Citation: Gregory, K.S.; Acharya, K.R. A Comprehensive Structural Analysis of Clostridium botulinum Neurotoxin A Cell-Binding Domain from Different Subtypes. Toxins 2023, 15, 92. https://doi.org/10.3390/ toxins15020092 Received: 1 December 2022 Revised: 11 January 2023 Accepted: 16 January 2023 Published: 18 January 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).toxins Review A Comprehensive Structural Analysis of Clostridium botulinum Neurotoxin A Cell-Binding Domain from Different Subtypes Kyle S. Gregory and K. Ravi Acharya * Department of Life Sciences, University of Bath, Claverton Down, Bath BA2 7AY, UK * Correspondence: bsskra@bath.ac.uk; Tel.: +44-(0)1225-386-238 Abstract: Botulinum neurotoxins (BoNTs) cause flaccid neuromuscular paralysis by cleaving one of the SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) complex proteins. BoNTs display high affinity and specificity for neuromuscular junctions, making them one of the most potent neurotoxins known to date. There are seven serologically distinct BoNTs (serotypes BoNT/A to BoNT/G) which can be further divided into subtypes (e.g., BoNT/A1, BoNT/A2 . . . ) based on small changes in their amino acid sequence. Of these, BoNT/A1 and BoNT/B1 have been utilised to treat various diseases associated with spasticity and hypersecretion. There are potentially many more BoNT variants with differing toxicological profiles that may display other therapeutic benefits. This review is focused on the structural analysis of the cell-binding domain from BoNT/A1 to BoNT/A6 subtypes (HC/A1 to HC/A6), including features such as a ganglioside binding site (GBS), a dynamic loop, a synaptic vesicle glycoprotein 2 (SV2) binding site, a possible Lys–Cys/Cys–Cys bridge, and a hinge motion between the HCN and HCC subdomains. Characterising structural features across subtypes provides a better understanding of how the cell-binding domain functions and may aid the development of novel therapeutics. Keywords: botulinum neurotoxin; cell-binding domain; ganglioside binding; synaptic vesicle glycoprotein 2; structural analysis Key Contribution: The cell-binding domain of botulinum neurotoxin A displays significant func- tional variability across subtypes. A comprehensive analysis of the available structural data on the cell-binding domains BoNT/A1 to BoNT/A6 is presented here; detailing the six structural features present across the subtypes; a ganglioside binding site; a dynamic loop; an SV2 protein binding site; an SV2 glycan binding site; a Lys–Cys/Cys–Cys bridge; and a hinge region. 1. Introduction Clostridium botulinum neurotoxin (commonly referred to as BoNT) is widely utilised within the pharmaceutical industry to treat a range of diseases associated with muscular and endocrine overactivity. There are over 100 approved medicinal applications of BoNT, such as the treatment of spasticity, blepharospasm and dystonia [1], as well as for aesthetic indications [2]. Historically, all bacteria producing botulinum neurotoxin were classified as Clostridium botulinum, and this metabolically diverse species was divided into four distinct phenotypes (Groups I, II, III, and IV) [3]. BoNT can cause a potentially fatal disease, botulism, that is characterised by flaccid paralysis due to inhibition of acetylcholine neurotransmission. C. botulinum groups I (proteolytic) and II (non-proteolytic) are responsible for botulism in humans [4], whereas group III causes animal botulism [5,6] and C. botulinum group IV does not cause botulism. There are many BoNT variants which have been categorised into distinct serotypes due to neutralisation by different antibodies. A total of seven serotypes [7] (BoNT/A to /G) have been identified as well as a novel serotype (BoNT/X) that is not neutralised by any known BoNT antibody [8,9]. Advancements in genome Toxins 2023, 15, 92. https://doi.org/10.3390/toxins15020092 https://www.mdpi.com/journal/toxins
Toxins 2023, 15, 92 2 of 22 sequencing have also led to the discovery of mosaic BoNTs (BoNT/FA(HA), /CD, and /DC) [10–13] as well as BoNT-like proteins from Weissella oryzae (‘BoNT/Wo’) [14–16], Enterococcus faecium (‘BoNT/En’) [17,18], and Paraclostridium bifermentans (‘PMP1’) [19]. Toxicity across serotypes varies significantly, with BoNT/A, /B, /E, and /F associated with human disease [20]. Each serotype within BoNT is further divided into subtypes (e.g., BoNT/A1-/A10, BoNT/B1-/B8, BoNT/E1-/E12, BoNT/F1-/F9) due to differences within their amino acid sequence [21]. The similarity cut-off for any newly identified BoNT being defined as a subtype has been arbitrarily set to >2.5% [22], meaning within the subtypes themselves there is additional variation. This huge diversity of BoNTs suggests the toxin may have a viral origin [23], and at the domain level subtypes appear chimeric as they have evolved at different rates [24]. Most subtypes, within the same serotype, bind the same antibody and cleave the same substrate at the same site; however, they display different binding characteristics to monoclonal and polyclonal antibodies [25–28]. One exception is BoNT/F5, which cleaves its substrate at a different site compared to the other BoNT/F subtypes [29]. Therefore, understanding the functional and structural differences across different subtypes is important for the development of novel BoNT-based therapies. The molecular structure of BoNT/A (Figure 1A) consists of both a light chain (LC) of ~50 kDa and a heavy chain (HC) of ~100 kDa that are held together by a single disul- phide bond [30]. BoNTs are expressed as a single polypeptide chain that is cleaved post-translationally to form an active di-chain molecule by either a native or host pro- tease [31,32]. The LC possesses a catalytic domain, a Zn2+-dependent endopeptidase, whereas the HC comprises two domains: a translocation domain (HN) and a cell-binding domain (HC). The HN resembles a coiled-coil structure with a ~50 residue long ‘belt’ region that wraps around the LC [30] (Figure 1A) inhibiting endopeptidase activity by behaving as a pseudo-substrate [33]. The cell-binding domain contains an N-terminal β-jelly-roll fold and C-terminal β-trefoil fold which are referred to as the HCN and HCC subdomains, respec- tively [34] (Figure 1A). Most BoNT serotypes are thought to adopt a domain arrangement in what can be described as an ‘open-wing’ conformation based on the crystal structures of BoNT/A (Figure 1A) [30] and BoNT/B [35]. The exception to this is BoNT/E, which adopts a ‘closed-wing’ conformation [36] (Figure 1B).Toxins 2023, 15, 92 2 of 23 is not neutralised by any known BoNT antibody [8,9]. Advancements in genome sequenc- ing have also led to the discovery of mosaic BoNTs (BoNT/FA(HA), /CD, and /DC) [10– 13] as well as BoNT-like proteins from Weissella oryzae (‘BoNT/Wo’) [14–16], Enterococcus faecium (‘BoNT/En’) [17,18], and Paraclostridium bifermentans (‘PMP1’) [19]. Toxicity across serotypes varies significantly, with BoNT/A, /B, /E, and /F associated with human disease [20]. Each serotype within BoNT is further divided into subtypes (e.g., BoNT/A1-/A10, BoNT/B1-/B8, BoNT/E1-/E12, BoNT/F1-/F9) due to differences within their amino acid se- quence [21]. The similarity cut-off for any newly identified BoNT being defined as a sub- type has been arbitrarily set to >2.5% [22], meaning within the subtypes themselves there is additional variation. This huge diversity of BoNTs suggests the toxin may have a viral origin [23], and at the domain level subtypes appear chimeric as they have evolved at different rates [24]. Most subtypes, within the same serotype, bind the same antibody and cleave the same substrate at the same site; however, they display different binding char- acteristics to monoclonal and polyclonal antibodies [25–28]. One exception is BoNT/F5, which cleaves its substrate at a different site compared to the other BoNT/F subtypes [29]. Therefore, understanding the functional and structural differences across different sub- types is important for the development of novel BoNT-based therapies. The molecular structure of BoNT/A (Figure 1A) consists of both a light chain (LC) of ~50 kDa and a heavy chain (HC) of ~100 kDa that are held together by a single disulphide bond [30]. BoNTs are expressed as a single polypeptide chain that is cleaved post-transla- tionally to form an active di-chain molecule by either a native or host protease [31,32]. The LC possesses a catalytic domain, a Zn2+-dependent endopeptidase, whereas the HC com- prises two domains: a translocation domain (HN) and a cell-binding domain (HC). The HN resembles a coiled-coil structure with a ~50 residue long ‘belt’ region that wraps around the LC [30] (Figure 1A) inhibiting endopeptidase activity by behaving as a pseudo-sub- strate [33]. The cell-binding domain contains an N-terminal β-jelly-roll fold and C-termi- nal β-trefoil fold which are referred to as the HCN and HCC subdomai
Toxins 2023, 15, 92 3 of 22Toxins 2023, 15, 92 3 o Figure 1. Structure of botulinum neurotoxin and a model of dual receptor binding at the neurom cular junction. (A) Crystal structure of BoNT/A (PDB 3BTA) and schematic of linear protein Model of BoNT/A (PDB 3BTA) [30] and BoNT/E (PDB 3FFZ) [36] binding to a neuromuscular j tion membrane by superimposition of BoNT/A and BoNT/E with the AlphaFold2 model of S (synaptic vesicle glycoprotein 2 isoform ‘C’) (uniprot Q496J9) and the SV2C-bound structur HC/A1 [37] (PDB 5JLV). The two models illustrate the different domain organisation betw BoNT/A and BoNT/E, where the latter is thought to display enhanced translocation ability by p tioning both the LC and HN more closely to the membrane [36,38]. SV2C is coloured purple, th domain grey, HN domain green, LC red, and ganglioside orange. The three domains have specific functions that contribute to the mechanism of icity (Figure 2). The first step involves target-cell binding by the HCC subdomain [39] u arrival at the neuromuscular junction (NMJ) (Figure 2A). This subdomain (with the ception of BoNT/C1) binds to both a ganglioside and protein receptor present on the face of motor neurons (Figure 1B) [40]. Two protein receptors have been identified BoNT—three isoforms of synaptic vesicle glycoprotein 2 (SV2A-C) which are recogn by BoNT/A [41], BoNT/D [42], BoNT/E [43], and BoNT/F [44], and two isoforms of syn totagmin (sytI-II) which are recognised by BoNT/B [45,46] and BoNT/G [47]. Following binding, BoNT is internalised through receptor-mediated endocyt [48]. Subsequent acidification of the endosome causes a conformational change of the domain and triggers translocation [49]. Although the precise molecular basis of transl tion has not yet been elucidated, the conformational change to the HN is widely belie to grant the passage of the LC through the endosomal membrane into the cytosol 38,49–56] (Figure 2B). Reduction of the disulphide bond connecting the LC and HC cytosolic thioredoxin (Trx) and thioredoxin reductase [57], is required for release of LC from the membrane-bound HC. Upon release, the LC is then free to cleave its ta SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) pro (Figure 2C). SNARE proteins are involved in the formation of the SNARE complex facilitates vesicular fusion and exocytosis of neurotransmitters and hormones from c BoNT/A, /C, and /E cleave the SNARE protein SNAP-25 [58], whereas BoNT/B, /D, /F, /G cleave VAMP [59–62]; BoNT/C also cleaves syntaxin [63]. Cleavage of any one of th Figure 1. Structure of botulinum neurotoxin and a model of dual receptor binding at the neuro- muscular junction. (A) Crystal structure of