A Novel P1B-type Mn2+-transporting ATPase Is Required for Secreted Protein Metallation in Mycobacteria*

Background: CtpC is an uncommon metal transport ATPase required for Mycobacterium tuberculosis virulence. Results: CtpC shows Mn2+-ATPase activity. Mutations in ctpC alter Mn2+ homeostasis, increase sensitivity to redox stress, and decrease Mn-superoxide dismutase activity. Conclusion: CtpC is a Mn2+ transport ATPase required for homeostasis and the assembly of secreted metalloproteins in mycobacterium. Significance: CtpC provides a novel mechanism for Mn2+ metallation of secreted proteins. Transition metals are central for bacterial virulence and host defense. P1B-ATPases are responsible for cytoplasmic metal efflux and play roles either in limiting cytosolic metal concentrations or in the maturation of secreted metalloproteins. The P1B-ATPase, CtpC, is required for Mycobacterium tuberculosis survival in a mouse model (Sassetti, C. M., and Rubin, E. J. (2003) Genetic requirements for mycobacterial survival during infection. Proc. Natl. Acad. Sci. U.S.A. 100, 12989–12994). CtpC prevents Zn2+ toxicity, suggesting a role in Zn2+ export from the cytosol (Botella, H., Peyron, P., Levillain, F., Poincloux, R., Poquet, Y., Brandli, I., Wang, C., Tailleux, L., Tilleul, S., Charriere, G. M., Waddell, S. J., Foti, M., Lugo-Villarino, G., Gao, Q., Maridonneau-Parini, I., Butcher, P. D., Castagnoli, P. R., Gicquel, B., de Chastellièr, C., and Neyrolles, O. (2011) Mycobacterial P1-type ATPases mediate resistance to zinc poisoning in human macrophages. Cell Host Microbe 10, 248–259). However, key metal-coordinating residues and the overall structure of CtpC are distinct from Zn2+-ATPases. We found that isolated CtpC has metal-dependent ATPase activity with a strong preference for Mn2+ over Zn2+. In vivo, CtpC is unable to complement Escherichia coli lacking a functional Zn2+-ATPase. Deletion of M. tuberculosis or Mycobacterium smegmatis ctpC leads to cytosolic Mn2+ accumulation but no alterations in other metals levels. Whereas ctpC-deficient M. tuberculosis is sensitive to extracellular Zn2+, the M. smegmatis mutant is not. Both ctpC mutants are sensitive to oxidative stress, which might explain the Zn2+-sensitive phenotype of the M. tuberculosis ctpC mutant. CtpC is a high affinity/slow turnover ATPase, suggesting a role in protein metallation. Consistent with this hypothesis, mutation of CtpC leads to a decrease of Mn2+ bound to secreted proteins and of the activity of secreted Fe/Mn-superoxide dismutase, particularly in M. smegmatis. Alterations in the assembly of metalloenzymes involved in redox stress response might explain the sensitivity of M. tuberculosis ctpC mutants to oxidative stress and growth and persistence defects in mice infection models.

Transition metals are essential for life; however, at high concentrations, they can become toxic due to either adventitious metal binding to various biomolecules or the promotion of oxidative stress through Fenton chemistry (1,2). To prevent metal toxicity, chaperone and chelating molecules tightly bind free ion species, and transmembrane transporters prevent excess accumulation in the cytosol. Together, these mechanisms maintain the aqueous milieu essentially free of uncomplexed metals (3)(4)(5). As a consequence of tightly regulated metal homeostasis, metalloproteins acquire the necessary cofactors through specific protein-protein interactions (4).
The involvement of transition metals in host-pathogen interaction is highlighted by the requirement of various transition metal transporters as well as transition metal-responsive transcriptional regulators for bacterial virulence (3,6). Intracellular pathogens, such as Mycobacterium tuberculosis, must cope with transition metal starvation or excess stress during infection. For example, the natural resistance-associated macrophage protein 1 (Nramp1), a divalent cation-proton antiporter, participates in bacterial killing by driving Fe 2ϩ efflux from the phagosome and starving the enclosed bacteria of this essential nutrient (7). Whether or not transition metals other than Fe 2ϩ are depleted in the phagosome after infection is unclear (8,9). In addition to the transition metal starvation model, it has been proposed that transporter recruitment to the phagosomal membrane may generate stress by increasing the concentration of free transition metal, such as Cu ϩ and Zn 2ϩ (9 -11). In particular, there is evidence that Zn 2ϩ accumulation during infection hampers the growth of intracellular pathogens (11). Supporting this, expression of the lysosome-associated Zn 2ϩ transporter ZIP8 in IFN␥-stimulated macrophages and T-cells has been reported (12,13).
Transition metals also appear to be relevant for bacterial virulence through their participation in redox detoxification by metalloenzymes. For instance, M. tuberculosis survives in a highly oxidative environment by means of intracellular and secreted superoxide dismutases that participate in reactive oxygen species and reactive nitrogen species detoxification mechanisms (14 -16). M. tuberculosis SodA, is a secreted Fe 2ϩ / Mn 2ϩ -dependent superoxide dismutase protein, but under in vitro growth conditions, it contains Fe 2ϩ as a primary cofactor (14,16,17). SodA is exported in an unfolded state via a Sec-dependent mechanism (18), probably acquiring its metal cofactor in the extracytoplasmic milieu.
Heavy metal transport P 1B -ATPases 3 drive the efflux of a range of cytoplasmic metal ions, such as Cu ϩ , Zn 2ϩ , and Co 2ϩ , using the energy of ATP hydrolysis and are involved in maintaining cellular metal quotas (3,19). The metal specificity of most P 1B -ATPase can be predicted based on the conserved signature sequences present in transmembrane segments 6, 7, and 8 (TM6, TM7, and TM8) 4 (3, 19 -23). There are seven transition metal-transporting P 1B -ATPases in M. tuberculosis (20). Their signature transmembrane metal binding sites indicate that three (CtpA (cation transporter protein A), CtpB, and CtpV) are Cu ϩ -ATPases and that two others are likely Co 2ϩ -ATPases (CtpD and CtpJ), whereas CtpC and CtpG appear to be novel transition metal transporters. Large scale genetic screens have predicted that ctpC and ctpD are specifically required for the optimal in vivo growth or survival of M. tuberculosis in the mouse model of tuberculosis (24). Microarray analyses of immune response induced genes have shown that ctpC and ctpV are expressed more robustly during infection of BALB/C mice than in the immune-defective SCID mouse strain or during in vitro growth (25). Similarly, induction of ctpC, ctpG, and ctpV in M. tuberculosis during macrophage infection has been reported (26). Linking CtpC activity with metal homeostasis during infection, early studies on M. tuberculosis Fe 2ϩregulated genes showed that ctpC expression was induced in low Fe 2ϩ medium (27).
