Initiation and maintenance of infection by mycobacteria in susceptible hosts are not well understood. A screen of Mycobacterium marinum transposon mutant library led to isolation of eight mutants that failed to cause haemolysis, all of which had transposon insertions in genes homologous to a region between Rv3866 and Rv3881c in Mycobacterium tuberculosis, which encompasses RD1 (Rv3871–Rv3879c), a known virulence gene cluster. The M. marinum mutants showed decreased virulence in vivo and failed to secrete ESAT-6, like M. tuberculosis RD1 mutants. M. marinum mutants in genes homologous to Rv3866-Rv3868 also failed to accumulate intracellular ESAT-6, suggesting a possible role for those genes in synthesis or stability of the protein. These transposon mutants and an ESAT-6/CFP-10 deletion mutant all showed reduced cytolysis and cytotoxicity to macrophages and significantly decreased intracellular growth at late stages of the infection only when the cells were infected at low multiplicity of infection, suggesting a defect in spreading. Direct evidence for cell-to-cell spread by wild-type M. marinum was obtained by microscopic detection in macrophage and epithelial monolayers, but the mutants all were defective in this assay. Expression of M. tuberculosis homologues complemented the corresponding M. marinum mutants, emphasizing the functional similarities between M. tuberculosis and M. marinum genes in this region that we designate extRD1 (extended RD1). We suggest that diminished membranolytic activity and defective spreading is a mechanism for the attenuation of the extRD1 mutants. These results extend recent findings on the genomic boundaries and functions of M. tuberculosis RD1 and establish a molecular cellular basis for the role that extRD1 plays in mycobacterial virulence. Disruption of the M. marinum homologue of Rv3881c, not previously implicated in virulence, led to a much more attenuated phenotype in macrophages and in vivo, suggesting that this gene plays additional roles in M. marinum survival in the host.
Despite more than a century of research on prevention and treatment, tuberculosis remains one of the world's most significant infectious diseases, with more than 2 million deaths each year (Dolin et al., 1994). The availability of the complete genome sequence of Mycobacterium tuberculosis (Cole et al., 1998) and of methods for forward genetics (Bardarov et al., 1997; Pelicic et al., 1997; Camacho et al., 1999; Cox et al., 1999; Rubin et al., 1999) has led to important advances in understanding the pathogenesis of tuberculosis. Comparative genomics of M. tuberculosis, Mycobacterium bovis, Mycobacterium leprae, the vaccine strain M. bovis BCG (BCG) and Mycobacterium microti has revealed several chromosomal regions that are absent from the genomes of the less virulent organisms (Mahairas et al., 1996; Behr et al., 1999; Gordon et al., 1999; Brodin et al., 2002). This analysis has focused particular attention on one region, termed region of difference 1 (RD1), which encompasses nine genes (from Rv3871–Rv3879c), because it is missing from the less virulent BCG (Mahairas et al., 1996; Behr et al., 1999; Gordon et al., 1999) (see Fig. 1D). In addition, an overlapping deletion has been found in M. microti that involves 13 ORFs (Rv3864–Rv3876) when compared to the M. tuberculosis genome (Brodin et al., 2002) (see Fig. 1D). The overlap between RD1 and the M. microti deletion includes CFP-10 (Rv3874) and ESAT-6 (Rv3875), whose secreted protein products are prominent T and B cell antigens in patients infected with M. tuberculosis. An important role for RD1 in mycobacterial pathogenesis has been demonstrated experimentally in several ways. Incorporation of M. tuberculosis genes extending through the M. microti and BCG deletions into the BCG or M. microti chromosome enhances virulence of these organisms (Pym et al., 2002); conversely, a deletion of RD1 genes from M. tuberculosis greatly decreases its capacity to cause disease (Lewis et al., 2003). Recently, by using signature-tagged mutagenesis (Stanley et al., 2003) and transposon site hybridization (Sassetti and Rubin, 2003) screens and targeted gene deletions (Hsu et al., 2003; Guinn et al., 2004), several individual RD1 genes of M. tuberculosis have been shown to be essential for virulence in mice. One major clue about the role of RD1 genes has come from recent discoveries that demonstrate these genes may constitute an alternative secretion system required for ESAT-6 and CFP-10 secretion (Hsu et al., 2003; Pym et al., 2003; Stanley et al., 2003; Guinn et al., 2004). ESAT-6 itself enhances permeability of artificial membranes, suggesting a possible direct role in cytolysis/cytotoxicity (Hsu et al., 2003). These data have led to the hypothesis that ESAT-6 and CFP-10, which can form a 1:1 complex (Renshaw et al., 2002), may be effectors of the enhanced pathogenicity of the mycobacteria that contain RD1. However, how exactly ESAT-6 and CFP-10 potentiate mycobacterial virulence at the molecular and cellular level is not fully understood. Moreover, the functions of M. tuberculosis genes missing in M. microti that do not overlap with RD1 (Fig. 1D) are not understood. As a result, the virulence function of the genetic region containing RD1 remains an area of intense investigation.
