Today, an estimated one-third of the world's population is infected with Mycobacterium tuberculosis, the causative agent of tuberculosis [1, 2], and tuberculosis causes about 2 million deaths annually [1, 2]. Many aspects of the interactions between M. tuberculosis and its human host remain unclear. Not only is this bacterium able to evade the defenses of the host's immune system, it is also able to persist in the body for years and may reactivate to cause disease decades after the initial infection. A better understanding of the interaction between M. tuberculosis and its human host is critical to developing new strategies to control the tuberculosis epidemic.

A key feature of the pathogenicity of M. tuberculosis is its ability to evade the antimicrobial processes of the macrophage and replicate intracellularly. Mycobacteria enter macrophages primarily by conventional receptor-mediated phagocytic pathways [3]. Following phagocytosis, phagosomes containing viable tubercle bacilli fail to acidify, apparently because of failure to insert a proton-ATPase pump into the phagosomal membrane [4–6]. This leads to altered vacuolar maturation such that phagosome-lysosome fusion is blocked and the mycobacteria-containing vacuoles end up with an internal pH of about 6 and markers of phagosomes and early and late endosomes [7–11]. The mycobacterial genes responsible for these processes are largely unknown.

One method for identifying bacterial genes involved in pathogenesis is to express these genes in a nonpathogenic host and isolate bacteria with increased virulence. This technique was first used to identify a gene that enables Yersinia pseudotuberculosis to invade HEp-2 cells [12]. In these studies, Escherichia coli bacteria expressing Y. pseudotuberculosis genes were used to infect HEp-2 cells. Only E. coli bacteria expressing the Y. pseudotuberculosis inv gene were able to invade the animal cells [12]. Using a similar approach, Arruda et al.[13] identified an M tuberculosis gene responsible for invasion of HeLa cells. We have previously used a similar technique to identify M. leprae genes involved in intracellular survival [14]. That is, by expressing M. leprae genes in E. coli, we were able to isolate recombinant bacteria that exhibited increased resistance to killing by murine bone-marrow derived macrophages [14].

A similar approach was used by Wei et al [15] to isolate 21 Mycobacterium smegmatis recombinant clones that displayed a greater than 2-fold enhancement in survival after 48 hours. M. smegmatis is a fast-growing, nonpathogenic species of Mycobacterium in which M. tuberculosis genes can be efficiently expressed [16–18]. An in depth analysis of one recombinant clone revealed that M. smegmatis recipients carrying the M. tuberculosis eis gene on an extrachromosomal multicopy plasmid displayed 2.4- to 5.3-fold greater survival in U937 macrophages than wild-type M. smegmatis bacteria at 24 to 48 hours post-infection [15]. The 42-kDa eis gene product has been shown to be associated with the mycobacterial cell surface and is released into extracellular medium, but its precise function is not yet known [19].

In the studies reported here, we used a similar enrichment scheme to identify two additional recombinant M. smegmatis clones, sur2 and sur3, that demonstrated enhanced survival during infections of THP-1 human monocyte-derived macrophages. The M. tuberculosis ORFs responsible for increased survival were shown to be Rv0365c for the sur2 clone and Rv2235 for the sur3 clone. These ORFs encode proteins with as-of-yet unknown functions.

Cosmids carrying M. tuberculosis genes [20] were electroporated into M. smegmatis LR222 to create a library of M. smegmatis transformants, each of which carries a pYUB178::H37Rv cosmid integrated into its chromosome. The library was generated from about 4000 independent transformants. This represents about 20 M. tuberculosis genome-equivalents given that ~225 cosmids contain one genome-equivalent of M. tuberculosis[20].

To enrich for clones with increased survival in human macrophages, the library was passed through THP-1 macrophages as shown schematically in Figure 1. THP-1 is a human monocyte-derived macrophage cell line [21] and wild-type M. smegmatis bacteria are rapidly killed by THP-1 macrophages [22]. After three rounds of enrichment, the resulting clones were analyzed individually by Southern blot to evaluate the number of different clones present as previously described [20]. Out of the 3000 colonies recovered, genomic DNAs from the bacteria of 90 randomly chosen colonies were digested with Pst I and hybridized with the pYUB178 vector. Strains carrying different cosmids should exhibit a different pattern of hybridizing bands because of the presence of different M. tuberculosis H37Rv inserts. Two clones, designated sur2 and sur3, were each present three times and were chosen for further study. The remaining 84 clones each displayed unique patterns.

