Cytotoxicity of the Vibrio vulnificus MARTX toxin effector DUF5 is linked to the C2A subdomain

Abstract

The multifunctional-autoprocessing repeats-in-toxin (MARTX) toxins are bacterial protein toxins that serve as delivery platforms for cytotoxic effector domains. The domain of unknown function in position 5 (DUF5) effector domain is present in at least six different species' MARTX toxins and as a hypothetical protein in Photorhabdus spp. Its presence increases the potency of the Vibrio vulnificus MARTX toxin in mouse virulence studies, indicating DUF5 directly contributes to pathogenesis. In this work, DUF5 is shown to be cytotoxic when transiently expressed in HeLa cells. DUF5 localized to the plasma membrane dependent upon its C1 domain and the cells become rounded dependent upon its C2 domain. Both full-length DUF5 and the C2 domain caused growth inhibition when expressed in Saccharomyces cerevisiae. A structural model of DUF5 was generated based on the structure of Pasteurella multocida toxin facilitating localization of the cytotoxic activity to a 186 amino acid subdomain termed C2A. Within this subdomain, an alanine scanning mutagenesis revealed aspartate-3721 and arginine-3841 as residues critical for cytotoxicity. These residues were also essential for HeLa cell intoxication when purified DUF5 fused to anthrax toxin lethal factor was delivered cytosolically. Thermal shift experiments indicated that these conserved residues are important to maintain protein structure, rather than for catalysis. The Aeromonas hydrophila MARTX toxin DUF5(Ah) domain was also cytotoxic, while the weakly conserved C1-C2 domains from P. multocida toxin were not. Overall, this study is the first demonstration that DUF5 as found in MARTX toxins has cytotoxic activity that depends on conserved residues in the C2A subdomain.

Keywords: MARTX; Vibrio vulnificus; cytotoxin; effector domains; yeast screen.

FIGURE 1

DUF5 from V. vulnificus MARTX toxin is cytotoxic to HeLa cells. (A) Scale drawing of V. vulnificus MARTX and P. multocida PMT protein toxins with enlarged region showing C1, C2A, and C2B domains that are shared between the two toxins. (B–F) Epifluorescent and DIC images (200x) of HeLa epithelial cells transfected with pEGFP-N3 plasmid clones expressing EGFP (B), DUF5_Vv-EGFP (C–E), or C1C2_Pm-EGFP (F). Panels (D) and (E) enlarged 200% to show detail of localization of DUF5_Vv-EGFP and cell blebbing, respectively. Arrows in panel E indicate EGFP-positive cells in DIC only image. Percent of rounded cells in each cell type is quantified from three independent experiments (G) and expression of protein in transfected cells is shown by western blot detection using an anti-GFP antibody (H).

FIGURE 2

The C1 MLD of DUF5_Vv is necessary only for efficient cell rounding. (A) Structural model of DUF5_Vv generated in HHPRED and Modeller based on published structure of PMT. C1, C2A, and C2B subdomains are indicated. (B–D) Epifluorescent and DIC images (200x) of HeLa epithelial cells transfected with pEGFP-N3 plasmid clones expressing EGFP (B), DUF5_Vv-EGFP (C), or C2--GFP (E) and C2B-EGFP (F). Average of percent rounded cells in each cell type is quantified from three independent experiments (G) and expression of protein in transfected cells is shown by western blot detection using and anti-GFP antibody (H). Note that C2A-EGFP could not be detected due to consistent poor sample recovery from plates due to toxicity of this domain. (I–J) DIC images of HeLa cells intoxicated with 7 nM PA in combination with 3 nM purified unmodified LF_N (I), LF_N fused to DUF5_Vv (J) and LF_N fused to only the C1-C2A subdomains of DUF5_Vv (K). LF_N-DUF5_Ah was also tested for ability to round cells as compared to LF_N control samples in panels (L) and (M). Protein purity was assessed via SDS-PAGE (N). Three independent experiments were performed and the number of round cells quantified in panel O.

