MntP and YiiP Contribute to Manganese Efflux in Salmonella enterica Serovar Typhimurium under Conditions of Manganese Overload and Nitrosative Stress

ABSTRACT The divalent transition metal cation manganese is important for protein function, particularly under conditions of iron limitation, nitrosative stress, and oxidative stress, but can mediate substantial toxicity in excess. Salmonella enterica serovar Typhimurium possesses multiple manganese importers, but the pathways for manganese efflux remain poorly defined. The S. Typhimurium ATCC 14028s genome was analyzed for putative manganese export pathways, which identified a previously uncharacterized homologue of the Escherichia coli manganese exporter mntP, stm1834, and two cation diffusion facilitator family transporters, zitB (stm0758) and yiiP (stm4061). Manganese acquisition by S. Typhimurium has been shown to occur in response to nitric oxide, an important chemical mediator of the mammalian innate immune response. However, cellular manganese can rapidly return to prechallenge levels, strongly suggesting that one or more S. Typhimurium exporters may contribute to this process. Here, we report that mntP and yiiP contribute to manganese resistance and export in S. Typhimurium. YiiP, also known as FieF, has previously been associated with zinc and iron transport, although its physiological role remains ambiguous due to a lack of zinc-sensitive phenotypes in yiiP mutant strains of S. Typhimurium and E. coli. We report that S. Typhimurium ΔmntP ΔyiiP mutants are exquisitely sensitive to manganese and show that both YiiP and MntP contribute to manganese efflux following nitric oxide exposure. IMPORTANCE Transition metal cations are required for the function of many proteins but can mediate toxicity when present in excess. Identifying transporters that facilitate metal ion export, the conditions under which they are expressed, and the role they play in bacterial physiology is an evolving area of interest for environmental and pathogenic organisms. Determining the native targets of metal transporters has proved challenging since bioinformatic predictions, in vitro transport data, and mutant phenotypes do not always agree. This work identifies two transporters that mediate manganese efflux from the Gram-negative pathogen Salmonella enterica serovar Typhimurium in response to manganese overload and nitric oxide stress. While homologues of MntP have been characterized previously, this is the first observation of YiiP contributing to manganese export.

manganese inhibits heme biosynthesis while chronic manganese stress ultimately leads to iron depletion and impaired formation of Fe-S cluster proteins (15,16). Together, these effects can result in inhibition of energy-generating and biosynthetic pathways. In Bacillus subtilis, which requires manganese for growth, excess manganese has been associated with impaired function of the cytochrome aa 3 heme-copper menaquinol oxidase (QoxABCD) of the electron transport chain (17). Therefore, while increased cellular manganese may benefit bacteria when challenged with a specific physiological or chemical stress, efflux of manganese may be necessary upon alleviation of the aforementioned stress.
Three types of manganese efflux systems have been identified in prokaryotes to date. Transporters from the cation diffusion facilitator (CDF) family are widely prevalent among prokaryotic species with family members implicated in transport of zinc, cadmium, cobalt, nickel and manganese, depending on the sequence motifs present in metal binding regions (18). The prokaryotic CDF manganese exporter MntE was first identified in Streptococcus pneumoniae and has since been studied in Staphylococcus aureus, Enterococcus faecalis, and Streptococcus pyogenes (19)(20)(21)(22). B. subtilis relies on two CDF family transporters for manganese efflux with MneP functioning as the primary exporter and MneS playing a secondary role (23). P 1B -ATPases have been shown to export a range of transition metal ions, with manganese export first established for CtpC from Mycobacterium tuberculosis (24). MntP, which lacks homology to other established classes of manganese exporters, was first characterized in Xanthomonas oryzae, E. coli, and Neisseria meningitidis (as MntX) (25)(26)(27). This architecturally distinct transporter has since been shown to have orthologs in additional species.
Previously, we showed that manganese acquisition by MntH, SitABCD, and ZupT is important for S. Typhimurium nitrosative stress resistance. Furthermore, total cellular manganese was restored to prechallenge levels following the resolution of the stress (14). In this study, we sought to identify and characterize S. Typhimurium efflux systems that contribute to manganese homeostasis in response to stress. We show that orthologs of the E. coli transporters MntP and YiiP protect S. Typhimurium against manganese intoxication and mediate manganese efflux during the late-stage response to nitrosative stress.