As a first step in the analysis of the sur2 clone, the time course of its survival in THP-1 macrophages was determined by infecting THP-1 macrophages, lysing the infected macrophages at various times after phagocytosis, and enumerating viable intracellular bacteria by plating on solid medium. Both parental and the sur2 bacteria were rapidly killed during the first few hours after phagocytosis (Figure 2). The sur2 bacteria appeared to survive slightly better at the 9 hour time point, but the difference was not statistically significant.

To compare directly the relative ability of the sur2 and wild-type bacteria to survive in macrophages, THP-1 macrophages were infected with a mixture of a genetically marked control strain and the sur2 strain and the survival of each strain was followed independently as previously described [22]. The wild-type strain expresses the xylE gene product, catechol 2,3-dioxygenase, such that when its colonies are sprayed with catechol, they turn bright yellow, while wild-type colonies remain white. In essence, each well of the experiment contains an internal standard (the wild-type bacteria) to which to compare the survival of the recombinant bacteria. In experiments in which THP-1 cells were infected at a MOI of 50:1 (results in ~1 phagocytosed M. smegmatis bacterium per macrophage) with a mixture containing equal numbers of bacteria of the xylE-expressing control strain and a strain carrying the cosmid vector pYUB178, the ratio of recovered white colonies to yellow colonies was 1:1 at all time points (data not shown). This indicates that the survival of the xylE-expressing strain was the same as that of the wild-type and could be used an internal reference by which to measure the survival of other clones.

In coinfection experiments with xylE-expressing bacteria, both sur2 and sur3 bacteria exhibited increased survival (Figure 3). Immediately after the 2 hour phagocytosis period (0 hr time point) the ratio of sur2 colonies to xylE-expressing colonies was 1:1 and by 12 hours it was about 7:1. By 24 hours, the ratio was approximately 20:1. The differences between the ratios at the zero time point and the subsequent time points were statistically significant (p < 0.005) for the 9, 12, and 24 hr time points. The ratio of sur3 colonies to control colonies increased from 1:1 at time 0 to 4:1 at 12 hours and to 5:1 at 24 hours (p < 0.005).

Because the recombinant clones contain integrated pYUB178::H37Rv cosmids, the following strategy was used to isolate cosmids corresponding to those in the sur2 and sur3 clones. First, genomic DNA from the sur2 clone was digested with Pst I, and genomic DNA from the sur3 clone was digested with Bam HI to generate fragments of each integrated cosmid carrying oriE, aph, and a portion of the M. tuberculosis genomic DNA insert. The presence of oriE allows the recombinant to replicate as a plasmid in E. coli. The digestion products were treated with T4 DNA ligase and transformed into E. coli XL 1-Blue. Plasmid DNA from the resulting kanamycin-resistant colonies were analyzed by restriction site mapping and partial sequencing of the M. tuberculosis genomic DNA insert as described in Materials and Methods. For the sur2 recombinant, PCR primers were designed to amplify a 0.8 kb region of the M. tuberculosis insert from M. tuberculosis genomic DNA for use as a probe in colony blot experiments. Probing colony blots of the E. coli (pYUB178::H37Rv) library with the 0.8 kb PCR fragment led to the isolation of a 4 kb plasmid. This plasmid contained an intact oriE and aph gene, a portion of the integrase gene, and 1.1 kb of M. tuberculosis genomic DNA (Figure 4). PCR, DNA sequence, and Southern blot data indicated that the recovered plasmid was the same as the cosmid integrated in the sur2 genome (data not shown). For example, sequencing of PCR amplicons of the junctions between the mycobacterial sequences and vector sequences revealed that the junctions in the recovered plasmid were identical to those in the sur2 genomic DNA.

