Host cell interactions of outer membrane vesicle-associated virulence factors of enterohemorrhagic Escherichia coli O157: Intracellular delivery, trafficking and mechanisms of cell injury

Abstract

Outer membrane vesicles (OMVs) are important tools in bacterial virulence but their role in the pathogenesis of infections caused by enterohemorrhagic Escherichia coli (EHEC) O157, the leading cause of life-threatening hemolytic uremic syndrome, is poorly understood. Using proteomics, electron and confocal laser scanning microscopy, immunoblotting, and bioassays, we investigated OMVs secreted by EHEC O157 clinical isolates for virulence factors cargoes, interactions with pathogenetically relevant human cells, and mechanisms of cell injury. We demonstrate that O157 OMVs carry a cocktail of key virulence factors of EHEC O157 including Shiga toxin 2a (Stx2a), cytolethal distending toxin V (CdtV), EHEC hemolysin, and flagellin. The toxins are internalized by cells via dynamin-dependent endocytosis of OMVs and differentially separate from vesicles during intracellular trafficking. Stx2a and CdtV-B, the DNase-like CdtV subunit, separate from OMVs in early endosomes. Stx2a is trafficked, in association with its receptor globotriaosylceramide within detergent-resistant membranes, to the Golgi complex and the endoplasmic reticulum from where the catalytic Stx2a A1 fragment is translocated to the cytosol. CdtV-B is, after its retrograde transport to the endoplasmic reticulum, translocated to the nucleus to reach DNA. CdtV-A and CdtV-C subunits remain OMV-associated and are sorted with OMVs to lysosomes. EHEC hemolysin separates from OMVs in lysosomes and targets mitochondria. The OMV-delivered CdtV-B causes cellular DNA damage, which activates DNA damage responses leading to G2 cell cycle arrest. The arrested cells ultimately die of apoptosis induced by Stx2a and CdtV via caspase-9 activation. By demonstrating that naturally secreted EHEC O157 OMVs carry and deliver into cells a cocktail of biologically active virulence factors, thereby causing cell death, and by performing first comprehensive analysis of intracellular trafficking of OMVs and OMV-delivered virulence factors, we provide new insights into the pathogenesis of EHEC O157 infections. Our data have implications for considering O157 OMVs as vaccine candidates.

Fig 1. EHEC O157 OMVs carry a cocktail of virulence factors.

(A) Electron microscopy of ultrathin cryosections of LB agar cultures of strains 5791/99 and 493/89 stained with anti-E. coli O157 LPS antibody and Protein A Gold or Protein A Gold alone (control). Examples of OMVs (v) and bacteria (b) are indicated. Frames depict OMVs located in other microscopic fields than the producing bacteria. Scale bars are 150 nm. (B) Distribution of virulence factors in OMVs and OMV-free supernatants determined by immunoblot with antibodies against OmpA (an OMV marker) and the indicated virulence proteins. (C) Distribution of virulence factors in OptiPrep density gradient fractions (1 to 12, collected from top to bottom) of O157 OMVs determined by immunoblot. The lanes designated OMV contain non-fractionated OMVs (positive control). (D) Localizations of virulence factors within 5791/99 OMVs visualized by electron microscopy of ultrathin cryosections of OptipPrep-purified OMVs stained with antibodies against the indicated virulence factors and Protein A Gold (panels 1 and 2) or Protein A Gold alone (panels 3; control). Bars are 100 nm. Note that by electron microscopy of immunostained ultrathin cryosections only 10%–15% of the total antigen present in the section can be detected [31] explaining relatively low numbers of signals observed for most virulence factors. (E) Immunoblots of proteinase K (PK)-untreated (PK-) and PK-treated (PK+) O157 OMVs either intact (EDTA-) or lysed with 0.1 M EDTA (EDTA+) with the indicated antibodies. (Anti-CdtV-C antibody is not suitable for electron microscopy but detects CdtV-C by immunoblot).

Fig 2. OMVs are taken up by target cells via dynamin-dependent endocytosis.

