Mechanism of Substrate Recognition by Botulinum Neurotoxin Serotype A*

Botulinum neurotoxins (BoNTs) are zinc proteases that cleave SNARE proteins to elicit flaccid paralysis by inhibiting neurotransmitter-carrying vesicle fusion to the plasma membrane of peripheral neurons. Unlike other zinc proteases, BoNTs recognize extended regions of SNAP25 for cleavage; however, the molecular basis for this extended substrate recognition is unclear. Here, we define a multistep mechanism for recognition and cleavage of SNAP25 by BoNT/A. SNAP25 initially binds along the belt region of BoNT/A, which aligns the P5 residue to the S5 pocket at the periphery of the active site. Although the exact order of each step of recognition of SNAP25 by BoNT/A at the active site is not clear, the initial binding could subsequently orient the P4′-residue of SNAP25 to form a salt bridge with the S4′-residue, which opens the active site allowing the P1′-residue access to the S1′-pocket. Subsequent hydrophobic interactions between the P3 residue of SNAP25 and the S3 pocket optimize alignment of the scissile bond for cleavage. This explains how the BoNTs recognize and cleave specific coiled SNARE substrates and provides insight into the development of inhibitors to prevent botulism.


Cleavage of SNAP25 by BoNT-LCs
Linear Velocity Reaction-Reactions (10 l) performed were as follows. 5 M SNAP25 was incubated with the indicated concentrations of LC/A or LC/A derivatives in 10 mM Tris-HCl, pH 7.6, with 20 mM NaCl. Reactions were resolved by SDS-PAGE, and the amount of SNAP25 cleavage was determined by densitometry.
Time Course Assay-5 M SNAP25 was incubated with 2 nM wild type LC/A or LC/A derivatives. Aliquots were withdrawn at specified times and processed as described above.
Kinetic Parameters-K m and k cat determinations were performed for WT-LC/A and LC/A derivatives using SNAP25 as substrate. LC concentrations were adjusted to cleave Ͻ10% substrate at several concentrations of substrate ranging from 3 to 24 M. The reaction was carried out in 10 l of toxin reaction buffer, incubated at 37°C for 10 min, and stopped by adding 10 l of 2ϫ SDS-PAGE buffer. The cleaved products were separated by SDS-PAGE, and the amount of cleaved product was calculated by densitometry. Reaction velocity versus substrate concentration was fit to the Michaelis-Menten equation, and kinetic constants were derived from the Lineweaver-Burk plot using the SigmaPlot program (Chicago, IL).
Mutation Complementation Assay-5 M wild type SNAP25, SNAP(D193A), SNAP(R198A), or SNAP(I171A) were incubated with WT-LC/A (0.1-10 nM) or LC/A derivatives (0.01-10 M). Reactions were subjected to SDS-PAGE analysis, and the amount of SNAP25 cleaved was determined by densitometry. The concentrations of WT-LC/A or LC/A derivatives that cleaved 50% of SNAP25 derivatives were calculated.
L-Arginine Hydroxamate Inhibition of LC/A Cleavage of SNAP25-WT-LC/A or LC/A derivatives, at concentrations that cleaved ϳ50% SNAP25-(141-206) in a linear velocity reaction, were incubated with 10, 1, or 0.1 mM ArgHX for 30 min on ice, followed by the addition of 5 M SNAP25-(141-206). The reaction was subjected to SDS-PAGE to determine the amount of SNAP25 cleaved.

Trypsin Digestion of LC/A
5 M WT-LC/A or LC/A derivative was incubated with 0.1 M trypsin. The reactions were subjected to SDS-PAGE and stained to visualize the partial tryptic digestion. Mutations within LC/A had identical trypsin digestion profiles as WT-LC/A (Fig. 1), which indicated that these single mutations in LC/A had limited effects on the overall structural configuration of LC/A.

GST Pulldown Assay
GST pulldown assay (100-l reaction) was performed by preincubating 3 M GST-SNAP25(R198E) with 2 and 5 M LC/A WT or LC/A(D370A, L175A) and then adding 30 l of gluta-thione-Sepharose beads, preblocked in 10% bovine serum albumin. The beads were pelleted and washed, and proteins in the pellet were detected by Western blotting using anti-LC/A and anti-GST antibody (Sigma).

