Evidence for Pore Formation in Host Cell Membranes by ESX-1-Secreted ESAT-6 and Its Role in Mycobacterium marinum Escape from the Vacuole^▿

Bacteria and media.

M. marinum strain M was cultured and maintained as described previously (16). The M. marinum ESX-1 mutants were produced as described previously (17). The mh3877::tn mutant was recently isolated from an M. marinum transposon mutant library (16). The mh3881c::tn mutant and its complementation were described previously (50).

Generation of M. marinum Δesat-6 mutant.

Δesat-6 mutant M. marinum was generated by allelic exchange. The left flanking fragment was amplified by PCR with primers DelESAT-F1 (5′CCGCTCTAGACCTGGTTGCAGACCGCCTCGAC3′) and DelESAT-R1 (5′GCCCGAATTCAGAAGCCCATTTGCGAGGACAGCGC3′). The right flanking fragment was amplified by PCR with primers DelESAT-F2 (5′CGGGAATTCGCGTAGAATACCGAAGCACGAGATCGGG3′) and DelESAT-R2 (5′CCGCAAGCTTCTAGATTCATGCCGGTTTGGCGTGGC3′). The left flanking fragment was digested with XbaI and EcoRI, and the right flanking fragment was digested with EcoRI and HindIII. The left and right flanking fragments and a kanamycin resistance cassette (kan^r) (cut with EcoRI from pUC4K) were ligated into pBluescript. The entire sequence containing the flanking sequences and kan^r was cut from the pBluescript clone and ligated into pLYG304.zeo (19) to generate the esat-6 knockout plasmid. The plasmid was electroporated into wild-type (WT) M. marinum, and homologous recombinants were selected as described previously (19). Confirmation of the Δesat-6 mutation was carried out by PCR with two primer pairs. One pair confirms recombination within the left flanking sequence, in which a primer that anneals to a sequence upstream from the flanking sequence (5′CGTGGACCGGAGGCGGCAGCGAGAAAG3′) and another that anneals to a sequence in kan^r (5′CACCTTCTTCACGAGGCAGACCTCAGCGCC3′) were used. The other pair confirms recombination within the right flanking sequence, in which a primer that anneals to a sequence downstream from the flanking sequence (5′GGATTCAGCCTCCGGTGGCCCTGGAG3′) and another that anneals to a sequence in kan^r (5′GGCAATGTAACATCAGAGATTTTGAGACACAACGTGGC3′) were used.

Complementation of M. marinum Δesat-6 mutant.

To complement the Δesat-6 mutant with both the esat-6 and cfp-10 genes together, a fragment containing both genes was amplified by PCR from the M. marinum genome with primers 5′CAGAGATGAAGACCGATGCCGCTACCCTCG3′ and 5′GGCCGGATCCTTAGTGATGGTGATGGTGATGAGCAAACATCCCCGTGACGTTGCC C3′. The reverse primer contains a six-His tag fused to the C terminus of ESAT-6. The PCR product was cloned into pLYG206.Zeo (18) to generate the complementation plasmid. This plasmid was electroporated into Δesat-6 to obtain the complementation strain.

Macrophages.

J774.A1 (ATCC TIB67) or Raw264.7 (ATCC TIB-71) murine macrophage-like cells were cultured and maintained as described previously (16). Bone marrow-derived macrophages (BMDMs) were obtained from C57BL/6 mice as previously described (38). Cells were harvested 8 to 10 days after plating and allowed to adhere to fibronectin-coated coverslips (Becton Dickinson) for infection with M. marinum the next day.

DiI labeling of MCV membranes.

BMDMs on glass coverslips were infected with M. marinum at a multiplicity of infection (MOI) of 2 for 2 h, followed by three washes with phosphate-buffered saline (PBS) and 1 h of incubation with 200 μg/ml amikacin to kill the extracellular bacteria. At the end of the antibiotic incubation, the cells were washed two times with PBS and incubated at 32°C in 5% CO₂ for 48 to 72 h. CM-DiI (Molecular Probes) was added to the cells at a 2 μM final concentration and incubated for 1 h. The cells were then washed two times with PBS to remove excess DiI, incubated for 1 h in the cell culture medium, and fixed with 4% paraformaldehyde. The fixed cells were washed three times with PBS, mounted on a glass slide with Prolong Antifade (Molecular Probes), and imaged.

