Intracellular protein transport routes can be studied using toxins that exploit these to enter cells. BoNTA (botulinum neurotoxin type A) is a protease that binds to peripheral nerve terminals, becomes endocytosed and causes prolonged blockade of transmitter release by cleaving SNAP-25 (synaptosome-associated protein of 25 kDa). Retrograde transport of the toxin has been suggested, but not of the transient muscle relaxant, BoNTE (botulinum neurotoxin type E). In the present study, dispersal of these proteases in compartmented cultures of rat sympathetic neurons was examined after focal application of BoNTA or BoNTE to neurites. A majority of cleaved SNAP-25 was seen locally, but some appeared along neurites and accumulated in the soma over several weeks. BoNTE yielded less cleaved SNAP-25 at distal sites due to shorter-lived enzymic activity. Neurite transection prevented movement of BoNTA. The BoNTA protease could be detected only in the supernatants of neurites or cell body lysates, hence these proteases must move along neuronal processes in the axoplasm or are reversibly associated with membranes. Substitution into BoNTE of the BoNTA acceptor-binding domain did not alter its potency or mobility. Spontaneous or evoked transmission to cell bodies were not inhibited by retrogradely migrated BoNTA except with high doses, concurring with the lack of evidence for a direct central action when used clinically.
- botulinum neurotoxin
- intracellular protein transport
- soluble N-ethylmaleimide-sensitive factor-attachment protein receptor (SNARE)
- synaptosome-associated protein of 25 kDa (SNAP-25)
Elucidation of the largely unknown processes for the intracellular trafficking and localization of proteins has been aided by bacterial and plant di-chain toxins that bind susceptible cells via one of their polypeptides and deliver inside an associated enzymatic chain . In this regard, intense interest has been kindled in BoNTA (botulinum neurotoxin type A) because its targeting and internalization into peripheral nerves culminates in proteolytic inactivation of a protein essential for neuroexocytosis, leading to potent and selective inhibition of acetylcholine release. Moreover, this array of unique properties has been exploited with astounding success for relaxing overactive muscles (e.g. in dystonias, spasticity and overactive bladder) and normalizing secretory disorders (e.g. hyperhydrosis and sialorrhoea) for 3–12 months after a single treatment . Seven serotypes of BoNT (A–G) from Clostridium botulinum contain a 100 kDa HC (heavy chain) linked by a disulfide and non-covalent bonds to a 50 kDa LC (light chain). Their HCs mediate binding to neuronal ecto-acceptors and internalization, whereas the metalloprotease activities of the LCs inactivate SNARE (soluble N-ethylmaleimide-sensitive factor-attachment protein receptor) proteins essential for synaptic vesicle fusion . Another advantage of BoNTs as probes for investigating intracellular protein trafficking is that their intriguing uptake route involves translocation of the inhibitory LC across the limiting membrane to the cytosolic site of action via a channel created by HN (N-terminal half of the HC) .
Research on BoNTA is increasing due to its numerous successful clinical applications noted above, particularly its remarkable ability to cause prolonged neuroparalysis. It deletes nine C-terminal residues from SNAP-25 (synaptosome-associated protein of 25 kDa). Another serotype, E, cleaves the same SNARE target further from the C-terminus, removing 26 amino acids . Despite evidence that BoNTE (botulinum neurotoxin type E) disables SNAP-25 more effectively than BoNTA in vitro [5–7], it is rarely used clinically due its neuroparalysis recovering rapidly . Transfection of neuroendocrine cells with a gene fragment for LC of BoNTA fused to that of a fluorescent protein provided evidence that it associates with the plasmalemma via, in part, its unique dileucine motif  and its N-terminus binding to SNAP-25 , although susceptibility of BoNTE to ubiquitination has been reported recently to contribute to a faster degradation than BoNTA . Concurrently, the dileucine motif in BoNTA was shown to underlie its extraordinarily long duration of action because substituting both of the leucine residues dramatically shortened the persistence of neuromuscular paralysis in vivo . However, selective and complete retention of exogenously applied BoNTs at the site of uptake into neurons has not been established. Although the majority of 125I-labelled BoNTA localized within a few microns of the non-myelinated synaptic terminals after intramuscular injection, a trace of radiolabel could be detected inside the myelinated nerve trunk  and spinal cord [14,15]; nevertheless, transport of disconjugated iodine or BoNT-degradation products were not excluded. Later, cleaved SNAP-25 was visualized distant from neuronal sites of uptake and attributed to long-distance axonal transport of BoNTA  inside vesicles . It has been speculated that transfer of BoNTA in this way may reach the CNS (central nervous system) and cause central effects in peripherally injected patients [17–19], although indications of this have never been observed in humans [20,21]. Instead, alteration by BoNTA of neuron excitability or synaptic transmission centrally is usually attributed to indirect consequences of its peripheral action [22–24].
In the present study, the intraneuronal distribution of BoNTA (as well as BoNTE) and possible distal effects on synaptic transmission, together with features that influence these, were investigated in compartmented cultures of rat SCGNs (superior cervical ganglion neurons). Following application of BoNTA or BoNTE to neurites, predominantly slow migration of their proteases throughout the axoplasm to the soma was observed, in broadly equivalent proportions, an outcome not influenced by the acceptor-binding domain exploited to enter the cells. Thus the longer persistence and wider distribution of BoNTA-cleaved SNAP-25 in neurons seem to be largely a consequence of greater stability of BoNTA protease [11,12] compared with BoNTE, rather than differences in intracellular mobility. Surprisingly, electrophysiological recordings in SCGNs could not uncover evidence for inhibition of synaptic transmission to cell bodies by BoNT migrating to the soma after being applied distally to neurites at high concentration (104 pM). This is indicative of insufficient protease reaching the somatic area or crossing presynaptically to inhibit exocytosis. Indeed, partial inhibition resulted from distal application of >105 pM BoNTA, a clinically irrelevant concentration (see the Discussion).
