SGM Journals
HOST RESPONSE

Mycobacterium tuberculosis strains disrupted in mce3 and mce4 operons are attenuated in mice

Ryan H. Senaratne, Ben Sidders, Patricia Sequeira, Grainne Saunders, Kathleen Dunphy, Olivera Marjanovic, J. Rachel Reader, Patricia Lima, Stephen Chan, Sharon Kendall, Johnjoe McFadden, Lee W. Riley

  • 1School of Public Health, University of California, Berkeley, CA 94720, USA
  • 2School of Biomedical and Molecular Sciences, University of Surrey, Guildford GU2 7XH, UK
  • 3Department of Pathology and Infectious Disease, Royal Veterinary College, Royal College Street, London NW1 0TU, UK
  • 4Comparative Pathology Laboratory, School of Veterinary Medicine, University of California, Davis, CA 95616, USA
  • Correspondence
    Lee W. Riley
    lwriley{at}berkeley.edu

    • Journal of Medical Microbiology 2008; 57(2):164–170 · https://doi.org/10.1099/jmm.0.47454-0

    View at publisher PubMed

    Abstract

    The Mycobacterium tuberculosis genome contains four copies of an operon called mce (mce1–4). Previously we reported that M. tuberculosis disrupted in the mce1 operon is more virulent than wild-type M. tuberculosis in mice. We generated single deletion mutants in mce3 (Δmce3) and mce4 (Δmce4) operons and a double deletion mutant (Δmce3/4). Similar doubling times and growth characteristics were observed for all mutants and the wild-type (parent) M. tuberculosis H37Rv strain in culture and in macrophages. In addition, similar bacterial burdens were detected in organs from mice infected with Δmce3 and the parent strain. However, the bacterial burdens of mice infected with Δmce4 and Δmce 3/4 were less than those of mice infected with the parent strain. The median survival times of mice infected with wild-type M. tuberculosis, Δmce3, Δmce4 and Δmce3/4 were 40.5, 46, 58 and 62 weeks, respectively. Histopathological examination of lungs at 15 weeks post-infection showed that the extent of the lung lesions was less prominent in mice infected with Δmce4 and Δmce 3/4 mutants than in mice infected with the other two strains. These observations suggest that the mce3 and mce4 operons have a role distinct from that of mce1 for in vivo survival of M. tuberculosis.

    • ↵†These authors contributed equally to this work.

    INTRODUCTION

    It is estimated that 2 billion people worldwide are latently infected with Mycobacterium tuberculosis, the aetiologic agent of tuberculosis (Dye et al., 2002), and 2–23 % of these latently infected persons ultimately develop active disease in their lifetime (Parrish et al., 1998). We previously showed that an M. tuberculosis strain disrupted in an operon called mce1 (Δmce1) was more virulent in mice than its parent wild-type (WT) strain, H37Rv; the mutant was also diminished in its ability to induce a Th1-type immune response (Shimono et al., 2003). We thus suggested that mce1 genes may temper the ability of M. tuberculosis to cause overt disease and play a role in establishing latent infection in mice (Shimono et al., 2003).

