De novo design of potent inhibitors of clostridial family toxins

Abstract

Clostridioides difficile remains a leading cause of hospital-acquired infections, with its primary virulence factor, toxin B (TcdB), responsible for severe colitis and recurrent disease. The closely related toxin, TcsL, from Paeniclostridium sordellii, causes a rarer but often fatal toxic shock syndrome, particularly in gynecological and obstetric contexts. We report the de novo design of small protein minibinders that directly neutralize TcdB and TcsL by preventing their entry into host cells. Using deep learning and Rosetta-based approaches, we generated high-affinity minibinders that protect cells from intoxication with picomolar potency and, in the case of TcsL, prolonged survival following lethal toxin challenge in mice. The designed proteins against TcdB demonstrate exceptional stability in proteolytic and acidic environments, making them well-suited for oral delivery-a valuable feature for treating C. difficile infections localized to the gastrointestinal tract. For TcsL, potent inhibitors were identified from 48 initial designs and 48 optimized designs, highlighting the potential of computational design for rapidly developing countermeasures against life-threatening bacterial toxins.

Keywords: C. difficile; cryo-EM; protein design; tcdb; tcsl.

Fig. 1.

Composite models of clostridial toxin-host cell receptor complexes. Inset boxes highlight the residues targeted during design. (A) TcdB [PDB 6OQ5 (20)] with Frizzled-2 [PDB 6C0B (8)] and TFPI [PDB 7V1N (7)]. (B) TcdB [PDB 6OQ5 (20)] with CSPG4 [PDB 7ML7 (6)]. (C) TcsL [PDB 8JB5 (21)] with SEMA6A [PDB 6WTS (9)].

Fig. 2.

Design of anti-TcdB Frizzled-blocking minibinders. (A) Design models of three high affinity minibinder families. (B) Corresponding single cycle kinetic analysis of a design from each family (fzd13 from group 1, fzd48 from group 2 and fzd24 from group 3) with TcdB RBD captured on chip and a 6-step 4-fold dilution series starting at 100 nM. (C) Disulfide stabilization of fzd84 minibinder with the location of the disulfide introduced (Left) and a coomassie stained SDS-PAGE after a time course incubation in simulated intestinal fluid (with 0.1 mg/mL of trypsin and chymotrypsin) at 37 °C (Right). (D) Design model of fzd48 in complex with the full-length toxin from PDB 6OQ5. (E) Segmented cryoEM map of TcdB in complex with fzd48 (in pink). (F) Example class averages with arrows indicating additional density due to fzd48 binding.

Fig. 3.

Design of anti-TcdB CSPG4-blocking minibinders. (A) Design models of the two CSPG4-blocking minibinder families. (B) Single cycle kinetic analysis of a group 1 (design cspg4) and group 2 (design ss2cspg18) minibinder amine conjugated to the surface across a 4-fold, 6-step dilution series of full-length TcdB with an upper concentration of 100 nM. (C) The design model docked into the CryoEM density map showing agreement between the observed density and the design. Lower Inset boxes highlight specific side chains on the design model resolved at the target:binder interface. (D) Disulfide stabilization of CSPG4-blocking minibinders with the location of the two disulfides introduced to make ss2cspg18 (Left) and a coomassie stained SDS-PAGE after a time course incubation in simulated intestinal fluid (with 0.1 mg/mL of trypsin and chymotrypsin) at 37 °C with cspg18 and the dual disulfide version (ss2cspg18) (Right).

Fig. 4.

Minibinder neutralization of TcdB in Vero cells. (A) Overview of the three experimental set-ups used to generate the neutralization data for the different classes of binders alone and in combination. (B) Sequence optimized fzd binders from group 1 (fzd5) and group 2 (fzd48, fzd84). (C) Protease resistant, sequence optimized fzd binder ssfzd84 compared with the protease susceptible designs. (D) Sequence optimized cspg designs from group 1 (cspg4) and group 2 (cspg18, cspg27, cspg35). (E) Titration of disulfide stabilized ss2cspg18 (x-axis) in the presence of four concentrations of ssfzd84 in the CSPG4-dependent system (F) Titration of ss2cspg18 (x-axis) with four concentrations of ssfzd84 in the dual receptor system. (G) Neutralization activity of fzd-cspg binder fusions secreted by S. boulardii. Fusion constructs between ss2cspg18 (c18) and ss2cspg27 (c27) fused to ssfzd84 (f84) with either a short GSG linker that should not allow simultaneous receptor engagement (dashed) or a longer (G4S)₄ linker that should enable simultaneous binding at the two sites (solid). In all cases, a single example from independent replicates is shown. Reported IC50 values (in text) are the average across independent replicates. All response curves plotted on the same axes were run in the same experiment. The circled number corresponds to the experiment system in panel A, while Sb indicates the use of S. boulardii secreted protein.

Fig. 5.

Design and characterization of TcsL-blocking minibinders. (A) RFdiffusion denoising trajectory starting from random noise placed above the target site at T = 50 to the fully denoised design at T = 0. (B) Design model of F4 binding to the RBD of TcsL. The inset box shows the hydrophobic pocket that accounts for the bulk of the interface. (C) Single cycle kinetic analysis with SPR of two of the optimized designs to TcsL (black) and TcdB (gray) with toxin RBD captured and a 6-step fourfold dilution series of the minibinder starting at 100 nM. (D) Neutralization of TcsL by parental (F4) and optimized (B4 and B10) designs on HCT116 cells (Left) and HULEC5a cells (Right). A single example from independent replicates is shown. Reported IC50 values (in text) are the average across independent replicates. All response curves plotted on the same axes were run in the same experiment. (E) Administration of a single bolus of TcsL alone or coincubated with a 1,000x molar excess of B4-M79. B4-M79 extended survival time from a median of 4.25 h to 8 h P = 0.0027 (log-rank test) (Upper). Administration of TcsL alone or coincubated with a 1,000x molar excess of B4-M79, followed by dosing with the same 1,000x molar excess (0.06 mg/kg) every 2 h, P = 0.0002 (Middle). Administration of TcsL followed by vehicle or B4-M79 1 h later, and repeated dosing every 2 h, P = 0.0012 (Lower). (F) Example histological images of the mouse lung from treatment group B with H&E stain showing the presence of edema in the vehicle control mice, characteristic of TcsL toxic shock. (G) Quantification of fluid in the lung of control mice vs treated mice in group B. Lines indicate mean ± SE. P < 0.0001 from a two-tailed unpaired t test.