BoNT/A (PDB 3BTA) and schematic of linear protein. (B) Model of BoNT/A (PDB 3BTA) [30] and BoNT/E (PDB 3FFZ) [36] binding to a neuromuscular junction membrane by superimposition of BoNT/A and BoNT/E with the AlphaFold2 model of SV2C (synaptic vesicle glycoprotein 2 isoform ‘C’) (uniprot Q496J9) and the SV2C-bound structure of HC/A1 [37] (PDB 5JLV). The two models illustrate the different domain organisation between BoNT/A and BoNT/E, where the latter is thought to display enhanced translocation ability by positioning both the LC and HN more closely to the membrane [36,38]. SV2C is coloured purple, the HC domain grey, HN domain green, LC red, and ganglioside orange. The three domains have specific functions that contribute to the mechanism of tox- icity (Figure 2). The first step involves target-cell binding by the HCC subdomain [39] upon arrival at the neuromuscular junction (NMJ) (Figure 2A). This subdomain (with the exception of BoNT/C1) binds to both a ganglioside and protein receptor present on the surface of motor neurons (Figure 1B) [40]. Two protein receptors have been identified for BoNT—three isoforms of synaptic vesicle glycoprotein 2 (SV2A-C) which are recognised by BoNT/A [41], BoNT/D [42], BoNT/E [43], and BoNT/F [44], and two isoforms of synaptotagmin (sytI-II) which are recognised by BoNT/B [45,46] and BoNT/G [47]. Following binding, BoNT is internalised through receptor-mediated endocytosis [48]. Subsequent acidification of the endosome causes a conformational change of the HN domain and triggers translocation [49]. Although the precise molecular basis of translocation has not yet been elucidated, the conformational change to the HN is widely believed to grant the passage of the LC through the endosomal membrane into the cytosol [37,38,49–56] (Figure 2B). Reduction of the disulphide bond connecting the LC and HC, by cytosolic thioredoxin (Trx) and thioredoxin reductase [57], is required for release of the LC from the membrane-bound HC. Upon release, the LC is then free to cleave its target SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) protein (Figure 2C). SNARE proteins are involved in the formation of the SNARE complex that facilitates vesicular fusion and exocytosis of neurotransmitters and hormones from
Toxins 2023, 15, 92 4 of 22 underpinning post-translocation trafficking of BoNT LC are yet to be established, however, recent evidence suggests LC/A1 is co-localised to the plasma membrane along with SNAP- 25 [65].Toxins 2023, 15, 92 4 of 23 SNARE proteins at the NMJ prevents release of acetylcholine [64]. The precise mecha- nisms underpinning post-translocation trafficking of BoNT LC are yet to be established, however, recent evidence suggests LC/A1 is co-localised to the plasma membrane along with SNAP-25 [65]. Figure 2. Modular mechanism of BoNT/A activity. (A) On arrival at the neuromuscular junction (NMJ), the cell-binding domain binds to both a ganglioside and SV2 on the surface of motor neurons where it is then internalised by receptor-mediated endocytosis. (B) The acidic environment of the endosome results in conformational change of the translocation domain which facilitates the pas- sage of the LC through the endosomal membrane into the cytosol. Cytosolic thioredoxin (Trx) and thioredoxin reductase (TrxR) reduce the disulphide bond connecting the LC to the HC, releasing the LC into the cytosol. (C) The LC then cleaves its target SNARE protein (SNAP-25), preventing vesicular fusion and acetylcholine (ACh) secretion into the neuromuscular junction. BoNT/A is the most well-characterised serotype, with currently ten subtypes identi- fied through a blast pairwise alignment [66]. The first eight subtypes vary in amino acid sequence between 3 and 16% [67]. Although the differences in amino acid sequence across BoNT/A subtypes are small, they have been associated with significant variation in tox- icity. For BoNT subtypes /A1 to /A6, both cell entry rates and enzyme kinetics have been shown to differ, but the duration of action (measured by the time in which SNAP-25 hy- drolysis in primary rat spinal cord cells exposed to BoNT stops) is due to the LC domain alone [68]. BoNT subtypes /A1, /A2, /A4, /A5, and /A6 action persists for >10 months, whereas /A3 lasts much shorter, only 3 months [69,70]. This long duration of action is thought to be partly due to the ubiquitin proteasome system, which removes ubiquitin from the LC inside neuronal cells, preventing ubiquitin-dependent protein degradation [71]. Although the duration of action is comparable across the subtypes (with the excep- tion of A3), the extent of SNAP-25 hydrolysis, and therefore potency, has been shown to vary [72–75]. As the catalytic residues are 100% conserved, this difference in potency is likely driven by exosites within the LC or by other domains [67,76]. Compared to BoNT/A1, BoNT/A2 displays a similar potency in mouse models [77], however, it enters cells faster due to differences in both the HN and HC domains [38,77]. Figure 2. Modular mechanism of BoNT/A activity. (A) On arrival at the neuromuscular junction (NMJ), the cell-binding domain binds to both a ganglioside and SV2 on the surface of motor neurons where it is then internalised by receptor-mediated endocytosis. (B) The acidic environment of the endosome results in conformational change of the translocation domain which facilitates the passage of the LC through the endosomal membrane into the cytosol. Cytosolic thioredoxin (Trx) and thioredoxin reductase (TrxR) reduce the disulphide bond connecting the LC to the HC, releasing the LC into the cytosol. (C) The LC then cleaves its target SNARE protein (SNAP-25), preventing vesicular fusion and acetylcholine (ACh) secretion into the neuromuscular junction. BoNT/A is the most well-characterised serotype, with currently ten subtypes identi- fied through a blast pairwise alignment [66]. The first eight subtypes vary in amino acid sequence between 3 and 16% [67]. Although the differences in amino acid sequence across BoNT/A subtypes are small, they have been associated with significant variation in toxicity. For BoNT subtypes /A1 to /A6, both cell entry rates and enzyme kinetics have been shown to differ, but the duration of action (measured by the time in which SNAP-25 hydrolysis in primary rat spinal cord cells exposed to BoNT stops) is due to the LC domain alone [68]. BoNT subtypes /A1, /A2, /A4, /A5, and /A6 action persists for >10 months, whereas /A3 lasts much shorter, only 3 months [69,70]. This long duration of action is thought to be partly due to the ubiquitin proteasome system, which removes ubiquitin from the LC inside neuronal cells, preventing ubiquitin-dependent protein degradation [71]. Although the duration of action is comparable across the subtypes (with the exception of A3), the extent of SNAP-25 hydrolysis, and therefore potency, has been shown to vary [72–75]. As the catalytic residues are 100% conserved, this difference in potency is likely driven by exosites within the LC or by other domains [67,76].
Toxins 2023, 15, 92 5 of 22 Compared to BoNT/A1, BoNT/A2 displays a similar potency in mouse models [77], however, it enters cells faster due to differences in both the HN and HC domains [38,77]. BoNT/A2 also has a higher affinity for gangliosides than BoNT/A1 [39] and remains more localised with faster onset of paralysis compared to BoNT subtypes /A1, /A3, /A4, and /A5 [78]. Although the potency of BoNT/A1 and BoNT/A2 are similar in vivo, the catalytic activity of BoNT/A2 in vitro is five times lower than that of BoNT/A1 [72]. Not only does BoNT/A3 display the shortest duration of action, it also has the lowest potency in cells, compared to the other subtypes [69,72,74]. Compared to BoNT/A1, both BoNT/A3 and BoNT/A4 display a 2-fold and 1000-fold lower activity in mouse models due to less efficient cell entry [69,72]. BoNT/A5 showed similar potency in mice but was less potent in human cell models compared to BoNT/A1 [72], suggesting a possible variance in activity within subtypes across species. For BoNT/A6, potency was higher than BoNT/A1 with faster entry kinetics, and like BoNT/A2 [69,72,78] the onset of SNAP-25 cleavage by BoNT/A6 occurred more rapidly than by /A1 in neurons [70]. However, unlike both BoNT/A1 and BoNT/A2, BoNT/A6 cleaved more SNAP-25 [70]. Considering that the LC activities of BoNT/A6 and BoNT/A1 in vitro are similar, the earlier onset of BoNT/A6 activity is likely due to more efficient cell entry [70]. However, it is interesting to note that, although BoNT/A2 and BoNT/A6 have a similar and increased potency compared to BoNT/A1 in rat and human cell models, they required a two times higher toxin dose upon local intramuscular injection to be lethal in mice [70,75]. Both BoNT/A7 and BoNT/A8 show similar lethal activity in mice compared to BoNT/A1 [79]; however, the latter was also shown to have reduced ganglioside binding and lower enzymatic activity in vitro compared to /A1 [80]. These studies highlight the importance of determining precise differences in function- ality across subtypes that arise due to minor changes in amino acid sequence. The HC in particular, has been shown to be a ‘hotspot’ of this variation (Figure 3) [81], with the largest variation occurring within the N- and C-termini. X-ray crystallography is one strategy that may help deduce the cause of these functional differences.Toxins 2023, 15, 92 5 of 23 BoNT/A2 also has a higher affinity for gangliosides than BoNT/A1 [39] and remains more localised with faster onset of paralysis compared to BoNT subtypes /A1, /A3, /A4, and /A5 [78]. Although the potency of BoNT/A1 and BoNT/A2 are similar in vivo, the catalytic activity of BoNT/A2 in vitro is five times lower than that of BoNT/A1 [72]. Not only does BoNT/A3 display the shortest duration of action, it also has the lowest potency in cells, compared to the other subtypes [69,72,74]. Compared to BoNT/A1, both BoNT/A3 and BoNT/A4 display a 2-fold and 1000-fold lower activity in mouse models due to less effi- cient cell entry [69,72]. BoNT/A5 showed similar potency in mice but was less potent in human cell models compared to BoNT/A1 [72], suggesting a possible variance in activity within subtypes across species. For BoNT/A6, potency was higher than BoNT/A1 with faster entry kinetics, and like BoNT/A2 [69,72,78] the onset of SNAP-25 cleavage by BoNT/A6 occurred more rapidly than by /A1 in neurons [70]. However, unlike both BoNT/A1 and BoNT/A2, BoNT/A6 cleaved more SNAP-25 [70]. Considering that the LC activities of BoNT/A6 and BoNT/A1 in vitro are similar, the earlier onset of BoNT/A6 ac- tivity is likely due to more efficient cell entry [70]. However, it is interesting to note that, although BoNT/A2 and BoNT/A6 have a similar and increased potency compared to BoNT/A1 in rat and human cell models, they required a two times higher toxin dose upon local intramuscular injection to be lethal in mice [70,75]. Both BoNT/A7 and BoNT/A8 show similar lethal activity in mice compared to BoNT/A1 [79]; however, the latter was also shown to have reduced ganglioside binding and lower enzymatic activity in vitro compared to /A1 [80]. These studies highlight the importance of determining precise differences in func- tionality across subtypes that arise due to minor changes in amino acid sequence. The HC in particular, has been shown to be a ‘hotspot’ of this variation (Figure 3) [81], with the largest variation occurring within the N- and C-termini. X-ray crystallography is one strat- egy that may help deduce the cause of these functional differences. Figure 3. Cont.