(A, D, G) Kinetics of R-18 OMV uptake by indicated cell types. Fluorescence of cells incubated with OMVs was normalized to that of OMVs without cells (net fluorescence intensity). (C, F) CLSM visualization of OMV uptake by Caco-2 cells (C) and HBMEC (F) after 30 min and 24 h of incubation. Green, OMVs; red, actin; blue, nuclei. Confocal Z-stack projections are included at upper/right sides. Crosshairs show the position of the xy and yz planes. Scale bars are 10 μm. (B, E, H) Uptake of R-18 OMVs by Caco-2 cells (B), HBMEC (E), and HRGEC (H) which were pretreated with the indicated endocytosis inhibitors. OMV uptake in the presence of inhibitors was expressed as the percentage of OMV uptake by inhibitor-untreated cells (100%). Data in A, D, G, and B, E, H are means ± standard deviations from three independent experiments. ** p < 0.01, and *** p < 0.001 compared to inhibitor-untreated cells (one-way ANOVA).

Fig 3. OMV-associated virulence proteins differentially separate from OMVs during intracellular trafficking.

(A) CLSM of HBMEC exposed to 5791/99 OMVs for 30 min at 4°C (OMV binding; panels 0 min) followed by 15 min to 20 h at 37°C (OMV internalization). The indicated single fluorescence channels are shown in the right panels (enlargements are displayed in S9 and S10 Figs) and the merged images in the left panels (red, OMVs; green, virulence factors; blue, nuclei; yellow, colocalized red and green signals). The percentages of colocalizations between OMVs and virulence proteins were determined with the BioImageXD6 tool (white numbers). Scale bars are 10 μm. (B) Graphical summary of colocalizations of virulence factors with OMVs during time based on CLSM data shown in A. (Means of colocalizations from at least five different samples are shown in A and B; standard deviations and significance analysis see in S8A Fig).

Fig 4. Intracellular trafficking of O157 OMVs.

(A) CLSM of HBMEC preincubated with OMVs 5791/99 for 30 min at 4°C, and postincubated at 37°C for 30 min to 20 h. The indicated single fluorescence channels are shown in the right panels (enlargements are displayed in S12 and S13 Figs) and the merged images in the left panels (green, OMVs; red, compartment-specific marker proteins; blue, nuclei; yellow, colocalized green and red signals). The percentages of OMV colocalizations with compartment-specific marker proteins (white numbers) and with nucleus (blue numbers in panels ER) were calculated with the BioImageXD6 tool. Scale bars are 10 μm. (B) Graphical summary of CLSM data shown in A. (Means of colocalizations from at least five different samples are shown in A and B; for standard deviations and significance analysis see S8B Fig). (C) Immunoblot detection of OMVs (anti-OmpA antibody) in isolated subcellular fractions of HBMEC which were postincubated for the times indicated with 5791/99 OMVs or for 72 h without OMVs or with TRIS-HCl OMV buffer (negative controls); 5791/99 OMVs without cells (OMV ctrl) were a positive control. (D) Densitometric quantification of OmpA lysosomal signals shown in C. Ee, early endosomes; Le/Lyso, late endosomes/lysosomes; ER, endoplasmic reticulum; Mito, mitochondria; Nucl, nucleus; Cyto, cytosol; DU, densitometric unit.

Fig 5. Intracellular trafficking of OMV O157-delivered Stx2a.

(A) CLSM of HBMEC preincubated with OMVs 5791/99 for 30 min at 4°C, and postincubated at 37°C for 30 min to 20 h. The indicated single fluorescence channels are shown in the right panels (enlargements are displayed in S15 and S16 Figs) and the merged images in the left panels (green, Stx2a; red, compartment-specific marker proteins; blue, nuclei; yellow, colocalized green and red signals). The percentages of Stx2a colocalizations with compartment-specific marker proteins (white numbers) and with nucleus (blue numbers in panels ER) were calculated with the BioImageXD6 tool. Scale bars are 10 μm. (B) Graphical summary of Stx2a colocalizations with subcellular compartments based on CLSM data shown in A, and with OMVs (based on data shown in Fig 3A). (Means of colocalizations from at least five different samples are shown in A and B; for standard deviations and significance analysis see S8C Fig). (C) Immunoblot detection of Stx2a A subunit in isolated subcellular fractions of HBMEC which were postincubated for the times indicated with 5791/99 OMVs or for 72 h without OMVs or with TRIS-HCl OMV buffer (negative controls); 5791/99 OMVs without cells (OMV ctrl) were a positive control. Molecular weights of Stx2a A subunit (32 kDa) and Stx2a A1 fragment (27.5 kDa) are shown on the right side. (D) Densitometric quantification of Stx2a A/A1 signals shown in C. For abbreviations see legend to Fig 4.