    The mce1 operon is a member of a family of related operons comprising mce1, 2, 3 and 4 containing homologous genes arranged similarly (Cole et al., 1998; Tekaia et al., 1999). Six of the mce1 genes (mce1A–F) encode proteins that localize to the cell wall (Chitale et al., 2001; Shimono et al., 2003; Tekaia et al., 1999). One of the mce1 proteins (Mce1A) was previously shown to confer upon a nonpathogenic Escherichia coli strain an ability to enter HeLa cells (Arruda et al., 1993; Chitale et al., 2001). The corresponding protein in the mce2 operon (Mce2A), which is 67 % similar in amino acid sequence to Mce1A, did not exhibit this activity. Santangelo et al. (2002) have shown that mce3 genes are negatively regulated by a gene (Rv1963) belonging to the tetR family. In contrast, mce1 has a negative transcriptional regulator (mce1R) belonging to the GntR family at the corresponding position (Casali et al., 2006; Cole et al., 1998). Recently we have shown that the M. tuberculosis strain disrupted in mce1R exhibits enhanced virulence in mice, indicating that the level of expression of the mce1 products has a profound effect on the clinical outcome of the infection (Uchida et al., 2007). Gioffre et al. (2005) have studied Δmce1, Δmce2 and Δmce3 in BALB/c mice. The above authors used two methods of infection: intratracheal and intraperitoneal routes. When mice were infected via the intratracheal route, all three mce mutants showed reduced bacterial counts; when mice were infected via the intraperitoneal route, Δmce1 showed 50 % increased bacterial counts while the counts decreased for Δmce2 and remained unchanged for Δmce3. The mouse survival study was carried out only up to 20 weeks (Gioffre et al., 2005). Joshi et al. (2006) studied Δmce4 in C57BL/6 mice using intravenous infection. They co-infected mice with Δmce4 and WT H37Rv and studied the infection up to 100 days. Thus these other studies have not assessed the long-term survival of mce2–4 operon mutations in mice. The long-term effects of mce mutations on mice are important to assess the suggested role of the mce operons in latency. We therefore studied the long-term survival (>60 weeks) of mice infected with mutants disrupted in mce3 and mce4 using a low-dose aerosol infection model.

    METHODS

    Mycobacteria and culture conditions.

    WT M. tuberculosis H37Rv and its derivative strains (mce mutants) were grown in Middlebrook 7H9 broth containing 10 % ADC (Becton Dickinson), 0.2 % glycerol and 0.05 % Tween 80 (7H9-ADCT) or on Middlebrook 7H11 agar containing OADC (Becton Dickinson), 0.5 % glycerol and the antifungal agent cycloheximide (100 μg ml−1) (Sigma-Aldrich). Bacteria were passed through a 5 μm pore filter to prepare single cell suspensions prior to mouse infection and measuring growth curves.

    Generation of mce mutants.

    The M. tuberculosis H37Rv mce mutants were constructed by the method of Parish & Stoker (2000). Deleted alleles were created by amplifying (approx.) 1 kb regions upstream and downstream of each operon. Following digestion with the appropriate restriction enzymes, these PCR products were subcloned into the p2NIL vector in tandem. Mutant selection and additional vector information have been previously described by Parish & Stoker (2000). Deletion mutations were confirmed by Southern blot hybridization.

    Mouse infections.

    Eight-week-old C57BL/6 mice (Jackson Laboratories) were infected with the M. tuberculosis strains via inhalation by the Inhalation Exposure System (Glas-col). The inoculum doses were assessed from harvest of the right lungs of three mice (per infection) 24 h post-infection (p.i.). The dose of infection was 72–108 bacilli per lung for all infections. Lungs were homogenized and plated onto 7H11 agar, followed by enumeration of c.f.u. 21 days later. At different time points p.i., the right lung, liver and spleen from three mice were collected, homogenized in PBS-Tween (0.05 %), appropriately diluted, and plated onto 7H11 agar plates (supplemented as above). At 21 days, the bacterial load of each organ was determined by c.f.u. enumeration.

    Determination of mouse morbidity.

    Four groups of eight to nine mice each, infected with each of the above strains, were followed until they exhibited moribund features that occur just before death. At this point, mice were anaesthetized with a mixture of ketamine HCl, xylazine and acepromazine injected subcutaneously, and then euthanized by cervical dislocation. Loss of weight accompanied by failure to groom, ruffled fur and lethargy were used to assess morbidity, in addition (in some instances) to the recommendation of the veterinary staff of the North Animal Facility of UC Berkeley. The health of the mice was monitored daily by the above veterinary staff.

    Determination of survival and induction of cytokines in macrophages by mce mutants.