Toxins 2023, 15, 92 6 of 22Toxins 2023, 15, 92 6 of 23 Figure 3. Sequence alignment of BoNT/A1 to /A8 cell-binding domains. Regions shown in white text with a red background indicate 100% identity. Regions/residues with high sequence similarity are shown in red text with a white background. Residues that lack both identity and similarity are shown in black text with a white background. Secondary structural elements are labelled α for α- helix, β for β-strand, and TT for turn. The residue numbering and secondary structural features are with respect to BoNT/A1 (3BTA), meaning the residue numbering for BoNT/A3 and /A4 is offset by +4 and -6, respectively. UniprotKB accession numbers in parentheses for the following BoNT/A sub- types: BoNT/A1 (P0DPI1), BoNT/A2 (Q45894), BoNT/A3 (D3IV24), BoNT/A4 (Q3LRX8), BoNT/A5 (C7BEA8), BoNT/A6 (C9WWY7), BoNT/A7 (K4LN57), and BoNT/A8 (A0A0A7PDB7). The wealth of structural data available on BoNT/A subtype cell-binding domains (HC/A1 to HC/A6) alone and in complex with their receptors have revealed features related to SV2 and ganglioside receptor binding [37,82–92]. In particular, the structure of HC/A1 in complex with human glycosylated SV2C [37] and of HC/A1 to HC/A6 in complex with the ganglioside GD1a (GM1b for HC/A5) [85–90] (Table 1) have identified six structural features that appear to be common to the cell-binding domain (Figure 4). Figure 3. Sequence alignment of BoNT/A1 to /A8 cell-binding domains. Regions shown in white text with a red background indicate 100% identity. Regions/residues with high sequence similarity are shown in red text with a white background. Residues that lack both identity and similarity are shown in black text with a white background. Secondary structural elements are labelled α for α-helix, β for β-strand, and TT for turn. The residue numbering and secondary structural features are with respect to BoNT/A1 (3BTA), meaning the residue numbering for BoNT/A3 and /A4 is offset by +4 and -6, respectively. UniprotKB accession numbers in parentheses for the following BoNT/A subtypes: BoNT/A1 (P0DPI1), BoNT/A2 (Q45894), BoNT/A3 (D3IV24), BoNT/A4 (Q3LRX8), BoNT/A5 (C7BEA8), BoNT/A6 (C9WWY7), BoNT/A7 (K4LN57), and BoNT/A8 (A0A0A7PDB7). The wealth of structural data available on BoNT/A subtype cell-binding domains (HC/A1 to HC/A6) alone and in complex with their receptors have revealed features related to SV2 and ganglioside receptor binding [37,82–92]. In particular, the structure of HC/A1 in complex with human glycosylated SV2C [37] and of HC/A1 to HC/A6 in complex with the ganglioside GD1a (GM1b for HC/A5) [85–90] (Table 1) have identified six structural features that appear to be common to the cell-binding domain (Figure 4). Table 1. List of BoNT/A subtype cell-binding domains and associated RCSB PDB codes. A compre- hensive analysis of the structures listed here forms the basis of this review. Structure PDB Code Resolution HC/A1 2VUA 1.70 Å HC/A1:GT1b 2VU9 1.60 Å HC/A1:GD1a 5TPC 2.00 Å
Toxins 2023, 15, 92 7 of 22 Table 1. Cont. Structure PDB Code Resolution HC/A1:gSV2C 5JLV 2.00 Å HC/A1:SV2C 4JRA 2.30 Å HC/A2 7Z5T 1.63 Å HC/A2:GD1a 7Z5S 2.10 Å HC/A2:SV2 6ES1 2.00 Å HC/A2:SV2 5MOY 2.30 Å HC/A3 6F0O 1.60 Å HC/A3:GD1a 6THY 1.75 Å HC/A4 6F0P 1.34 Å HC/A4:GD1a 7QPT 2.30 Å HC/A5 6TWP 1.15 Å HC/A5:GM1b 7QPU 2.40 Å HC/A6 (crystal form I) 6TWO 1.35 Å HC/A6 (crystal form II) 8ALP 1.50 Å HC/A6:GD1a 8AGK 1.50 ÅToxins 2023, 15, 92 7 of 23 Table 1. List of BoNT/A subtype cell-binding domains and associated RCSB PDB codes. A compre- hensive analysis of the structures listed here forms the basis of this review. Structure PDB code Resolution HC/A1 2VUA 1.70 Å HC/A1:GT1b 2VU9 1.60 Å HC/A1:GD1a 5TPC 2.00 Å HC/A1:gSV2C 5JLV 2.00 Å HC/A1:SV2C 4JRA 2.30 Å HC/A2 7Z5T 1.63 Å HC/A2:GD1a 7Z5S 2.10 Å HC/A2:SV2 6ES1 2.00 Å HC/A2:SV2 5MOY 2.30 Å HC/A3 6F0O 1.60 Å HC/A3:GD1a 6THY 1.75 Å HC/A4 6F0P 1.34 Å HC/A4:GD1a 7QPT 2.30 Å HC/A5 6TWP 1.15 Å HC/A5:GM1b 7QPU 2.40 Å HC/A6 (crystal form I) 6TWO 1.35 Å HC/A6 (crystal form II) 8ALP 1.50 Å HC/A6:GD1a 8AGK 1.50 Å Figure 4. Features of BoNT/A cell-binding domain. The crystal structures of binding domain sub- types, HC/A1 to HC/A6, have allowed for the identification of six features of cell binding: The gan- Figure 4. Features of BoNT/A cell-binding domain. The crystal structures of binding domain subtypes, HC/A1 to HC/A6, have allowed for the identification of six features of cell binding: The ganglioside binding site (yellow), SV2 binding site (blue), Lys–Cys/Cys–Cys bridge (magenta), two dynamic loop regions (1220-1240 (red) and 1268-1276 (cyan)), and a hinge rotation axis between the HCN and HCC domains. The structure of HC/A1 in complex with GT1b (PDB 2VU9) [90] and SV2 (PDB 5JLV) [37] is used here to show dual receptor binding and illustrate the features.
Toxins 2023, 15, 92 8 of 22 2. Cell-Binding Domain 2.1. Ganglioside Binding Site Gangliosides are ubiquitous within cell membranes, consisting of a membrane-bound lipid tail and an extracellular glycan ‘head-group’ [93]. Four gangliosides (GM1, GD1a, GT1b, and GD1b) (Figure 5) constitute 90% of all gangliosides in mammalian cells [94], and HC/A is known to bind both GT1b and GD1a [95].glioside binding site (yellow), SV2 binding site (blue), Lys–Cys/Cys–Cys bridge (magenta), two dy- namic loop regions (1220-1240 (red) and 1268-1276 (cyan)), and a hinge rotation axis between the HCN and HCC domains. The structure of HC/A1 in complex with GT1b (PDB 2VU9) [90] and SV2 (PDB 5JLV) [37] is used here to show dual receptor binding and illustrate the features. 2. Cell-binding Domain 2.1. Ganglioside Binding Site Gangliosides are ubiquitous within cell membranes, consisting of a membrane- bound lipid tail and an extracellular glycan ‘head-group’ [93]. Four gangliosides (GM1, GD1a, GT1b, and GD1b) (Figure 5) constitute 90% of all gangliosides in mammalian cells [94], and HC/A is known to bind both GT1b and GD1a [95]. Figure 5. Common neuronal gangliosides. Gangliosides comprise a hydrophobic lipid tail and a hydrophilic sugar moiety (represented by 1-6). The schematic representation is of the most common Figure 5. Common neuronal gangliosides. Gangliosides comprise a hydrophobic lipid tail and a hydrophilic sugar moiety (represented by 1–6). The schematic representation is of the most common neuronal gangliosides, GT1b, GD1a, GM1, and GD1b, illustrating the difference in sialic acid composition. Blue circle = glucose, yellow circle = galactose, yellow square = N-acetylglucosamine, purple diamond = sialic acid. Binding occurs within a β-hairpin at the C-terminus of the HCC subdomain (Figures 4 and 6) [85–90], a region referred to as the ganglioside binding site (GBS), which is formed partly by a conserved ‘H . . . SxWY . . . G’ peptide motif [67] (Figures 3 and 6).
Toxins 2023, 15, 92 9 of 22position. Blue circle = glucose, yellow circle = galactose, yellow square = N-acetylglucosamine, pur- ple diamond = sialic acid. Binding occurs within a β-hairpin at the C-terminus of the HCC subdomain (Figures 4 and 6) [85–90], a region referred to as the ganglioside binding site (GBS), which is formed partly by a conserved ‘H…SxWY…G’ peptide motif [67] (Figures 3 and 6). Figure 6. Ganglioside binding site of BoNT subtypes /A1 to /A6. (A) Table of ganglioside binding residues for HC/A1 through to HC/A6, with equivalent residues aligned. Green highlights binding residues that are observed across all subtypes, and red highlights non-binding residues. The under- lined tryptophan indicates 𝜋-stacking interaction with the ganglioside rather than a hydrogen bond. Hydrogen bond distances (Å ) are shown in parentheses, where for water-mediated interac- tions the first length is the protein-water distance and the second is the water-oligosaccharide dis- tance. * The number of hydrogen bonds includes water-mediated interactions as a ‘single’ hydrogen bond. (B) Structural changes that occur at the amino acid level upon binding ganglioside (sugar moieties are as in Fig 5). The apo structures of each subtype are displayed in green and the gangli- oside-bound structures in magenta. The arrows in HC/A2, HC/A3, HC/A5, and HC/A6 indicate a wid- ening of the 1268–1276 loop and associated flip between residues 1277 (1273A3) and 1278 (1274A3) Figure 6. Ganglioside binding site of BoNT subtypes /A1 to /A6. (A) Table of ganglioside binding residues for HC/A1 through to HC/A6, with equivalent residues aligned. Green highlights bind- ing residues that are observed across all subtypes, and red highlights non-binding residues. The underlined tryptophan indicates π-stacking interaction with the ganglioside rather than a hydrogen bond. Hydrogen bond distances (Å) are shown in parentheses, where for water-mediated interactions the first length is the protein-water distance and the second is the water-oligosaccharide distance. * The number of hydrogen bonds includes water-mediated interactions as a ‘single’ hydrogen bond. (B) Structural changes that occur at the amino acid level upon binding ganglioside (sugar moieties are as in Figure 5). The apo structures of each subtype are displayed in green and the ganglioside-bound structures in magenta. The arrows in HC/A2, HC/A3, HC/A5, and HC/A6 indicate a widening of the 1268–1276 loop and associated flip between residues 1277 (1273A3) and 1278 (1274A3) upon ganglioside binding. PDB codes are in parentheses for the following structures: HC/A1 (2VUA) [90], HC/A1:GT1b (2VU9) [90] HC/A1:GD1a (5TPC) [89], HC/A2 (7Z5T) [85], HC/A2:GD1a (7Z5S) [85], HC/A3 (6F0O) [82], HC/A3:GD1a (6THY) [86], HC/A4 (6F0P) [82], HC/A4:GD1a (7QPT) [87], HC/A5 (6TWP) [84], HC/A5:GM1b (7QPU) [87], HC/A6 (6TWO) [84], HC/A6:GD1a (8AGK) [88].