Mycobacterium marinum causes a systemic tuberculosis-like disease in fish and frogs, its natural hosts, and a localized disease in immunocompetent humans. Both the human and animal infections are marked by the presence of a granulomatous host response, a hallmark of diseases caused by M. tuberculosis (Ramakrishnan et al., 1997; Talaat et al., 1998; Prouty et al., 2003). M. marinum is genetically closely related to M. tuberculosis (Tonjum et al., 1998) and has been used increasingly as a model for understanding the pathogenesis of tuberculosis (Ramakrishnan et al., 1997; 2000; Talaat et al., 1998; Davis et al., 2002; Gao et al., 2003a,b; Prouty et al., 2003). Like M. tuberculosis, M. marinum grows and persists in host macrophages in vivo and in vitro. Infection by either leads to lysis of host cells, suggesting the existence of a membranolytic activity in both species that may be important in dissemination of infection (Ramakrishnan and Falkow, 1994; Valdivia et al., 1996; Bouley et al., 2001; Gao et al., 2003a). Although M. tuberculosis has been shown convincingly to reside exclusively in macrophage phagosome (Russell, 2001; Clemens et al., 2002), M. marinum can escape from the phagosome into the cytosol and polymerize actin (Stamm et al., 2003), suggesting another potential role for membranolytic activity in this organism. In the present study we have used transposon mutagenesis (Gao et al., 2003a) to understand the haemolytic activity that is characteristic of pathogenic mycobacteria including M. tuberculosis and M. marinum and that has been associated with virulence (Rudnicka et al., 1999). Haemolysis may also be related to mycobacteria-mediated cytolysis of nucleated cells, which also has been suggested to have a role in pathogenesis of tuberculosis (Maslow et al., 1999). By screening an M. marinum transposon mutant library, we identified eight mutants with abolished haemolytic activity and found that all of the mutants had transposon insertions in M. marinum genes homologous to M. tuberculosis RD1 and surrounding genes, a gene cluster that we refer to as extended RD1 (extRD1). These mutants and others we created by homologous recombination demonstrate that haemolytic, cytolytic and cytotoxic activities of M. marinum are tightly linked, require multiple individual genes of extRD1 and correlate with secretion of ESAT-6. Like BCG, M. microti and M. tuberculosis strains missing genes in RD1, M. marinum mutants with mutations in individual extRD1 genes had significantly decreased virulence in vivo. Furthermore, these mutants showed diminished spreading in macrophage and epithelial cell monolayers and markedly reduced dissemination of infection in vivo. Because bacterial spreading is a characteristic linked to pathogenicity for several intracellular pathogens including Listeria and Shigella (Goldberg, 2001), we propose that this defect may account for the decreased virulence of the M. marinum extRD1 mutants. Thus, this work demonstrates that genes in the extRD1 region are required for mycobacterial expression of a membranolytic activity necessary for efficient bacterial spreading among host cells. These results extend the recent knowledge of the genomic boundaries and functions of RD1 and establish a cellular basis for the role that extRD1 plays in mycobacterial virulence.
A screen for M. marinum haemolysis mutants identifies RD1 and surrounding genes required for haemolysis
We screened mutants from an M. marinum transposon mutant library for diminished contact-dependent haemolysis of sheep red blood cells (sRBC). Wild type (WT) M. marinum was haemolytic to sRBC, which could be detected after 1 h coincubation of the bacteria with sRBC at a ratio as low as 1:1, and haemolysis was augmented at increasing ratios of bacteria to sRBC (Fig. 1A). Human erythrocytes also were lysed efficiently by M. marinum (data not shown). We identified eight mutants (M1–M8) that reproducibly showed markedly reduced haemolytic activity, four of which (M2, M3, M4 and M8) were completely non-haemolytic (Fig. 1B). We sequenced chromosomal DNA flanking individual transposon insertions in those mutants and found that all of the insertions were clustered in a genomic region containing ORFs homologous to M. tuberculosis RD1 genes (Fig. 1C and D). Alignment of an M. marinum assembled contig sequence (mar478c10.p1k, http://www.sanger.ac.uk/Projects/M_marinum/) to this region of the M. tuberculosis chromosome revealed that both the ORF sequences and their genomic organization are highly conserved between the two mycobacterial species (Fig. 1C and D). For this reason, we have used the prefix Mh to denote the M. marinum homologue of each M. tuberculosis gene, based on the sequence of H37Rv (Cole et al., 1998). Four of the transposon insertions were located in three individual M. marinum ORFs (Mh3866–Mh3868) in a region homologous to M. tuberculosis genes missing in M. microti; another three insertions were in M. marinum ORFs (Mh3876, Mh3878 and Mh3879c) homologous to M. tuberculosis RD1. One insertion was in a nearby ORF (Mh3881c) homologous to Rv3881c. The insertions in Mh3866, Mh3867, Mh3868 and Mh3881c were all beyond the so-far-defined boundaries of M. tuberculosis RD1 but produced similar abolition of the haemolytic phenotype. These results indicate that production of haemolytic activity requires multiple genes on the mycobacterial chromosome that include RD1 and surrounding genes, and we refer to this entire gene cluster (Mh3866–Mh3881c) as the extended RD1 (extRD1, Fig. 1C and D). BCG, which lacks RD1, lacks haemolytic activity (King et al., 1993), suggesting that RD1 is important in haemolysis by M. bovis and probably other mycobacteria as well.
M. marinum extRD1 mutants are complemented by expression of homologous genes from M. tuberculosis
We expressed M. tuberculosis homologues in four of the M. marinum extRD1 mutants to determine whether these could complement the mutants’ defect in haemolysis. As shown in Fig. 1B, expression of Rv3866, Rv3879c and Rv3881c genes, respectively, in M1 (Mh3866::Tn), M7 (Mh3879c::Tn) and M8 (Mh3881c::Tn) resulted in a complete recovery of haemolysis, and there was no significant difference in complementation of M8 by Rv3881c compared to the Mh3880c–Mh3881c operon from M. marinum. Complementation was specific to expression of the appropriate M. tuberculosis gene, for neither an empty plasmid nor a plasmid expressing a non-corresponding gene (Rv3866) restored haemolysis by Mh3879c::Tn (Fig. 1B). Therefore, extRD1 genes are conserved between M. marinum and M. tuberculosis not only for structure but also for function, further evidence for involvement of this region in the known haemolytic activity of other pathogenic mycobacteria (King et al., 1993).
extRD1 genes are required for rapid cytolysis to macrophages
To explore potential biological functions for the haemolytic activity during mycobacterial infection, we first examined whether haemolytic activity of M. marinum was related to cytolysis to macrophages. J774 murine macrophage-like cells were incubated with M. marinum strains, and cytolysis was detected by permeabilization of the plasma membrane to ethidium homodimer-1. As shown in Fig. 2A and E, WT M. marinum at a bacterium:cell ratio of 50:1 produced massive cytolysis to J774 cells within 1 h of incubation. Cytolysis to J774 cells was detected at a bacterium:cell ratio as low as 3 and augmented at increasing ratios. In contrast, none of the extRD1 mutants exhibited significant cytolysis to J774 cells (Fig. 2B, C and E), even at a bacterium:cell ratio up to 200. Cytolysis by M1, M7 and M8 was fully complemented by expression of Rv3866, Rv3879c and Rv3881c (or Mh3880c–Mh3881c) genes respectively (Fig. 2C–E). Thus, the extRD1 genes required for haemolysis are also necessary for rapid cytolysis of macrophages, suggesting a tight genetic linkage between the two effects.