    The RAW 264.7 murine macrophage-like cell line (ATCC) was cultured and maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco) supplemented with 10 % fetal bovine serum (Omega Scientific) at 37 °C in a 5 % CO2 humidified incubator. Cells were plated at 2×105 cells per well in 24-well tissue culture plates. The macrophages were incubated (i.e. infected) with either 2×105, 2×106 or 2×107 bacteria for 6 h. After 6 h of infection, macrophages were washed three times with DMEM to remove extracellular bacteria. To examine the intracellular invasion, growth and survival of bacteria, we lysed the macrophages with 1 ml PBS-0.5 % Triton X-100 and their serial dilutions were plated onto 7H11 agar. This was performed 6, 48 and 72 h p.i. c.f.u. on plates were enumerated 21 days after plating. Supernatants of the above cell cultures were used to measure TNFα, IL-12, IL-6, IL-10 and monocyte chemoattractant protein 1 (MCP-1) produced by macrophages in response to infection with WT or the mce mutant strains. Lipopolysaccharide (1 μg ml−1) (Sigma) was used as a positive control and uninfected macrophages served as negative controls for each experiment. The above cytokines and MCP-1 were measured by ELISA with the reagents purchased from eBioscience.

    Histology.

    The mouse left lung fixed in 10 % neutral (PBS) buffer formalin was embedded in paraffin, sectioned and stained for histology with either haematoxylin and eosin (H&E) or the Ziehl–Neelsen technique. Sectioning and staining were performed by Histology Consultation Services, Everson, Washington, USA. For comparative purposes, sections were obtained from the same regions of all lungs; three sections were obtained from each lung from three mice. Sections obtained from the top and the bottom parts of the lung were stained by H&E, while the section obtained from the middle region was stained by the Ziehl–Neelsen technique. The histopathology of each lung was assessed for two to four H&E-stained sections and one to two sections stained by the Ziehl–Neelsen technique. Histopathology of 6 weeks and 15 weeks p.i. lungs was analysed. Pathological analysis was done by a veterinary pathologist from the School of Veterinary Medicine UC Davis, USA.

    Statistics.

    Mouse survival was compared by Kaplan–Meier curves. The mean c.f.u. counts of M. tuberculosis recovered from organs of each mouse group (three per group) were compared by the Student's t-test.

    RESULTS

    Generation of mce mutants

    We generated three mutants by in-frame deletion of the mce3, mce4 and both mce3 and mce4 operons in M. tuberculosis H37Rv (Δmce3, Δmce4 and Δmce3/4, respectively). The mutations were designed so that eight genes (yrbE3A–mce3F) within mce3 (Tekaia et al., 1999) and eight genes and the first 250 bp of the ninth gene (yrbE4A to the first 250 bp of Rv3493c) within mce4 (Tekaia et al., 1999) would be deleted (Figs 1⇓ and 2⇓). Deletions of the genes were confirmed by Southern blot analysis (Figs 1⇓ and 2⇓) and PCR (data not shown). The resulting mutants, Δmce3, Δmce4 and Δmce3/4, displayed in vitro growth characteristics similar to those of the WT parent strain (data not shown).

    Figure image not available in archive
    Fig. 1.

    Southern blot analysis of the mce3 operon deletion. (a) Genomic organization of the region surrounding the mce3 operon in the WT (M. tuberculosis H37Rv). Black arrows depict the deleted genes [yrbE3A–mce3F: TubercuList coordinates () 2207697–2216573] of the mce3 operon. Two black boxes (labelled 5′ probe and 3′ probe) indicate regions of DNA probes used in Southern blot analysis. (b) Genomic organization of the region surrounding the mce3 deletion in Δmce3 and Δmce3/4. White arrows in (a) and (b) depict the genes neighbouring the region of the deletion: mce3R (5′ end) and Rv1972 (3′ end). Genomic DNA was digested with PvuI for Southern blot analysis. (c) Southern blot analysis of genomic DNA from WT (lane 2), Δmce3 (lane 3) and Δmce3/4 (lane 4). DIG-labelled molecular mass standards II (lane 1) and VII (lane 5) are from Roche Diagnostics. Arrows indicate the molecular mass standards in kbp.

