|Home | About | Journals | Submit | Contact Us | Français|
The pe and ppe genes are unique to mycobacteria and are widely speculated to play a role in tuberculosis pathogenesis. However, little is known about how expression of these genes is controlled. Elucidating the regulatory control of genes found exclusively in mycobacteria, such as the pe and ppe gene families, may be key to understanding the success of this pathogen. In this study, we used a transposon mutagenesis approach to elucidate pe and ppe regulation. This resulted in the identification of Rv0485, a previously uncharacterized transcriptional regulator. Microarray and quantitative real-time PCR analysis confirmed that disruption of Rv0485 reduced the expression of the pe13 and ppe18 gene pair (Rv1195 and Rv1196), defined the Rv0485 regulon, and emphasized the lack of global regulation of pe and ppe genes. The in vivo phenotype of the Rv0485 transposon mutant strain (Rv0485::Tn) was investigated in the mouse model, where it was demonstrated that the mutation has minimal effect on bacterial organ burden. Despite this, disruption of Rv0485 allowed mice to survive for significantly longer, with substantially reduced lung pathology in comparison with mice infected with wild-type Mycobacterium tuberculosis. Infection of immune-deficient SCID mice with the Rv0485::Tn strain also resulted in extended survival times, suggesting that Rv0485 plays a role in modulation of innate immune responses. This is further supported by the finding that disruption of Rv0485 resulted in reduced secretion of proinflammatory cytokines by infected murine macrophages. In summary, we have demonstrated that disruption of a previously uncharacterized transcriptional regulator, Rv0485, results in reduced expression of pe13 and ppe18 and attenuation of M. tuberculosis virulence.
Mycobacterium tuberculosis is a highly successful human pathogen, responsible for 8 million new cases of tuberculosis each year. M. tuberculosis employs a number of strategies for evasion of host immune defenses, which allow it to persist within the human host for extended periods, possibly even decades. While some of these mechanisms have been well characterized, others are less well understood. It has been widely speculated that the pe and ppe gene families may contribute to immune evasion by M. tuberculosis. These extensive gene families are named for the proline-glutamic acid (PE) and proline-proline-glutamic acid (PPE) motifs at the N termini of their encoded proteins. The pe and ppe gene families are exclusive to mycobacteria (10) and are particularly numerous in pathogenic mycobacteria (29), and emerging data suggest that that they may play diverse roles in mycobacterial pathogenesis (4, 6, 11, 12, 20, 37, 38, 48, 60).
A subset of the pe and ppe families is closely associated with the esx loci of M. tuberculosis. The ESX-1 region encodes a secretory apparatus that is a major contributor to M. tuberculosis virulence (31, 36, 45). Comparative genomic analysis has revealed that orthologues of the ESX-1 system are present in a number of pathogens (28, 41), with multiple copies present in M. tuberculosis. However, it is only in mycobacteria that these regions also include members of the pe and ppe gene families. It is known that some PE/PPE proteins are cell wall associated or secreted into the extracellular environment (1, 5, 18, 19, 26, 53), which may be mediated by the specialized ESX secretion systems (1, 26).
Very few pe and ppe genes are thought to be essential (55, 56), likely due to extensive functional redundancy. However, a number of studies have demonstrated immune recognition by infected hosts (9, 13, 14, 17, 22, 24, 25, 33, 39, 42, 67), indicating that they are expressed during infection. Although the precise function of the PE/PPE families has yet to be determined, their close genomic and evolutionary association with the ESX regions (29), together with their highly immunogenic nature, points to an integral role in the infectious process.
A fundamental step in understanding the role of pe and ppe genes is elucidating how their expression is regulated. Although various studies have demonstrated that pe and ppe genes are expressed under a range of in vitro and in vivo conditions, they have not revealed any obvious indication of global pe and ppe gene regulation (65). Some pe and ppe genes appear to be constitutively expressed, some variably so, and some have not been identified under any experimental conditions (65). Investigation of various M. tuberculosis two-component regulators and sigma factors has demonstrated indirectly that the expression of subsets of the pe and ppe genes is under the control of particular members of these regulatory proteins (27, 30, 32, 35, 49, 50, 63). In other work, the iron-dependent regulator IdeR has been shown to interact directly with the predicted promoter region of one ppe gene, Rv2123 (51). A more recent study demonstrated that two members of the pe gene family are inversely regulated in vivo, suggesting specific regulatory mechanisms (21). However, it is unknown how this regulation may be orchestrated by the bacterium, and no studies have specifically focused on identifying and characterizing regulators of these genes. Elucidating the regulatory control of genes unique to mycobacteria may be key to understanding novel aspects of tuberculosis pathogenesis, as recently highlighted by the characterization of EspR, an ESX-1-associated transcription factor (47).
