Neuronal entry and high neurotoxicity of botulinum neurotoxin A require its N-terminal binding sub-domain

Abstract

Botulinum neurotoxins (BoNTs) are the most toxic proteins known, due to inhibiting the neuronal release of acetylcholine and causing flaccid paralysis. Most BoNT serotypes target neurons by binding to synaptic vesicle proteins and gangliosides via a C-terminal binding sub-domain (H_CC). However, the role of their conserved N-terminal sub-domain (H_CN) has not been established. Herein, we created a mutant form of recombinant BoNT/A lacking H_CN (rAΔH_CN) and showed that the lethality of this mutant is reduced 3.3 × 10⁴-fold compared to wild-type BoNT/A. Accordingly, low concentrations of rAΔH_CN failed to bind either synaptic vesicle protein 2C or neurons, unlike the high-affinity neuronal binding obtained with ¹²⁵I-BoNT/A (K_d = 0.46 nM). At a higher concentration, rAΔH_CN did bind to cultured sensory neurons and cluster on the surface, even after 24 h exposure. In contrast, BoNT/A became internalised and its light chain appeared associated with the plasmalemma, and partially co-localised with vesicle-associated membrane protein 2 in some vesicular compartments. We further found that a point mutation (W985L) within H_CN reduced the toxicity over 10-fold, while this mutant maintained the same level of binding to neurons as wild type BoNT/A, suggesting that H_CN makes additional contributions to productive internalization/translocation steps beyond binding to neurons.

Figure 1. Expressed, purified rAΔH_CN and rA showed similar proteolytic activity.

(a) Schematic of rA and rAΔH_CN. Nucleotides encoding I₈₇₄-Q₁₀₉₁were deleted and replaced with 6 nucleotides GGCGGT to generate rAΔH_CN. S-S denotes inter-chain disulphide, GG between H_N and H_CC in rAΔH_CN represent glygly residues. H₆ indicates 6xHis and (↑) represent consensus site for thrombin. (b) rAΔH_CN and rA expressed in E. coli, purified and nicked before being subjected to SDS-PAGE in the presence or absence of DTT, followed by Coomassie staining. Note that HC and H_N-H_CC were separated from LCs only under reducing conditions. (c) rAΔH_CN and rA exhibited similar protease activity towards a model GFP-SNAP-25C₇₃-His₆ substrate.

Figure 2. rAΔH_CN caused minimal cleavage of neuronal SNAP-25 and displayed greatly reduced lethality.

Rat cultured CGNs (a,b) or TGNs (c,d) were incubated with various concentrations of rA or rAΔH_CN in medium for 24 h before harvesting in LDS sample buffer. Samples were separated by SDS-PAGE followed by Western blotting with an antibody recognising both intact and cleaved SNAP-25. The arrow and arrowhead in panel a and c indicate intact and cleaved SNAP-25, respectively. Data are mean ± S.E.M, n = 3. (e) Removal of the H_CN sub-domain from rA dramatically reduced its lethality in mice. (f) DAS values were recorded over 28 days after injecting rA or rAΔH_CN into the right gastrocnemius muscle of mice; SEM values are shown from 7 mice for each toxin. Two-way ANOVA with Bonferroni post hoc test analysis highlights that there is no significant difference (p > 0.05 at all time points) between the curves for rA and a 33,400-fold higher quantity of rAΔH_CN.

Figure 3. Deletion of H_CN from rA diminished its binding to SV2C-L4 and intact neurons.

(a) Western blot from the in vitro pull-down assay (see Materials and Methods) shows that rAΔH_CN at 10 and 1 nM concentrations failed to bind non-glycosylated GST-SV2C-L4, in contrast to rA. At 100 nM, a lower amount of rAΔH_CN than rA was retained by immobilised SV2C-L4. Note that samples were reduced by DTT before SDS-PAGE. (b) CGNs were incubated with increasing concentrations of ¹²⁵I-rA alone (○) or with 1 μM BoTIM/A (•) for 1 h at 4 °C followed by three washes before γ counting of the pellets. Subtracting non-saturable binding (•) from the total binding (○) yielded the saturable component (□). Inset: Scatchard plot of the saturable binding of ¹²⁵I-BoNT/A to CGNs. (c) Binding of ¹²⁵I-labelled rAΔH_CN to CGNs, measured as in (b) was non-significant. Data are mean ± S.E.M from two independent experiments performed in duplicates.