Molecular Modeling of the LC/A Active Site Domain-Satu-
ration mutagenesis defined the organization and recognition of SNAP25 by LC/A, where an AS region defined substrate cleavage and a B region that was distanced from the AS defined high affinity binding by LC/A (14). Analysis of molecular models predicted a series of discontiguous interactions that first aligned the B region of SNAP25 to the active site and then optimized scissile bond cleavage through a series of interactions that involve four pockets (S5, S4Ј, S1Ј, and S3) that comprise the LC/A AS (Fig. 2, A-C). An overview of this model was shown by a stereo image (Fig. 2D).
Initial Interactions That Align the P5 Residue of SNAP25 to the LC/A Active Site-In the holotoxin structure, the N-terminal heavy chain (HC) loop forms a belt that wraps around the substrate binding cleft of the LC; unexpectedly, the region of the HC loop that bound LC/A aligned well with the region of SNAP25 that bound to LC/A (Fig. 3A). This prompted a comparison of SNAP25 and the HC loop of BoNT/A that revealed spatial and sequence homology between the two proteins. The alignment extended along a stretch of 38 amino acids of SNAP25 that abruptly ended at the P5 residue (Fig. 3, A and B). The residues that played an important role in substrate binding and recognition in SNAP25 with LC/A aligned with HC loop residues that interacted with LC/A (Fig. 3B). This suggested that the initial recognition of SNAP25 with LC/A was mimicking the binding of the HC loop within the binding cleft of LC/A in the native holotoxin. The biological function of the HC loop in BoNT/A may be as a pseudo-substrate that blocks the active site of LC/A to prevent auto-cleavage (16). Analysis of the LC/A and SNAP25 complex structure revealed that LC/A residues Ile 115 , Lys 41 , Cys 134 , and Val 129 directly interacted with residues within the SNAP25 B region. Ala mutations to Ile 115 , Lys 41 , Cys 134 , and Val 129 each had a ϳ4-fold increase in K m and a ϳ5-fold decrease in k cat for the cleavage of SNAP25 (Table 1). Thus, the B region of SNAP25 performs a dual role as the initial site of LC/A recognition and as the first active site interaction between the P5 residue of SNAP25 and the S5 pocket of the LC/A active site. However, defining the significance of slight changes in K m as being due to direct or indirect interactions is difficult. An Eadie Hofstee plot of wild type LC/A cleavage of SNAP25 is shown in Fig. 4. This interaction packs loop250 to loop370 and broadens the active site cavity of LC/A, which allows LC/A to dock the P1Ј-site into S1Ј-pockets. C, recognition of P1Ј-Arg 198 by the LC/A S1Ј-pocket. The S1Ј-pocket is formed by Asp 370 , Thr 220 , Phe 194 , and Phe 163 . Asp 370 and Phe 194 contribute to the direct recognition of the P1Ј-Arg 198 . These structural models were made using the complex structure of LC/A⅐SNAP25 (1XTG) as template and prepared by PyMol. (Green structure represents LC/A; red structure, SNAP25; blue residues, basic residues; red residues, acidic residues). D, stereo image of LC/A bound to SNAP25(Lys 189 Ͼ Lys 201 ) of SNAP25. Note the large acidic cavity representing the S1Ј-pocket of LC/A (red, acidic; blue, basic). The image was generated by Z. Fu as described in Ref. 13. Interaction between the S5 Pocket Residue of LC/A and SNAP25 P5 Site Residue-The LC/A-SNAP25 structure showed that Leu 175 , Thr 176 , and Arg 177 were organized as a pocket that surrounded the P5 residue, Asp 193 ( Fig. 2A). Muta-tion of each residue to Ala reduced LC/A hydrolysis activity between 60-and 100-fold (Table 1) with a greater effect on k cat than K m . This suggested that the S5 pocket contributed to the proper alignment of the scissile bond for cleavage rather than contributing to substrate affinity. The loss of catalytic activity by S5 pocket residue mutations can be complemented by mutation of the P5 site of SNAP25, Asp 193 , supporting a direct role for S5 pocket residues on P5 recognition. While LC/A(L175A), LC/A(T176A), and LC/A(R177A) possessed a reduced capacity to cleave WT-SNAP25, ϳ75-, ϳ50-, and ϳ50-fold reduction in cleavage activities, respectively, these three mutated LC/As cleaved SNAP25(D193A) at ϳ3to 5-fold lower than WT-LC/A. These results suggest that P5 Asp 193 was recognized by S5 pocket residues, but this recognition may not be a direct interaction between S5 pocket residues and Asp 193 where electrostatic interactions between the basic S5 pocket and negatively charged Asp 193 contribute to this interaction ( Table 2). The S5 pocket residues appeared specific, because mutation of other LC/A residues adjacent to the P5 residue of SNAP25, LC/A(Q162A) and LC/A(H269A), did not affect hydrolytic activity (Table 1). Interactions between the S5 pocket of LC/A and P5 residue of SNAP25 orient the next step in SNAP25 recognition, the binding of the P4Ј residue of SNAP25 to the LC/A active site.