In this study, we sought to identify regulators of the esx-associated gene pair, pe13 and ppe18. PPE18 was previously identified as the immunogen Mtb39A (22) and forms part of the polyprotein vaccine candidate Mtb72f (58), which is currently under evaluation in Phase II clinical trials (http://clinicaltrials.gov/ct2/show/NCT00600782). Elucidating the regulation of this immunogen therefore has implications for vaccine development as well as for a more general understanding of pe and ppe biology. We describe here the use of a transposon mutagenesis approach to identify the transcriptional regulator Rv0485. We demonstrate that the pe13 and ppe18 (Rv1195 and Rv1196) gene pair is part of the Rv0485 regulon. Furthermore, we show that Rv0485 is required for virulence of the pathogen during in vivo growth in mice.
Mouse-passaged laboratory stocks of Mycobacterium tuberculosis H37Rv were grown in Middlebrook 7H9 liquid medium supplemented with 10% oleic acid-albumin-dextrose-catalase (OADC), 0.2% glycerol, and 0.05% Tween 80 or on solid Middlebrook 7H10 or 7H11 medium supplemented with 10% OADC, 0.5% glycerol, and 100 μg ml−1 cycloheximide. Recombinant strains constructed in this study (Table (Table1)1) were grown on the same media, supplemented with the following antibiotics as necessary: hygromycin at 50 μg ml−1, kanamycin at 25 μg ml−1, and apramycin at 50 μg ml−1. M. tuberculosis on solid medium was cultured for 21 to 28 days at 37°C and in liquid culture for 5 to 8 days at 37°C with shaking. For screening transposon mutant libraries, 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) was added to the plates at a concentration of 80 μg ml−1.
Electrocompetent Escherichia coli DH10B cells were purchased from Invitrogen. Recombinant E. coli was cultured in LB-Miller broth supplemented with the following antibiotics as appropriate: 180 μg ml−1 hygromycin and 50 μg ml−1 apramycin.
Plasmids and cloning primers used in this study are listed in Tables Tables11 and and2.2. The plasmid pUPSpe13 was derived from pCV129 (kind gift from A. Steyn) and contains a 539-bp region upstream of pe13 transcriptionally fused to lacZ. pCV129 was first modified to incorporate a PacI site to facilitate subsequent cloning steps. This was achieved by amplification of a 910-bp PCR product from pCV129 with the primers pCV129FOR and pCV129REV, which incorporate a PacI site overlapping an existing AciI site. The resulting PCR product was digested with AciI and NcoI and cloned into AciI/NcoI-restricted, dephosphorylated pCV129 to generate pHIL. The SacB gene was then excised from pMP7 with NotI and NdeI, Klenow filled, and cloned into NotI-restricted, Klenow-filled, dephosphorylated pHIL to generate pSHIL. Finally, the 539-bp region upstream of pe13 was amplified with the primers 1195UFOR and 1195UREV, which incorporate PacI and NcoI, respectively. The PacI/NcoI-digested PCR product was cloned into the PacI/NcoI-digested, dephosphorylated pSHIL. The resulting plasmid, pUPSpe13, contains the region upstream of pe13 transcriptionally fused to lacZ, a hygromycin resistance cassette, the SacB gene, and an attP attachment site.
Plasmid pRV0485 (used to complement M. tuberculosis Rv0485::Tn) was based on the shuttle vector pMP349 (a kind gift from M. S. Pavelka, Jr. ). The full-length Rv0485 open reading frame (ORF; plus 200 bp upstream predicted to incorporate the promoter region of Rv0485) was amplified with the primers 0485COMF1 and 0485COMR2, which incorporate XbaI and EcoRI sites into the resulting product. The PCR product was digested with XbaI and EcoRI and cloned into XbaI/EcoRI-restricted, dephosphorylated pMP349. The resulting plasmid also contains an apramycin resistance marker and mycobacterial and E. coli origins of replication.
Mycobacterial genomic DNA was extracted according to a modified version of that by Belisle and Sonnenberg (7). Briefly, mycobacterial cultures were pelleted, resuspended in Tris-EDTA buffer, pH 8.0, and extracted with chloroform-methanol. The bacterial mass was dried, resuspended in Tris-EDTA buffer containing 0.1 M Tris, pH 9.0, and then treated overnight with 100 μg ml−1 lysozyme. Following digestion with 100 μg ml−1 of proteinase K, the solution was phenol-chloroform extracted and genomic DNA was recovered by isopropanol precipitation.