RESULTS
STM1834 (MntP) protects S. Typhimurium against excess manganese. E. coli and S. Typhimurium share similar genetic sequences at many loci, so the S. Typhimurium genome was searched for proteins with homology to MntP from E. coli. One match, at locus stm1834 (Fig. 1A), was identified with 91% identity and 96% similarity (over 188 amino acids) to E. coli MntP. To determine the function of the putative S. Typhimurium mntP ortholog, a deletion mutant was generated. The mutant was then grown in the presence of 0.5 mM MnSO 4 and the phenotype compared to the wild-type parental strain. The S. Typhimurium DmntP strain was delayed for growth in excess manganese compared to the wild-type (Fig. 1B). Constitutive expression of S. Typhimurium mntP from a lowcopy-number plasmid (pMntP) complemented the growth defect of a DmntP mutant (Fig. 1C). Consistent with the growth phenotype data, spot plate assays on manganese supplemented media showed that the S. Typhimurium DmntP strain had decreased growth on 0.5 mM MnSO 4 and decreased survival on 1 mM MnSO 4 , while expression of mntP complemented these phenotypes (Fig. 1D). Taken together, these data are consistent with previous studies of MntP function in E. coli and suggest that stm1834 encodes a manganese exporter homologous to E. coli MntP (26).
MntP is not solely responsible for manganese efflux following NOÁ treatment. To determine if manganese export by MntP is responsible for returning cellular manganese levels to pretreatment levels following NOÁ exposure, we compared the phenotypes of wildtype and DmntP S. Typhimurium by inductively coupled plasma-mass spectrometry (ICP-MS). Here, we used conditions defined in our prior studies of S. Typhimurium wherein the application of NOÁ stress alters metal homeostasis and induces manganese accumulation (14,28). Accordingly, cultures were treated with 2 mM diethylamine NONOate (DEANO), a fast-release NOÁ donor, and the cellular manganese content was monitored over the course of 60 min. Consistent with prior observations, manganese levels increased by 30 min posttreatment and then returned to pretreatment levels by 60 min (Fig. 2) (14). Notably, manganese levels in the DmntP strain were not significantly different than in the wild-type at any time. These data indicate either that manganese efflux does not occur via MntP following NOÁ treatment or that, in the absence of mntP, manganese efflux occurs via another transporter in S. Typhimurium.
YiiP expression enhances zinc toxicity in S. Typhimurium. Since the S. Typhimurium genome contained only one MntP homologue, the genome was next searched for proteins with homology to the CDF family manganese transporter MntE. Two S. Typhimurium (B) Growth phenotypes of S. Typhimurium wild-type (WT) and DmntP strains in LB with or without 0.5 mM MnSO 4 supplementation. The DmntP strain was delayed exiting lag phase relative to WT in the presence of 0.5 mM MnSO 4 (P , 0.001). (C) Growth phenotypes of S. Typhimurium empty vector strains WT pJK724 and DmntP pJK724 compared to plasmid-based complementation strain DmntP pMntP. In 0.5 mM MnSO 4 the DmntP pJK724 strain was delayed exiting lag phase compared to both WT pJK724 and DmntP pMntP (P , 0.001). Data for (B) and (C) are the mean of 4 independent experiments. Statistical significance of differences between strains was determined by the time (hr) to reach 50% maximum growth (OD 600 ; dashed line) by unpaired two-tailed t test. (D) Growth of strains from (B) and (C) assessed using spot assays. Dilutions of OD 600 = 0.3 cultures were spotted onto LB agar with and without MnSO 4 supplementation. A representative spot assay for each condition is shown, selected from 3 independent biological replicates.
E. coli YiiP, also known as FieF, was first reported to serve as an iron efflux transporter. Subsequent in vitro studies showed that YiiP also had the ability to interact with zinc ions, although its physiological contribution to zinc homeostasis remains to be defined (30)(31)(32)(33). Three metal binding regions, the A-site, B-site, and C-site (comprised of C1 and C2 plus a linker) have been identified based on the E. coli YiiP crystal structure (31,32,34). The A-site has been established as the primary motif determining metal selectivity (34,35). S. pneumoniae MntE has an ND-DD A-site motif, but DD-DD A-site motifs are also common in manganese exporting CDF proteins (36). By contrast, the zinc exporting CDF from S. pneumoniae, CzcD, has an HD-HD motif, which is the most common motif in CDF family zinc transporters (36). S. Typhimurium YiiP has a DD-HD A-site motif, which precludes bioinformatic prediction of the native ligand but is suggestive of the potential to export ions other than zinc.