Fig 6. The role of Gb3 in the uptake and retrograde transport of OMV-delivered Stx2a.

(A, D) FACS analysis of Gb3 content in (A) PPMP-untreated (no PPMP) and PPMP-treated (+PPMP) HBMEC and (D) DLD-1 cells stained with fluorescein isothiocyanate (FITC)-conjugated anti-CD77/Gb3 antibody or unstained (control). One representative experiment is shown for each cell line (geometric mean fluorescence ± standard deviations from three independent experiments are shown in S18A and S18B Fig). (B, E) CLSM analysis of trafficking of OMV-delivered and free Stx2a (control) into the endoplasmic reticulum (ER) and late endosomes/lysosomes (Le-Lyso) in (B) PPMP-untreated and PPMP-treated HBMEC (20 h) and (E) DLD-1 cells (4 h). The indicated single fluorescence channels are shown in the right panels (enlargements are displayed in S19 Fig) and the merged images in the left panels (green, Stx2a or OMVs, as indicated; red, PDI or CD63, as indicated; blue, nuclei; yellow, colocalized green and red signals). The percentages of colocalizations of the respective signals (white numbers) were calculated with the BioImageXD6 tool (means of colocalizations from three different samples are shown). (C) Uptake of 5791/99 OMVs by DLD-1 cells after 4 h of incubation. Untreated cells (no OMV) were a negative control. Green, OMVs; red, actin; blue, nuclei. Confocal Z-stack projections are included at upper/right sides. Crosshairs show the position of the xy and yz planes. Scale bars in all panels are 10 μm. (F, G) Immunoblot detection of Stx2a in isolated endoplasmic reticulum (ER) and lysosomal (Lyso) fractions of (F) PPMP-treated HBMEC and (G) DLD-1 cells after 20 h and 4 h of incubation, respectively, with the indicated samples or left untreated (no OMV); 5791/99 OMVs without cells (OMV ctrl) were a positive control.

Fig 7. Retrograde transport of OMV-delivered Stx2a involves the toxin´s interactions with DRM-associated Gb3.

(A, B, C) Colocalization of Stx2a with Gb3 (A), and colocalization of Gb3 with the Golgi marker GM130 (B), and the endoplasmic reticulum marker BiP (C) in cells exposed to 5791/99 OMVs or free Stx2a (control) for 4 h (A, C) or 90 min (B) and processed for CLSM after extraction with a Triton X-100-containing buffer (1 min on ice). Green, Gb3; red, Stx2a (A) or GM130 (B) or BiP (C); blue, nuclei; yellow, colocalized green and red signals. (D) Colocalization of OMV-delivered Stx2a and free Stx2a (control) with the Golgi complex (90 min) and the endoplasmic reticulum (4 h) in HBMEC untreated (no Triton) or pretreated with Triton X-100 (+Triton) before being processed for CLSM. Green, Stx2a; red, GM130 or BiP, as indicated; blue, nuclei; yellow, colocalized green and red signals. In all pictures, the indicated single fluorescence channels are shown in the right panels (enlargements are displayed in S20 Fig) and the merged images in the left panels. The percentages of colocalizations of the respective signals (white numbers) were calculated with the BioImageXD6 tool (means of colocalizations from three different samples are shown). Scale bars are 10 μm.

Fig 8. Intracellular trafficking of OMV O157-delivered CdtV-B.