Mycobacterial RNA was extracted with TRIzol (Life Technologies), followed by Qiagen RNeasy kit column purification. Briefly, mid-exponential-phase broth culture (usually 10 ml) was added directly to 3 vol guanidium thiocyanate solution containing 5 M guanidinium thiocyanate, 0.5% sodium N-lauryl sarcosine, 0.1 M β-mercaptoethanol, and 0.5% Tween 80. The bacteria were pelleted by centrifugation and resuspended in 0.1 vol TRIzol. The TRIzol suspension was added to FastPrep blue tubes (MP Biomedicals) containing 0.1-mm silica beads and processed by reciprocal shaking in a FastPrep reciprocal shaker (MP Biomedicals) at 6.5 m·s−1 for 45 s. The supernatant was removed to a fresh tube containing an equal volume of chloroform, mixed, and centrifuged at 13,000 rpm for 5 min. The aqueous phase was reextracted with chloroform. The RNA was isopropanol precipitated, washed with 70% ethanol, and dissolved in RNAsecure (Ambion). The RNA was column purified using the RNeasy kit in combination with on-column DNase I digestion (Qiagen). RNA quality and quantity were assessed using Agilent RNA 6000 Nano Chips and the Agilent 2100 Bioanalyzer according to the manufacturer's instructions.
M. tuberculosis H37Rv transformed with pUPSpe13 (wild type; forms blue colonies on X-Gal plates) was transduced with the transposon donor phagemid ΦMycoMarT7 (a kind gift from Eric Rubin, Harvard School of Public Health) to generate MycoMarT7 transposon mutant libraries, as described elsewhere (54). The MycoMarT7 transposon integrates randomly throughout the genome, introducing a kanamycin resistance marker into disrupted ORFs. To ensure comprehensive coverage of the genome, >10,000 colonies were screened. Colonies demonstrating color changes relative to the parent strain (from the blue color of M. tuberculosis H37Rv::pUPSpe13 to either dark blue or white) were picked and replated onto solid medium containing X-Gal to verify the color change. Genomic DNA was extracted from colonies that demonstrated consistent color changes relative to the parent strain, and the site of MycoMarT7 integration was determined by DNA sequencing.
Whole-genome microarray expression analysis was performed as described elsewhere (61). Briefly, RNA was reverse transcribed with SuperScript II reverse transcriptase (Invitrogen), and cDNA was labeled by direct incorporation of either Cy3 or Cy5 dCTP (Amersham). The relevant pairs of Cy3- and Cy5-labeled cDNA were mixed and purified using a Qiagen MinElute kit. Two technical replicates (affected by dye swap design) were performed for each of three biological replicates.
Whole-genome microarrays included PCR products from 4,410 predicted ORFs of the sequenced strains of M. tuberculosis H37Rv, M. tuberculosis CDC1551, and Mycobacterium bovis AF2122/97 (accession no. A-BUGS-23; Bacterial Microarray Group at St. Georges, Array version 2.1.1; array design is available in BμG@Sbase [http://bugs.sgul.ac.uk/A-BUGS-23]) and also ArrayExpress (accession no. A-BUGS-23). Microarrays were hybridized with labeled cDNA as described elsewhere (43, 61) and then scanned with an Affymetrix 428 scanner. The scanned images were analyzed with ImaGene4.1, the median spot intensities were calculated, and then data were analyzed with GeneSpring software.
cDNA was generated from 1 μg total RNA using random hexamers in combination with SuperScript III reverse transcriptase (Invitrogen) according to the manufacturer's instructions. The reaction mix was incubated at 25°C for 5 min and 55°C for 1 h, followed by a 15-min incubation at 70°C. cDNA was quantified with the QuantiTect Sybr green PCR kit (Qiagen). A total of 0.75 μl of cDNA was added to each reaction mixture with 0.5 μM of each primer and QuantiTect Sybr green master mix as recommended by the manufacturer. Reactions were performed on a Rotor-Gene 3000 thermal cycler (Corbett Research), and samples were incubated at 95°C for 15 min, followed by 40 cycles of 95°C for 20 s, 60°C for 15 s, and 72°C for 20 s. Melt curve analysis between 72°C and 99°C was performed at the end of the cycling. Each reaction was performed in duplicate (on at least two biological replicates), and no template controls and reverse-transcriptase-negative controls were included for each run.
Rotor-Gene software was used to calculate the reaction efficiency for each primer pair using standard curves generated from 10-fold serial dilutions of genomic DNA. For each amplification run, the calculated threshold cycle of the gene of interest was normalized to the threshold cycle of the reference gene, sigA, amplified from the corresponding sample. Under the growth conditions investigated, sigA levels remain constant, and it is thus an appropriate reference gene. The change in expression was calculated using the relative expression software tool (REST) (43).
