Polarly Localized EccE₁ Is Required for ESX-1 Function and Stabilization of ESX-1 Membrane Proteins in Mycobacterium tuberculosis

Tuberculosis (TB), the world’s leading cause of death of humans from an infectious disease, is caused by the intracellular bacterium Mycobacterium tuberculosis. The development of successful strategies to control TB requires better understanding of the complex interactions between the pathogen and the human host. We investigated the contribution of EccE₁, a membrane protein, to the function of the ESX-1 secretion system, the major virulence determinant of M. tuberculosis. By combining genetic analysis of selected mutants with eukaryotic cell biology and proteomics, we demonstrate that EccE₁ is critical for ESX-1 function, secretion of effector proteins, and pathogenesis. Our research improves knowledge of the molecular basis of M. tuberculosis virulence and enhances our understanding of pathogenesis.

ABSTRACT

Mycobacterium tuberculosis is a slow-growing intracellular bacterium with the ability to induce host cell death and persist indefinitely in the human body. This pathogen uses the specialized ESX-1 secretion system to secrete virulence factors and potent immunogenic effectors required for disease progression. ESX-1 is a multisubunit apparatus with a membrane complex that is predicted to form a channel in the cytoplasmic membrane. In M. tuberculosis this complex is composed of five membrane proteins: EccB₁, EccCa₁, EccCb₁, EccD₁, and EccE₁. In this study, we have characterized the membrane component EccE₁ and found that deletion of eccE₁ lowers the levels of EccB₁, EccCa₁, and EccD₁, thereby abolishing ESX-1 secretion and attenuating M. tuberculosis ex vivo. Surprisingly, secretion of EspB was not affected by loss of EccE₁. Furthermore, EccE₁ was found to be a membrane- and cell wall-associated protein that needs the presence of other ESX-1 components to assemble into a stable complex at the poles of M. tuberculosis. Overall, this investigation provides new insights into the role of EccE₁ and its localization in M. tuberculosis.

IMPORTANCE Tuberculosis (TB), the world’s leading cause of death of humans from an infectious disease, is caused by the intracellular bacterium Mycobacterium tuberculosis. The development of successful strategies to control TB requires better understanding of the complex interactions between the pathogen and the human host. We investigated the contribution of EccE₁, a membrane protein, to the function of the ESX-1 secretion system, the major virulence determinant of M. tuberculosis. By combining genetic analysis of selected mutants with eukaryotic cell biology and proteomics, we demonstrate that EccE₁ is critical for ESX-1 function, secretion of effector proteins, and pathogenesis. Our research improves knowledge of the molecular basis of M. tuberculosis virulence and enhances our understanding of pathogenesis.

INTRODUCTION

Mycobacterium tuberculosis is a slow-growing intracellular pathogen with the ability to infect and survive inside macrophages. Protein secretion is essential for mycobacterial virulence and host-pathogen interactions (1). The M. tuberculosis genome encodes five specialized secretion systems, referred to as ESX or type VII systems, termed ESX-1 to ESX-5 (2). ESX systems are multisubunit apparatuses that have similar structures and secrete related proteins but have different functions (3, 4).

ESX-1 is involved in virulence factor secretion and pathogenesis. It is essential for phagosomal rupture, thereby allowing translocation to the cytosol, where bacteria can induce host cell death and subsequently spread to neighboring cells (5–8). Deletion or inactivation of ESX-1 does not affect growth in vitro but causes attenuation of virulence in infection models (9, 10). Indeed, loss of eight genes (RD1) from the esx-1 locus is the primary attenuating deletion of the live tuberculosis vaccine, Mycobacterium bovis BCG (11). The locus encoding the ESX-1 secretion system spans 20 genes and codes for secreted factors and structural components. Additionally, the distal espACD operon, encoding three proteins (EspA, EspC, and EspD), is also required for full ESX-1 activity (12, 13).

The main ESX-1 substrates, EsxA and EsxB, are major virulence factors and among the most potent T-cell antigens of M. tuberculosis (4, 14, 15). Besides these two substrates, other proteins are also secreted through the ESX-1 machinery, such as EspB, EspA, and EspC. Importantly, EsxA/EsxB and EspA/EspC, but not EspB, are codependent for secretion (12, 16, 17). Depletion of any of these proteins leads to severe attenuation in cellular and animal models of infection.

ESX-1 has a set of conserved components, with paralogs in all mycobacterial ESX systems, that have been shown to contribute to the ESX secretion process in at least one mycobacterial species (6, 9, 10, 18–23). The five conserved elements EccB₁, EccCa₁, EccCb₁, EccD₁, and EccE₁ are likely to be membrane proteins and the core components of the ESX-1 machinery. Paralogs of these proteins encoded by the esx-5 locus form the membrane complex of the ESX-5 secretion system in Mycobacterium bovis, Mycobacterium marinum, and Mycobacterium xenopi (6, 23, 24). New structural insight into the membrane complex indicated that the four components Ecc(BCDE)₅ are present in equimolar amounts and adopt a hexameric arrangement around a central pore (6, 23, 24). Complex formation of ESX-1 membrane proteins has also been analyzed in M. marinum, showing the same composition and size as the ESX-5 complex (23). Considering the conserved nature of the Ecc proteins, the structure of the ESX-1 membrane complex should resemble that of ESX-5.

