Zn2+ Intoxication of Mycobacterium marinum during Dictyostelium discoideum Infection Is Counteracted by Induction of the Pathogen Zn2+ Exporter CtpC

Microelements are essential for the function of the innate immune system. A deficiency in zinc or copper results in an increased susceptibility to bacterial infections.

M ycobacterium marinum is a pathogen that causes a tuberculosis-like infection in fish and amphibians, in which it generates granulomas very similar to those caused in humans by its close relative Mycobacterium tuberculosis (Mtb). M. marinum also opportunistically infects humans, but these infections are limited to the skin and extremities due to their lower temperature (1). At the level of the host macrophage, the course of infection by both M. marinum and Mtb is strikingly similar, making M. marinum an experimentally versatile model to study pathogenic mechanisms of tuberculosis (2). Apart from fish and mammalian phagocytes, another model host for M. marinum is the soil amoeba Dictyostelium discoideum (3), the endocytic and cell-autonomous defense pathways of which are well conserved with humans (4).
Upon phagocytic uptake, M. marinum resides inside a phagosome that is actively modified by the bacterium to rapidly divert from the canonical maturation pathway, to become a replicative niche, the so-called Mycobacterium-containing vacuole (MCV). Like Mtb, M. marinum was thought to be a vacuolar pathogen. However, accumulating evidence indicates that both bacteria also reside and proliferate in the cytosol, where access to nutrients is not limited but bacteria are exposed to cytosolic defenses (3,5,6). M. marinum starts damaging the MCV in the first hours of infection, before full rupture of the compartment releases the bacteria into the host cytosol at around 24 to 36 h. As also shown for Mtb, perforation of the MCV and escape to the cytosol are achieved, among other means, by secretion of the membrane-damaging peptide ESAT-6 through the mycobacterial ESX-1 type VII secretion system (T7SS) (3,(7)(8)(9)(10)(11)(12)(13).
Infected cells use a vast repertoire of strategies to fight intracellular mycobacteria. For instance, M. marinum infection has been shown to induce inducible nitric oxide synthase (iNOS) gene expression in fish macrophages (14), and neutrophils kill these bacilli through reactive oxygen species (ROS)-dependent mechanisms (15). In addition, M. marinum is targeted for digestion by the autophagy machinery of murine macrophages (16) and D. discoideum (3,8). However, these mycobacteria have also evolved mechanisms of counterdefense against their hosts. They downregulate iNOS levels (17) and suppress the production of reactive nitrogen intermediates, decrease the expression of NADPH oxidase components, and reduce the production of ROS (14). Moreover, M. marinum avoids phagosome maturation by (i) modulating the composition of its cell wall (18) or the MCV content in phosphatidylinositol-3-phosphate (PtdIns3P [19]), (ii) impairing the recruitment of the endosomal sorting complex required for transport (ESCRT)-0 component Hrs (20), and (iii) avoiding accumulation of lysosomal enzymes (3,16,19,21,22). It also blocks the autophagic flux, which would eventually deliver the bacteria into autolysosomes for killing (3,16).
Apart from the defense mechanisms mentioned above, little is known about the chemical warfare between M. marinum and its hosts, and especially about the manipulation of transition metals. In the case of Mtb, immune cells deprive the bacteria of essential nutrients such as iron and manganese, while they intoxicate the mycobacteria by accumulating copper and zinc inside the MCV (23)(24)(25)(26). However, Mtb resists this metal-centered fight with an arsenal of metal-binding proteins, oxidases, and efflux transporters. For instance, Mtb captures Fe 31 from macrophages through its siderophore mycobactin (27), but it keeps low intracellular Cu 21 levels with its CtpV and MctB transporters and the Cu 21 -binding metallothionein MymT (24,28,29). Mtb also resists the toxic free Zn 21 burst induced in human macrophages by exporting Zn 21 through the P-type ATPase CtpC (30), an efflux pump also present in two isoforms in M. marinum (31). On the host side, the mRNA levels of various Zn 21 transporters (i.e., ZIP8, ZnT1, ZIP1, and ZIP10) are altered in fish granulomas compared to resting macrophages (32). However, little is known about the conservation of these host-versus-pathogen strategies, and especially how the two mycobacterial CtpC and host Zn 21 transporters impact M. marinum infection in D. discoideum.