Zinc sensitivity phenotypes have not been observed for S. Typhimurium or E. coli strains with yiiP deleted alone or in combination with other known zinc exporters (28,29,37). However, these studies could have failed to detect zinc sensitivity phenotypes if yiiP was not expressed under the conditions tested. To address this, we expressed yiiP constitutively from a low-copy-number plasmid (pYiiP) in the S. Typhimurium DzntA DzitB background to ascertain whether this could decrease the zinc sensitivity of this mutant strain. We observed that the DzntA DzitB strain was delayed for growth in the presence of 0.125 mM ZnSO 4 compared to the wild-type. Expression of yiiP in the DzntA DzitB genetic background abrogated bacterial growth (Fig. 3B). Notably, expression of mntP elicited a similar phenotype in the DzntA DzitB background (Fig. 3C). Spot assays revealed that expression of either YiiP or MntP in the DzntA DzitB genetic background led to decreased survival in the presence of 0.0625 mM ZnSO 4 and little to no survival on plates with 0.125 mM ZnSO 4 (Fig. 3D). These data indicate that YiiP does not facilitate zinc export. It therefore follows that the increased sensitivity of the yiiP-expressing DzntA DzitB strain suggests that YiiP may export a different metal ion that results in enhanced susceptibility to zinc intoxication.
Metal availability results in altered expression of mntP but not yiiP. We next measured yiiP expression under conditions of metal limitation, metal stress, and NOÁ challenge compared to mntP. In the presence of general divalent cation chelator ethylenediaminetetraacetic acid (EDTA), expression of neither mntP nor yiiP was significantly altered (Fig. 4A). In response to metal stress, mntP was upregulated when the medium was supplemented with 0.5 mM iron or manganese, but not zinc (Fig. 4B). By contrast, yiiP expression did not change in response to iron, manganese, or zinc (Fig. 4B), which also differs from observations of the orthologous gene from E. coli (29,30). Expression of neither mntP nor yiiP was significantly altered at any time following treatment with 2 mM DEANO (Fig. 4C). YiiP contributes to manganese resistance in S. Typhimurium. Although expression of yiiP did not change in response to metal stress under our experimental conditions, the protein may be present due to constitutive expression and contribute to metal ion efflux. Building on the observations that YiiP shares homology with MntE ( Fig. 3A) and its expression enhanced the zinc sensitivity of a zinc efflux mutant strain in a similar fashion to expression of MntP ( Fig. 3B to D), a role in manganese efflux was investigated. To assess whether YiiP contributes to manganese resistance of S. Typhimurium, DyiiP and DmntP DyiiP strains were generated. The DyiiP strain grew similarly to the wild-type in the presence of 0.5 mM MnSO 4 (Fig. 5A). However, supplementation of the growth medium with 0.5 mM MnSO 4 delayed the growth of the DmntP strain and abrogated growth of the DmntP DyiiP strain (Fig. 5A). In spot assays, the DmntP DyiiP strain displayed decreased growth at 0.25 mM MnSO 4 and decreased survival at both 0.5 mM and 1 mM MnSO 4 . By contrast, the DmntP strain only showed a moderate decrease in survival at 1 mM MnSO 4 (Fig. 5C). Taken together, these data show an enhancement of manganese sensitivity when both transporters are absent. Plasmid-based complementation with either mntP or yiiP expressed from the native promoter attenuated the growth defect of a DmntP DyiiP mutant, although the DmntP DyiiP p n YiiP strain still had a minor growth delay relative to the wild-type and DmntP DyiiP p n MntP strains (Fig. 5B). Similar complementation results were obtained in spot assays (Fig. 5D).
