The Copper-Responsive RicR Regulon Contributes to Mycobacterium tuberculosis Virulence

The RicR regulon is presumed to be important for Cu resistance because a ricR null mutant, which constitutively expresses all of the genes in the RicR regulon (Fig. 1A), is resistant to high levels of Cu (11). However, the contributions of individual RicR-regulated genes to Cu resistance and virulence had not been determined. Therefore, we sought to quantify the Cu resistance of mutants lacking each RicR-regulated gene. Mutants with three RicR-regulated genes disrupted were identified in our lab collection of more than 10,000 ΦMycoMarT7 mutants in the M. tuberculosis H37Rv strain background (16). We also received a previously reported H37Rv mymT deletion-disruption strain (15). The genotypes of all of the strains used in this study are described in Table 1.

We used a quantitative liquid-based assay to measure the Cu sensitivity of lpqS, Rv2963, socA, and mymT mutants compared to that of wild-type (WT) M. tuberculosis (11). As previously reported, the mymT mutant was more sensitive to Cu than WT M. tuberculosis or the complemented strain was (Fig. 1B) (15). The socA transposon mutant showed WT Cu resistance, while the Rv2963 transposon mutant was slightly (and not always reproducibly) more resistant to Cu (Fig. 1B). Perhaps most interestingly, the lpqS transposon mutant was consistently extremely hyperresistant to Cu. We also used a semiquantitative agar plate assay (15, 17) that showed similar Cu susceptibility results for the mymT mutant but not the lpqS mutant (Fig. 1C).

We next assessed the phenotypes of several mutants in mice. No single mutation attenuated bacterial growth in mice (Fig. 2A). Interestingly, the Rv2963 and lpqS mutant bacteria showed increased growth in vivo (Fig. 2A). In experiments where we inadvertently used a moderately large inoculum of bacteria (~2,000 CFU/mouse), we unexpectedly observed that mice infected with the lpqS mutant were moribund within 4 weeks (Fig. 2B). However, neither gene could restore WT virulence to the respective mutant (data not shown), assuming that the introduction of the WT allele of either gene expressed from its native promoter resulted in appropriate protein synthesis. Thus, it is unclear how (or if) disruption of either gene resulted in the observed hypervirulence phenotypes.

Upon closer inspection of the lpqS locus, we hypothesized that the transposon insertion in the lpqS mutant somehow increased the expression of the divergently expressed gene mmcO (mycobacterial multicopper oxidase [MCO]) (17) (Fig. 1A and 3A). MCOs exist in all kingdoms of life and play a critical role in Cu and iron homeostasis (18–22). MCOs can oxidize substrates, including reduced metals such as Cu⁺ or Fe²⁺, as well as phenolic compounds (23–26). Recently it was shown that MmcO can oxidize Fe²⁺ to Fe³⁺ and perhaps can also convert Cu⁺ to Cu²⁺ (17). Additionally, Rowland and Niederweis observed the induction of MmcO production in M. tuberculosis upon Cu treatment (17). A microarray analysis determined that mmcO is more highly expressed in a ricR mutant than in WT M. tuberculosis, suggesting that it is Cu and RicR regulated (11). Consistent with these previously published data, we confirmed that MmcO levels are elevated upon Cu treatment and now show that it is also highly abundant in a ricR mutant (Fig. 3B) (11).

On the basis of the orientation of the neo gene in the ΦMycoMarT7 transposon in lpqS, we hypothesized that the neo promoter could increase mmcO expression in the absence of Cu (Fig. 3A), potentially resulting in the Cu hyperresistance and hypervirulence of this particular strain. We examined MmcO levels in the WT and lpqS mutant strains and found that MmcO was more abundant in the lpqS mutant than in the WT strain (Fig. 3C).

