A tripartite cytolytic toxin formed by Vibrio cholerae proteins with flagellum-facilitated secretion - pnas.org
Vibrio cholerae is known as the cause of cholera, a disease that can lead to fatal dehydration (1). The disease is caused by a few serogroups, and the main factor behind the symptoms is the cholera toxin (CT) encoded by genes located on a prophage mobile genetic element (CTX-φ) that induce severe disruption of intestinal cell function, leading to watery, secretory diarrhea (2). Most serogroups do not cause cholera, as they do not possess the genes for CT, but they cause other diseases [e.g., skin, wound, and gastrointestinal infections as well as bacteremia (3)]. The natural reservoirs of V. cholerae are aquatic sources such as rivers, brackish waters, and estuaries and are often associated with copepods, aquatic plants, and shellfish (4). The factors and mechanisms allowing V. cholerae and other Vibrionaceae to survive and thrive in harsh natural environments are of major interest to researchers (5).
V. cholerae is motile by virtue of a single polar flagellum. The flagellum export machinery and the virulence-associated type-III secretion system (fT3SS and vT3SS, respectively) are suggested to share a common ancestor (6), explaining their similar structure and molecular organization. The vT3SS allows the delivery of effector proteins through a hollow channel directly to the eukaryotic host cell (7), and flagellar proteins are delivered via the fT3SS channel during flagellum assembly. In the bacterial cytoplasm, effectors secreted by the vT3SS are stabilized by chaperones to prevent aggregation. These chaperones are often encoded by genes adjacent to those encoding the effectors (8). Flagellar proteins are similarly protected by chaperones before they are transported to the growing distal end of the flagellum (9).
We use Caenorhabditis elegans as a predatory organism model for identifying and assessing V. cholerae factors, other than CT, that may contribute to bacterial survival and persistence (10). With this model, we discovered a cytotoxin, MakA (motility-associated killing factor A), which we demonstrated to be an essential factor for the cytotoxic activity of V. cholerae in both C. elegans and Danio rerio (zebrafish) (11). We also demonstrated that secretion of MakA occurs via the flagellum in a manner that is undocumented in V. cholerae.
Our crystal structure of MakA revealed similarities to ClyA (11), the pore-forming toxin first identified in nonpathogenic Escherichia coli (12, 13) and, subsequently, also in Salmonella enterica (14). ClyA from E. coli is expressed from a monocistronic operon and oligomerizes into a dodecameric pore upon release via membrane vesicles (13, 15, 16). MakA is also structurally related to two proteins from Bacillus cereus, the hemolysin BL binding component B (HBL-B) and the NheA component of the Nhe nonhemolytic enterotoxin. Both of these are considered components of tripartite toxins (17). Recently, a tripartite toxin, AhlABC, was identified and structurally characterized as a pore-forming toxin in Aeromonas hydrophila, and the structure of soluble AhlB shares the general structure described for MakA (18). A similar toxin complex of three proteins, SmhABC from Serratia marcescens, was also reported (19). However, if and how the Ahl and Smh proteins are released during normal growth, or if there is a dedicated secretion system, remain unclear.
Here, we identify the proteins from the five V. cholerae genes, vca0880 through vca0884, that are coexpressed from the operon makDCBAE and analyze the crystal structures of MakA, MakB, and MakE. Our in vitro studies revealed that an equimolar combination of the MakA/B/E proteins acted as a tripartite cytotoxin causing lysis of red blood cells and cytotoxicity to epithelial cells. Examination of a large number of bacterial genomes revealed that the mak operon is present in many V. cholerae and other Vibrionaceae strains. These include Vibrio (Listonella) anguillarum, an inhabitant of estuarine and marine coastal ecosystems worldwide and the etiological agent of vibriosis in warm- and cold-water fish (20). The identification and structural characterization of the Mak proteins in V. cholerae presented here reveals a hitherto-unrecognized potential of many pathogenic Vibrionaceae strains to produce the tripartite Mak cytolytic toxin.
Results
Expression and Secretion of MakA, MakB, and MakE from V. cholerae.
We previously described a gene cluster in V. cholerae O1 strain A1552 comprising the genes makD, makC, makB, and makA (11). Later, an additional gene, makE, was identified downstream of makA. The five genes in the gene cluster are transcribed in the direction makD → makC → makB → makA → makE (Fig. 1A). By cloning and mutagenesis of each gene, we analyzed the encoded proteins. Previous tests with V. cholerae mutants defective in makA or makB demonstrated a clear attenuation of toxicity in C. elegans, whereas the effect of ΔmakD or ΔmakC was minimal (11).