Recognition of the P4Ј-Residue of SNAP25 by the S4Ј-Pocket Residue of LC/A-The loop250 of LC/A (residues 242-259) was initially identified as the site of auto-cleavage (17) and subsequently observed to pack next to LC/A loop370 (residues 359 -370) when LC/A bound SNAP25 (12). Whereas earlier studies proposed that the loop250 residue Met 202 represented the primary contact of SNAP25 to LC/A, subsequent analyses showed that mutations of SNAP25-Met 202 had a limited effect on catalysis (12) but that mutations at SNAP25-Lys 201 yielded poor substrates for cleavage by LC/A (14). Examination of the LC/A-SNAP25 structure showed a potential salt bridge between SNAP25-Lys 201 and LC/A-Glu 257 (Fig. 2B). Supporting a role of Glu 257 in SNAP25 recognition/cleavage was the determination that mutations at Glu 257 had a 60-to 600-fold slower rate of SNAP25 cleavage, reflected in a lower k cat , than WT-LC/A (Table 1). Sequential binding at the S5 and S4Ј-sites aligns SNAP25 within the active site such that the P1Ј-residue of SNAP25, Arg 198 , initiates interactions with the S1Ј-pocket of LC/A.
Recognition of the P1Ј-Site Residue of SNAP25 by S1Ј-Pocket Residues of LC/A-LC/A-Asp 370 forms a salt bridge with the inhibitor L-ArgHX (13). Examination of the structure of LC/A and L-ArgHX identified an additional LC/A residue that contacted the inhibitor Phe 163 and several residues that aligned Reactions were carried out in 10 l at 37°C for 10 min and stopped with SDS-PAGE buffer. The reaction mixture was separated by SDS-PAGE, and the amount of cleaved product was calculated by densitometry. Reaction velocity versus substrate concentration was fit to the Eadie Hofstee plot using Sig-maPlot (Chicago, IL).  Thus, in addition to a salt bridge of the guanidinium group of Arg 198 with the S1Ј-pocket residue, Asp 370 , there appear to be hydrophobic interactions between the aliphatic portion of the side chain of Arg 198 and the hydrophobic S1Ј-pocket residues, in particular Phe 194 (Fig. 2C). A direct role of Asp 370 and Phe 194 on substrate recognition was supported by the catalytic complementation of LC/A(Asp 370 ) and LC/A(Phe 194 ) by P1Ј-SNAP25 mutations. Whereas LC/A(D370A) and LC/A(F194A) cleaved WT-SNAP25 less efficiently than WT-LC/A, both mutated proteins cleaved the P1Ј-mutated SNAP25(R198A) at similar rates as WT-LC/A ( Table 3). The role of S1Ј-pocket residues on P1Ј-residue recognition was also supported by the change in sensitivity of S1Ј-pocket mutated LC/A to inhibition by L-ArgHX, an inhibitor of LC/A (18). Examination of the structure of the LC/A⅐L-ArgHX complex showed that the carbonyl-and N-hydroxyl groups of L-ArgHX formed a bidentate ligand with the zinc ion and the guanidinium group of the Arg side chain formed a salt bridge with the carbonyl R group of Asp 370 , suggesting that the inhibitor-bound structure mimicked a catalytic intermediate for the P1Ј-residue Arg 198 at the active site. The ability of L-ArgHX to inhibit catalysis by mutations in LC/A was used to probe the S1Ј-pocket. The residual hydrolytic activity of LC/A(D370A) and LC/A(D370R) for SNAP25 (Fig. 3) was not inhibited by 10 mM L-ArgHX. LC/A(F163A), LC/A (F194A), and LC/A(T220A) were also less sensitive to L-ArgHX inhibition than WT-LC/A and were only partially inhibited at 10 mM L-ArgHX (Fig. 5), which supported their role in P1Ј-residue interactions. In contrast, LC/A possessing mutations in S4Ј-and S5 pocket residues remained sensitive to the inhibition by L-ArgHX.