To determine if MmcO overproduction was responsible for the Cu hyperresistance and hypervirulence phenotypes of the lpqS mutant, we deleted and disrupted mmcO in this strain. We showed by immunoblotting that MmcO was absent from the double mutant strain (Fig. 4A). A Cu susceptibility assay revealed that deletion of mmcO from the lpqS mutant restored Cu sensitivity to WT levels, suggesting that MmcO contributed to the Cu hyperresistance of this strain (Fig. 4A). Importantly, the double mutant was no more sensitive to Cu than WT M. tuberculosis was, suggesting that lpqS itself has a little or no role in Cu resistance. Although Cu hyperresistance was eliminated upon the deletion of mmcO, the double mutant was as hypervirulent as the parental lpqS strain (Fig. 4B). Thus, MmcO overproduction was not responsible for the hypervirulence of this strain.

We do not understand the nature of the lpqS hypervirulence phenotype; complementation with lpqS or lpqS plus two downstream genes (cysK2, Rv0849) could not reduce bacterial growth to WT levels in mice (data not shown). Additionally, whole-genome sequencing of this strain did not reveal any differences from the parental WT strain except for the transposon insertion in lpqS.

Following the above observation that MmcO can confer Cu hyperresistance when overproduced, we wanted to determine the effect of deleting mmcO from WT M. tuberculosis. We deleted and disrupted mmcO in both the H37Rv and CDC1551 strains. Deletion was confirmed by PCR amplification of the region surrounding the deleted locus (data not shown) and immunoblot detection of MmcO protein (Fig. 5A). Interestingly, when we tested these strains for Cu sensitivity, we found that WT H37Rv was more sensitive to Cu than WT CDC1551 was and we thus had to use higher concentrations of CuSO₄ for the CDC1551 strains in our assays. Nonetheless, in both strain backgrounds, the mmcO mutants showed WT Cu resistance in our quantitative Cu susceptibility assay (Fig. 5A). However, during the preparation of this report, Rowland and Niederweis showed than an mmcO mutant is hypersensitive to Cu by using the semiquantitative agar-based assay (17). We tested our mutants in the agar plate assay and indeed observed that both of our mmcO mutants were more sensitive to Cu than the WT or complemented strain was (Fig. 5B).

To determine the importance of mmcO in vivo, we infected mice with the H37Rv WT, mmcO mutant, and mmcO-complemented strains. All of these strains displayed WT growth in the lungs and spleens of WT C57BL/6 mice (Fig. 5C). Thus, mmcO alone does not significantly contribute to the virulence of M. tuberculosis in a mouse model of infection.

Deletion of mmcO did not result in robust Cu hypersensitivity, suggesting that there might be other Cu-binding proteins that contribute to strong Cu resistance. Because mymT is the only other known RicR-regulated gene that significantly contributes to Cu resistance, we deleted and disrupted mmcO in the mymT deletion-disruption mutant and quantified the Cu resistance and virulence of this double mutant strain. We confirmed that MmcO protein was no longer produced by the double mutant strain (Fig. 6A). In the agar plate assay, the mmcO mymT double mutant showed greater Cu susceptibility than the mymT or mmcO single mutant, a phenotype that could be partially complemented with an integrative plasmid encoding mymT and mmcO expressed from their native promoters (Fig. 6B, top). However, in the liquid-based Cu assay, the double mutant showed Cu sensitivity similar to that of the mymT single mutant (Fig. 6B, bottom).

We next infected mice with the WT and single and double mutant strains. None of the mutants demonstrated an attenuated phenotype based on the bacterial burdens found in mouse lungs and spleens up to 8 weeks after aerosol infection (Fig. 6C). Thus, these data suggest that mmcO and mymT alone are not required for normal replication of M. tuberculosis in mice.