Natural BoNTs were purchased from Metabiologics; specific neurotoxicities in mice [×108 MLD50 (median lethal dose) per mg] were determined by the supplier; BoNTA (2.5) and BoNTE (0.6, after nicking). Recombinant BoNT chimaera EA and a model substrate for BoNT protease, a fusion of GFP (green fluorescent protein) and 73 C-terminal residues of SNAP-25, have been described previously . QX-314, an Na+-channel blocker, was supplied by Tocris; buffers, salts and tissue culture reagents were bought from Sigma.
Isolation of neurons, maintenance in culture and exposure to BoNTs or antibodies
Isolation of superior cervical ganglia from newborn rat pups, enzymatic dissociation of the neurons and their maintenance in culture were as described in . Neurons were seeded into compartmented cultures following published procedures , using chambers supplied by the Tyler Research Corporation. BoNTs, or an antibody against the p75 neurotrophin receptor (IgG192) conjugated to Cy3 (indocarbocyanine) (ATS Bio), were added into the growth medium, as detailed in the Figure legends.
Harvesting of cellular material and analysis by SDS/PAGE and Western blotting
Neuron cell bodies and/or neurites in their respective compartments were dissolved in sample buffer and heated to 80°C for 5 min) before SDS/PAGE (12% NuPAGE gels, Invitrogen), transfer on to PVDF membrane and blotting with a rabbit polyclonal antibody specific for residues 9–29 of SNAP-25 (Sigma). Bound IgGs were detected with an alkaline phosphatase conjugate of a species-reactive secondary antibody and visualized by development of a coloured product which was quantified . The amounts of intact and cleaved SNAP-25 were calculated from the requisite scanned signals as a percentage of the total SNAP-25 (i.e. the sum of signals for intact and cleaved SNAP-25). In some experiments, neurites were removed by hypotonic shock with deionized water, and their membranes were concentrated by sedimentation at 288000 g for 15 min, before dissolution in SDS sample buffer and processing, as above. In some experiments, datasets were compared by two-way ANOVA with Bonferroni post-tests, which were performed using GraphPad Prism version 4 for Windows.
Cell fractionation and in vitro assay of BoNT protease activity
Neuron cell bodies and/or neurites from their respective compartments were lysed using deionized water in the presence of protease inhibitors (Sigma–Aldrich) before sedimenting the membranes, as above. The pellet was dissolved in deionized water containing 1% (v/v) Triton X-100; the same final concentration of detergent was also added to the supernatant from a 10× concentrated stock. Aliquots of 25 μl were removed and probed for SNAREs by Western blotting, as described above. The remaining samples were incubated with a model substrate  for 30 min at 37°C before stopping the reaction with SDS sample buffer and heating to 80°C for 5 min. Proteolysis of this substrate was assessed by SDS/PAGE followed by Western blotting with an antibody that preferentially binds to BoNTA-cleaved SNAP-25 . For negative and positive controls, the substrate was incubated as above in the absence or presence of 100 nM BoNTA that had been pre-treated with 5 mM dithiothreitol.
Images were captured and tiled to create a multifield view using a 10× air objective on an Olympus IX71 inverted microscope, fitted with a digital camera and a computer-controlled motorized stage (Märzhäuser), operating under the command of ImagePro Plus software (Media Cybernetics).
Experiments were performed with >4-week-old cultures of SCGNs. After careful removal of the Campenot frame, each culture dish was mounted on the stage of a light microscope (Olympus BX51WI). Whole-cell recordings of spontaneous and evoked synaptic activity were obtained from individual neurons at 36–37°C under continuous perfusion with oxygenated ACSF (artificial cerebrospinal fluid) (125 mM NaCl, 2.5 mM KCl, 1.25 mM NaH2PO4, 25 mM NaHCO3, 3.5 mM CaCl2, 1.2 mM MgCl2 and 25 mM glucose, pH 7.3) at a rate of ~2–3 ml/min. Data were collected using an EPC10USB amplifier controlled with PatchMaster software (HEKA) at −75 mV holding potential (voltage-clamp experiments) or from ~70 mV (current-clamp). Recording electrodes with 3–5 MΩ in-bath input resistance were fabricated from borosilicate glass with a P-97 puller (Sutter Instruments) and were filled with a potassium methyl sulfate-based internal solution (140 mM KCH3O3S, 5 mM KCl, 5 mM NaCl, 2 mM MgATP, 0.01 mM EGTA and 10 mM Hepes, pH 7.3). QX-314 was routinely added to the internal solution (5 mM final concentration) to block the autaptic synaptic transmission caused by antidromic/direct activation by the stimulation of clamped neurons, while eliciting eEPSPs (evoked excitatory postsynaptic potentials). Current pulses (100 μs/<100 μA at 0.05–0.1 Hz frequency) generated with a stimulus isolator (A-365, WPI) under computer control (PatchMaster) were delivered by a bipolar electrode (tip diameter ~10 μm) made from theta-tubing (TST150-6, 1.5 mm outer diameter, WPI) and filled with ACSF, which was placed in close proximity to a cluster of SCGNs before patching a visually identified SCGN.