Recently, Botella et al. (11) have shown that Zn 2ϩ exposure induces the expression of ctpC and that a ctpC-null strain is hypersensitive to Zn 2ϩ in vitro. A priori, these phenotypic characteristics might suggest that CtpC is a Zn 2ϩ -ATPase. However, CtpC is structurally distinct from Zn 2ϩ -transporting ATPases (20) because some of the invariant Zn 2ϩ -coordinating residues involved in transmembrane translocation are absent from this protein. In addition, CtpC also lacks characteristic, although nonessential, regulatory cytoplasmic metal binding domains. These observations suggested that further characterization was necessary to define the biochemical function of this protein. We reexamined CtpC catalytic activity and its role in bacterial virulence to identify the substrate and consequent physiological role of this ATPase. Our results indicate that CtpC is a high affinity, slow turnover Mn 2ϩ -ATPase, which is required for M. tuberculosis virulence. CtpC appears involved in the loading of Mn 2ϩ into secreted metalloproteins, such as SodA. We propose that this function may be required for infection.

EXPERIMENTAL PROCEDURES
Bioinformatics Analysis-M. tuberculosis CtpC protein sequence was used for a BLAST search against all predicted sequence in the NCBI database. Among the returned hits, we selected those protein sequences lacking cytoplasmic metal binding domains (N-or C-MBD) and containing conserved transmembrane signatures, TM6 (CPC) and TM8 (HXXSS), probably involved in metal coordination during transport (20). Resulting homologous sequences were aligned with MUSCLE (28), Bioedit (29), and ESPript software (30). NCBI GI accession numbers were obtained, and the KEGG database nomenclature was used. The proteins that were not on KEGG were named with a three-letter key to denote the organismal origin as indicated in Fig. 1.
Bacterial Strains-M. tuberculosis H37Rv and Mycobacterium smegmatis mc 2 155 cells were used in these studies. These were grown at 37°C in 7H9 and 7H10 media (BD Biosciences) supplemented with 10% OADC (oleic acid, albumin, dextrose, catalase) enrichment or in Sauton's medium, as indicated in the figures. Escherichia coli BL21 (DE3) pLysS cells transformed with the plasmid pSJS1240 coding for rare tRNA (31) were grown at 37°C in 2XYT medium containing 50 g/ml spectinomycin, 34 g/ml chloramphenicol.
ctpC Cloning and Expression-The ctpC gene was amplified from M. tuberculosis H37Rv and M. smegmatis mc 2 155 genomic DNA using the primers Mtb ctpC Fwd plus Mtb ctpC Rvs and ctpC_smeg Fwd plus ctpC_smeg Rvs (Table 1). These introduced a tobacco etch virus protease site-coding sequence at the amplicon 3Ј-ends. The resulting amplicons were cloned into pBAD TOPO/His vector (Invitrogen) that introduces a C-terminal His 6 tag suitable for Ni 2ϩ affinity purification. The M. tuberculosis ctpC cDNA was used as a template to introduce the mutations coding for single substitution H697A and double substation S700A/S701A, employing the QuikChange site-directed mutagenesis kit (Stratagene). The primers used are listed in Table 1. Sequences were confirmed by automated DNA sequence analysis. E. coli BL21 (DE3) pLysS pSJS1240 cells were transformed with these constructs and grown at 37°C in 2XYT medium supplemented with 100 g/ml ampicillin, 50 g/ml spectinomycin, and 34 g/ml chloramphenicol. Protein expression was induced with 0.002% L-arabinose. Cells were harvested 4 h after induction; washed with 25 mM Tris, pH 7.0, and 100 mM KCl; and stored at Ϫ80°C. In alternative experiments, pBAD TOPO vector carrying M. tuberculosis ctpC was introduced in E. coli W3110 ⌬zntA cells (32) and selected with 100 g/ml ampicillin, 50 g/ml kanamycin.
Protein Purification-CtpC purification was performed according to Mandal et al. (33). All purification steps were carried out at 0 -4°C. Cells were suspended in buffer A (25 mM Tris, pH 7.0, 100 mM sucrose, 1 mM phenylmethylsulfonyl fluoride (PMSF; Sigma)) and disrupted with a French press at 20,000 p.s.i. Lysed cells were centrifuged at 8,000 ϫ g for 30 min. The supernatant was then centrifuged at 229,000 ϫ g for 1 h, and the pelleted membranes were resuspended in buffer A (10 -15 mg/ml). For protein solubilization and purification, membranes were diluted to a final concentration of 3 mg/ml in buffer B (25 mM Tris, pH 8.0, 100 mM sucrose, 500 mM NaCl, 1 mM PMSF) and solubilized with 0.75% dodecyl-␤-D-maltoside (DDM; Calbiochem). The preparation was incubated for 1 h at 4°C with mild agitation and centrifuged at 229,000 ϫ g for 1 h. The supernatant was incubated overnight at 4°C with Ni 2ϩnitrilotriacetic acid resin (Qiagen) pre-equilibrated with buffer B, 0.05% DDM, and 5 mM imidazole. The resin was washed with buffer B, 0.05% DDM containing 10 and 20 mM imidazole, and the protein was eluted with buffer B, 0.05% DDM, 250 mM imidazole. Fractions were pooled and concentrated, and buffer was replaced by 25 mM Tris, pH 8.0, 100 mM sucrose, 50 mM NaCl, and 0.01% DDM (buffer C) using 50 kDa cut-off centricons (Millipore). The proteins were aliquoted and stored in 20% (v/v) glycerol at Ϫ20°C until use. All protein determinations were performed in accordance with the Bradford method (34). Purified CtpC protein was analyzed by 10% SDS-PAGE followed by Coomassie Brilliant Blue (CBB) staining or Western blot using an anti-His 6 tag antibody (GenScript).