(A) CLSM of HBMEC preincubated with OMVs 5791/99 for 30 min at 4°C, and postincubated at 37°C for 30 min to 20 h. The indicated single fluorescence channels are shown in the right panels (enlargements are displayed in S21 and S22 Figs) and the merged images in the left panels (green, CdtV-B; red, compartment-specific marker proteins; blue, nuclei; yellow, colocalized green and red signals). The percentages of CdtV-B colocalizations with compartment-specific marker proteins (white numbers) and with nucleus (blue numbers in panels ER) were calculated with the BioImageXD6 tool. Scale bars are 10 μm. (B) Graphical summary of CdtV-B colocalizations with subcellular compartments based on CLSM data shown in A, and with OMVs (based on data shown in Fig 3A). (Means of colocalizations from at least five different samples are shown in A and B; for standard deviations and significance analysis see S8D Fig). (C) Immunoblot detection of CdtV-B in isolated subcellular fractions of HBMEC which were postincubated for the times indicated with 5791/99 OMVs, or for 72 h without OMVs or with TRIS-HCl OMV buffer (negative controls); 5791/99 OMVs without cells (OMV ctrl) were a positive control. (D) Densitometric quantification of CdtV-B signals shown in C. For abbreviations see legend to Fig 4.

Fig 9. Intracellular trafficking of OMV O157-delivered CdtV-A.

(A) CLSM of HBMEC preincubated with OMVs 5791/99 for 30 min at 4°C, and postincubated at 37°C for 30 min to 20 h. The indicated single fluorescence channels are shown in the right panels (enlargements are displayed in S23 and S24 Figs) and the merged images in the left panels (green, CdtV-A; red, compartment-specific marker proteins; blue, nuclei; yellow, colocalized green and red signals). The percentages of CdtV-A colocalizations with compartment-specific marker proteins (white numbers) and with nucleus (blue numbers in panels ER) were calculated with the BioImageXD6 tool. Scale bars are 10 μm. (B) Graphical summary of CdtV-A colocalizations with subcellular compartments based on CLSM data shown in A, and with OMVs (based on data shown in Fig 3A). (Means of colocalizations from at least three different samples are shown in A and B; for standard deviations and significance analysis see S8E Fig). (C) Immunoblot detection of CdtV-A in isolated subcellular fractions of HBMEC which were postincubated for the times indicated with 5791/99 OMVs or for 72 h without OMVs or with TRIS-HCl OMV buffer (negative controls); 5791/99 OMVs without cells (OMV ctrl) were a positive control. (D) Densitometric quantification of CdtV-A signals shown in C. For abbreviations see legend to Fig 4.

Fig 10. Intracellular trafficking of OMV O157-delivered CdtV-C.

(A) CLSM of HBMEC preincubated with OMVs 5791/99 for 30 min at 4°C, and postincubated at 37°C for 30 min to 20 h. The indicated single fluorescence channels are shown in the right panels (enlargements are displayed in S25 and S26 Figs) and the merged images in the left panels (green, CdtV-C; red, compartment-specific marker proteins; blue, nuclei; yellow, colocalized green and red signals). The percentages of CdtV-C colocalizations with compartment-specific marker proteins (white numbers) and with nucleus (blue numbers in panels ER) were calculated with the BioImageXD6 tool. Scale bars are 10 μm. (B) Graphical summary of CdtV-C colocalizations with subcellular compartments based on CLSM data shown in A, and with OMVs (based on data shown in Fig 3A). (Means of colocalizations from at least three different samples are shown in A and B; for standard deviations and significance analysis see S8F Fig). (C) Immunoblot detection of CdtV-C in isolated subcellular fractions of HBMEC which were postincubated for the times indicated with 5791/99 OMVs, or for 72 h without OMVs or with TRIS-HCl OMV buffer (negative controls); 5791/99 OMVs without cells (OMV ctrl) served as a positive control. (D) Densitometric quantification of CdtV-C signals shown in C. For abbreviations see legend to Fig 4.

Fig 11. Retrograde transport of OMV-delivered CdtV-B does not require CdtV-A and CdtV-C.

