Various human neurogenic hyper-excitability disorders are successfully treated with type A or B BoNT (botulinum neurotoxin). The BoNT/A complex is widely used because of its longer-lasting benefits; also, autonomic side-effects are more often reported for BoNT/B. To establish if this distinct effect of BoNT/B could be exploited therapeutically, BoNT/A was modified so that it would bind the more abundant BoNT/B acceptor in rodents while retaining its desirable persistent action. The advantageous protease and translocation domain of BoNT/A were recombinantly combined with the acceptor-binding moiety of type B [HC/B (C-terminal half of BoNT/B heavy chain)], creating the chimaera AB. This purified protein bound the BoNT/B acceptor, displayed enhanced capability relative to type A for intraneuronally delivering its protease, cleaved SNAP-25 (synaptosome-associated protein of 25 kDa) and induced a more prolonged neuromuscular paralysis than BoNT/A in mice. The BA chimaera, generated by substituting HC/A (C-terminal half of BoNT/A heavy chain) into BoNT/B, exhibited an extremely high specific activity, delivered the BoNT/B protease via the BoNT/A acceptor into neurons, or fibroblast-like synoviocytes that lack SNAP-25, cleaving the requisite isoforms of VAMP (vesicle-associated membrane protein). Both chimaeras inhibited neurotransmission in murine bladder smooth muscle. BA has the unique ability to reduce exocytosis from non-neuronal cells expressing the BoNT/A-acceptor and utilising VAMP, but not SNAP-25, in exocytosis.
Hyper-excitability disorders of cholinergically innervated skeletal and smooth muscles are treatable with BoNTs (botulinum neurotoxins) . There are seven serotypes of BoNT/(A–G), from Clostridium botulinum, consisting of a LC (light chain) protease which is linked to a HC (heavy chain) through a disulphide and non-covalent bonds. HC (C-terminal half of HC) binds to its specific acceptors expressed on susceptible neurons, whereas HN (N-terminal half of HC) forms a channel that allows the attached LC to translocate from ‘endosomal-like’ membrane vesicles into the cytosol. Thereafter, the LC cleaves intracellular SNAREs (soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptor) and negates their roles in neurotransmitter release (reviewed in [2,3]). BoNT/A binds SV2 (synaptic vesicle protein 2) isoforms [4,5] and removes nine amino acids from the C-terminus of SNAP-25 (synaptosome-associated protein of 25 kDa) resulting in a very prolonged neuroparalysis, one of the key features that underlies its widespread and effective treatment for various disorders of muscles (e.g. dystonias, dysphonias, spasticity, over-active bladder and certain gastrointestinal conditions) and glands (hyper-hydrosis and sialorrhea) due to their overly-active cholinergic nerves . Syt (synaptotagmin) I and II plus gangliosides serve as acceptors for BoNT/B [6–8]. There is a higher content of Syts in rat synaptic vesicles than SV2 , and murine motor nerve endings possess a higher density of binding sites for BoNT/B than BoNT/A . After internalization, VAMP (vesicle-associated membrane protein) is cleaved by BoNT/B ; its complex with haemagglutinin is sometimes injected intra-muscularly into patients not responding to BoNT/A, but much higher doses are required [12,13]. Neurologists have noticed that patients given BoNT/B tended to show autonomic side-effects (e.g. dry mouth, accommodation difficulties, dryness of eyes and reduced sweating) [14,15], an interesting finding in relation to SytI being mainly expressed in rat peripheral autonomic and sensory neurons rather than motor endplates . Hence, it seemed warranted to evaluate if a novel therapeutic could be engineered that encompasses the most advantageous properties of BoNT/A (e.g. its long-lived LC protease ), the more abundant BoNT/B acceptors in rodents and the, apparently, more pronounced action of BoNT/B on autonomic cholinergic nerves innervating secretory glands. One strategy for such an attractive improvement entailed creating the chimaera AB by replacing the C-terminal acceptor binding domain of BoNT/A (HC/A) with its counterpart HC/B. In a complementary approach, the acceptor binding domain of BoNT/B was replaced with its counterpart from BoNT/A in order to establish if the resultant protein has new therapeutic potential by delivering its VAMP-cleaving protease into BoNT/B-insensitive cells.