extRD1 genes are required for cytotoxicity to macrophages and epithelial cells following intracellular infection
To determine whether the RD1-dependent haemolytic/cytolytic activity is related to cytotoxicity that occurs during mycobacterial infection, we infected J774 cells with M. marinum WT and extRD1 mutants at a low multiplicity of infection (moi) of 0.1 (≈ 104 bacteria internalized by 105 cells). Cytotoxicity to the cells was examined by use of ALAMAR BLUE to measure the cell metabolic activity. As shown in Fig. 2F, WT bacteria produced massive cytotoxicity to J774 cells by 5 days after infection, with a loss of 92% of the metabolic activity of the cell monolayer. In contrast, all extRD1 mutants showed markedly reduced cytotoxicity to J774 cells. The cytotoxic defects of M1 and M7 were fully restored by expression of corresponding M. tuberculosis homologues (Fig. 2F). Diminished cytotoxicity by the M. marinum extRD1 mutants and complementation by expression of M. tuberculosis homologues were also observed during infections of murine bone marrow-derived macrophages (BMDMs) and human A549 epithelial-like cells (data not shown). These results demonstrate that extRD1 genes are required for cytotoxicity to macrophages and epithelial cells following intracellular infection that is initiated at a low inoculum. In addition, the close correlation between cytotoxicity and haemolysis/cytolysis argues that cytotoxicity resulting from intracellular growth of mycobacteria is probably related in mechanism to the rapid haemolytic and cytolytic activities that occur after addition of extracellular bacteria.
extRD1 genes are required for secretion of ESAT-6 and CFP-10
Genes in M. tuberculosis RD1 are required for secretion of ESAT-6 and CFP-10 (Hsu et al., 2003; Pym et al., 2003; Stanley et al., 2003; Guinn et al., 2004). To assess whether this was true for M. marinum as well, we examined secretion of these molecules in the mutants we had identified. As shown in Fig. 3A, WT M. marinum secreted a significant amount of ESAT-6 into the culture medium (≈ 5 ng protein/3 × 108 bacteria), which was approximately two times more than in the bacterial cell lysate. GroEL1, a cytosolic protein that has been used as a marker for bacteriolysis when present in the culture medium (Sonnenberg and Belisle, 1997; Stanley et al., 2003), was detected solely in the cell lysates of both WT and extRD1 mutants (Fig. 3A), demonstrating that ESAT-6 in the supernatant was secreted from WT M. marinum and not simply a result of cell lysis. In contrast to WT M. marinum, none of the extRD1 mutants secreted a detectable amount of ESAT-6 (Fig. 3A). Lack of ESAT-6 in the culture filtrate was a specific defect in secretion for several of the mutants (M5–M8), because the protein was detected in cell lysates in an amount similar to WT (Fig. 3A). Interestingly, mutants M1–M4 (Mh3866::Tn, Mh3867::Tn and Mh3868::Tn) had markedly reduced intracellular accumulation of ESAT-6, suggesting a role for these genes in either protein synthesis or protein stability (Fig. 3A).
Similar to ESAT-6, CFP-10 was secreted into the culture medium by WT M. marinum, which was approximately two to three times more than in the cell lysate (Fig. 3A). The secretion of CFP-10 was markedly reduced or abolished for several of the mutants (M1–M4 and M8), but it was only modestly decreased for the other mutants (M5–M7) (Fig. 3A). Therefore, secretion of CPF-10 and ESAT-6 are not tightly coupled, suggesting that they may not be secreted in a fixed 1:1 ratio. All extRD1 mutants produced CFP-10 in the cell extracts (Fig. 3A), although its production in M2–M4 was substantially reduced. These results suggest that failure of accumulation of CFP-10 in the culture filtrates from the non-haemolytic mutants probably resulted from failure of secretion, and that association with ESAT-6 is not necessary for stable production of the protein in the bacterial cells. The defects in secretion of ESAT-6 and CFP-10 were fully restored by expression of M. tuberculosis homologues in M1, M7 (Fig. 3A) and M8 (Fig. 3B).
To determine the specificity of the secretion defect in extRD1 mutants, we examined secretion of fibronectin attachment protein (FAP) and the antigen-85 complex (Ag85), both of which are well-characterized secreted proteins with signal sequences. There was no significant difference between M. marinum WT and extRD1 mutants in either the production or secretion of these proteins (Fig. 3A). Furthermore, there were no detectable differences between WT and the mutants in cell wall lipids, including mycolic acids, polar and apolar lipids (data not shown). Together, these results demonstrate that multiple individual genes in extRD1 are required for secretion of ESAT-6 and CFP-10. These data are consistent with the hypothesis (Gey Van Pittius et al., 2001; Pallen, 2002) and recent reports on M. tuberculosis (Hsu et al., 2003; Pym et al., 2003; Stanley et al., 2003; Guinn et al., 2004) that RD1 genes may encode a multicomponent secretion system, with at least ESAT-6 and CFP-10, both lacking signal peptides, as targets for secretion.
Secretion of ESAT-6 is essential for haemolysis, cytolysis and cytotoxicity
ESAT-6 and CFP-10 are targets of the immune response in tuberculosis and have been implicated in the pathophysiology of infection (Wards et al., 2000; Pym et al., 2002; 2003). Because both are secreted proteins that may act as effectors of RD1, we examined their role in haemolysis, cytolysis and cytotoxicity, which we have shown to require extRD1 genes. We created an M. marinum mutant deleted for both ESAT-6 and CFP-10 genes (ΔCE) by homologous recombination and confirmed that the mutant produced no ESAT-6 protein in either the bacterial cell or the culture medium (Fig. 3A). Compared to WT M. marinum, ΔCE was extremely defective in haemolysis (Fig. 1B), cytolysis (Fig. 2E) and cytotoxicity (Fig. 2F), similar to other extRD1 mutants. Expression of M. tuberculosis ESAT-6/CFP-10 genes completely restored the ability for ΔCE to produce and secrete both proteins (Fig. 3A) and to produce haemolysis, cytolysis and cytotoxicity (Figs 1B, 2E and F). Because ESAT-6 was not secreted by any of the extRD1 mutants, but CFP-10 was secreted at least to some extent by several of the mutants, ESAT-6 secretion is more closely associated with the effector functions of extRD1 genes.