HBMEC were preincubated (30 min, 4°C) with recombinant CdtV-B-containing OMVs from strain BL21(cdtV-B) or with CdtV-B-lacking OMVs from BL21(pET23) vector control and postincubated at 37°C for 90 min to 20 h. (A) OMV uptake was determined by CLSM after 4 h. Green, OMVs; red, actin; blue, nuclei. Confocal Z-stack projections are included at upper/right sides. Crosshairs show the position of the xy and yz planes. (B, C) Trafficking of BL21(cdtV-B) OMVs and OMV-delivered CdtV-B, and separation of CdtV-B from the vesicles during the trafficking were determined by CLSM. (B) Colocalization of CdtV-B with OMVs upon OMV cellular binding (time 0 min). (C) Colocalizations of BL21(cdtV-B) OMVs with late endosomes/lysosomes (panels OMV/Le-Lyso), of CdtV-B with OMVs (panels CdtV-B/OMV), and of CdtV-B with the indicated subcellular compartments after postincubation at 37°C for 90 min to 20 h. (D) CLSM of control cells postincubated for 20 h with BL21(pET23) OMVs. The indicated single fluorescence channels in B to D are shown in the right panels (enlargements are displayed in S29 and S30 Figs) and the merged images in the left panels (green, OMV or CdtV-B, as indicated; red, OMV or compartment-specific marker proteins, as indicated; blue, nuclei; yellow, colocalized green and red signals). The percentages of colocalizations of the respective signals (white numbers) were calculated with the BioImageXD6 tool; blue numbers in panels ER indicate colocalization of CdtV-B with nucleus (means of colocalizations from three different samples are shown). Scale bars in A to D are 10 μm. (E) Immunoblot detection of CdtV-B in isolated subcellular fractions of HBMEC which were postincubated with the indicated OMVs for 90 min to 20 h. BL21(cdtV-ABC) OMVs were a positive control and BL21(cdtV-ACΔB) OMVs, BL21(pET23) OMVs and untreated cells (no OMV) negative controls; lanes OMV ctrl contain 5791/99 OMVs without cells. (F) Densitometric quantification of CdtV-B signals shown in E. For abbreviations see legend to Fig 4.

Fig 12. OMV-delivered CdtV triggers DNA damage signaling and G2 arrest in epithelial and microvascular endothelial cells.

(A) Activation of DNA damage G2 checkpoint signaling in Caco-2 cells, HBMEC and HRGEC incubated with O157, TA153 (CdtV-positive control), or TA154 (vector control) OMVs for 20 h or left untreated (no OMV) as demonstrated by immunoblotting of cell lysates with antibodies against phosphorylated forms of the indicated proteins. Actin served as a loading control. (B, C, D) G2 arrest induced in the indicated cell cultures by O157, TA153, or TA154 OMVs during 24 h to 96 h demonstrated by flow cytometric detection of cells with 4n DNA content. Untreated cells and cells treated with OMV buffer (20 mM TRIS-HCl, pH 8.0) were negative controls. Data are means ± standard deviations from three independent experiments. *p < 0.05, **p < 0.01 or ***p < 0.001 for G2 arrest caused by the indicated OMVs compared to OMV buffer; ^#p < 0.01 for G2 arrest caused by Stx2a-containing versus Stx2a-lacking OMVs; ^×p < 0.01 for G2 arrest caused by 5791/99 and 493/89 OMVs after 24 h compared to 48 h (HRGEC), or after 48 h compared to 72 h (Caco-2, HBMEC). Calculations were performed using one-way ANOVA. (E) Cell distension after 72 h of incubation with CdtV-containing O157 or TA153 OMVs; cells treated with CdtV-negative TA154 OMVs and untreated cells (no OMV) were negative controls. Cells were stained with Giemsa (Caco-2, HBMEC) or photographed native (HRGEC). Scale bars are 20 μm. Note the presence of apoptotic cells (condensed nuclei, reduced cytoplasm) in cultures treated with Stx2a-containing OMVs (5791/99, 493/89).

Fig 13. CdtV-B is responsible for the biological effects caused by OMV-delivered CdtV holotoxin.