Helix-turn-helix (HTH) prediction software was accessed via the Pôle Bioinformatique Lyonnais site (http://npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_hth.html) (23). Mycobacterial sequence information was accessed via the TubercuList (http://genolist.pasteur.fr/TubercuList/) (15) and the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/) Web servers. Sequence alignments were performed using ClustalW2 software (http://www.ebi.ac.uk/Tools/clustalw2/index.html) (34). Protein solubility prediction software (PROSO) was accessed via the Expropriator Web server (http://mips.helmholtz-muenchen.de/proso/proso.seam) (59). Proteins of known structures were queried via the RCSB Protein Data Bank Web server (http://www.rcsb.org/pdb/home/home.do).
Barrier-bred, 6- to 8-week-old female BALB/cJ or BALB/cJ SCID mice were purchased from Jackson Laboratories and housed in microisolator cages or in individually ventilated cages in a biosafety level 3 animal facility in compliance with institutional protocols. Mice were infected through the lateral tail vein with (i) the parental M. tuberculosis H37Rv strain, carrying an intact copy of Rv0485 (wild type); (ii) the mutant strain with a disruption in Rv0485, M. tuberculosis H37Rv Rv0485::Tn (Rv0485::Tn); or (iii) the complemented strain, M. tuberculosis H37Rv Rv0485::Tn plus Rv0485 (Rv0485::Tn + Rv0485). Immunocompetent BALB/cJ mice were infected with 1 × 106 CFU of each strain, and immunocompromised SCID mice with 1 × 103 CFU of each strain. Bacterial burdens in organs were determined by serial dilution plating of lung and spleen homogenates at various time points postinfection. Moribund animals were euthanized according to institutional protocols.
Tissues were fixed in 10% phosphate-buffered formalin and then embedded in paraffin. Sections were stained with hematoxylin and eosin (H&E) to visualize mammalian cells or Ziehl-Neelsen stained to visualize acid-fast bacilli. Stained sections were visualized using a Nikon Eclipse E600 microscope, and images were captured with a Nikon DXN1299 digital camera and ACT-1 software.
J774A.1 macrophages were plated onto 24-well tissue culture plates at a density of 5 × 105/well in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum (FCS). Mycobacterial strains were grown to mid-log phase, briefly ultrasonicated to disperse clumps, washed with phosphate-buffered saline-0.05% Tween 80, and then diluted in DMEM-10% FCS prior to addition to macrophage monolayers (1 × 105 CFU/well). After 4 h, noninternalized extracellular bacteria were removed by treating with 200 μg ml−1 amikacin for 1 h, followed by three washes with DMEM. Monolayers were incubated in DMEM-10% FCS for up to 72 h, and then cell-free supernatants were removed and cytokine levels determined by enzyme-linked immunosorbent assay according to the manufacturer's instructions (BD Biosciences, Oxford, United Kingdom). Intracellular bacteria were enumerated by serial dilution plating after lysis of macrophages with phosphate-buffered saline-0.1% Triton X-100. Each strain was analyzed in triplicate wells in at least three separate experiments.
Statistical analyses were performed using GraphPad Prism 4 software. Unless otherwise stated, data were analyzed by an unpaired, two-tailed t test. Mouse survival data were analyzed by the log-rank test. P values of <0.05 were considered to be statistically significant.
Fully annotated microarray data have been deposited in BμG@Sbase (accession no. E-BUGS-91 [http://bugs.sgul.ac.uk/E-BUGS-91]) and also in Array-Express (accession no. E-BUGS-91).
We chose to screen for regulators of pe13 and ppe18 (Rv1195 and Rv1196). This gene pair is cotranscribed (data not shown) and precedes the esxK and esxL gene pair (Fig. (Fig.1A).1A). Although not predicted to be essential (55, 56), a number of reports indicate that PE13 and PPE18 are expressed and recognized during infection (22, 40). Furthermore, PPE18 constitutes part of the polyprotein vaccine candidate, Mtb72f (58).
To identify genes influencing the expression of pe13 and ppe18, we used a promoter-reporter fusion approach in conjunction with transposon mutagenesis. Briefly, we first constructed an integrating plasmid incorporating 539 bp upstream of pe13 and ppe18 (pUPSpe13) fused to lacZ. This was transformed into M. tuberculosis H37Rv to generate a hygromycin-resistant strain with the UPSpe13-lacZ reporter gene fusion integrated at the attB site of the genome. For brevity, this parental strain (with an intact copy of Rv0485) will be referred to as the “wild type.” Due to the presence of the integrated UPSpe13-lacZ reporter construct, the wild-type strain reproducibly yields blue colonies when streaked onto 7H10 medium containing X-Gal. The wild-type strain was subsequently transduced with ΦMycoMarT7 (54) and plated onto solid medium containing X-Gal to screen for color changes which may be associated with changes in expression. Genomic DNA was isolated from clones demonstrating reproducible color changes, and the MycoMarT7 integration sites were sequenced to identify genes disrupted by the transposon. In this way, an insertion into Rv0485, a putative transcriptional regulator, was identified (740 bp from the 5′ end of the gene) (Fig. (Fig.1B1B).
