is one of the most devastating microbial agents, as it infects nearly one-third of the world’s population and kills nearly two million people annually (http://www.who.int/en/
). Currently available chemotherapies are lengthy and potentially toxic (1
). In addition, the numbers of multiresistant, extensively resistant, and totally drug-resistant strains are rising (2–4
). Thus, an improved understanding of M. tuberculosis
pathogenesis is urgently needed in order to develop improved treatments for tuberculosis.
It has recently been determined that host-derived Cu is important for controlling M. tuberculosis
infections in two animal models of infection (5
). Cu is a well-known antimicrobial agent, but only in the last few years has its role been appreciated with regard to microbial infections in mammals (6
). Previous studies found that Cu levels transiently increase in gamma interferon-activated macrophages infected with mycobacteria (7
). In another study, it was shown that Cu accumulates within phagolysosomal compartments via the Cu-transporting ATPase ATP7A (8
). Additionally, in a guinea pig model of infection, Cu accumulates in the granulomatous lesions of infected lungs (5
). Perhaps because of this host response, it appears that M. tuberculosis
has acquired several independent mechanisms to defend itself against Cu toxicity (6
). These include mycobacterial Cu transport protein B (MctB) (5
), the Cu-sensitive operon repressor (CsoR) operon (9
), and the regulated in Cu repressor (RicR) regulon (11
The RicR regulon was discovered in an attempt to understand the link between M. tuberculosis
proteasome function and pathogenesis, as M. tuberculosis
strains defective for proteasomal degradation are highly attenuated in mice (11–14
). This regulon includes ricR
(encodes a transcriptional repressor), lpqS
(encodes a putative lipoprotein), mymT
(encodes a mycobacterial metallothionein), socAB
(small open reading frame induced by copper A and B), and Rv2963 (a putative permease gene) (11
). All five loci are transcriptionally repressed in strains defective for proteolysis by the M. tuberculosis
). Interestingly, with the exception of ricR
itself, all of these genes are found only in pathogenic mycobacteria, suggesting that they are important during infections of a vertebrate host. All of these genes have a palindromic motif in their promoters that is recognized by the transcriptional repressor RicR. Like its closely related paralog CsoR, RicR is presumed to bind Cu+
and is released from DNA (9
). The only previously characterized RicR-regulated gene other than ricR
itself is mymT
. Although a mymT
mutant is hypersensitive to Cu, this mutant has no virulence defect in mice (15
In this study, we sought to determine the contribution of every RicR-regulated gene to Cu resistance and virulence. We found that most of the genes conferred no to variable Cu resistance in vitro. Furthermore, none of the single mutants had an attenuated phenotype in mice. In contrast, repression of the entire RicR regulon resulted in a strong Cu-sensitive phenotype in vitro and severely attenuated growth in vivo. Thus, it appears that multiple members of the RicR regulon are required for Cu resistance during infections.
The discovery of several Cu-responsive regulons in a human-exclusive pathogen suggests that M. tuberculosis faces host-supplied Cu during infections. In this work, we sought to understand the contribution of the RicR regulon to Cu resistance and virulence in mice. We determined that, with the exception of mymT, the disruption of any single RicR-regulated gene was insufficient to sensitize M. tuberculosis to Cu. We also established in this study that mmcO is a member of the RicR regulon. Overexpression of mmcO resulted in Cu hyperresistance but did not impact virulence. We also determined that the contributions of individual RicR-regulated genes to pathogenesis appear to be minimal. Single mutations in mmcO, lpqS, Rv2963, and socA did not attenuate M. tuberculosis growth in mice, and curiously, lpqS and Rv2963 mutants were hypervirulent for reasons that remain to be determined. We also tested the idea that deletion of the two RicR-regulated genes directly implicated in Cu resistance (mymT, mmcO) might have a more robust effect on bacterial survival in vivo. However, an mmcO mymT mutant was as virulent as WT M. tuberculosis in mice.
We did not detect robust Cu-associated phenotypes with the lpqS and Rv2963 mutants. LpqS and Rv2963 are putative membrane proteins each predicted to have several histidines that localize just outside the cytoplasmic membrane. These residues may be potential candidates to coordinate metal ions. Rv2963 is predicted to be a permease the disruption of which may perhaps alter either the import or the export of Cu or other metal ions under certain conditions.