In the present study, the chimaera AB was constructed by protein engineering to harness the LC/A protease, including the contiguous translocation domain (HN) of BoNT/A together with the HC of BoNT/B. For its counterpart, BA, HC/A was combined with the VAMP-cleaving LC protease and HN/B. Both chimaeras were readily expressed in Escherichia coli as SC (single chain) His6-tagged proteins, completely purified and converted readily into their disulfide-linked DC (di-chain) forms by controlled proteolysis. AB was found to enter cultured neurons efficiently cleaving SNAP-25, and to block synaptic transmission in mouse phrenic nerve-muscle like BoNT/A. Most importantly, it produced a more extended neuromuscular paralysis than BoNT/A in mice. BA exhibited all the functional characteristics of BoNT/B, and with an exceptionally high specific neurotoxicity. An observed ability of BA to cleave VAMP in non-neuronal cells highlights its therapeutic potential for normalizing secretion from cells expressing the acceptor for BoNT/A (SV2), but not that for BoNT/B (SytI and II), and requiring VAMP for exocytosis.
Suppliers of reagents for production of chimaeric proteins were listed previously . Custom-made antibodies specific for LC/A, LC/B or BoNT/A were prepared by Zymed Laboratories; anti-BoNT/B IgG antibody was from Metabiologics. Natural purified BoNTs were purchased from List Laboratories (BoNT/A in the DC form) and Metabiologics (BoNT/B); the latter SC was proteolytically nicked to DC with TrypZean (8 μg/mg toxin for 40 min at 22°C). Mouse SytII cDNA clone was obtained from ImaGenes. Unless specified, all other reagents were from Sigma.
Female Tyler's Ordinary mice were purchased from Harlan UK and Sprague–Dawley rats were bred in an approved Bioresource Unit at Dublin City University. All experimental procedures involving animals were approved by the Institutional Ethics Committee and licensed by the Irish Government, Department of Health and Children.
Construction of BoNT AB and BA chimaeras
The cloning and expression of all BoNT variants were performed in accordance with European Union regulations, registered in Ireland with the Environmental Protection Agency and notified to the Health and Safety Authority. Codon-optimized genes encoding the SC of either BoNT/A or BoNT/B were synthesized and their sequences verified. Gene fragments encoding the LC plus the HN of BoNT/A (LC.HN/A) and that for the BoNT/B-binding domain (HC/B) were produced by PCR and cloned into pET29a vector to generate the chimaera AB (Figure 1A, upper panel). Nucleotides encoding a 9-residue linker (ELGGGGSEL) were inserted between the HN and HC to improve protein folding. The chimaera BA (see Figure 1A, lower panel) was, likewise, generated by ligation of the corresponding genes generated by PCR for LC.HN/B and HC/A via a linker encoding two extra residues (DI) into pET29a to create an expression vector containing the BA insert. This construct, unlike AB, contains two thrombin-cleavage sites generated by PCR using suitable primers followed by self-ligation. One site is in the loop region to facilitate specific nicking, whereas extra nucleotides were also added to encode a thrombin consensus site for cleaving the C-terminal His6 tag (Figure 1A, lower panel), a strategy successfully used before .
Arrangements of the functional domains in the BoNT chimaeras AB and BA, their expression and purification
Expression and purification of the chimaeric neurotoxins
All of the DNA sequences were verified and each new SC gene was transformed into E. coli BL21.DE3 cells, expressed and the protein purified by IMAC (immobilized metal-affinity chromatography) as described previously . The resultant AB chimaera was buffer exchanged into 50 mM Tris/HCl (pH 8.1) and further purified on a UNO Q1 column; after washing with 30 mM NaCl, elution was achieved with a stepwise gradient (up to 1 M NaCl in 50 mM Tris/HCl). Pooled fractions containing pure SC were either stored at −80°C or proteolytically nicked by TrypZean (1 μg/mg of toxin for 1 h at 22°C) before the addition of trypsin inhibitor (100 μg/mg of toxin) and storage (as for the SC). Material from the IMAC purification of chimaera BA was gel-filtered into 20 mM sodium phosphate buffer (pH 5.8) and further purified on a UNO S1 column, followed by washing with 100 mM NaCl and elution with a stepwise gradient (up to 1 M NaCl in the phosphate buffer). After buffer exchanging the eluted toxin into 20 mM Hepes and 145 mM NaCl (pH 7.8), purified SC toxin was either stored at −80°C or nicked by biotinylated thrombin (1 unit/mg for 1 h at 22°C) followed by removal of the thrombin by Streptavidin agarose, using the manufacturer's protocol before storage. The chimaeras were analysed by SDS/PAGE (4–12% precast Bis-Tris gel, Invitrogen) and Western blotting at each stage, as described previously . Note that all assays were performed using their DC forms.