The ORF Rv0485 is annotated in the M. tuberculosis sequence database TubercuList as a putative transcriptional regulator and a member of the NagC/XylR repressor family. Further bioinformatic analysis demonstrates that Rv0485 is highly conserved within the mycobacterial genus and the closely related Rhodococcus sp. and Nocardia farcinica (Fig. (Fig.2).2). HTH motif prediction software is only weakly predictive of an HTH DNA-binding domain at the N terminal of Rv0485. However, alignment with other conserved sequences with higher probability HTH predictions indicates that this region is highly conserved, particularly in the second, C-terminal helix (Fig. (Fig.2).2). Comparison with proteins of experimentally determined structure in the RCSB Protein Data Bank identifies a Vibrio cholerae transcriptional regulatory protein, VC_2007 (PDB ID, 1Z05; 23% identity; 42% positives; E value, 6.32 × 10−12), and the E. coli NagC-like transcriptional regulator (PDB ID, 3BP8; 19% identity; 39% positives; E value, 7.62 × 10−4). These proteins are both members of the ROK (repressor ORF kinase) family and have been shown to be involved in the regulation of proteins within phosphotransferase systems (8, 44). Both proteins exhibit a two-domain structure with an N-terminal DNA binding domain containing an HTH motif and a C-terminal glucokinase-like domain. Based on sequence alignments, we predict that Rv0485 will share a similar structure.
We verified that Rv0485 exerts an influence on the expression of pe13 and ppe18 by comparing mRNA expression levels (from mid-log-phase cultures) in the wild-type versus mutant (Rv0485::Tn) strains using real-time quantitative reverse transcription-PCR (qRT-PCR). This confirmed that disruption of Rv0485 led to an approximately eightfold decrease in pe13 and ppe18 mRNA levels (Fig. (Fig.3),3), suggesting that Rv0485 is a positive regulator of the pe13 and ppe18 gene pair. Further analysis revealed that complementation with an episomal copy of Rv0485 under the control of its natural promoter (Rv0485::Tn + Rv0485) restored nearly wild-type levels of pe13 and ppe18 mRNA.
For a more global overview of the impact of disruption of Rv0485 on gene expression, we performed whole-genome microarray analysis. Comparison of mRNA levels in the wild-type versus Rv0485::Tn strains provided further confirmation that disruption of Rv0485 leads to downregulation of pe13 (Table (Table3).3). A limited number of other genes were downregulated (P < 0.05; Benjamini and Hochberg ratio of <0.5) in Rv0485::Tn, including conserved hypotheticals, a transposase, and devS. Only six genes were upregulated (P < 0.05; Benjamini and Hochberg ratio of >2) in the mutant strain (Table (Table4).4). Five of these were immediately adjacent to each other, while the sixth apparently upregulated gene was Rv0485 itself.
The Rv0485 regulon was analyzed in more detail using qRT-PCR. First, we confirmed that genes identified as upregulated or downregulated in the wild type versus Rv0485::Tn by whole-genome microarray expression analysis showed similar patterns when analyzed by qRT-PCR (Tables (Tables33 and and4).4). We extended this analysis to the complemented strain, Rv0485::Tn + Rv0485, and demonstrated that complementation with intact Rv0485 restored expression to nearly wild-type levels for the majority of these genes.
Next, we examined more closely the observation from the microarray data that Rv0485 appeared to be upregulated in the mutant. For this purpose, we designed two sets of PCR primers to amplify products on either side of the transposon insertion in Rv0485 (Fig. (Fig.1B).1B). In this regard, it is important to note that the microarray-spotted PCR product corresponding to Rv0485 lies entirely 5′ of the transposon insertion, i.e., equivalent to the 5′ qRT-PCR product for Rv0485 (Table (Table2).2). Although qRT-PCR suggested a slight elevation in mRNA corresponding to the 5′ region, this was not to the same degree as that identified by microarray analysis. Levels of the qRT-PCR product corresponding to the 3′ region of Rv0485 were significantly decreased in the mutant relative to that in the wild type, suggesting that the transposon insertion has disrupted transcriptional read-through. As expected, levels of both Rv0485 qRT-PCR products were elevated in the complemented strain relative to those of the wild type.