After establishing whole-cell configuration in voltage-clamp mode (at −75 mV with correction for the liquid junction potential), Cm (membrane capacitance) was cancelled and series resistance compensated (~80%); readouts of Cm, input resistance (Rinput) and holding current (Ihold) values were tabulated for offline analysis. Only neurons with membrane resistance exceeding 100 MΩ were used for analysis. sEPSCs (spontaneous excitatory postsynaptic currents) were collected in continuous mode; eEPSPs were quantified with an episodic recording regime after switching the amplifier to current-clamp set-up. Stimulus intensity in control cultures was adjusted to reliably induce evoked synaptic responses (<10% failure rate). In toxin-treated cultures, if no evoked response was detectable at a comparable stimulus intensity as for controls, the pulse strength was increased up to 500% of the threshold stimulus (up to 0.5 mA) for non-intoxicated controls, to confirm a lack of neurotransmission in a given neuron.
Acquisition and analysis of electrophysiological data
Analogue signals were digitized at 20 kHz and analysed offline (FitMaster, HEKA; IgorPro, Wavemetrics). sEPSC (exceeding three times the amplitude of the noise band) frequency was estimated as a reciprocal of the interevent interval, measured as the time between the peaks of two consecutive sEPSCs. The kinetics of sEPSCs were quantified on the basis of the peak value of the first-order derivative of sEPSC waveform (dA/dt), whereas values for decay (single-exponential fitting) were taken as measures of sEPSC decay time constant. The difference between the base and the peak of synaptic responses represented the amplitudes. The PPR (paired-pulse response ratio) was determined from the second (P2) and first (P1) eEPSP, evoked by paired-pulse stimulation (interpulse interval 40 ms), using the formula PPR=P2/P1×100. Data are reported as means±S.E.M. and statistical significance was assessed using Student's t test, with P<0.05 defining a significant difference.
BoNTA and BoNTE applied to the neurites of SCGNs act locally at clinically relevant doses with little or no protease activity seen in the cell bodies unless higher doses are applied
For monitoring the diffusion of BoNTA and BoNTE proteases inside neurons, rat SCGNs were grown in compartmented cultures . This system permitted restricted application of toxin to the neurites and allowed individual harvesting of these components and cell bodies for biochemical analyses. After seeding newly isolated SCGNs in a water-tight compartment, created by attaching a Teflon divider to culture dishes with silicon grease, they elaborated processes that grew under the barrier into the adjacent chamber(s) (Figure 1A), as visualized by their uptake of a dye-conjugated antibody against the p75 neurotrophin receptor. Movement of BoNTA was examined in a three-compartment system (Figure 1B). SCGNs were seeded in the central chamber (Cr) and, after 7 days, neurites that had grown into peripheral chambers (P) were exposed to BoNTA for 24 h. Each compartment was harvested with SDS buffer and proteins were separated by electrophoresis before performing Western blotting, using an antibody that recognizes intact SNAP-25 and the larger N-terminal product of its cleavage by the BoNTA protease. Addition of 10 pM BoNTA to each peripheral chamber (Figure 1B; 300 pg in each peripheral chamber containing 0.2 ml of growth medium) resulted in the proteolysis of approximately one-third of the SNAP-25 in the neurites (Figure 1C), but with virtually no cleavage being found in the cell bodies in the central chamber (proximal portions of processes were also present). Note that this amount of BoNTA is equivalent to 75 MLD50 units (see the Experimental section) and exceeds the maximum recommended clinical dose (50 units) for a single-site injection of BoNTA complex . Increasing the amount added to each peripheral chamber revealed dose-dependent SNAP-25 proteolysis in the neurites that saturated at ~104 pM with ~60% of the SNAP-25 having been cleaved. Although the half-maximal effect in the neurites (EC50; in this case 30% truncation) was 6 pM (Figure 1C), no cleaved SNAP-25 was detected in cell bodies unless the neurites were exposed to 100 pM (750 units) or more BoNTA, much in excess of a clinical dose, with a large rightwards shift in the dose–response relationship to an EC50 of 104 pM. Assuming that the concentration-dependence for intracellular cleavage is the same in neurites and cell bodies, the EC50 values suggest that less than 0.1% of the protease taken up by neurites reached cell bodies within 24 h (direct evidence that the protease moves is presented below).
When neurites were exposed to BoNTE, proteolysis of the SNAP-25 in these processes also occurred (Figure 1D), but cleavage of 30% of the SNAP-25 required 300 pM; up to 95% became truncated with 3×105 pM, notably higher than the maximum for BoNTA. Again, much less cleavage product was detected in cell bodies. Comparing the value for 30% SNAP-25 cleavage by BoNTE in cell bodies (8×104 pM) and neurites (300 pM) indicates that >0.3% of the protease reached cell bodies in 24 h. Thus, even though BoNTE truncates SNAP-25 in neurites that are inaccessible to BoNTA, much greater amounts had to be applied to neurites to obtain equivalent cleavage of SNAP-25 in the cell bodies. These apparently contradictory characteristics arise from the distinct dose–response profiles for BoNTA and BoNTE; larger quantities of the latter can ultimately proteolyse a much greater fraction of the total SNAP-25, but BoNTA is more effective at low concentrations. The basis for these distinct profiles is not known.