Secretion of Mak proteins forming a tripartite cytolytic toxin. (A) Gene organization of the mak operon in V. cholerae O1 El Tor strain A1552. The gene loci vca0880, vca0881, vca0882, vca0883, and vca0884 have been denoted makD, makC, makB, makA, and makE, respectively. (B) Western immunoblotting was used to detect MakA, MakB, and MakE in whole-cell lysates and the supernatant. The study was done on wild-type V. cholerae A1552, ΔflhA, ΔmakA, ΔmakB, and ΔmakE mutants. The proteins were detected with antisera raised against MakA, MakB, and MakE, respectively. The asterisk indicates an unidentified protein also detected by the MakB antiserum. (C) C. elegans survival upon feeding on E. coli Top10 harboring makABE genes and mutants thereof. Data represent average survival percentage of three separate experiments. (D) Western blot analysis of MakA, MakB, or MakE secreted to the culture supernatant or in whole-cell lysates of E. coli Top10 expressing wild-type MakA or MakA mutants together with MakB and MakE. (E and F) Lysis of erythrocytes by Mak proteins in solution. A total of 250 nM MakA, MakB, and MakE was added to human erythrocyte (2% whole blood) suspensions and incubated at 37 °C for 120 min, either individually, sequentially, or in combination. ClyA (250 nM) was used as a positive control and phosphate-buffered saline (PBS) as the negative control. Hemolysis was examined using spectrophotometry. Data are from three independent experiments.
MakA, composed of 369 amino acids, is mainly secreted via the V. cholerae flagellum (11). To test if MakB and MakE (354 and 353 amino acids, respectively) also are secreted, we analyzed samples from cell lysates and supernatants using anti-MakA, anti-MakB, and anti-MakE antisera (Fig. 1B). Similar to the earlier findings with MakA, both MakB and MakE were detected in the supernatant. When MakA, MakB, and MakE were expressed by a flagellum-deficient (ΔflhA) mutant of V. cholerae, the secretion of all three proteins was drastically reduced (Fig. 1B). We therefore conclude that secretion of these Mak proteins is facilitated by the V. cholerae flagellum. It should be noted that MakA secretion occurred to a level of about 10% even from the ΔflhA mutant derivative (11).
We monitored the expression and secretion of Mak proteins in V. cholerae derivatives mutated in each of the mak genes (SI Appendix, Fig. S1). There was a strong reduction in the cellular MakA levels in the ΔmakB mutant, and, consequently, very little MakA was secreted. Also, the ΔmakC, ΔmakD, and ΔmakE mutants showed a somewhat lower cellular level of MakA, and, accordingly, the levels were lower in the supernatants (SI Appendix, Fig. S1A, lanes 3 and 9 through 12). The low MakA amount in the ΔmakB mutant remained also when the strain was complemented in trans with a plasmid expressing MakB (SI Appendix, Fig. S1E). Similarly, MakB expression and secretion levels were much reduced in the ΔmakC mutant (SI Appendix, Fig. S1 B and D, lanes 4 and 10). In addition, the amount of MakB in the ΔmakC mutant remained low in the strain complemented with a plasmid expressing MakC (SI Appendix, Fig. S1F). The results indicated that deletions of makB or makC caused strong polarity effects, in particular on their immediate downstream neighboring genes. We did not detect any secretion of MakC (SI Appendix, Fig. S1D), and we consider that both MakC and MakD have accessory roles and remain in the bacterial cytoplasm or periplasm.
The Vibrio flagellum is covered by a sheath that surrounds the filament (21, 22). Secretion of the antisigma factor FlgM through the filament was demonstrated, which led to the suggestion that there may be a sheath opening at the flagellar tip. Studies suggest that the H-ring is essential for outer membrane penetration and assembly of the flagellum (23). Vibrios lacking FlgT (an H-ring component) synthesize some flagella that failed to penetrate the outer membrane, forming periplasmic flagella, and it was considered that the H-ring might play a role in sheath formation. In the absence of any known mutant lacking the sheath per se, we tested if secretion of the MakA/B/E proteins was affected by a ΔflgT mutation. The ΔflgT mutation did not cause any difference in secretion of the MakA/B/E proteins compared to the wild-type strain (SI Appendix, Fig. S2). Evidently, the proposed role of the H-ring to facilitate outer membrane penetration of the sheathed flagellum was not essential for Mak secretion.