Fluorescent labeling of F-actin.

BMDMs on glass coverslips were infected at an MOI of 2 as described above. At 48 to 72 h postinfection, the cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, stained with Alexa Fluor phalloidin (Molecular Probes), and imaged. To artificially deliver the ESX-1 mutant bacteria into macrophage cytosol, the cells at 48 h postinfection were treated for 5 min with a hypotonic solution (4 parts phenol red-free Dulbecco modified Eagle medium [DMEM] to 1 part H₂O), incubated in cell culture medium at 32°C in 5% CO₂ for an additional 24 h, and then processed for actin staining. The procedures for the hypotonic shock treatment were similar to those described by Okada and Rechsteiner (31).

Electron microscopy.

BMDMs were infected at an MOI of 2 as described above. At 72 h after infection, the cells were processed for electron microscopic analysis as described previously (48).

Detection of pore formation in red blood cell membranes.

Induction of pore formation in red blood cell membranes by M. marinum was detected by hemolysis assay as previously described (15, 17). Briefly, M. marinum grown in 7H9 medium to mid-log phase was washed twice with PBS. A volume of 1.3 ml of M. marinum suspension (containing 2.5 × 10⁹ bacteria) was mixed with 400 μl of sheep red blood cells (sRBC; Quad Five) (containing 1 × 10⁸ cells) in a microcentrifuge tube and centrifuged at 8,000 × g for 2 min. The tubes were incubated at 32°C for designated periods of time. The pellets were then resuspended and centrifuged, and the A₄₀₅ of the supernatants was measured. To examine the role of energy-dependent secretion in membrane pore formation, M. marinum WT bacteria were pretreated for 15 min with 20 μM carbonyl cyanide m-chlorophenylhydrazone (CCCP) to uncouple the proton motive force and then incubated with sRBC in the presence of CCCP for 2 h. To observe the reversibility of CCCP in hemolysis, the pellet containing M. marinum and red blood cells was resuspended in PBS and then repelleted and further incubated for 2 h before measurement of hemolysis.

In the polyethylene glycol (PEG) osmoprotection experiment, PEG1000, PEG3350, PEG6000, and PEG8000 were resuspended in PBS and added to red blood cells with or without M. marinum to a final concentration of 30 mM. Hemolysis was measured after 2 h of incubation at 32°C. To determine if protection from hemolysis by PEG8000 is reversible, the pellet containing M. marinum and red blood cells was resuspended in PBS and then repelleted and further incubated for 2 h before measurement of hemolysis. To estimate membrane pore size, the various PEGs were resuspended in PBS at the following ranges of osmolarities: PEG1000, 1.0, 2.0, 3.0, and 4.0 M; PEG3350, 0.2, 0.4, 0.6, 0.8, and 1.0 M; PEG6000, 0.2, 0.3, 0.4, and 0.5 M; PEG8000, 0.1, 0.2, and 0.3 M. The osmolarity required for each PEG to provide 50% protection from hemolysis was determined and plotted against the Einstein-Stokes molecular diffusion radium, R_ES (40).

Expression and purification of recombinant ESX-1 proteins.

The rESAT-6-His₍₆₎, rCFP-10-His₍₆₎, and rMh3881c-His₍₆₎ proteins were affinity purified and had endotoxins removed. The rESAT-6-His₍₆₎ and rCFP-10-His₍₆₎ proteins were obtained from Colorado State University under the NIH TB Research Materials Contract. The inclusion bodies containing rESAT-6-His₍₆₎ were solubilized with 6 M urea, and the protein was purified with the His-Bind resin (Novagen). Endotoxins were removed by washing the column with 10 mM Tris-HCl, followed by 0.5% ASB-14. The protein was eluted with 10 mM Tris-HCl containing 1 M imidazole. The eluted protein was dialyzed against 10 mM ammonium bicarbonate. The residual concentration of endotoxins was ≤0.24 ng/mg protein. Two lots were obtained, one produced in 2001 and the other in 2007. Both lots were used for the analysis of membrane pore formation, and similar results were observed. We expressed and purified the rMh3881c-His₍₆₎ protein by using procedures similar to those described above.

Detection of pore formation by purified ESX-1 proteins.

To determine the ability of ESX-1-secreted proteins to induce membrane pore formation, the above-described purified rESAT-6-His₍₆₎, rCFP-10-His₍₆₎, or rMh3881c-His₍₆₎ in a 50-μl volume was mixed with 100 μl of sRBC (containing 1 × 10⁹ cells) in a microcentrifuge tube. The tubes were incubated at 32°C for designated periods of time. The cells were then resuspended and centrifuged at 4,000 rpm for 7 min. The supernatants were transferred to corresponding wells in a 96-well plate, and the A₄₀₅ was measured. In the PEG osmoprotection experiment, PEG1000, PEG3350, PEG6000, and PEG8000 were resuspended in PBS and added to the red blood cell-protein mixture at a final concentration of 30 mM. Hemolysis was measured after 2 h of incubation at 32°C.

Detection of pore formation in macrophage cell membranes.

M. marinum strains grown to mid-log phase were washed twice with DMEM, added to macrophage monolayers at an MOI of 50, centrifuged for 10 min at 1,500 × g to allow immediate bacterium-cell contacts, and incubated at 32°C in 5% CO₂ for designated periods of time. The release of lactate dehydrogenase (LDH) by the infected and noninfected cells was measured with a CytoTox-One Homogeneous Membrane Integrity Assay kit (Promega) at an excitation wavelength of 560 nm and an emission wavelength of 590 nm. Pore formation in macrophage cell membranes was determined by an osmoprotection assay with various PEGs at a final concentration of 30 mM. To examine the induction of pore formation in macrophage cell membranes by rESAT-6-His₍₆₎, rCFP-10-His₍₆₎, or rMh3881c-His₍₆₎, the proteins were dissolved in DMEM and added individually to Raw264 cells in a 96-well plate at designated concentrations. After 2 h of incubation at 32°C in 5% CO₂, the release of LDH was determined.

To detect membrane pore formation by a microscopic method, infected or noninfected macrophages were incubated with ethidium homodimer-1 (Molecular Probes) and penetration of the cell by ethidium homodimer-1 across the membranes was detected by fluorescence microscopy. In brief, macrophages were incubated for 40 min in phenol red-free culture medium containing ethidium homodimer-1 (4 μM) and calcein AM (2 μM), followed by imaging. Ethidium homodimer-1 only penetrates permeabilized cell membranes and stains the nuclei red. Calcein AM permeates every cell membrane and is only metabolized by live cells to produce green fluorescence.

Preparation of M. marinum short-term culture filtrate and cell lysate.

Preparation of M. marinum short-term culture filtrate and cell lysate was carried out as previously described (50). In brief, WT and mutant M. marinum cells were first grown in 7H9 medium to mid-log phase. The bacteria were then washed and diluted 10 times in Sauton's medium and cultured for 2 days to reach mid-log phase. The bacteria were washed again and diluted 10 times in Sauton medium and cultured for another 2 days. The culture supernatant was collected, filtered through a 0.2-μm filter, and concentrated 100 times with a Centricon centrifugal filter with a molecular weight cutoff of 3,000. The cell lysate was obtained by bead beating the bacterial pellet.

Western blotting.

Proteins were separated by 4 to 20% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane. After blocking with 2% bovine serum albumin, the membrane was incubated with a primary antibody diluted in 2% bovine serum albumin overnight, followed by incubation for 1 h with a horseradish peroxidase-conjugated secondary antibody (Bio-Rad). The membrane was developed by enhanced chemiluminescence (Pierce) and exposed to films. The following primary antibodies were used at the dilutions indicated: anti-ESAT-6 (monoclonal antibody HYB 76-8; Abcam), 1:3,000; anti-CFP-10 (Colorado State University, NIH contract NO1-AI-75320), 1:2,000; anti-Mh3881c (Michael Lodes, Corixa Corporation, Seattle, WA), 1:5,000.

ESX-1 is essential for M. marinum to escape from the vacuole.

M. marinum is able to escape from the MCV into the host cell cytosol, polymerize actin, and spread from cell to cell (17, 42, 43). Because mutations in various ESX-1 genes abolish M. marinum spreading (17), we hypothesized that the ESX-1 secretion system may play a role in either the escape of M. marinum from the MCV or initiation of actin polymerization. We first determined if ESX-1 is involved in M. marinum escape from the vacuole. We examined the association of WT or ESX-1 mutant bacteria with the vacuole membranes in live macrophages by using a fluorescent membrane dye, DiI. DiI is frequently used to label the live cell membranes and has worked well in previous studies labeling MCV membranes (42). As shown in Fig. 1A to C and Table 1, at 72 h postinfection only a small fraction (18%) of the WT bacteria colocalized with DiI, suggesting that the majority of the bacteria entered the cytosol. In sharp contrast, for all of the nine ESX-1 mutants examined, the majority of the bacteria (≥80%) colocalized with DiI (Fig. 1D to F and Table 1), indicating that they reside predominantly within MCV membranes. To confirm the above observations, we used transmission electron microscopy to examine in greater detail the association of M. marinum with vacuole membranes. This study shows that only a small fraction of the WT bacteria are bound with MCV membranes (36%), while most are free of the membrane (64%) (Fig. 2A and Table 1). On the other hand, for the four ESX-1 mutants examined, more than 98% of the bacteria are surrounded by MCV membranes (Fig. 2B and Table 1). The mutants used in the above assays contain mutations in either the ESX-1 secretion apparatus (such as mh3877::tn and mh3871::tn) or the secreted substrates (such as Δesat-6, Δcfp-10+esat-6, and mh3881c::tn). Because the two groups of mutants showed similar phenotypes (Table 1), these results indicate that the ESX-1 secretion system plays a critical role in the escape of M. marinum from the MCV.

M. marinum escape from the vacuole is required for the polymerization of actin.

Next, we determined if ESX-1 is required for M. marinum to initiate actin polymerization. At 72 h postinfection, 34% of the WT bacteria showed actin polymerization at one pole of the bacterium, forming the “actin comet tail” (Fig. 3A to C and Table 1), similar to previous observations (42, 43). In contrast, none of the nine ESX-1 mutants were able to polymerize actin (Fig. 3D to F and Table 1).

One possible explanation for the above results is that an ESX-1-secreted protein is directly involved in the recruitment and polymerization of actin. Alternatively, an ESX-1-secreted protein could be involved in compromising the integrity of MCV membranes to facilitate M. marinum escape into the host cell cytosol. To distinguish between these two possibilities, we treated the infected macrophages with a hypotonic solution to artificially deliver the ESX-1 mutant bacteria into the cytosol and then reexamined actin polymerization. Okada and Rechsteiner (31) showed elegantly that hypotonic shock treatment causes lysis of pinocytic/endocytic vesicle membranes without disrupting the plasma membrane. This finding supports that the hypotonic shock treatment could facilitate entry of the ESX-1 mutant bacteria into the host cell cytosol. Indeed, as shown in Fig. 4, after the treatment, mh3868::tn gained the ability to polymerize actin. Similar results were observed with mh3881c::tn (data not shown). These results indicate that the defect in the ability of the ESX-1 mutants to polymerize actin is due to their inability to escape from the vacuole rather than a deficiency in initiation of actin polymerization. They also provide direct evidence that M. marinum initiates actin polymerization only after it enters the host cell cytosol. Moreover, they suggest that ESX-1 is involved in secreting a pore-forming protein that may compromise the integrity of the MCV membranes to facilitate the escape of M. marinum.

Evidence for membrane pore formation by M. marinum ESX-1.