Migration of BoNTA protease from neurites to cell bodies occurs slowly over several days
Dose–response plots obtained 1 or 2 weeks after exposure to toxin unveiled a change in the protease location. At day 7, after the transient (24 h) incubation of neurites with BoNTA, there was still more cleaved SNAP-25 in the neuronal processes than in the cell bodies (Figure 2A), but the difference was much reduced compared with day 1 (cf. Figure 1C). Two-way ANOVA revealed that this increase in the amounts of cleaved SNAP-25 in the cell bodies was extremely significant (P<0.0001), with Bonferroni post-test analysis indicating that significant increases in the level of cleaved SNAP-25 occurred only in cells exposed to high concentrations (≥102 pM) of BoNTA. In contrast, there was no significant change in the amounts of cleaved SNAP-25 in neurites (P>0.05). By the second week (Figure 2B), the protease activity in neurites and cell bodies had become similar. Seemingly, the BoNTA protease had migrated slowly from neurites to the soma over 2 weeks; the amount of cleaved SNAP-25 in cell bodies 14 days after exposing neurites to BoNTA was very significantly greater than at 24 h (P<0.0001), whereas there was still no significant change in the neurites. Interestingly, ~40% of the SNAP-25 in neurites and ~25% in cell bodies remained uncleaved by BoNTA at day 14. A different pattern was observed with BoNTE. On day 3 after the transient exposure, the extent of SNAP-25 cleavage in both neurites and cell bodies had decreased, compared with the level 24 h after exposure (Figure 2C, cf. Figure 1D), consistent with a general reduction in protease activity due to its known shorter half-life in neurons ; such instability hindered reliable measurement of its redistribution inside the cells.
Peripherally applied BoNTA protease reaches cell bodies, persists there after neuritotomy and diffuses back into regenerated neurites
An alternative explanation for the appearance of truncated SNAP-25 in cell bodies of neurons exposed peripherally to large doses of BoNT could be migration of the cleaved protein. Therefore to ascertain whether the proteases themselves accumulate in cell bodies, advantage was taken of the fact that hypotonic shock removes neurites from SCGNs, leaving the cells to survive and regenerate processes . Neurites were exposed to 104 pM BoNTA per peripheral chamber for 24 h (a large dose known to produce cleaved SNAP-25 in the cell bodies) before washing the peripheral chambers with deionized water to remove the processes. These washes were collected and membranes were concentrated by sedimentation to facilitate analysis of SNAP-25 in the neurites; as expected, ~60% of the SNARE therein was cleaved (Figure 3A1). Although this treatment removed all of the neurites from the peripheral chambers, they regenerated after the washed compartments were refilled with medium and returned to normal growth conditions; in fact, 7 days after neuritotomy, robust regrowth of neurites had occurred into both peripheral chambers. Analysis of the regenerated neurites and cell bodies revealed the presence of cleaved SNAP-25 in both (Figure 3A2). Thus BoNTA protease was not removed from the neurons by simply washing off the neurites that had acquired it; some had obviously reached the cell bodies and subsequently migrated back into the new processes.
BoNTA-cleaved, but very little BoNTE-truncated, SNAP-25 appears in contralateral neurites after unilateral application of BoNTs: minimal movement seen with a clinically relevant dose
Importantly, when SCGNs were exposed unilaterally to 10 pM BoNTA (300 pg in 0.2 ml, a lower and more clinically relevant amount), no cleaved SNAP-25 was detected in the contralateral neurites or cell bodies 6 days later (Figure 3B). Clearly, at this low dose, little, if any, of the protease moved out of the exposed neurites for at least 6 days. In corroboration, when neurites were detached after 24 h of exposure to this amount of BoNTA, no cleaved SNAP-25 was found in the cell bodies or regenerated neurites on either side (Figure 3C), indicating that the protease had been removed with the original neurites. However, earlier experiments had implied that BoNTA (applied in large amounts) migrates both retro- and antero-gradely, raising the possibility that detectable enzyme may move through the cell bodies into processes that project towards the opposite distal chamber. To test this conjecture, 104 pM BoNTA was added to only one side (hatched) of a three-chamber dish before washing off the neurites 24 h later and measuring the amount of cleaved SNAP-25 (Figure 3D1) on the exposed (PI) and contralateral side (PC). The presence of BoNTA-product (black bars) on both sides confirmed that this protease not only reached the cell bodies, but also passed through to neurites on the other side. Detectable cleavage of SNAP-25 (~20%) in neurites could be produced by as little as 1 pM BoNTA (Figure 1C), i.e. by the transfer of as little as 0.01% of the BoNT applied unilaterally (Figure 3D1). When cells were allowed to regenerate processes for 7 days, similar levels of SNAP-25 proteolysis were observed in each peripheral chamber (Figure 3D2). In contrast, only traces of product were detected in the contralateral neurites removed from cells exposed to 105 pM BoNTE (Figure 3D1, white bars) and, as expected, very little was present 6 days later in the cell bodies or regenerated neurites on either side (Figure 3D2), consistent with the known faster decay of BoNTE activity compared with BoNTA [11,12,28]. If the neurites were not removed, cleavage of SNAP-25 increased progressively in the contralateral neurites of cells that had been exposed unilaterally for 24 h to 104 pM BoNTA, but not those treated with 105 pM BoNTE, as determined at either 2 (Figure 3E) or 6 (Figure 3F) days later. Collectively, these findings show that BoNTA, but little BoNTE, protease moves retrogradely into cell bodies and then anterogradely into both growing and mature processes, although detectable migration required the application of large concentrations.