The fact that MakB and MakE were also secreted in a flagella-facilitated manner prompted us to test if they also could be secreted from E. coli as was shown earlier for MakA (11). Using a plasmid with an inducible promoter comprising the makB+, makA+, and makE+ genes, we tested if it mediated a toxic effect on C. elegans, similar to what we previously observed for MakA. As shown in Fig. 1C and D (lane 2), when MakA/B/E were expressed in E. coli, all three proteins were readily secreted into the supernatant and caused killing of C. elegans. We therefore conclude that the other two proteins, MakC and MakD from the mak operon, are not essential for the secretion of MakA/B/E and their toxin activity.
The MakA/B/E Tripartite Has Hemolytic and Cytotoxic Activity.
We tested if recombinant MakA, MakB, and MakE as individual proteins or in bipartite/tripartite combinations would cause erythrocyte lysis by applying them on blood agar plates (SI Appendix, Fig. S3A). As positive control, we used the cytolytic protein ClyA (12, 13). We observed a clear hemolytic effect of the equimolar combination of the tripartite MakA/B/E but none for the individual proteins or bipartite combinations.
The cytolytic activity was further assessed by quantitative hemolysis assays with erythrocytes in solution (Fig. 1 E and F). The proteins (each at 250 nM) were introduced in different sequential orders over 120 min (Fig. 1 E and F). Maximum hemolytic activity was observed with the order MakA → MakB → MakE with intervals of 15 min (Fig. 1E, column 1) and when all three components were premixed prior to their addition to the erythrocytes (Fig. 1E, column 7). Notably, the hemolytic activity of the equimolar MakA/B/E tripartite was similar to that observed with 250 nM ClyA (Fig. 1E, column 11). Hemolysis, to about a 10 to 15% lower level, was also observed when MakA and MakB were first introduced as a premix and MakE was added last (Fig. 1E, column 2). Interestingly, when MakB was added first, followed by MakA and, finally, MakE, no hemolysis was detected within the 120-min assay (Fig. 1E, column 3). Similarly, when MakB and MakE were first introduced as a premix and MakA was added 15 min later, there was no detectable hemolysis (Fig. 1E, column 4). A low level of hemolysis (5 to 10%) was detected when the proteins were added separately in the order MakE → MakA → MakB or in the order MakAE → MakB (Fig. 1, columns 5 and 6, respectively). The individual Mak proteins did not cause any detectable hemolysis (Fig. 1, columns 8 through 10). We conclude that all three proteins were required for cytolytic activity and that there was a clear dependence on the order of addition. Only when MakA was added first, alone or in a premix with MakB, did the tripartite yield high-level hemolysis.
The MakA/B/E tripartite on surfaces of red blood cells was visualized using confocal microscopy (Fig. 2A). The binding kinetics of the Mak tripartite to a lipid membrane in real time was studied using supported synthetic lipid membrane bilayers of complex composition. The analysis was performed in a quartz crystal microbalance with dissipation monitoring (QCM-D) of the protein adsorption (SI Appendix, Fig. S3B), confirming binding to the bilayer surface. The proteins were stably bound and were not washed off by rinsing.
Activity of the tripartite MakA, MakB, and MakE cytotoxin on human cells. (A) Human red blood cells (1% whole blood) in PBS were treated with an Alexa568-labeled MakA, MakB, and MakE tripartite combination (125 nM of each). Binding and membrane accumulation of the Alexa568-labeled proteins were assessed by confocal laser scanning microscopy. Arrowheads (black) indicate membrane accumulation of Alexa568-MakA/B/E tripartite complex. (Scale bars, 10 µm.) (B) Liposomes from E. coli total LE were treated with vehicle (Tris 20 mM), MakA, MakB, MakE, or the MakA/B/E combination for 90 min and stained with 1.5% uranyl acetate solution. Micrographs were captured with TEM. Inset indicates star-shaped oligomers of MakA and larger complexes of the MakA/B/E tripartite complex. (Scale bars, 100 nm.) (C) Caco-2 cells were treated with Alexa568-labeled MakA, MakB, and MakE (250 nM of each) either individually or in the tripartite combination at equimolar concentrations. The cell-associated Alexa568-labeled protein was assessed after 24 h by flow cytometry analysis. (D) Caco-2 cells were treated with MakA, MakB, and MakE proteins (250 nM of each), either individually or in the tripartite combination, for 24 h, and toxicity was assessed by the MTS cell viability assay. Data points represent four biologically independent experiments; bar graphs show mean ± SD. Significance was determined from biological replicates using one-way ANOVA with Dunnett's multiple comparisons test. **P < 0.01, ns = not significant.
Transmission electron microscopy (TEM) was used to investigate if the MakA/B/E tripartite or the individual proteins would form recognizable oligomeric or pore-like assemblies on liposomes prepared from E. coli total lipid extracts. Assemblies were indeed observed with the MakA/B/E tripartite (Fig. 2B) and among the individual Mak proteins; only MakA formed seemingly well-organized star-shaped oligomers.