D. discoideum possesses four Zn 21 efflux transporters of the ZnT family (Znts) and 11 Zn 21 influx transporters of the ZIP family (Zpls). ZntA locates to the contractile vacuole (CV), an organelle that regulates the osmotic and metal balances in the cell, and ZntB to organelles of the endosomal pathway (33,34). While ZntA does not have any close homologue, ZntB shares homology with human ZNT1 and ZNT10, the latter being also present at early endosomes (4). Similarly to their human homologs ZNT6 and ZNT7, located in the early secretory pathway (4), D. discoideum ZntC and ZntD localize to the Golgi complex or recycling endosomes (33). We wondered whether these ZnTs were part of the host defense mechanisms against M. marinum infection. We demonstrate here that D. discoideum intoxicates M. marinum by inducing the expression and recruitment of ZntA and ZntB, but not ZntC nor ZntD, to the MCV. However, M. marinum resists the D. discoideum Zn 21 immune response by specifically expressing its Zn 21 exporter CtpC.

RESULTS
Free Zn 2+ accumulates within intact M. marinum-containing vacuoles. In D. discoideum, free Zn 21 accumulates inside (i) the CV network, (ii) zincosomes (endosomal compartments with lysosomal and postlysosomal characteristics), and (iii) phagosomes containing beads and nonpathogenic bacteria such as Escherichia coli or Mycobacterium smegmatis (33, 35) (summarizing scheme is in Fig. 1A). We wondered whether vacuoles containing the pathogenic bacterium M. marinum or M. marinum DRD1, a mutant lacking a functional ESX-1 secretion system and thus strongly attenuated in inducing membrane damage and escaping to the cytosol (8,9), also accumulate Zn 21 . We monitored infected cells expressing the endosomal and MCV marker AmtA-mCherry (36) and labeled with the pH-independent Zn 21 probe NBD-TPEA (33) (Fig. 1B). Intact MCVs of the DRD1 mutant accumulated Zn 21 at all the stages of infection ( Fig. 1B and C). In contrast, numerous compartments of M. marinum wild type (wt) appeared devoid of Zn 21 , likely because it leaked out of either subtly or visibly broken MCVs (Fig. 1B). Quantification of NBD-TPEA-labeled MCVs confirmed that, because damage increases as infection progresses, fewer and fewer compartments containing wt bacteria were positive for Zn 21 (Fig. 1C).
M. marinum senses and reacts to toxic levels of Zn 2+ by inducing its CtpC Zn 2+ efflux pump. During infection of macrophages, Mtb is exposed to a burst of transition metals that induces the expression of efflux P-type ATPases such as CtpC, CtpG, and CtpV (30,37). In particular, CtpC expression helps Mtb detoxify excess Zn 21 , which the host transports inside the MCV (30). The expression of Mtb ctpV is also induced during macrophage infection (28), which is essential for virulence and survival in response to elevated copper levels in the MCV (28). The genome of M. marinum encodes 20 P-type ATPases that are phylogenetically close to their Mtb homologues (see Fig. S1 in the supplemental material). Interestingly, M. marinum has two homologues of CtpC, CtpC (MMAR_1271) and CtpC-like (MMAR_2147). Inspection of our parallel RNA-sequencing (RNA-Seq) results of M. marinum and D. discoideum during infection (38) revealed that, as shown for Mtb, expression of the ctpC gene increased at all time points and reached its maximum 36 h postinfection (hpi) compared to M. marinum grown in broth ( Fig. 2A,  Fig. S2, and Table S1A). Expression of the ctpc-like gene was increased only later during infection and also peaked at 36 hpi ( Fig. 2A, Fig. S2, and Table S1A). These results suggest a possible specific role of the two CtpC P-type ATPases in the detoxification of Zn 21 during infection. It is noteworthy that other P-type ATPases were differentially expressed during infection, including the three M. marinum CtpV homologues, suggesting a complex scenario in which M. marinum faces elevated levels of other ions ( Fig. S2 and Table S1A).
We next attempted to evaluate whether the expression of ctpC is influenced by Zn 21 . To this end, the expression levels of selected P-type ATPases were determined by qRT-PCR analysis. Supplementing the growth medium with increasing concentration of ZnSO 4 strongly enhanced the transcription of ctpC but not ctpC-like, ctpV, or ctpV-like (Fig. 2B). These results suggest that CtpC is exclusively involved in detoxification of Zn 21 in M. marinum and that CtpC-like might function under different circumstances than those tested.
