|Home | About | Journals | Submit | Contact Us | Français|
The success of Mycobacterium tuberculosis (Mtb) as a human pathogen relies on its ability to resist eradication by the immune system of healthy individuals. Identification of mechanisms that enable Mtb to persist is key to limiting latent tuberculosis, which affects one third of the world’s population. Here we demonstrate that conditional gene silencing makes it possible to determine if an Mtb gene that is required for optimal growth in vitro is also important for virulence, and during which phase of an infection it is required. Application of this approach to prcBA, which encode the core of the mycobacterial proteasome, reveals an unpredicted essentiality of the core proteasome for persistence of Mtb during the chronic phase of infection in mice. Proteasome depletion also attenuated Mtb in interferon-γ-deficient mice, pointing to a function of the proteasome beyond defence against the adaptive immune response. Genes that are essential for growth in vitro and/or in vivo account for approximately 20% of Mtb’s genome. Conditional gene silencing could therefore facilitate the validation of up to 800 potential Mtb drug targets and improve our understanding of host-pathogen dynamics.
Many life-threatening infections are due to pathogens that commonly establish long-lasting latent infections before causing clinical disease. Approximately 1.8 billion people are estimated to be latently infected with one such pathogen, Mycobacterium tuberculosis (Mtb), and at least 5% of these infected individuals will eventually develop tuberculosis and require chemotherapy1. Mtb survives in an infected host despite the development of strong cell-mediated immunity. In mice, persistence of Mtb is observed as prolonged survival in lungs at nearly constant bacterial titres resulting in a chronic infection2. Mtb mutants that are attenuated during chronic infections in experimental animals can identify pathways that may be targeted for the eradication of persistent bacteria. Unfortunately, relatively little is known about the genetic repertoire that Mtb requires to avoid clearance by the immune system and to persist in its host. So far, persistence mutants - including those affected in intermediary metabolism3, cell envelope modification4, or heat shock control5 - have primarily been identified using deletion or transposon mutagenesis. However, this approach can neither determine which of the many genes that are essential for growth in vitro6 are required in vivo, nor identify which of the genes that are essential for growth in vivo7 are also required for persistence.
To facilitate the analysis of genes that are essential for growth we previously generated regulated mycobacterial expression systems that are controlled by tetracycline repressors (TetRs)8–10. These expression systems allowed the construction of conditional knockdown mutants of M. smegmatis and M. bovis BCG, in which expression of a single target gene can be silenced either by the addition (Tet-OFF system) or removal of tetracyclines (Tet-ON system)8, 10, 11. For this study we aimed to determine if Tet-ON and Tet-OFF systems also allow generating conditional knockdown mutants for virulent Mtb and if silencing of Mtb genes can be achieved during mouse infections. As the target for gene silencing we selected the mycobacterial 20S proteasome.
The Mtb 20S proteasome core consists of 14 α subunits encoded by prcA and 14 β subunits encoded by prcB, which assemble in an α7β7β7α7 complex12,13. Bacterial proteasomes, which are unique to actinomycetes14, 15, are not essential for in vitro growth of Streptomyces coelicolor, S. lividans, or M. smegmatis16–18. In contrast, prcA and prcB were predicted to be essential for optimal in vitro growth of Mtb6. Inactivation of two proteasome-associated genes, mpa and pafA, reduced Mtb virulence in mice, suggesting that the proteasome is also important for pathogenicity19, 20. Mpa encodes a proteasomal ATPase, which is homologous to ATPases found in the regulatory cap of the eukaryotic 26S proteasome and likely involved in substrate recognition, unfolding and translocation into the proteasome core15,21,22. A biochemical activity for PafA remains to be discovered, but it appears to play a role similar to that of Mpa in protein degradation23. Here we report the construction of mutants that allow conditional silencing of the proteasome α and β subunits and analyzed the role of the Mtb proteasome for in vitro growth, for survival under protein damaging conditions and during in vivo growth and persistence in mice.