    Figure image not available in archive
    Fig. 2.

    Southern blot analysis of the mce4 operon deletion. (a) Genomic organization of the region surrounding the mce4 operon in the WT (M. tuberculosis H37Rv). Black arrows depict the completely or partially deleted genes [first 250 bp of Rv3493c to yrbE4A: TubercuList () coordinates 3911426–3920856] of the mce4 operon. Two black boxes (labelled 5′ probe and 3′ probe) indicate the regions of DNA probes used in Southern blot analysis. (b) Genomic organization of the region surrounding the mce4 deletion in Δmce4 and Δmce3/4. The black arrow depicts the partially deleted Rv3493c (479 bp of the 3′ end). In (a) and (b), white arrows depict the genes neighbouring the region of deletion: Rv3492c (5′ end) and Rv3502c (3′ end). Genomic DNA was digested with PvuII for Southern blot analysis. (c) Southern blot analysis of genomic DNA from WT (lane 2), Δmce3/4 (lane 3) and Δmce4 (lane 5). DIG-labelled molecular mass standards II (lane 1) and VII (lanes 4 and 6) are from Roche Diagnostics. Arrows indicate the molecular mass standards in kbp.

    Survival and induction of cytokines in RAW macrophages by mce mutants

    Previously we reported that, upon infection, in comparison to WT, the mce1 mutant induces less TNF-α, IL-6 and MCP-1 in RAW macrophages (Shimono et al., 2003). Similarly, we investigated the induction of the above cytokines and MCP-1 by RAW cells infected with Δmce3, Δmce4 and Δmce3/4 in comparison to WT. In addition to the above two cytokines and chemokine, we also investigated the induction of two additional cytokines, IL-10 and IL-12. Macrophages were infected with three different inoculum doses (1 macrophage to either 1, 10 or 100 bacilli) of M. tuberculosis strains. In all three inoculum doses, we did not observe any reproducible difference in cytokine or chemokine induction by Δmce3, Δmce4 and Δmce3/4 compared to WT (data not shown). All three mce mutants replicated similarly to WT in RAW macrophages (data not shown).

    Bacterial burden and survival of mice

    Similar bacterial burdens were detected in organs from mice infected with Δmce3 or the WT at all time points (Fig. 3⇓). However, at 15 weeks p.i., recovery of Δmce4 and Δmce3/4 c.f.u. from mouse lungs was significantly less than that of the WT (P <0.05). All other c.f.u. recoveries (from all three organs) at 15 weeks p.i. from mouse organs infected with mce mutants were not significantly different compared to c.f.u. recoveries from mouse organs infected with the WT (Fig. 3⇓).

    Figure image not available in archive
    Fig. 3.

    Bacterial burdens of mouse organs infected with the mce mutants. Recovery of c.f.u. from the right lung at different time points p.i. (a), and recovery of c.f.u. at 105 days p.i. from the right lung [re-representation of (a) at 105 p.i.] (b), spleen (c) and liver (d) for WT, Δmce3, Δmce4 or Δmce3/4 from C57BL/6 mice after aerosol infection (n=3 mice per group per time point).

    The long-term survival time of mice infected with Δmce3/4 or Δmce4 was significantly longer than that of mice infected with the WT or Δmce3. The median survival time of WT-infected mice (n=8) was 40.5 weeks compared to 46 weeks (P=0.02) for Δmce3-infected mice (n=8). By comparison with WT infected mice, the median survival times for Δmce3/4-infected mice (n=9) and Δmce4-infected mice (n=9) were 62 weeks (P <0.0001) and 58 weeks (P <0.0001), respectively. The statistical differences between the survival times of mice infected with Δmce3 versus Δmce3/4 and Δmce3 versus Δmce4 strains were also significant (P=0.0003 and 0.003, respectively). Additionally, the differences between the survival times of mice infected with Δmce4 versus Δmce3/4 strains were significant (P=0.029) (Fig. 4⇓).