Another conserved component of the ESX-1 apparatus is MycP₁, a membrane protein, which is not part of the core membrane complex but is loosely associated with it (23). MycP₁ is a subtilisin-like serine protease that cleaves EspB in the periplasmic space (23). Besides its role in substrate processing, MycP₁ plays a second role in the secretion process by stabilizing the ESX-1 membrane complex (23). In the esx-1 locus of M. tuberculosis, mycP₁ is situated upstream of eccE₁, and the two genes are cotranscribed (25). EccE₁ is the conserved element that has been the least explored, especially in M. tuberculosis, where no experimental work has been reported yet. In Mycobacterium smegmatis, the homolog of EccE₁ is required for EsxA and EsxB secretion (19). Previous work on the composition and structure of the ESX-5 membrane complex demonstrated that EccE₅ is located at the perimeter of the membrane complex and is important for its formation and stability (6, 24).

Although the molecular mechanisms underlying the ESX secretion process are not fully understood, studying individual components has provided important insight. Localizing active ESX systems can also offer valuable information about the mechanism of secretion. ESX-1-related proteins have been visualized at the cell poles of M. marinum and M. smegmatis (26, 27). Nonetheless, information about the localization of the entire ESX-1 system in M. tuberculosis is missing.

In this study, we explored the role of EccE₁ in M. tuberculosis. We show that EccE₁ is required for ex vivo virulence, for stabilizing ESX-1 membrane proteins, and for secretion of EsxA, EsxB, EspA, and EspC. Moreover, we investigated the localization of the protein and its incorporation into a functional ESX-1 system. Our findings indicate that EccE₁ is a membrane protein that requires other ESX-1 components to form a stable complex at the poles of M. tuberculosis.

RESULTS

Mutant construction.

To assess the role of EccE₁ in ESX-1-related functions, we first constructed an eccE₁ deletion mutant of M. tuberculosis. As eccE₁ is in the same transcriptional unit as mycP₁ (25), we initially deleted the entire mycP₁-eccE₁ region from the chromosome by allelic exchange. Whole-genome sequencing confirmed the complete deletion of the mycP₁-eccE₁ coding sequences (CDS) and the absence of other mutations in the genome. The resulting mycP₁-eccE₁ double mutant was then partially or fully complemented with an integrative plasmid bearing either the single gene mycP₁ or eccE₁ or the entire mycP₁-eccE₁ locus under the control of the PTR promoter. As a result, we obtained four strains: a ΔmycP₁-eccE₁ double mutant, a fully complemented ΔmycP₁-eccE₁/mycP₁-eccE₁ strain, and two partially complemented strains, ΔmycP₁-eccE₁/mycP₁ and ΔmycP₁-eccE₁/eccE₁, which correspond to the ΔeccE₁ and ΔmycP₁ single mutants, respectively.

The four strains were subsequently analyzed by RNA sequencing and the transcriptional profile compared to that of the H37Rv wild-type (WT) strain. As anticipated, no deregulated genes were found except for the deleted genes, suggesting that the excision of eccE₁-mycP₁ and their complementation in trans did not impact the transcription of other genes. In the complemented derivatives, the expression levels of mycP₁ and eccE₁ were very similar to those measured in the WT strain (see Table S1 in the supplemental material).

The ability of the mutant strains to grow in synthetic media was explored. The ΔeccE₁ mutant, as well as the ΔmycP₁ and the ΔmycP₁-eccE₁ strains, exhibited growth kinetics similar to that of the WT strain (see Fig. S1 in the supplemental material). These results indicate that eccE₁ and mycP₁ are not required for in vitro growth of M. tuberculosis.

EccE₁ and MycP₁ are not involved in susceptibility to drugs targeting the cell envelope.

Previous studies on the ESX-5 system demonstrated that disruption of this apparatus in M. tuberculosis reduces cell wall integrity (18). Consequently, susceptibility to various antibiotics targeting several steps in the cell wall biosynthetic pathway greatly increased (18). To test whether its paralog, the ESX-1 system, is also involved in maintaining the stability of the cell envelope, the susceptibility of the mutants to various cell wall-targeting antibiotics was evaluated using the resazurin-based microdilution assay (REMA) (28). The ΔeccE₁ mutant and the WT strain displayed similar MICs for all antibiotics tested, as did the ΔmycP₁ and the ΔmycP₁-eccE₁ mutants (see Table S2 in the supplemental material). We concluded that the absence of either eccE₁ or mycP₁ did not impact susceptibility to β-lactams (penicillins and cephalosporins) or to glycopeptides (vancomycin). In contrast to the case for ESX-5, a secretion system previously described as essential for cell wall integrity in M. tuberculosis, ESX-1 inactivation did not impact susceptibility to the antibiotics tested.