To generate mutants that allow conditional silencing of the Mtb proteasome, we integrated Pmyc1tetO, a tetracycline responsive mycobacterial promoter8, 5′ of prcBA, generating strain H37Rv Pmyc1tetO:prcBA (Fig. 1a, b). Placement of Pmyc1tetO 5′ of prcBA did not alter proteasome expression or growth of Mtb (Supplementary Fig. 1). Next, we transformed H37Rv Pmyc1tetO:prcBA with plasmids expressing wild type Tet repressor (wtTetR) or reverse TetR (revTetR)10. wtTetR represses Pmyc1tetO in the absence of tetracyclines, whereas revTetR requires tetracyclines as cofactors for binding to and silencing of Pmyc1tetO24. Addition of anhydrotetracycline (atc) thus induces expression of prcBA in H37Rv Pmyc1tetO:prcBA/wtTetR (hereafter, PrcBATet-ON) and silences expression of prcBA in H37Rv Pmyc1tetO:prcBA/revTetR (PrcBATet-OFF).
We analyzed efficiencies and kinetics of prcBA silencing in PrcBATet-ON and PrcBATet-OFF using immunoblot, mRNA quantification and proteasome activity assays (Fig. 1c-f). By immunoblot, the β subunit was reduced to levels below the limit of detection following removal of atc from PrcBATet-ON cultures or addition of atc to PrcBATet-OFF cultures (Fig. 1c). PrcBA transcripts were reduced 15- to 20-fold after silencing in both mutants (Fig. 1d). Atc did not affect PrcB protein or prcBA transcript levels in the absence of wtTetR or revTetR (Fig. 1c, d), demonstrating that silencing of prcBA was TetR-dependent. To analyze proteasome activity we measured the peptidolytic activity of Mtb lysates against succinyl-Leu-Leu-Val-Tyr-aminomethyl coumarin (Suc- LLVY-Amc), a fluorescent peptide substrate of the Mtb proteasome13. Four days after silencing was initiated, proteasome activity was reduced ~100 and 10-fold in the PrcBATet-OFF and PrcBATet-ON mutants, respectively (Fig 1e). By 7 days proteasome activity was below the level of detection in both mutants. A similar difference in the kinetics of silencing proteasome expression between the Tet-ON and Tet-OFF strains was found in immunoblot analyses (Fig. 1f). Together, these experiments demonstrate that PrcBATet-ON and PrcBATet-OFF allowed rapid and efficient silencing of the mycobacterial proteasome by removing (wtTetR, PrcBATet-ON) or adding (revTetR, PrcBATet- OFF) atc, respectively. Silencing was slightly more efficient in the Tet-ON mutant but somewhat faster in the Tet-OFF mutant. These differences are likely due to the lower activity of revTetR compared to wtTetR in mycobacteria10 and residual atc bound to Mtb after transfer of a Tet-ON strain into medium not containing atc.
We next sought to determine the impact of prcBA silencing on in vitro growth and survival under nitrosative stress because previous studies predicted that the Mtb proteasome is essential for optimal growth on agar plates and required for defence against reactive nitrogen intermediates (RNI)6,19. Proteasome depletion impaired growth of Mtb on agar plates and, to a small degree, in liquid culture (Fig. 2a), and increased susceptibility to RNI (Fig. 2b). After exposure to acidified nitrite for 3 days, silencing in the PrcBATet-ON mutant led to a 10-fold decrease in survival (p=0.005, Fig. 2b, left panel). PrcBA silencing in the Tet-OFF mutant did not significantly reduce the number of bacteria recovered after RNI stress (p=0.116, Fig. 2b, middle panel) suggesting that silencing in this mutant was not sufficient to cause decreased survival in this assay.
Mutants in the proteasome-associated genes mpa and pafA were more resistant to H2O2 than wild type Mtb19 and similarly, depletion of the proteasome resulted in increased resistance to oxidative stress (p=0.002 for PrcBATet-ON and p=0.007 for PrcBATet-OFF, Fig. 2c). The proteasome did not confer resistance or susceptibility to another physiological stress, acidic pH (data not shown). Atc did not affect survival of the control strain (H37Rv Pmyc1tetO:prcBA/no TetR) in the presence of acidified nitrite or H2O2 (Fig. 2b, right panel). In summary, these data support the proposed essentiality of prcBA for optimal growth on agar plates and the proposed functional association of Mpa and PrcBA.