BoNTA and BoNTE proteases traffic through the axoplasm of SCGNs
By further compartmentation in larger five-chamber dishes (Figure 4A), the cell bodies (seeded in the central chamber) were separated by intermediate neurite domains [in medial chambers (M)] from the distal growing tips (in peripheral chambers). Notably, although cleavage of SNAP-25 in distal tips reached a plateau when ~103 pM BoNTA was applied in the peripheral chambers, the proportion truncated in medial neurites and cell bodies increased significantly upon raising the dose to 105 pM (Figure 4B). This indicates that incomplete proteolysis of SNAP-25 is not due to saturation of the cellular machinery exploited by BoNTs to enter the cell. It is also noteworthy that cleaved SNAP-25 occurred all along neurites and the extent proteolysed decreased progressively with distance from the site of toxin application. As SNAP-25 resides on the cytosolic surface of the plasma membrane and synaptic vesicles, active BoNT protease would need to have migrated through the cytosol rather than inside transport organelles. The protease must travel through neurites because their transection in the medial chambers prevented the appearance of all but a trace of cleaved SNAP-25 in the central chamber (Figure 4C), despite application of a very high concentration of BoNTA (105 pM) in peripheral chambers (note that the neurites in peripheral chambers, segregated from their cell bodies, degenerated after transection). Its movement through the axoplasm was confirmed by subcellular fractionation. On day 7 after 24 h of exposure to BoNTA, the neurites and cell bodies were harvested by hypotonic shock and the membranes were sedimented by centrifugation. Western blotting for SNAREs showed that all were in the pellet, as expected (Figure 4D). The resultant fractions were assayed for BoNTA protease in vitro, using a model substrate. Cleaved product was detected by Western blotting (Figure 4E) with an antibody that selectively binds BoNTA-cleaved SNAP-25 . Notably, BoNTA protease activity appeared only in the soluble cell lysates and occurred all along the neurites as well as in cell bodies; it was more prevalent in cells exposed to higher concentrations of BoNTA (a longer exposure was required to detect the product in cells exposed to less toxin).
As expected, more BoNTE, relative to BoNTA, was needed to produce cleavage of SNAP-25 in neurite tips (Figure 4F). Even so, the appearance of protease in medial neurites and cell bodies could be detected with high concentrations of BoNTE, >104 pM. As with BoNTA, the fraction of SNAP-25 cleaved diminished progressively with distance from the neurite tips. This pattern is in accordance with a small proportion of BoNTE protease migrating through the cytosol from neurite tips towards the cell bodies.
Movement of BoNTE protease is not influenced by ecto-acceptor selectivity
Although BoNTA and BoNTE both bind to SV2 (synaptic vesicle protein 2), they interact with different isoforms (SV2C preferentially for BoNTA, and glycosylated forms of SV2A and SV2B for BoNTE [30,31]). To determine whether acceptor-selectivity influences the intracellular distribution of the BoNTE protease in compartmented sympathetic neurons, the experiments were repeated with an EA chimaera of BoNTA and BoNTE that has the acceptor-binding HC (C-terminal half of the HC) domain of BoNTA fused to the HN-LC of BoNTE, and shown to bind to the BoNTA acceptor [7,25]. Despite this change, the susceptibility of SCGNs to EA (Figure 4G) remained similar to the pattern observed for BoNTE (Figure 4F): although EA proved slightly more potent, and the proportion of BoNTE protease migrating into medial neurites and cell bodies was not altered. Thus the delivery of BoNTE protease into cells, and its intracellular mobility therein, are not influenced by changing the toxin's acceptor-binding domain.
Globally applied BoNTA or BoNTE virtually abolish synaptic transmission between SCGNs with no effect on passive membrane properties
For examining the functional consequences (if any) for synaptic transmission at cell bodies arising from the movement of BoNTs through neurites, the ability of cultured SCGNs to form synapses with neighbouring cells was exploited . As synaptic activity increases as cultures mature, SCGNs were grown for 4–5 weeks before carefully dismantling the Campenot chambers such that the cell bodies (with their proximal neurites intact) remained attached to the culture dish, but became accessible for electrophysiological recording (Figure 5). Notably, during maturation in Campenot chambers, SCGN cell bodies migrate together to form clusters (e.g. as in Figure 1A). Thus any cell patched would be surrounded by many others. sEPSCs were recorded in 14 of 15 SCGNs examined (93.3%; Figure 5A1) under continuous voltage clamp at a holding potential of −75 mV. These events manifested rapid onset (tpeak=1.10±0.03 ms; dA/dt=133.4±7.02 nA/s) with a high decay time constant (Figures 5B and 5C; τ=3.7±0.07 ms). The frequency of sEPSCs varied broadly between cells [Figure 5D; mean value=1.3±0.25 Hz; variation coefficient (CV)=0.72; n=15). Additionally, eEPSPs induced by field stimulation of SCGN clusters were recorded at membrane potentials between −70 and −75 mV (Figure 5E). Interference from antidromic activation of individual neurons by field stimulation was excluded by the addition of a membrane-impermeant inhibitor of voltage-activated Na+ channels (QX-314) in the internal solution. eEPSPs were recorded in 11 of 12 (91.6%) control SCGNs, with an average amplitude of 10.9±1.18 mV for responders (Figure 5F). Consistent with previous studies , eEPSPs of lower amplitude (8.6±1.4 mV) were elicited by a second stimulus delivered 40 ms later, representing a paired-pulse depression of 21.9% (Figures 5G and 5H).