S. Typhimurium mutants lacking mntP and yiiP are neither sensitive to zinc nor disrupted for zinc homeostasis. Previous experiments investigating the metal binding properties of recombinant E. coli YiiP revealed a capacity to interact with zinc ions in vitro, but not manganese (33,38). Thus, the observed impact of manganese on the growth and survival of the S. Typhimurium DmntP DyiiP strain may reflect an indirect effect on zinc homeostasis. Interplay between zinc and manganese homeostasis has been shown to occur in several Gram-positive pathogens, such as S. pneumoniae where zinc has been established to disrupt manganese uptake and increase sensitivity to oxidative stress (39)(40)(41)(42). However, this phenomenon has not been reported for the Enterobacteriaceae and, in S. Typhimurium, this may be attributable to the presence of the manganese-transporting natural resistance-associated macrophage protein (NRAMP) transporter MntH, which is not susceptible to zinc inhibition (9). By contrast, the impact of manganese on zinc homeostasis in S. Typhimurium has not been determined. Accordingly, we investigated whether the observed manganese sensitivity of the S. Typhimurium DmntP DyiiP strain was due to pleiotropic effects of perturbed zinc homeostasis.
Here, we examined the impact of zinc stress on the wild-type DyiiP, DmntP, and DmntP DyiiP strains. This revealed that zinc supplementation had no impact on the growth phenotype of any strain ( Fig. 6A and B). These data indicate that S. Typhimurium zinc homeostasis is not dysregulated in the absence of manganese efflux. Furthermore, neither YiiP nor MntP contributes to substantially to zinc homeostasis. To further probe the impact on the zinc regulatory network of S. Typhimurium, the expression of genes controlled by the zinc uptake regulator, Zur, and zinc export regulator, ZntR, were analyzed in the wild-type and DmntP DyiiP strains during exposure to excess manganese. The sensitivity of these metalloregulators is in the femtomolar to nanomolar range; thus, they provide a highly sensitive insight into cellular zinc homeostasis (43)(44)(45). Here, we monitored the expression of the Zur-regulated zinc uptake transporter znuABC, the primary S. Typhimurium pathway for zinc acquisition (46)(47)(48); the ZntR-regulated zinc exporter zntA, the major S. Typhimurium zinc efflux system (28,49,50); and zupT, which has been implicated in zinc and manganese import, although its regulatory control remains to be defined (14,51,52). These data show that there was no significant difference in expression of zntA, znuC, or zupT between the wildtype and DmntP DyiiP strains at 20 min (Fig. 6C) or 60 min (Fig. 6D) in the presence of YiiP contributes to manganese efflux in S. Typhimurium. We next investigated whether YiiP participates in control of cellular manganese levels following exposure to NOÁ. It is important to note that in this experiment, direct comparisons are confined to the wild-type and the mutant strains within each analysis. This is due to differences in medium manganese concentrations that differed between the analyses (i.e., Fig. 7A vs. Fig. 7B), which influenced the absolute cellular abundances in the bacterial strains (43). Changes in cellular manganese followed generally similar patterns in the wild-type and DyiiP strains following treatment with 2 mM DEANO. However, manganese levels were elevated in the DyiiP strain relative to the wild-type strain at 5, 15, and 60 min (Fig. 7A). Manganese levels were greater in the DmntP DyiiP strain than in the wild-type at 5, 15, 45, and 60 min after treatment with 2 mM DEANO, but both strains reached a similar peak at 30 min (Fig. 7B). In the wild-type strain, cellular manganese decreased at 45 min compared to the 30 min peak and fell to pretreatment levels by 60 min. In the DmntP DyiiP strain, there was no significant difference in manganese levels at 30, 45, and 60 min posttreatment, suggesting that efflux was impaired when both genes were deleted (Fig. 7B).