    Figure image not available in archive
    Fig. 4.

    Survival of mice infected with mce mutants. Survival of C57BL/6 mice after aerosol infection with WT and mce mutants; n=8 for WT, n=8 for Δmce3, n=9 for Δmce4 and n=9 for Δmce3/4-infected groups.

    Histopathological analysis

    Histopathological examination of lungs at 15 weeks p.i. showed that the extent of the lung lesions (granulomatous interstitial pneumonia) was less prominent in Δmce4- and Δmce3/4-infected mice than in mice infected with the other two strains (Fig. 5a⇓). At 15 weeks p.i., the average area of the lung parenchymal lesions in mice infected with WT, Δmce3, Δmce4 or Δmce3/4 was 53 %, 54 %, 19 % and 22 %, respectively. In all mouse groups, granulomatous interstitial pneumonia was observed, which initially (at 6 weeks p.i.) began as areas of interstitial expansion and demarcated nodules, sometimes progressing to coalescing nodules. In the WT- and Δmce3-infected mouse groups, the lesions progressed over time to become more diffuse and less well demarcated. In the Δmce4- and Δmce3/4- infected groups, the lesions remained mostly as nodules and did not progress significantly. The inflammatory infiltrate in all four groups was similar and was predominantly a mixture of foamy and epithelioid macrophages intermingled with minimal numbers of neutrophils (Fig. 5b⇓). Although there was a small but significant difference in the survival times between the WT- and Δmce3-infected groups, no differences could be detected in the histopathological lesions between them.

    Figure image not available in archive
    Fig. 5.

    Histology of pulmonary lesions in infected mice. Sections of C57BL/6 mouse lungs magnified ×20 (a) and ×200 (b) stained with H&E. Bars, 300 μm (a) and 30 μm (b). Mouse lungs from all groups were harvested at 15 weeks p.i. All the mice were infected as in Fig. 3. The photographs were taken by a veterinary pathologist so as to represent each group. Δmce4- and Δmce3/4-infected mice had pulmonary lesions that were nodular and more contained, involving a smaller area of pulmonary parenchyma, than those of Δmce3- and WT-infected mice, where the lesions were more diffuse and involved a larger percentage of the parenchyma.

    DISCUSSION

    This study found that the mce3 and mce4 operon mutants were attenuated in mice compared to WT M. tuberculosis H37Rv, as evidenced by the longer survival times of mice infected with the mutants. This observation suggests that the in vivo function of these operons is distinct from that of the mce1 operon, whose mutation was previously shown to cause increased mortality in mice (Shimono et al., 2003). It should be noted that the current study used C57BL/6 mice instead of BALB/c mice used by Shimono et al. (2003). Since C57BL/6 mice are relatively more resistant than BALB/c mice to M. tuberculosis infection, the differences in outcome could have been influenced by mouse species differences. However, a recent study showed that the mce1 operon mutant has the same virulence phenotype in C57BL/6 mice as it does in BALB/c mice (Lima et al., 2007).

    The longer survival times of mice infected with Δmce4 suggest that the attenuation of Δmce4 is greater than that of Δmce3. This is also indicated by the differences in c.f.u. recovery at 15 weeks p.i. from mouse lungs (Fig. 3⇑) and the corresponding histopathology (Fig. 5⇑). At this time point, the above two parameters did not indicate any attenuation of Δmce3 compared to WT in mice. However, the survival times of mice infected with Δmce3 were significantly longer than those of the mice infected with WT. There was also a small but significant difference between the survival times of Δmce3/4- versus Δmce4-infected mice. Therefore, it is possible that the greater attenuation of the double mce3/4 mutant is due to the combined effect of mutations in both mce3 and mce4 operons. However, the mouse survival, lung pathology and bacterial burden data suggest that the attenuation of Δmce4 in mice is much more prominent than that of Δmce3. Nevertheless, all of the above observations indicate that mce3 and mce4 operon mutants behave differently from the mce1 operon mutant in mice, suggesting that, despite similarity in gene sequences and arrangement, the functions of the mce3 and mce4 operons are distinct from that of the mce1 operon.