Protease activities of new BoNT variants
A model recombinant substrate for assay of SNAP-25 cleavage by BoNT/A and the chimaera AB was used as described previously [18,20]. For analysis of VAMP2 cleavage by BoNT/B and chimaera BA, DNA-encoding GFP (green fluorescent protein) was fused to nucleotides encoding rat VAMP2 (residues 2–94) and a His6 tag. The GFP–VAMP2(2–94)–His6 fusion protein, expressed in E. coli and purified by IMAC, acted as substrate in the fluorescence assay following an established protocol [18,20].
In vitro acceptor-binding assay
GST (glutathione transferase)-tagged intra-luminal fragments of acceptors for BoNT/A [GST–rat SV2C(454–579)]  and BoNT/B [GST–mouse SytII(1–63)–His6] were expressed/purified as described previously  and employed for measuring the binding characteristics of the two chimaeras. GST–SytII(1–63)–His6 was generated from the mouse SytII gene using PCR, cloned into the pET-41a vector (Novagen) and expressed in E. coli cells (BL21.DE3). The binding assay was performed as described previously [4,21]. Briefly, the proteins (~100 μg) were immobilized using 100 μl of slurry of glutathione Sepharose-4B Fast Flow resin (GE Healthcare) and incubated with 100 nM of each toxin in total volume of 100 μl of binding buffer [50 mM Tris/HCl, 150 mM NaCl and 0.5% Triton X-100 (pH 7.6)]. Beads were then collected by centrifugation (500 g for 5 min at 4°C) and washed five times with >10 bed volumes of the same buffer for 15 min at 4°C. Bound proteins were eluted from the washed beads by adding SDS sample buffer. Less than 5% of bound material was subjected to electrophoresis on 4–12% precast Bis-Tris gels (Invitrogen). Toxins were detected by Western blotting with the antibodies indicated.
Cell-based SNARE cleavage assay
Preparation and maintenance of rat CGNs (cerebellar granule neurons) followed standard methods . After 7 days in vitro, the cells were exposed for 24 h at 37°C to a series of toxin concentrations. Spinal neurons were prepared from mouse spinal cords removed at gestation day 13, as described previously . Cultures were exposed to 400 pM of toxin in stimulation buffer  for the time indicated, washed twice with toxin-free medium and incubated for a further 20 h before harvesting. Fibroblast-like synoviocytes were prepared from the knee joint synovium of Sprague–Dawley rats (8–10-weeks-old), as reported previously [22,23]. After four passages, the cells were incubated at 37°C for 20 h with or without 100 nM toxin in the presence of 1 μM substance P in complete DMEM (Dulbecco's modified Eagle's medium) [22,23]. The cells were washed twice before harvesting, solubilization in SDS sample buffer and SDS/PAGE (12% precast Bis-Tris gel) followed by Western blotting; cleavage of endogenous SNAREs was quantified as described previously [17,24].
Neuromuscular paralytic activities and lethalities of BoNTs
Mouse phrenic-nerve hemi-diaphragms were set up as described previously . The whole bladder was removed from freshly decapitated rats or mice, processed and stimulated, as established by others . The times taken for each toxin, at a range of concentrations, to reach 50% paralysis were recorded. For clarity, data for a single representative concentration have been plotted. Comparison of the recovery times from neuromuscular paralysis in vivo was based on the DAS (digit abduction score) assay [18,26]. For each BoNT, the TDmax. (maximal tolerated dose) was established ; ten mice per sample were injected with the doses summarized in the Figure legends. Toxins' lethalities were determined using a LD50 assay after intraperitoneal injection into mice as described previously . Groups of four mice were used for each concentration; specific neurotoxicities are expressed as the number of mLD50 (mouse LD50) units/mg of toxin.
Statistical analysis and data presentation
Average data are presented with the S.E.M. and relevant sample size. All calculations and graphs were done using GraphPad Prism 4.0 and P values calculated as indicated in the Figure legends; P<0.05 was considered statistically significant.