Recombineering, Mutant, and Complemented Strain Preparation-The M. smegmatis ctpC mutant strain was constructed following the procedures described previously (36,37). For mutation of M. tuberculosis ctpC, a 1000-bp fragment corresponding to the 500 5Ј-most and 500 3Ј-most bp of ctpC was constructed. An insertion cassette containing the restriction sites for NotI-HpaI-AscI was added between the 5Ј-and 3Ј-most 500-bp regions. The resulting synthesized fragment was then inserted between HindIII sites in a pUC57 expression vector (GeneScript), resulting in plasmid pEL1a. Vector pKM342 contains a hygR cassette flanked by NotI-AscI sites. To insert the hygR cassette in pEL1a, both pEL1a and pKM342 were digested with NotI-AscI. The 1.2-kbp hygromycin fragment was then ligated into pEL1a, resulting in pEL2a. To generate the ctpC mutant, the resulting 2.2-kbp ctpC-hygR-ctpC fragment from digestion of pEL2a with HindIII was transformed into the M. tuberculosis H37Rv recombineering strain. Briefly, the M. tuberculosis H37Rv recombineering strain bearing plasmid pNIT:ET (38) was induced for 18 h with 1 M isovaleronitrile. The culture was treated with 0.2 M glycine for 8 h before making electrocompetent cells and transformed. After selection on 7H10 plates containing hygromycin (50 g/ml), the presence of the ctpC insertional mutation was assayed by PCR amplification of the hygR cassette flanked by the N-and C-terminal junctions. All primers used are listed in Table 1.
Constructs for mutant complementation assays were made by amplifying the M. tuberculosis and M. smegmatis ctpC from genomic DNA. The resulting PCR fragments were digested and ligated into pJEB402 (39), resulting in pJEB402-tbC and pJEB402-smC. The ligation reactions were transformed into DH5␣ cells, and the presence of the insert was verified by colony PCR and restriction digests. The plasmids were then purified and transformed into the mutant strains. Transformants showing kanamycin resistance were analyzed for the presence of the genes by PCR.
Metal and Redox Stressor Sensitivity Assays-M. tuberculosis H37Rv wild type, ctpC::hyg, and complemented strains were grown in 7H9-OADC medium to A 600 ϭ 1.0. Metal sensitivity was assayed by spotting dilution series of wild type, ctpC::hyg, and complemented strains on 7H10-OADC agar containing increasing concentrations of CuCl 2 , CoCl 2 , MnCl 2 , ZnSO 4 , or CdCl 2. For tert-butyl hydroperoxide sensitivity, 7H10-OAD was used instead of 7H10-OADC. Colony-forming units (CFU) were assessed after 18 days of incubation at 37°C. M. smegmatis mc 2 155 wild type, ctpC::hyg, and complemented strains were grown in 7H9-OADC medium with increasing concentrations of ZnSO 4 or MnCl 2 . A 600 was determined after 48 h of culture. For analyzing the sensitivity to extracellular redox stress, these strains were grown in 7H9 medium until A 600 ϭ 1.0. The cells were incubated for 0 (before addition), 30, 60, 90, and 120 min with 0.1 unit/ml xanthine oxidase and 250 M hypoxanthine (Sigma) diluted 1:100 in phosphate-buffered saline as described previously (40). Serial dilutions were plated at different time points, and CFU were quantified. The percentage of survival (CFU at 30 -120 min/CFU at 0 min) was calculated for three independent experiments.
Gene Expression Analysis-M. tuberculosis H37Rv and M. smegmatis mc 2 155 cells in exponential phase were supplemented with 100 M various metals and redox stressors as indicated in Figs. 4 and 5 and incubated for 2 h. Chelexed Sauton's medium was used to evaluate ctpC expression under metal starvation conditions. Sauton's medium, pH 7.4, containing 6% (v/v) glycerol, 3.6 mM KH 2 PO 4 , 11.4 mM citric acid, 30 mM asparagine, 0.1 M FeCl 3 , 4.2 mM MgCl 2 , and 0.05% (v/v) Tween 80 was incubated overnight at 4°C in constant stirring, in the presence (chelexed medium) or the absence (non-chelexed medium) of 1 g of Chelex beads (Sigma) per 100 ml of medium. Subsequently, 0.1 M FeCl 3 and 4.2 mM MgCl 2 were added to the media. Manganese content was determined in 7H9, nonchelexed, and chelexed Sauton's media by furnace atomic absorption spectroscopy (Varian SpectrAA 880/GTA 100). 7H9 medium contained 0.88 Ϯ 0.02 nM, non-chelexed Sauton's medium contained 0.36 Ϯ 0.01 nM, and chelexed Sauton's medium contained 0.27 Ϯ 0.01 nM Mn 2ϩ . Cells were pelleted and rinsed three times with chelexed medium and incubated 2 h with either 7H9 or non-chelexed Sauton's medium as controls and chelexed Sauton's medium. Cells were harvested, resuspended in 1 ml of TRIzol reagent (Invitrogen), and disrupted using lysing matrix B (MP Biomedicals) in a cell disrupter (FastPrep FP120, Qbiogene). RNA pellets were air-dried and redissolved in 50 l of diethyl pyrocarbonate-treated ultrapure water. The remaining DNA was removed with the RNeasy minikit and an on-column DNase I kit (Qiagen). The RNA samples (1 g) were used as templates for cDNA synthesis with random primers and SuperScript III reverse transcriptase (Invitrogen). Quantitative RT-PCR was performed with iQ SYBR Green supermix (Bio-Rad). Transcript analysis of ctpC was performed by quantitative RT-PCR with the primers listed in Table 1 with cycler conditions as described previously (41). The RNA polymerase factor (sigA) was used as an internal reference. Determinations were carried out with RNA extracted from three independent biological samples, with the threshold cycle (C t ) determined in triplicate. The relative levels of transcription were calculated by using the 2 Ϫ⌬⌬ C t method (42). The mock reverse transcription reactions, containing RNA and all reagents except reverse transcriptase, confirmed that the results obtained were not due to contaminating genomic DNAs (data not shown).