To investigate the physiological role of CtpC, we constructed a deletion mutant of ctpC (MMAR_1271; Fig. 2B). M. marinum wt and its isogenic mutant DctpC grew similarly in broth and on agar, indicating no constitutive growth defect ( Fig. S3A and E to G). However, the growth of M. marinum DctpC in broth was impaired in the presence of Zn 21 in a dose-dependent manner ( Fig. 2C and Fig. S3B to E). In addition, its growth was impaired on solid medium containing elevated levels of Zn 21 but not of Mn 21 or Cu 21 (Fig. S3E to G), demonstrating that the absence of ctpC renders M. marinum specifically susceptible to Zn 21 . Strikingly, the intracellular growth of M. marinum DctpC was notably impaired (Fig. 2D). Most importantly, both the growth defect of DctpC in broth supplemented with Zn 21 (Fig. 2C) and the intracellular growth (Fig. 2D) were fully rescued by constitutive expression of the Mtb CtpC orthologue (Rv3270). Together, these results demonstrate that the susceptibility of M. marinum DctpC to Zn 21 is solely due to the lack of CtpC and that MMAR_1271 is the functional orthologue of Mtb CtpC. We conclude that, during infection of D. discoideum, M. marinum is exposed to toxic concentrations of Zn 21 and that the bacteria counteract this intoxication by specifically inducing the Zn 21 efflux pump CtpC.
The D. discoideum Zn 2+ transporters ZntA and ZntB localize to the M. marinum MCV. We wondered which of the four D. discoideum Zn 21 efflux transporters ZntA, ZntB, ZntC, and ZntD (33) localized to the MCV and led to Zn 21 accumulation. To ensure optimal preservation of membranes and organelles including the MCV and CV for immunofluorescence, we performed fixation/permeabilization in ultracold methanol (39,40). Only ZntA and ZntB ( Fig. 3A and B), but neither ZntC nor ZntD ( Fig. S4A and B), localized to the MCV. While 20% of the MCVs were positive for ZntA (Fig. 3C), ZntB was present at around 50 to 60% of the MCVs throughout the infection cycle (Fig. 3D). Because ZntA is localized exclusively at the CV membrane in noninfected cells (33), we confirmed its localization at the MCV by live imaging. Strikingly, besides ZntA (Fig. S4C) other CV resident proteins such as Rab11a and Rab11c (41) were present at the MCV ( Fig. S4D and E), emphasizing a possible cross talk between the two To confirm the successful deletion of ctpC, the mRNA levels in M. marinum DctpC were tested as well. Shown are mean and standard deviations from two independent experiments. Statistical differences were calculated with an unpaired t test (*, P , 0.05). (C) Dose-response curves of M. marinum wt, DctpC, and DctpC::ctpC grown in broth supplemented with increasing concentrations of ZnSO 4 (0, 0.125, 0.25, 0.5, and 1 mM). The fluorescence intensities of GFP and E2-Crimson were used as a proxy for bacterial growth. The areas under the curve were calculated for the three strains, and the values were plotted as a function of ZnSO 4 concentrations. Error bars indicate the SD from eight technical replicates from two independent biological replicates. Statistical differences were calculated with a Fisher LSD test after two-way ANOVA (****, P , 0.0001). (D) D. discoideum was infected with GFP-or E2-Crimson-expressing M. marinum wt, DctpC (GFP expressing), or DctpC::ctpC (E2-Crimson expressing). The intracellular bacterial growth (in relative fluorescence units [RFUs]) was monitored every hour. Shown is the fold increase in bacterial fluorescence over time of two independent biological replicates. Error bars indicate the SD from eight technical replicates. Statistical differences were calculated with Dunnett's multiple-comparison test after two-way ANOVA (****, P , 0.0001).
Host-Directed Zn 21 Intoxication of M. marinum ® organelles. CV proteins were also found to be enriched in the MCV proteome ( Fig. S4F and Table S2) (42). In addition, differential expression of genes encoding many CV proteins (38) suggests a role for this organelle during infection ( Fig. S4F and Table S1B) (38).
In summary, the data suggest that ZntB is the main MCV Zn 21 transporter but that ZntA might also contribute to the import of Zn 21 into the MCV (summarizing scheme is in Fig. 3E).
The expression of Dictyostelium ZnTs is altered during infection with M. marinum. M. marinum infection in zebrafish modulates the expression of various Zn 21 transporters in granulomas (32). Further inspection of our dual RNA-Seq results of D. discoideum during infection with M. marinum (38) showed that the expression of zntA and zntB was upregulated at very early, mid-, and late stages of infection, with the highest levels found for zntB at 1 hpi ( Fig. 3F and Table S1C) (38). In contrast, the expression of zntC and zntD was not or only weakly affected. We conclude that  Table S1C. Cells were either infected with GFP-expressing M. marinum wt or mock infected, and samples were collected at different hpi. The time points with statistically significant differential expression are marked with asterisks (*, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001). Colors indicate the strength of expression of zntA to zntD (in logFC) in infected cells compared to mock infected: from dark red (highest expression) to dark blue (lowest expression). expression of ZntA and ZntB increases during M. marinum infection and also that their localization at the MCV contributes to the accumulation of Zn 21 inside the compartment. On the other hand, ZntC and ZntD, which are absent from the MCV, either are not relevant or might only play an indirect role by contributing to an accumulation of Zn 21 in other organelles that are linked to the MCV by membrane transport.