Inter-changing functional domains between BoNT/A and BoNT/B generated chimaera AB and BA with unaltered levels of the requisite protease activities
It was highly desirable to create a therapeutic encompassing the most advantageous features of the two clinically used serotypes, BoNT/A and BoNT/B. In the first instance, the long-acting LC/A protease and associated translocation moiety (HN) was combined recombinantly with the binding domain of BoNT/B, HC/B, which targets the relatively abundant acceptors, SytI and II in rodents (see the Introduction section). This chimaera AB (LC.HN/A–HC/B) was created by fusing portions of the genes encoding the protease and translocation domains of BoNT/A with the C-terminal acceptor-binding moiety of BoNT/B, on the basis of their crystal structures [28,29]. A linker of nine exogenous residues between the HN/A and HC/B domains was added to facilitate protein folding and functioning of these moieties (Figure 1A, upper panel). Generation of the counterpart chimaera was achieved in a similar manner. In this case, DNA encoding the LC.HN of BoNT/B was fused to the sequence for the HC/A, creating a contiguous open reading frame for the BA chimaera (Figure 1A, lower panel). In the present study, the native loop of BoNT/B was partially substituted with a specific consensus sequence and a six-residue non-structured linker for efficient cleavage by thrombin. Both proteins were C-terminally tagged with His6 for affinity purification; a C-terminal recognition site for thrombin was also introduced in chimaera BA to facilitate tag removal.
Both chimaeric toxins were successfully expressed in E. coli and purified from the lysed bacteria to ≤80% purity by IMAC via their His6 tag. Following elution with imidazole, proteins of the expected size (~150 kDa) were detected by SDS/PAGE and Coomassie Blue staining, together with contaminants of lower molecular mass (Figures 1B and 1D). Nearly all impurities were removed by subsequent ion-exchange chromatography; chimaera AB was eluted from an anion-exchange column at ≥70 mM NaCl (~1 mg of pure toxin per litre of culture) (Figure 1C), whereas BA could be readily purified using cation-exchange chromatography (Figure 1E), eluting at ≥180 mM NaCl, with a typical yield of ~8 mg of pure toxin per litre of culture. SDS/PAGE analysis of peak fractions, followed by protein staining, revealed a single band of expected size for both chimaera AB (Figure 2A) and BA (Figure 2C) in the presence or absence of DTT (dithiothreitol). Controlled nicking of chimaera AB with TrypZean resulted in near-complete proteolysis of SC to a disulfide-linked DC (Figure 2A). Virtually all of chimaera BA was nicked with thrombin (Figure 2C). The appearance of HC and LC from both chimaeras in the presence of reducing agent not only confirms efficient nicking, but also indicates that the inter-chain disulfide bond had been successfully formed in each (Figures 2A and 2C). Presence of the requisite functional domains in AB and BA was confirmed by Western blotting, using specific IgGs (Figures 2B and 2D). Unlike visualization of His6 in the SC and DC of chimaera AB, and in the HC in the presence of DTT (Figure 2B), this signal could not be detected in BA after treatment with thrombin, confirming complete removal of the tag (Figure 2D).
Activation of SC chimaeric BoNTs by specific proteolytic nicking
It was necessary to verify that the HC substitution of BoNT/A and BoNT/B did not in any way hinder protease function, using an assay with two model synthetic substrates (see the Experimental section and ). Both chimaeras showed protease activities comparable with their parents (Table 1), demonstrating that these were not altered by the binding domains being swapped. Therefore, any differences between the performance of chimaeras and their parents could be ascribed to the translocation and/or acceptor-binding domain.
|Toxin||EC50 (nM) for cleavage of GFP–SNAP-25(134–206)–His6 or GFP–VAMP2(2–94)-His6*||mLD50 units/mg‡|
|Chimaera AB||0.53±0.10 (n=3)||0.3×108§|
|Chimaera BA||3.81±2.18 (n=5)†||6×108|
|Toxin||EC50 (nM) for cleavage of GFP–SNAP-25(134–206)–His6 or GFP–VAMP2(2–94)-His6*||mLD50 units/mg‡|
|Chimaera AB||0.53±0.10 (n=3)||0.3×108§|
|Chimaera BA||3.81±2.18 (n=5)†||6×108|
Proteolytic activities of chimaeric and parental DC toxins were determined using model substrates (13.5 μM GFP–SNAP-25(134–206)–His6 or 13.5 μM GFP–VAMP2(2–94)–His6). Values represent the amount of each toxin needed to cleave 50% of substrate within 30 min at 37°C.
There are no significant differences between BoNT/B and chimaera BA (P>0.05 using unpaired Student's t test).
The lowest dose of toxin that killed 50% of a group of four mice within 4 days after intraperitoneal injection was 1 mLD50 unit.
A similar value was obtained for AB devoid of the His6 tag.
