Reengineering the specificity of the highly selective Clostridium botulinum protease via directed evolution

Abstract

The botulinum neurotoxin serotype A (BoNT/A) cuts a single peptide bond in SNAP25, an activity used to treat a wide range of diseases. Reengineering the substrate specificity of BoNT/A's protease domain (LC/A) could expand its therapeutic applications; however, LC/A's extended substrate recognition (≈ 60 residues) challenges conventional approaches. We report a directed evolution method for retargeting LC/A and retaining its exquisite specificity. The resultant eight-mutation LC/A (omLC/A) has improved cleavage specificity and catalytic efficiency (1300- and 120-fold, respectively) for SNAP23 versus SNAP25 compared to a previously reported LC/A variant. Importantly, the BoNT/A holotoxin equipped with omLC/A retains its ability to form full-length holotoxin, infiltrate neurons, and cleave SNAP23. The identification of substrate control loops outside BoNT/A's active site could guide the design of improved BoNT proteases and inhibitors.

Figure 1

Directed evolution overview. (A) SNAP25 sub-family proteins and their isoforms (CLUSTAL 2.1). (B) Co-crystal structure (PDB: 1XTG) of LC/A (white) and SNAP25 (dark gray). Eight substitutions in LC/A drive SNAP23 specificity (teal) through substrate control loops (pink) alongside prior substitutions (light gray). (C) Platform for the directed evolution of SNAP23 substrate specificity. 1. Random or site-directed mutagenesis (e.g., site-saturation); 2. QC by DNA sequencing; 3. High-throughput protein production; 4. Measure V₀²³ and V₀²⁵ for substrate specificity; 5. Confirmation screens. The most specific and consistent variant from the DARET assay entered the next round of directed evolution. (D) Sequence alignment of substrates used for screening (UniProt: P60880, O00161). The SNAP binding exosites in LC/A (residue numbers above) and cleavage site (scissors) are shown. The gradient of color indicates homology from identical (white, *), to strongly similar (light gray, :), weakly similar (teal, .), or dissimilar (dark teal, space).

Figure 2

Directed evolution of a SNAP23-specific LC/A. (A) Improvement in specificity index through directed evolution in salt-free buffer. The most specific variant from each round was assayed multiple times (replicates shown). The average specificity index (horizontal bars) and standard deviation (error bars) are shown. The final dilution of cell lysates for screening is indicated by color code. (B) Increased specificity index achieved through directed evolution in salt buffer. (C) Roadmap for directed evolution showing the fold increase in specificity indices over eight rounds (R1 to R8). As shown in Fig. S1, the substrate specificity of the qmLC/A variant is sensitive to screening dilution; therefore, each point represents the variant’s average specificity normalized to the qmLC/A average specificity at the corresponding screening dilution. Advancement to the next round (solid line) weighed specificity index, protein solubility, and stability.

Figure 3

Characterization of omLC/A cleavage. (A) Rates of SNAP cleavage by batch-expressed, purified qmLC/A and (B) omLC/A at the indicated DARET substrate concentrations (n = 3). (C) Deconvoluted ESI–MS of recombinant, full-length SNAP23 (fl-S23) treated with DTT and iodoacetamide to carbamidomethylate its six cysteines (6 × CAM). The mass spectrum of intact fl-S23 incubated with buffer (top, black) is compared to that of fl-S23 incubated with omLC/A (bottom, teal). Intact fl-S23 (1) and cleaved fl-S23 (cl. fl-S23, 2) peaks are labeled. Additional marked peaks correspond to the masses of peaks 1 and 2 plus one additional CAM (+ 57 Da), likely resulting from overalkylation by iodoacetamide. The cleaved peptide was not directly observed, but inferred from the mass difference of peaks 1 and 2. The deconvolution error is +/- 2 Da. (D) Recombinant omBoNT/A was purified by IMAC followed by anion exchange (AEX) chromatography. The omBoNT/A is ≈95% nicked upon DTT reduction as demonstrated by the presence of the HC/A and omLC/A bands. (E) In vitro cleavage of recombinant fl-S23 by two independent preparations of omBoNT/A (1 and 2) visualized with a C-terminal anti-SNAP23 antibody; before proteolysis, omBoNT/A was reduced with TCEP. The untreated, wtLC/A, and wtLC/E lanes provide negative controls. (F) In cellulo cleavage of SNAP23 and SNAP25 in SiMA-H1 cells infected with adenovirus delivering DNA encoding mCherry/SNAP23. Cells were treated with omBoNT/A or wtBoNT/A or without toxin (ct). Proteins, full-length (fl) or cleaved (cl), were identified by Western blotting with anti-SNAP23, -SNAP25, or -mCherry antibodies (M, MW marker). Full-length images of these gels with multiple exposures where necessary are shown in Fig. S12 and S13).

Figure 4

Substrate control loops are structurally present in other BoNT LC serotypes. (A) Multiple BoNT holotoxins were structurally aligned with UCSF Chimera MatchMaker (PDB IDs: 3BTA, 1S0G, 3FFZ, 1ZB7, & 2A8A). The heavy chain (HC) of each BoNT is colored dark grey, the light chain (LC) is colored in white, and the substrate control loops are colored as indicated in the key. (B) Larger image of the boxed section in A highlighting the substrate control loops. (C) The primary sequences of LC serotypes A through G were aligned using Clustal Omega. The substrate control loops are highlighted and annotated in teal. Identical residues are color-coded black (*), strongly similar residues in dark grey (:), weakly similar residues in light grey (.), and dissimilar residues are transparent. The poor primary sequence conservation in these structurally conserved loops suggests they help inform the diverse substrate specificities of other LC serotypes.