EccE₁ and MycP₁ are required for M. tuberculosis-mediated cell death.

EccE was identified as a peripheral component of the ESX membrane complex, but further information about its role in M. tuberculosis is missing. To investigate the involvement of EccE₁ in ESX-1-mediated function, we exploited the ability of M. tuberculosis to induce host cell death, which is linked to ESX-1-related secretion (5, 7). To this end, THP-1 human macrophages were infected with the mycP₁ and eccE₁ mutants as well as with the WT strain, and the survival of macrophages was monitored at 72 h postinfection. In the presence of WT M. tuberculosis, the survival of macrophages significantly decreased compared to that for the uninfected control. In contrast, when macrophages were infected with strains lacking eccE₁ and/or mycP₁, the macrophages survived to levels similar to that for the noninfected control. Importantly, the complemented mutant restored the WT phenotype (Fig. 1). This result indicates that M. tuberculosis needs both EccE₁ and MycP₁ for cytotoxicity and that the absence of either of these ESX-1 components leads to complete attenuation of M. tuberculosis ex vivo.

FIG 1.

EccE₁ and MycP₁ are required to induce M. tuberculosis-mediated cell death. Cell viability of THP-1 human macrophages at 72 h postinfection is shown. Uninfected cells were used as a control for macrophage survival and the H37Rv WT strain as a positive control for M. tuberculosis-mediated cell death. The ΔeccE₁, ΔmycP₁-eccE₁, and ΔmycP₁ mutants did not affect survival of THP-1 macrophages, in contrast to the case for the complemented strain. Bars represent mean values from three independent biological replicates, and error bars show the standard deviation. One-way ANOVA followed by Tukey’s multiple-comparison test was used for statistical comparison. ****, P < 0.001.

To exclude an infection defect that could impact macrophage survival and to validate the attenuated phenotype displayed in the absence of eccE₁ and/or mycP₁, we determined the levels of intracellular bacteria at 4 h postinfection by CFU enumeration. The bacterial burden was similar for the WT and the mutants, indicating that all strains were phagocytosed with similar efficiency (see Fig. S2 in the supplemental material). These results demonstrated that M. tuberculosis does not require EccE₁ or MycP₁ to infect macrophages, but it does need both components to induce host cell death.

EccE₁ and MycP₁ are essential for ESX-1 protein secretion.

To determine if EccE₁ and MycP₁ are crucial to the function of the ESX-1 machinery, we checked whether the deletion of eccE₁ and/or mycP₁ impacts ESX protein secretion. To this end, cell lysates and culture filtrates of WT and mutant bacteria were prepared and analyzed by immunoblotting. The presence of Ag85B, an ESX-1-independent secreted protein, and the absence of the cytosolic GroEL2 protein in the culture filtrates served as controls. As expected, EsxA and EsxB were present in the secretome of the WT M. tuberculosis strain. However, EsxA and EsxB were not detected in culture filtrates of strains lacking eccE₁ and/or mycP₁ although both substrates were present in the cell lysates. Furthermore, secretion of EsxA and EsxB was restored in the fully complemented ΔmycP₁-eccE₁/mycP₁-eccE₁ strain (Fig. 2A). These data indicate that both EccE₁ and MycP₁ are required for secretion of the two main ESX-1 substrates. In contrast, EspB, another known ESX-1 substrate, was found in the cell lysates and culture filtrates of the WT, mutant, and complemented strains (Fig. 2A), suggesting that secretion of EspB is independent of EccE₁ or MycP₁. Overall, we conclude that expression of EsxA, EsxB, and EspB is not affected by the absence of EccE₁ or MycP₁. However, both ESX-1 membrane proteins are essential for EsxA and EsxB secretion but dispensable for the release of EspB into the culture filtrate.

FIG 2.

EccE₁ and MycP₁ are required for secretion of EsxA, EsxB, EspA, and EspC but not EspB. (A) Immunoblots of culture filtrates (CF) (15 μg per well) and whole-cell lysates (CL) (10 μg per well). Detection of Ag85B was used as a loading control for the CF and detection of GroEL2 as a loading control in the CL. Secretion of EsxA and EsxB but not EspB is disrupted in the ΔeccE₁, ΔmycP₁-eccE₁, and ΔmycP₁ mutants. (B) Proteomic analysis of the secreted fraction, comparing the ΔmycP₁-eccE₁ mutant with the H37Rv WT strain. Each dot corresponds to an identified protein. Proteins are represented using a volcano plot-based strategy where t-test P values on the −log base 10 scale are combined with ratio information on the log base 2 scale. The dashed lines represent the significance curve (SO value of 0.5) and delineate the differentially quantified proteins. Three independent biological replicates per strain were analyzed, and a t test (significance level of 0.05) was used for statistical comparison. The abundance of EsxA, EsxB, EspA, and EspC is highly affected in the ΔmycP₁-eccE₁ double mutant compared to the WT strain. EsxA, EsxB, EspA, and EspC are the only significantly reduced proteins.