None of the genes of the RicR regulon seemed individually important for virulence. Therefore, we hypothesized that either several of the RicR-regulated genes are needed or the entire regulon is needed to play a significant role in virulence. Ideally, we would construct an M. tuberculosis strain that has all five RicR-regulated loci mutated. Because of the arduous process of deleting multiple genes from M. tuberculosis, we developed an alternative method to repress all of the genes in the RicR regulon by producing a “Cu-blind” allele of RicR in a ricR null mutant strain. RicR is a homologue of CsoR and has conserved residues that are predicted to bind Cu⁺ (9). Cysteine 38 of RicR is predicted to be required for Cu⁺ binding (D. Giedroc, personal communication); thus, conversion of cysteine 38 to alanine (RicR_C38A) would prevent RicR from sensing Cu⁺ and detaching from DNA. We previously showed by using a DNA affinity chromatography assay that recombinant RicR binds to a specific sequence in vitro under low-Cu conditions and elutes from DNA with increasing amounts of Cu (11) (Fig. 7A, top). In contrast, RicR_C38A was unresponsive to Cu and could only be eluted from DNA with a high-salt buffer (Fig. 7A, bottom). Because ricR is autoregulated, RicR_C38A could also constitutively repress its own production, leading to reduced repressor levels. Therefore, two point mutations that disrupt RicR binding to DNA (11) were introduced into the ricR promoter (ricRp) of the ricR_C38A construct to allow the constitutive expression of the ricR_C38A allele (p^c-ricR_C38A) (see Table 1 and Materials and Methods). This allele was introduced into a CDC1551 ricR mutant on a plasmid that integrates into the attB site of the M. tuberculosis chromosome. As a control, we also introduced a plasmid expressing WT ricR from ricRp^c (p^c-ricR⁺) into the ricR mutant.

Expression of p^c-ricR_C38A resulted in extreme sensitivity to Cu (Fig. 7B, far right). Interestingly and in contrast, the p^c-ricR⁺-expressing strain was hyperresistant to Cu (Fig. 7B). Because RicR is a Cu-binding protein, it is possible that it sequesters Cu and protects against Cu toxicity when constitutively overproduced. Importantly, we found that the p^c-ricR_C38A-expressing strain, but not the p^c-ricR⁺-expressing strain, was highly attenuated in mice (Fig. 7C). Taken together, our data support a model where several gene products of the RicR regulon, perhaps including RicR itself, are likely critical for WT Cu resistance and full virulence in a mouse model of infection.

It is worth noting that although a ricR null mutant is hyperresistant to Cu in vitro, it is not hypervirulent in mice. This mutant had a subtle growth defect in the lungs of infected mice compared to the WT and ricR-complemented strains (Fig. 7C, right panel).

Finally, we tested if M. tuberculosis strains defective in proteasome function were more sensitive to Cu. Proteasomal degradation of a protein requires PafA (proteasome accessory factor A), which ligates the posttranslational modifier Pup (prokaryotic ubiquitin-like protein) to protein substrates, and Mpa (mycobacterial proteasome ATPase), which delivers pupylated protein substrates into the proteasome core for degradation (reviewed in reference 27). Mutations that reduce proteasomal degradation have repressed the expression of all RicR regulon genes; however, RicR itself does not appear to be a proteasome substrate (11). Nonetheless, because the RicR regulon is repressed in proteasome-defective M. tuberculosis strains, we predicted that these strains would be more sensitive to Cu than WT bacteria are. Indeed, we found that M. tuberculosis strains lacking mpa or pafA were more sensitive to Cu than the WT and complemented strains were (Fig. 7D). Taken together, this suggests that the in vivo attenuated phenotype of proteasomal degradation mutants could be partly due to reduced Cu resistance.