Finally, we wished to establish whether the effect of disruption of Rv0485 was confined to pe13 and ppe18 or extended more generally to other pe and ppe genes. As evidenced by the failure to detect a reduction in ppe18 mRNA levels in Rv0485::Tn by a microarray-based approach, the high degree of similarity between different pe and ppe genes may confound analysis. This is a particular concern with a PCR-based array such as the one used in this study; ppe18, for example, demonstrates a high degree of similarity to ppe19 and ppe60 (>90% identity to the PCR products used). We therefore performed a directed qRT-PCR analysis on a subset of pe and ppe genes, with primer pairs designed and tested to confirm specific amplification of individual pe and ppe genes. For this, we focused on the pe and ppe genes with genomic arrangements similar to those of pe13 and ppe18, i.e., those encoded as closely situated gene pairs which are frequently associated with ESX loci (see Table S1 in the supplemental material) (29). This classification included ppe60, and we also investigated the expression of ppe19; these are the two ppe genes with the highest degree of similarity to ppe18, as discussed above. For all the additional 35 pe and ppe genes analyzed, no change in expression was associated with disruption of Rv0485 (see Table S2 in the supplemental material). This indicates that, despite apparent similarities in terms of sequence homologies, genomic arrangement, and association with ESX loci, individual pe and ppe genes function within distinct regulatory networks.
To assess whether the effects of Rv0485 on pe13 and ppe18 expression are mediated by direct binding of Rv0485 to putative promoter regions upstream of pe13 and ppe18, we attempted to purify recombinant Rv0485. However, although we were able to express and purify recombinant protein, this proved to be extremely insoluble and underwent rapid precipitation. Despite testing a wide range of purification methods, buffer conditions and additives, different tags and tag positions, and alternative hosts for protein expression (data not shown), we were unable to obtain soluble purified protein for further analysis. This result is consistent with a high calculated probability of Rv0485 insolubility using PROSO (59). We were therefore unable to experimentally determine whether Rv0485 is able to directly bind DNA. Nonetheless, bioinformatic investigation together with examination of mRNA expression levels suggests that Rv0485 is a transcription factor which regulates the expression of a defined subset of M. tuberculosis genes.
Since disruption of regulatory networks can frequently have an impact on strain phenotype (2, 27, 32, 62, 63), we went on to investigate the biological effect of disruption of Rv0485. First, we assessed the in vivo growth phenotype of the wild-type versus Rv0485::Tn strains in the mouse model. BALB/cJ mice were infected intravenously with equivalent doses of the wild type or Rv0485::Tn strain, and bacterial burdens in organs were determined at different time points following infection (Fig. 4A and B). In both lungs and spleens, there was no discernible difference in bacterial burden in organs at 3 weeks postinfection (P values of 0.4322 and 0.5468 for lungs and spleen, respectively). At 8 weeks postinfection, the mutant strain demonstrated only a moderate growth defect in lungs and spleens (P values of 0.0038 and 0.0123, respectively). The two strains therefore demonstrated equivalent rates of increase in spleen and lung in early infection (up to 3 weeks). Although the subsequent growth rate of the mutant was slightly reduced, the difference was very moderate and was not observed in repeat experiments. Notably, the mutant did not exhibit any inherent in vitro growth defect (data not shown). In contrast to the similar growth profiles, a remarkable reduction in lung pathology and extension of survival was associated with disruption of Rv0485 (Fig. 4C and D). Mice infected with the mutant strain demonstrated significantly less severe lung pathology than wild-type-infected mice. Larger and more numerous lesions were evident on gross examination of wild-type-infected versus mutant-infected lungs (not shown). Lungs from wild-type-infected mice demonstrated a reduced proportion of intact alveolar space relative to the Rv0485::Tn group (<50% in wild-type-infected versus >90% in mutant-infected mice).
To determine the longer-term consequences of the observed differences in immunopathology resulting from disruption of Rv0485, we assessed survival times of mice infected with the wild-type versus Rv0485::Tn strains. Mice infected with wild-type M. tuberculosis succumbed to disease significantly sooner (P < 0.0004), with a median survival time of 98.5 days (Fig. (Fig.4E).4E). In contrast, the median survival time of Rv0485::Tn-infected animals was 419.5 days. To rule out the possibility that this difference was the result of clearance of the mutant strain, the bacterial burden in organs of mice from the mutant-infected group was enumerated at 150 days postinfection (Fig. 4A and B). High numbers of bacteria were detected in lungs and spleens in all animals, with negligible changes in organ bacterial burden between 2 months and >5 months postinfection. These findings suggest that disruption of Rv0485 alters the immunomodulatory characteristics of the mutant strain rather than impacting growth or strain clearance.
To verify that the observed phenotype is attributable to disruption of Rv0485 and not to another unintended mutation, we repeated the shorter (8-week) infection experiment with the inclusion of the complemented strain. Growth kinetics (Fig. (Fig.5A)5A) and subsequent plating onto selective media (with and without apramycin) (data not shown) suggest that the complementing plasmid is not stably maintained in all bacteria during in vivo growth. Nonetheless, even suboptimal complementation of the Rv0485::Tn mutant with full-length Rv0485 restored the ability of the strain to cause tissue damage (Fig. 5B to D and and6).6). This confirmed that disruption of Rv0485 is indeed responsible for the observed phenotype.