Toxins 2023, 15, 92 10 of 22 The HC/A1:GT1b, HC/A1:GD1a, HC/A2:GD1a, HC/A3:GD1a, HC/A4:GD1a, HC/A5:GM1b, and HC/A6:GD1a structures [85–90] (Table 1) have revealed the precise molecular inter- actions present across the HC/A:ganglioside interface. Six binding residues across the subtypes analysed here (HC/A1 to HC/A6) are conserved (Figure 6A), with some variation depending on the ganglioside. These residues are also conserved for HC/A7 and HC/A8, which suggests that they may also be involved with ganglioside binding (Figure 3) [83]. HC/A1 has been co-crystallised with both GT1b and GD1a, and these structures show a difference in binding residues (Figure 6A). Although the monosaccharides involved in binding are identical (Figures 5 and 6A), HC/A1 has a higher affinity for GT1b than GD1a (GT1b > GD1a > GM1) [95]. Given that GT1b and GD1a differ by only one monosaccharide (Figure 5), the variability in binding has been proposed to be due to the additional sialic acid (Sia7) in GT1b enhancing the rigidity of Sia6 and altering the oligosaccharide torsion angles [89]. In addition, HC/A1 forms 15 hydrogen bonds (including water-mediated hydrogen bonds) with GT1b, but only 10 with GD1a [89,90] (Figure 6A). Following on from this, the relative affinity that each HC/A subtype has for GD1a may be inferred from the number of hydrogen bonds present across the interface, which suggests that HC/A1 and HC/A2 possess the highest affinity for GD1a (A1/A2 > A3/A4 > A6) [85–89]. Interestingly, two separate investigations of HC/A1 ganglioside binding showed variable affinity for GD1a—0.6 μM for the entire GD1a molecule [95] and 1 μM for the sugar moiety only [89]. This indicates that the lipid itself may contribute to the binding affinity of HC/A for gangliosides. The GBS can be divided into three subsites ‘A’, ‘B’, and ‘C’ (Figure 6A) [89]. Subsite ‘A’ is occupied by Sia5 but the orientation of this sugar group within the subsite varies across the HC/A subtypes. In the HC/A2:GD1a, HC/A3:GD1a, HC/A4:GD1a, HC/A5:GM1b, and HC/A6:GD1a structures (Table 1), Sia5 is oriented perpendicular to Gal4, whereas in HC/A1:GD1a it is linear [85–90] (Figure 6B). This difference in Sia5 positioning can be attributed to several changes across the subtypes. HC/A1:GT1b forms five hydrogen bonds with sia5 [90], whereas HC/A1:GD1a [89] and HC/A4 [87] form four, HC/A2 [85], HC/A3 [86], and HC/A5 [87] form three, and HC/A6 forms two [88] (Figure 6A). Subsite ‘A’ also has amino acid variations at two sites corresponding to Tyr 1117 of HC/A1 (Tyr 1117A5, Tyr 1123A4, Phe 1117A2/A6, Phe 1113A3) and Gln 1254 of HC/A1 (Gln 1254A5, Gln 1260A4, Leu 1254A2, Leu 1250A3, Lys 1254A6) (Figures 3 and 6A). The presence of a Tyr residue in the first site explains the higher hydrogen bond count for HC/A1, HC/A4, and HC/A5, because it forms a hydrogen bond with Sia5 through its side-chain hydroxyl group. The variation in hydrogen bonding to Arg 1276A1/A2/A5/A6/1272A3/1282A4, may also contribute to this difference in Sia5 orientation. In the HC/A2:GD1a, HC/A3:GD1a, HC/A5:GM1b, and HC/A6:GD1a structures (Table 1), this Arg residue does not contribute to hydrogen bonding, whereas for HC/A4:GD1a, the hydrogen bond involves the amine nitrogen atom, and for HC/A1:GD1a and HC/A1:GT1b, the hydrogen bond is via the carbonyl oxygen atom of Arg 1276 through a water-mediated interaction [85–88]. Considering that this Arg residue is conserved in all subtypes (Figure 3) but does not contribute to ganglioside binding for each subtype, ganglioside-binding residues cannot be inferred from amino acid sequence comparison alone. Further to these specific amino acid variations, a dynamic loop region corresponding to HC/A1 residues 1268-1276 also contributes to differences in binding across the subtypes. This loop forms part of subsite ‘A’ and has the highest sequence variability within the GBS (Figure 3). It has been shown to widen upon ganglioside binding for HC/A2, HC/A3, HC/A5, and HC/A6 (Figure 6B, arrows), along with an associated flip at residues Thr 1277A2/A5/A6/1273A3 and Phe 1278A2/A5/A6/1274A3. The latter rotates towards the GBS and contributes to the formation of a hydrophobic pocket with Phe 1117A2/A6/1113A3 (Tyr 1117A5) and Phe 1252A2 (Figure 6B). The equivalent residue in HC/A1 and HC/A4 (Leu 1278A1/1284A4) does not flip; however, the position of this residue does differ between the subtypes. Prior to binding GD1a, Leu 1284A4 is already positioned within subsite ‘A’, contributing to the hydrophobic pocket, whereas for HC/A1, the Leu 1278A1 does not
Toxins 2023, 15, 92 11 of 22 alter conformation upon binding GD1a and does not contribute to the hydrophobic pocket. Interestingly, in the HC/A1:GT1b structure Leu 1278A1 does flip towards the binding site [90]. Together, these differences demonstrate both the flexibility of the ganglioside moiety and variability of subsite ‘A’ across HC/A subtypes. Five of the six totally conserved residues form subsite ‘B’ which is occupied by Gal4 upon ganglioside binding (Figure 6A). Gal4 is sandwiched between His 1253A1/A2/A5/A6/ 1249A3/1259A4 and Trp 1266A1/A2/A5/A6/1262A3/1272A4 forming π-stacking interactions, and hydrogen bonds to Glu 1203A1/A2/A5/A6/1199A3/1209A4, Phe 1252A1/A2/A5/A6/1248A3/ 1258A4, His 1253A1/A2/A5/A6/1249A3/1259A4, and Ser 1264A1/A2/A5/A6/1260A3/1270A4 (Figure 6B). An additional hydrogen bond to Gal4 is observed in the HC/A1:GT1b, HC/A2:GD1a, and HC/A3:GD1a structures through a water-mediated interaction with Gln 1254A1 (Leu 1254A2/1250A3). This large number of hydrogen bonding with its sugar group and low average B-factor across all HC/A:ganglioside structures [85–90] makes subsite ‘B’ the most stable and tightly bound pocket for a ganglioside monosaccharide. Together, subsites ‘A’ and ‘B’ appear to provide high specificity for glycans containing a terminal Sia–Gal moiety [89]. Subsite ‘C’ binds GalNAc3, forming one hydrogen bond via Glu 1203A1/A2/A5/A6/1199A3/ 1209A4 in the GD1a bound structures, and an additional water-mediated interaction in the HC/A1:GT1b structure (Figure 6A). The monosaccharide, Sia6, does not occupy a subsite but it does form a weak hydrogen bond with Trp 1266 in the HC/A1:GD1a/GT1b and HC/A2:GD1a structures. It is possible that this interaction occurs for binding sites HC/A3 to HC/A6 but has not been observed in the crystal structures because of the weak nature of the indole–hydrogen bond [96]. 2.2. SV2 Binding Site Synaptic vesicle glycoprotein 2 isoforms (SV2A, B, and C) are a small family of secre- tory vesicle glycoproteins present in all synapses, they consist of 12 transmembrane helices and a large extracellular domain with three N-linked glycosylation sites [97]. SV2 proteins are multifunctional as they are involved in the modulation of exocytosis and are essential for normal function of the nervous and endocrine systems [98]. Overexpression of SV2A and SC2C in insulin-secreting cells reduces glucose-induced secretion [99], whereas loss of SV2B results in a reduction in neurotransmission [100]. BoNT/A can bind to the extracellular domain of all three isoforms of SV2 (SV2A, SV2B, and SV2C), recognising both the protein (Figure 7) and glycan moiety [41] (Figure 8). HC/A has the highest affinity for SV2C [101], and binding of HC/A1 and HC/A2 to the protein moiety has been characterised through X-ray crystallography [37,91,92,102]. Three residues of HC/A1 and HC/A2 (1142, 1144, and 1146) contribute to an extended β-sheet stacking interaction with the SV2C protein moiety (Phe 557, Asn 559, Cys 560, and Phe 562) (Figure 7A) [37,91,92,102]. Considering that these residues are not conserved across the HC/A subtypes, (Figures 3 and 7A), they are unlikely to contribute to any significant difference in SV2 binding. There are, however, two highly conserved residues in all HC/A subtypes (Thr 1145A1/A2/A5/A6/1141A3/1151A4 and Thr 1146A1/A2/A5/A6/1142A3/1152A4), that are essential for HC/A1- and HC/A2-binding of the SV2C protein moiety via their side-chain atoms (Figure 7B) [102]. Away from the extended β-sheet stacking interaction, there are additional residues that form hydrogen bonds through their side-chain atoms, and some of these residues are conserved across the other subtypes (Figures 3 and 7A). In HC/A1, Arg 1156A1 forms a unique π-stacking interaction with Phe 563 of SVC, which has been shown to contribute to SV2C binding affinity [102]. In HC/A2:SV2C, Arg 1156A1 is substituted by Glu 1156A2, which forms a hydrogen bond with His 564 instead (Figure 7B). However, these few and varying interactions that HC/A1 and HC/A2 have with the SV2C protein moiety do not explain their high affinity and specificity for SV2 [37,102].
Toxins 2023, 15, 92 12 of 2223, 15, 92 12 of 23 Glu 1156A2, which forms a hydrogen bond with His 564 instead (Figure 7B). However, these few and varying interactions that HC/A1 and HC/A2 have with the SV2C protein moiety do not explain their high affinity and specificity for SV2 [37,102]. Figure 7. SV2 protein-moiety binding site. (A) Table listing the SV2 residues that interact with the binding site in HC/A1 and HC/A2. Interacting residues are aligned. The equivalent residues in HC/A3 to HC/A6 are also shown to highlight a putative SV2 binding site (conserved residues are displayed in green). (B) Structural differences of all interacting residues (ball and stick) between HC/A1, and HC/A2, the apo structures are shown in green (HC/A1 [PDB 2VUA] and HC/A2 [PDB 7Z5T]), and the SV2-bound structures in pink (HC/A1:gSV2C (PDB 5JLV) and HC/A1:SV2C [PDB 5MOY]) and or- ange (HC/A1:SV2 [PDB 4JRA], and HC/A2 [PDB 6ES1]). Putative SV2 protein-moiety binding site of HC/A3 to HC/A6 are also shown. PDB codes are in parentheses for the following structures: HC/A1 Figure 7. SV2 protein-moiety binding site. (A) Table listing the SV2 residues that interact with the binding site in HC/A1 and HC/A2. Interacting residues are aligned. The equivalent residues in HC/A3 to HC/A6 are also shown to highlight a putative SV2 binding site (conserved residues are displayed in green). (B) Structural differences of all interacting residues (ball and stick) between HC/A1, and HC/A2, the apo structures are shown in green (HC/A1 [PDB 2VUA] and HC/A2 [PDB 7Z5T]), and the SV2-bound structures in pink (HC/A1:gSV2C (PDB 5JLV) and HC/A1:SV2C [PDB 5MOY]) and orange (HC/A1:SV2 [PDB 4JRA], and HC/A2 [PDB 6ES1]). Putative SV2 protein-moiety binding site of HC/A3 to HC/A6 are also shown. PDB codes are in parentheses for the follow- ing structures: HC/A1 (2VUA) [90], HC/A1:gSV2C (5JLV) [37], HC/A2 (7Z5T) [85], HC/A2:SV2C (5MOY/6ES1) [91,92], HC/A3 (6F0O) [82], HC/A4 (6F0P) [82], HC/A5 (6TWP) [84], and HC/A6 (6TWO) [84].