Metal Content Analysis-Whole cell metal contents were measured in M. tuberculosis strains grown to the late exponential phase and incubated in the presence or absence of 100 M ZnSO 4 for 1 h. Metals bound to M. tuberculosis secreted proteins were measured in strains grown in Sauton's medium for 2 weeks at 37°C. Cells were pelleted, and the supernatant was filtered through 0.2-m filters (Corning Inc.) and concentrated using 3 kDa cut-off centricons (Millipore). Protein content was determined, and samples were prepared for metal content analysis. In the case of M. smegmatis metal-bound secreted pro-teins, cells were grown for 48 h in the presence or absence of 50 M MnCl 2 and pelleted, and supernatants were processed as described above. To obtain cytosolic and membrane fractions, cells were grown as described and disrupted in a French press at 20,000 p.s.i., and homogenates were centrifuged at 8,000 ϫ g for 30 min. Supernatant was centrifuged at 229,000 ϫ g for 1 h. The supernatant of this centrifugation was considered the cytosolic fraction and concentrated with 3 kDa cut-off centricons. The pelleted membrane fractions were resuspended in 25 mM Tris, pH 7.0, 100 mM KCl, and protein content was determined.
All samples were acid-digested with HNO 3 (trace metal grade) for 1 h at 80°C and overnight at 20°C. Digestions were concluded by adding one-eighth volume of 30% (v/v) H 2 O 2 and a 1:5 dilution with water. Metal content in digested samples was measured by furnace atomic absorption spectroscopy (Varian SpectrAA 880/GTA 100).
Enzymatic Assay of Superoxide Dismutase Activity-M. smegmatis strains were grown in Sauton's medium without the addition of metals or in the presence of 50 M MnCl 2 , ZnSO 4 , CoCl 2 , or FeCl 2 , and secreted proteins were collected as indicated above. M. tuberculosis strains were grown in Sauton's medium, and secreted proteins were collected as stated above. Secreted fractions were resolved in 10% non-denaturing PAGE,  and gels were incubated for 30 min in 2.5 mM nitro blue tetrazolium (Sigma), followed by a 20-min incubation with 30 mM potassium phosphate, 30 mM TEMED, 30 mM riboflavin, pH 7.8 (43). Superoxide dismutase activity was visualized by illuminating gels with white light for 10 min on a transilluminator. Areas of activity were visible as white bands against a dark background. As a control, samples were run in independent gels and CBB-stained. Liquid Chromatography Mass Spectroscopy-Supernatants obtained from M. tuberculosis and M. smegmatis wild type, ctpC::hyg, and complemented strain cultures were separated by 10% non-denaturing PAGE and CBB-stained. The corresponding bands from parallel gels showing superoxide dismutase activity were excised, destained by several dehydration-rehydration steps, and trypsin-digested for subsequent MS/MS analysis. Briefly, samples were dehydrated with acetonitrile and rehydrated in 100 mM NH 4 HCO 3 , 50% (v/v) acetonitrile. When most of the dye was removed, the samples were reduced with 10 mM DTT in 100 mM NH 4 HCO 3 , incubated for 35 min at 56°C in a shaking incubator, and vacuum-dried. Alkylation was conducted by incubating for 30 min with 55 mM iodoacetamide in 100 mM NH 4 HCO 3 . After wash using 100 mM NH 4 HCO 3 , 50 l of 100 mM NH 4 HCO 3 , 50% (v/v) acetonitrile were added to the pellet before the final drying step. The sample was digested by adding 30 -50 l of sequencing grade modified trypsin (Promega) and incubated overnight at 37°C. Soluble peptides were extracted and combined with the result of three washes of 50 ml of 50% (v/v) acetonitrile, 5% (v/v) formic acid (15 min). Finally, digested samples were resolved in an Accurate-Mass Q-TOF LC/MS 6520, and peptide identification from collision-induced fragmentation patterns was performed with Spectrum Mill software (Agilent).
Mouse Infection-C57BL/6 female mice (8 -10 weeks old) were infected with ϳ1 ϫ 10 3 CFU of wild type, ctpC::hyg, or ctpC complemented M. tuberculosis via the aerosol route. Groups of three mice were sacrificed at the indicated time points, and the bacterial burden in the lung homogenates was obtained by plating on 7H10 agar medium. In alternative competition experiments, two M. tuberculosis H37Rv strains (wild type and ctpC::hyg mutant or complemented) were mixed in a 3:1 ratio (6 ϫ 10 5 CFU in a 200-l final volume) and inoculated into the tail vein of female C57BL/6J mice. Groups of three mice were sacrificed at the indicated time points, and the bacterial burden in the lung homogenates was obtained by plating on 7H10 agar medium with or without 100 g/ml hygromycin for mutant CFU and total CFU counting, respectively. Mice were housed under specific pathogen-free conditions and in accordance with University of Massachusetts Medical School Institutional Animal Care and Use Committee guidelines.

RESULTS
CtpC Is a Novel ATPase-CtpC belongs to the family of P 1B -ATPases. As such, it is characterized by the presence of eight TMs in a typical membrane topology, the cytoplasmic actuator, ATP binding, and phosphorylation domains (Fig. 1A) (19,20). This last cytosolic loop contains the DKTGT sequence common to all P-type ATPases, where the conserved Asp is phosphorylated during catalysis. Considering the conservation of major catalytic domains and transmembrane overall structure, it is not surprising that the primary sequence of CtpC is similar (26 -36% identity) to that of Cu ϩ -ATPases and Zn 2ϩ ATPases, leading to a hypothesis suggesting Cu ϩ and Zn 2ϩ as likely substrates (11,20). BLAST searches revealed the presence of proteins with structural characteristics similar to those of CtpC only in a few mycobacterial species, Rhodococcus opacus, Kineococcus radiotolerans, and Nakamurella multipartite (Fig. 1B). CtpC homologs lack the well described regulatory cytoplasmic N-MBDs that are ubiquitous in Cu ϩ -and Zn 2ϩ -ATPases (20). More important, CtpC proteins present differences in metal-coordinating transmembrane residues. Invariant amino acids in TMs flanking the ATP binding domain are critical for the metal specificity of P 1B -ATPases because these coordinate the metal substrates during transport (19 -23). Alignment of CtpC sequences corresponding to the sixth, seventh, and eighth TM segments showed the presence of two conserved Cys residues in TM6, Asn and Tyr in TM7, and His, Asn, and two Ser residues in TM8 (Fig. 1, A and B). Comparison with ligand donor amino acids in substrate binding sites of Cu ϩ -, Co 2ϩ -, and particularly Zn 2ϩ -ATPases shows significant differences, suggesting that CtpC might be involved in the transport of an alternative metal (Fig. 1C) (19 -23).