Residue Asp45 is required for YiiP to alleviate manganese toxicity. Prior work has indicated that the A-site motif of CDF transporters dictates metal specificity (34,35). For YiiP, the aspartate residue at position 45 has been implicated in facilitating transport of both cadmium and zinc in in vitro assays (35). When the aspartate residue was mutated to histidine (D45H), creating an HD-HD A-site motif in YiiP, the mutant protein transported zinc at similar rates as wild-type YiiP in in vitro assays, but no longer transported cadmium (35). In this study, our data indicate that S. Typhimurium YiiP is associated with physiological manganese export and do not support a role in zinc homeostasis; however, this does not preclude a capacity for interaction with zinc in vitro. Here, we sought to determine the contribution of Asp45 to YiiP manganese transport. We generated a point mutation, substituting histidine for aspartate (D45H) in the p n YiiP plasmid (p n YiiP D45H). The mutant derivative was then investigated in a DmntP DyiiP background grown in medium supplemented with 0.5 mM MnSO 4 . We observed that the growth phenotype of the DmntP DyiiP p n YiiP D45H strain was no different than the DmntP DyiiP pRB3 strain (empty vector) in the presence of manganese, indicating a lack of complementation (Fig. 8A). To confirm that lack of complementation was due to lack of transporter function rather than protein instability, FLAG-tagged YiiP and YiiP D45H were expressed constitutively in the DmntP DyiiP background, subjected to SDS-PAGE and visualized by Western blot. Single bands of similar intensity were observed for both proteins at ;26 kDa (Fig. 8B). Although this is smaller than the predicted mass for YiiP (32.9 kDa), integral membrane proteins are known to show altered mass profiles in SDS-PAGE analyses. To confirm that the FLAG-tagged YiiP proteins were functional, spot assays were performed. Expression of YiiP:FLAG complemented the growth and survival defects of the DmntP DyiiP strain while expression of pYiiP D45H:FLAG did not (Fig. 8C). Collectively, these results show that Asp45 is required for manganese transport activity by S. Typhimurium YiiP.

DISCUSSION
Manganese efflux by MntP and the CDF transporter MntE has been studied in a variety of prokaryotic species, but thus far the presence of multiple manganese efflux systems within a single prokaryote has only been characterized in B. subtilis. S. Typhimurium encodes both an MntP homologue and YiiP, a CDF transporter. Identifying the target metal of CDF family transporters can be challenging despite efforts to establish the motifs that provide ion selectivity. YiiP has homology within the metal coordinating A-site to both zinc and manganese-transporting members of the CDF family (Fig. 3A) (36). YiiP from E. coli has been characterized as a zinc transporter using in vitro and structural methods, while an in vivo study linked its function to iron homeostasis. A physiological role for YiiP in zinc homeostasis has not yet been established in S. Typhimurium or E. coli, with DyiiP mutant strains not demonstrating zinc susceptible phenotypes (28,29,37). Here we show that expression of YiiP enhances the zinc sensitivity of DzntA DzitB S. Typhimurium rather than reducing it (Fig. 3B). These data, along with those showing enhanced manganese sensitivity and lack of manganese export in DmntP DyiiP S. Typhimurium (Fig. 5, 7B), suggest that zinc is not the cognate cargo of S. Typhimurium YiiP in vivo. Manganese transport phenotypes have been demonstrated in vivo for YiiP homologues from Sinorhizobium meliloti, Deinococcus radiodurans, and Rhizobium etli (53-55). These proteins share a clade with YiiP from E. coli that is phylogenetically distinct from the clades containing the well-characterized CDF zinc transporters CzcD and ZitB (54,56). The variety of proposed substrates for transporters within this clade is diverse while relatively few transporters have been experimentally characterized. Further work will be required to gain an accurate understanding of the biological functions of this group of transporters.
While function of MntP alone was sufficient to provide protection from manganese toxicity under the conditions of this study, deletion of both mntP and yiiP enhanced the sensitivity of S. Typhimurium to manganese intoxication ( Fig. 5A and C). Function of either MntP or YiiP could restore manganese homeostasis following NOÁ treatment, and only the DmntP DyiiP strain was abrogated for manganese efflux during the latestage response to NOÁ exposure (Fig. 7B). Selectivity for manganese depends on residue Asp45 within the A-site metal binding region of YiiP. A change in transport activity in response to mutation of Asp45 is consistent with similar results arising from the mutation of A-sites in E. coli YiiP and other CDF transporters (34,35). While we cannot exclude the possibility that YiiP might facilitate the transport of other transition metal cations under certain conditions, the results of this study support a role for YiiP as a manganese exporter in S. Typhimurium. Regulation of yiiP expression