Toxins 2023, 15, 92 13 of 22Toxins 2023, 15, 92 14 of 23 Figure 8. SV2 glycan-moiety binding site. (A) Table listing the HC/A1 glycan-binding residues, and equivalent residues in HC/A2 to HC/A6. Conserved residues are shown in green. (B) The SV2 glycan binding sites of HC/A1, and equivalent for HC/A2 to HC/A6. For HC/A1 the unbound (green) and SV2 bound (magenta) structures are overlayed. PDB codes are in parentheses for the following struc- tures: HC/A1 (2VUA) [90], HC/A1:gSV2C (5JLV) [37], HC/A2 (7Z5T) [85], HC/A2:SV2C (5MOY/6ES1) [91,92], HC/A3 (6F0O) [82], HC/A4 (6F0P) [82], HC/A5 (6TWP) [84], and HC/A6 (6TWO) [84]. Most of these interacting residues are conserved across the subtypes with the excep- tion of Asn 954, Ser 957, His 1064, and Gly 1292 (Figures 3 and 8). These differences in residues at the glycan binding site may alter binding and contribute to variation in affinity across BoNT/A subtypes. In particular, the replacement of His 1064A1 with an arginine in HC/A2, HC/A3, HC/A6, and HC/A8 (Arg1064A2/A6/A8/1060A3) makes them unable to form a π- stacking interaction with one of the GlcNAc molecules. Figure 8. SV2 glycan-moiety binding site. (A) Table listing the HC/A1 glycan-binding residues, and equivalent residues in HC/A2 to HC/A6. Conserved residues are shown in green. (B) The SV2 glycan binding sites of HC/A1, and equivalent for HC/A2 to HC/A6. For HC/A1 the unbound (green) and SV2 bound (magenta) structures are overlayed. PDB codes are in parentheses for the follow- ing structures: HC/A1 (2VUA) [90], HC/A1:gSV2C (5JLV) [37], HC/A2 (7Z5T) [85], HC/A2:SV2C (5MOY/6ES1) [91,92], HC/A3 (6F0O) [82], HC/A4 (6F0P) [82], HC/A5 (6TWP) [84], and HC/A6 (6TWO) [84]. The glycan-moiety binding site is crucial for high affinity binding of BoNT/A for SV2, because binding required glycosylated SV2A/B and enhanced binding for glycosylated SV2C [37,41]. The crystal structure of the BoNT/A1 binding domain in complex with glycosylated human SV2C (HC/A1:gSV2C) revealed a high number of interactions with an N-linked glycan on Asn-559 of gSV2C [37] (Figure 8A). This residue is conserved in all
Toxins 2023, 15, 92 14 of 22 vertebrate SV2-homologs. Two key residues involved in HC/A1 glycan binding are Phe 953 and His 1064, which form π-stacking interactions with two GlcNAc molecules of the Asn 559 glycan (Figure 8B). A further 12 residues (Phe 953, Asn 954, Ser 957, Asp 1062, His 1064, Arg 1065, Thr 1145, Tyr 1155, Asp 1288, Asp 1289, Gly 1292, Glu 1293) form a network of hydrogen bonds with the glycan through water-mediated interactions (Figure 8A). Most of these interacting residues are conserved across the subtypes with the exception of Asn 954, Ser 957, His 1064, and Gly 1292 (Figures 3 and 8). These differences in residues at the glycan binding site may alter binding and contribute to variation in affinity across BoNT/A subtypes. In particular, the replacement of His 1064A1 with an arginine in HC/A2, HC/A3, HC/A6, and HC/A8 (Arg1064A2/A6/A8/1060A3) makes them unable to form a π-stacking interaction with one of the GlcNAc molecules. Although the binding site of HC/A2 is fairly rigid, two different structures of HC/A2:SV2C revealed alternative ways of binding to the SV2C protein moiety (Figure 7B) [91,92]. This is likely due to the interaction involving predominantly main chain backbone atoms of HC/A2. Whether this plasticity is due to the lack of glycosylation remains unclear. It is possible that glycosylated Asn 559 may restrict the interaction that HC/A subtypes have with the SV2 protein moiety, therefore, non-glycosylated forms of SV2 may sometimes be misleading in how HC/A binds. In order to obtain a full structural understanding of receptor binding across all BoNT/A subtypes, detailed structures of HC/A2 through to HC/A8 in complex with glycosylated SV2 are required. 2.3. The Hinge Region A comparison of the HC/A1 to HC/A6 structures to their ganglioside-bound struc- tures [82–90] revealed a subtle hinge rotation between the HCN and HCC subdomains. Additionally, a new crystal form (crystal form II) of HC/A6 [88] revealed a much larger rotation of ~16.8◦ (Figure 9), which appears to prevent the formation of a β-stacking interaction between a portion of the HCN subdomain and the SV2 binding site of the HCC subdomain.Toxins 2023, 15, 92 15 of 23 Although the binding site of HC/A2 is fairly rigid, two different structures of HC/A2:SV2C revealed alternative ways of binding to the SV2C protein moiety (Figure 7B) [91,92]. This is likely due to the interaction involving predominantly main chain backbone atoms of HC/A2. Whether this plasticity is due to the lack of glycosylation remains unclear. It is possible that glycosylated Asn 559 may restrict the interaction that HC/A subtypes have with the SV2 protein moiety, therefore, non-glycosylated forms of SV2 may some- times be misleading in how HC/A binds. In order to obtain a full structural understanding of receptor binding across all BoNT/A subtypes, detailed structures of HC/A2 through to HC/A8 in complex with glycosylated SV2 are required. 2.3. The Hinge Region A comparison of the HC/A1 to HC/A6 structures to their ganglioside-bound structures [82–90] revealed a subtle hinge rotation between the HCN and HCC subdomains. Addition- ally, a new crystal form (crystal form II) of HC/A6 [88] revealed a much larger rotation of ~16.8° (Figure 9), which appears to prevent the formation of a β-stacking interaction be- tween a portion of the HCN subdomain and the SV2 binding site of the HCC subdomain. Figure 9. Hinge motion of HCN and HCC subdomains. Two crystal forms of HC/A6 revealed a 16.8° hinge motion between the HCN and HCC subdomains [88]. The HCN from both structures were aligned independently to highlight the difference in HCC orientation between HC/A6 crystal form I (green) (PDB 6TWO) [84] and HC/A6 crystal form II (magenta) (PDB 8ALP) [88]. The image on the left is orientated perpendicular to the rotation axis, with the image on the right a view at 90°. For the image on the right, the rotation axis is shown by the dotted line. The hinge residues are indicated by the red arrow and the difference in SV2 binding site-loop positioning between crystal form I (fully mod- elled) and II (partially modelled) is shown by the black arrow. This suggested that the flexibility of the SV2 binding site increased upon rotation, allowing it to protrude out towards the surface of the protein (Figure 9, black arrow) [88]. Although a large hinge rotation has only been observed in HC/A6, such a feature may exist across the subtypes [85,87,88]. The function of the hinge rotation is yet to be established, but it may contribute to the enhancement of SV2 binding or translocation. Upon binding to the dual receptor complex (Figure 1B), both the LC and HN need to reposition to be in Figure 9. Hinge motion of HCN and HCC subdomains. Two crystal forms of HC/A6 revealed a 16.8◦ hinge motion between the HCN and HCC subdomains [88]. The HCN from both structures were aligned independently to highlight the difference in HCC orientation between HC/A6 crystal form I (green) (PDB 6TWO
Toxins 2023, 15, 92 15 of 22 This suggested that the flexibility of the SV2 binding site increased upon rotation, allowing it to protrude out towards the surface of the protein (Figure 9, black arrow) [88]. Although a large hinge rotation has only been observed in HC/A6, such a feature may exist across the subtypes [85,87,88]. The function of the hinge rotation is yet to be established, but it may contribute to the enhancement of SV2 binding or translocation. Upon binding to the dual receptor complex (Figure 1B), both the LC and HN need to reposition to be in contact with the membrane to initiate translocation [36,49,51,90,103–105]. Additionally, the ‘BoNT-switch’, which is a stretch of 47 residues that alters conformation from an α-helix to a β-finger at acidic pH, would need to reorientate to come into contact with the cell membrane [106]. The hinge may therefore offer a region of flexibility that could accommodate this repositioning. Further structural studies of the toxin in the presence of a membrane may aid our understanding of the function of this hinge motion. Other than the new crystal form of HC/A6, crystallographic and in-solution structural studies of BoNT have revealed very little motion within the molecule [107]. In full-length structures of BoNT this is perhaps due to the regulatory function that each domain exhibits on one another, for example, the HC has been shown to reduce the pH range in which the HN domain forms channels in phospholipid bilayers [54,108], and the HN moderates the LCs catalytic activity [33]. It has been established that a change in pH alone is not sufficient to elicit a conformational change within the BoNT/B [109], and if this holds true for BoNT/A, elucidating binding of BoNT with its dual receptor complex at low pH may reveal more biologically relevant conformational changes. 2.4. Lys–Cys/Cys–Cys Bridge Between the Nζ of Lys 1236 and the Sγ of Cys 1280 in HC/A5, a continuation of electron density was observed, characteristic of a Lys–Cys bridge between the two atoms [84]. Similar findings were later observed in the crystal structures of HC/A2 [85] and HC/A6 [88]. A recent meta-analysis of structures deposited with the protein data bank revealed a large number of unreported Lys–O–Cys (Nζ-O-Sγ, NOS) bridges [110]; the first of which was identified in a transaldolase enzyme from Neisseria gonorrhoeae that demonstrated a redox- dependent ‘switch’-like nature [111]. Consequently, the bridging interaction in HC/A2 and HC/A6 (crystal form II) was characterised as a NOS bridge due to the apparent redox-dependency in the available crystal structures of HC/A2 [85] (Table 1). Although sub-angstrom resolution data are required to unequivocally determine the identity of the bridging atom [112], the current reported literature indicates that ‘O’ is the most probable connecting atom [110,111]. Synchrotron radiation (generally used for X-ray data collection in the crystal structure determination process) is known for its reducing properties, therefore, the presence of NOS bridges in HC/A subtypes is unlikely to be an artefact of data collection [111]. In fact, the opposite is true, radiation damage is likely to break NOS bridges [110]. It is theoretically possible for all HC/A subtypes to form a NOS bridge because both Lys 1236A1/A2/A5/A6/1232A3/1242A4 and Cys 1280A1/A2/A5/A5/A6/1276A3/1286A4 are con- served (Figure 3). To check if the NOS bridge has gone previously undetected in BoNT/A structures, we reviewed the presence of the NOS bridge in all the available crystal structures of HC/A subtypes (Table 1). If we consider that a continuation of electron density between Lys 1236A1/A2/A5/A6/1232A3/1242A4) and Cys 1280A1/A2/A5/A6/1276A3/1286A4 is indica- tive of a NOS bridge, the NOS bridge would appear to be the most common interaction (Figure 10A). This NOS bridge is likely to be dynamic because the Cys 1280A1/A2/A5/A6/ 1276A3/1286A4 residue may also form a disulphide bond with Cys 1235A1/A2/A5/A6/1231A3/ 1241A4 or not form a bridge at all (Figure 10B). In total, 57% of HC/A structures show a potential NOS bridge, 35% a Cys–Cys bridge, and 8.0% no bridge at all (Figure 10A). Interestingly, two crystal forms of HC/A1 were obtained from the same crystallisation condition [83]. One showed a Cys–Cys bridge and the other no bridge [83], indicating that the crystallisation condition does not dictate which interaction is observed in the crystal structure. The location of this bridging interaction is close to the GBS and con-
Toxins 2023, 15, 92 16 of 22 nects two flexible regions of the molecule: the ‘dynamic loop’ (1268-1276) within subsite ‘A’, and a β-hairpin at residues 1220-1240 (Figure 10B). This β-hairpin adopts different conformations across the subtypes depending on whether a NOS or a Cys–Cys bridge is present (Figure 10B). This difference is likely due to the different lengths of the bridging interaction—a Cys–Cys bridge appears to ‘pull’ the β-hairpin loop closer to the dynamic loop of subsite ‘A’ (Figure 10B). Residues corresponding to Lys 1236A1, Cys 1235A1, and Cys 1280A1 are conserved across the BoNT/A subtypes (Figure 3); however, only the equivalent Cys residues are conserved across other serotypes (BoNT/B, /E, /F, and /G) [83], that have been shown to affect humans. The NOS/Cys–Cys bridge, therefore, appears to be unique to all BoNTs with a HC/A binding domain. The function of the NOS/Cys–Cys bridge in BoNT/A is unknown, but its existence close to the GBS suggests it may have a biological function.023, 15, 92 17 of 23 Figure 10. The Lys–Cys/Cys–Cys bridge and associated structural change. (A) Table of available crystal structures of HC/A subtypes categorised by the possible presence of Lys 1236A1/A2/A5/A6/1232A3/1242A4-Cys 1280A1/A2/A5/A6/1276A3/1286A4, Cys 1280A1/A2/A5/A6/1276A3/1286A4-Cys 1235A1/A2/A5/A6/1231A3/1241A4, or no bridge at all. Structures possessing a Lys–Cys bridge were ob- served in the deposited coordinates for HC/A2 (7ZSQ/7Z5S), HC/A5 (6TWP) and HC/A6 (8ALP), whereas the other remaining structures were re-analysed and reported to show electron density between the Lys–Cys residue and structural changes consistent with the Lys–Cys bridge. (B) Struc- tural comparison of loops 1220-1240A1/A2/A5/A6 (1216-1236A3 and 1226-1246A4) and 1268-1276A1/A2/A5/A6 (1264-1272A3 and 1274-1282A4) associated with occurrence of either a Lys–Cys (green) or Cys–Cys (magenta) bridge. A Cys–Cys bridge appears to ‘pull’ the 1220-1240A1/A2/A5/A6 loop (1216-1236A3 and 1226-1246A4) (black arrows) towards the 1268-1276A1/A2/A5/A6 loop (1264-1272A3 and 1274-1282A4) (red arrow) and alter the direction of the β-hairpin slightly. The position of bound ganglioside for each subtype is shown to illustrate the proximity of the Lys–Cys/Cys–Cys bridge to the GBS (sugar moi- Figure 10. The Lys–Cys/Cys–Cys bridge and associated structural change. (A) Table of available crystal structures of HC/A subtypes categorised by the possible presence of Lys 1236A1/A2/A5/A6/ 1232A3/1242A4- Cys 1280A1/A2/A5/A6/1276A3/1286A4, Cys 1280A1/A2/A5/A6/1276A3/1286A4-Cys 1235A1/A2/A5/A6/ 1231A3/1241A4, or no bridge at all. Structures possessing a Lys–Cys bridge were observed in the deposited
Toxins 2023, 15, 92 17 of 22 coordinates for HC/A2 (7ZSQ/7Z5S), HC/A5 (6TWP) and HC/A6 (8ALP), whereas the other re- maining structures were re-analysed and reported to show electron density between the Lys–Cys residue and structural changes consistent with the Lys–Cys bridge. (B) Structural comparison of loops 1220-1240A1/A2/A5/A6 (1216-1236A3 and 1226-1246A4) and 1268-1276A1/A2/A5/A6 (1264-1272A3 and 1274-1282A4) associated with occurrence of either a Lys–Cys (green) or Cys–Cys (magenta) bridge. A Cys–Cys bridge appears to ‘pull’ the 1220-1240A1/A2/A5/A6 loop (1216-1236A3 and 1226-1246A4) (black arrows) towards the 1268-1276A1/A2/A5/A6 loop (1264-1272A3 and 1274-1282A4) (red arrow) and alter the direction of the β-hairpin slightly. The position of bound ganglioside for each subtype is shown to illustrate the proximity of the Lys–Cys/Cys–Cys bridge to the GBS (sugar moieties are as in Figure 5). The structures used to illustrate structural differences in the presence of either a Lys–Cys or Cys–Cys bridge are highlighted in Figure 10A: HC/A1 Lys–Cys (5TPC) [89], HC/A1 Cys–Cys (5MK7) [83], HC/A2 Lys–Cys (7Z5S) [85], HC/A2 Cys–Cys (5MOY) [92], HC/A3 Lys–Cys (6THY) [86], HC/A4 Cys–Cys (7QPT) [87], HC/A5 Lys–Cys (7QPU) [87], HC/A6 Lys–Cys (8ALP) [88], HC/A6 Cys–Cys (8AGK) [88]. 3. Conclusions and Future Perspectives The structural data of BoNT/A1 to /A6 subtype cell-binding domains (HC/A1 to HC/A6) have provided a wealth of information, not only on subtype-to-subtype variability, but also on how the cell-binding domain recognises its receptors and the implication of this on functional features. The crystal structures have allowed for characterisation of six features of cell binding: The ganglioside binding site (GBS), a dynamic loop, a possible NOS/Cys–Cys bridge, the SV2 protein and glycan binding sites, and the HCN– HCC subdomain hinge. The GBS consists of three subsites (‘A’, ‘B’, and ‘C’) that are able to accommodate gangliosides containing a terminal Sia–Gal–GalNAc moiety, with preference for GT1b and GD1a. A dynamic loop forms part of subsite ‘A’ and shows variation across the subtypes, appearing to widen upon GD1a (GM1b for HC/A5) binding in HC/A2, HC/A3, HC/A5, and HC/A6. Close to the GBS is a possible alternating bridging interaction unique to BoNT/A that switches between Lys–Cys and Cys–Cys, with some evidence of redox-dependency. The two cysteines are conserved across mammalian-targeting BoNT serotypes (BoNT/A, /B, /E, and /F), which may indicate a functional role in the mechanism of intoxication. Further experimental investigation is needed to establish if there is a definitive biological role. BoNT/A recognises its protein receptor (SV2) through both a protein and glycan moiety, with the latter portion appearing to contribute significantly to the high specificity and affinity of binding observed across BoNT/A subtypes. Each of these features is located within the HCC subdomain, whereas the function of the HCN still remains unclear. There appears to be a rotation axis between the two subdomains that might aid in SV2 binding or subsequent translocation. We highlight the importance of more biologically relevant structural data, in particular with reference to non-glycosylated SV2-bound structures that show limited information regarding the true nature of binding. With newly emerging serotypes and subtypes, the expansion of the available atomic data on receptor binding among BoNTs would prove beneficial for both the development of antibodies against BoNT, and the engineering of BoNT molecules for therapeutic purposes. Author Contributions: K.S.G. wrote the manuscript; K.R.A. edited and reviewed the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: K.S.G. is supported by a joint postgraduate studentship between the University of Bath and Ipsen Bioinnovation Limited. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Not applicable.