We used human colon cancer cells (Caco-2 cells) to investigate if Alexa568-labeled Mak proteins would bind to epithelial cells. Caco-2 cells are human colon adenocarcinoma cells isolated from a primary colonic tumor (24). Flow cytometry was used to monitor binding and/or uptake. The results indicated that a distinctly higher number of cells were labeled with the MakA/B/E tripartite than with any of the three proteins tested separately (Fig. 2C). Of the proteins tested individually, MakA showed higher binding/uptake than MakB or MakE. We also assessed the viability of the Caco-2 cells upon 24-h treatment with the Mak proteins. The MakA/B/E tripartite displayed the most pronounced cytotoxic activity, resulting in a loss of viability in about 80% of the Caco-2 cells (Fig. 2D). A lower, but still significant, degree of cytotoxicity was observed with MakA alone.
To determine if the MakA/B/E tripartite disrupted intracellular structures of epithelial cells, Caco-2 cells were stained for 1) actin filaments using phalloidin-fluorescein isothiocyanate (FITC), 2) the cis-Golgi marker, GM130, or 3) the mitochondrial marker, Tom20. The tripartite combination induced distinct changes in the cellular distribution of the three markers (Fig. 3 A–C). Importantly, the MakA/B/E tripartite disrupted the actin filaments (Fig. 3A) and induced Golgi fragmentation (Fig. 3B). It also caused rounding of mitochondria, as evidenced by redistribution of Tom20 staining from filamentous mitochondria to round structures (Fig. 3C). MakB or MakE failed to induce any detectable change of the cellular distribution of the actin, Golgi or mitochondrial markers, whereas MakA had a weak effect on Golgi and mitochondria (SI Appendix, Fig. S4 A–C). Results with Caco-2 cells stained for the mitochondrial potential marker tetramethylrhodamine methyl ester (TMRM) indicated that the MakA/B/E tripartite caused depolarization of mitochondria (Fig. 3 D and E). The individual components had very little to no effect on the mitochondrial potential (Fig. 3 D and E). Similar results were obtained with HCT8 cells (SI Appendix, Fig. S4 D and E) that are human adenocarcinoma cells from the ileocecal region (25). The effect on total cellular adenosine triphosphate (ATP) was measured in Caco-2 cells and showed that the tripartite caused the most severe ATP depletion. (Fig. 3F). Importantly, the tripartite caused a time-dependent decrease in total cellular ATP content (Fig. 3G). Together, these results suggest that the MakA/B/E tripartite mediated disruption of cell organelles, leading to dysfunction of mitochondria.
Effect of the MakA/B/E tripartite on intracellular structures and organelles. Caco-2 cells treated with vehicle or MakA/B/E (250 nM, equimolar concentration) for 24 h were examined by confocal laser scanning microscopy. Nuclei were counterstained with DAPI or Hoechst 33342. (Scale bars, 10 µm.) (A) Effect on actin filaments. Cells were stained with phalloidin-FITC to visualize actin filaments. Arrowheads (white) indicate disruption of actin filaments. (B) Effect on the Golgi apparatus. Cells were stained with a cis-Golgi marker, GM130. Arrowhead (white) indicates changes in the cellular distribution of the Golgi complexes. (C) Effect on mitochondria. Immunofluorescence detection was performed with antibodies against Tom20 to visualize mitochondria. Arrowhead (white) indicates swelling of mitochondria. (D) Caco-2 cells treated with vehicle, individual components of the tripartite complex, or MakA/B/E (24 h) and stained with mitochondrial potential marker, TMRM (250 nM, 30 min). (Scale bars, 10 µm.) (E) Quantification of fluorescence intensity displayed upon treatment of Caco-2 cells as shown in D. Bar graphs show mean ± SEM. Data points represent fluorescence intensity of 68 to 81 individual cells. (F) Effect on cellular ATP content. Caco-2 cells were treated with 250 nM MakA, MakB, and MakE proteins individually and the tripartite combination at equimolar concentration for 48 h. Histogram represents data from four biologically independent experiments. Bar graphs show mean ± SD. (G) Caco-2 cells were treated with an equimolar concentration of the MakA/B/E tripartite (250 nM) in a time-dependent manner. Histogram represents data from three independent experiments. Bar graphs show mean ± SD. In E and F, the significance was determined from replicates using one-way ANOVA with Dunnett's multiple comparisons test. *P < 0.05, **P < 0.01, ns = not significant.
A tripartite cytolytic toxin formed by Vibrio cholerae proteins with flagellum-facilitated secretion - pnas.org
Comments
Post a Comment