FIGURE 3

C2A is the cytotoxic subdomain of DUF5_Vv. (A). Schematic of proteins expressed in the panel. (B–H) Epifluorescent and DIC images (200x) of HeLa epithelial cells transfected with pEGFP-N3 plasmid clones expressing EGFP (B), DUF5_Vv-EGFP (C), C2-EGFP (D), C2A-EGFP (E) and C2B-EGFP (F). Average of percent rounded cells in each cell type is quantified from three independent experiments (G) and expression of protein in transfected cells is shown by western blot detection using and anti-GFP antibody (H). Note that C2A-EGFP could not be detected due to consistent poor sample recovery from plates due to toxicity of this domain. (I–J) DIC images of HeLa cells intoxicated with 7 nM PA in combination with 3 nM purified unmodified LF_N (I), LF_N fused to DUF5_Vv (J) and LF_N fused to only the C1-C2A subdomains of DUF5_Vv

FIGURE 4. DUF5 from A. hydrophila

is cytotoxic and causes cell rounding. LF_N-DUF5_Ah caused cell rounding when delivered to HeLa cells (A–B). Protein purity was assessed with SDS-PAGE (C). Rounding efficiency was comparable to DUF5_Vv, at all concentrations tested (D, E). Finally, release of LDH from intoxicated cells was measured (F, G), showing that there is no appreciable lysis when cells are intoxicated with either DUF5 protein.

FIGURE 5

Amino acid alignment generated in MacVector 12.6.0 of only the C2A and C2B subdomains. Grey shading indicates 100% identical residues. Triangles indicate that sections of sequence were removed during alignment calculations. Asterisk indicated residues changed to alanine via site directed mutagenesis and boxes indicate two aa identified as important for growth inhibition in yeast. Larger asterisks indicate residues G3948 and V3906 which were mutated to stop codons, while the last large asterisk indicates S3986, which was not targeted in the initial mutagenesis but was later found by structural modeling to potentially interact with R3841.

FIGURE 6

DUF5_Vv and the C2 domain cause growth inhibition when expressed in yeast. S. cerevisiae strain InvSc1 was transformed with pYC2 NT/A plasmid expressing proteins indicated at left or with empty vector (EV), actin crosslinking domain from V. cholerae MARTX (ACD), or the C2 domain with stop codons introduced at V3906 or G3048. Panels show 5 μl of 10-fold serial dilutions spotted to SC agar without uracil supplemented with either glucose (non-inducing) or galactose and raffinose (inducing). Panels at right show 12 h growth curves of cultures in SC broth with galactose.

FIGURE 7

D3721 and R3841 are important residues for growth inhibition of yeast. Growth inhibition in yeast for C2-D3721A and C2-D3721E (A) and DUF5_Vv-D3721A, DUF5_Vv-R3841A, DUF5 with both residues mutated to alanine (DARA) and DUF5 with swapped residues (DRRD) in panel B. See figure legend 4 for details. (C, D) Structural model of DUF5_Vv showing polar contacts of D3721 and R3841 and potential cross association of R3841 with S3986 in C2B. Color-coding is the same as in Fig. 2A. Panel E shows the purified 6 x His-tagged DUF5 and DUF5 D3721A proteins that were used in FTS experiments to determine melting temperature in panel F.

FIGURE 8

D3721 and R3841 are important residues for intoxication of HeLa cells. (A–E) DIC images of HeLa cells intoxicated for 24 h (upper) or 48 h (lower) with 7 nM PA in combination with 3 nM purified unmodified LF_N (A), LF_N fused to DUF5_Vv (B) and LF_N fused to only the C1-C2A subdomains of DUF5_Vv (C), or LF_N fused to only the C1-C2A subdomains of DUF5_Vv with D3721 (D) or R3841 (E) point mutations. Protein purity was assessed by SDS-PAGE in panel F. Three independent experiments were performed and cells were manually counted (G).

FIGURE 9

MARTX toxin undergoes autoprocessing upon entry into the host cell. Autoprocessing by the inositol hexakisphosphate bound cysteine protease domain releases other effector domains and allows them to perform their functions. DUF5 has been shown to be a stable protein when all the subdomains are present, and is able to efficiently round cells when the C2 domain is intact. When C2B subdomain in removed from the protein, leaving only C1-C2A, cell rounding is less efficient, presumably due to protein turnover. Therefore, C2B is hypothesized to be involved in stabilizing the interaction between DUF5 and the cellular target. C2A and C2B are required for a stable interaction with the target protein, but C2A alone is sufficient for cytotoxic activity.