remains an open question. Unlike mntP, it is not part of the MntR regulon and has not been shown to be part of the Fur, Zur, or ZntR regulons (26,57). Expression of mntP is regulated by both MntR and the ybbP-ykoY manganese-sensing riboswitch in E. coli (58). The promoter region of mntP from S. Typhimurium shows homology to that of E. coli, suggesting that similar regulatory mechanisms may apply to mntP expression in both organisms. By contrast, the promoter region of yiiP lacks this homology. Induction of mntP expression occurred in response to manganese and iron stress, whereas yiiP expression was unaffected by the conditions tested (Fig. 4B). Previous work using b-galactosidase fusions to the yiiP promoter in E. coli showed upregulation in response to zinc and a modest response to iron after several hours of incubation (29,30). Since none of the known metal-responsive regulators have been implicated in yiiP expression, any expression change in response to metal excess may be indirect and more time may be necessary for the cellular conditions driving regulation to develop. Alternatively, expression of yiiP may be constitutive in S. Typhimurium under standard laboratory growth conditions. Constitutive expression would be consistent with the more rapid accumulation of manganese observed in strains lacking yiiP (Fig. 7), while a strain lacking only the inducible mntP acquires manganese similarly to the wild-type (Fig. 2). Expression of mntE, the CDF manganese transporter from S. pneumoniae, does not respond to manganese abundance and has been proposed to be expressed constitutively (19). Manganese efflux activity in E. coli is further regulated by the small protein MntS. E. coli that overexpress MntS display enhanced manganese accumulation and have the same manganese-sensitive phenotype as DmntP mutants, suggesting that this small protein may function to prevent the efflux activity of MntP (16). Overexpression of MntS in a DmntP strain did not enhance the manganese sensitivity of this mutant, suggesting that it does not inhibit the function of additional manganese exporters such as YiiP, but this possibility has not yet been investigated directly.
While efflux by either MntP or YiiP is sufficient to restore baseline manganese levels in S. Typhimurium following NOÁ exposure, and both contribute to protecting S. Typhimurium from excess manganese in culture, the broader functional significance of these transporters in S. Typhimurium biology and pathogenesis has yet to be determined. Manganese acquisition is required for S. Typhimurium infection and virulence, but what role, if any, efflux might play is not currently known (10,11). Manganese efflux has been shown to play an important role in the virulence of other pathogens such as S. pneumoniae and S. aureus and in colonization by E. faecalis (19)(20)(21). Although manganese has primarily been understood to act as an antioxidant, and absence of mntE rendered S. pneumoniae more resistant to both NOÁ and oxidative stress in the form of methyl viologen, this has not been the case for all organisms and oxidants (19). S. aureus lacking mntE was more sensitive to sodium hypochlorite (NaOCl), while S. pyogenes was more sensitive to peroxide (H 2 O 2 ) (20,22). Manganese excess has been shown to affect electron transport chain function and function of Fe-S cluster enzymes in central energy-generating pathways. (15)(16)(17). These proteins are also targets of oxidants, which could explain, at least in part, why enhanced sensitivity to oxidants is seen for some species, but the mechanisms underlying the requirement for manganese efflux remain incompletely understood. In light of these data, the dynamic flux of multiple metals that occurs in response to NOÁ, and the varied environments and stresses encountered by pathogens within the host, characterizing the biological roles of microbial manganese efflux systems, is an open and potentially complex area of investigation. Based on the findings of this study, we suggest that any future work concerning the biological significance of manganese efflux in S. Typhimurium should address both the function of MntP and the newly established manganese export function of YiiP.

MATERIALS AND METHODS
Bioinformatic analysis. S. Typhimurium homology matches to MntP from Escherichia coli MG1655 and MntE and CzcD from Streptococcus pneumoniae TIGR4 were determined using BLASTP from the Kyoto Encyclopedia of Genes and Genomes (KEGG) website with default settings. Alignments were generated using the KEGG ClustalW tool (59). Alignment graphics were generated using BoxShade hosted by ExPASy, the Swiss Institute of Bioinformatics Resource Portal (https://embnet.vital-it.ch/software/BOX_form.html).
Growth conditions. Salmonella enterica serovar Typhimurium and Escherichia coli were grown in Luria-Bertani medium (LB; Fisher) at 37°C with shaking at 250 rpm unless otherwise specified. Antibiotic selection was used for strain construction only at the following concentrations: 100 mg mL 21 ampicillin (Amp), 50 mg mL 21 kanamycin (Kan), and 20 mg mL 21 chloramphenicol (Cm).