HBMEC were incubated with OMVs from the indicated recombinant strains containing the CdtV holotoxin (BL21(cdtV-ABC) OMVs), CdtV-B subunit (BL21(cdtV-B) OMVs), or CdtV-AC subunits (BL21(cdtV-ACΔB) OMVs). OMVs from BL21(pET23) vector control and untreated cells (no OMV) were negative controls. (A) DNA damage response was determined by immunoblotting of cell lysates with antibodies against phosphorylated forms of H2AX (γ-H2AX) and cdc2 (p-cdc2) after 20 h of exposure. Actin served as a loading control. (B) G2 arrest was measured by flow cytometric detection of cells with 4n DNA content after 48 h of exposure. Positions of the G1 (2n DNA) and G2 (4n DNA) peaks are indicated in the first histogram, and the proportions (%) of cells in the G1 and G2 cell cycle phase, respectively, are shown in all histograms. (C) Cell distension was visualized by light microscopy of Giemsa-stained cells after 72 h of exposure. Scale bars are 20 μm.

Fig 14. CdtV-mediated G2 arrest is followed by apoptosis induced by Stx2a and CdtV.

(A, B, C) Time course of apoptosis induced in (A) Caco-2 cells, (B) HBMEC, and (C) HRGEC by CdtV-containing O157 OMVs possessing (5791/99, 493/89) or lacking (493/89Δstx_2a) Stx2a, by control CdtV-containing or CdtV-lacking OMVs TA153 or TA154, respectively, and by purified Stx2a (460 ng/ml) as determined by flow cytometric quantification of cells with hypodiploid nuclei. Staurosporine (1 μM) was a positive and OMV buffer and untreated cells negative controls. *p < 0.05, **p < 0.01 or ***p < 0.001 for apoptosis caused by the indicated OMVs, purified Stx2a or staurosporine compared to OMV buffer; ^#p < 0.01 or ^##p < 0.001 for apoptosis caused by Stx2a-lacking versus Stx2a-containing OMVs; ^×p < 0.01 or ^××p < 0.001 for apoptosis caused by OMVs 5791/99, 493/89, purified Stx2a or staurosporine after 48 h compared to 24 h (HRGEC), or after 72 h compared to 48 h (Caco-2, HBMEC). (D) Apoptosis and necrosis caused by O157 OMVs and controls in the indicated cell types after 96 h as determined by the Cell Death Detection ELISA. Enrichment factors were calculated by dividing OD₄₀₅ absorbance values of sample-treated cells with those of untreated cells; *p < 0.05, **p < 0.01 or ***p < 0.001 for apoptosis caused by OMVs or purified Stx2a compared to OMV buffer; ^#p < 0.01 for apoptosis caused by Stx2a-lacking compared to Stx2a-containing OMVs. (E) Activities of caspase-9 and caspase-8 in the indicated cell lysates after 48 h of incubation with O157 OMVs or with positive (purified Stx2a, OMV T153) or negative (TA154 OMVs, OMV buffer) controls determined with colorimetric substrates (LEHD-pNA and IETD-pNA, respectively). The caspase activities in O157 OMV-treated or control-treated cells were expressed as a fold-increase of those in untreated cells. Inhibitor of caspase-9 (z-LEHD-fmk) was added to cells 30 min before samples; *p < 0.05 compared to untreated cells. Data in all panels are means ± standard deviations from three independent experiments; calculations were performed using one-way ANOVA.

Fig 15. Summary of intracellular trafficking of O157 OMVs and OMV-delivered toxins.

After uptake via dynamin-dependent endocytosis, O157 OMVs carrying the toxin cocktail enter the endosomal compartments of target cells (1). Stx2a holotoxin and CdtV-B subunit separate from OMVs in early endosomes (2) and are retrogradely transported to the Golgi complex (3) and the endoplasmic reticulum (4). From the endoplasmic reticulum, CdtV-B is translocated to the nucleus to target DNA (5), and Stx2a A1 catalytic fragment to the cytosol to reach ribosomes (6). CdtV-A and CdtV-C subunits and EHEC-Hly are sorted with OMVs to late endosomes/lysosomes (7). Here EHEC-Hly separates from OMVs, escapes from the lysosomes (8), and is transported to the mitochondria (9). CdtV-A and CdtV-C remain OMV-associated and are degraded with OMVs in lysosomes (10). Moreover, residual subsets of CdtV-B and Stx2a, which did not separate from OMVs in early endosomes, are sorted with OMVs to lysosomes for degradation (depicted by semi-transparent symbols).