To obtain deeper knowledge about the impact of EccE₁ and MycP₁ on M. tuberculosis secretion and to identify potential EccE₁-dependent substrates, the whole secretomes of the mutants were analyzed and compared to that of the WT strain (see Table S5 in the supplemental material). Bacteria were grown under the same conditions as described above for immunoblotting and the secreted fraction analyzed by mass spectrometry. We found that EsxA, EsxB, EspA, and EspC were the only proteins that were significantly reduced in the culture filtrates (Fig. 2B). Moreover, in the WT strain, EsxA was in the top 10 most abundant proteins. On the other hand, the secretion of this protein greatly decreased (log₂ ratio, −5.1) when EccE₁ and MycP₁ were missing (Table S5). The same trend was observed for EsxB, EspA, and EspC, whose secretion was also greatly decreased (log₂ ratio, [−7.8, −3.6]) in the deletion mutants compared to the WT strain (Table S5). It was noteworthy that the levels of the other ESX-1-related proteins, including EspB, were not different in the absence or presence of EccE₁ and MycP₁ (Fig. 2B). Proteins using other secretion systems, such as Ag85B, were secreted at similar levels in the WT and in the mutants (Fig. 2B). Taken together, the secretome analysis demonstrated that only secretion of EsxA, EsxB, EspA, and EspC is dependent on EccE₁ and MycP₁. This analysis also confirmed the previous results obtained by immunoblotting.

EccE₁, but not MycP₁, impacts the levels of various ESX-1 membrane proteins.

In order to evaluate whether the absence of EccE₁ and MycP₁ could affect the stability of other ESX-1 membrane proteins, we analyzed the intracellular proteomes of the mutants (ΔeccE₁, ΔmycP₁, and ΔmycP₁-eccE₁) and compared them to that of the WT strain (see Table S6 in the supplemental material). As expected, EccE₁ and MycP₁ were present in the WT and absent in the corresponding mutants. In addition, the intracellular control proteins GroEL2 and RpoB were found at similar levels in all strains. No significant differences were observed for EsxA, EsxB, or EspB (Fig. 3; Table S6), thus corroborating the data obtained by immunoblotting (Fig. 2A). However, in the absence of EccE₁, the levels of EccCa₁, EccB₁, and EccD₁ were reduced (Fig. 3A and C; Table S6). Interestingly, the abundance of these membrane proteins did not change in the ΔmycP₁ mutant compared to the WT strain (Fig. 3B; Table S6). Considering the transcriptome sequencing (RNA-seq) data, which showed no deregulation of eccCa₁, eccB₁, and eccD₁ transcription (Table S1), these data suggest that the absence of EccE₁, but not MycP₁, impacted EccCa₁, EccB₁, and EccD₁ in a posttranscriptional manner.

FIG 3.

Deletion of EccE1, but not MycP1, impacts stability of various ESX-1 membrane proteins. Proteomic analysis of the cell lysate fractions, comparing different M. tuberculosis mutants to the WT strain, is shown. Each dot corresponds to an identified protein. Proteins are represented using a volcano plot-based strategy where t-test P values on the −log base 10 scale are combined with ratio information on the log base 2 scale. The dashed lines represent the significance curve (SO value of 0.5) and delineate the differentially quantified proteins. Three independent biological replicates per strain were analyzed, and a t test (significance level of 0.05) was used for statistical comparison. Pink, ESX-1 membrane proteins; blue, ESX-1 secreted proteins and intracellular protein controls. (A) ΔmycP₁-eccE₁/mycP₁ versus WT. (B) ΔmycP₁-eccE₁/eccE₁ versus WT. (C) ΔmycP₁-eccE₁ versus WT.

EccE₁ is a membrane- and cell wall-associated protein.

Since antibodies against EccE₁ of M. tuberculosis were not available, a hemagglutinin (HA) tag was inserted in the C-terminal part of EccE₁ by genetic engineering, as recently applied to other ESX-1-related proteins with success (29). For this purpose, the ΔmycP₁-eccE₁ double mutant was complemented with an integrative plasmid carrying mycP₁ and eccE₁-HA (ΔmycP₁-eccE₁/mycP₁-eccE₁HA). Both genes were present in single copy and expressed under the control of the same promoter, PTR. To ensure that the tag did not interfere with the activity of EccE₁, the ability of the HA-tagged complemented mutant to induce cell lysis was tested. The THP-1 ex vivo model of infection confirmed that the EccE₁-HA-expressing mutant was as cytolytic as the WT strain, thereby implying the presence of a functional ESX-1 system and thus a functional EccE₁ protein (Fig. 4A).

FIG 4.