extRD1 genes are required for virulence to zebrafish
Mycobacterium tuberculosis RD1 has been shown to be required for full virulence in the mouse infection model (Pym et al., 2002; Hsu et al., 2003; Lewis et al., 2003; Stanley et al., 2003; Guinn et al., 2004). A number of fish species are natural hosts for M. marinum, and zebrafish has recently been established as a relevant laboratory model that manifests both acute disseminated disease and chronic persistent infection (Davis et al., 2002; Prouty et al., 2003). To determine whether RD1 is required for M. marinum virulence in vivo, we infected zebrafish intraperitoneally (i.p.) with M. marinum WT and three of the extRD1 mutants and examined severity of infection by monitoring fish survival, mycobacterial growth in vivo and histopathology. At 105 CFU per fish, WT-infected fish all died 4–5 weeks after infection (Fig. 4A). In contrast, even after 8 weeks, none of the fish infected by this dose of M8 (Mh3881c::Tn) died, and only one of four fish infected by M3 (Mh3868::Tn) died (Fig. 4A). At a lower inoculum (103 CFU per fish), none of the fish died following infection by either WT or the mutants for the entire 9 weeks of the experiment. With respect to M. marinum growth in vivo, at 105 CFU per fish, WT CFU increased about 1000-fold in fish livers during the first 3 weeks of infection, after which the fish began to die (Fig. 4B). In contrast, CFUs for M3 and ΔCE showed a much slower increase during the first 3 weeks (≈ 1% of WT) and simply sustained or declined slightly thereafter (Fig. 4B). The extent of this growth defect is similar to that reported for RD1-deleted M. tuberculosis in mice (Lewis et al., 2003; Guinn et al., 2004); thus the functional homology between M. tuberculosis and M. marinum extRD1 includes effects on in vivo virulence as well as secretion of ESAT-6 and CFP-10. Mutation of the Mh3881c gene, not a part of RD1, caused the most severe attenuation in vivo, with minimal growth during the first week and a decrease in CFU thereafter (Fig. 4B).
Histopathological studies of infected fish supported attenuation of the mutants. At 3 weeks after infection, fish infected by 105 WT M. marinum developed massive necrotic damage of the livers; large numbers of mycobacteria were observed in damaged tissues, and granulomas were rarely detected (Fig. 5A and B). In contrast, no significant tissue damage was observed in fish infected by M3, M8 or ΔCE; instead, granulomas containing few mycobacteria were observed (data not shown). After inoculation with 103 CFU per fish, fish infected by both WT and the mutants developed granulomas, but significantly fewer granulomas and fewer mycobacteria per granuloma were detected in the livers of fish infected by M3 (Fig. 5E and F) compared to WT (Fig. 5C and D). No granulomas were detected in the livers of fish infected by M8; a few granulomas were seen in tissues of the peritoneal space, which was the original site of the infection (Fig. 5G and H). Together, these results indicate that extRD1 genes affect dissemination of infection by M. marinum in zebrafish and suggest that Mh3881c may be particularly important for growth in vivo.
extRD1 genes potentiate M. marinum growth in macrophage culture
Macrophages are primary sites for intracellular growth of mycobacteria (Russell, 2001). Because extRD1 genes are required for full virulence of M. marinum to fish, we examined whether they are also required for M. marinum growth in macrophages. When BMDMs were infected by WT M. marinum at moi 0.01 (≈ 103 CFU internalized by 105 cells), the bacteria replicated extensively in these cells and the CFU increased by ≈ 500-fold over the 6 days of intracellular infection (Fig. 6A). In contrast, although most of the extRD1 mutants showed initial growth similar or close to WT, they had markedly reduced growth at later time points, and by 6 days after infection their CFUs were ≈ 2-logs less than WT (Fig. 6A). M8 was the most attenuated, with a considerable loss of viability soon after infection in BMDMs and complete sterilization by 6 days (Fig. 6A). The growth defects of M1, M7 and M8 were restored by complementation with the appropriate M. tuberculosis genes (Fig. 6B). There were no significant differences between M. marinum WT and extRD1 mutants for growth in the bacterial culture medium, except that ΔCE showed slightly delayed exit from the initial lag phase of culture (Fig. S1).
extRD1 genes are required for M. marinum spreading in cell culture
Because (i) the majority of M. marinum extRD1 mutants showed an intracellular growth defect that was most significant at late time points of infection and (ii) these mutants lacked cytolysis and cytotoxicity, we hypothesized that they might exhibit defective spreading in cell culture. To test this possibility, we directly observed M. marinum spreading in confluent BMDM monolayers infected with M. marinum strains expressing green fluorescent protein (GFP) at exceedingly low moi 0.001 (103 CFU internalized by 106 cells). As illustrated in Fig. 7A–C, WT M. marinum developed fluorescent foci of infection (FFI) that increased in size with the time-course of the infection. FFI were detectable by 48 h as individual macrophages filled with replicating GFP-expressing bacteria. M. marinum spreading was apparent by 72 h (Fig. 7A), when FFI contained 4–10 cells filled with the bacteria, indicating M. marinum spreading into neighbouring cells. At 120 h, each FFI contained 20–40 cells (Fig. 7B and C), demonstrating further M. marinum spreading. When observed at a higher magnification, single WT bacteria were detected in cells surrounding the heavily infected cells that formed the FFI (Fig. 7G and H). In contrast, all extRD1 mutants lacked the ability to spread from the initially infected cells, although they (except M8) achieved a high bacterial density in these cells, as shown by the intense GFP fluorescence throughout the cytoplasm of individual infected cells (Fig. 7D–F, I and J). M8 was unable to form even the single cell-based FFI (Fig. S2). The spreading defects of M1, M7, M8 and ΔCE were fully complemented by expressing corresponding M. tuberculosis or M. marinum genes (Fig. S2).
