Insertion-trigger residues differentially modulate endosomal escape by cytotoxic necrotizing factor toxins

Abstract

The cellular specificity, potency, and modular nature of bacterial protein toxins enable their application for targeted cytosolic delivery of therapeutic cargo. Efficient endosomal escape is a critical step in the design of bacterial toxin-inspired drug delivery (BTIDD) vehicles to avoid lysosomal degradation and promote optimal cargo delivery. The cytotoxic necrotizing factor (CNF) family of modular toxins represents a useful model for investigating cargo-delivery mechanisms due to the availability of many homologs with high sequence identity, their flexibility in swapping domains, and their differential activity profiles. Previously, we found that CNFy is more sensitive to endosomal acidification inhibitors than CNF1 and CNF2. Here, we report that CNF3 is even less sensitive than CNF1/2. We identified two amino acid residues within the putative translocation domain (E374 and E412 in CNFy, Q373 and S411 in CNF3) that differentiate between these two toxins. Swapping these corresponding residues in each toxin changed the sensitivity to endosomal acidification and efficiency of cargo-delivery to be more similar to the other toxin. Results suggested that trafficking to the more acidic late endosome is required for cargo delivery by CNFy but not CNF3. This model was supported by results from toxin treatment of cells in the presence of NH₄Cl, which blocks endosomal acidification, and of small-molecule inhibitors EGA, which blocks trafficking to late endosomes, and ABMA, which blocks endosomal escape and trafficking to the lysosomal degradative pathway. These findings suggest that it is possible to fine-tune endosomal escape and cytosolic cargo delivery efficiency in designing BTIDD platforms.

Keywords: bacterial toxin; drug delivery system; fusion protein; molecular evolution; protein chimera; protein deamidation; protein engineering; protein translocation; small GTPase; structure–function.

Figure 1

Sensitivity of wild-type CNF toxins to endosomal acidification. Shown are dose–response curves to NH₄Cl treatment of the wild-type CNF toxins in the SRE-luciferase assay, as described in the Experimental procedures. HEK293T cells were treated with NH₄Cl for 30 min prior to treatment with the indicated wild-type CNF toxins and assayed for Firefly/Renilla activity after 6-h incubation at a concentration of (A) 100 ng/ml or (B) corresponding to their respective EC₅₀ values (0.01 nM CNF1 (green), 0.10 nM CNF2 (blue), 0.02 nM CNF3 (purple), and 0.80 nM CNFy (red)). Relative activity indicates the fold activation compared with no-inhibitor treatment. Corresponding scatter plots with all data points used to derive the best fit lines and mean values are shown in Fig. S2.

Figure 2

Sensitivity of CNF3y chimeric toxins to endosomal acidification.A, shown is a schematic depicting the joining sites of the chimeric constructs tested in (B–D), where the functional domains of CNF3 delivery vehicle and CNFy cargo are indicated. B–D, shown are dose–response curves to NH₄Cl treatment of the wild-type CNF3 and CNFy toxins and CNF3y chimeric toxins in the SRE-luciferase assay, as described in the Experimental procedures. HEK293T cells were treated similarly as in Figure 1B, where the toxin concentration used was at its EC₅₀ value. Relative activity indicates the fold activation compared with no-inhibitor treatment. CNF3 (purple circles); CNFy (red circles); CNF3y-519 (teal diamonds); CNF3y-223 (gray diamonds); CNF3y-428 (pink diamonds); CNF3y-317 (orange-brown diamonds); CNF3y-412 (green diamonds); CNF3y-375 (black diamonds); CNF3y-349 (light blue diamonds). Corresponding scatter plots with all data points used to derive the best fit lines and mean values are shown in Fig. S3. B2, secondary binding domain; DUF4765, Domain of Unknown Function.