EccE₁ localizes to the membrane and cell wall fractions of M. tuberculosis. (A) THP-1 macrophage survival at 72 h postinfection. The ΔmycP₁-eccE₁/mycP₁-eccE₁HA mutant induces cell death ex vivo, indicating the presence of a functional ESX-1 system and therefore that the HA tag at the C-terminal part of the protein did not interfere with EccE₁ function. Uninfected cells were used as a control for macrophage survival, and the H37Rv WT strain and the ΔmycP₁-eccE₁ mutant served as positive and negative controls, respectively, for M. tuberculosis-mediated cell death. Bars represent mean values from three independent replicates, and error bars show the standard deviation. One-way ANOVA followed by Tukey’s multiple-comparison test was used for statistical comparison. ****, P < 0.001. (B) Subcellular fractionation of the ΔmycP₁-eccE₁/mycP₁-eccE₁HA mutant followed by immunoblotting of each fraction (20 μg per well). Detection of RpoB represented a control for lysis, that of EsxB for secretion, and that of Rv3852 for the membrane fraction. Anti-HA antibodies localized EccE₁.

EccE₁ is predicted to be a membrane protein in M. tuberculosis. Subcellular fractionation of the mutant carrying eccE₁-HA was therefore performed and the cytosolic, membrane, cell wall, capsular, and secreted fractions probed by immunoblotting using control proteins of known subcellular location to validate the procedure. RpoB, an RNA polymerase subunit, was found mainly in the cytosol and partly in the membrane. Rv3852, a membrane protein, was present only in the membrane fraction, as previously described (30). EsxB, a substrate of the ESX-1 system, was detected in the cytosol and to a lesser extent in the capsule, although it was found mainly in the culture filtrate. Finally, EccE₁ was identified in the membrane and cell wall but not in other compartments of M. tuberculosis (Fig. 4B).

EccE₁ assembles with other ESX-1 proteins at the poles of M. tuberculosis.

Fluorescent fusion proteins have been successfully employed to study ESX-1-related proteins in M. smegmatis (27). Overexpression of reporter proteins is usually required for visualization; however, production of a large amount of protein can eventually lead to aberrant phenotypes or artifacts. To avoid this potential problem, we constructed a fluorescent fusion protein of EccE₁ with mNeon, a 27-kDa monomeric fluorescent protein reported to be three to five times brighter than green fluorescent protein (GFP) and export competent (31–33). To generate the EccE₁-mNeon-expressing strain, the same strategy used for generating the ΔmycP₁-eccE₁/mycP₁-eccE₁-HA strain was followed. The complemented derivative (ΔmycP₁-eccE₁/mycP₁-eccE₁mNeon) has a single copy of the mycP₁ and eccE₁mNeon genes under the control of the PTR promoter, which provides expression of EccE₁-mNeon close to physiological levels. To check whether fusion to mNeon impacted EccE₁ function, the cytotoxic phenotype of the complemented mutant was analyzed using THP-1 macrophages. The EccE₁-mNeon-expressing mutant induced cell death similarly to the WT strain (Fig. 5B), suggesting that this mutant can form a functional ESX-1 secretion system. Since the WT copy of EccE₁ is not present in the mutant and this protein is required for ESX-1 function in M. tuberculosis, all active ESX-1 apparatuses in the fluorescent mutant should contain EccE₁-mNeon proteins.

FIG 5.

EccE₁-mNeon and the functional ESX-1 system localize to the poles of M. tuberculosis. (A) EccE₁-mNeon-expressing cells (ΔmycP₁-eccE₁/mycP₁-eccE₁-mNeon) display a fluorescent signal at the poles of the bacilli. When mNeon is expressed without EccE₁ (ΔmycP₁-eccE₁/mNeon), the fluorescent signal distributes all over the bacterium. Expressing EccE₁-mNeon in the ΔΔRD1 mutant, which lacks most of the ESX-1 genes, did not result in polar localization, showing that EccE₁ requires another ESX-1 protein to localize at the poles. Scale bars, 3 μm. (B) THP-1 survival at 48 h postinfection. The ΔmycP₁-eccE₁/mycP₁-eccE₁-mNeon mutant induces cell death ex vivo, suggesting the presence of functional ESX-1 systems and therefore indicating that the addition of mNeon did not interfere with EccE₁ function. Uninfected cells were used as a control for macrophage survival, and the H37Rv WT strain and the ΔmycP₁-eccE₁ mutant were used as positive and negative controls, respectively, for M. tuberculosis-mediated cell death. Bars represent mean values from three independent replicates, and error bars show the standard deviation. One-way ANOVA followed by Tukey’s multiple-comparison test was used for statistical comparison. ****, P < 0.001.