To further explore how disruption of Rv0485 may modulate the immune response, we utilized the BALB/cJ SCID mouse model. Although these mice possess an intact innate immune response, they lack T cells and B cells and therefore cannot coordinate a functional adaptive immune response. BALB/cJ SCID mice were therefore used to test whether the observed phenotypic differences were attributable to the adaptive immune response. SCID mice were infected with equivalent doses of the wild-type and Rv0485::Tn strains, and organ bacterial burden and survival were assessed. Once again, despite statistically equivalent bacterial burdens in organs (Fig. 7A and B), a significant difference in survival times was observed (P < 0.0001), with a median survival of 28.0 days in wild-type-infected mice compared to 33.5 days in Rv0485::Tn-infected animals (Fig. (Fig.7C).7C). Although the relatively small difference in survival times does not rule out a role for the adaptive immune response, this finding suggests that disruption of Rv0485 alters the ability of M. tuberculosis to elicit innate immune responses.
To further explore whether Rv0485 contributes to the subversion of early immune responses, we investigated the response of the murine macrophage-like J774A.1 cell line to infection with wild-type, mutant, or complemented strains (Fig. (Fig.8).8). First, we demonstrated that disruption of Rv0485 did not compromise the ability of the mutant strain to survive in macrophages (not shown). Second, we showed that macrophages infected with Rv0485::Tn secreted reduced amounts of the proinflammatory cytokines tumor necrosis factor alpha (TNF-α) (P = 0.0728) and interleukin-6 (IL-6) (P = 0.0101) relative to wild-type-infected cells. Conversely, infection of macrophages with Rv0485::Tn + Rv0485 elicited higher levels of TNF-α and IL-6 production than by wild-type-infected macrophages. This finding supports a role for the Rv0485 regulon in modulating innate immune responses.
The PE and PPE families of Mycobacterium tuberculosis have been the subject of much speculation since their formal identification in 1998, when they were proposed to play a role in antigenic variation (15). Since then, studies have suggested that individual pe and ppe gene family members may be involved in different aspects of mycobacterial pathogenesis, for example, in granuloma and macrophage persistence (48), inhibition of antigen processing (17), acid resistance, vacuole acidification (37), and induction of apoptosis and proinflammatory cytokine secretion (6). It has been established that a number of PE/PPE proteins are either cell wall associated or secreted into the extracellular milieu (1, 5, 18, 19, 26, 53), ideally placing them to interact with host factors. Their secretion may be mediated by the ESX secretory apparatus (1) or other undetermined mechanisms. Emerging data thus increasingly support a role for these proteins in interaction with components of the innate immune response, although a coherent picture has yet to emerge.
To move toward an understanding of the precise role of PE/PPE proteins in the infectious process, we sought to identify and characterize regulators of pe and ppe expression. As a starting point, we screened for regulators of the pe13 and ppe18 gene pair. High-density transposon mutagenesis screening suggests that neither of these genes is essential for in vitro growth or in vivo survival (55, 56). However, it has been demonstrated that the proteins encoded by these genes are expressed and recognized in M. tuberculosis-infected and M. bovis-infected hosts, suggesting that they do play a role during infection (22, 40). Of particular note, PPE18 (also known as Mtb39A and Rv1196) was identified by serological expression screening (22) and is undergoing evaluation as part of the promising polyprotein vaccine candidate Mtb72f (http://clinicaltrials.gov/ct2/show/NCT00600782) (58, 64).
Using a transposon mutagenesis approach in conjunction with a LacZ reporter fused to the region upstream of the pe13 and ppe18 operon, we identified the putative transcriptional regulator, Rv0485. This protein is predicted to be a member of the NagC/XylR family of transcriptional regulators and is highly conserved in mycobacteria and other closely related species.
We confirmed that disruption of Rv0485 resulted in reduced expression of the pe13 and ppe18 gene pair, and this effect was reversed upon complementation with full-length Rv0485. Whole-genome microarray and real-time qRT-PCR expression analysis demonstrates that a limited number of other genes were affected by the Rv0485 mutation. Only six other unique genes were downregulated, and five adjacent genes (likely a single operon) were upregulated in the mutant strain. Interestingly, no other pe or ppe genes were affected, supporting previous suggestions that there is no global regulation of these gene families (65). This suggests that pe and ppe regulatory networks are very complex, which may allow for a high degree of flexibility in the PE/PPE expression repertoire.
The highly insoluble nature of the purified recombinant protein precluded DNA binding and structural studies. However, sequence comparisons with other known structures show strong similarity to the E. coli DNA-binding protein Mlc (44). In addition, sequence alignments show a high degree of conservation within a predicted HTH region. Although this remains to be experimentally verified, bioinformatic analysis suggests that Rv0485 may well bind DNA directly. In addition, expression analysis demonstrates that disruption of Rv0485 impacts on the transcriptional profile of the resultant mutant.