We next used PrcBATet-ON and PrcBATet-OFF to determine the role of the proteasome during growth and persistence of Mtb in mice. Tet-ON systems have been applied to infection models for pathogens such as Staphylococcus aureus and Yersinia pestis25,26 but have not heretofore allowed analysis of Mtb mutants during mouse infections. Moreover, application of revTetR-dependent Tet- OFF systems to the in vivo analysis of a bacterial pathogen has not been reported. We first constructed an Mtb reporter strain containing gfp (encoding green fluorescent protein) controlled by Pmyc1tetO and wtTetR (GFPTet-ON). Feeding GFPTet-ON infected mice doxycycline, a tetracycline derivative, induced green fluorescence in pulmonary Mtb, while no fluorescent bacteria were detected in mice that had not received doxycycline (Fig. 3). This demonstrated that the mycobacterial Tet-ON system allows control of Mtb gene expression during mouse infections. Next, we infected mice with the prcBA mutants and initiated proteasome silencing at day 1 (PrcBATet-ON), day 7 (PrcBATet-OFF) or day 21 (PrcBATet-OFF) in different mice (Fig. 4). In C57BL/6 mice, the onset of the adaptive immune response at around 3 weeks post infection results in control of Mtb replication, such that titres remain stable for many months27 (Fig. 4d). In contrast, bacterial counts in lungs measured 112 days post infection were reduced by 140-fold (PrcBATet-ON) and 500- fold (PrcBATet-OFF) as a consequence of prcBA silencing (p=0.028 and p=0.0038) (Fig. 4a, b). Even when prcBA silencing was only initiated at the onset of the chronic phase of the infection (day 21, PrcBATet-OFF), bacterial loads in lungs and spleens were 36-fold and 22-fold lower than those of the control groups at day 112 (Fig. 4b, c). Thus, the core proteasome is essential for persistence of Mtb in mice.
A less dramatic effect of prcBA silencing was observed for growth of Mtb during the acute phase. Colony forming units obtained at day 21 were 2-fold reduced after silencing of prcBA in the PrcBATet-ON mutant. This difference was larger (7-fold) for the PrcBATet-OFF mutant, presumably due to the faster kinetics of proteasome silencing in this strain (Fig. 4a, b). The reduced growth of proteasome-depleted strains during the acute phase suggested that depletion of the Mtb proteasome during an infection of immune-compromised mice may extend their lifespan. IFNγ−/− mice cannot control growth of Mtb28. We infected these mice with the PrcBATet-OFF mutant, allowed bacterial replication for 10 days and then induced prcBA silencing. This extended the mean survival time of the mice from 54 days to 81 days (p< 0.0001) (Fig. 4f). The Mtb proteasome is thus also essential for optimal in vivo growth and its function goes beyond providing defence against the IFNγ-dependent elements of the host’s immune response.
The genes encoding the Mtb core proteasome, prcB and prcA, were among the approximately 600 genes of the Mtb genome predicted to be essential or required for optimal growth in vitro6. This suggested that prcBA deletion mutants cannot be obtained and we therefore used conditional gene silencing to determine if prcBA are important for virulence of Mtb. Both Tet- ON and Tet-OFF systems allowed efficient prcBA silencing and resulted in significant decreases in proteasome transcript and protein levels as well as proteasome activity. The silencing kinetics of the Tet-OFF system were slightly faster than those achieved with the Tet-ON system even though the final proteasome level was lower after silencing with the Tet-ON system. Silencing of prcBA significantly delayed growth of Mtb on agar plates, which confirmed that the proteasome is required for optimal growth on this medium. However, prcBA silencing hardly affected growth in liquid culture even though prcBA transcript levels were reduced more than 10-fold and proteasome activity was reduced approximately 100-fold. This suggests that prcBA may not be essential for growth in liquid medium.
We further validated the efficiency of prcBA silencing by demonstrating that depletion of the proteasome rendered Mtb more susceptible to RNI and increased its resistance to H2O2. These results are consistent with the phenotypes of mutants of the proteasome associated genes mpa and pafA19. Increased resistance to organic peroxide has also been reported for S. coelicolor proteasome mutants and was accompanied by increased levels of haloperoxidase29. The mechanisms by which the core proteasome protects Mtb against nitrosative damage may include removal of nitrosylated proteins, as demonstrated in mammalian cells30. However, our and previous data suggest that the role of the Mtb proteasome extends beyond defence against RNI. Infection of IFNγ-deficient mice with proteasome-depleted Mtb extended their life span compared to mice infected with proteasome expressing Mtb. Similarly, an Mtb mpa mutant was not fully virulent in iNOS-deficient mice19.