Knockout of ZntA and ZntB impact on the concentration of Zn 2+ inside the MCV. We had previously shown that in the absence of ZntA, lumenal Zn 21 increases in acidic zincosomes and phagosomes, while in cells lacking ZntB, zincosomes contain low levels of Zn 21 (33). To determine the relevance of ZntA and ZntB in the restriction of M. marinum by Zn 21 , we infected cells lacking one or the other transporter and monitored the levels of Zn 21 inside early, intact MCVs (Fig. 4A to D). Similarly to bead-  Fig. 4E and F).
Zn 2+ restricts M. marinum during infection. Accumulation of higher Zn 21 levels in phagosomes of D. discoideum zntA KO cells results in a more efficient killing of the nonpathogenic bacterium E. coli (33). Plaque formation by D. discoideum on lawns of "food" bacteria spiked with pathogenic bacteria allows monitoring of either the virulence of the pathogen or the resistance/susceptibility of D. discoideum to these pathogens (43). Monitoring of plaque formation on lawns containing M. marinum confirmed that the zntA KO was less affected by the presence of M. marinum than wt D. discoideum ( Fig. S5A and B). Interestingly, both D. discoideum wt and zntA KO grew better on lawns containing the attenuated M. marinum DctpC mutant ( Fig. S5A and  B), confirming a Zn 21 -dependent defense process. In contrast, D. discoideum zntB KO cells did not benefit from a growth advantage on M. marinum-containing lawns ( Fig. S5C and D), corroborating that the uptake, killing, and digestion of food bacteria are unaffected in these cells, although they have lower lumenal Zn 21 concentrations. Importantly, plaque formation on a variety of nonpathogenic bacteria was comparable for zntA KO, zntB KO, and wt cells (Fig. S5E and F). We conclude that the resistance of D. discoideum is proportional to both the level of Zn 21 within the MCV (which is higher in the zntA KO) and the capacity of M. marinum to detoxify its own intracellular Zn 21 (33).
Next, we monitored the impact of znt KOs on M. marinum intracellular growth. Strikingly, absence of ZntA led to a decrease in the intracellular load of M. marinum (Fig. 5A), suggesting that the bacteria encounter noxious levels of Zn 21 during infection ( Fig. 4A and C). This was further corroborated by the fact that compared to M. marinum wt, the intracellular load of M. marinum DctpC, which is more sensitive than wt to high Zn 21levels (Fig. 2C), was decreased in zntA KO cells (Fig. 5A). This epistatic interaction between M. marinum DctpC and D. discoideum zntA KO (i.e., the higher growth restriction of DctpC in the zntA KO compared to wt host) is strong evidence for the role of Zn 21 during infection.
Intriguingly, although the concentration of Zn 21 inside the MCVs was lower in cells lacking ZntB (Fig. 4B and D), the intracellular bacterial load also decreased with time in these mutant cells (Fig. 5B), comparably to what happens inside zntA KO cells (Fig. 5A). But we also noticed that, while bacterial growth leveled off inside zntA KO cells only after 24 hpi (Fig. 5A), the growth of M. marinum in zntB KO cells was impaired from the beginning of the infection (Fig. 5B). One possible explanation is that phagosomal escape of M. marinum is altered in the zntB KO cells, exposing them either to a stronger restriction by cytosolic defenses in the case of an earlier escape (9,44) or to a prolonged restriction by vacuolar Zn 21 is the case that MCV escape is inhibited.