Figure 3

The effect of swapping amino acid residues in the putative HLH region of CNF3 and CNFy on dose response and sensitivity to endosomal acidification.A, shown is the alignment of the putative HLH region of the indicated CNF toxin homologs. The scale bar shown on top corresponds to residue numbers for CNF1 and CNFy in this region. The four red arrowheads indicate the acidic residues critical for CNF1 activity. The red arrows indicate two critical acidic residues that distinguished between the CNF3 and CNFy sensitivity to endosomal acidification inhibitors. The blue bars at the bottom indicate the alpha helices found in the structure of CNFy (PDB 6YHK). B–G, the mutant toxins were analyzed by the SRE-luciferase assay, as described in the Experimental procedures. B–D, dose–response curves comparing point mutants of CNFy and CNF3, as indicated. E–G, effect of NH₄Cl on the dose–response curves of mutant and wild-type CNF3 and CNFy at their respective EC₅₀ concentrations, as described in Figure 1B and Experimental procedures. Corresponding scatter plots with all data points used to derive the best fit lines and mean values are shown in Fig. S4.

Figure 4

Effect of EGA and ABMA on wild-type CNFy and CNF3 toxin activity. HEK293T cells were treated with EGA or ABMA for 60 min prior to treatment with the indicated toxin concentration for 6 h. Cells were then lysed and analyzed by SRE-luciferase assay, as described in the Experimental procedures. A and B, shown are scatter plots for the effect of EGA on the dose–response curves for (A) CNFy (0 μM EGA black, 0.5 μM red, 1 μM green, 1.5 μM dark blue, 2 μM light blue, 3 μM pink, 4 μM orange, 5 μM gray) and (B) CNF3 (0 μM EGA black, 0.5 μM red, 5 μM green, 10 μM dark blue, 15 μM light blue, 20 μM pink, 25 μM gray). C and D, shown are scatter plots for the effect of ABMA on the dose–response curves for (C) CNFy (0 μM ABMA black, 1 μM red, 5 μM green, 10 μM light blue, 15 μM dark blue, 20 μM pink) and (D) CNF3 (0 μM ABMA black, 1 μM red, 5 μM green, 10 μM light blue, 15 μM dark blue, 20 μM pink). E, shown are the effects of EGA treatment on the activity of CNF3 toxin (purple) and CNFy (red) at toxin concentrations of 0.3 nM and 6 nM, respectively. F, shown are the effects of ABMA treatment on the activity of CNF3 toxin (purple) and CNFy (red) at toxin concentrations of 0.25 nM and 10 nM, respectively. Concentrations of inhibitor higher than 40 μM for ABMA and 50 μM for EGA were toxic to the cells (data not shown). Corresponding scatter plots with all data points used to derive the best fit lines and mean values are shown in Fig. S5.

Figure 5

Structures of the proposed insertion trigger regions in CNFy and CNF3.A, shown is a ribbon diagram representation of the structure of CNFy (PDB 6YHK) generated using ChimeraX. Cyan, activity domain (residues 718–1014). Pink, domain of unknown function (residues 522–700). Blue, subdomain of translocation module (residues 424–522). Beige, N-terminal translocation and receptor-binding module (residues 1–424). The electrostatic surface is shown for residues 344 to 423. B, shown is a bottom view of the structure in (A) with acidic residues labeled. C, shown is a similar view of the structure of CNF3 generated by HHpred-Modeller using the CNFy structure as the template, with the residues corresponding to those in (B) labeled.

Figure 6

Proposed mechanism of cytosolic cargo delivery by CNF3 and CNFy toxins. Shown is a diagram of a proposed mechanism for intoxication and cytosolic cargo delivery by CNF3 and CNFy. Based on results from their differential response to NH₄Cl treatment, endosomal escape of CNF3 cargo is proposed to occur at a higher endosomal pH than for CNFy cargo. At high concentrations of inhibitor, both EGA and ABMA blocked CNF3 and CNFy activity, with EGA inhibiting CNFy more than CNF3 and ABMA blocking both toxins equally. However, at low inhibitor concentrations, CNF3 and CNFy toxin activities were enhanced by both inhibitors, presumably due to the inhibitors preventing trafficking to the lysosomal degradative pathway, which in each case enabled more toxin to escape from the endosome. Since CNF3 can escape the endosome at a higher pH than CNFy, EGA blockade of early to late endosomal trafficking enhanced CNF3 activity more than CNFy activity.