Silencing of prcBA transcription during mouse infections demonstrated that the Mtb proteasome is not only required for optimal in vivo growth but is also essential during the chronic phase of the infection, when the pathogen replicates slowly or not at all31,32. The eukaryotic proteasome participates in protein turnover, transcription and DNA repair33. It is unclear by which of these processes Mtb’s proteasome allows the pathogen to persist in its host. Mutants of the proteasomal ATPase Mpa suggested that the Mtb core proteasome is required for growth of Mtb in mice19,20 but did not predict a role for the proteasome during the chronic phase of infection.
Analyses of prcBA provided proof-of-principle that transcriptional silencing permits inactivation of Mtb genes during different stages of an infection in mice. Reporter gene studies and experiments with other mutants demonstrated that TetR-mediated gene silencing is atc-dose dependent8,10. This suggests that gene silencing could determine to which extent a gene has to be inactivated before mycobacterial growth or survival are impaired. Both of these features, conditionality and dose-responsiveness, distinguish gene silencing from deletion and transposon mutagenesis and are not only important for functional analyses but also relevant to the development of new drugs against tuberculosis. Chemotherapy must be administered during different phases of an infection, including chronic disease, and is likely most effective if it inhibits processes that are essential during all stages of an infection34. Conditional gene silencing facilitates the genetic identification of such processes. In addition, the development of drugs that completely inactivate a target is difficult. Experiments that use gene silencing to partially inactivate a gene might therefore help to focus drug development on targets whose inactivation is effective even if it is incomplete.
In addition to the opportunities that gene silencing offers, the approach also has its challenges. The true impact of target inactivation might be masked by compensatory mutations, epigenetic changes or the residual transcription that occurs despite silencing. Careful analysis of bacterial populations after gene silencing, for example with respect to the frequency of suppressor mutations and the residual expression of the target protein, are required to interpret experiments in which gene silencing does not lead to a phenotypic effect. In addition, for some genes it might be necessary to tailor the activity of the regulated promoter to the target gene’s native expression level to achieve promoter replacement. This could be accomplished with a library of well-regulated promoters that differ in their induced activities. The continued improvement of gene silencing tools for mycobacteria should therefore further facilitate the validation of targets for the development of new drugs against tuberculosis.
Wild type Mtb (H37Rv) was obtained from Dr. Robert North, Trudeau Institute. Mycobacteria were grown in Middlebrook 7H9 medium (Difco) with 0.2% glycerol and 0.05% Tween 80, supplemented with 0.5% BSA, 0.2% dextrose and 0.085% sodium chloride (ADN). Recombinant mycobacteria were selected on 50 μg/ml hygromycinB (hyg) and 15 μg/ml kanamycin (kan) as required. Anhydrotetracycline (atc) was used at 100 ng/ml and replenished every 4 days in liquid culture.
Pmyc1tetO was integrated 5′ of prcBA on the Mtb chromosome via homologous recombination following transduction with temperature sensitive mycobacteriophage35. To achieve this, 768 bp 5′ of the start codon of prcB and the first 675 bp of prcB were amplified by PCR from H37Rv genomic DNA. The DNA fragments were cloned to flank the hygromycin resistance gene in pJSC284, a derivative of pYUB854 containing a lambda cos site, and a unique PacI site35 (gift of J.S. Cox). Pmyc1tetO was cloned 5′ of the start codon of prcB. A transcriptional terminator was cloned 3′ of the hygromycin resistance gene. The plasmid was ligated with temperature-sensitive phage Φ874,35 and the resulting phage was used to infect Mtb as previously described4,36. Hyg-resistant transductants selected on 7H11 agar plates with 50 μg/ml hyg for 3 weeks were analyzed by Southern blot. Recombinants were transformed by electroporation with plasmids encoding wtTetR, revTetR, or no TetR and transformants selected on hyg and kan, in the presence (wtTetR) and absence (revTetR, no TetR) of atc.