Chimaera AB specifically binds to a BoNT/B acceptor, SytII, whereas BA interacts with SV2C, the acceptor for BoNT/A
Previous studies established that the binding of BoNT/B to SytI and II leads to uptake into cells, whereas the entry of BoNT/A is mediated by SV2 (see the Introduction section). To ensure successful delivery of LC/B via the BoNT/A-binding domain and of LC/A via the acceptor for HC/B, it was first necessary to ascertain if both chimaeras had acquired the ability to bind the acceptors dictated by their HC domains. Evaluating their binding in a pull-down assay utilized purified intra-luminal fragments of acceptors for BoNT/A [GST–rat SV2C(454–579)] or BoNT/B [GST–mouse SytII(1–63)–His6]. SytII rather than I was chosen because the former has been confirmed to be present at the neuromuscular junction and does not require the presence of gangliosides for BoNT binding [6,8,30]. Binding to the protein acceptors was established by confirming the specificities of antibodies used for detecting the toxins (Figures 3A and 3B, left-hand panels). As expected, the BoNT/A acceptor pulled down only toxins containing HC/A (BoNT/A and BA) (Figure 3A, right-hand panel). This interaction has to be attributed to the binding of HC/A to SV2C as this is the only domain shared by these two toxins. The binding also appears to be specific as HC/B-containing toxins (BoNT/B and AB) were not detected (Figure 3A, right-hand panel). In other experiments, where the toxins were pulled down using the BoNT/B acceptor, only BoNT/B and chimaera AB were found to bind to the SytII fragment (Figure 3B, right-hand panel). This reaffirms that acceptor binding can be mediated through the HC/B in BoNT/B or chimaera AB. Neither chimaera nor parental BoNTs displayed any binding to GST–agarose (results not shown).
Exclusive interaction of chimaera AB with SytII and BA with SV2C in vitro
AB and BA cleave their requisite SNAREs, SNAP-25 and VAMP, in neurons and block transmission in both skeletal and smooth muscles
To assess the toxins' abilities to undergo the multiple steps of binding, translocation and cleavage of their respective substrates in situ, they were incubated with rat cultured CGNs for 24 h and cleavage of SNAP-25 and VAMP2 was monitored by immunoblotting. Chimaera AB and BoNT/A proved equipotent, with the ratio between intact and cleaved SNAP-25 being the same for both of these BoNT/A protease-containing toxins (Figure 4A). Immunoprobing of cell lysates for VAMP2 cleavage showed BA matched the activity of BoNT/B (Figure 4B). As the cause of death by BoNT is asphyxiation due to blockade of neurotransmission in the diaphragm, uptake of these toxins into the phrenic nerve was examined. Both chimaeras took similar times to paralyse mouse nerve hemi-diaphragms as their parental toxins (Figure 4C). Hence successful delivery of the respective LCs into motor neurons, via different acceptors, followed by translocation to the cytosol had occurred. The versatility of these chimaeras was assessed in the autonomic nervous system. BA and AB displayed characteristics comparable with the parental toxins when tested on mouse bladder strips; 1 nM of each achieved 50% paralysis within 3 h (Figure 4D). Contraction of bladder from rat was inhibited only by BoNT/A but not BoNT/B (Figure 4D), observations reflected by their chimaeras where only AB induced muscle paralysis. An insensitivity of rat bladder to BoNT/B or BA may be attributable to the known non-susceptibility of VAMP1 to cleavage ; this observation suggests an apparent preferential use of isoform 1 for neurotransmission in smooth muscle rather than an absence of the Syt acceptors (see the Discussion section).
Both chimaeras potently and specifically cleave their requisite substrates in intact cultured neurons and block neuromuscular transmission in vitro
Unique characteristics of BA offering potential for new therapeutic avenues
The neurotoxicity of chimaera BA, measured by intraperitoneal injection into mice, gave a very high value (6×108 mLD50/mg), 20-fold greater than that for AB (Table 1). Such findings could indicate more optimal folding and/or interactive compatibility of the constituent domains of BA than is the case for AB, even though the latter contains a linker which does slightly improve the potency (results not shown). Special features of this new chimaera could allow additional unexplored applications in the treatment of diseases. Such prospects are exemplified by attempts to cleave the particular SNARE isoforms found in fibroblast-like synoviocytes, with the possibility of reducing their exocytosis of cytokines and, in that way, alleviating symptoms of arthritis [31,32]. These cells were found by Western blotting to contain VAMP3 and SNAP-23 (synaptosome-associated protein of 23 kDa) rather than SNAP-25 (Figures 5A and 5C); SNAP-23 is not cleaved by BoNT/A (Figure 5A). Despite VAMP3 being a substrate for BoNT/B, the latter failed to cleave this target in the synoviocytes (Figures 5A and 5B); encouragingly, 100 nM BA cleaved ~60% of VAMP3 in the presence of substance P (Figure 5A and 5B), a pain mediator whose level is known to be elevated under arthritic conditions . These contrasting findings are explained by the demonstrated presence of the BoNT/A acceptor, SV2A, in these cells and an absence of SytI and II, in contrast with their occurrence in CGNs (Figure 5C). This example highlights the utility of BA which could be applicable to other non-neuronal cells lacking SNAP-25, but possessing the SV2, the BoNT/A acceptor, and utilising VAMP(1–3) in exocytosis.