CtpC Is a Unique Mn 2ϩ -ATPase-To avoid the complexity of in vivo systems, the substrates of M. tuberculosis CtpC were biochemically characterized using purified protein. M. tuberculosis ctpC was cloned and expressed in E. coli. The resulting protein was solubilized and purified by metal affinity chromatography, and the engineered His 6 tag was removed by tobacco etch virus protease treatment ( Fig. 2A). A functional preparation was obtained by reconstituting the protein in lipid/detergent micelles. The central characteristic of the transport mechanism of all P-ATPases is the coupling of transmembrane substrate transport to ATP hydrolysis (3,19). Stimulation of ATP hydrolysis by substrates is a proven approach to evaluate the substrate specificity of these enzymes. M. tuberculosis CtpC ATPase activity was stimulated by the presence of various metals in the assay media (Fig. 2B). Under metal-saturating conditions, CtpC exhibited a maximum ATPase activity in the presence of Mn 2ϩ and to a smaller extent (25-30%) by Co 2ϩ , Cu 2ϩ , and Zn 2ϩ . It is equally significant that other metals, such as Cu ϩ , Ni 2ϩ , Fe 2ϩ , and Cd 2ϩ , did not activate CtpC (Fig. 2B) because Zn 2ϩ -ATPases are also activated by Cd 2ϩ (44,45), and a role in Fe 2ϩ transport might be suggested by certain phenotypic observations (see below). Cleavage of the His 6 tag or the presence of thiols in the assay media had no effect on metal activation patterns (data not shown). Fig. 2C shows the dependence of CtpC ATPase activity on the Mn 2ϩ and Zn 2ϩ concentration. Interestingly, CtpC turnover (V max ) ( Table 2) is remarkably slow when compared with enzymes involved in metal detoxification and tolerance, such as E. coli CopA (10fold higher V max ) (46), Pseudomonas aeruginosa CopA1 (4-fold higher) (41), E. coli ZntA (10-fold higher) (44), or Archaeoglobus fulgidus CopA (4-fold higher) (33). In fact, CtpC V max is more similar to P. aeruginosa CopA2, a slow transporting Cu ϩ -ATPase involved in cytochrome c oxidase assembly (41). The apparent high affinity of CtpC for Mn 2ϩ is also consistent with this function. Although the estimation of relative metal affini-ties requires equilibrium binding experiments and the low ATPase activity hampers precise activation K1 ⁄ 2 determinations, it is clear that CtpC interacts with Mn 2ϩ and Zn 2ϩ with very high affinities ( Table 2). The CtpC estimated K1 ⁄ 2 value for activation is ϳ1-2 orders of magnitude lower than those seen in other P 1B -ATPases (33,44,46,47). These observations were further confirmed by analysis of the homologous M. smegmatis CtpC. The heterologously expressed and purified M. smegmatis protein showed metal activation and biochemical parameters identical to those of the M. tuberculosis CtpC (Table 2).
Invariant amino acids located in TMs determine metal specificity and, as metal ligands, are required for enzyme function (19 -23). Although a full analysis of metal coordination by CtpC is beyond this report, the importance of the distinct HXXSS sequence in TM8 of M. tuberculosis CtpC was explored. As expected, proteins carrying point mutant H699A and neighbor-ing S700A/S701A showed no ATPase activity at saturating Mn 2ϩ levels (Fig. 2D). This supports the link between the observed Mn 2ϩ -dependent ATPase activity and the binding of this metal to a transmembrane transport site distinct from those in Cu ϩ -or Zn 2ϩ -ATPases.
The observed biochemical properties, low transport rate, and high apparent metal affinity suggest that CtpC might perform a role other than detoxifying the cells by exporting cytoplasmic metals into the extracellular media (3,41). However, the modest activation of CtpC by Zn 2ϩ and the previously proposed role for this protein as a Zn 2ϩ -ATPase controlling cytoplasmic levels of this metal (11) prompted us to confirm that this protein's specificity was not altered in vivo. To test this, we complemented a zntA::kan RW3110 E. coli strain, lacking the functional Zn 2ϩ -ATPase (32), with M. tuberculosis CtpC. Fig. 3A shows that, although Zn 2ϩ tolerance is restored by comple- menting E. coli with ZntA, M. tuberculosis CtpC introduced in a similar construct was unable to rescue the E. coli zntA mutant when grown in the presence of high Zn 2ϩ levels. Fig. 3B shows the expression of CtpC under the tested conditions. These results indicate that CtpC is unlikely to play a role in controlling cytoplasmic Zn 2ϩ levels.
CtpC Deficiency Renders M. tuberculosis Sensitive to Zn 2ϩ and Oxidative Stress-It has been reported that deletion of ctpC decreases the Zn 2ϩ tolerance of the M. tuberculosis GC1257 strain (11). We observed a similar phenotype in M. tuberculosis H37Rv ctpC::hyg cells when grown at Zn 2ϩ concentrations as low as 5 M (Fig. 4A). However, no changes were observed in the sensitivity to Cd 2ϩ , a metal that usually shows similar toxicity in Zn 2ϩ mutants or to the enzyme substrate Mn 2ϩ (Fig. 4,  B and C). Similarly, the presence of Co 2ϩ or Cu 2ϩ in the medium had no effect on the growth of these cells (data not shown). Reasoning that Zn 2ϩ could impose oxidative stress, we determined if the ctpC mutant was sensitive to other oxidants. Indeed, we found that this mutant was hypersensitive to both tert-butyl hydroperoxide (TBHP; Fig. 4D) and superoxide (generated with hypoxanthine/xanthine oxidase; Fig. 4E). In both cases, the wild type phenotype was restored in the complemented strain. Correlating with the alterations in Zn 2ϩ tolerance, a significant induction of ctpC expression in response to Zn 2ϩ was detected (Fig. 4F) (11). This induction was not observed when cells were challenged with other tested stressors (Fig. 4F). These data imply a complex system in which ctpC expression is neither induced by nor confers resistance to its substrate Mn 2ϩ . However, the enzyme is required for tolerance to Zn 2ϩ and oxidative stress. To explain these observations, the role of CtpC in M. smegmatis was studied. Surprisingly, the M. smegmatis ctpC::hyg strain showed no higher sensitivity to Zn 2ϩ (Fig. 5A), although it behaved as the M. tuberculosis mutant in the presence of Mn 2ϩ (no effect; Fig. 5B) and redox stress (increased sensitivity; Fig. 5C). Furthermore, contrary to the observation that Zn 2ϩ stimulates M. tuberculosis ctpC expression, no induction was observed in M. smegmatis when exposed to the metal (Fig. 5D). Interestingly, a 15-fold increase in ctpC expression was observed when M. smegmatis was exposed to the extracellular superoxide generator hypoxanthine/xanthine oxidase.