Strain and plasmid construction. All strains and plasmids are listed in Table 1. All primers are listed in Table 2. S. Typhimurium strains were generated in the ATCC 14028s genetic background, which also served as the wild-type strain in this study (JK237). Deletion mutants were constructed using l-RED mediated recombination with either pKD3 or pKD4 as the template (60). Expression of Flp-FRT recombinase from pCP20 was used to remove the antibiotic resistance cassettes from strains EF657 and EF635 to generate strains EF725 and EF755 (60). Combination mutants were created using P22HT105/int bacteriophage transduction (61). All mutants were verified by PCR.
E. coli strain TB1 was used as the cloning host strain. Purified genomic DNA from S. Typhimurium 14028s was used as the PCR template unless otherwise specified. Plasmids pEF101 and pEF102 were generated by amplifying the upstream promoter region and coding sequence of mntP using primers EFP334 and EFP335 and yiiP using primers EFP336 and EFP340. Amplified fragments were digested with KpnI and HindIII then ligated into pRB3-273C digested with the same enzymes in reverse orientation to the multiple cloning site promoter (62). Constructs were confirmed by sequencing with JKP227 and JKP244 then transformed into strain EF657 to generate strains EF744 and EF745. To generate plasmids pEF103 and pEF106, the coding sequence of yiiP was amplified using primers EFP319 and EFP320 and the coding sequence of mntP was amplified using primers EFP346 and EFP335. Amplified fragments were digested with NcoI and HindIII then ligated into pJK724 digested with the same enzymes (63). Constructs were sequenced using primers EFP21 and JKP777 then transformed into zinc efflux mutant strain EF528 to generate strains EF746 and EF762. To generate plasmid pEF108, primers EFP336 and EFP347 were used to amplify a short fragment containing the target mutation (D45H). The fragment was purified from an agarose gel then used in a second amplification reaction with EFP340. Full-length product was digested with KpnI and HindIII then ligated into pRB3-273C digested with the same enzymes in reverse orientation to the multiple cloning site promoter. The construct was verified by sequencing with JKP227 and JKP244 then transformed into strain EF657 to generate strain EF773. Plasmids pEF115 and pEF116 were generated by amplifying the coding sequences of plasmids pEF102 and pEF108 using primers EFP319 and EFP397. Amplified fragments were digested with NcoI and HindIII then ligated into pJK724 digested with the same enzymes. Constructs were sequenced using primers EFP21 and JKP777 then transformed into strain EF657 to generate strains EF829 and EF830. Metal sensitivity growth curve assays. To determine manganese sensitivity, wild-type S. Typhimurium (JK237) and transporter mutants (EF561, EF635, EF657) were grown overnight in LB, normalized to optical density at 600 nm (OD 600 ) of 1, and diluted in triplicate 1:1000 into LB or LB with 0.5 mM MnSO 4 for a final volume of 300 mL in a flat-bottom 96-well nontreated tissue culture microtiter plate (Midwest Scientific). Cultures were grown aerobically with shaking at 567 rpm (3 mm) in a Biotek Synergy HTX multimode 96-well plate reader at 37°C. Growth was monitored by recording OD 600 every 15 min for 40 h. Statistical significance was determined by comparing the time required to reach 50% maximum OD 600 by unpaired two-tailed t test using Microsoft Excel. Complementation of DmntP was assessed using strains JK895, EF737, and EF738. Complementation of DmntP DyiiP was assessed using JK895, EF739, EF740, and EF741. Complementation of DmntP DyiiP with the YiiP D45H mutant construct pEF108 was assessed using strains JK895, EF739, EF741, and EF773. Growth assays were carried out as described above except MnSO 4 was added at both 0.5 mM and 0.25 mM final concentration.
Strains used to determine zinc sensitivity in response to YiiP and MntP expression were EF564, EF745, EF746, and EF762. Zinc sensitivity of manganese transporter mutants was assessed using strains JK237, EF561, EF635, and EF657. Zinc sensitivity assays were carried out as described above but the LB was supplemented with 0.125 mM or 0.5 mM ZnSO 4 respectively.
Metal sensitivity spot assays. Strains were grown overnight in LB, diluted 1:1000 in 5 mL fresh medium then grown with shaking at 37°C for 3 h to OD 600 = 0.3. Cultures were serially diluted 10-fold in PBS, then 3 mL were spotted onto LB agar plates with or without metal sulfate supplementation. Plates were grown 14-16 h at 37°C prior to imaging. Spot assays utilized the same strains as growth curve assays. Complementation of manganese sensitivity by expression of FLAG-tagged YiiP proteins was assessed using strains EF564, EF747, EF829, and EF830.