The bacterial strains, plasmids, and primers used in this work are listed in Table 1. M. tuberculosis strains were grown in 7H9 broth (Difco) supplemented with 0.2% glycerol, 0.05% Tween 80, 0.5% bovine serum albumin, 0.2% dextrose, and 0.085% sodium chloride (ADN) or Sauton’s minimal medium (3.7 mM potassium phosphate, 2.5 mM magnesium sulfate, 30 mM l-asparagine, 3.5 mM zinc sulfate, 9.5 mM citric acid, 6.0% glycerol, 0.005% ferric ammonium citrate, 0.05% Tween 80). Cultures were grown at 37°C without agitation in vented flasks. For M. tuberculosis growth on solid medium, Middlebrook 7H11 agar (Difco) was prepared with Middlebrook OADC (oleic acid, albumin, dextrose, and catalase; BBL) supplementation. M. tuberculosis strains were grown in 50 µg ml⁻¹ kanamycin, 50 µg ml⁻¹ hygromycin, and/or 25 µg ml⁻¹ streptomycin when necessary. Escherichia coli cultures were grown in Luria-Bertani (LB) broth (Difco) or on LB agar at 37°C. Antibiotics were added at final concentrations of 100 µg ml⁻¹ kanamycin, 150 µg ml⁻¹ hygromycin, and 50 µg ml⁻¹ streptomycin.

To clone the p^c-ricR_C38A allele into pMV306, site-directed mutagenesis was performed by splicing by overlap extension by PCR (29) (the primers used are described in Table 1). Cysteine 38 of RicR was first changed to alanine. A second mutation was then introduced into the ricR_C38A plasmid; 2 nucleotides (nt) were changed in the ricR promoter (ricRp) (TACCCCGCTGGGTA → TACCCCGCTGTTTA = ricRp^c) to reduce repressor binding.

With the exception of the biotin tetraethylene glycol (BioTEG; Operon)-labeled primer, all primers were purchased from Invitrogen (Table 1). Phusion polymerase and restriction enzymes from New England Biolabs were used for cloning. All plasmid inserts generated by PCR were sequenced by GENEWIZ (Plainfield, NJ).

For antibody production, MmcO-His₆ was overproduced in E. coli and purified under denaturing conditions by following the manufacturer’s instructions (Qiagen). Polyclonal rabbit antibodies were generated by Covance (Denver, PA). DlaT (used for loading controls) and RicR antibodies were described previously (11, 30, 31).

For immunoblot analysis, equivalent cell numbers based on optical density at 580 nm (OD₅₈₀) were collected for each relevant strain at the same growth phase. For Cu-treated protein samples, bacteria were treated with CuSO₄ for 4 h at a final concentration of 500 µM. Bacteria were washed once with phosphate-buffered saline (PBS) with 0.05% Tween 80 and resuspended in 300 µl of lysis buffer (100 mM Tris-Cl, 1 mM EDTA, pH 8). Bacteria were lysed by bead beating with zirconia beads three times for 30 s each. Whole-cell lysates were mixed with 4× sodium dodecyl sulfate (SDS) sample buffer and boiled for 10 min at 100°C. Proteins were separated by 10% SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, incubated with rabbit polyclonal antibodies to MmcO at a dilution of 1:1,000 in 3% BSA in Tris-buffered saline with Tween 20, and visualized with horseradish peroxidase coupled with anti-rabbit secondary antibodies (Thermo Scientific). Detection of horseradish peroxidase was performed with SuperSignal West Pico or West Femto Chemiluminescent Substrate (Thermo Scientific).

We made M. tuberculosis mutants by using a method previously described by our lab (11). Briefly, pYUB854, which was originally developed for specialized transduction mutagenesis (32), was used to clone ~700 bp of sequence both upstream (5′) and downstream (3′) of a gene of interest. The 5′ and 3′ sequences, including the start and stop codons, respectively, were cloned to flank the hygromycin resistance cassette in pYUB854. We also constructed a kanamycin resistance-marked version of this plasmid to make the ΔmmcO::kan mutant. The hygromycin resistance cassette in pYUB854 was replaced with the kanamycin resistance cassette from pMV306.kan to generate pYUB854.kan (the primers and plasmids used are described in Table 1). Because the ricR mutant is in the CDC1551 background, we chose to mutate mmcO in both the H37Rv and CDC1551 strains.