To assess whether the MCV integrity is compromised in D. discoideum lacking zntA or zntB, we monitored selected readouts for MCV membrane damage and M. marinum cytosolic access (36,44). In brief, membrane damage induced by M. marinum leads to bacterial and MCV ubiquitination and recruitment of the autophagy machinery (44). Consequently, the fraction of ubiquitin-positive mycobacteria is a proxy for damage. In addition, Plin, the D. discoideum homologue of mammalian perilipins, binds the hydrophobic mycobacterial cell wall when the mycobacteria are exposed to the cytosol (36). Interestingly, upon infection with M. marinum wt, the fractions of bacteria ubiquitinated or associated with Plin were comparable in all host cells (Fig. S6A to D), suggesting that membrane damage but also the detection of cytosolic bacteria is not affected in these mutants. In addition, no significant difference in ubiquitination was observed for the DctpC M. marinum mutant, neither in wt nor in the zntA and zntB KO cells ( Fig. S6A and B). However, a strong reduction of Plin-positive DctpC bacteria was observed in zntB KO cells, suggesting that DctpC might escape less efficiently to the cytosol in these cells (Fig. S6D). Nevertheless, the epistatic interaction observed between M. marinum DctpC and D. discoideum zntB KO during the assessment of intracellular growth (Fig. 5B) suggests that Zn 21 plays a major role in the restriction of these bacteria also in the absence of D. discoideum ZntB.
D. discoideum is a model to study the role of transition metals in the fight against intracellular pathogens (35,(52)(53)(54). For example, while Fe 21 participates in the defense against Mycobacterium avium or Legionella pneumophila (35,(52)(53)(54), Cu 21 or Zn 21 does not affect the infection by Legionella (35). On the other hand, to access poorly available transition metals such as iron, M. marinum stimulates the synthesis and release of small chelating molecules known as siderophores (55).
In contrast to the 24 transporters encoded in humans, D. discoideum possesses only 11 Zn 21 transporters: seven ZIP-like proteins (ZplA to -G) and four ZnTs (ZntA to -D) (4). ZntA is the main transporter in the CV and a key regulator of Zn 21 homeostasis in D. discoideum. The transporter is not found in the endocytic and phagocytic pathways in noninfected cells. When ZntA is knocked out, cells compensate for its absence by increasing pumping of Zn 21 inside the compartments of the endocytic pathway through the other transporters ZntB to ZntD (33). ZntB is the main Zn 21 importer in lysosomes and recycling endosomes (33). Interestingly, throughout the intracellular infection cycle of M. marinum, both ZntA and ZntB were located at MCVs (Fig. 3A to E), Host-Directed Zn 21 Intoxication of M. marinum ® although at different levels. While ZntA localized at detectable levels only to a minor fraction of MCVs (around 20% [Fig. 3C]), ZntB was present at the majority of them (50 to 60% [Fig. 3D]). This, together with the fact that the expression of ZntB notably increased during M. marinum infection (Fig. 3F) and that MCVs accumulate less Zn 21 in cells lacking ZntB (Fig. 4B, D, and F), led us to hypothesize that ZntB is the main Zn 21 efflux transporter during infection by M. marinum, directly pumping Zn 21 into the MCVs. Since accumulation of Zn 21 within MCVs was not completely abolished upon ZntB depletion, we propose that Zn 21 can also be transferred into MCVs by direct transport via ZntA, especially considering that the levels of the zntA mRNA increase at all stages of infection ( Fig. 2A). In addition, fusion of MCVs with zincosomes, as shown to occur to the M. smegmatis-containing compartment (33), might also contribute to elevated lumenal Zn 21 levels. As ZntC and ZntD are localized to other compartments of the endocytic and secretory pathways, they could indirectly contribute to lumenal Zn 21 levels (33). But a direct role of these transporters can likely be disregarded, because ZntC and ZntD were not detected at the MCV at any time postinfection (Fig. 3E and see also Fig. S4A and B in the supplemental material), and the levels of their mRNA were not altered, or only slightly reduced, during infection with M. marinum ( Fig. 3F and Table S1C). Previous studies have shown that the expression of ZNT1, the homologue of ZntB, increases in human macrophages after 18 and 72 h of infection with Mtb (26,30,32). This suggests that D. discoideum reacts to the invasion by pathogenic mycobacteria in a manner similar to mammalian macrophages.
The complex Zn 21 homeostasis is regulated by the balance of ZnT efflux transporters and ZIP influx transporters. In other words, higher ZIP-mediated pumping of Zn 21 to the cytosol is compensated for by detoxification, via ZnT-mediated import of Zn 21 inside various compartments, including the MCV. On the other hand, decreased ZIPmediated transport from endosomes to the cytosol might also increase lumenal Zn 21 . In keeping with these scenarios, the expression of many of the D. discoideum Zpl influx pumps is affected in complex manners during infection with M. marinum (Fig. S7 and Table S1D).