Toxins 2023, 15, 92 18 of 22 Acknowledgments: We thank Sai Man Liu from Ipsen Bioinnovation Ltd. for critical reading and constructive comments on the manuscript. Conflicts of Interest: K.S.G. and K.R.A. from the University of Bath declare no competing financial interests. All experimentally determined 3D structural data used in the analysis presented in this review were obtained from the RCSB Protein Data Bank and are based on published work from K. Ravi Acharya’s group (University of Bath, UK) and other researchers in the field. Appropriate citations and PDB codes are provided. Abbreviations ACh Acetylcholine BoNT/x Botulinum neurotoxin (/serotype) Gal Galactose GalNAc N-acetylglucosamine GBS Ganglioside binding site Glc Glucose HC/x Heavy chain of botulinum neurotoxin (/serotype or /subtype) HC/x Cell-binding domain of botulinum neurotoxin (/serotype or /subtype) HCC/x C-terminal cell-binding subdomain of botulinum neurotoxin (/serotype or /subtype) HCN/x N-terminal cell-binding subdomain of botulinum neurotoxin (/serotype or /subtype) HN/x Translocation domain of botulinum neurotoxin (/serotype or /subtype) LC/x Light chain of botulinum neurotoxin (/serotype or /subtype) NMJ Neuromuscular junction Sia Sialic acid SNAP-25 Synaptosomal-associated protein of 25 kDa SNARE Soluble N-ethylmaleimide-sensitive factor attachment protein receptor SV2x Synaptic vesicle glycoprotein 2 (x = A/B/C) Syt I/II Synaptotagmin (isoform I/II) Trx Thioredoxin TrxR Thioredoxin reductase VAMP Vesicle-associated membrane protein References 1. Fonfria, E.; Maignel, J.; Lezmi, S.; Martin, V.; Splevins, A.; Shubber, S.; Kalinichev, M.; Foster, K.; Picaut, P.; Krupp, J. The Expanding Therapeutic Utility of Botulinum Neurotoxins. Toxins 2018, 10, 208. [CrossRef] 2. Satriyasa, B.K. Botulinum Toxin (Botox) A for Reducing the Appearance of Facial Wrinkles: A Literature Review of Clinical Use and Pharmacological Aspect. Clin. Cosmet. Investig. Dermatol. 2019, 12, 223–228. [CrossRef] 3. Collins, M.D.; East, A.K. Phylogeny and Taxonomy of the Food-borne Pathogen Clostridium Botulinum and Its Neurotoxins. J. Appl. Microbiol. 1997, 84, 5–17. [CrossRef] 4. Carter, A.T.; Peck, M.W. Genomes, Neurotoxins and Biology of Clostridium Botulinum Group I and Group II. Res. Microbiol. 2015, 166, 303–317. [CrossRef] 5. Woudstra, C.; Mäklin, T.; Derman, Y.; Bano, L.; Skarin, H.; Mazuet, C.; Honkela, A.; Lindström, M. Closing Clostridium Botulinum Group III Genomes Using Long-Read Sequencing. Microbiol. Resour. Announc. 2021, 10, e0136420. [CrossRef] 6. Fillo, S.; Giordani, F.; Tonon, E.; Drigo, I.; Anselmo, A.; Fortunato, A.; Lista, F.; Bano, L. Extensive Genome Exploration of Clostridium Botulinum Group III Field Strains. Microorganisms 2021, 9, 2347. [CrossRef] 7. Masuyer, G.; Chaddock, J.A.; Foster, K.A.; Acharya, K.R. Engineered Botulinum Neurotoxins as New Therapeutics. Annu. Rev. Pharmacol. Toxicol. 2014, 54, 27–51. [CrossRef] 8. Zhang, S.; Masuyer, G.; Zhang, J.; Shen, Y.; Henriksson, L.; Miyashita, S.I.; Martínez-Carranza, M.; Dong, M.; Stenmark, P. Identification and Characterization of a Novel Botulinum Neurotoxin. Nat. Commun. 2017, 8, 14130. [CrossRef] 9. Masuyer, G.; Zhang, S.; Barkho, S.; Shen, Y.; Henriksson, L.; Košenina, S.; Dong, M.; Stenmark, P. Structural Characterisation of the Catalytic Domain of Botulinum Neurotoxin X-High Activity and Unique Substrate Specificity. Sci. Rep. 2018, 8, 4518. [CrossRef] 10. Fan, Y.; Barash, J.R.; Conrad, F.; Lou, J.; Tam, C.; Cheng, L.W.; Arnon, S.S.; Marks, J.D. The Novel Clostridial Neurotoxin Produced by Strain IBCA10-7060 Is Immunologically Equivalent to BoNT/HA. Toxins 2019, 12, 9. [CrossRef] 11. Nakamura, K.; Kohda, T.; Seto, Y.; Mukamoto, M.; Kozaki, S. Improved Detection Methods by Genetic and Immunological Techniques for Botulinum C/D and D/C Mosaic Neurotoxins. Vet. Microbiol. 2013, 162, 881–890. [CrossRef] [PubMed] 12. Gonzalez-Escalona, N.; Thirunavukkarasu, N.; Singh, A.; Toro, M.; Brown, E.W.; Zink, D.; Rummel, A.; Sharma, S.K. Draft Genome Sequence of Bivalent Clostridium Botulinum Strain IBCA10-7060, Encoding Botulinum Neurotoxin B and a New FA Mosaic Type. Genome. Announc. 2014, 2, e01275-14. [CrossRef] [PubMed]
Toxins 2023, 15, 92 19 of 22 13. Nakamura, K.; Kohda, T.; Umeda, K.; Yamamoto, H.; Mukamoto, M.; Kozaki, S. Characterization of the D/C Mosaic Neurotoxin Produced by Clostridium Botulinum Associated with Bovine Botulism in Japan. Vet. Microbiol. 2010, 140, 147–154. [CrossRef] [PubMed] 14. Tanizawa, Y.; Fujisawa, T.; Mochizuki, T.; Kaminuma, E.; Suzuki, Y.; Nakamura, Y.; Tohno, M. Draft Genome Sequence of Weissella Oryzae SG25T, Isolated from Fermented Rice Grains. Genome Announc. 2014, 2, e00667-14. [CrossRef] [PubMed] 15. Zornetta, I.; Azarnia Tehran, D.; Arrigoni, G.; Anniballi, F.; Bano, L.; Leka, O.; Zanotti, G.; Binz, T.; Montecucco, C. The First Non Clostridial Botulinum-like Toxin Cleaves VAMP within the Juxtamembrane Domain. Sci. Rep. 2016, 6, 30257. [CrossRef] [PubMed] 16. Košenina, S.; Masuyer, G.; Zhang, S.; Dong, M.; Stenmark, P. Crystal Structure of the Catalytic Domain of the Weissella Oryzae Botulinum-like Toxin. FEBS Lett. 2019, 593, 1403–1410. [CrossRef] 17. Brunt, J.; Carter, A.T.; Stringer, S.C.; Peck, M.W. Identification of a Novel Botulinum Neurotoxin Gene Cluster in Enterococcus. FEBS Lett. 2018, 592, 310–317. [CrossRef] 18. Zhang, S.; Lebreton, F.; Mansfield, M.J.; Miyashita, S.I.; Zhang, J.; Schwartzman, J.A.; Tao, L.; Masuyer, G.; Martínez-Carranza, M.; Stenmark, P.; et al. Identification of a Botulinum Neurotoxin-like Toxin in a Commensal Strain of Enterococcus Faecium. Cell Host Microbe 2018, 23, 169–176. [CrossRef] 19. Contreras, E.; Masuyer, G.; Qureshi, N.; Chawla, S.; Dhillon, H.S.; Lee, H.L.; Chen, J.; Stenmark, P.; Gill, S.S. A Neurotoxin That Specifically Targets Anopheles Mosquitoes. Nat. Commun. 2019, 10, 2869. [CrossRef] 20. Chen, Z.P.; Glenn Morris, J.; Rodriguez, R.L.; Shukla, A.W.; Tapia-Núñez, J.; Okun, M.S. Emerging Opportunities for Serotypes of Botulinum Neurotoxins. Toxins 2012, 4, 1196–1222. [CrossRef] 21. Hill, K.K.; Xie, G.; Foley, B.T.; Smith, T.J. Genetic Diversity within the Botulinum Neurotoxin-Producing Bacteria and Their Neurotoxins. Toxicon 2015, 107, 2–8. [CrossRef] 22. Hill, K.K.; Smith, T.J. Genetic Diversity within Clostridium Botulinum Serotypes, Botulinum Neurotoxin Gene Clusters and Toxin Subtypes. In Current Topics in Microbiology and Immnology; Springer: Berlin/Heidelberg, Germany, 2012; Volume 364, pp. 1–20; ISBN 9783540921646. 23. DasGupta, B.R. Botulinum Neurotoxins: Perspective on Their Existence and as Polyproteins Harboring Viral Proteases. J. Gen. Appl. Microbiol. 2006, 52, 1–8. [CrossRef] [PubMed] 24. Arndt, J.W.; Jacobson, M.J.; Abola, E.E.; Forsyth, C.M.; Tepp, W.H.; Marks, J.D.; Johnson, E.A.; Stevens, R.C. A Structural Perspective of the Sequence Variability within Botulinum Neurotoxin Subtypes A1-A4. J. Mol. Biol. 2006, 362, 733–742. [CrossRef] [PubMed] 25. Gibson, A.M.; Modi, N.K.; Roberts, T.A.; Hambleton, P.; Melling, J. Evaluation of a Monoclonal Antibody-Based Immunoassay for Detecting Type B Clostridium Botulinum Toxin Produced in Pure Culture and an Inoculated Model Cured Meat System. J. Appl. Bacteriol. 1988, 64, 285–291. [CrossRef] [PubMed] 26. Gibson, A.M.; Modi, N.K.; Roberts, T.A.; Shone, C.C.; Hambleton, P.; Melling, J. Evaluation of a Monoclonal Antibody-Based Immunoassay for Detecting Type A Clostridium Botulinum Toxin Produced in Pure Culture and an Inoculated Model Cured Meat System. J. Appl. Bacteriol. 1987, 63, 217–226. [CrossRef] 27. Kalb, S.R.; Lou, J.; Garcia-Rodriguez, C.; Geren, I.N.; Smith, T.J.; Moura, H.; Marks, J.D.; Smith, L.A.; Pirkle, J.L.; Barr, J.R. Extraction and Inhibition of Enzymatic Activity of Botulinum Neurotoxins/A1, /A2, and /A3 by a Panel of Monoclonal Anti-BoNT/A Antibodies. PLoS ONE 2009, 4, e5355. [CrossRef] 28. Kalb, S.R.; Santana, W.I.; Geren, I.N.; Garcia-Rodriguez, C.; Lou, J.; Smith, T.J.; Marks, J.D.; Smith, L.A.; Pirkle, J.L.; Barr, J.R. Extraction and Inhibition of Enzymatic Activity of Botulinum Neurotoxins /B1, /B2, /B3, /B4, and /B5 by a Panel of Monoclonal Anti-BoNT/B Antibodies. BMC Biochem. 