To gain more insight into the localization of EccE₁ and of the ESX-1 system in M. tuberculosis, the fluorescent ΔmycP₁-eccE₁/mycP₁-eccE₁mNeon mutant was used to image live bacteria. The ΔmycP₁-eccE₁ double mutant expressing mNeon alone (ΔmycP₁-eccE₁/mNeon) and the H37Rv ΔΔRD₁ mutant (lacking the extended esx-1 locus) (34) expressing EccE₁-mNeon (ΔΔRD₁/mycP₁-eccE₁mNeon) were used as controls. We observed bipolar foci in all ΔmycP₁-eccE₁/mycP₁-eccE₁mNeon cells examined (see S3A in the supplemental material). However, cells expressing mNeon alone displayed a diffuse signal throughout the bacterium (Fig. 5A), as did the ΔΔRD₁ mutant expressing EccE₁-mNeon (Fig. 5A and S3B). These results indicate that most of the EccE₁ proteins localize to the poles of M. tuberculosis in the presence of a functional ESX-1 system and suggest that the bipolar localization may correspond not only to the EccE₁ protein but also to the assembled ESX-1 complex.

DISCUSSION

In the present study, we constructed and employed mutants and genetic tools to understand the localization and functional role in M. tuberculosis of EccE_1, an as-yet-unexplored ESX-1 membrane protein. We demonstrate that the absence of EccE₁ does not impact antibiotic susceptibility or bacterial growth in vitro (see Fig. S1 in the supplemental material), whereas EccE₁ is essential for macrophage lysis, a function required for M. tuberculosis pathogenesis (Fig. 1). To the best of our knowledge, this is the first time that disruption of the core component EccE₁ in M. tuberculosis has proved to lead to complete attenuation ex vivo.

Analysis of the ESX-1 secretome explains this phenotype, as secretion of EsxA, EsxB, EspA, and EspC, which is essential to promote virulence (12, 13, 20), was found to be strongly dependent on EccE₁ (Fig. 2). Proteomic data showed no significant variations in the intracellular levels of EsxA, EsxB, EspA, and EspC when EccE₁ and MycP₁ were missing (Fig. 3), indicating that the reduced secretion of these ESX-1 substrates is due to disruption of the secretion process and not to decreased protein production. The same phenotype was also observed in the ΔmycP₁ mutant, confirming that MycP₁ is essential for EsxA/B secretion and virulence in M. tuberculosis, as previously reported (21). However, secretion of EspB was not affected by the absence of EccE₁ and/or MycP₁, suggesting that secretion of this protein is not ESX-1 dependent (Fig. 2) and that another mechanism may be involved, as postulated earlier for EspD (35). In contrast with our findings, EspB secretion was previously reported to require MycP₁ in both M. tuberculosis and M. marinum (21, 23). In our investigation, secretion of EspB was analyzed in the M. tuberculosis H37Rv strain, whereas in previous studies, an Erdman background was used (21). However, EspB was also secreted by an attenuated H37Rv ESX-1 mutant lacking EspL, while secretion of the main ESX-1 substrates, EsxA, EsxB, EspA, and EspC, was significantly reduced (36). Strain differences may thus explain this discrepancy in EspB secretion.

A further strain-dependent difference occurs in the EspB secretion pattern. While EspB is secreted mainly as a cleaved form by the M. tuberculosis Erdman strain (21) (see Fig. S4 in the supplemental material), both cleaved and full-length EspB proteins are secreted in the same proportions by M. tuberculosis H37Rv (Fig. S4). Previous studies have also reported differences in ESX-1 secretion between strains H37Rv and Erdman, as the Erdman strain was shown to secrete larger amounts of EsxA than the H37Rv strain. A polymorphism found in the promoter region of the ESX-1 transcriptional regulator WhiB6 was the underlying cause of this difference (37). Altogether, this indicates that the genetic background of the M. tuberculosis strains should be considered when analyzing ESX-1-related proteins, as it may impact the resulting data and our understanding of the secretion mechanism.

The analysis of the intracellular proteome also revealed the impact of EccE₁ on the stability of ESX-1 membrane proteins and, therefore, formation of the ESX-1 membrane complex (Fig. 3). It is plausible that in the absence of EccE₁, the EccB₁, EccCa₁, EccD₁, and EccCb₁ proteins may not properly assemble or fold and hence may become more vulnerable to degradation. In line with these results, Bosserman and colleagues (38) recently demonstrated that the lack of EccCb₁ in M. marinum led to reduced levels of various ESX-1 membrane proteins, including EccE₁. Together, these results support the hypothesis that loss of a single Ecc protein destabilizes the ESX-1 complex. In contrast, the absence of MycP₁ did not affect the intracellular levels of the ESX-1 membrane proteins, suggesting that the membrane complex is probably formed in the protease-deficient mutant. Consistent with this, previous work on M. marinum reported the formation of the ESX-1 core membrane complex in the absence of the MycP₁ mycosin (23), which is not itself an integral part of the ESX-1 apparatus but nonetheless is crucial for its integrity and function (23).