We show here that, in vitro, M. marinum senses and reacts to elevated levels of Zn 21 by specifically inducing the expression of the Zn 21 efflux pump CtpC (Fig. 2B). Accordingly, growth of M. marinum DctpC was inhibited at elevated concentrations of Zn 21 (Fig. 2C). Importantly, this growth inhibition was fully rescued by expressing the Mtb CtpC (Rv3270), confirming that CtpC (MMAR_1271) is its functional orthologue in Zn 21 efflux in M. marinum ( Fig. 2C and D). Increasing concentrations of Mn 21 did not affect growth of M. marinum wt, nor that of the DctpC mutant (Fig. S3F), excluding that M. marinum CtpC is a manganese exporter as was previously claimed for Mtb CtpC (56).
As shown for Mtb, M. marinum ctpC is also induced during infection of D. discoideum ( Fig. 2A), and deletion of ctpC strongly attenuates intracellular growth of M. marinum (Fig. 2D). This indicates that, inside amoebae, M. marinum experiences toxic concentrations of Zn 21 , which it resists by exporting Zn 21 through CtpC. Our results further demonstrate that MCV damage leads to the efflux of Zn 21 to the cytosol at early infection stages (Fig. 1), suggesting that the transient exposure to high Zn 21 levels is sufficient to stimulate ctpC expression. Therefore, it appears that M. marinum makes use of an integrated strategy to counteract the intoxication by Zn 21 in the MCV, first by increasing CtpC-mediated efflux into the MCV lumen and second by inducing ESAT-6-dependent MCV damage, resulting in leakage of ions (H 1 , Zn 21 , and others) to the cytosol. This model is in line with a previous study (57) showing that accumulation of Zn 21 in Mtb-infected fibroblasts leads both to the upregulation of CtpC and to an increased secretion of ESAT-6 that results in phagosomal damage and ion efflux.
Strikingly, the intracellular growth of M. marinum wt was similarly impaired in amoebae lacking ZntA or ZntB (Fig. 5), although the levels of Zn 21 inside MCVs were higher in the D. discoideum zntA KO but lower in the zntB KO, compared to wt (Fig. 4).
We propose the following model to account for this apparently counterintuitive result.
First, the fact that an M. marinum mutant in a Zn-related gene (ctpC) is even more attenuated for intracellular growth in a D. discoideum mutant itself affected in a Zn-related gene (zntA or zntB) is a demonstration that the two mutants interact genetically in a way that is best explainable by a dual perturbation in Zn 21 homeostasis. Second, the increased attenuation of the DctpC mutant in the zntA KO host is most likely due to the elevated MCV Zn 21 concentration, while the increased attenuation in the zntB KO is likely due to inefficient MCV escape and prolonged exposure to the residual lumenal Zn 21 concentration.
In summary, we conclude that D. discoideum restricts an infection with M. marinum by using one or more Zn 21 -dependent host defense mechanisms, which are counteracted by the pathogen via the specific induction of its Zn 21 exporter CtpC. To our knowledge, this is the first study reporting the role of CtpC in the resistance of M. marinum to Zn 21 -mediated host defense.

MATERIALS AND METHODS
D. discoideum strains, culture, and plasmids. The D. discoideum material used in this study is listed in Table S3A in the supplemental material. The wt strains Ax2(Ka) and Ax4 were grown at 22°C in HL5-C medium (Formedium) supplemented with 100 U/ml penicillin and 100 mg/ml streptomycin. Overexpressers and KO cell lines were grown in medium with hygromycin (25 mg/ml), G418 (5 mg/ ml), and/or blasticidin (5 mg/ml).
Mycobacterium strains, culture, and plasmids. The M. marinum strains and plasmids used in the present study are listed in Table S3B. M. marinum DctpC was generated by specialized phage transduction as previously described (Table S3B) (19) with modifications. The flanking regions of ctpC were amplified using the primer pairs oHK143/oHK144 and oHK145/oHK146 (Table S3B). The PCR products were digested with AlwNI and Van91I, respectively, and ligated with Van91I-digested p0004S vector fragments. The resulting plasmid pHK42 and the DNA of the temperature-sensitive phage phAE159 were ligated after previous linearization with PacI, gaining pHK50. High-titer phages of pHK50 were prepared in M. smegmatis to perform specialized transduction in M. marinum, finally yielding the DctpC::Hyg r mutant. The phAE7.1 phage was used to remove the Hyg r cassette (19). Primers used for mutant verification are listed in Table S3B (oHK159, oHK160, oHK33, and oHK36). A complementation vector was generated by the FX-cloning strategy (58) using the backbone pFLAG plasmid (59) to constitutively expresses the Mtb orthologue of ctpC (Rv3270) from the medium-strength tetracycline promoter. The vector is then stably integrated into the genomic attB site. Mtb ctpC was amplified from genomic DNA using the primer pair oHK375/oHK376 to yield pHK124. The correct insertion was verified by sequencing using the primers oHK328, oHK340, and oHK377. Mycobacteria were cultured at 150 rpm at 32°C in Middlebrook 7H9 (Difco) supplemented with 10% oleic acid-albumin-dextrose-catalase (OADC), 0.2% glycerol, and 0.05% Tween 80. Bacterial clumping was minimized by adding 5-mm glass beads during cultivation. Mutants and plasmid carriers were grown in medium supplemented with hygromycin (100 mg/ml), kanamycin (25 mg/ml), and/or ampicillin (100 mg/ml).