2011, 12, 58. [CrossRef] 29. Kalb, S.R.; Baudys, J.; Webb, R.P.; Wright, P.; Smith, T.J.; Smith, L.A.; Fernández, R.; Raphael, B.H.; Maslanka, S.E.; Pirkle, J.L.; et al. Discovery of a Novel Enzymatic Cleavage Site for Botulinum Neurotoxin F5. FEBS Lett. 2012, 586, 109–115. [CrossRef] 30. Lacy, D.B.; Tepp, W.; Cohen, A.C.; DasGupta, B.R.; Stevens, R.C. Crystal Structure of Botulinum Neurotoxin Type A and Implications for Toxicity. Nat. Struct. Biol. 1998, 5, 898–902. [CrossRef] 31. DasGupta, B.R.; Sugiyama, H. Role of a Protease in Natural Activation of Clostridium Botulinum Neurotoxin. Infect. Immun. 1972, 6, 587–590. 32. Fischer, A.; Montal, M. Crucial Role of the Disulfide Bridge between Botulinum Neurotoxin Light and Heavy Chains in Protease Translocation across Membranes. J. Biol. Chem. 2007, 282, 29604–29611. [CrossRef] [PubMed] 33. Brunger, A.T.; Breidenbach, M.A.; Jin, R.; Fischer, A.; Santos, J.S.; Montal, M. Botulinum Neurotoxin Heavy Chain Belt as an Intramolecular Chaperonef
Toxins 2023, 15, 92 20 of 22 37. Yao, G.; Zhang, S.; Mahrhold, S.; Lam, K.H.; Stern, D.; Bagramyan, K.; Perry, K.; Kalkum, M.; Rummel, A.; Dong, M.; et al. N-Linked Glycosylation of SV2 Is Required for Binding and Uptake of Botulinum Neurotoxin A. Nat. Struct. Mol. Biol. 2016, 23, 656–662. [CrossRef] 38. Sun, S.; Tepp, W.H.; Johnson, E.A.; Chapman, E.R. Botulinum Neurotoxins B and E Translocate at Different Rates and Exhibit Divergent Responses to GT1b and Low PH. Biochemistry 2012, 51, 5655–5662. [CrossRef] 39. Kroken, A.R.; Blum, F.C.; Zuverink, M.; Barbieri, J.T. Entry of Botulinum Neurotoxin Subtypes A1 and A2 into Neurons. Infect. Immun. 2017, 85, e00795-16. [CrossRef] 40. Foster, K.A. The Dual-Receptor Recognition of Botulinum Neurotoxins. In Molecular Aspects of Botulinum Neurotoxin; Springer: New York, NY, USA, 2014; pp. 129–149; ISBN 9781461494546. 41. Dong, M.; Yeh, F.; Tep, W.H.; Chapman, P.; Dean, C.; Johnson, E.A.; Janz, R.; Chapman, E.R. SV2 Is the Protein Receptor for Botulinum Neurotoxin A. Science 2006, 312, 592–596. [CrossRef] 42. Peng, L.; Tepp, W.H.; Johnson, E.A.; Dong, M. Botulinum Neurotoxin D Uses Synaptic Vesicle Protein SV2 and Gangliosides as Receptors. PLoS Pathog. 2011, 7, e1002008. [CrossRef] 43. Dong, M.; Liu, H.; Tepp, W.H.; Johnson, E.A.; Janz, R.; Chapman, E.R. Glycosylated SV2A and SV2B Mediate the Entry of Botulinum Neurotoxin E into Neurons. Mol. Biol. Cell 2008, 19, 5226–5237. [CrossRef] [PubMed] 44. Fu, Z.; Chen, C.; Barbieri, J.T.; Kim, J.J.P.; Baldwin, M.R. Glycosylated SV2 and Gangliosides as Dual Receptors for Botulinum Neurotoxin Serotype F. Biochemistry 2009, 48, 5631–5641. [CrossRef] [PubMed] 45. Dong, M.; Richards, D.A.; Goodnough, M.C.; Tepp, W.H.; Johnson, E.A.; Chapman, E.R. Synaptotagmins I and II Mediate Entry of Botulinum Neurotoxin B into Cells. J. Cell Biol. 2003, 162, 1293–1303. [CrossRef] [PubMed] 46. Nishiki, T.-I.; Kamata, Y.; Nemotos, Y.; Omoris, A.; Itos, T.; Takahashit, M.; Kozakio, S. Identification of Protein Receptor for Clostridium Botulinum Type B Neurotoxin in Rat Brain Synaptosomes. J Biol Chem. 1994, 269, 10498–10503. [CrossRef] 47. Rummel, A.; Karnath, T.; Henke, T.; Bigalke, H.; Binz, T. Synaptotagmins I and II Act as Nerve Cell Receptors for Botulinum Neurotoxin G. J. Biol. Chem. 2004, 279, 30865–30870. [CrossRef] 48. Solabre Valois, L.; Wilkinson, K.A.; Nakamura, Y.; Henley, J.M. Endocytosis, Trafficking and Exocytosis of Intact Full-Length Botulinum Neurotoxin Type a in Cultured Rat Neurons. Neurotoxicology 2020, 78, 80–87. [CrossRef] 49. Pirazzini, M.; Tehran, D.A.; Leka, O.; Zanetti, G.; Rossetto, O.; Montecucco, C. On the Translocation of Botulinum and Tetanus Neurotoxins across the Membrane of Acidic Intracellular Compartments. Biochim. Biophys. Acta Biomembr. 2016, 1858, 467–474. [CrossRef] 50. Schmid, M.F.; Robinson, J.P.; Dasgupta, B. Direct Visualization of Botulinum Neurotoxin-Induced Channels in Phospholipid Vesicles. Nature 1993, 364, 827–830. [CrossRef] 51. Koriazova, L.K.; Montal, M. Translocation of Botulinum Neurotoxin Light Chain Protease through the Heavy Chain Channel. Nat. Struct. Biol. 2003, 10, 13–18. [CrossRef] 52. Oblatt-Montal, M.; Yamazaki, M.; Nelson, R.; Montal, M. Formation of Ion Channels in Lipid Bilayers by a Peptide with the Predicted Transmembrane Sequence of Botulinum Neurotoxin A. Protein Sci. 1995, 4, 1490–1497. [CrossRef] 53. Fischer, A.; Sambashivan, S.; Brunger, A.T.; Montal, M. Beltless Translocation Domain of Botulinum Neurotoxin A Embodies a Minimum Ion-Conductive Channel. J. Biol. Chem. 2012, 287, 1657–1661. [CrossRef] [PubMed] 54. Fischer, A.; Montal, M. Molecular Dissection of Botulinum Neurotoxin Reveals Interdomain Chaperone Function. Toxicon 2013, 75, 101–107. [CrossRef] [PubMed] 55. Cai, S.; Kukreja, R.; Shoesmith, S.; Chang, T.W.; Singh, B.R. Botulinum Neurotoxin Light Chain Refolds at Endosomal PH for Its Translocation. Protein J. 2006, 25, 455–462. [CrossRef] [PubMed] 56. Montal, M. Translocation of Botulinum Neurotoxin Light Chain Protease by the Heavy Chain Protein-Conducting Channel. Toxicon 2009, 54, 565–569. [CrossRef] [PubMed] 57. Pirazzini, M.; Tehran, D.A.; Zanetti, G.; Megighian, A.; Scorzeto, M.; Fillo, S.; Shone, C.C.; Binz, T.; Rossetto, O.; Lista, F.; et al. Thioredoxin and Its Reductase Are Present on Synaptic Vesicles, and Their Inhibition Prevents the Paralysis Induced by Botulinum Neurotoxins. Cell Rep. 2014, 8, 1870–1878. [CrossRef] 58. Vaidyanathan, V.V.; Yoshino, K.-I.; Jahnz, M.; Dörries, C.; Bade, S.; Nauenburg, S.; Niemann, H.; Binz, T. Proteolysis of SNAP- 25 Isoforms by Botulinum Neurotoxin Types A, C, and E: Domains and Amino Acid Residues Controlling the Formation of Enzyme-Substrate Complexes and Cleavage. J. Neurochem. 1999, 72, 327–337. [CrossRef] 59. Chen, S.; Hall, C.; Barbieri, J.T. Substrate Recognition of VAMP-2 by Botulinum Neurotoxin B and Tetanus Neurotoxin. J. Biol. Chem. 2008, 283, 21153–21159. [CrossRef] 60. Schiavo, G.; Shone, C.C.; Rossetto
Toxins 2023, 15, 92 21 of 22 64. Lu, B. The Destructive Effect of Botulinum Neurotoxins on the SNARE Protein: SNAP-25 and Synaptic Membrane Fusion. PeerJ 2015, 3, e1065. [CrossRef] [PubMed] 65. Gardner, A.P.; Barbieri, J.T.; Pellett, S. How Botulinum Neurotoxin Light Chain A1 Maintains Stable Association with the Intracellular Neuronal Plasma Membrane. Toxins 2022, 14, 814. [CrossRef] [PubMed] 66. Rossetto, O.; Pirazzini, M.; Montecucco, C. Botulinum Neurotoxins: Genetic, Structural and Mechanistic Insights. Nat. Rev. Microbiol. 2014, 12, 535–549. [CrossRef] 67. Davies, J.R.; Liu, S.M.; Acharya, K.R. Variations in the Botulinum Neurotoxin Binding Domain and the Potential for Novel Therapeutics. Toxins 2018, 10, 421. [CrossRef] 68. Pellett, S.; Bradshaw, M.; Tepp, W.H.; Pier, C.L.; Whitemarsh, R.C.M.; Chen, C.; Barbieri, J.T.; Johnson, E.A. The Light Chain Defines the Duration of Action of Botulinum Toxin Serotype a Subtypes. mBio 2018, 9, e00089-18. [CrossRef] 69. Whitemarsh, R.C.M.; Tepp, W.H.; Johnson, E.A.; Pellett, S. Persistence of Botulinum Neurotoxin a Subtypes 1-5 in Primary Rat Spinal Cord Cells. PLoS ONE 2014, 9, e90252. [CrossRef] 70. Moritz, M.S.; Tepp, W.H.; Bradshaw, M.; Johnson, E.A.; Pellett, S.; Krantz, B. Isolation and Characterization of the Novel Botulinum Neurotoxin A Subtype 6. mSphere 2018, 3, 466–484. [CrossRef] 71. Tsai, Y.C.; Kotiy, A.; Kiris, E.; Yang, M.; Bavari, S.; Tessarollo, L.; Oyler, G.A.; Weissman, A.M. Deubiquitinating Enzyme VCIP135 Dictates the Duration of Botulinum Neurotoxin Type A Intoxication. Proc. Natl. Acad. Sci. USA 2017, 114, E5158–E5166. [CrossRef] 72. Whitemarsh, R.C.M.; Tepp, W.H.; Bradshaw, M.; Lin, G.; Pier, C.L.; Scherf, J.M.; Johnson, E.A.; Pellett, S. Characterization of Botulinum Neurotoxin A Subtypes 1 through 5 by Investigation of Activities in Mice, in Neuronal Cell Cultures, and in Vitro. Infect. Immun. 