Substrate recognition by the S1Ј-residue has been proposed based upon characterization of BoNT-SNARE interactions (14,19,20) and solved crystal structures of BoNTs (17, 20 -27), which share overall structure of the active site regions of the LCs. Within the active site, zinc is coordinated by the HEXXH motif, and a conserved Glu and a conserved Arg and Tyr lie in close proximity to the scissile peptide bond. However, conservation of amino acids within the active site domain is not extended to the S1Ј-pocket, which is consistent with the diverse chemical nature of P1Ј-residues of the substrates of various BoNT serotypes. Overall, S1Ј-pocket residues correlated in size and hydrophobicity with the cognate P1Ј-residue. In LC/A, each of the four residues that form the S1Ј-pocket appear to play different roles in substrate recognition. Asp 370 is located underneath the S1Ј-pocket and forms a salt bridge with the guanidinium group of Arg 198 , whereas Phe 194 , Phe 163 , and Thr 220 have hydrophobic interactions with the aliphatic chain of the Arg 198 side chain. S1Ј-pocket residue recognition appears to be the primary mechanism of substrate recognition in LC/A and other neurotoxins (14,28,29).
Alignment of P3 Residue of SNAP25 to the S3 Pocket Residues of LC/A-Mutation of the P3 residue of SNAP25(A195S) had a 1000-fold reduction in hydrolysis relative to the cleavage of WT-SNAP25 by LC/A (14). Examination of the co-crystal structure of SNAP25 and LC/A revealed that the methyl side chain on Ala 195 tightly fits into a pocket within LC/A (the S3 pocket) without enough space for the ϪOH of serine (12). Because of spatial constraints of the S3 pocket main chain interactions there were no conserved substitutions that were predicted to maintain the cavity of the S3 pocket, which discouraged subsequent manipulation of this site. The role of the S3 pocket residues of LC/A in the substrate recognition involves the alignment of the P3 residue of SNAP25, Ala 195 , to the S3 pocket of LC/A, which sets up an optimal orientation for the interactions of the P1Ј-residue of SNAP25, Arg 198 , to Asp 370  and Phe 194 of LC/A, resulting in the precise alignment of the scissile bond in the active site.

Effect of Double Pocket Mutations on SNAP25 Cleavage by LC/A-
The functional relationships between residues within a single pocket or different pockets were assessed to test the predicted organization of LC/A for substrate recognition. The double pocket mutations LC/A(Leu 175 ,Asp 370 ) and LC/A(Leu 175 ,Phe 194 ) showed only residual capacity to cleave SNAP25 (reductions in specific activity were Ͼ40,000fold), whereas double mutations within a single predicted pocket, LC/A(Phe 194 ,Thr 220 ) and LC/A(Leu 175 ,Lys 177 ), cleaved SNAP25 at rates similar to the respective individual mutations at these residues. GST pulldown experiments showed that LC/A(L175A,D370A) and LC/A(L175A,F194A) had similar affinity for SNAP25(R198E), a non-cleavable form of SNAP25, as WT LC/A (Fig. 6), supporting that the roles for these residues are in catalytic action, not in substrate binding. LC/A(F194A,T220A)and LC/A(L175A,K177A) had reduced k cat for SNAP25 cleavage as observed for individual pocket mutations (Table 1). Experiments were conducted at concentrations of LC/A and SNAP25 to yield a dose response for the amount of wild type LC/A in the pulldown (data not shown). Together, the data indicated that mutation of residues that lay in different pockets caused synergistic reduction in catalysis, whereas mutation of residues located within a single pocket had reduced catalytic activity that was similar to individual mutations within the respective pockets.