PrcBATet-ON and PrcBATet-OFF mutants were grown in the presence or absence of atc to OD580nm of 0.8–1.0. RNA was prepared from 10 ml cultures and cDNA was synthesized using random hexamers. A reaction lacking reverse transcriptase was performed for every RNA sample. Real-time PCR was performed as described36 using specific TaqMan probes for prcA, sigA, 16S rRNA, and rpoB. For immunoblots, cell lysates were prepared by bead-beating in PBS containing protease inhibitor cocktail (Complete Mini, Roche). 15μg cell lysates were subjected to SDS237 PAGE, followed by transfer to a nitrocellulose membrane. Blots were probed with anti-PrcB and anti-DlaT rabbit sera and developed using Immobilon Western Chemiluminescent HRP substrate (Millipore).
Bacteria were grown to OD580nm 1.0 in the presence or absence of atc and cell pellets from 10 ml cultures were lysed in PBS containing protease inhibitor cocktail (Complete Mini, Roche) using a bead beater. Clarified lysates were sterilized by passage through a 0.2 μm filter. Proteasome activity was determined as described elsewhere13. Briefly, 50 μg of lysate was incubated with 100 μM Succinyl-Leu-Leu-Val-Tyr-aminomethyl coumarin (Suc-LLVY-AMC) in 20 mM HEPES, 0.5 mM EDTA buffer and fluorescence was monitored at excitation of 370 nm and emission of 430 nm at 37°C over 60 minutes.
Mtb cultures were grown for 7 days to log phase (OD580nm 0.6) in the presence and absence of 100 ng/ml atc. Single cell suspensions of cultures were prepared by centrifugation at 800 rpm for 10 minutes and subsequently diluted to OD580nm 0.01. To test susceptibility to RNI, cultures were diluted and incubated at pH 5.5 with or without 3mM NaNO2 for 3 days at 37°C. To test susceptibility to H2O2, cultures were incubated at pH 6.8 with or without 5mM H2O2 for 4h at 37°C. Cultures containing atc were maintained in medium with atc during the course of the experiment. To determine viability, serial dilutions of cultures were plated on 7H11 plates with atc for H37Rv PrcBATet-ON and on plates without atc for all other strains.
The PrcBAtet-ON mutant was grown in the presence of atc, the PrcBAtet-OFF mutant and H37Rv/revTetR were grown in the absence of atc to early log phase and washed single cell suspensions were used to infect female wild type and IFNγ-deficient (Jackson Laboratory) C57/BL6 mice via aerosol, as described previously36. Groups of infected mice received 1 mg/ml doxycycline (doxy) in the drinking water containing 5% sucrose. Doxy-containing water was kept in light-protected bottles and changed twice a week. Bacterial burden in lungs and spleen was determined by plating for colony forming units (cfu) at indicated times. The log-rank test was used to determine statistical significance of survival differences observed in PrcBAtet-OFF mutant infected IFNγ−/− mice (GraphPad Prism 4.0).
Mice were infected by aerosol with 100 cfu of Mtb transformed with Pmyc1tetO-gfp or Pmyc1tetO-gfp/wtTetR. Two weeks post infection one group of mice received 1mg/ml doxy in the drinking water containing 5% sucrose while the control groups were kept without doxy. Six days later mice were sacrificed, lung tissues were fixed for 24–36 hrs in 10% formalin and frozen 10 mm thick sections of lung tissues were analyzed microscopically. Images were acquired with a Leica DMIRB widefield microscope using a 40× 1.25 and a 100× 1.25 numerical aperture objective and a 1.5× magnification changer.
We thank G. Lin and C. Nathan for anti-PrcB antiserum; R. Bryk and C. Nathan for anti-DlaT antiserum, L. M. Pierini for help with fluorescent microscopy; E. Hwang for technical support; C. Nathan and G. Lin for helpful discussions; and K.H. Darwin and C. Nathan for review. Supported by NIH AI63446 (S.E.), the I.T. Hirschl Trust (S.E.), the Ellison Medical Foundation (D.S.), Deutsche Forschungsgemeinschaft through SFB 473 (W.H.), Fonds der Chemischen Industrie (W.H.), a grant from the Bill and Melinda Gates Foundation and the Wellcome Trust through the Grand Challenges in Global Health Initiative (S.E., D.S.). The Dept. of Microbiology and Immunology acknowledges the support of the William Randolph Hearst Foundation.
Author Contributions S.G. performed experiments. D.S. guided experimental design and analyses. M.M. helped with initial experiments. W.H. provided constructs and advice. S.E. helped with experiments and guided the study. S.G., D.S., S.E. wrote the paper. All authors discussed the results and commented on the manuscript.
The authors declare no competing financial interest.