We confirmed that EccE₁, with its two N-terminal transmembrane regions (39), is indeed a membrane-anchored protein (Fig. 4) with its hydrophilic C-terminal part predicted to be in the periplasm. However, detection of the EccE₁-mNeon protein in the cell wall fraction as well (Fig. 4) suggests that EccE₁ also localizes there and thus behaves similarly to EccB₁, another core component of ESX-1 with two localizations. Indeed, EccB₁, a periplasmic protein with a single transmembrane region, was found in both the plasma membrane and cell wall fractions (40). These observations are consistent with the structure of the ESX-5 membrane complex, where the soluble domain of EccE₅ was predicted to be located in the periplasmic space (24).

Two previous studies demonstrated the presence of ESX-1 core components at a single cell pole in M. marinum and M. smegmatis using immunofluorescence and microscopy analysis of single cells overexpressing fluorescent fusion proteins (26, 27). In this study, we localized EccE₁-mNeon at both poles of M. tuberculosis, using low-level expression (Fig. 5A), but only when a functional ESX-1 apparatus was present. In its absence, the EccE₁-mNeon signal was diffuse and was observed throughout the cell (Fig. 5). These observations suggest that since EccE₁ is localized at the poles in M. tuberculosis, the ESX-1 apparatus may be there too. To test this, the EccE₁-mNeon protein constructed here may prove to be a useful tool for visualizing the nanomachine in situ by correlative light-electron microscopy.

MATERIALS AND METHODS

Bacterial culture conditions.

M. tuberculosis was routinely grown in 7H9 broth (supplemented with 0.2% glycerol, 10% albumin-dextrose-catalase [ADC], and 0.05% Tween 80) or on 7H10 agar (supplemented with 0.5% glycerol and 10% oleic acid-albumin-dextrose-catalase [OADC]). Escherichia coli TOP10 or chemically competent E. coli DH5α cells were used for cloning and plasmid propagation and were grown on LB broth or agar.

Mutant construction.

Deletion of the mycP₁-eccE₁ region was accomplished in the H37Rv strain by two-step homologous recombination using the pJG1100-derived vector (41). Two fragments of ∼900 bp corresponding to the up- and downstream regions of mycP₁-eccE₁ were PCR amplified and inserted in pJG1110 vector. The first recombination event was selected on 7H10 plates with hygromycin (50 μg/ml) and kanamycin (20 μg/ml). Positive colonies identified by PCR were grown in 7H9 medium with no antibiotics and subjected to a second selection on 7H10 plates supplemented with 2.5% sucrose. The absence of mycP₁-eccE₁ in the resulting clones was scored by PCR and further confirmed by Southern blotting.

DNA extraction and whole-genome sequencing.

The M. tuberculosis ΔmycP₁-eccE₁ mutant was grown in 7H9 broth to an optical density at 600 nm (OD₆₀₀) of 0.8. Bacteria were harvested by centrifugation, and DNA was extracted using the QIAamp UCP pathogen kit (Qiagen) as described previously (42). Illumina libraries were prepared using the Kapa Hyper prep kit as described previously (42) and quantified using the Qubit double-stranded DNA (dsDNA) BR assay kit (Thermo Fisher Scientific). Fragment size was assessed on a fragment analyzer (Advanced Analytical Technologies). Finally, libraries were multiplexed and sequenced as 100-base-long single-end reads on an Illumina HiSeq 2500 instrument. Reads were adapter and quality trimmed with Trimmomatic v0.33 (43) and mapped onto the M. tuberculosis H37Rv reference genome (RefSeq NC_000962.3) using Bowtie2 v2.2.5 (50).

Construction of complemented derivatives.

The integrative pGA44 vector (44) was used to construct all complemented derivates. All plasmids used in this study are listed in Table S3 in the supplemental material. Briefly, the genes of interest were PCR amplified from M. tuberculosis H37Rv genomic DNA (see Table S4 in the supplemental material) and cloned in frame under control of the PTR promoter. The monomeric yellow-green fluorescent protein mNeonGreen (32) and the HA tag sequences were cloned at the C-terminal part of eccE₁. All plasmids were checked by Sanger sequencing and were further transformed into competent M. tuberculosis ΔmycP₁-eccE₁ cells together with pGA80 (44), which provided the integrase. All strains, plasmids, and oligonucleotides used in this study are listed in Tables S3 and S4.

RNA sequencing.

M. tuberculosis strains were grown in 25 ml of Sauton’s medium without detergent to an OD₆₀₀ of ∼0.5. Cultures were then harvested by centrifugation, and the resulting pellets were resuspended in 1 ml TRIzol reagent (Thermo Fisher) and stored at –80°C until further processing. Bacteria were disrupted by bead beating, and total RNA was isolated as previously described (45). Two independent cultures for each strain were used for this experiment. Total RNA concentration was measured using the Qubit RNA HS assay kit. Library preparation and Illumina high-throughput sequencing and analysis were performed as described previously (36).

In vitro growth curves.

All strains were diluted to an initial OD₆₀₀ of 0.05 in Middlebrook 7H9 medium, and the OD₆₀₀ was recorded at different time points over a period of 7 days. Data from three independent experiments were used for growth representation.

Antibiotic susceptibility.