Chimaera BA enters rat cultured synoviocytes and cleaves VAMP 3, unlike BoNT/B which fails to cleave its substrate
AB enhanced the delivery of LC into spinal cord neurons and induced longer neuromuscular paralysis than BoNT/A in vivo
It was hypothesized that the expression of a higher content of Syt I/II than SV2 in rat synaptic vesicles , and a higher density of binding sites for BoNT/B than BoNT/A in murine motor nerve endings , would allow more uptake of AB than BoNT/A which, in turn, might culminate in an extension of the duration of action. Mouse spinal cord neurons were used initially for addressing this possibility. These cultured neurons were briefly bathed in stimulation buffer containing 400 pM toxin for selected periods, followed by extensive washing; subsequent incubation for 20 h allowed extrapolation of the amounts of toxin bound from the extents of SNARE cleavage. Under this paradigm, there seemed to be less binding of BoNT/A than BoNT/B over short times as reflected in a minor fraction of SNAP-25 being cleaved (Figure 6A), compared with the extent of VAMP2 disappearance due to BoNT/B (Figure 6B). There was only enough of BoNT/A bound to the acceptor in 5 min to truncate ~10% of SNAP-25 after incubation for 20 h (Figure 6A), whereas BoNT/B caused cleavage of over 80% of VAMP2 (Figure 6B). AB cleaved significantly more SNAP-25 than BoNT/A (Figure 6A), even though it possesses the same protease, more closely reflecting the binding time course of BoNT/B. As the AB chimaera shares the same protease and HN regions as its BoNT/A parent, it is reasonable to deduce that these differences are attributable to acceptor HC-mediated binding. This deduction was reinforced by comparing BA with BoNT/A. Even though BA contains the VAMP2-cleaving LC/B, shown to cause significant cleavage when part of BoNT/B (Figure 6B), the protease activity of this chimaera deduced from this assay was significantly retarded, registering ~10% cleavage after 5 min of binding compared with BoNT/B, which cleaved ~80% of VAMP2 (Figure 6B). Again, their near-identical protease activities towards recombinant VAMP2 substrate (Table 1) rule out any of these differences being derived from LC.
Cleavage of SNAREs by the chimaeras in spinal cord neurons approximates to that of the parental toxins providing their HC: AB induces the longest neuroparalysis in vivo
The above-noted enhanced acceptor-binding capabilities of AB compared with BoNT/A in spinal cord neurons might alter the persistence of its neuromuscular paralysis in vivo. Therefore, this was measured using the DAS assay in mice (Figure 6C). The TDmax. of each toxin was established empirically (Figure 6C, inset) to avoid any local and systemic side-effects . As found previously , full muscle paralysis (DAS=4) was observed in mice injected with TDmax. of BoNT/A, lasting up to 28 days (when DAS=0). The amount of BoNT/B that could be injected without inducing systemic effects was limited by its high LD50 value and when the smaller permitted amount of BoNT/B (1.5 units/~2 pg) was injected intramuscularly, near-complete paralysis ensued within 2 days and full recovery occurred by day 14 (Figure 6C). Chimaera BA containing LC/B proved to be highly effective in inducing muscle weakening; again the limiting dose (1.5 units/~2 pg) injected caused paralysis followed by recovery within ~10 days. On the other hand, the neuromuscular paralysis induced by chimaera AB (6 units/200 pg) outlasted that of the other toxins tested in rodents, exceeding the previously published record set by the clinically used type A complex . This result seems to support the hypothesis that the more efficient binding of AB, apparently facilitated by a greater abundance of SytI/II acceptors, lengthens the duration of neuromuscular paralysis (up to 50 days) in this test system.