To ascertain whether synaptic transmission between SCGNs is susceptible to blockade by BoNTs, the neurons were exposed globally (i.e. in all chambers) for 24 h to 104 pM BoNTA or 105 pM BoNTE in each chamber before recording sEPSCs (Figures 5A2 and 5A3) and eEPSPs (see below). After treatment with BoNTA, fewer cells exhibited sEPSCs (three of seven; 42.8%) relative to non-treated controls, with the frequency in the three responders (Figure 5D) reduced dramatically to 0.01±0.007 Hz (CV=1.5, P=0.004). Small, but significant, changes in sEPSC kinetics were also detected (tpeak=1.3±0.17 ms; dA/dt=119±5.7 nA/s; τ=4.2±0.2 ms; n=3) compared with non-treated controls (P=0.042; P=0.025; P=0.046). Global exposure to BoNTE for 24 h produced an even stronger block of sEPSCs, which were only detected in one of nine SCGNs examined (10.1%) and with very low frequency (Figure 5A3) relative to control (0.002±0.002 Hz; CV=3; P=0.001).
Likewise, the proportion of SCGNs showing eEPSPs was reduced to only 14.3% (two of 14) by the treatment with BoNTA, and the mean EPSP amplitude in responders (1.9±0.3 mV) was much decreased relative to control values (Figure 5F; P<0.0001). Interestingly, the amplitude of a second eEPSP elicited by paired-pulse stimulation (Figures 5G and 5H) showed a small enhancement relative to the first, unlike the paired-pulse depression seen in control cells, perhaps due to a reduction in release probability  that can be overcome by increased intracellular [Ca2+] . Increasing the stimulus intensity did not alter the proportion of cells exhibiting eEPSP responses (results not shown). Even more striking, eEPSPs were not detected in any of the 14 neurons that had been globally exposed to BoNTE (Figure 5F). Importantly, passive membrane properties (Table 1) were not significantly different between non-treated control and SCGNs exposed to either BoNTA or BoNTE. Clearly, intact SNAP-25 is required for spontaneous and evoked synaptic transmission between SCGNs, thereby implicating SNAP-25-dependent exocytosis. The more extensive inhibition by BoNTE than BoNTA may be attributed to the use of a 10-fold higher concentration, cleavage of a larger proportion of the SNAP-25 (even though BoNTA was used at a supersaturating concentration), removal of a bigger peptide fragment from the C-terminus  or a combination thereof.
Exposing SCGN distal neurites to 104 pM BoNTA failed to block synaptic transmission at their cell bodies and 105 pM gave only partial inhibition
Having established that the synaptic inputs to cell bodies of SCGNs are susceptible to BoNTA, attention turned to cells exposed to this protein via their neurites only, specifically addressing whether a sufficient amount of BoNTA protease can move from the neurites to produce a functional blockade at distant synaptic sites. Given that the protease requires several days to migrate through neurites and accumulate in the cell body compartment (Figures 2 and 3), SCGN neurites were exposed transiently (24 h) to BoNTA (104 pM in each peripheral chamber) and the neurons were maintained for a further 7 days before electrophysiological examination of synaptic transmission. Notably, in SCGNs treated distally with this concentration of BoNTA, neither the frequency (1.12±0.39 Hz; CV=1.1; P=0.68, n=10; Figures 5A4 and 5D) nor kinetic parameters (tpeak=1.13±0.01 ms; P=0.73; dA/dt=141.8±8.2 nA/s; P=0.44; τ=3.82±0.1 ms; P=0.41; Figure 5C) of sEPSCs differed significantly from their respective values in control cells. Furthermore, eEPSPs were detected in 88.2% (n=17) of SCGNs 7 days after exposure of their neurites to this high dose of BoNTA in the peripheral chambers, with an average amplitude (9.6±0.85 mV) not significantly different from values recorded in untreated control cells (Figure 5F; P=0.44). These neurons also showed paired-pulse depression (19%; amplitude of second eEPSP=7.8±0.98 mV) similar to the extent in control cells (Figure 5H). Thus, despite biochemical evidence that 1 week after exposure to BoNTA a fraction of its protease had migrated along the axons towards their cell bodies, electrophysiological recordings showed no significant functional impact on synaptic transmission between cells. It therefore appears that BoNTA does not reach the sites of transmitter release in sufficient amounts even using 104 pM, a concentration extrapolated to greatly exceed the doses used clinically (see below). Only in cells exposed distally to even higher concentrations was any significant reduction in eEPSPs observed. In cultures treated with 105 pM BoNTA (Figure 5F), eEPSPs were observed only in four of seven neurons examined with a mean amplitude (3.25±0.49 mV; P<0.05 with respect to non-treated control). Also, the latter cells showed a paired-pulse enhancement (8.7±2.8%) of eEPSPs like those observed in cells exposed globally to this toxin. Furthermore, only one of five cells treated with 3×105 pM BoNTA yielded any eEPSPs, with amplitude only 1.8 mV (results not shown).