Plasmids were digested with PacI, and ~1 µg of linearized, gel-purified DNA was used for electroporation into M. tuberculosis. M. tuberculosis strains were grown to an OD₅₈₀ of ~0.4 to 1, washed, and resuspended in 10% glycerol to make electrocompetent cells as described in detail elsewhere (33). Bacteria were inoculated onto 7H11 solid medium with 50 µg ml⁻¹ hygromycin or 50 µg ml⁻¹ kanamycin as needed. A no-DNA control electroporation was always done to check for spontaneous antibiotic-resistant mutants. After 2 weeks, colonies were picked and inoculated into 200 µl of 7H9 medium with antibiotics. The 200-µl 7H9 starter cultures were inoculated into 5-ml cultures for further analysis. Potential mutants were tested by immunoblot analysis, PCR (Taq polymerase; Qiagen), and sequencing.

Experiments were performed as described previously (11). DNA probes were made by amplifying the lpqS promoter with a forward primer containing 5′ BioTEG (Operon) modifications. Clarified lysates of E. coli producing either WT RicR or RicR_C38A were incubated with a DNA-Dynabead mixture. Protein interacting with the coupled DNA was sequentially eluted with increasing concentrations of CuSO₄ as indicated and finally with high-salt buffer (1 M NaCl in 50 mM Tris, pH 7.5). RicR immunoblotting was performed as previously described (11).

MHD131 (lpqS::ΦMycoMarT7) and the parental H37Rv strain were sequenced with an Illumina GenomeAnalyzer IIx. Approximately 5 µg of DNA was processed by the standard Illumina sample preparation protocol for genomic DNA (Illumina, Inc.), and the samples were sequenced in paired-end mode with a read length of 51 nt. The genomes were assembled by mapping reads to the public H37Rv reference sequence (GenBank accession no. NC_000962) and single-nucleotide polymorphisms and insertions/deletions as described in reference 34. The mean depths of coverage (number of reads covering each site) were 59.6 and 63.1 times, respectively, for the two strains.

CuSO₄ powder was dissolved in water and sterilized through 0.45-µm-pore-size filters to make a 1 M stock solution. M. tuberculosis strains were grown in 7H9-ADN medium to an OD₅₈₀ of ~0.5 to 1. Bacteria were harvested, washed once with Sauton’s medium (no added Cu), and then subjected to a low-speed spin (800 g for 8 min) to remove clumped cells. Bacteria were diluted to an OD₅₈₀ of 0.08 in Sauton’s medium. Each culture was placed into the wells of a 96-well plate at 194 µl per well and treated with 6 µl of CuSO₄ at different concentrations, and the plate was incubated for 10 days. Bacteria were then diluted and inoculated onto 7H11 agar and CFU were enumerated 2 to 3 weeks later. Because we noticed some variability in the minimum bactericidal Cu concentration between experiments, we always used a range of Cu concentrations for every experiment and drew conclusions based on reproducible trends. All experiments were done at least in duplicate.

The agar plate assay was performed as described previously (15, 17). M. tuberculosis strains were grown in 7H9-ADN medium to an OD₅₈₀ of ~0.5 to 1. Cells were harvested by centrifugation, washed once with PBS-Tween (0.05%), and then spun slowly at 800 × g. Declumped cell suspensions were diluted to an OD₅₈₀ of 0.1 with PBS-Tween. Three-microliter samples of serial dilutions were spotted onto M. tuberculosis 7H11-OADC agar plates containing either no additional CuSO₄ or different concentrations thereof. Plates were incubated at 37°C for ~2 weeks.

Mouse infections were performed essentially as described previously (12). M. tuberculosis strains were grown in 7H9-ADN medium to an OD₅₈₀ of ~0.4. Bacterial clumps were removed from the cultures as described above. Female C57BL6/J mice (Jackson Laboratories) were infected with a Glas-Col Inhalation Exposure System to inoculate ~200 to 400 CFU/mouse. Three (day 1) or four (days 21 and ~56) mice were humanely sacrificed, and their lungs and spleens were removed, homogenized, and inoculated onto 7H11 agar. CFU were enumerated after 2 to 3 weeks. All procedures were performed with the approval of the NYU Institutional Animal Care and Use Committee.