M. marinum growth on agar and in broth. M. marinum wt and DctpC were grown in liquid 7H9-OADC-glycerol-Tween at 32°C in shaking until reaching an optical density at 600 nm (OD 600 ) of approximately 0.5. For growth assays on agar, serial 10-fold dilutions were performed and 5 ml of each dilution was plated on 7H11-OADC-glycerol-Tween containing different concentrations (0, 0.125, 0.250, 0.500, 1.000, and 2.50 mM) of ZnSO 4 , MnCl 2 , and CuSO 4 . Pictures were taken after 7 days of incubation at 32°C. For growth in broth, overnight cultures of green fluorescent protein (GFP)-expressing M. marinum wt and DctpC as well as of E2-Crimson (pTEC19, Addgene)-expressing wt and DctpC::ctpC were diluted to 5 Â 10 5 bacteria/ml, centrifuged, and resuspended in 7H9 medium containing increasing ZnSO 4 concentrations as indicated. Bacteria were distributed in 96-well plates at 10 5 /well (Perkin Elmer). Growth at 32°C under shaking conditions was assessed by measuring fluorescence at 509 nm for the GFP signal and 646 nm for the E2-Crimson signal in a plate reader (Synergy MX; BioTek). The growth of the DctpC (GFP) and DctpC::ctpC (E2-Crimson) strains was normalized to the M. marinum wt strains expressing GFP and E2-Crimson, respectively.
Infection of D. discoideum. Infections were performed as previously described (21,60). After spinoculation and washing off the extracellular bacteria, the infected cells were resuspended at a density of 1 Â 10 6 cells/ml in filtered HL5-C. Five micrograms per milliliter of streptomycin and 5 U/ml of penicillin were added to prevent extracellular bacterial growth. The infected cells were then incubated at 25°C at 130 rpm, and samples were taken for analysis at the indicated time points.
Live cell imaging. To monitor the subcellular localization of Zn 21 during infection, infected cells were plated on m-dishes (ibidi), medium was exchanged to Soerensen buffer, and intracellular Zn 21 was stained with 5 mM NBD-TPEA (Sigma catalog no. N1040 [61]) for 30 min in the dark (33). To stain unlabeled bacteria, Vybrant DyeCycle Ruby stain (ThermoFisher) was used as previously described (3). Escape of M. marinum from MCVs was assessed in cells expressing mCherry-Plin. Images of live cells were taken with an inverted 3i Marianas spinning disc confocal microscope using the 63Â glycerol or 100Â oil objectives. Fluorescent proteins or probes were excited using the 488-nm (GFP and NBD-TPEA), 561-nm (mCherry), and 640-nm (Vybrant Ruby) laser lines, respectively. To quantify the integrated intensity of NBD-TPEA inside MCVs, cells were infected with mCherry-expressing M. marinum. At the indicated Host-Directed Zn 21 Intoxication of M. marinum ® time points, samples were stained with NBD-TPEA and images were taken using the ImageXpress spinning disc confocal microscope (Molecular Devices). Here, mCherry-expressing bacteria were excited using the Texas Red laser line and NBD-TPEA-expressing bacteria were excited using the GFP laser line. The integrated intensity per MCV area was assessed using an analysis pipeline created in MetaXpress Custom Module Editor (Molecular Devices). Briefly, MCVs were defined as a sphere around mCherry-labeled bacteria that were detected using the Texas Red channel. NBD-TPEA inside MCVs was detected using the settings for the GFP channel. The normalized integrated intensity of NBD-TPEA was divided by the area of each MCV, leading to the integrated intensity per MCV area.