Mycobacterium marinum spreading was also observed for WT, but not the extRD1 mutants, in A549 lung epithelial cell monolayers, even in the presence of 20 µg ml−1 amikacin, a concentration sufficient to kill > 95% of extracellular WT M. marinum in 4 h (Fig. 8). Compared to cultures without amikacin, the presence of 20 µg ml−1 of the antibiotic considerably reduced but did not eliminate formation of FFI by WT M. marinum (Fig. 8A, B, D and E). Formation of multicellular FFI with no apparent cell lysis and in presence of amikacin (Fig. 8B and E) suggests that M. marinum is capable of direct cell-to-cell spread without significant exposure to the extracellular milieu. Because the absence of amikacin led to more robust spreading accompanied by significant cell lysis (Fig. 8A and D), it is possible that spreading of bacteria to uninfected cells through lysis of primarily infected cells with subsequent uptake of bacteria by neighbouring cells also occurred. None of the M. marinum extRD1 mutants were able to spread in A549 cell monolayer even in the absence of the antibiotic, and the defects of M1, M7, M8 and ΔCE were abolished after expression of the corresponding M. tuberculosis or M. marinum genes (Figs 8 and S3). The defects of the mutant strains for spreading in A549 cells was confirmed in a plaque assay, in which the mutants’ defects were again complemented by expression of the corresponding M. tuberculosis or M. marinum genes (Fig. 9). Together, these data demonstrate that extRD1 genes are required for mycobacteria cell-to-cell spread during infection of macrophages and lung epithelial cells.
The possibility that genes within the RD1 region in M. tuberculosis and M. bovis contribute to virulence first received attention when it was realized that all strains of BCG, the attenuated M. bovis used for vaccination against tuberculosis, have a deletion in this region of the chromosome (Mahairas et al., 1996; Behr et al., 1999; Gordon et al., 1999). An important role for RD1 in pathogenesis was demonstrated directly by enhancement of the virulence of BCG and M. microti with expression of this region of the M. tuberculosis chromosome and by reduction of the virulence of M. tuberculosis with its deletion. Very recently, several groups have reported that M. tuberculosis RD1 genes other than ESAT-6 and CFP-10 are required for secretion of these two immunodominant antigens and for virulence in mice (Hsu et al., 2003; Pym et al., 2003; Stanley et al., 2003; Guinn et al., 2004). Numerous questions remain about how this region affects virulence: for example, which of the genes in this region are required for, and how do they participate in, mycobacterial infection of host cells; which of the genes are involved in secretion of ESAT-6 and CFP-10; and whether and how the secretion of ESAT-6 and CFP-10 potentiates the virulence of mycobacteria. This study demonstrates that multiple genes in the M. marinum chromosomal region considerably larger than RD1 (extRD1) spanning Mh3866–Mh3881c are required for virulence both in vivo and in vitro and for secretion of ESAT-6 and CFP-10. The extRD1 genes are required for disseminated M. marinum infection in vivo and cell-to-cell spread in vitro in cultured macrophages and epithelial cells, both of which are genetically associated with the secretion of ESAT-6. We propose that extRD1 genes potentiate virulence by contributing to secretion of a membranolytic activity that involves ESAT-6 and that this activity is required for mycobacteria cell-to-cell spread in the host. The ability of mycobacteria to spread from cell-to-cell probably facilitates dissemination of infection among macrophages and dendritic cells and may be involved in early events in infection when inhaled or ingested bacilli come in contact with both epithelial and immune cells.
Spreading of mycobacteria in cultured cell monolayers has been observed previously for M. tuberculosis, in which two potentially different modes of cell-to-cell spread are described (Byrd et al., 1998; Castro-Garza et al., 2002). In one case, addition of extracellular antibiotic did not affect spreading, suggesting that the spreading bacteria were not exposed to the extracellular milieu sufficient for antibiotic effect, and in the other, addition of antibiotic abolished spread, suggesting a requirement for release into the extracellular medium before uptake by uninfected host cells for spreading. In our study, continuous incubation with amikacin at 20 times the MIC for M. marinum only moderately reduced the size of FFI compared to the absence of antibiotic (Fig. 8). This suggests that the bacteria are not exposed to extracellular milieu long enough to be killed by amikacin before rapid uptake into uninfected cells and may suggest that spreading in the presence of amikacin is predominantly a result of direct cell-to-cell spread similar to Listeria or Shigella. Our recent demonstration that M. marinum can escape the phagosome and polymerize actin (Stamm et al., 2003) suggests a cellular mechanism for this direct cell-to-cell spread. While this may not be directly relevant to cell-to-cell spread by M. tuberculosis, which does not escape the phagosome, it certainly is intriguing that M. tuberculosis RD1 genes are thought to control host cell cytolysis (Hsu et al., 2003) and cell-to-cell spread (Guinn et al., 2004). The otherwise very close parallels between M. tuberculosis and M. marinum in organization, structure and function of the genes in extRD1 raise the intriguing question of why M. tuberculosis does not lyse phagosome membranes. The answer to this question will probably require a more detailed molecular understanding of the mechanisms of membranolysis for the two closely related organisms.