Deletion of ctpC Leads to Cytoplasmic Mn 2ϩ Accumulation and a Decrease in Secreted Mn 2ϩ -bound Proteins-Based on the role of well described P 1B -ATPases in controlling cellular    (Table 3). Under basal conditions, no changes in transition metal levels were observed except for a modest but significant increase of Mn 2ϩ content. Considering the described Zn 2ϩ accumulation in M. tuberculosis GC1237 ctpC mutant cells (11), Zn 2ϩ levels under similar Zn 2ϩ stress conditions were also analyzed. No Zn 2ϩ accumulation was detected in our system (Tables 3 and  5). Mn 2ϩ homeostasis was further analyzed in cells grown in the presence of 50 M Mn 2ϩ . Taking into account the proposed function of P 1B -ATPases on the metallation of periplasmic/secreted proteins, Mn 2ϩ bound to periplasmic/secreted protein fractions was measured. To avoid interference from the presence of albumin in 7H9 medium, these experiments were performed in the more defined Sauton's medium. A 4-fold increase in cellular Mn 2ϩ content was observed in the M. tuberculosis ctpC mutant, along with a significant decrease in the Mn 2ϩ bound to secreted proteins ( Fig. 6A and Table 4). On the contrary, the levels of Zn 2ϩ , Fe 2ϩ , or Cu 2ϩ bound to secreted proteins were not affected in the M. tuberculosis or M. smegmatis ctpC mutant strain (Tables 4 and 5). M. smegmatis ctpC mutant cells allowed further analysis of subcellular fractions (Fig. 6B). A large increase of Mn 2ϩ bound to cytosolic proteins was observed, along with a significant decrease in Mn 2ϩ bound to secreted proteins. These results confirmed the function of CtpC as a Mn 2ϩ -ATPase involved not only in maintaining cytoplasmic metal quotes but probably participating in the metallation of secreted Mn 2ϩ proteins. Moreover, these observations suggest that CtpC might not be involved in conferring Zn 2ϩ tolerance but rather in an alternative process, such as the response to redox stress. In the case of M. tuberculosis, this response to redox stress might be triggered by high Zn 2ϩ levels.
CtpC Is Required for Functional Secreted SodA-The metallation of secreted Mn 2ϩ -bound proteins appeared to be deficient in ctpC mutants, and some of these may be required for overcoming redox stress. Mn-SodA was a logical candidate for CtpC-mediated metallation. In mycobacteria, the SodA protein is secreted in the apo-form via the Sec pathway (18). Consequently, SodA probably acquires its metal in the extracellular milieu. The metal specificity of SodA-like enzymes is not predictable based on primary protein sequence. Although M. tuberculosis SodA appears to contain Fe 2ϩ , other highly homologous SodAs are Mn 2ϩ enzymes (17,48,49). Moreover, cambialistic properties have been observed in these enzymes, suggesting that distinct metals might be used by the same enzyme under different conditions. To test the putative role of CtpC in the metallation of secreted SodA, in-gel superoxide dismutase activity assays were performed. The secreted fraction from M. tuberculosis ctpC::hyg strain showed a 37% decrease in the activity of SodA compared with wild type (Fig. 7,  A and B). The SodA activity was partially recovered in the CtpC complement strain when compared with wild type, probably reflecting the levels of CtpC expression in the complemented cells. Interestingly, when analyzing the secreted SodA from M. smegmatis ctpC::hyg mutant, the activity decreased up to 80% compared with the SodA obtained from the wild type strain (Fig. 7, C and D). CBB staining of identical gels showed no decrease SodA protein secretion (Fig. 7C, bottom). The identity of secreted SodA was verified by liquid chromatography/mass spectrometry (LC/MS; Table 6). Because mycobacteria also have cytoplasmic Cu/Zn-SodCs, it is also relevant that no differences were found when analyzing total superoxide dismutase activity in whole cell homogenates or cytosolic or membrane fractions obtained from M. smegmatis wild type, ctpC::hyg, and the complemented strains (data not shown).
Because of the mentioned cambialistic properties of SodA (50 -52), we determined if M. smegmatis SodA activity could be restored by other metal co-factors independently of CtpC activity. M. smegmatis wild type, ctpC::hyg, and the complemented strains were grown in Sauton's medium supplemented with 50 M MnCl 2 , FeCl 2 , ZnSO 4 , or CoCl 2 until late log phase. In-gel superoxide dismutase activity assays showed that SodA activity was restored when adding MnCl 2 or FeCl 2 to the culture medium of the M. smegmatis ctpC::hyg mutant strain (Fig. 7C). The addition of ZnSO 4 and CoCl 2 had no effect on this activity. No differences were observed in the activity of the secreted SodA from the wild type and complemented strains (data not shown). These data further support an important role for CtpC in the Mn 2ϩ metallation of mycobacterial SodA during metal starvation stress. Fe 2ϩ metallation appears to be unaffected, and under certain culture conditions, this prosthetic group may be used (17). The alternative use of these metal cofactors by SodA is addressed under "Discussion." Considering the requirement of CtpC for Mn 2ϩ secretion and SodA metallation and the lack of ctpC induction under Mn 2ϩ excess (Figs. 4F and 5D), it is tempting to hypothesize that under Mn 2ϩ starvation, ctpC expression should be induced. Although all of the media used in these studies have no Mn 2ϩ formally included, interestingly, a 3.0 Ϯ 0.3-and a 2.12 Ϯ 0.06-fold increase in ctpC expression was observed when M. smegmatis cells were cultured in chelexed (0.3 nM Mn 2ϩ ) and non-chelexed (0.4 nM Mn 2ϩ ) Sauton's media, respectively, compared with those maintained in 7H9 medium (0.9 nM Mn 2ϩ ). Although small changes in Mn 2ϩ levels might affect ctpC transcription, these are very small and unlikely to be of physiological significance.