Figure 4. Generation of probes for visualising the cellular location of rAΔH_CN and rA.

(a) Schematic of the BoNT/A probes generated. A short length of nucleotides encoding the HA tag was inserted before the thrombin cleavage site in the loop region of rA and rAΔH_CN to yield constructs encoding rA LC.HA-HC and rA LC.HA-H_NH_CC. For simplicity, the latter two were termed rA-HA and rAΔH_CN-HA, respectively. After purification, nicked rA-HA and rAΔH_CN-HA were subjected to SDS-PAGE in the presence or absence of DTT, followed by Coomassie staining (b) and Western blotting, using the indicated antibodies (c). rA was loaded for comparison. HA antibody only recognised LCs containing the HA tag. Note that due to the insertion of HA tag, LC-HA migrated slightly slower than WT LC.

Figure 5. A majority of rAΔH_CN-HA failed to enter cultured TGNs unlike rA-HA.

Rat TGNs on coverslips were incubated with 100 nM of rA-HA (a), rA (b), or rAΔH_CN-HA (c) for 24 h at 37 °C in culture medium. Washed cells were fixed with paraformaldehyde, permeabilised and blocked with BSA. Paired primary antibodies [rabbit monoclonal anti-HA and mouse monoclonal anti-VAMP2] were added for 1 h. Washed samples were incubated with Alexa Fluor 488 goat anti mouse IgG and Alexa Fluor 568 goat anti-rabbit IgG for 1 h. Images of cell bodies and fibres were captured with a confocal microscope. Representative images from three independent experiments are shown. Bars, 10 μm.

Figure 6. Effects of mutations in the H_CN of BoNT/A on its multi-functional activities in rat CGNs as monitored by intracellular cleavage of SNAP-25.

(a) The expressed and purified mutants were nicked and subjected to SDS-PAGE and Coomassie staining in the presence of DTT; note that thrombin converted the majority of the SC to DC forms. (b) After 24 h exposure of CGNs to the different concentrations of the toxin variants in culture medium, the samples were subjected to SDS-PAGE followed by Western blotting. (c) Extents of SNAP-25 cleavage in CGNs by W₉₈₅L and W₉₈₅L/L₉₈₇A were decreased > 10-fold compared to that of rA WT. Changing W₉₇₄ to L did not affect the functioning of BoNT/A, whereas mutant L₉₈₇A exhibited a mild drop in its activity. Note that in some cases error bars are encompassed by symbols. (d,e) CGNs were treated with 0.5 nM W₉₈₅L or its WT in 5 mM K⁺ (LK) buffer for 8 min at 37 °C. Unbound toxin was removed by three washes before incubation in medium for 5 h to allow internalised toxin to cleave SNAP-25. The arrow and arrowhead in panel b and d indicate intact and cleaved SNAP-25, respectively. Data in panel c and e are mean ± S.E.M, n = 3. ***P < 0.001. (f) A representative protein stained gel showing BoNT/A WT and W₉₈₅L mutant have similar protease activities in cleaving GFP-SNAP-25C₇₃-His₆ substrate. The arrow and arrowhead indicate intact and cleaved substrate, respectively. (g) Non-glycosylated GST-SV2C pulled down BoNT/A WT and W₉₈₅L variant to similar extents, revealed by Western blotting using antibodies indicated. (h) Binding of ¹²⁵I-rA(W₉₈₅L) to CGNs was performed as in Fig. 3b. Subtracting non-saturable binding in the presence of 1 μM unlabelled rA(W₉₈₅L) SC (•) from the total (○) yielded the saturable values (□). Inset: Scatchard plot analysis. Data are mean ± S.E.M from two independent experiments performed in duplicates.