It has been reported that a ΔlpqS
mutant is hypersensitive to Cu in vitro
). As in our study, the authors of that previous study could not complement their mutation, suggesting that their Cu-sensitive phenotype might be unlinked to LpqS. A possibility is that the disruption of lpqS
in the study of Sakthi and Narayanan was polar on the expression of genes that are important for Cu resistance. Two uncharacterized genes, cysK2
and Rv0849, are cotranscribed with lpqS
and may perhaps have a role in Cu resistance. Another possibility is that disruption of lpqS
results in the dysregulation of the divergently expressed gene mmcO
, which is important for Cu resistance. The mechanism of RicR repression of mmcO
expression is not fully understood but may involve the bending of DNA between mmcO
to simultaneously repress both genes with a single RicR-binding site. On the basis of the published lpqS
data and our data, we strongly hypothesize that LpqS itself is not critical for Cu resistance.
socAB is perhaps the most mysterious RicR-regulated locus; these genes are found only in the M. tuberculosis complex and do not resemble sequences in any other organism sequenced to date. Because of the lack of robust phenotypes associated with disruptions in this locus, it is unclear what role it plays, if any, in Cu homeostasis.
Although the lpqS
and Rv2963 transposon mutants lacked clear Cu resistance phenotypes, both had very intriguing hypervirulence phenotypes in mice; however, we could not complement these mutations to restore WT virulence. We hypothesized that overexpression of mmcO
was responsible for the hypervirulence of the lpqS
strain, but this was not the case (Fig. 4
). Currently, we can only speculate as to why the lpqS
and Rv2963 mutants are hypervirulent. A possibility is that the absence of these putative membrane proteins permits the bacteria to grow more rapidly in vivo
, which appears to be the reason for the increased virulence of these strains. Alternatively, it is possible that mutant or truncated proteins that alter the course of infection are produced by these strains. Yet another possibility is that the transposon insertions in these mutants change the expression of other genes that increase the growth of the bacteria in vivo
. Needless to say, we are very interested in understanding why these mutants rapidly kill their hosts and are testing several of these hypotheses.
None of the genes of the RicR regulon individually showed a role in promoting virulence. It is possible that in the absence of one or more of the RicR-regulated genes, other genes encoding Cu-binding proteins or efflux systems could be induced to compensate for their absence. Nonetheless, a ricR
mutant that constitutively represses all RicR-dependent promoters was highly attenuated in mice. These data strongly suggest that the RicR regulon, either in its entirety or in part, is required for the full virulence of M. tuberculosis
. Our in vitro
and in vivo
data also suggest that RicR itself may sequester Cu like a metallothionein because the constitutive overexpression of WT ricR
(as opposed to ricRC38A
, which is expected not to bind Cu) resulted in increased Cu resistance (Fig. 7B
). Another possible reason that single mutations had little to no impact in vivo
is that mice may not be the best model for testing the role of these genes; some genes may show importance in models of infection that more closely resemble human disease.
It is notable that we observed considerable differences in Cu susceptibility, depending on the assay we used. We observed robust differences in Cu resistance when using the liquid-based quantitative assay for the mymT (hypersensitive), ricR, and lpqS (hyperresistant) mutant strains, while in contrast, we could detect a phenotype for mmcO mutants only by using an agar plate-based assay. Interestingly, both assays revealed that M. tuberculosis strain CDC1551 is inherently more resistant to Cu than H37Rv is. The results from the different assays suggest that different Cu-binding proteins are important under different conditions. Furthermore, it has yet to be determined which Cu regulon, CsoR or RicR, responds first to Cu stress. It is also possible that the repressors respond to different concentrations of Cu. In the agar-based assay, bacteria were exposed to Cu throughout the experiment (14 to 21 days), whereas in the liquid-based assay we exposed the bacteria for a defined time period (10 days) before inoculation onto agar. Additionally, the oxygen tension, which has a critical impact on the redox status of Cu, could impact the effective Cu+ concentration during the experiment. Finally, the media used for agar versus broth cultures are also slightly different and may impact Cu susceptibility in unknown ways.
At the forefront of our remaining questions is what the link is between RicR regulon expression and M. tuberculosis proteasomal degradation. A simple explanation would be that RicR is a proteasome substrate and that the accumulation of this repressor in proteasome degradation-defective mutants results in repression of the regulon. However, we have no evidence that RicR accumulates in proteasome-defective mutants. Another possibility is that one or several Cu-binding proteins are proteasome substrates the accumulation of which in a proteasome degradation-defective strain sequesters Cu away from RicR, leading to gene repression. These and other hypotheses are currently being tested.
Our work supports previous observations that Cu homeostasis is critical for the pathogenesis of M. tuberculosis
). As in other organisms, too little accessible Cu is detrimental while too much Cu can be toxic. Taken together, our findings affirm that the careful control of Cu homeostasis is essential for M. tuberculosis
virulence and that the RicR regulon plays an important and nonredundant role in this process.