CtpC Is Important for M. tuberculosis Virulence-Large scale genetic studies have predicted that ctpC is required for the growth or survival of M. tuberculosis H37Rv in the C57BL/6 mouse model of tuberculosis (24). More recent studies have reported that mutation of the ctpC gene impairs the intracellular growth of M. tuberculosis strain GC1237 in human macrophages but does not alter the virulence of this strain in BALB/C    or SCID mice (11). To verify the importance of CtpC for M. tuberculosis virulence, we assessed the growth of the ctpC::hyg mutant strain in two different models of tuberculosis. First, M. tuberculosis H37Rv, ctpC::hyg, and ctpC complemented strains were used to infect the relatively resistant C57BL/6 mouse strain via the low dose aerosol route. The ctpC::hyg strain grew similarly to wild type M. tuberculosis during the initial phase of infection that is governed by innate immune mechanisms. However, a significant decrease in viable ctpC::hyg bacteria was noted after a month of infection (Fig.  8A). This specific persistence defect is similar to that of mutants that are unable to resist the adaptive immune response (53). The attenuation of this mutant was confirmed using a more sensitive competitive model in which the ratio of mutant versus wild type bacteria was monitored in animals that were co-infected via the intravenous route. As we observed in the singlestrain infections, the ctpC::hyg mutant was significantly underrepresented after 42 days of infection (Fig. 8B). The attenuation of this strain was observed at earlier time points in the competitive model (21 days), probably due to the earlier onset of adap-tive immunity after intravenous challenge (54). No differences in growth were observed between wild type and complemented strains in vitro (not shown), and genetic complementation reversed the in vivo growth/survival defects of this mutant. Thus, ctpC is required for the bacterium to adapt to the host environment and appears to be specifically important for resisting the adaptive immune response.

DISCUSSION
Large scale genetic screens implicated ctpC in the adaptation of M. tuberculosis to the host environment (24). Recent studies have shown that an M. tuberculosis ctpC mutant is sensitive to Zn 2ϩ (11), a metal that appears to accumulate in the phagosome (9). Taking into account these phenotypes and CtpC homology with Cu ϩ -and Zn 2ϩ -ATPases, it was proposed that CtpC was a Zn 2ϩ -ATPase involved in the efflux of cytoplasmic Zn 2ϩ (11). However, CtpC lacks signature residues that are part of the transmembrane metal binding sites of Cu ϩ -and Zn 2ϩ -ATPases, suggesting that it might transport alternative metals. Considering its putative importance for virulence and unique structure, the function of CtpC in M. tuberculosis and M. smegmatis was examined using a combination of biochemical and genetic approaches. Here, we describe that CtpC shows a preference for Mn 2ϩ , controls the Mn 2ϩ cytoplasmic quota, and is involved in the uploading of Mn 2ϩ into secreted metalloproteins. We propose that these activities of CtpC are important for M. tuberculosis virulence.
CtpC is a metal transport P 1B -ATPase with characteristic phosphorylation domains and membrane topology, although it lacks the regulatory N-MBDs ubiquitous in enzymes with specificity for Cu ϩ or Zn 2ϩ (19,20). It is now well established that metal specificity is conferred by conserved residues in the TMs flanking the large cytoplasmic ATP binding and hydrolysis domain (19 -23). Consequently, despite the significant similarity between CtpC and Cu ϩ /Zn 2ϩ -ATPases, it lacks the sets of invariant metal-coordinating residues found in well characterized P 1B -ATPases, particularly the Lys and Asp in TM7 and TM8 that characterize Zn 2ϩ -ATPases (20,22). The analysis of the 13 sequences of homologous CtpC proteins showed that although these have retained the CPC signature in TM4, they have a unique NY in TM7 and HNASS in TM8. The presence of these unique residues is not trivial because their replacement leads to inactive proteins. It is interesting that although these residues resemble the Cu ϩ binding sequences (19 -21), the replacement of Met for His (a harder Lewis base) is consistent with selectivity for Mn 2ϩ (a hard Lewis acid), the substrate of CtpC identified in our studies. Moreover, the Ser (a hard Lewis base) conserved in CtpC provides further specificity toward Mn 2ϩ . This Ser is absent in Zn 2ϩ sites (55,56).  Direct biochemical analysis of CtpC activity showed that maximum activation of the transporter occurs in the presence of Mn 2ϩ . However, Zn 2ϩ , Co 2ϩ , and even Cu 2ϩ can act as alternative substrates although at slower turnover rates. This is not surprising because many Zn 2ϩ -ATPases and Cu ϩ -ATPases can transport alternative metals sharing similar chemical properties (33,44,47,57). Furthermore, it is significant that CtpC is not activated by Cd 2ϩ , a well characterized activator of Zn 2ϩ -ATPases (44,58). However, it is perhaps more important to consider the high apparent affinity of CtpC for Mn 2ϩ and that its activity is relatively slow compared with that of enzymes responsible for maintaining cytoplasmic metal quotas. If the reported activity of E. coli Zn 2ϩ -ATPase is considered, the Zn 2ϩ -ATPase activity of CtpC is 2 orders of magnitude lower when measured under practically identical conditions. This kinetic difference probably explains the inability of CtpC to confer Zn 2ϩ tolerance and functionally complement a ⌬zntA E. coli mutant. We hypothesize that in M. tuberculosis, the necessary Zn 2ϩ efflux is mediated by the cation diffusion facilitator coded by Rv2025. This is homologous to the M. smegmatis Zn 2ϩ transporter ZitA (59) and E. coli ZitB (60).