2013, 81, 3894–3902. [CrossRef] 73. Wang, D.; Krilich, J.; Pellett, S.; Baudys, J.; Tepp, W.H.; Barr, J.R.; Johnson, E.A.; Kalb, S.R. Comparison of the Catalytic Properties of the Botulinum Neurotoxin Subtypes A1 and A5. Biochim. Biophys. Acta Proteins Proteom. 2013, 1834, 2722–2728. [CrossRef] 74. Henkel, J.S.; Jacobson, M.; Tepp, W.; Pier, C.; Johnson, E.A.; Barbieri, J.T. Catalytic Properties of Botulinum Neurotoxin Subtypes A3 and A4. Biochemistry 2009, 48, 2522–2528. [CrossRef] [PubMed] 75. Moritz, M.S.; Tepp, W.H.; Inzalaco, H.N.t.; Johnson, E.A.; Pellett, S. Comparative Functional Analysis of Mice after Local Injection with Botulinum Neurotoxin A1, A2, A6, and B1 by Catwalk Analysis. Toxicon 2019, 167, 20–28. [CrossRef] [PubMed] 76. Gardner, A.P.; Barbieri, J.T. Light Chain Diversity among the Botulinum Neurotoxins. Toxins 2018, 10, 268. [CrossRef] 77. Kohda, T.; Tsukamoto, K.; Torii, Y.; Kozaki, S.; Mukamoto, M. Translocation Domain of Botulinum Neurotoxin A Subtype 2 Potently Induces Entry into Neuronal Cells. Microbiol. Immunol. 2020, 64, 502–511. [CrossRef] [PubMed] 78. Pellett, S.; Tepp, W.H.; Whitemarsh, R.C.M.M.; Bradshaw, M.; Johnson, E.A. In Vivo Onset and Duration of Action Varies for Botulinum Neurotoxin A Subtypes 1-5. Toxicon 2015, 107, 37–42. [CrossRef] [PubMed] 79. Morineaux, V.; Mazuet, C.; Hilaire, D.; Enche, J.; Popoff, M.R. Characterization of Botulinum Neurotoxin Type A Subtypes by Immunocapture Enrichment and Liquid Chromatography–Tandem Mass Spectrometry. Anal. Bioanal. Chem. 2015, 407, 5559–5570. [CrossRef] 80. Kull, S.; Schulz, K.M.; Strotmeier, J.W.; Kirchner, S.; Schreiber, T.; Bollenbach, A.; Dabrowski, P.W.; Nitsche, A.; Kalb, S.R.; Dorner, M.B.; et al. Isolation and Functional Characterization of the Novel Clostridium Botulinum Neurotoxin A8 Subtype. PLoS ONE 2015, 10, e0116381. [CrossRef] 81. Tian, R.; Widel, M.; Imanian, B. The Light Chain Domain and Especially the C-Terminus of Receptor-Binding Domain of the Botulinum Neurotoxin (BoNT) Are the Hotspots for Amino Acid Variability and Toxin Type Diversity. Genes 2022, 13, 1915. [CrossRef] 82. Davies, J.R.; Rees, J.; Liu, S.M.; Acharya, K.R. High Resolution Crystal Structures of Clostridium Botulinum Neurotoxin A3 and A4 Binding Domains. J. Struct. Biol. 2018, 202, 113–117. [CrossRef] 83. Davies, J.R.; Hackett, G.S.; Liu, S.M.; Acharya, K.R. High Resolution Crystal Structures of the Receptor-Binding Domain of Clostridium Botulinum Neurotoxin Serotypes A and FA. PeerJ 2018, 6, e4552. [CrossRef] [PubMed] 84. Davies, J.R.; Britton, A.; Liu, S.M.; Acharya, K.R. High-Resolution Crystal Structures of the Botulinum Neurotoxin Binding Domains from Subtypes A5 and A6. FEBS Open Bio 2020, 10, 1474–1481. [CrossRef] [PubMed] 85. Gregory, K.S.; Mahadeva, T.B.; Liu, S.M.; Acharya, K.R. Structural Features of Clostridium Botulinum Neurotoxin Subtype A2 Cell Binding Domain. Toxins 2022, 14, 356. [CrossRef] 86. Gregory, K.S.; Liu, S.M.; Acharya, K.R. Crystal Structure of Botulinum Neurotoxin Subtype A3 Cell Binding Domain in Complex with GD1a Co-Receptor Ganglioside. FEBS Open Bio
Toxins 2023, 15, 92 22 of 22 91. Gustafsson, R.; Zhang, S.; Masuyer, G.; Dong, M.; Stenmark, P. Crystal Structure of Botulinum Neurotoxin A2 in Complex with the Human Protein Receptor SV2C Reveals Plasticity in Receptor Binding. Toxins 2018, 10, 153. [CrossRef] 92. Benoit, R.M.; Schärer, M.A.; Wieser, M.M.; Li, X.; Frey, D.; Kammerer, R.A. Crystal Structure of the BoNT/A2 Receptor-Binding Domain in Complex with the Luminal Domain of Its Neuronal Receptor SV2C. Sci. Rep. 2017, 7, 43588. [CrossRef] 93. Yu, R.K.; Tsai, Y.-T.; Ariga, T.; Yanagisawa, M. Structures, Biosynthesis, and Functions of Gangliosides-an Overview. J. Oleo Sci. 2011, 60, 537–544. [CrossRef] [PubMed] 94. Yu, R.K.; Saito, M. Structure and Localization of Gangliosides. In Neurobiology of Glycoconjugates; Springer: Boston, MA, USA, 1989; pp. 1–34; ISBN 9781475759570. 95. Yowler, B.C.; Schengrund, C.L.; Pennsyl, V. Botulinum Neurotoxin A Changes Conformation upon Binding to Ganglioside GT1b. Biochemistry 2004, 43, 9725–9731. [CrossRef] [PubMed] 96. Scheiner, S.; Kar, T.; Pattanayak, J. Comparison of Various Types of Hydrogen Bonds Involving Aromatic Amino Acids. J. Am. Chem. Soc. 2002, 124, 13257–13264. [CrossRef] [PubMed] 97. Stout, K.A.; Dunn, A.R.; Hoffman, C.; Miller, G.W. The Synaptic Vesicle Glycoprotein 2: Structure, Function, and Disease Relevance. ACS Chem. Neurosci. 2019, 10, 3927–3938. [CrossRef] 98. Ciruelas, K.; Marcotulli, D.; Bajjalieh, S.M. Synaptic Vesicle Protein 2: A Multi-Faceted Regulator of Secretion. Semin. Cell Dev. Biol. 2019, 95, 130–141. [CrossRef] 99. Iezzi, M.; Theander, S.; Janz, R.; Loze, C.; Wollheim, C.B. SV2A and SV2C Are Not Vesicular Ca2+ Transporters but Control Glucose-Evoked Granule Recruitment. J. Cell Sci. 2005, 118, 5647–5660. [CrossRef] [PubMed] 100. Morgans, C.W.; Kensel-Hammes, P.; Hurley, J.B.; Burton, K.; Idzerdal, R.; McKnight, G.S.; Bajjalieh, S.M. Loss of the Synaptic Vesicle Protein SV2B Results in Reduced Neurotransmission and Altered Synaptic Vesicle Protein Expression in the Retina. PLoS ONE 2009, 4, e5230. [CrossRef] 101. Mahrhold, S.; Rummel, A.; Bigalke, H.; Davletov, B.; Binz, T. The Synaptic Vesicle Protein 2C Mediates the Uptakeof Botulinum Neurotoxin A into Phrenic Nerves. FEBS Lett. 2006, 580, 2011–2014. [CrossRef] [PubMed] 102. Benoit, R.M.; Frey, D.; Hilbert, M.; Kevenaar, J.T.; Wieser, M.M.; Stirnimann, C.U.; McMillan, D.; Ceska, T.; Lebon, F.; Jaussi, R.; et al. Structural Basis for Recognition of Synaptic Vesicle Protein 2C by Botulinum Neurotoxin A. Nature 2014, 505, 108–122. [CrossRef] 103. Zuverink, M.; Bluma, M.; Barbieri, J.T. Tetanus Toxin Cis-Loop Contributes to Light-Chain Translocation. mSphere 2020, 5, e00244-20. [CrossRef] 104. Fujinaga, Y.; Popoff, M.R. Translocation and Dissemination of Botulinum Neurotoxin from the Intestinal Tract. Toxicon 2018, 147, 13–18. [CrossRef] [PubMed] 105. Araye, A.; Goudet, A.; Barbier, J.; Pichard, S.; Baron, B.; England, P.; Pérez, J.; Zinn-Justin, S.; Chenal, A.; Gillet, D. The Translocation Domain of Botulinum Neurotoxin A Moderates the Propensity of the Catalytic Domain to Interact with Membranes at Acidic PH. PLoS ONE 2016, 11, e0153401. [CrossRef] 106. Lam, K.; Matsui, T.; Perry, K.; Rummel, A.; Bowen, M.E.; Guo, Z.; Weisemann, J.; Krez, N.; Jin, R. A Viral-Fusion-Peptide-like Molecular Switch Drives Membrane Insertion of Botulinum Neurotoxin A1. Nat. Commun. 2018, 9, 5367. [CrossRef] 107. Davies, J. Structure-Function Studies on Clostridium Botulinum Neurotoxins. Ph.D. Thesis, University of Bath, Bath, UK, 2018. 108. Fischer, A.; Mushrush, D.J.; Lacy, D.B.; Montal, M. Botulinum Neurotoxin Devoid of Receptor Binding Domain Translocates Active Protease. PLoS Pathog. 2008, 4, e1000245. [CrossRef] [PubMed] 109. Sun, S.; Suresh, S.; Liu, H.; Tepp, W.H.; Johnson, E.A.; Edwardson, J.M.; Chapman, E.R. Receptor Binding Enables Botulinum Neurotoxin B to Sense Low PH for Translocation Channel Assembly. Cell Host Microbe 2011, 10, 237–247. [CrossRef] 110. Rabe von Pappenheim, F.; Wensien, M.; Ye, J.; Uranga, J.; Irisarri, I.; de Vries, J.; Funk, L.M.; Mata, R.A.; Tittmann, K. Widespread Occurrence of Covalent Lysine–Cysteine Redox Switches in Proteins. Nat. Chem. Biol. 2022, 18, 368–375. [CrossRef] [PubMed] 111. Wensien, M.; von Pappenheim, F.R.; Funk, L.M.; Kloskowski, P.; Curth, U.; Diederichsen, U.; Uranga, J.; Ye, J.; Fang, P.; Pan, K.T.; et al. A Lysine–Cysteine Redox Switch with an NOS Bridge Regulates Enzyme Function. Nature 2021, 593, 460–464. [CrossRef] [PubMed] 112. Neumann, P.; Tittmann, K. Marvels of Enzyme Catalysis at True Atomic Resolution: Distortions, Bond Elongations, Hidden Flips, Protonation States and Atom Identities. Curr. Opin. Struct. Biol. 2014, 29, 122–133. [CrossRef] [PubMed] Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibilit