A New ESX-1 Substrate in Mycobacterium marinum That Is Required for Hemolysis but Not Host Cell Lysis

Both Mycobacterium tuberculosis, the cause of human tuberculosis (TB), and Mycobacterium marinum, a pathogen of ectotherms, use the ESX-1 secretion system to cause disease. There are many established similarities between the ESX-1 systems in M. tuberculosis and in M. marinum. Yet the two bacteria infect different hosts, hinting at species-specific functions of the ESX-1 system. Our findings demonstrate that MMAR_2894 is a PE protein secreted by the ESX-1 system of M. marinum. We show that MMAR_2894 is required for the optimal secretion of mycobacterial proteins required for disease. Because the MMAR_2894 gene is not conserved in M. tuberculosis, our findings demonstrate that MMAR_2894 may contribute to a species-specific function of the ESX-1 system in M. marinum, providing new insight into how the M. marinum and M. tuberculosis systems differ.

ABSTRACT

The ESX-1 (ESAT-6 system 1) secretion system plays a conserved role in the virulence of diverse mycobacterial pathogens, including the human pathogen Mycobacterium tuberculosis and M. marinum, an environmental mycobacterial species. The ESX-1 system promotes the secretion of protein virulence factors to the extracytoplasmic environment. The secretion of these proteins triggers the host response by lysing the phagosome during macrophage infection. Using proteomic analyses of the M. marinum secretome in the presence and absence of a functional ESX-1 system, we and others have hypothesized that MMAR_2894, a PE family protein, is a potential ESX-1 substrate in M. marinum. We used genetic and quantitative proteomic approaches to determine if MMAR_2894 is secreted by the ESX-1 system, and we defined the requirement of MMAR_2894 for ESX-1-mediated secretion and virulence. We show that MMAR_2894 is secreted by the ESX-1 system in M. marinum and is itself required for the optimal secretion of the known ESX-1 substrates in M. marinum. Moreover, we found that MMAR_2894 was differentially required for hemolysis and cytolysis of macrophages, two lytic activities ascribed to the M. marinum ESX-1 system.

IMPORTANCE Both Mycobacterium tuberculosis, the cause of human tuberculosis (TB), and Mycobacterium marinum, a pathogen of ectotherms, use the ESX-1 secretion system to cause disease. There are many established similarities between the ESX-1 systems in M. tuberculosis and in M. marinum. Yet the two bacteria infect different hosts, hinting at species-specific functions of the ESX-1 system. Our findings demonstrate that MMAR_2894 is a PE protein secreted by the ESX-1 system of M. marinum. We show that MMAR_2894 is required for the optimal secretion of mycobacterial proteins required for disease. Because the MMAR_2894 gene is not conserved in M. tuberculosis, our findings demonstrate that MMAR_2894 may contribute to a species-specific function of the ESX-1 system in M. marinum, providing new insight into how the M. marinum and M. tuberculosis systems differ.

INTRODUCTION

Pathogenic mycobacteria are a severe burden to human health. More than 10 million people develop tuberculosis (TB) or nontuberculous diseases due to mycobacterial infection annually (1, 2). Approximately 1.7 million people die each year from TB, making TB a leading cause of death globally (1). The high burden of mycobacterial disease is due to a combination of poor vaccine efficacy, diagnostics based on immunoreactivity rather than detecting infection with live bacteria, and resistance to currently available therapeutics (3). One way to develop new approaches to preventing and treating mycobacterial disease is to focus on understanding the basic mechanisms used by mycobacteria to manipulate the host during active infection (3, 4). Pathogenic mycobacteria use protein secretion as a key mechanism to modulate the host environment early during infection. In addition to Sec and Tat systems, pathogenic mycobacteria use specialized ESX (ESAT-6 system) systems to promote the secretion of protein substrates to extracytoplasmic locations (5–7).

In Mycobacterium tuberculosis, the cause of the human disease tuberculosis, the ESX-1, ESX-3, and ESX-5 systems promote mycobacterial virulence (5, 8–19). The ESX-1 system regulates the outcome of mycobacterial infection of the macrophage. Following uptake by the macrophage, pathogenic mycobacteria reside within the phagosome, which they remodel to promote survival (reviewed in reference 20). The ESX-1 system promotes phagosomal lysis (21–24). Lysis of the phagosome allows the bacterium and its secreted products to access the macrophage cytosol, which triggers the host immune response to infection (19, 20, 24–27). The host response includes, but is not limited to, the induction of interferon beta (IFN-β) and, eventually, lysis of the macrophage (13, 19–21, 25, 26, 28, 29). Many of the secreted mycobacterial proteins directly combat the host response to infection (reviewed in references 6, 9, and 20). In the absence of a functional ESX-1 system, M. tuberculosis is attenuated because it is retained in the phagosome (21–25, 30). The precise mechanism of phagosomal lysis is elusive, yet it is presumed that proteins secreted by ESX-1, either on the mycobacterial cell surface or in the phagosome, promote membrane damage (21, 22, 24, 26, 31–37).

Mycobacterium marinum is an environmental mycobacterial species that infects poikilothermic fish and occasionally humans (38, 39). M. marinum is a well-accepted model for several mechanisms of pathogenesis used by M. tuberculosis, including the ESX-1 system (40–43). The ESX-1 system is functionally conserved between M. tuberculosis and M. marinum (31, 44). Mutations in ESX-1 genes in M. marinum can be functionally complemented by the expression of ESX-1 genes from M. tuberculosis (31). Indeed, the esx-1 locus, which is highly conserved between M. marinum and M. tuberculosis, includes genes encoding the core components of the secretory apparatus and genes encoding several substrates. Despite the conservation of ESX-1 function between M. marinum and M. tuberculosis, there are significant differences between these pathogens, including growth temperature and host range (38, 41–43). Moreover, it is thought that M. marinum can fine-tune secretion by the ESX-1 system in different host environments (45). M. marinum and M. tuberculosis may use the ESX-1 system to secrete proteins that promote species-specific functions. Finally, the M. marinum ESX-1 system can promote the lysis of red blood cells (RBCs) in vitro, although the specific mechanism of hemolysis has not been defined (31, 46).

PE and PPE proteins are highly immunogenic proteins that are characterized by conserved N-terminal proline-glutamate (PE) or proline-proline-glutamate (PPE) motifs (47–50). Genes encoding proteins in the PE and PPE families make up 7% and 9.1% of the M. tuberculosis and M. marinum genomes, respectively (47, 51). PE and PPE proteins are found on the mycobacterial cell surface and are secreted from the bacterial cell, both in vitro and in the host phagocyte (37). Several PE and PPE proteins are secreted by ESX systems and are involved in bacterial persistence and interaction with the host immune response (15, 50, 52–57). Data from a recent study (58) and our unpublished proteomic data suggested that the PE protein MMAR_2894 is an M. marinum-specific ESX-1 substrate. The MMAR_2894 gene does not have a clear ortholog in the M. tuberculosis genome. In this study, we test the hypothesis that MMAR_2894 is an M. marinum-specific substrate of the ESX-1 system using both genetic and proteomic approaches. We found that MMAR_2894 is secreted by the ESX-1 system and is required for the optimal secretion of the other known ESX-1 substrates. Interestingly, we found that MMAR_2894 was dispensable for ESX-1-dependent lysis of macrophages but was required for ESX-1-mediated hemolytic activity. Our findings provide a new substrate of the ESX-1 system in M. marinum and what is, to our knowledge, the first example of an ESX-1 substrate that is differentially required for the hemolytic and cytolytic activities of the ESX-1 system.

RESULTS

MMAR_2894 is secreted in an ESX-1-dependent manner.

MMAR_2894 is annotated as a PE family protein (59). Genes encoding members of the PE family of proteins are widely distributed across mycobacterial genomes. In ESX-associated regions, PE genes are frequently found adjacent to PPE genes. Although the MMAR_2895 gene is predicted to produce a protein from the PPE family, there is a 1.47-kb intergenic region between the genes that is not annotated (Fig. 1A) (59). The MMAR_2894 gene is not associated with an ESX locus in the M. marinum genome, and there is no obvious ortholog in M. tuberculosis.

FIG 1.

MMAR_2894 is secreted from M. marinum in an ESX-1-dependent manner. (A) The MMAR_2894 genomic locus in M. marinum. Coordinates in base pairs are shown below the genes. Annotations are from the MycoBrowser database (59). (B) Western blot analysis of cell-associated (P) and secreted (S) protein fractions. A single blot representative of data from three biological replicates is shown. RpoB is a control for bacterial lysis. Mpt-32 is secreted independently of ESX-1 and is a control for loading. Eleven micrograms of protein was loaded in each lane.

Our unpublished data, as well as those from Phan et al., suggest that MMAR_2984 is an ESX-1 substrate (58). To test the hypothesis that MMAR_2894 is secreted by the ESX-1 system, we introduced a C-terminal Strep-tag into the MMAR_2894 gene (MMAR_2894-ST). We constitutively expressed the MMAR_2894-ST gene from the mycobacterial optimal promoter (Mop) on an integrating plasmid in both the wild-type (WT) M strain and the ΔeccCb₁ strain, in which the ESX-1 system is genetically inactivated. We collected secreted and cell-associated protein fractions and detected the MMAR_2894-ST protein in each fraction using Western blot analysis. As shown in Fig. 1B, the MMAR_2894-ST protein was present in the secreted protein fraction only in the presence of a functional ESX-1 secretion system. The MMAR_2894-ST protein was detected in the cell-associated fractions from both the WT and ΔeccCb₁ strains as a pair of bands running at ∼25 kDa and ∼15 kDa (Fig. 1B, lanes 5 and 7). We detected MMAR_2894 as several bands in the secreted protein fraction from the WT but not from the ΔeccCb₁ strain (compare lanes 6 and 8). The expected size of the MMAR_2894-ST protein is approximately 23 kDa. Therefore, the band near the 25-kDa molecular weight marker likely corresponds to the full-length protein. Importantly, the presence of MMAR_2894-ST followed the same pattern as that for the EsxA protein (Fig. 1B, bottom, αESAT-6, lanes 5 to 8), which is a well-characterized substrate of the ESX-1 system (17, 18, 60). Notably, the bands corresponding to the MMAR_2894-ST protein were not detected in protein fractions from strains lacking the MMAR_2894-ST expression plasmid (lanes 1 to 4). From these data, we conclude that the MMAR_2894 protein is secreted by the ESX-1 system in M. marinum.

MMAR_2894 is required for ESX-1 secretion.

It has been well established that the EsxA and EsxB substrates are also required for ESX-1 function (18, 60). In strains lacking the esxA or esxB gene, additional protein substrates of the ESX-1 system are no longer secreted into the culture supernatant in vitro (46, 61–64). To test if MMAR_2894 is similarly required for ESX-1 function, we generated an MMAR_2894 deletion strain using allelic exchange (Fig. 2A). We confirmed the deletion of the MMAR_2894 gene using PCR (Fig. 2B) and targeted DNA sequencing. We demonstrated that the ΔMMAR_2894 strain grew similarly to the WT, the ΔeccCb₁ strain, and the ΔMMAR_2894 strain bearing the MMAR_2894-ST expression plasmid (complemented strain) under standard conditions in vitro (see Fig. S1 in the supplemental material). We isolated the secreted and cell-associated protein fractions from the WT, ΔRD1 (region of difference 1) (which lacks several genes required for ESX-1 secretion, including the esxA gene), ΔMMAR_2894, and complemented strains. We measured the expression and secretion of the EsxA substrate and of the MMAR_2894-ST protein. As shown in Fig. 3A, EsxA was produced and secreted from the WT strain (lanes 1 and 2) and absent from the ΔRD1 strain (lanes 3 and 4), as previously reported (18, 44, 46, 60). The EsxA protein was present in the cell-associated fraction (lane 5), indicating that EsxA was produced by the ΔMMAR_2894 strain. The EsxA protein was absent from the secreted fraction from the ΔMMAR_2894 strain, demonstrating that MMAR_2894 is required for the secretion of the EsxA substrate from M. marinum. Expression of the MMAR_2894-ST protein in the ΔMMAR_2894 strain (lanes 7 and 8) restored the secretion of EsxA (lane 8).

FIG 2.

Generation and confirmation of the ΔMMAR_2894 M. marinum strain. (A) The MMAR_2894 genomic locus in M. marinum. The unmarked deletion was generated using allelic exchange. The deletion left 10 codons in the MMAR_2894 gene, including the stop codon. The orb150/orb151 primer pair anneals to the upstream and downstream genomic regions. The predicted size for PCR products from each strain is shown. NT, N terminus. (B) The deletion was confirmed using PCR with the orb150/orb151 primer pair. The resulting PCR product was analyzed using targeted DNA sequencing.

FIG 3.

MMAR_2894 promotes ESX-1 secretion in vitro. (A) Western blot analysis of cell-associated and secreted protein fractions from M. marinum strains. A blot representative of data from three biological replicates is shown. RpoB is a control for bacterial lysis. Mpt-32 is secreted independently of ESX-1 and is a control for loading. Thirteen micrograms of protein was loaded in each lane. (B) Volcano plot from label-free quantitative proteomics with fold changes (log₂) in levels of proteins differentially expressed in the secreted fractions of WT M. marinum and the ΔMMAR_2894 strain (see Table S2 in the supplemental material). Boxed areas show ≥2-fold changes and a P value of <0.05. EspJ and MMAR_2894 (blue) are noted as examples of proteins identified in high abundance in the WT but not detected in the secreted fraction from the ΔMMAR_2894 strain. (C) Log₂ fold changes of the levels of selected ESX-1 proteins in the supernatant fractions from the ΔMMAR_2894 and complemented strains relative to the levels of these proteins in supernatant fractions generated from the WT strain. “ns” indicates that there is no significant difference in the levels of the protein secreted by the WT and complemented strains (Table S3). (D) Functional analysis of proteins whose levels are reduced in the supernatant fraction generated from the ΔMMAR_2894 strain relative to the supernatants generated from the WT and complemented strains (Table S4).

Our findings suggested that deletion of the MMAR_2894 gene negatively impacted the secretion of the EsxA protein in vitro. EsxA is required for the secretion of several additional ESX-1 substrates. We therefore hypothesized that deletion of the MMAR_2894 gene would impact the secretion of additional proteins from M. marinum. We isolated both cell-associated and secreted proteins from the WT, ΔMMAR_2894, and complemented strains and performed label-free quantitative proteomics (for a complete list of proteins identified, see Table S1 in the supplemental material). As shown in Fig. 3B (and Fig. S2), we identified 685 proteins that were present in the secreted fraction from the ΔMMAR_2894 strain at levels different from those in the secretion fraction generated from the WT and complemented strains (Table S2 in the supplemental material includes the proteins that met the criteria for label-free quantification [LFQ], and these are plotted on the volcano plot in Fig. 3B). Proteins present at significantly reduced or increased levels in the secreted fraction from the ΔMMAR_2894 strain relative to the secreted fraction from the WT strain are boxed in green or magenta, respectively, in Fig. 3B (with a −log₁₀ P value cutoff of 1.3 and an average log₂ fold change of 1 [gray]). As expected, the MMAR_2894 protein was not detected in the ΔMMAR_2894 strain. The MMAR_2894 protein was detected in the both the WT and complemented strains. However, there was significantly less MMAR_2894 protein in the complemented strain than in the WT strain (ratio of log₂ fold changes of WT/complemented strain of 3.02; −log₁₀ P value of 5.22 [Table S3]). We detected all of the known ESX-1 substrates (18, 34, 61, 62, 64–66) in the cell-associated fractions from all of the M. marinum strains tested, with the exception of PE35. When the secreted protein fraction generated from the ΔMMAR_2894 strain was compared to those of the WT and complemented strains, we found that the levels of several known ESX-1 substrates were not completely abolished but rather were significantly reduced (Fig. 3B and C and Tables S2 and S3). The exception was the EspJ substrate, which was undetected in the secreted protein fraction from the ΔMMAR_2894 strain, similar to the MMAR_2894 protein. Notably, complementation with the MMAR_2894-ST expression plasmid restored the levels of the ESX-1 proteins in the secreted protein fraction (Fig. 3C and Table S3). From these data, we conclude that the loss of the MMAR_2894 gene results in a reduced-secretion phenotype for the majority of ESX-1 substrates in M. marinum. We identified 76 proteins (log₂ fold change cutoff of 1 and −log₁₀ P value of 1.3) with levels that were significantly reduced in the secreted fraction generated from the ΔMMAR_2894 strain compared to both the WT and complemented strains (Table S4, proteins whose levels were significantly decreased in the ΔMMAR_2894 strain compared to the WT and complemented strains by LFQ). These proteins could be directly or indirectly affected by the loss of ESX-1, some of which could represent additional new ESX-1 substrates. We analyzed these proteins using the functional annotations from the MycoBrowser M. marinum database (59). Three major functional categories were represented, as shown in Fig. 3D. ESX-1 proteins made up 16% of the proteins that were secreted at reduced levels in the absence of MMAR_2894. Thirty-two percent of the proteins with reduced levels were annotated as being associated with the cell wall and cell processes. This category includes proteins associated with the cell wall, such as lipoproteins, previously demonstrated exported and secreted proteins, and hypothetical and conserved membrane-associated proteins. Interestingly, 22% of the proteins were annotated as “PE/PPE family” proteins, similar to MMAR_2894. Several of these proteins were reproducibly undetected in the secreted fraction of the MMAR_2894 strain, including MMAR_2894 and the EspJ ESX-1 substrate, but present in the secreted protein fractions generated from the WT and complemented strains (Table S4). Proteins whose secreted levels increased in an MMAR_2894-dependent manner are listed in Table S5 (proteins whose levels were significantly increased in the ΔMMAR_2894 strain compared to the WT and complemented strains by LFQ), and the functional analysis is presented in Fig. S3. Proteins annotated as “PE/PPE family” were notably absent. Together, these findings indicate that MMAR_2894 is required to promote optimal ESX-1 secretion from M. marinum, and the loss of MMAR_2894 may broadly impact protein secretion.

MMAR_2894 promotes hemolysis but not macrophage cytotoxicity.

The functional role of the ESX-1 system is to damage the phagosomal membrane following uptake by host phagocytes (21–25). Phagosomal damage allows M. marinum, as well as its secreted products, access to the host cell cytosol, where it triggers and combats the host cell response to infection (20). One measure of the lytic activity of the ESX-1 system is its activity against sheep red blood cells (sRBCs). M. marinum lyses sRBCs in an ESX-1-dependent manner (46, 67). We tested if the MMAR_2894 gene is required for the hemolytic activity of M. marinum. As shown in Fig. 4A, the WT strain lyses sRBCs. Hemolysis was almost completely abolished in the ΔRD1 strain, as established previously (46). Consistent with the reduction of EsxA secretion (Fig. 3A to C), the ΔMMAR_2894 strain exhibited hemolytic activity that was comparable to those of the ΔRD1 strain and the negative control (no bacteria). Expression of the MMAR_2894-ST gene restored hemolytic activity to the ΔMMAR_2894 strain to levels that were not significantly different from those of the WT strain. Together, these data demonstrate that MMAR_2984 is required for hemolytic activity in M. marinum, further linking this gene to the ESX-1 system.

FIG 4.

The ΔMMAR_2894 strain promotes host cell cytolysis but not hemolysis. (A) Hemolysis of sRBCs by M. marinum. Distilled water (dH₂O) and PBS are positive and negative controls for lysis. Both controls lack bacteria. Means of data from three biological replicates, in technical triplicate, are shown. Significance was determined by ordinary one-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test. Significance relative to lysis by the WT strain is shown. ****, P < 0.0001; ***, P = 0.0008. The hemolytic activities between the WT and complemented strains were not statistically significant. There were no significant differences between the PBS control and the hemolytic activity of the ΔRD1 and the ΔMMAR_2894 strains. (B) For RAW 264.7 infection, EthD-1-stained cells were counted from 10 fields from each of 3 wells (total of 30 fields). Significance was determined using one-way ordinary ANOVA (P < 0.0001) followed by Dunnett’s multiple-comparison test relative to the WT strain. ****, P < 0.001. There were no significant differences between the WT, ΔMMAR_2894, and complemented strains. The data are representative of results from three independent infections.

One reflection of phagosomal lysis is the ability of M. marinum to cause macrophage cell death (21). To further understand the role of MMAR_2894 in ESX-1 secretion and virulence, we infected macrophage-like cells and measured cytolysis (Fig. 4B). To measure cytolysis, we infected RAW 264.7 murine macrophage-like cells with M. marinum strains at a multiplicity of infection (MOI) of 5 for 24 h. We measured cytolysis using ethidium homodimer 1 (EthD-1) staining, which occurs only following cell perforation, as described previously (35, 68, 69). As shown in Fig. 4B, infection with WT M. marinum caused cytolysis of RAW 264.7 cells. The number of permeabilized cells was significantly lower in the uninfected controls and in cells infected with the ΔeccCb₁ or ΔesxBA strain (which lacks a functional ESX-1 system) than in cells infected with the WT strain (P < 0.0001). Notably, infection of RAW 264.7 cells with the ΔMMAR_2894 strain or the complemented strain resulted in levels of cytolysis that were not significantly different from those measured during infection with the WT strain. We found no significant differences in uptake or survival of the WT, ΔeccCb₁, and ΔMMAR_2894 strains at 24 h postinfection (Fig. S4). From these data, we conclude that although the MMAR_2894 gene is required for the optimal secretion of ESX-1 substrates in vitro and for hemolysis, MMAR_2894 is dispensable for cytolysis of RAW 264.7 cells.

DISCUSSION

From the present study, we conclude the following. We demonstrate that MMAR_2894 is an ESX-1-secreted protein in M. marinum. We found that MMAR_2894 is required for the optimal secretion of the established ESX-1 substrates. Finally, we found that MMAR_2894 is required for ESX-1-mediated hemolysis but dispensable for ESX-1-dependent macrophage lysis. These conclusions are based upon our findings that MMAR_2894 is not secreted from an M. marinum strain lacking a functional ESX-1 system. Moreover, deletion of the MMAR_2894 gene caused a significant reduction in the levels of all of the established ESX-1 substrates from M. marinum. The ΔMMAR_2894 M. marinum strain was nonhemolytic, similar to other ESX-1-deficient strains. Yet, in contrast to ESX-1-deficient strains, the ΔMMAR_2894 M. marinum strain caused WT levels of lysis of RAW 264.7 cells during infection. The observed changes in ESX-1 secretion and hemolysis were complemented by the expression of the MMAR_2894 gene in the ΔMMAR_2894 M. marinum strain, definitively linking MMAR_2894 to the ESX-1 system. Collectively, our experimental findings indicate that MMAR_2894 is a new ESX-1 substrate in M. marinum that is not conserved in M. tuberculosis, which may play specific roles in the lytic activity of the ESX-1 system. Moreover, this is one of the first reports of a PE protein outside the extended RD1 region needed for ESX-1 export in M. marinum.

Previous work suggested that MMAR_2894 is an ESX-1 substrate. Phan et al. identified the MMAR_2894 protein as a candidate ESX-1 substrate using proteomics in a recent report (58). We also observed a loss of MMAR_2894 secretion using proteomics-based approaches in our previous unpublished work. Our findings confirm this hypothesis and provide the first characterization of the role of MMAR_2894 in ESX-1 secretion and function. In support of our findings, M. marinum and M. tuberculosis have been reported to secrete several PE proteins, particularly in the PE_PGRS family (15, 47, 50, 53, 54, 56, 70–73). Both the ESX-1 system and a related ESX-type exporter, ESX-3, secrete PE and PPE family proteins encoded by genes in the same genomic locus as those encoding the secretory apparatus (62, 74). PE and PPE protein export has largely been associated with a related ESX-type exporter, ESX-5, which is also required for mycobacterial pathogenesis (15, 56, 57, 73). The PE and PPE proteins exported by the ESX-5 system are encoded by genes found throughout the mycobacterial genome and not necessarily associated with the ESX-5 region (15, 56, 57, 75, 76). Likewise, the MMAR_2894 gene and those encoding the PE35 and PPE68_1 substrates (34, 77) are not linked to the ESX-1 locus in the M. marinum genome.

In support of MMAR_2894 as an ESX-1 substrate, MMAR_2894 shares certain characteristics with other ESX-secreted PE and PPE proteins. BLAST analysis of the MMAR_2894 sequence against the M. marinum genome revealed three paralogous proteins with significant similarity, including MMAR_5447 (PE35_1), MMAR_5286 (PE34, an ortholog of the M. tuberculosis PE35 protein), and MMAR_0185 (PE35) (59). PE35 is an ESX-1 substrate encoded at the ESX-6 locus (54, 78, 79). PE35_1 is encoded at the ESX-1 locus. Interestingly, MMAR_2894 is larger than these paralogous proteins, with all of the similarity within the first 100 amino acids (aa) of MMAR_2894 (see Fig. S5 in the supplemental material). This raises the interesting possibility that PE34 is another ESX-1 substrate. MMAR_2894 contains a YXXXD/E motif, which is shared by other ESX-secreted proteins (77). MMAR_2894 is larger than the 100-amino-acid ESAT-6/WXG100 superfamily proteins secreted by ESX systems (80), such as EsxA and EsxB, which are approximately 10 kDa in size but smaller than other ESX-1-secreted proteins such as EspB (47 kDa) and EspA (40 kDa) (17, 18, 61, 64, 81). Larger PE proteins, like LipY, have C-terminal domains with enzymatic activities (56). However, the C-terminal region of MMAR_2894 was not predicted to have any domains by BLAST analysis. Like other previously described PE and PPE proteins, a C-terminal tag did not hinder export through ESX systems (56, 82). Finally, other PE, PPE (LipY), and ESX (EspB) proteins have shown processing upon secretion (56, 64, 81, 83). We observed at least two bands by Western blot analysis that corresponded to the MMAR_2894-ST protein, which may indicate processing or modification. However, processing or modification may occur independently of secretion because we observed at least two bands corresponding to MMAR_2894-ST in the cell-associated protein fraction.

Our findings are consistent with reports of PE/PPE family proteins impacting the secretion of specific proteins in pathogenic mycobacteria. For example, mutations in the PPE38 protein have been shown to hinder the secretion of PE_PGRS proteins in M. tuberculosis (84). Likewise, the ESX-5a accessory region includes PE/PPE substrate pairs that are required for the secretion of a subset of ESX-5 substrates (85, 86). Our findings indicate that MMAR_2894 is required for the optimal secretion of the established ESX-1 substrates. We found, however, that the loss of MMAR_2894 impacted the secretion of some substrates more severely than others. For example, we measured a log₂ fold reduction of 2.42 for the secretion of the EsxB substrate in the absence of the MMAR_2894 gene. Yet we measured a log₂ fold change reduction of 8.62 for the EspE substrate. We were unable to detect the EspJ substrate in the secreted fraction in the absence of the MMAR_2894 gene. Further work is required to delineate how MMAR_2894 differentially impacts substrate secretion, but MMAR_2894 may indeed be part of an ESX-1 accessory region, similar to those found for ESX-5.

In addition to ESX-1 substrates, we found that the secretion of several PE/PPE family proteins was reduced in the absence of the MMAR_2894 gene (Fig. 3D and Tables S4 and S6 [Table S6 includes proteins that we could identify but for which we could not quantify changes]). Interestingly, five of these proteins (MMAR_0382, MMAR_4999, MMAR_5013, MMAR_3290, and MMAR_2460) are secreted to the M. marinum cell surface in an ESX-5-dependent manner (76). The remaining PE/PPE family proteins that we identified and quantified have not previously been linked to an ESX system. Therefore, MMAR_2894 may impact the secretion of a subset of substrates from several ESX systems, including ESX-1 and ESX-5. Moreover, the additional PE/PPE proteins that we identified may be substrates of ESX systems.

A major finding in our study is the differential requirement for MMAR_2894 for hemolytic and cytolytic activities of the ESX-1 system (Fig. 5). The ESX-1 system is thought promote membrane destruction (31). Early studies of M. tuberculosis linked the cytolysis of pneumocyte cell lines to ESX-1 function (17). In M. marinum, the ESX-1 system is required for contact-dependent hemolysis (31, 46). In a genetic screen to identify the requirements for hemolysis, all of the nonhemolytic strains had transposon insertions in the genetic locus encoding the ESX-1 system (46). The nonhemolytic strains were deficient in ESX-1 secretion and were attenuated in macrophages (46). Importantly, ESX-1-deficient strains of both M. marinum and M. tuberculosis fail to lyse infected macrophages. It was later shown that the ESX-1 systems of both M. marinum and M. tuberculosis permeabilized the phagosomal membrane following the uptake of the bacteria by the macrophage (21, 22, 24). Lysis of the phagosomal membrane by the ESX-1 system is required for macrophage lysis by both M. tuberculosis and M. marinum (21). As such, we and others have used hemolysis by M. marinum to reflect ESX-1 lytic activity against the phagosome (31, 32, 35, 46, 68, 79, 87). Macrophage lysis has been widely used in both M. tuberculosis and M. marinum to reflect phagosomal lysis and cytosolic access (17, 19, 21, 22, 24, 35, 46, 88, 89). It is our understanding that ESX-1-dependent hemolysis and macrophage lysis have not been separable in previous reports. We have not seen another report of a mycobacterial strain which is nonhemolytic but can still lyse macrophages.

FIG 5.

Differential requirement for MMAR_2894 in hemolytic and cytolytic activities of the ESX-1 system. Our findings demonstrate that MMAR_2894 is a new substrate of the ESX-1 system that is required for hemolysis but is dispensable for macrophage cytolysis. MMAR_2984 may do this directly or may impact ESX-1 function by impacting the secretion of additional ESX-1 substrates. These data suggest for the first time that the molecular mechanisms underlying these two lytic functions of the ESX-1 system, hemolysis and phagosomal lysis leading to cell death, are separable.

We can think of three possible explanations consistent with our findings, which are not mutually exclusive. First, MMAR_2894 may play a direct role in hemolysis but not phagosome damage and, consequently, macrophage lysis. The significance of the hemolytic activity of M. marinum is not well understood (31, 46). Moreover, the precise ESX-1 substrates required for phagosomal lysis remain unknown (31). The mechanisms underlying hemolysis and phagosomal lysis may be distinct. It is entirely possible that the ESX-1 substrates that promote hemolysis and phagosomal lysis differ. MMAR_2894 may be a substrate involved in hemolysis that is dispensable for phagosomal lysis, perhaps due to redundancy.

Second, it is possible that the ESX-1 system secretes different sets of effectors under different conditions. Consistent with our findings, MMAR_2894 may be secreted in vitro and during interaction with RBCs but may not be secreted during macrophage infection. Indeed, M. marinum has a broad host range compared to M. tuberculosis (38, 40–43, 45). A transposon-directed insertion site (TraDIS) sequencing-based study demonstrated that mycobacteria tailor virulence factors to specific hosts, supporting the idea that different substrates could be secreted under different infection conditions (45). Importantly, the requirement of specific ESX-1 genes for virulence or antivirulence (insertion in this gene makes M. marinum survive better than the WT strain) was host specific. Those authors concluded that ESX systems are fine-tuned for the survival of M. marinum in specific hosts (45).

Finally, our findings could be explained by the intermediate levels of secreted ESX-1 substrates in the absence of MMAR_2894. We demonstrated that the loss of MMAR_2894 reduced but did not abrogate the secretion of ESX-1 substrates and that substrate secretion was differentially impacted. The loss of MMAR_2894 may result in levels of ESX-1 secretion that are insufficient to promote hemolysis but sufficient to promote phagosomal damage and, consequently, macrophage lysis. Support for this idea comes from studies by us and others that demonstrate that the levels of ESX substrate secretion required to promote function are demonstrably lower that those measured in vitro (12, 35, 90). For example, M. tuberculosis and M. marinum with significantly reduced ESX-1 secretion in vitro were still able to promote virulence in infection models (35, 90). The same trend was observed in the nonpathogenic species Mycobacterium smegmatis. Secretion by the ESX-3 system is required to promote survival under low-iron conditions. M. smegmatis strains with significantly reduced ESX-3 secretion in vitro survived under low-iron conditions (12). Thus, the secretion of ESX-1 substrates in vitro, which may also include hemolysis, may not reflect the levels of secretion required to function in the environment (host, water, and soil, etc.) or to permeabilize the phagosomal membrane (91). Further experimentation is required to distinguish between these possibilities.

Our studies also contribute a new method to directly identify proteins whose secretion is dependent on individual ESX-1 substrates. In this study, we took advantage of the fact that ESX-1 substrates are often required for the secretion of other substrates (61). We were able to identify new potential ESX-1 substrates not by genetically inactivating the secretion system but rather by deleting a gene encoding a potential substrate. By using the proteogenomic workflow described here, we found that the EspJ substrate is not secreted at detectable levels in the absence of MMAR_2984. Moreover, we identified several additional proteins (Table S5) that may represent additional proteins that directly or indirectly depend on MMAR_2894, some of which may represent new ESX-1 substrates. These types of approaches may lead to a new understanding of the total catalog of ESX-1-secreted effectors and may provide new insight into the relationship between and the function of secreted effectors.

Together, our findings define a new secreted protein of the ESX-1 system in M. marinum, demonstrating that there are more substrates to be found. Moreover, our findings begin to phenotypically delineate hemolysis and macrophage lysis, two lytic activities which require a functional ESX-1 system in M. marinum.

MATERIALS AND METHODS

Growth of mycobacterial strains.

All M. marinum strains used in this study were derived from the M. marinum M strain (ATCC BAA-535) and are listed in Table 1. M. marinum strains were grown at 30°C in Middlebrook 7H9 liquid broth (Sigma-Aldrich, St. Louis, MO) supplemented with 0.5% glycerol and 0.1% Tween 80 (Amresco, Solon, OH). Kanamycin (IBI Scientific, Peosta, IA) (20 μg/ml) or hygromycin (EMD Millipore, Billerica, MA) (50 μg/ml) was added where appropriate. These strains were then diluted to an optical density at 600 nm (OD₆₀₀) of 0.2 in technical duplicate and biological triplicate. For in vitro growth curves, OD₆₀₀ values were obtained every 12 h using an Eppendorf BioPhotometer Plus instrument (see Fig. S1 in the supplemental material). For hemolysis assays and infection assays, it was estimated that 1 OD₆₀₀ unit equals 7.7 × 10⁷ cells/ml.

TABLE 1.

Strains used in this study

Strain	Genotype or description	Source or reference
M	WT M. marinum; parental strain for all strains in this study	ATCC BAA-535
ΔRD1	Unmarked deletion of ESX-1 genes eccCb1′-espK′	44
M/pMMAR_2894-ST	Wild-type strain with pMMAR_2894 Strep-tag integrated at the att site	This study
ΔeccCb1	Unmarked deletion of eccCb1	84
ΔeccCb1/pMMAR_2894-ST	eccCb1 deletion strain with pMMAR_2894 Strep-tag integrated at the att site	This study
ΔMMAR_2894	Unmarked deletion of MMAR_2894	This study
ΔMMAR_2894/pMMAR_2894-ST	MMAR_2894 deletion strain complemented with pMMAR_2894 Strep-tag integrated at the att site	This study
ΔesxBA	Deletion of the esxBA genes	46

Generation of the ΔMMAR_2894 M. marinum strain.

All of the primers used in this study were purchased from Integrated DNA Technologies (IDT) (Coralville, IA) and are listed in Table 2. M. marinum M genomic DNA was isolated as described previously (66). The 1,501 bp of DNA upstream of the annotated MMAR_2894 gene (59) were amplified from M. marinum M genomic DNA using oligonucleotide primers orb130 and orb131. The 1,495 bp of DNA downstream of the MMAR_2894 gene was amplified from M. marinum genomic DNA using oligonucleotide primers orb132 and orb133. The primers were designed to retain the first five and the last five codons of the MMAR_2894 protein. The p2NIL-ΔMMAR_2894 plasmid was generated using FastCloning (92). The pGOAL cassette was added to the p2NIL-ΔMMAR_2894 plasmid as described previously (87). The resulting p2NIL-ΔMMAR_2894 suicide plasmid was UV treated and introduced into the wild-type strain by electroporation as previously described (87). Resolved strains were screened for the loss of the MMAR_2894 gene using PCR (Fig. 2) and verified by targeted DNA sequencing of the PCR product with oligonucleotide primers orb150 and orb151 at the Genomics and Bioinformatics Core Facility at the University of Notre Dame.

TABLE 2.

Primers used in this study^a

^aAll primers were generated in this study.

Generation of the p_MOP-MMAR_2894 Strep-tag plasmid.

The p_MOP-MMAR_0039-ST plasmid was used as the parent plasmid for the plasmid expressing the MMAR_2894-ST gene. The p_MOP-MMAR_0039 (35) plasmid was amplified with oligonucleotide primers omf003 and omf004. These primers resulted in a vector backbone including the Step-tag and a novel SpeI restriction site. The amplified plasmid was DpnI (NEB) treated and introduced into chemically competent Escherichia coli as described previously (87). The resulting parental plasmid was verified by SpeI (NEB) restriction enzyme digestion, confirming the introduced SpeI site.

The p_MOP-MMAR_2894-ST plasmid was constructed by amplifying the MMAR_2894 gene from M. marinum genomic DNA using oligonucleotide primers orb74 and orb75. The p_MOP-MMAR_0039-ST plasmid, excluding the MMAR_0039 gene, was amplified using oligonucleotide primers omf057 and omf004. The resulting plasmid, bearing the MMAR_2894 gene behind the mycobacterial optimal promoter with a 3′ Strep-tag (5′-GGATGGAGCCACCCGCAGTTCGAAAAATGA-3′, encoding the affinity tag GWSHPQFEK), was constructed by FastCloning (68, 92). The p_MOP-MMAR_2894 Strep-tag plasmid was confirmed using DNA sequencing with the MOPs forward and the MOPs reverse primers (at the Genomics and Bioinformatics Facility at the University of Notre Dame [described in reference 69]).

Hemolysis assay.

Sheep red blood cell (sRBC) hemolysis assays were performed exactly as previously described (79). The data shown in Fig. 3 are the results of three independent biological replicates, each with three technical replicates.

Protein preparation and analysis.

ESX-1 protein secretion assays were performed exactly as previously described (79). In short, M. marinum strains were grown in 7H9 broth and diluted to an OD₆₀₀ of 0.8 in Sauton’s broth. Following 48 h of growth, the M. marinum cells were collected by centrifugation. The cells were lysed using a Biospec Mini-BeadBeater-24 instrument, and the lysate was clarified by centrifugation. The resulting proteins are the cell-associated protein fraction. The supernatant was filtered and concentrated, yielding the secreted protein fraction. The protein concentration was defined using the MicroBCA kit from Pierce. The amount of protein loaded is indicated in each figure legend. RpoB (RpoB) (1:10,000) (catalog number ab12087; Abcam, Cambridge, UK) and ESAT-6 (EsxA) (1:5,000) (catalog number HYB 076-08-02; Thermo Fisher, Waltham, MA) were detected as described previously (79). The following reagents were obtained through BEI Resources, NIAID, NIH, and used as previously described (79, 87): polyclonal anti-Mycobacterium tuberculosis CFP10 (gene Rv3874) (antiserum, rabbit) (NR-13801) and polyclonal anti-Mycobacterium tuberculosis Mpt32 (gene Rv1860) (antiserum, rabbit) (NR-13807). C-terminally Strep-tagged MMAR_2894 was detected using an NWSHPQFEK rabbit polyclonal antibody (GenScript, Piscataway, NJ), which was resuspended in water to a final concentration of 0.5 mg/ml and used at a 1:5,000 dilution.

RAW 264.7 cytotoxicity assay.

RAW 264.7 cells (ATCC TIB-71) were cultured in high-glucose, high-pyruvate Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Dublin, Ireland) with 10% heat-inactivated fetal bovine serum (FBS) (VWR, Radnor, PA). Macrophages were washed with sterile 1× phosphate-buffered saline (PBS) (Gibco) and harvested by cell scraping for passage as needed. All experiments were performed at 37°C with 5% CO₂.

RAW 264.7 cells were seeded in 1 ml DMEM plus 10% FBS per well at 5 × 10⁵ cells/ml in a 24-well plate (Greiner Bio-One, Germany) and allowed to grow for 24 h. Bacteria were added at an estimated MOI of 5 (2.5 × 10⁶ cells/ml) in technical triplicate and mixed. Infections were allowed to proceed for 2 h before gentamycin (RPI Corporation, Mt. Prospect, IL) was added at 100 μg/ml. Infection mixtures were incubated for an additional 2 h before washing three times with sterile 1× PBS and adding fresh DMEM plus 10% FBS. Infections were then allowed to proceed for 24 h. To assay cytotoxicity, medium was aspirated, and 250 μl of an EthD-1 (1 μl/ml)–Calcein-AM (0.25 μl/ml) (Live/Dead viability/cytotoxicity kit; Life Technologies, Carlsbad, CA) solution in 1× PBS was added. Cells were incubated for an additional 30 min and imaged using a Zeiss AxioObserver A1 inverted microscope with phase-contrast, rhodamine (red), and green fluorescent protein (GFP) filters. Ten images were taken per well, and ImageJ was used to quantify the number of dead cells exactly as described previously (68). For CFU determination at 24 h postinfection (Fig. S4), bacteria were added at an estimated MOI of 1 (5 × 10⁵ cells/ml). After 24 h, medium was removed, and 0.5 ml of filter-sterilized lysis buffer (H₂O plus 0.1% [vol/vol] Tween 80) was added. Following the addition of lysis buffer, plates were incubated at 37°C for 10 min before scraping and pipetting up and down. Cells were plated at 1:100 and 1:1,000 dilutions using filter-sterilized dilution buffer (1× PBS plus 0.05% [vol/vol] Tween 80). Fifty microliters of each dilution was plated on Middlebrook 7H11 plates supplemented with 10% (vol/vol) oleic acid-albumin-dextrose-catalase (OADC) and 0.5% (vol/vol) glycerol. Colonies were counted following a 1-week incubation at 32°C.

For entry assays (Fig. S4A), infections were performed as described above at an estimated MOI of 5. Infection mixtures were incubated for 2 h and then washed three times with sterile 1× PBS to remove extracellular bacteria. PBS was removed, and 0.5 ml of filter-sterilized lysis buffer (H₂O plus 0.1% Tween 80) was added. Following the addition of lysis buffer, plates were incubated at 37°C for 10 min before scraping and pipetting up and down. Cells were plated at 1:100, 1:1,000, and 1:10,000 dilutions using filter-sterilized dilution buffer (1× PBS plus 0.05% Tween 80). Fifty microliters of each dilution was plated on Middlebrook 7H11 (Sigma-Aldrich, St. Louis, MO) plates supplemented with 10% (vol/vol) OADC and 0.5% (vol/vol) glycerol. Colonies were counted following a 1-week incubation at 32°C.

Proteomic analysis of secreted proteins.

Cell-associated and secreted protein fractions of the wild-type, ΔMMAR_2894, and complemented M. marinum strains were prepared for quantitative bottom-up proteomics similar to our previously reported work (87). Fifty micrograms of each sample in biological duplicate was denatured with 5% SDS, reduced with 15 mM dithiothreitol (DTT) at 60°C for 45 min, and then alkylated with a 2-fold molar excess of iodoacetamide (IAA) (Sigma) for 20 min in the dark. The samples were acidified by the addition of 10% (vol/vol) with 12% phosphoric acid and diluted by a 7-fold addition of 95:5 methanol–50 mM ammonium bicarbonate (ABC) in water. The protein solution was loaded onto an S-Trap spin column (Protifi, Huntington, NY) and washed 3 times with the 95:5 solution, 1.2 μg of sequencing-grade trypsin (Promega, Fitchburg, WI) was added to each sample trap, and the samples were incubated overnight at 37°C with the addition of ∼100 μl 50 mM ABC. Peptides were eluted with two additions of an 80:20 mixture of water-acetonitrile plus 0.1% formic acid (FA) and further acidified with FA as needed. Peptide digests were desalted using 10 mg HLB solid-phase extraction cartridges (Waters, Milford, MA) according to the manufacturer’s instructions, dried, and resuspended in 100 μl of 0.1% FA prior to analysis.

Liquid chromatography-tandem mass spectrometry (LC-MS/MS) acquisition and analysis were performed essentially as previously described (87, 93). Label-free quantification (LFQ) was performed on secreted and cell-associated samples separately in technical duplicate (and biological replicate) using MaxQuant (v1.6.2.3) against the MycoBrowser M. marinum protein database (59) downloaded as of August 2018. Data were normalized using the nearest neighbor and median fold change, as previously described (87, 94–96).

Data visualization/reduction and volcano plots were performed using LEMMA analysis modified for proteomics data via Proteosign software (97) and visualized with Perseus and Plotly as described previously (87, 98). Integrated quantitative data are in the supplemental material, and statistical volcano plot data are in Table S2. Multiple-hypothesis P values were capped at a −log₁₀ value of 6.1 (10⁻⁶) to reflect the dynamic range of the instrument between an average abundant protein and the lowest detected average protein.

Functional analysis of secreted proteins was performed using functional categories as annotated by the MycoBrowser M. marinum database (59). Levels of proteins included in the analysis were significantly increased or decreased (log₂ fold change of ≥1; −log₁₀ P value of ≥1.3) in the secreted fractions of the ΔMMAR_2894 strain compared to both the wild-type and complemented strains.