These collective findings demonstrate clearly that BoNT chimaeras can be constructed through recombination of functional domains from different serotypes and, also, highlight how these can be expressed efficiently, simply purified to homogeneity, and readily converted into active DC forms. Both proteins retained functional characteristics of their parents; most importantly, AB displays a persistent BoNT/A-like property, but of even longer duration, in mice, whereas BA has the potential to substitute for BoNT/B because of its high specific neurotoxicity and demonstrated ability to deliver LC/B into cells insensitive to BoNT/B, but susceptible to VAMP cleavage.
Engineered BoNTs with potential as improved and more versatile inhibitors of exocytosis
Development of novel forms of BoNTs with extended duration of paralytic activity and wider applications is very desirable for treating human neurogenic hyper-activity disorders. For the first time, it is shown in the present study that the duration of BoNT/A can be extended substantially in a murine model by replacing the acceptor-binding domain with its counterpart from BoNT/B, resulting in the chimaera AB. Due to utilizing a different acceptor to initiate entry, it could provide an additional and/or improved form active on certain neuron types. On the other hand, BA, which harnesses the binding domain of BoNT/A, can enter neurons, block neuromuscular transmission and has an extremely high potency, giving a similar duration of neuroparalysis as BoNT/B. Its ability to target an alternative SNARE in BoNT/A-sensitive neurons, or other cells, creates the exciting potential for being an innovative therapeutic applicable to cells whose exocytosis involves VAMP, but not SNAP-25 (see below).
Chimaera AB causes the most persistent neuromuscular weakening in a mouse model
Proof of principle was obtained for generating a longer-lasting chimaeric neurotoxin by selectively combining the most advantageous domains of two BoNTs, based on an understanding of their multi-step mechanisms of action. As BoNT/A causes the most prolonged muscle weakness in human therapy, LC/A was selected for this purpose, as the life-times of LCs determine durations of action [17,18,35,36]. Murine motor endplates possess more acceptors for 125I-labelled BoNT/B than BoNT/A , consistent with findings from a proteomic demonstration that more copies of Syt than SV2 are present on small synaptic clear vesicles from rat brain . Additionally, evidence from rat cultured neurons suggests that SytI and II can accumulate at the cell surface, forming a stockpile of acceptor available for binding HC/B, which is not the case for SV2C, whose distribution is skewed towards the synaptic vesicle rather than the cell surface [37,38]. Recently, in-vitro-translated 35S-labelled HC/B was reported to have a higher affinity for rat brain synaptosomes than HC/A, although the sensitivity of motor nerve endings was not measured . Transfer of BoNT/B acceptor-binding domain to BoNT/A could, it was speculated, allow delivery of more BoNT/A protease to the rodent nerve terminals and, thereby, cause an extended duration of its action. Moreover, as BoNT/B appears to show a more pronounced action on autonomic cholinergic nerves in human secretory glands (see the Introduction section), transfer of this feature to BoNT/A could increase its scope for targeting the BoNT/A long-lived protease to certain nerve types. For these purposes, the engineered AB, expressed in E. coli as a SC and purified to homogeneity, followed by activation by controlled proteolysis, was shown to retain characteristics of the respective functional domains from BoNT/A and BoNT/B, i.e. binding the luminal portion of murine SytII, entering into neurons, cleaving SNAP-25 and blocking nerve–muscle transmission.
Experiments on time-dependency of acceptor binding to murine spinal neurons indicated that AB does, indeed, deliver its LC more efficiently than BoNT/A into the cytosol, as reflected by a greater extent of SNAP-25 cleavage at each time tested. This difference may underlie the notably prolonged neuroparalysis resulting from AB, which even exceeds that of BoNT/A. Such an extended action of AB seems to be due to binding to distinct acceptors via its modified HC domain. Notably, chimaera AB is equipotent to native BoNT/A in blocking neurotransmission in mouse hemi-diaphragm in vitro, but it exhibits much lower specific toxicity in a mouse lethality assay. This could be due to fact that the in vitro assay only reflects the speed of onset of neuroparalysis, rather than true potency. Therefore, the recent report of a 4-fold greater potency on the diaphragm for the SC of chimaera AB compared with SC BoNT/A , may not necessarily translate into a higher specific lethality in vivo; no comparative biological data for BoNT/A were provided . Nevertheless, ~1.7-fold increase in the duration of muscle weakness in mice observed for AB relative to BoNT/A could further extend the therapeutic action in patients as the time courses observed in mice are known to be far longer in humans. Although ethical issues pose difficulties, it will be necessary eventually to evaluate AB in humans, especially as human SytII was recently excluded as a high-affinity acceptor for BoNT/B due to a single amino acid difference in the interaction site compared with the rodent protein , which leaves Syt I as the protein constituent of the acceptor for mediating entry of the toxin into motor endings in humans. This and the lower specific neurotoxicity of AB compared with BoNT/A may require injection of a higher protein dose, which could increase the possibility of triggering the production of neutralising antibodies. It may be possible to overcome the latter by protein engineering an improved version of the toxin via site-directed mutagenesis in its binding site for human SytII, thereby raising the affinity for the acceptor, as suggested previously . The duration of action of AB on autonomic neurons should also be evaluated, as the autonomic side-effects reported for BoNT/B could be due to the higher abundance of SytI in autonomic neurons as shown in rodents  or by the higher amounts injected.