Whereas biochemical determinations provide a direct evaluation of heavy metal transport ATPase substrates, characterization of cellular roles might provide further insight into the enzyme selectivity. Determination of metal contents and tolerance to metal stress as well as induction of gene expression by putative substrates are commonly used indicators of substrate specificity. We observed little or no sensitivity of the ctpC::hyg cells to various metal stressors, except for the described lack of tolerance to Zn 2ϩ (11). This phenotype was not observed in M. smegmatis, suggesting that M. tuberculosis is particularly sensitive to Zn 2ϩ or to an associated stress (see below). Interestingly, the lack of Zn 2ϩ tolerance was correlated with a lack of induction of the ctpC gene by Zn 2ϩ in M. tuberculosis. On the other hand, it was also remarkable that the Cd 2ϩ did not induce ctpC expression, suggesting that ctpC expression is not driven by Zn 2ϩ -sensing transcriptional regulators (61).
We observed that mutation of ctpC led to an increase in intracellular Mn 2ϩ levels but did not affect the cellular Zn 2ϩ , Co 2ϩ , or Fe 2ϩ content. Mn 2ϩ is apparently nontoxic to the cells because high (millimolar) extracellular Mn 2ϩ is necessary to affect cell growth. Noticeably, the Mn 2ϩ accumulation was observed even in basal medium conditions; it was more prominent in cells grown in 50 M Mn 2ϩ . These findings are consistent with our biochemical data, although we did not observe the previously described Zn 2ϩ accumulation in ctpC mutant cells under Zn 2ϩ stress conditions (11). These differences might be associated with alternative Zn 2ϩ efflux systems present in the different strains of M. tuberculosis used in these experiments. Previous work employed the M. tuberculosis GC1237 strain, whereas we characterized the ctpC::hyg mutant in the H37Rv background. Still, our in vivo observations support the conclusion that this CtpC functions primarily as a Mn 2ϩ transporter.
Given this biochemical function, the Zn 2ϩ sensitivity and the Zn 2ϩ -dependent ctpC expression observed in M. tuberculosis are unexpected. It might be considered that metal homeostasis is linked to other cellular phenomena (e.g. the redox equilibrium). In particular, the association of Zn 2ϩ , Mn 2ϩ , and Fe 2ϩ transporters in several redox-responsive regulons has been documented (62,63). Expression of the ctpC gene is induced in low Fe 2ϩ conditions (27), and it is possible that high Zn 2ϩ concentrations might mimic Fe 2ϩ deficiency. Alternatively, a link between Zn 2ϩ and factors has been proposed in M. tuberculosis (64), and reactive oxygen species have been shown to affect the signaling pathways involved in the activation of multiple transcription factors (65). Considering the observation that ctpC::hyg cells are also sensitive to redox stressors, it is tempting to speculate that the sensitivity to Zn 2ϩ is related to alteration in Fe 2ϩ or redox homeostasis rather than to an increase in cytosolic Zn 2ϩ . The increased susceptibility of both mycobacterial ctpC::hyg mutants to the oxidative stress produced by xanthine oxidase supports these ideas.
An alternative link between Zn 2ϩ and redox stress might also be considered. In M. tuberculosis, metallothioneins are primary barriers against Cu ϩ toxicity (66). There is evidence that these proteins also bind Zn 2ϩ and Cd 2ϩ with high affinity. Both ions could generate an imbalance in cell redox through mechanisms such as mycothiol consumption, thioredoxins inhibition, or Cu ϩ displacement by competition when present at high con-  centrations (67). The addition of Zn 2ϩ to the media might release Cu ϩ from oxidized metallothioneins, the former generating reactive oxygen species by means of the Fenton reaction. Contrary to Zn 2ϩ , 50 M Cd 2ϩ induces the transcription of metallothioneins in M. tuberculosis (66). The induction of metallothioneins by Cd 2ϩ might explain why M. tuberculosis ctpC is not sensitive to Cd 2ϩ when compared with the wild type strain H37Rv. CtpC is required for M. tuberculosis growth and survival in the mouse model. The importance of Mn 2ϩ for the virulence of other bacterial pathogens has been described, although Mn 2ϩ levels in the phagosome might not change dramatically during infection (9,49). Our biochemical and phenotypic analyses indicate that CtpC, although it partially influences cytoplasmic Mn 2ϩ levels, is a slow ATPase, its expression is not induced by its substrate, and CtpC-deficient cells exhibit normal tolerance to Mn 2ϩ metal excess. These are all characteristics similar to those observed in Cu ϩ -ATPases responsible for assembly of cytochrome c oxidase instead of cytosolic detoxification (41). In agreement with this putative role, the ctpC::hyg mutant showed a decrease in the amount Mn 2ϩ -bound to secreted proteins. Further experiments showed that deletion of ctpC reduced secreted Mn 2ϩ /Fe 2ϩ -SodA activity without altering the abundance of the protein, suggesting that the enzyme might not be metallated. Supporting this hypothesis, M. smegmatis SodA activity was restored by supplementing the growth medium with either Mn 2ϩ or Fe 2ϩ . Secreted metalloproteins, such as SodA, have been shown to be required for virulence (14,16). Both M. tuberculosis and M. smegmatis SodA are highly homologous and lack a classical signal sequence for protein export (68), although M. tuberculosis SodA is exported via the SecA2 pathway, which exports unfolded apoproteins (18). Interestingly, one important difference between them is the metal co-factor either SodA acquires in vitro. M. tuberculosis SodA has been characterized as an Fe-superoxide dismutase, whereas M. smegmatis SodA contains Mn 2ϩ (17). However, M. smegmatis SodA has a cambialistic behavior in vitro, being activated by Fe 2ϩ when the pH is acidic (48). These enzymes differ in one of the metal-coordinating residues; M. tuberculosis contains a His instead of a Gln at position 145, which is present in M. smegmatis Mn 2ϩ -SodA. However, replacing His-145 with Gln does not affect the metal binding in M. tuberculosis, as determined by x-ray crystallography (17). The specificity for iron of the tuberculosis SodA seems to be counterintuitive, considering that upon macrophage infection, M. tuberculosis cells must face and overcome iron starvation (8). Thus, it is possible that M. tuberculosis SodA is preferentially loaded with Mn 2ϩ in the iron-limited in vivo environment.
In summary, the data presented indicate that CtpC is a unique Mn 2ϩ -ATPase predominantly present in mycobacteria. The enzyme appears to be required for loading Mn 2ϩ into secreted metalloproteins. In this role, the CtpC is a key element for virulence.