In view of widespread clinical uses of BoNTA in treating neuromuscular overactivity, autonomic disorders and chronic pain , there are pressing needs to establish the extent of its putative intra- and inter-neuronal movement, particularly considering the possible functional consequences. It has been suggested that differences in the subcellular distribution of BoNTA and BoNTE proteases may underlie their distinct lifetimes inside cells . Also, there are reports of long-distance trafficking of BoNTA inside nerves, but not BoNTE, leading to changes in nerve activity at distal sites [16,19]. Hence the present study examined the fate of these enzymes in SCGNs, chosen as a model of peripheral autonomic neurons because they innervate secretory glands in vivo , can be readily grown in vitro with long neurite processes elaborated in compartmented cultures and are amenable to electrophysiological recording of synaptic transmission. They are also commonly used for studies of slow and fast axonal transport [37,38].
To reveal the distribution of their proteases inside neurons after focal application of BoNTs to neurites, or even just their distal tips, the presence of both cleaved and intact SNAP-25 was measured. This is a significant improvement over immunoassay of the cleaved product only [16,19] because it gives a measurement of the fraction of total SNAP-25 cleaved and thus facilitated a semi-quantitative determination of the concentration-dependence for protease activity. Although specific immunodetection of cleaved SNAP-25 is undoubtedly an extremely sensitive method for demonstrating the presence of minute traces of BoNTA protease activity, it can mask an overwhelming background of uncleaved and, presumably, fully functional SNAP-25.
High concentrations are required to unveil intra-axonal migration of BoNTA and BoNTE proteases
Advantageously, the dose–response relationships described in the present paper reflect the relative amounts of protease in different regions of the BoNT-treated neurons. Such analyses indicated that most of the BoNTA and BoNTE enzymes remained close to the site of uptake, but a small fraction moved retrogradely along processes. Indeed, within 24 h of addition to neurites, only 0.3% of the applied BoNTE had reached the somatic compartment and <0.1% of the BoNTA. It must be stressed that even when the clinical dose (50 MLD50 units) of BoNTA was exceeded, by adding 10 pM (75 MLD50 units) into each peripheral chamber, virtually no cleaved SNAP-25 appeared in cell bodies within 24 h (Figure 1C); in the case of BoNTE, much higher concentrations (103 pM) could be added to neurites without product appearing in the soma. Even when higher amounts of BoNTA were applied, accumulation of cleaved SNAP-25 in cell bodies took days to weeks (Figure 2). The latter can be attributed to the movement of BoNTA protease through the neurites because (i) neurite transection almost completely prevented its appearance there, and (ii) the protease activity could be released from cell bodies and detected by an in vitro assay 7 days after transient application of BoNTA to the neurite extremities. Rapid deterioration of the BoNTE protease prevented a build-up of its product (Figure 2), probably due to its ubiquitination and targeting to the proteasome, a process largely avoided by BoNTA . Thus loss of activity precluded the continuous accumulation of BoNTE-cleaved SNAP-25 at distal sites rather than any lower axoplasmic mobility than BoNTA, as proposed previously .
Migrant protease detected in the supernatant of lysates is not tightly associated with membranes
BoNTA protease has been reported to bind both intact and BoNTA-cleaved SNAP-25 via its N-terminus . It also contains a dileucine motif  through which, theoretically, it could interact with transport vesicles via adaptor protein complex 2 . Nevertheless, subcellular fractionation herein revealed that the BoNTA protease was only detectable in the supernatant of the cell lysate from the neurites and the cell bodies, clearly indicating migration of the proteases through the axoplasm in active form. This notion is corroborated by the presence of membrane-associated BoNTA- and BoNTE-cleaved SNAP-25 in the cell membrane all along the neurites; although BoNTA LC must bind SNAP-25 during the proteolysis reaction, this interaction must be reversible as no detectable protease activity co-sedimented with its substrate in the membrane fraction. These observations are consistent with some BoNTA protease migrating retrogradely from sites of uptake [13,16], but not exclusively inside vesicles. Rather, BoNT captured within recycling synaptic vesicles after binding intraluminal acceptors [31,40] must translocate its LC across the endosome membrane  with subsequent breakage of the disulfide link to the HC (an essential step for activation of the BoNTA LC protease ) before beginning the journey towards the cell body. Accordingly, replacing the HC domain in BoNTE with that of BoNTA had minimal effect on the migration of the BoNTE protease, revealing that the different acceptors selected by BoNTA and BoNTE to enter neurons [30,31] do not influence the movement of active protease to cell bodies. Instead, this appears to be determined by the potency and stability of the LC after translocation out of the vesicles, consistent with the proposal that preferential proteolytic degradation underlies the shorter lifetime of BoNTE . Thus the mechanism for intracellular redistribution of BoNTs seems to be quite different from intravesicular trafficking, widely held to be the mechanism by which the distinct tetanus toxin is transported towards the cell soma [18,42,43]. This accords with the disparate sites of action for BoNTs and tetanus toxin, which cause flaccid and spastic paralysis respectively. By showing that cleaved SNAP-25 is produced along neuronal processes and at distal sites several millimetres away from the cell regions originally exposed in fluidically isolated chambers, and that this is due to the slow migration of the protease through the axoplasm, the findings preclude the proposal that cleaved SNAP-25 observed distal to exposure sites is entirely due to BoNTA being “loaded on to the retrograde transport machinery”  of “retrogradely transported organelles” . Although integral membrane proteins and luminal constituents mostly undergo fast axonal transport in specialized organelles, cytosolic proteins generally move slowly through the axoplasm [44,45]. The possibility cannot be excluded that some of the cleaved SNAP-25 is produced by BoNTA that had been transported along neurites inside vesicles, but, even if so, in the sympathetic model used in the present study this cannot amount to anything more than a minute fraction of the protease activity (below detection limits of the in vitro assay). Moreover, although it is also theoretically possible that vesicular transport of BoNTs is more prevalent in other neuron types or in vivo, this study highlights slow extravesicular axoplasmic migration as an alternative mechanism to explain the apparently long-range intraneuronal distribution of BoNTA protease activity that has been reported [16,19]. Importantly, movement of the protease away from neurites did not diminish significantly the extent of SNAP-25 cleaved in the exposed regions, even after 2 weeks (Figure 2). Thus migration of some portion of the protease along neuron processes is reconcilable with the retention of sufficient BoNTA enzyme at nerve terminals for a prolonged duration of action, as observed clinically.