    The decreased lung bacterial burden in mice infected with Δmce4 or Δmce3/4 mutants compared to that in mice infected with WT and Δmce3 at 15 weeks p.i. may be because either (1) the disruption of mce4 causes bacteria to replicate more slowly in host cells or (2) the mce4 operon mutant is killed more rapidly when the bacteria first encounter the host adaptive immune response. Since Δmce4 can replicate similarly to other strains in RAW macrophages (data not shown), it is unlikely that Δmce4 is more susceptible to the antibacterial activity inside the host cell. Δmce4-infected mouse lung has less extensive granulomatous pneumonia (i.e. fewer immune cells). One explanation is that the mce4 operon-related products attract proinflammatory cells to the lung. However, as observed in RAW cells, the absence of the mce4 operon did not have any effect on the ability of M. tuberculosis to induce or suppress TNF-α, IL-6, IL-10, IL-12 and MCP-1 in ex vivo-infected macrophages. Thus the diminished proinflammatory cell response in lungs of mice infected with Δmce4 and Δmce3/4 and their reduced c.f.u. counts may be due to a decreased replicative ability of these strains under the host adaptive immune response.

    Recently, Kumar et al. (2003) detected the expression of mce1, 3 and 4 operons in tubercle material collected from infected animals (guinea pigs and rabbits). These observations support our findings that mce operons other than mce1 are expressed during the disease state. In addition, Ahmad et al. (2004) demonstrated the expression of several mce3 genes during natural infection of humans infected with M. tuberculosis.

    Our results support the findings of a recent study by Gioffre et al. (2005) that reported that the mce3 operon mutant was attenuated in mice. They did not examine Δmce4. Their survival study, however, was carried out only up to 20 weeks. Hence the long-term effect of their Δmce3 cannot be compared with our results. However, in that study, attenuation was observed only when mice were infected via the intratracheal route, and not when the intraperitoneal route was used. They suggested that the route of infection may make a difference in infection outcome. This study used a more physiologically relevant aerosol route of infection. The route of infection (aerosol vs intravenous) with another mutant (Δmce1R) in mice, however, did not make any difference to clinical outcome or survival (Uchida et al., 2007). Additionally, Joshi et al. (2006) studied the mice infected with mce mutants up to 100 days only, while we studied the survival of mice infected with Δmce3 and Δmce4 for >60 weeks. As with the study of Gioffre et al. (2005), we were unable to complement the mce2 and 3 operon mutants due to the large size of the deleted region.

    In conclusion, the studies described in this paper on mce3 and mce4 operons suggest that despite their similarity with the mce1 operon in gene organization and sequences, the mce3 and mce4 operon-encoded proteins have a role markedly distinct from mce1 operon-encoded products.

    Acknowledgments

    We thank Sally Cantrell, Nicola Casali and Lisa Morici for helpful discussion. This project was supported in part by the Ellison Medical Foundation and NIH R21AI063350. G. S. was supported by European Union grant ‘TB Vaccine Cluster’ Contract QLK2-CT-1999-01093.