Listeria monocytogenes represents a group of bacteria that are able to disrupt the vacuole membranes to enter the host cell cytosol by the secretion of pore-forming proteins (41). We hypothesized that the ESX-1 secretion system may play a similar role to facilitate M. marinum escape into the host cell cytosol. A direct test of this hypothesis would require the analysis of pore formation in the MCV membranes of infected cells, which is technically challenging. In an attempt to address this hypothesis, we took an alternative approach. We incubated M. marinum with host cells at a relatively high MOI and examined pore formation in the cell plasma membranes. Membrane pore formation was determined by an osmoprotection assay (40) which has been used in a number of studies to demonstrate membrane pore formation and estimate pore size. This assay has been used to detect membrane pores and estimate the size of pores induced by pathogens such as Gardnerella vaginalis (30), Legionella pneumophila (26), Shigella flexneri (3), and Pseudomonas aeruginosa (8). We first determined if the hemolysis of red blood cells induced by M. marinum could be blocked by PEGs of various molecular weights. The lysis of red blood cells (hemolysis) by pore-forming proteins occurs through osmotic shock, which can be prevented by osmoprotectants that have larger sizes than the membrane pores (3, 8, 26, 29). As shown in Fig. 5A, WT M. marinum caused contact-dependent hemolysis of red blood cells, which was blocked completely by PEG8000 (6.4 nm in diameter) and 60% by PEG6000 (5.0 nm) but not blocked by PEG3350 (3.8 nm). Importantly, the blocking effect of PEG8000 is reversible, as indicated by the recovery of hemolysis after the removal of PEG8000 (Fig. 5A). We then examined pore formation in macrophage cell membranes with a similar assay in which different-size PEGs were used to block the release of LDH from infected macrophages. We found that WT M. marinum caused the release of LDH from macrophages, which was blocked completely by PEG8000 but not by PEG3350. Together, the results of both assays suggest that M. marinum causes contact-dependent pore formation in host cell membranes.

To estimate the size of the membrane pores produced by M. marinum, PEGs of different sizes were used at various concentrations to determine the osmolarity required for each PEG to provide 50% protection from hemolysis (26, 40). Figure 5B shows the osmolarity of each PEG that provides 50% protection as a function of its Einstein-Stokes molecular diffusion radium, R_ES (40). The response curve is hyperbolic and approaches a membrane stabilization limit asymptotically. By using a similar method developed by Scherrer and Gerhardt (40), we estimated pore size by extrapolating to the zero abscissa the linear regression between PEG1000 and PEG3350, and the intercept at R_ES of 2.25 is believed to represent the radius of the membrane pore.

We then determined if ESX-1 plays a role in membrane pore formation. As shown in Fig. 6A and B, mutations in either the ESX-1 secretion apparatus (mh3877::tn and mh3871::tn) or the secreted substrate (Δesat-6) completely abolished the ability of M. marinum to induce membrane pore formation, demonstrating that ESX-1 secretion plays an essential role in this process. To determine if continuous ESX-1 secretion or predeposition of ESX-1-secreted proteins on the bacterial surface is necessary to cause pore formation, we treated WT M. marinum with CCCP, a membrane deenergizer that blocks energy-dependent pathways including ESX-1. Figure 6C shows that the CCCP treatment abolished hemolysis completely and in a reversible manner, indicating that continuous energy- dependent ESX-1 secretion is required for M. marinum to induce membrane pore formation.

Evidence for membrane pore formation by ESX-1-secreted ESAT-6.

Three known ESX-1-secreted proteins, ESAT-6, CFP-10, and Mh3881c, are codependent for secretion (50). To determine the relative role of each individual protein in pore formation, we compared the hemolysis levels induced by different M. marinum strains producing various amounts of these proteins. We have shown previously that Mh3881c is cleaved during secretion to produce an N-terminal 50-kDa and a C-terminal 11-kDa fragment (apparent molecular masses) (50; Fig. 7, lane 1). The 50-kDa fragment is relatively stable and present at abundant levels in the culture supernatant, while the majority of the 11-kDa fragment is degraded proteolytically (50; data not shown). As shown in Fig. 7, mh3881c::tn fails to secrete these three proteins and is defective in pore formation (lane 2). Both the secretion and pore formation defects of this mutant are almost fully restored by expression of the WT mh3881c gene (lane 3). On the other hand, when only the N-terminal 50-kDa fragment of Mh3881c was expressed in this mutant, it was secreted by the mutant at levels comparable to those produced by WT bacteria, but ESAT-6 secretion was not detected and CFP-10 secretion was minimal (lane 4). Because this strain shows a complete defect in pore formation, the data suggest that Mh3881c, or at least the 50-kDa fragment, does not contribute directly to pore formation. To determine if the secretion of ESAT-6 or CFP-10 plays a role, we examined pore formation by Δesat-6 and the complementation strain. Δesat-6 fails to secrete these three proteins and is defective in pore formation (lane 5). Expressing esat-6-His₍₆₎ in Δesat-6 substantially restored the secretion of ESAT-6 and Mh3881c, although CFP-10 secretion was only recovered by a marginal level (lane 6). The reason that we used esat-6-His₍₆₎ instead of esat-6 was to express this protein in a form that is exactly the same as the recombinant protein used in other assays (see below). Because this strain shows a substantial increase in hemolysis, which correlates with the much increased ESAT-6 secretion, the results suggest that ESAT-6 secretion plays a more important role in pore formation.

To determine if ESAT-6 has a direct role in membrane pore formation, we examined the pore-forming activity of recombinant ESAT-6 [rESAT-6-His₍₆₎] purified from Escherichia coli and compared it to that of rCFP-10-His₍₆₎ or rMh3881c-His₍₆₎. These recombinant proteins were affinity purified and free of detergents and had endotoxins removed (see Materials and Methods for details). As shown in Fig. 8A, rESAT-6-His₍₆₎ from M. tuberculosis produced dose-dependent hemolysis after a 2-h incubation, i.e., partial hemolysis at 15 μg/ml and complete hemolysis at 30 μg/ml, almost equivalent to lysis with H₂O. In contrast, neither rCFP-10-His₍₆₎ from M. tuberculosis nor rMh3881c-His₍₆₎ from M. marinum caused hemolysis, even at a higher concentration of 60 μg/ml (Fig. 8B) or 120 μg/ml (data not shown). The combination of rESAT-6-His₍₆₎ with rCFP-10-His₍₆₎ or with rMh3881c-His₍₆₎ produced hemolysis at levels similar to those produced by rESAT-6-His₍₆₎ alone (data not shown). Hemolysis induced by rESAT-6-His₍₆₎ was not due to the residual endotoxins present in the recombinant protein preparations (≤0.24 ng/mg protein), since lipopolysaccharide at a concentration of 0.007 ng/ml [equivalent to the level of endotoxins present in rESAT-6-His₍₆₎ at a concentration of 30 μg/ml] failed to induce a detectable level of hemolysis (data not shown). Consistent with the hemolysis results, rESAT-6-His₍₆₎ at 60 μg/ml, but not rCFP-10-His₍₆₎ or rMh3881c-His₍₆₎ (even at 120 μg/ml), caused release of LDH from macrophages (Fig. 8C). Permeation of macrophage cell membranes was similarly observed by microscopic detection of penetration of ethidium homodimer-1 across the plasma membrane into the cytosol to stain the nuclei red (Fig. 8D). These results together suggest that ESAT-6 may play a direct role in membrane pore formation.

To provide a more direct demonstration that ESAT-6 by itself can induce pore formation in cell membranes, we performed an osmoprotection assay similar to that described above. As shown in Fig. 9A, the hemolysis induced by rESAT-6-His₍₆₎ was blocked completely by 30 mM PEG8000 and PEG6000, ∼40% by PEG3350, and to a small extent by PEG1000. The facts that hemolysis is induced by rESAT-6-His₍₆₎ and that it can be blocked by PEGs of increasing sizes indicate that ESAT-6 indeed plays a direct role in membrane pore formation. It was noticed that the membrane pores induced by rESAT-6-His₍₆₎ are somewhat smaller than those produced by the bacteria (see Discussion). We further characterized the ESAT-6-induced membrane pores by determining the kinetics of hemolysis produced by rESAT-6-His₍₆₎. Figure 9B shows that rESAT-6-His₍₆₎ at 30 μg/ml caused 75% hemolysis after a 5-min incubation, which increased to 93% after 10 min. This is almost equivalent to the hemolysis produced by H₂O, which caused 95% and complete hemolysis after 5 and 20 min of incubation, respectively.