Comparison of the sequence of the cloned 1.1 kb fragment with the M. tuberculosis H37Rv genome sequence [23] revealed that it contains portions of the Rv0366c and Rv0365c genes (Figure 4). In the sur2 clone, the amino-terminal 970 bp of the 1128 bp Rv0365c ORF is fused to 14 bp from the pYUB178 vector to generate an ORF encoding 328 amino acids (aa), compared to 376 aa encoded by the full-length Rv0365c ORF. The insert also contains 104 bp of the Rv0366c ORF fused to 220 bp of the L5 integrase ORF. This ORF could encode a 108 aa protein which contains the 73 amino-terminal amino acids of the 344 aa L5 integrase protein fused to the 34 carboxyl-terminal amino acids of the Rv0366c ORF.

To determine if the Rv0365c gene was responsible for the increased survival of sur2 bacteria, the full-length M. tuberculosis Rv0365c ORF as well as a truncated ORF similar to that present in the insert in the sur2 clone were subcloned into the pHIP vector downstream of the M. tuberculosis hsp65 promoter. The truncated ORF contained the 969 bp of Rv0365c present in the sur2 insert followed by a stop codon but did not contain the 5 amino acids corresponding to the vector contribution to the ORF. In coinfection experiments, the ratio of M. smegmatis bacteria expressing the full-length Rv0365c ORF to xylE-expressing wild-type bacteria was 1:1 at time zero and increased to approximately 10:1 at 12 hours and to 11:1 at 24 hours (Figure 5). The clone expressing the truncated Rv0365c ORF exhibited slightly less, but not statistically significantly different, increased survival (1:1 at time zero, 8.6:1 at 24 hrs).

To investigate the difference in survival at 24 hours of the full-length Rv0365c ORF expressing bacteria and sur2 bacteria, a strain expressing Rv0365c and xylE was constructed and used in THP-1 coinfections with the original sur2 clone. In this coinfection, the ratio of white (sur2) to yellow (xylE and Rv0365c expressing) colonies remained 1:1 through 12 hours and then increased to ~3:1 at 24 hrs, consistent with the above-described observations.

A cosmid corresponding to the one in the sur3 clone was isolated from the E. coli (pYUB178::H37Rv) library by probing colony blots with the 4.5 kb Bam HI/Eco Rl fragment of the M. tuberculosis genomic DNA insert in the plasmid recovered from the sur3 clone. PCR, DNA sequence, and Southern blot data indicated that the recovered cosmid was the same as the cosmid integrated in the sur3 genome (data not shown). The ~10.7 kb cosmid in the sur3 chromosome contains a 5.76 kb fragment of M. tuberculosis H37Rv genomic DNA (Figure 6) [23]. This region encodes eight potential ORFs designated Rv2233-Rv2240c as well as the tRNA for valine.

To determine which ORF(s) was responsible for the increased survival exhibited by sur3 bacteria, each ORF was subcloned individually into the expression vector pHIP. The two potential operons, Rv2233-Rv2235 and Rv2238c-Rv2240c, were also subcloned into pHIP. The recombinant bacteria were examined for survival in the macrophage in coinfection experiments at 0, 6, and 12 hours post phagocytosis (Figure 7). Each of the recombinants tested exhibited a 1:1 ratio of white-to-yellow colonies at 0 hours. Two of the recombinants demonstrated an increase in the ratio of white-to-yellow colonies over time, while the ratio remained at 1:1 for the other eight recombinants. One of the two recombinants contained the potential operon of Rv2233-Rv2235, and the other contained Rv2235 only. The ratio of white-to-yellow colonies for bacteria expressing ORFs Rv2233-Rv2235 was ~2:1 at 6 hours and ~6:1 at 12 hours. For bacteria expressing Rv2235, the ratio of white-to-yellow colonies was ~2:1 at 6 hours and >3:1 at 12 hours. The survival of recombinant bacteria expressing the Rv2235 ORF relative to wild-type was more directly compared to the relative survival of sur3 by doing the coinfections in parallel. The ratios of white-to-yellow for both recombinant bacteria and sur3 bacteria were 1:1 at 0 hours, ~5:1 at 12 hours, and >5:1 at 24 hours. The differences in the relative survival of the three strains were not statistically significant.

A PCR fragment containing the M. tuberculosis ORFs Rv2233, Rv2234, and Rv2235 without the upstream promoter region, was cloned into the pBPhin vector, which does not contain a promoter to express the inserted DNA. In coinfections with the xylE-expressing bacteria, the recombinant bacteria containing Rv2233, Rv2234, and Rv2235 did not exhibit the same increase in macrophage survival as the sur3 clone (data not shown), suggesting that the Rv2235 ORF is expressed using signals upstream of the Rv2233 ORF.