    References

    1. Ahmad, S., El-Shazly, S., Mustafa, A. S. & Al-Attiyah, R. (2004). Mammalian cell-entry proteins encoded by the mce3 operon of Mycobacterium tuberculosis are expressed during natural infection in humans. Scand J Immunol 60, 382–391.
    2. Arruda, S., Bomfim, G., Knights, R., Huima-Byron, T. & Riley, L. W. (1993). Cloning of an M. tuberculosis DNA fragment associated with entry and survival inside cells. Science 261, 1454–1457.
    3. Casali, N., White, A. M. & Riley, L. W. (2006). Regulation of the Mycobacterium tuberculosis mce1 operon. J Bacteriol 188, 441–449.
    4. Chitale, S., Ehrt, S., Kawamura, I., Fujimura, T., Shimono, N., Anand, N., Lu, S., Cohen-Gould, L. & Riley, L. W. (2001). Recombinant Mycobacterium tuberculosis protein associated with mammalian cell entry. Cell Microbiol 3, 247–254.
    5. Cole, S. T., Brosch, R., Parkhill, J., Garnier, T., Churcher, C., Harris, D., Gordon, S. V., Eiglmeier, K., Gas, S. & other authors (1998). Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature 393, 537–544.
    6. Dye, C., Williams, B. G., Espinal, M. A. & Raviglione, M. C. (2002). Erasing the world's slow stain: strategies to beat multidrug-resistant tuberculosis. Science 295, 2042–2046.
    7. Gioffre, A., Infante, E., Aguilar, D., Santangelo, M. P., Klepp, L., Amadio, A., Meikle, V., Etchechoury, I., Romano, M. I. & other authors (2005). Mutation in mce operons attenuates Mycobacterium tuberculosis virulence. Microbes Infect 7, 325–334.
    8. Joshi, S. M., Pandey, A. K., Capite, N., Fortune, S. M., Rubin, E. J. & Sassetti, C. M. (2006). Characterization of mycobacterial virulence genes through genetic interaction mapping. Proc Natl Acad Sci U S A 103, 11760–11765.
    9. Kumar, A., Bose, M. & Brahmachari, V. (2003). Analysis of expression profile of mammalian cell entry (mce) operons of Mycobacterium tuberculosis. Infect Immun 71, 6083–6087.
    10. Lima, P., Sidders, B., Morici, L., Reader, R., Senaratne, R., Casali, N. & Riley, L. W. (2007). Enhanced mortality despite control of lung infection in mice aerogenically infected with a Mycobacterium tuberculosis mce1 operon mutant. Microbes Infect
    11. Parish, T. & Stoker, N. G. (2000). Use of a flexible cassette method to generate a double unmarked Mycobacterium tuberculosis tlyA plcABC mutant by gene replacement. Microbiology 146, 1969–1975.
    12. Parrish, N. M., Dick, J. D. & Bishai, W. R. (1998). Mechanisms of latency in Mycobacterium tuberculosis. Trends Microbiol 6, 107–112.
    13. Santangelo, M. P., Goldstein, J., Alito, A., Gioffre, A., Caimi, K., Zabal, O., Zumarraga, M., Romano, M. I., Cataldi, A. A. & Bigi, F. (2002). Negative transcriptional regulation of the mce3 operon in Mycobacterium tuberculosis. Microbiology 148, 2997–3006.
    14. Shimono, N., Morici, L., Casali, N., Cantrell, S., Sidders, B., Ehrt, S. & Riley, L. W. (2003). Hypervirulent mutant of Mycobacterium tuberculosis resulting from disruption of the mce1 operon. Proc Natl Acad Sci U S A 100, 15918–15923.
    15. Tekaia, F., Gordon, S. V., Garnier, T., Brosch, R., Barrell, B. G. & Cole, S. T. (1999). Analysis of the proteome of Mycobacterium tuberculosis in silico. Tuber Lung Dis 79, 329–342.
    16. Uchida, Y., Casali, N., White, A., Morici, L., Kendall, L. V. & Riley, L. W. (2007). Accelerated immunopathological response of mice infected with Mycobacterium tuberculosis disrupted in the mce1 operon negative transcriptional regulator. Cell Microbiol 9, 1275–1283.