In recent years, it has been demonstrated that perturbing M. tuberculosis regulatory pathways can have striking consequences in terms of in vivo phenotype. Examples include mutations in the sigma factors SigC, SigE, SigF, and SigH and in the transcription factor WhiB3, which all result in decreased immunopathology and prolonged host survival despite apparently normal bacterial replication and survival within the host (2, 27, 32, 62, 63). In this study, we demonstrated that the Rv0485 mutant is also a member of this phenotypic class. The mutant is able to replicate normally and persist within host tissues, but it is defective in its ability to invoke a deleterious immunopathological response. This results in a striking and substantially prolonged survival of the infected hosts.
Although the adaptive immune response may play a role in containing infection with the Rv0485::Tn strain, this is not the only contributing factor, as shown by infection of immunocompromised SCID mice. These mice lack T cells and B cells and effectively do not possess a functional adaptive immune response. In this model, we observe similar bacterial replication rates but, once again, a significant difference in host survival times. This suggests that the observed in vivo phenotype is at least partly attributable to differences in the ability of the Rv0485::Tn mutant to invoke or respond to innate immune mechanisms. This is consistent with the finding that the Rv0485::Tn mutant induces lower levels of the proinflammatory cytokines TNF-α and IL-6 upon macrophage infection. The findings described above do not exclude a role for the adaptive immune response. Indeed, the Rv0485::Tn mutant is able to overcome the innate immune response in SCID mice within a relatively short time. Rv0485 may therefore influence the ability of the innate immune system to launch an appropriate adaptive immune response and tip the balance of the host response toward an environment favorable for bacterial persistence with minimal tissue damage.
Previously published microarray data suggest that Rv0485 is upregulated upon macrophage infection and in iron-limited media (3, 46). This may suggest that genes usually activated by Rv0485 are involved in inducing the typical macrophage response to infection. Interestingly, ppe18 is also upregulated under conditions of iron limitation (3). It is tempting to speculate that the dampening of macrophage responses and reduced tissue pathology in response to infection with Rv0485::Tn is due to reduced expression of PE13/PPE18. Admittedly, it would be surprising for disruption of only two genes to have such a marked effect, particularly in such large gene families where there is expected to be considerable functional redundancy. Nevertheless, available expression data suggest that only small subsets of these genes are affected by any one condition (65). It may well be the case that their expression is tightly regulated in response to changing environmental stresses, with different subsets of the gene families playing distinct roles at different stages of the infection.
Additional genes downregulated in the Rv0485::Tn mutant include the following three members of the so-called dormancy regulon (66): Rv2625, Rv2626, and Rv3132. Rv2626 has been shown to elicit T-cell and B-cell responses in M. tuberculosis-infected mice (52), and Rv2625 and Rv3132 are predicted to encode cell wall-associated proteins. Diminished expression of these may also act to reduce inflammatory responses and antigen-specific lymphocyte recruitment, with attendant reduction in tissue immunopathology. These three genes have been shown to be upregulated under conditions mimicking mycobacterial dormancy (66). This raises the interesting question of whether the Rv0485::Tn mutant will be compromised in its ability to respond to and survive physiological conditions approximating dormancy.
Naturally, genes with increased expression in the Rv0485::Tn mutant may also play a role in the observed phenotype. For example, Rv2392 (CysH; 5′ adenosinephosphosulfate reductase) has been shown to play a role in resistance to reactive nitrogen and oxygen species (57). Upregulation of this gene along with other members of the operon may render the mutant more capable of minimizing the deleterious effects of oxidative and nitrosative stress. It should be noted that the expression analysis was performed using mRNA isolated from a single experimental condition (i.e., mid-log-phase growth), and as such may not reflect the entirety of the Rv0485 regulon.
In summary, we have demonstrated that the predicted transcription factor Rv0485 impacts the expression of pe13 and ppe18 and plays a critical role in M. tuberculosis virulence. The precise contribution of pe13 and ppe18 and other genes in the Rv0485 regulon to virulence remains to be determined and will form the focus of future studies.
We are grateful to Bo-Shiun Yan and Jyothi Rengarajan for helpful discussions, Cassandra Krone and Ilona Breiterene for technical assistance, and Jason Hinds and Denise Waldron (Bacterial Microarray Group at St. Georges) for assistance with microarray analysis.
This work was supported by National Institutes of Health grant AI023545 and a Wellcome Trust Research Career Development Fellowship to S.L.S.
Editor: J. L. Flynn
Published ahead of print on 3 August 2009.
§Supplemental material for this article may be found at http://iai.asm.org/.