The MIC values were determined by using the resazurin-based microdilution assay (REMA) (28) as previously described (46). MIC values were determined by nonlinear fitting of the data to the Gompertz equation using GraphPad Prism.

Macrophage survival.

Human monocytic THP-1 cells were grown in RPMI medium supplemented with 10% fetal bovine serum (FBS) and 1% sodium pyruvate. Monocytes at 2 × 10⁶ cell/ml were differentiated into macrophages using phorbol 12-myristate-13-acetate (PMA) at a final concentration of 4 nM and subsequently pipetted at 10⁵ cells/well into a 96 well-plate. The plate was sealed and incubated at 37°C with 5% CO₂ for 18 h. Medium containing PMA was removed and replaced with new RPMI medium. Bacteria were grown to an OD₆₀₀ of 0.4 to 0.8, washed, resuspended in 7H9 broth to an OD₆₀₀ of 1 (3 × 10⁸ bacteria/ml), and diluted in RPMI medium at a concentration of 10⁷ bacteria/ml. THP-1 cells were then infected at a multiplicity of infection (MOI) of 5 and incubated for 2 or 3 days at 37°C with 5% CO₂. Afterwards, the PrestoBlue assay (Life Technologies) was used to evaluate cell viability according to the manufacturer’s instructions. Fluorescence was measured using a Tecan M200 instrument (excitation/emission wavelength of 560/590 nm). Fluorescence units of three biological replicates were analyzed in GraphPad Prism using one-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test.

Bacterial uptake by THP-1 macrophages.

THP-1 macrophages were infected with M. tuberculosis strains at an MOI of 5 and incubated at 37°C with 5% CO₂ in the presence of RPMI. At 4 h postinfection, extracellular bacteria were removed by washing twice with warm phosphate-buffered saline (PBS) and macrophages lysed with 0.1% Triton X-100 in PBS. Lysates were diluted, plated on 7H10 medium, and incubated for 3 weeks before the CFU were determined. Results from three independent replicates are represented.

Protein preparation, secretion analysis, and immunodetection.

Culture filtrates and cell lysates were prepared and immunoblotted as described previously (16, 35). GroEL2 was used as a loading control for cell lysates, and Ag85B, an ESX-1-independent secreted protein, was used as a loading control for culture filtrate. The following reagents were purchased from Abcam: monoclonal anti-ESAT6 (Ab26246), polyclonal anti-CFP10 (Ab45074), and polyclonal anti-Ag85B (ab43019). Polyclonal anti-EspB antibodies were produced by Ida Rosenkrands (Statens Serum Institut, Copenhagen, Denmark). The following reagent was obtained through BEI Resources, NIAID, NIH: monoclonal anti-M. tuberculosis GroEL2 (gene Rv0440), clone IT-56 (CBA1) (produced in vitro), NR-13655. Monoclonal anti-HA tag antibodies conjugated to horseradish peroxidase (HRP) (no. 2999) were purchased from Cell Signaling. Polyclonal anti-Rv3852 antibody was produced by Eurogentec (30).

Mass spectrometry-based analysis.

Proteins for secretion analysis (culture filtrates) were prepared as described above. To prepare the cell lysate, the cell pellet was washed once with PBS, resuspended in 100 mM Tris (pH 8)–2% SDS buffer with Roche protease inhibitor cocktail tablets, disrupted by sonication for 15 min at 4°C, clarified by centrifugation, and heat inactivated at 100°C for 1 h. The total protein concentration in all preparations was determined using bicinchoninic acid (BCA) assays. Mass spectrometry analysis was performed as described previously (36). Significant hits were determined by a volcano plot-based strategy, combining t-test P values with ratio information (47). Significance curves in the volcano plot corresponding to a small positive constant (SO) value of 0.5 and a permutation-based false-discovery rate (FDR) of 0.05 were determined by a permutation-based method. The FDR was determined after 250 random permutations of the data, allowing separation of biologically significant hits from those randomly obtained (47, 48).

Cell fractionation.

The M. tuberculosis ΔeccE₁/eccE₁HA mutant was grown in 100 ml of Sauton’s medium without Tween 80 for 4 days with a starting OD₆₀₀ of 0.5. Bacteria were harvested by centrifugation, and the fractions were collected as described previously (30).

Fluorescence microscopy.

Bacteria expressing mNeon were grown in 7H9 broth to log phase (OD₆₀₀ of 0.4 to 0.8) and then mounted on a microscope slide with 1.5% agarose pads and directly examined with an Olympus IX81 microscope under a 100× objective. Images were analyzed using ImageJ and normalized for brightness, contrast, and resolution.

Data availability.

The RNA-seq data were deposited at the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE119582) under accession number GSE119582. Raw data obtained from mass spectrometry experiments have been deposited to the ProteomeXchange Consortium via the PRIDE (49) partner repository (https://www.ebi.ac.uk/pride/archive/projects/PXD012584 ) with the data set identifier PXD012584.