(V m -V rev )) 54 . For this correction we used experimentally determined values of V m (the resting membrane potential) and 0 mV for the V rev (reversal potential). For each hemidiaphragm, corrected EPP amplitudes from 5 endplates were averaged to obtain a single value.
Naïve recordings were performed after 30 minutes of equilibration post-dissection. Hemidiaphragms were intoxicated by bath addition of 670 pM of BoNT/A, or 210 pM BoNT/E coupled with suprathreshold phrenic nerve stimulation at 0.5 Hz for 30 min, followed by persistent 0.05 Hz stimulation. These concentrations and conditions were sufficient to produce rapid and comprehensive synapse failure within 3-4 hours. Comprehensive intoxication was determined to be the point where consistent stimulation failure was observed in six stimulations (0.05 Hz) per endplate, observed across 5 consecutive endplates. To characterize the effects of 3,4-DAP and GV-58 on endplate function, the hemidiaphragm preparations were incubated with 2 µM of 3,4-DAP and/or 50 µM GV-58 in oxygenated Tyrode's for 45 minutes prior to recordings.
For all electrophysiology experiments, current measurements were filtered online at 2.9 kHz and digitized at 10 kHz. Data analysis and graphing were performed in Prism v6.1 (GraphPad software, La Jolla, CA).

Statistical analyses.
For each patched neuron, evoked amplitudes were averaged among the first three consecutive stimulations to produce the mean evoked current per neuron (in pA). Mean evoked amplitude values were then averaged among neurons within each condition and presented as percentages of age-and lot-matched control neurons. mEPSC frequencies were calculated by determining the mean frequency (in Hz) of events measured during 180 s of continuous voltage-clamp recording. Spontaneous release rates measured in BoNT-intoxicated neurons were normalized to mean frequencies observed in age-and lot-matched, vehicle-treated neurons and presented as percentages of control values. Statistical significance among means was calculated either using Student's t-test analysis (for two-sample comparisons), 1-way ANOVA testing (significance from control determined using the Dunnett's test against the vehicle control population), or 1-way ANOVA testing with Bonferroni's multiple comparison test. Quantitative data are presented as mean ± the standard error of the mean, with statistical significance listed, or represented using the following indicators: * indicates p < 0.05; ** indicates p < 0.01; *** indicates p < 0.001. Data availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information files).