Our study confirms recent work in M. tuberculosis that has demonstrated that individual genes near those encoding ESAT-6 and CFP-10 are required for their secretion and that these genes and products are required for full virulence in macrophages and in the mouse model of infection (Hsu et al., 2003; Sassetti and Rubin, 2003; Stanley et al., 2003; Guinn et al., 2004). Our work extends these data in several important ways. First, we show that the chromosomal region (Mh3866–Mh3881c) required for secretion of ESAT-6 and CFP-10 extends beyond the known boundaries of M. tuberculosis RD1 (Rv3871–Rv3879c). Mutations in M. marinum genes on both sides of the M. tuberculosis RD1 homologues led to loss of ESAT-6 and CFP-10 secretion. Thus, the putative secretion system is probably more complex than previously considered (Pallen, 2002; Hsu et al., 2003; Stanley et al., 2003; Guinn et al., 2004). Moreover, we demonstrate that in addition to the other recently characterized RD1 genes (Hsu et al., 2003; Stanley et al., 2003; Guinn et al., 2004), another two genes in RD1, Mh3878 and Mh3879c, also are required for secretion of ESAT-6. Second, we found discordance between the secretion of ESAT-6 and CFP-10 upon mutation of several extRD1 genes, including Mh3876, Mh3878 and Mh3879c. This demonstrates that the complex between these two molecules is not required for secretion of CFP-10. However, this does not exclude the possibility that the complex may facilitate ESAT-6 secretion. Our data show a very close linkage between failure to secrete ESAT-6 (but not CFP-10) and loss of virulence and are compatible with the suggestion that ESAT-6 is required for cytolysis (Hsu et al., 2003) and that this requirement is reflected in the haemolysis and cytotoxicity assays and in cell-to-cell spread of infection. Our data do not rule out the possibility that CFP-10 is necessary but not sufficient to produce these phenotypes nor do they exclude the possibility that as yet unidentified effectors of RD-1 are required for host membrane lysis and/or virulence. Third, we demonstrate that the extRD1 genes are differentially involved in the intracellular accumulation and secretion of ESAT-6. extRD1 mutants M1–M4, disrupted for genes Mh3866–Mh3868 upstream of RD1, failed to accumulate ESAT-6 in the bacterial cell (Fig. 3A), which cannot be a simple secondary consequence of loss of secretion, because intracellular ESAT-6 did accumulate in other mutants that failed to secrete the protein. This is not a result of a polar effect of the transposon insertions on the ESAT-6 gene, because ESAT-6 secretion was restored to the Mh3866 mutant by expression of Rv3866 alone. These genes therefore may be involved in stabilization of ESAT-6 in the bacteria before secretion and may be chaperones for this secretion system. Alternatively, these genes may be involved in transcription or translation of ESAT-6. Fourth, we have genetically linked the phenotypes of haemolysis, nucleated cell cytolysis and cytotoxicity to each other, to secretion of ESAT-6, to mycobacteria cell-to-cell spread and to virulence. We believe that the previously reported defects for BCG in haemolysis, cytolysis and cell-to-cell spread (King et al., 1993; Dobos et al., 2000; Castro-Garza et al., 2002) result from the deletion of RD1 genes. Fifth, we have provided direct evidence for mycobacteria cell-to-cell spread that involves the function of extRD1 genes, by microscopic detection of FFI and examination of plaque formation. The recent studies by Guinn et al. (2004) and Hsu et al. (2003) have suggested a role for M. tuberculosis RD1 genes in cell-to-cell spread by evaluation of the percentage of cells infected and the average CFU per cell, at different times after infection. While these conclusions were drawn based on statistical analysis of the entire culture monolayer, our visualization of these events adds considerable support to the inferences of these previous studies. Our assay for FFI is particularly useful, for it allows dynamic observation of individual spreading events at multiple times after infection. Finally, we demonstrate that M. marinum extRD1 genes are functionally conserved with their M. tuberculosis homologues both for biochemical and virulence functions in the bacterium's natural host, zebrafish.
Some of our newly identified extRD1 genes are predicted to encode proteins with possible roles in protein synthesis or secretion. For example, Mh3868 or Rv3868 is predicted to be a member of the AAA-ATPase protein family that carries out diverse cellular functions (Lupas and Martin, 2002). A common cheperone-like activity underlies this functional diversity, which involves structural remodelling, unfolding and disassembly of proteins and protein complexes (Lupas and Martin, 2002). Because there is no transmembrane segment predicted by the sequence of Mh3868 or Rv3868, the encoded protein may serve as a cytosolic chaperone to facilitate protein translocation across the cytoplasmic membrane, perhaps functionally interacting with the other two AAA-ATPases encoded in RD1 that have been shown to be required for secretion of ESAT-6 and CFP-10 (Stanley et al., 2003). Consistent with this hypothesis, Mh3866 (interrupted in M1) and its homologue Rv3866 are predicted to encode a periplasmic substrate-binding protein for high affinity transport of bacterial proteins across the cytoplasmic membrane. Alternatively, Mh3868 or Rv3868 could be involved in protein synthesis and/or stability, because its absence resulted in decreased intracellular accumulation as well as secretion of ESAT-6. Consistent with a possible role for these genes in protein synthesis at some step, Mh3867 is predicted to contain a RNA-binding domain. Clearly more detailed biochemical analysis of the products of these genes will be required to determine precisely how they are involved in the putative multicomponent secretion system that also requires multiple genes within RD1.
Finally, we demonstrate that M8 (Mh3881c::Tn) is far more attenuated in vivo and in macrophages than any other extRD1 mutant, including ΔCE. This does not result from a polar mutation of downstream genes, because M8 is complemented by expression of either Rv3881c or Mh3881c. This suggests that the Mh3881c/Mh3880c operon gene products perform additional function(s) required for mycobacterial survival during infection in addition to the secretion of ESAT-6 and CFP-10. However, we have been unable to find structural motifs in the predicted protein that suggest the nature of these additional roles or why Mh3881c is required for ESAT-6 and CFP-10 secretion. Further study of this operon to understand function and potential importance as a drug or vaccine target is clearly warranted.
Bacterial strains and media
Mycobacterium marinum strain M and the DH5α strain of E. coli were cultured and genetically manipulated as described (Gao et al., 2003a,b). Unless specified, M. marinum strains were cultured in 7H9 liquid with Tween 80 and ADC supplement or 7H10 agar with OADC supplement.
Isolation of M. marinum mutants defective in contact-dependent haemolysis
An M. marinum transposon mutant library was generated by the M4 (mariner transposon mutagenesis in M.marinum) procedure as previously described (Gao et al., 2003a). Briefly, the mutagenesis vector harbouring the mariner transposon was transformed into M. marinum WT and the mutants selected by a two-step procedure: first for single cross-over by kanamycin resistance, and second for double cross-over by resistance to sucrose and kanamycin. Approximately 1000 individual mutants were screened for haemolysis to sRBC. Contact-dependent haemolysis was assayed as previously described (King et al., 1993; Gao et al., 1999; Alli et al., 2000) with modifications. Briefly, M. marinum grown in 7H9 medium to mid- to late-log phase were washed twice with phosphate buffered saline (PBS) and mixed with sRBC (Remel). For the initial mutant screen, 120 µl of M. marinum suspension (containing 5 × 108 bacteria) was mixed with 120 µl of sRBC (containing 2 × 107 cells) in each well of the round-bottomed 96-well plates and centrifuged at 2600 g for 10 min. After incubation at 32°C, 5% CO2 for 1–2 h, the pellets were resuspended, centrifuged, and absorbance of the supernatant was measured at 405 nm. Defibrinated human RBCs (Jain, 1989) were similarly tested for haemolysis by M. marinum strains.