Southern blots of genomic DNA from M. tuberculosis, M. smegmatis, Mycobacterium leprae, and Mycobacterium avium were probed with Rv0365c and Rv2235. Rv0365c hybridized to bands in M. tuberculosis, M. avium, and M. smegmatis under high stringency conditions (data not shown). Rv2235 hybridized to a band in M. tuberculosis, and under low stringency conditions recognized a band in M. leprae (data not shown).

Database searches revealed that Rv0365c encodes an ~41-kDa protein which displays significant homology only to a truncated Corynebacterium glutamicum hypothetical protein of unknown function (42% identity in a 296 aa overlap) located upstream of a gene encoding a fructose-bisphosphate aldolase. Rv2235 encodes a conserved hypothetical membrane protein of about 30-kDa and shares a motif with the SURF-1 family of proteins. The other two ORFs in the putative Rv2233-Rv2235 operon, Rv2233 and Rv2234, share homology with several proteins in data base searches. Rv2233 may encode a putative phosphatase, and Rv2234 may encode a low molecular weight protein tyrosine phosphatase [23].

There are several potential limitations of isolating M. tuberculosis genes involved in intracellular survival using an enrichment procedure. One limitation is that this protocol is biased towards the recovery of clones with the greatest increase in survival relative to wild-type [14]. Thus, the recovered clones are not a random collection of genes involved in intracellular survival and hence the number of genes involved for survival cannot be calculated. This approach is also biased towards identifying genes expressing proteins that directly interfere with the antimicrobial processes of the macrophage. Some types of genes involved in resistance to killing, such as genes that are part of a multi-enzyme pathway, may not be isolated. Also, genes necessary for intracellular survival as opposed to resistance to killing are not likely to be isolated.

Two general classes of clones might be recovered following enrichment of the M. smegmatis recombinant library for clones with increased intracellular survival. One type might be clones that carry M. tuberculosis genes that confer enhanced resistance to the antimicrobial processes of the macrophage. Another type might be clones whose M. tuberculosis gene products are involved in attachment or invasion or increase phagocytosis. The two genes that were isolated in the studies reported here confer enhanced resistance rather than increased uptake. That is, the ratios of bacteria expressing either Rv0365c or Rv2235 to wild-type bacteria were 1:1 in both the initial mixture and inside the macrophages at the end of the phagocytosis period. Differences in survival compared to wild-type did not become apparent until about 9 hours post-phagocytosis for bacteria expressing Rv0365c and about 6 hours post-phagocytosis for those expressing Rv2235.

In this study, the enrichment process resulted in the isolation of two strains containing small, integrated plasmids rather than the expected 30–50 kb cosmids [20]. Small plasmids corresponding to the integrated ones were present in the original E. coli (pYUB178:H37Rv) cosmid library. PCR, sequencing, and Southern blot analysis demonstrated that the plasmid integrated in the sur2 clone was approximately 4 kb containing 1.1 kb of M. tuberculosis DNA and the plasmid integrated in the sur3 clone was approximately 10.7 kb containing about 5.7 kb of M. tuberculosis DNA. Cosmid libraries frequently contain clones without DNA inserts [24], so it is not too surprising that clones with small DNA fragments were present in the E. coli library after infection with λ phage. The small plasmids may have a growth or replication advantage causing them to be over-represented in the library following the various amplification steps.

The ~2 kb deletion of vector DNA in the sur2 clone results in a fusion of the ORF of the L5 integrase with that of the M. tuberculosis insert such that the fused ORF encodes a protein that contains only the amino-terminal 73 aa of the 344 aa L5 integrase. The observation that the sur2 clone contains an integrated plasmid corresponding to the plasmid isolated from the cosmid library suggests that either a) the hybrid protein retains integrase activity, b) the plasmid inserted into the attachment site by homologous recombination between attP and attB, or c) a functional integrase was provided in trans by a second cosmid transiently present in the original transformant or by a gene present in the M. smegmatis LR222 genome. However, repeated attempts to electroporate the 4 kb plasmid into M. smegmatis LR222 did not generate any stable kanamycin-resistant transformants (data not shown). These results suggests that the hybrid protein does not retain integrase activity, that integration by homologous recombination into attP is unlikely, and that provision of integrase activity by an M. smegmatis gene chromosomal does not occur. The simplest explanation is that integrase was provided in trans by a second cosmid transiently present in the original transformant, and indeed, M. smegmatis bacteria carrying an integrated copy of the 4-kb plasmid can be readily isolated following electroporation of a mixture of the 4-kb plasmid and a plasmid that expresses integrase (unpublished results).