Peripherally applied BoNTA failed to block neurotransmission at cell bodies, except at concentrations far exceeding clinical doses
Accumulation of BoNTA protease in motor or sensory neuron cell bodies or its transfer into central projections has been put forward as a likely cause for minor central effects [18,46]. However, the present study using patch-clamp recordings found no evidence for BoNTA (up to 104 pM) inducing significant changes in the excitability of sympathetic neurons in vitro (Table 1) or for inhibiting neurotransmission at distally located synapses after being applied to neurites. This method measures neurotransmission directly, unlike extracellular recordings  which can be influenced by both intrinsic and extrinsic firing activity. Nevertheless, the lack of inhibition of synaptic transmission observed was somewhat surprising considering the presence of detectable cleaved SNAP-25 in the somatic compartment. A likely explanation is that BoNTA protease cleaved some SNAP-25 in the proximal neurites and soma, but had not reached neurotransmitter release sites in the processes synapsing on to cell bodies. It is clear that neurotransmission between cultured SCGNs is susceptible to BoNTA, as direct application in the central compartment resulted in near-complete inhibition of synaptic activity with extensive, albeit incomplete, cleavage of SNAP-25. Presumably, when applied to the cell bodies, the toxin is taken up by recycling SV2-containing vesicles near release sites in fibres surrounding the soma and is released locally into the cytosol where it blocks exocytosis. Thus it can be deduced that BoNTA protease (104 pM) migrating from distal neurites does not reach the release sites in sufficient amounts to inhibit neurotransmission and, also, indicates a lack of trans-synaptic transfer in this model system. Indeed, given that the migrant protease is extravesicular, it is unclear how trans-synaptic transfer could occur; certainly not by the mechanism proposed for tetanus toxin in which it must be released from retrogradely transported organelles . The lack of functional effect of migrant BoNTA is all the more noteworthy in view of the large concentration (104 pM) applied bilaterally to neurites (Figure 5); these cells were exposed to a total of 6×105 pg of this protein, equivalent to 150000 MLD50 units. Although such exposure of culture cells to BoNTA in vitro does not replicate conditions under which it is used clinically, it is, nevertheless, remarkable that the latter is ~3000-fold more than the recommended 50 unit maximum clinical dose for single-site injection of BOTOX® , yet still proved incapable of producing any functional effect that could be detected electrophysiologically. Only at even higher concentrations (≥105 pM) was any significant blockade of neurotransmission detected. Given the much greater distance from injection sites to neuronal cell bodies in human recipients compared with the few millimetres between compartments in the SCGN cultures, even if similar migration of BoNTA is replicated within the axons of patients, extrapolation from our model system of sympathetic neurons predicts that central accumulation of protease would be very slow, negligible and innocuous, especially with a clinical dose. Secondary responses to muscle paralysis are the most cogent explanation for central effects observed after peripheral injection of BoNTA, such as persistent alteration of cortical activity following injection of as little as 2.5 units of BoNTA into the extensor digitorum brevis . The notion that sufficient protease from such a minuscule dose could traffic through nerves from the toe to the spinal cord, transfer to corticospinal fibres and migrate further along axons into the brain to block neurotransmission for >12 weeks , similar to the time course of neuromuscular block, lacks experimental support. Ideally, clinical evaluation, if possible, would be desirable to confirm in vivo the negligible retrograde migration and lack of trans-synaptic transfer of BoNTA.
Gary Lawrence and Saak Ovsepian performed experimental procedures. Jiafu Wang manufactured the EA chimaera. Gary Lawrence, Saak Ovsepian, Jiafu Wang, Roger Aoki and Oliver Dolly contributed to the preparation of the paper.
This work was supported in part by a Principal Investigator award from Science Foundation Ireland (to J.O.D.) and under the Programme for Research in Third Level Institutions (PRTLI) Cycle 4. The PRTLI is co-funded through the European Regional Development Fund (EDRF), part of the European Union Structural Funds Programme 2007–2013.
Abbreviations: ACSF, artificial cerebrospinal fluid; BoNT, botulinum neurotoxin; BoNTA, BoNT type A; BoNTE, BoNT type E; CV, variation coefficient; Cy3, indocarbocyanine; eEPSP, evoked excitatory postsynaptic potential; HC, heavy chain; HC, C-terminal half of the HC; HN, N-terminal half of the HC; LC, light chain; MLD50, median lethal dose; PPR, paired-pulse response ratio; SCGN, superior cervical ganglion neuron; sEPSC, spontaneous excitatory postsynaptic current; SNAP-25, synaptosome-associated protein of 25 kDa; SNARE, soluble N-ethylmaleimide-sensitive factor-attachment protein receptor; SV2, synaptic vesicle protein 2
- © The Authors Journal compilation © 2012 Biochemical Society