DISCUSSION
Although the precise mechanism for peptide bond cleavage by the BoNTs remains to be resolved, cleavage of the scissile peptide bond appears to follow a general base-catalyzed mechanism. Arg 362 and Tyr 365 interact with the carbonyl oxygens of the P1 and P1Ј-residues of SNAP25, respectively, and stabilize the oxyanion in the transition state (30,31). Peptide bond cleavage is initiated by a water molecule that is polarized by the Glu within the zinc binding motif (HEXXH) and Zn 2ϩ , which causes a nucleophilic attack on the carbonyl carbon of the scissile bond to form an oxyanion.
Peptide bond cleavage is likely achieved by a proton transfer from the attacking water mediated by the carboxyl group of the downstream Glu to form a protonated amine. Rate of catalysis will be affected by the relative contribution of particular residues in the docking and stabilization of substrate to the active site cavity. Upon substrate cleavage, the P4Ј-residue-S4Ј-residue interaction is disrupted, which initiates the dissociation of the C-terminal product of SNAP25 from LC/A (32,33). Upon dissociation of the P4Ј-residue the AS returns to the original conformation that has a lower affinity for the P1Јresidue, facilitating the dissociation of the C-terminal product of SNAP25 from LC/A. The N-terminal product of LC/A cleavage can associate with syntaxins, which yields unproductive syntaxin⅐SNAP-25 complexes that impede vesicle exocytosis, resulting in BoNT/A poisoning (34 -36).
The current study provides a molecular mechanism of LC/A recognition and cleavage of SNAP25 that involves sequential steps representing SNAP25 recognition and active site organization (Fig. 7). Initial   A and B, at the plasma membrane LC/A initially binds to SNAP25 through discontiguous surface interactions between residues within the belt region of LC/A and the B region residues of SNAP25. The Velcro-like binding of SNAP25 to LC/A aligns the P5 residue Asp 193 to form a salt bridge with Arg 177 , an S5 pocket residue, at the periphery of one side of the active site. C, this orients SNAP25 for the formation of a salt bridge between the P4Ј-residue Lys 201 and the S4Ј-residue LC/A(Asp 257 ). D, these interactions broaden the LC/A active site cavity and dock Arg 198 , the P1Ј-residue, via electrostatic interactions with Asp 370 within the S1Ј-pocket. The fine-tuning of the alignment of Arg 198 into the S1Ј-pocket is facilitated by the binding of SNAP25-Ala 195 to P3 residues in the hydrophobic S3 pocket of LC/A. The proper alignment of the P1Ј-P3 sites into the Zn 2ϩ active motif (E) facilitates the substrate cleavage (F). After the cleavage, the C-terminal product dissociates from LC/A, which returns the AS to the original conformation (G).
interactions involve discontiguous surfaces between residues within the belt region of LC/A and the B region residues of SNAP25. The Velcro-like binding of SNAP25 to LC/A aligns the P5 residue Asp 193 to form a salt bridge with Arg 177 , an S5 pocket residue at the periphery of one side of the active site. Although the exact order of each step of recognition of SNAP25 by BoNT/A at the active site is not clear, the initial binding could subsequently orient SNAP25 for the formation of a salt bridge between the P4Ј-residue SNAP25(Lys 201 ) and the S4Јresidue LC/A(Asp 257 ). These interactions (12) broaden the LC/A active site cavity and dock Arg 198 , the P1Ј-residue, via electrostatic and hydrophobic interactions within the S1Јpocket. The fine tuning of the alignment of Arg 198 into the S1Ј-pocket resulting in the precise alignment of the scissile bond is facilitated by the binding of the P3 residue, SNAP25-Ala 195 , into the hydrophobic S3 pocket of LC/A. The proper docking of the P1Ј-P1 sites into the AS initiates substrate cleavage. After cleavage, the P4Ј-residue dissociates from the S4Јresidue of LC/A, which converts the AS to a smaller conformation, facilitating dissociation of the P1Ј-residue from the AS.
Understanding the mechanism of substrate recognition may provide insight into the development of serotype-specific inhibitors against botulism. In addition, BoNT is the most widely utilized protein for human therapy (37) for numerous neurological disorders from migraines to muscle trauma to physical disabilities (38). BoNT serotype A is used in these therapies, based upon the ability to produce relatively pure amounts of the holotoxin and the longevity of LC action in neurons (39,40). Thus, understanding the substrate specificity of LC/A should also provide insight into the modification of BoNT to optimize therapeutic potential.