Metal content analyses. Inductively coupled plasma-mass spectrometry (ICP-MS) analyses were conducted using strain variants that were also flagellar mutants (EF610, EF697, EF698, EF701) to increase pelleting efficiency as in previous studies (14,28). Overnight cultures were diluted 1:1000 into 100 mL fresh LB medium and then grown to OD 600 ; 1. Cultures were divided into 5 mL aliquots in 18 Â 150 mm glass tubes, treated with 2 mM diethylamine NONOate (DEANO) and returned to shaking at 37°C. At 0, 5, 15, 30, 45, and 60 min posttreatment, 4.5 ml of culture was pelleted by centrifugation then washed twice with ultrapure water. Pellets were resuspended in analytical grade nitric acid, boiled, then diluted 1:10 with ultrapure water for analysis on an Agilent 8900x QQQ ICP-MS. Bacterial numbers, defined as CFU, were enumerated at each time point for calculation of relative metal concentrations. Statistical significance was determined by two-tailed t test in Microsoft Excel.
Gene expression analysis. Primer sequences used for expression analysis are listed in Table 2. For expression under metal chelation, wild-type S. Typhimurium was grown overnight, diluted 1:1000 in 2 Â 5 mL fresh LB medium, and grown to OD 600 ; 1. Ethylenediaminetetraacetic acid (EDTA) at a concentration of 3 mM was added to one of the cultures for 20 min. 1.5 ml of each culture was pelleted by centrifugation then resuspended in 800 mL TRIzol for RNA isolation.
To monitor expression in response to nitrosative stress, wild-type S. Typhimurium was grown overnight, diluted 1:1000 in 100 mL fresh LB medium, and grown to OD 600 ; 1. The culture was subdivided into 5 duplicate pairs of 5 mL cultures in 18 Â 150 mm glass tubes. One set was treated with 2 mM DEANO and one set left untreated. Cultures were returned to shaking at 37°C. At 5, 15, 30, 45, and 60 min posttreatment, 1.5 ml of treated and untreated culture was pelleted by centrifugation then resuspended in 800 mL TRIzol for RNA isolation.
To determine whether supplementation of growth medium with manganese disrupts zinc homeostasis and expression of zinc-regulated zinc transporters, JK237 and EF657 S. Typhimurium were grown overnight, diluted 1:1000 in 2 Â 5 mL fresh LB medium, and grown to OD 600 ; 0.5. One culture was left untreated and the other was supplemented with 0.5 mM MnSO 4 before cultures were returned to shaking at 37°C. At 20 and 60 min posttreatment, 1.5 mL of treated and untreated culture was pelleted by centrifugation then resuspended in 800 mL TRIzol for RNA isolation.
For all analyses, RNA and cDNA were prepared as described previously (64). Quantitative PCR (qPCR) was carried out using SYBR green on a Bio-Rad CFX96 real-time system with rpoD as the internal control for normalization. Fold change values (treated/untreated) were log 2 transformed prior to plotting. Statistical significance was determined by one-sample t test compared to a hypothetical means of 1 or 21 using GraphPad Prism.
Western blotting. Strains EF829 and EF830 were grown as for spot assays, then 3 mL of culture was pelleted, resuspended in 100 mL PBS, and diluted 1:1 in 2Â Laemmli sample buffer (Bio-Rad) with DTT (100 mM final). Samples were incubated 60 min at 55°C before 7.5 mL was loaded on a 4-20% Tris-glycine gel for separation in Tris-glycine-SDS buffer (Bio-Rad). Separated proteins were transferred to nitrocellulose membranes, which were blocked with EveryBlot Blocking Buffer (Bio-Rad) then probed with 1:000 monoclonal anti-FLAG M2-peroxidase (HRP) antibody (Millipore Sigma). Blots were visualized using an ECL Western blotting analysis system (Amersham) on an ImageQuant LAS 4000 imaging system (GE Healthcare) that captures both chemiluminescent and visible images. The visible image of the protein size standards was aligned with the chemiluminescent image of the same blot.

ACKNOWLEDGMENTS
We thank Sarah Hasty for technical assistance with ICP-MS sample preparation and Joyce Karlinsey for helpful discussion and assistance with construction of the DyiiP mutant.