RNA-sequencing data. RNA-Seq data were gathered from the work of Hanna et al. (38). Briefly, infection was carried out using M. marinum expressing GFP as described above. In brief, at each time point, a homogeneous population of infected cells was obtained by GFP fluorescence-activated cell sorting (FACS). The total RNA was extracted, and the rRNA from host and bacteria was depleted, before generation of the cDNA libraries using a Bio-Rad iScript kit and sequencing. The resulting reads were mapped in parallel against the D. discoideum and M. marinum genome. Differential expression analysis of the infection time course was performed using the R package limma, by comparing on one hand infected D. discoideum cells to mock (i.e., buffer)-infected cells, and on the other hand intracellular M. marinum to M. marinum grown in broth.
qRT-PCR sample collection and analysis. To assess the effect of Zn 21 on the expression of ctpC, ctpC-like, ctpV, and ctpV-like, M. marinum strains were exposed to various concentrations of ZnSO 4 (0.00, 0.05, 0.10, and 0.50 mM) for 2 h. Bacteria were harvested, RNA was extracted, and cDNA was synthesized using a Bio-Rad iScript kit. For each gene tested, the mean calculated threshold cycles (C T ) were averaged and normalized to the C T of a gene with constant expression (sigA). The normalized C T was used for calculating the fold change using the DDC T method. Briefly, relative levels of target mRNA, normalized with respect to an endogenous control (sigA), were expressed as 22DDC T (fold), where DC T = C T of the target gene 2 C T of the control gene (sigA), and DDC T = DC T of the studied set of conditions 2 DC T of the calibrator conditions, as previously described (63).
Measurement of intracellular bacterial growth. Intracellular growth of M. marinum expressing fluorescent reporters or bacterial luciferase was measured as previously described (64). Briefly, three different dilutions of infected cells (between 0.5 Â 10 5 and 2.0 Â 10 5 cells) were plated on nontreated, white F96 MicroWell plates (Nunc) or black 96-well plates (Perkin Elmer) covered with a gas-permeable moisture barrier seal (BioConcept). The course of infection was monitored by measuring either luminescence or fluorescence (at 509 and 646 nm) as a proxy of bacterial growth using a Synergy Mx microplate reader (BioTek) at a constant temperature of 25°C for around 70 h with 1-h intervals. The growth of the DctpC (GFP) and DctpC::ctpC (E2-Crimson) strains was normalized to M. marinum wt strains expressing GFP and E2-Crimson, respectively.
Phagocytic plaque assay. Phagocytic plaque assays in the presence of nonpathogenic bacteria as food source were performed as described previously (43). Briefly, 50 ml of an overnight culture of various nonpathogenic bacteria was added to the wells of a 24-well plate containing 2 ml SM-agar. Then, 10, 100, 1,000, or 10,000 D. discoideum cells were added to the bacterial lawn and plates were incubated at 22°C for 4 to 7 days until plaques were visible. Quantification was performed by scoring the appearance of plaques in at least three independent experiments. The logarithmic plaquing score was defined as follows: plaque formation in wells with 10 amoebae yielded a score of 1,000; in the cases where cells did not grow at lower dilutions, they obtained the corresponding lower scores of 100, 10, and 1. The ability of D. discoideum Ax2(Ka), zntA KO, Ax4, and zntB KO to form plaques on M. marinum wt and DctpC was assessed as described previously (43,65). Briefly, 5 Â 10 8 mycobacteria were harvested and resuspended in 1.2 ml of 7H9-OADC-glycerol-Tween containing a 1:10 5 dilution of an overnight culture of Klebsiella pneumoniae. Fifty microliters of the suspension was added to the wells of a 24-well plate containing 2 ml of 7H11-OADC-glycerol-Tween. Various dilutions of D. discoideum (i.e., 10, 100, 1,000, and 10,000 cells; for strains in the Ax4 background, 3-fold more cells were used to achieve similar temporal evolution of the plaques) were added to the bacterial lawn, and plates were incubated at 25°C for 7 days until plaques were visible. The scoring was performed as described above.
Phylogenetic tree. Putative P-type ATPases of M. marinum were identified by performing a protein BLAST search using NCBI PSI-BLAST and the sequence of Mtb CtpC. The results were then cross-compared to the annotations of M. marinum P-type ATPases available on Mycobrowser (66). The protein sequences of Mtb and M. marinum P-type ATPases were aligned with the online version of MAFFT (https://mafft.cbrc.jp/alignment/server/) using the G-INS-1 strategy and "leave gappy regions" to generate a phylogenetic tree in phylo.ilo using NJ conserved sites and the JTT substitution model (67).
Data availability. Information on the differential expression of M. marinum P-type ATPases (Table S1A) as well as D. discoideum CV genes (Table S1B), zntA to -D transporters (Table S1C), and zplA to -G transporters (Table S1D) was extracted from the preprint resource at https://doi.org/10.1101/ 590810 (38). Information on the enrichment of CV proteins at the MCV (Table S2) was extracted from the preprint resource at https://doi.org/10.1101/592717 (42).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.