Analysis of rapid cytolytic activity of M. marinum to macrophages
The murine J774 macrophage-like cells (ATCC TIB67) were maintained as described (Gao et al., 2003a). M. marinum strains grown to mid- to late-log phase and washed twice with DME medium were added to the cell monolayers, centrifuged at 1500 g for 10 min and incubated at 32°C, 5% CO2 for 1 h. Cytolysis was examined at the end of the incubation by fluorescence microscopy after staining the cells for 40 min with 2 µM calcein AM and 4 µM ethidium homodimer-1 (Molecular Probes). Calcein AM stains live cells with green fluorescence after being metabolized by the cells, while ethidium homodimer-1 penetrates through permeabilized plasmic membrane of dead cells to stain the nuclei with red fluorescence.
Analysis of M. marinum intracellular growth and cytotoxicity
Mouse BMDMs and A549 human alveolar epithelial-like cells were cultured as previously described (Gao and Abu Kwaik, 1999; Gao et al., 2003b). Infection of BMDMs, J774 and A549 cells and enumeration of intracellular M. marinum CFUs were also as described (Gao et al., 2003a,b). For examination of intracellular growth, the cells were infected by M. marinum strains at moi 0.01 (103 CFU internalized by 105 cells) and incubated at 32°C, 5% CO2 for 6 days. For detection of cytotoxicity, J774 cells were infected by M. marinum strains at moi 0.1 (104 CFU internalized by 105 cells) and incubated at 32°C, 5% CO2 for 4 days. Cytotoxicity was determined by replacing the culture medium with fresh medium containing 10% ALAMAR BLUE (Trek Dignosis) that measures the metabolic activity of only the live cells (Gao et al., 1997). In addition, microscopic examination of live and dead cells in the monolayers was performed by staining with calcein AM and ethidium homodimer-1, as described above. Statistical analysis of the results was based on the ALAMAR BLUE assay, because a portion of the cells in the monolayers detached, particularly in WT-infected monolayer, and cytolysis could not be accurately determined by microscopic scoring.
Analysis of M. marinum cell-to-cell spread
BMDMs or A549 cells were infected by M. marinum strains at very low moi 0.001 (103 CFU internalized by 106 cells). Extracellular bacteria were killed by incubation of the infected monolayers with 200 µg ml−1 amikacin for 2 h. Then, the monolayers were overlaid with 0.8% agarose and incubated in culture medium with or without 20 µg ml−1 amikacin (≈ 20 × MIC). M. marinum strains expressing GFP were used to detect cell-to-cell spread, in which FFI were detected by fluorescence microscopy. The GFP-expressing M. marinum strains were generated by transformation with plasmid pmsp12.GFP.Apr (Chan et al., 2002) (generously provided by Lalita Ramakrishnan, University of Washington) and selection on 7H10 plates containing 30 µg ml−1 apramycin. In addition, a plaque assay, in which cells were fixed with 10% formalin and stained with crystal violet, was used to detect cell-to-cell spreading, exactly as previously described (Fernandez et al., 1989).
Analysis of production and secretion of ESAT-6 and CFP-10
Mycobacterium marinum strains grown to mid-log phase in 7H9 medium were washed and then grown in Sauton's medium (Bryan, 1998) for 3–5 days at 30°C. Culture filtrates were prepared by pelleting the bacteria at 2600 g and concentrating the supernatants 100-fold using centriplus concentrators (Amicon) with a 3 kDa cut-off. Pellets were washed twice with Sauton's and weighed, following which a portion were plated for quantification of CFUs and the rest subjected to bead-beating to obtain cell lysates. Cell lysates were obtained from disrupted bacteria by centrifugation for 5 min at 2300 g. Culture filtrates and cell lysates were normalized to the weight of the bacterial pellets. The samples were separated by SDS-PAGE on a 4–20% gradient gel, and the proteins were detected by Western blotting, developed by enhanced chemiluminescence and exposed to films. Anti-ESAT-6 (Mab HYB 76-8) and anti-CFP-10 (K8493) antibodies were generous gifts from Peter Andersen (Statens Serum Institut, Denmark); the anti-Ag85 serum was from Marcus Horwitz (University of California, Los Angeles); anti-GroEL1 (CS-44) and purified recombinant ESAT-6 protein were from John Belisle (Colorado State University, NIH contract NO1-AI-75320). Protein bands on films are quantified by densitometry that determines relative protein amount.
Extraction and analysis of lipids
Mycobacterium marinum cell wall lipids were metabolically labelled by incubating with [14C]-acetic acid (1 µCi ml−1) for 2 h. Extraction and separation of M. marinum cell wall lipids to polar lipids, apolar lipids and mycolic acid methyl esters were performed as described (Besra, 1998). Polar and apolar lipids were resolved in one or two dimensions on high performance thin layer chromatography (HPTLC) plates (EM Science) on either silica or reverse phase surface. Mycolic acids were first separated to α-, methoxy- and keto-mycolates on silica HPTLC plates by hexanes:ethyl acetate (95:5, v/v), and the purified fractions from preparative plates were further resolved on reverse phase HPTLC plates by chloroform:methanol (40:60, v/v). [14C]-labelled lipids on HPTLC plates were visualized by exposure to a Cyclone phosphor screen, and analysed by the Cyclone Phosphor System software (Packard Instrument Company).