BA offers scope for attenuating secretion from a variety of cell types (e.g. fibroblast-like synoviocytes) via two sites of action
Notably, both BA and BoNT/B gave a much shorter duration of neuromuscular paralysis than BoNT/A, further confirming that the protease life-time determines the normal duration of action, as reported previously [17,18,35,36]. Nevertheless, BA proved as effective as BoNT/B in cleaving VAMP in cultured neurons and inhibiting neuromuscular transmission; of special note is its extremely high specific neurotoxicity. These attractive features highlight that BA was properly folded in E. coli, forming all of the fully functional multiple domains. As BA targets LC/B to BoNT/A-susceptible neurons by binding to SV2 rather than Syt in humans, it has the potential to be applied as an alternative to BoNT/B, but with much lower doses required. Moreover, because BA targets VAMP, it could complement BoNT/A-based neurotherapy. The further utility of BA has been highlighted with synoviocytes, that express BoNT/A-acceptor SV2A but not SytI or II (making them inaccessible to BoNT/B), in which cleavage of VAMP3 occurred following stimulation by substance P, a pain mediator in the development of arthritis . BoNT/A also can reduce arthritis-associated pain in humans, an effect most probably mediated through direct inhibition of peripheral nociceptive nerve activation and thereby prevents sensitization . Our collective observations suggest that BA could offer an advantage of counteracting the symptoms of arthritis at two points: directly upstream at the neuronal component like BoNT/A and, indirectly, in synoviocytes.
Involvement of VAMP 1 isoform in exocytosis from autonomic nerves revealed by BoNT chimaeras
The therapeutic effectiveness of BoNT/A in autonomic cholinergic diseases (reviewed in ) prompted evaluation of the performance of these chimaeras in the bladder. Although neuroparalysis observed in mouse bladder confirmed their functionality, reminiscent of that of their parents, experiments on rat bladder implicated VAMP1 in neurotransmission. The observation that BoNT/B had no effect on rat bladder suggested that there might be a BoNT/B-insensitive VAMP 1  mediating transmitter release or a lack of the acceptor for BoNT/B. The latter notion was excluded because chimaera AB blocked the transmission like BoNT/A, establishing the presence of the Syt I/II acceptor, as it successfully delivered the LC/A protease culminating in paralysis. Likewise, the inability of BA to block transmission in the rat bladder reaffirmed the notion that insensitivity to BoNT/B in rat is due to the resistant VAMP1 mediating transmitter release, as demonstrated for sensory neurons  and motor endplates .
cerebellar granule neuron
digit abduction score
green fluorescent protein
C-terminal half of HC
N-terminal half of HC
immobilized metal-affinity chromatography
synaptosome-associated protein of 23 kDa
synaptosome-associated protein of 25 kDa
soluble N-ethylmaleimide-sensitive factor attachment protein receptor
synaptic vesicle protein 2
maximal tolerated dose
vesicle-associated membrane protein
Jiafu Wang designed and manufactured the AB and BA chimaeras. Jiafu Wang, Tomas Zurawski, MacDara Bodeker, Jianghui Meng and Sanjay Boddul performed experimental procedures. Jiafu Wang, Tomas Zurawski and MacDara Bodeker prepared the paper; Roger Aoki and Oliver Dolly edited the paper prior to submission.
This work was supported by Science Foundation Ireland, IRCSET (Irish Research Council for Science, Engineering and Technology) and the Neuroscience section of the Programme for Research in Third Level Institutions (PRTLI) Cycle 4. The PRTLI is co-funded through the ERDF (European Regional Development Fund), part of the European Union Structural Funds Programme 2007–2013.