Data base searches did not reveal any homologies that could be used to predict functions for the gene products of ORFs Rv0365c or Rv2235. Rv0365c encodes a hypothetical protein of 376 amino acids which displays significant homology only to a Corynebacterium glutamicum hypothetical protein of unknown function (42% identity in a 296 aa overlap) [23]. Rv2235 encodes a hypothetical protein of 271 aa with three putative transmembrane domains and which displays significant homology only to hypothetical protein MLCB1243.32c in M. leprae, to which it is 74% identical [23, 25]. This M. leprae homologue was evident in Southern blot experiments done under low stringency conditions. Rv2235 also contains the SURF-1 signature sequence [26] and modest homology to members of the SURF-1 family such as the SURF-1 protein of Caulobacter crescentus (25% identity; 40% similarity) [27]. SURF-1 proteins are ~33-kDa, integral membrane proteins whose precise function is not known. In eukaryotic cells, SURF-1 proteins are involved in the assembly and maintenance of mitochondrial respiratory chain complexes including cytochrome oxidase [26, 28]. The homology to SURF-1 proteins raises the possibility that Rv2235 could play a role in resisting the antimicrobial activities of macrophages by helping to maintain the stability or function of an important cellular process, akin to the stabilizing role of chaperonins during a heat shock.

In addition to the genes described in this report, other studies with M. smegmatis recombinants and/or mutants have implicated 11 other M. tuberculosis genes in intracellular survival. Genes identified using enrichment or screening protocols include Rv2962c and Rv2958c (probable glucuronsyl transferases), Rv2220 (glutamine synthetase A1), Rv3913-Rv3914 (thioredoxin, thioredoxin reductase) and Rv2416c (eis, unknown function) [14, 15, 22, 29]. By screening insertional mutants of M. smegmatis, Lagier et al [30] isolated 8 mutants with impaired ability to survive in human peripheral blood monocyte-derived macrophages and identified the M. tuberculosis gene corresponding the mutated M. smegmatis gene for five of them. The genes included: Rv3052c (probable nrdI) which is postulated to be involved in deoxynucleotide production under stressed conditions; Rv0101 which is a nonribosomal peptide synthetase that displays strong homology with a Pseudomonas nonribosomal peptide synthetase required for the synthesis of the pyoveridine, a siderophore involved in iron uptake; Rv3420c which displays homology with the S18 ribosomal protein acetyltransferase which behaves as a heat shock protein in Chlamydia trachomatis; and Rv0497 and Rv3604c which are hypothetical conserved membrane proteins of unknown function.

It should be noted that these M. tuberculosis genes generally confer only a limited enhancement of the survival of M. smegmatis bacteria in the human macrophages. That is, usually only a few per cent of the recipients are viable 24 to 48 hours post-infection. This reinforces the concept that the intracellular survival of a pathogenic mycobacteria is a complex multifactoral process. The precise role(s) of any of the identified genes in the intracellular survival of mycobacteria is not yet known, although several of the identified genes have features of stress response genes. This is not unexpected given the relatively small enhancements of the survival of the M. smegmatis recombinants and the numerous environmental stresses encountered in the macrophage. Additional studies, such as the construction and characterization of targeted knock-out mutants, will be needed to determine the roles of the proteins encoded by these genes in the survival of M. tuberculosis in human macrophages.

Using an enrichment and screening procedure, two M. tuberculosis genes, Rv0365c and Rv2235, were identified that could confer an enhanced ability to survive in human macrophages to normally susceptible M. smegmatis recipients. The functions of these two proteins are not known. This study brings the number of M. tuberculosis genetic loci that have been implicated in enhancing the intracellular survival of M. smegmatis cells to 13. The precise role(s) of any of the identified genes in the intracellular survival of mycobacteria remain to be elucidated.