Analysis of transposon insertion junction sequences and similarity searches
Recovery and sequencing of the transposon insertion sites were performed as described (Gao et al., 2003a). Similarity searches were performed using blastp and blastn at the NCBI site (http://www.ncbi.nlm.nih.gov/BLAST/) and the incomplete M. marinum genome database at the Sanger Center site (http://www.sanger.ac.uk/Projects/M_marinum/). Domain homology searches for predicted protein sequences were also performed at the SMART site (http://smart.embl-heidelberg.de/). The complete genome sequences for M. tuberculosis, M. leprae and M. bovis were obtained from the NCBI site using Accession No. NC_000962, NC_002677 and NC_002945 respectively.
Generation of M. marinum CFP-10/ESAT-6 deletion mutant
The vector for allelic exchange was constructed by cloning the ZEOCIN-resistant gene under the mycobacterial groEL promoter (Gao et al., 2003a) into the blunt-ended BamHI site of pPR27 (Pelicic et al., 1997) to create pLYG304.Zeo. To generate M. marinum CFP-10/ESAT-6 deletion mutant (ΔCE), a left arm fragment was polymerase chain reaction (PCR) amplified from M. marinum genomic DNA using primers DeltaCE-LF (5′-GCCCTCTAGAAGTCAGCGACGGCGCAG CCAG-3′) and DeltaCE-LR (5′-GGCGGAATTCGTGTTGACG ACGCTCACCAGTCGTC-3′), and a right arm fragment using primers DeltaCE-RF (5′-CGGGAATTCGCGTAGAATACCGA AGCACGAGATCGGG-3′) and DeltaCE-RR (5′-CCGCAAG CTTCTAGATTCATGCCGGTTTGGCGTGGC-3′). The left and right arms and the kanamycin-resistant gene (kanr) (derived from pUC4K, Pharmacia) were ligated into the XbaI and HindIII sites of pBluescript (Stratagene) to generate pBS.CE/kan. The XbaI fragment of pBS.CE/kan containing the kanr gene was cloned into the XbaI site of pLYG304.Zeo to generate the plasmid for homologous recombination. Correct construction of the plasmid was confirmed by sequencing the junctions between the arms and the kanr gene and between the arms and the plasmid backbone. The plasmid was introduced into M. marinum WT by electroporation and the transformants were sequentially selected first for kanr, then for resistance to sucrose and kanamycin, and finally for sensitivity to ZEOCIN. Deletion of CFP-10 and ESAT-6 from the M. marinum chromosome was confirmed by PCR that amplifies chromosomal sequence encompassing the junctions of the arms using primers DeltaCE.check-F1 (GGCG GCTCAGCGGGAGATCCACCGAG) and DeltaCE.check-R1 (CACCTTCTTCACGAGGCAGACCTCAGCGCC). The PCR products were sequenced to confirm chromosomal disruption of the target genes.
Complementation of M. marinum extRD1 mutants with M. tuberculosis or M. marinum genes
Complementation was performed for four of the M. marinum extRD1 mutants: M1 (Mh3866::Tn), M7 (Mh3879c::Tn), M8 (Mh3881c::Tn) and ΔCE. The coding sequences of Rv3866, Rv3879c, Mh3880c–Mh3881c, Rv3881c, and M. tuberculosis CFP-10 and ESAT-6 genes (Cole et al., 1998; http://www.sanger.ac.uk/Projects/M_marinum/) were PCR amplified using primers pair one (5′-CGGAATTCGCCAGCAG GTCTGTGCCATAGCGAGTCG-3′ and 5′-CGGTCGACGGA GATGTGAGGGGTCAGCCTCGG-3′), pair two (5′-GGGTAC GTAGTATTACCAGGCCGACGGGCAGC-3′ and 5′-GGGAAT TCACGCAGCGCCTGTTGCTGTCCTGGC-3′), pair three (5′-GGCGCGGAGGCCAAGGTGGTGT-3′ and 5′-TCAGACC ATCCCGTTGTTCATTG-3′), pair four (5′-GTAAAGGTGGCG TGGCCGAATGCGA-3′ and 5′-TCACTTCGACTCCTTACTG TCCTGGC-3′) and pair five (5′-GCGAATTCCCGTAATGAC AACAGACTTCCCGGCCACCC3′ and 5′-GGAAGCTTGCC CTATGCGAACATCCCAGTGACGTTGCC-3′), and cloned into pLYG206.Zeo (Gao et al., 2003b) to generate p1, p2, p3, p4 and p5. M. marinum mutant strains were transformed by electroporation with corresponding complementing plasmids and selected for resistance to ZEOCIN.
Infection of zebrafish and histopathological studies
Zebrafish AB strain was maintained as described (Lee et al., 2003). For infection, fish were anaesthetized in 0.02% tricaine (ethyl 3-aminobenzoate methanesulphonate) (Sigma-Aldrich) for 3–5 min and injected i.p. with 5–10 µl of M. marinum suspension in PBS containing a known number of bacteria. The same volume of PBS alone was injected into control fish. For survival studies, four fish were infected with each M. marinum strain. To evaluate CFUs in fish organs, eight fish were infected by each M. marinum strain and two fish were sacrificed at 1 day and 1, 3 and 5 weeks after infection. Fish were anaesthetized and sacrificed at a higher concentration of tricaine (0.05%) for a longer period of time (>20 min) and sterilized for 2 min in 70% ethanol before dissection. Organs were homogenized in 0.1% Triton X-100, and serially diluted samples were plated onto 7H10 plates for enumeration of CFUs. Incubation in 0.1% Triton X-100 for 30 min did not affect the viability of both M. marinum WT and the extRD1 mutants. For histopathology, fish were fixed in 10% formalin for 24 h before embedding in paraffin and sectioning.
We thank Marcus Horwitz (University of California, Los Angeles) for anti-Ag85 antibody and for helpful comments and suggestions. We also thank Peter Andersen (Statens Serum Institut, Denmark) for anti-ESAT-6 and anti-CFP-10 antibody and John Belisle (Colorado State University) for anti-GroEL antibody and rESAT-6 protein. We thank Lalita Ramakrishnan (University of Washington) for the GFP-expressing plasmid pmsp12.GFP.Apr. We also thank Susie Lee for helpful assistance with zebrafish maintenance. We thank Jeff Cox and members of the Brown lab for suggestions and comments. This work is supported by RO1 AI055614 and P01 33348; L.Y.G. was supported in part by T32 AI07334.