ZccE is a Novel P-type ATPase That Protects Streptococcus mutans Against Zinc Intoxication

Zinc is a trace metal that is essential to all forms of life, but that becomes toxic at high concentrations. Because it has both antimicrobial and anti-inflammatory properties and low toxicity to mammalian cells, zinc has been used as a therapeutic agent for centuries to treat a variety of infectious and non-infectious conditions. While the usefulness of zinc-based therapies in caries prevention is controversial, zinc is incorporated into toothpaste and mouthwash formulations to prevent gingivitis and halitosis. Despite this widespread use of zinc in oral healthcare, the mechanisms that allow Streptococcus mutans, a keystone pathogen in dental caries and prevalent etiological agent of infective endocarditis, to overcome zinc toxicity are largely unknown. Here, we discovered that S. mutans is inherently more tolerant to high zinc stress than all other species of streptococci tested, including commensal streptococci associated with oral health. Using a transcriptome approach, we uncovered several potential strategies utilized by S. mutans to overcome zinc toxicity. Among them, we identified a previously uncharacterized P-type ATPase transporter and cognate transcriptional regulator, which we named ZccE and ZccR respectively, as responsible for the remarkable high zinc tolerance of S. mutans. In addition to zinc, we found that ZccE, which was found to be unique to S. mutans strains, mediates tolerance to at least three additional metal ions, namely cadmium, cobalt, and copper. Loss of the ability to maintain zinc homeostasis when exposed to high zinc stress severely disturbed zinc:manganese ratios, leading to heightened peroxide sensitivity that was alleviated by manganese supplementation. Finally, we showed that the ability of the ΔzccE strain to stably colonize the rat tooth surface after topical zinc treatment was significantly impaired, providing proof of concept that ZccE and ZccR are suitable targets for the development of antimicrobial therapies specifically tailored to kill S. mutans.


Introduction
The first row d-block metal ions cobalt (Co), copper (Cu), iron (Fe), manganese (Mn), nickel (Ni) and zinc (Zn) are essential trace metals due to their roles in intracellular signaling and incorporation to structural and catalytic domains of proteins that perform the biological processes required for life [1,2]. On the flip side, these metals are toxic when present in excess, such that the ability to maintain metal homeostasis is absolutely critical for the survival of all forms of life [1][2][3]. To prevent or fight infections, the host activates a number of metal ion withholding strategies, collectively termed nutritional immunity, that ultimately starve pathogens of essential biometals such as Fe, Mn, and Zn [4]. For example, Fe in host fluids and tissues is sequestered by transferrin and lactoferrin, or stored within hepatocytes (reviewed in [5]). Both Mn and Zn are primarily sequestered by calprotectin, a heterodimer of the S100A8 and S100A9 proteins produced and secreted in large quantities by neutrophils (reviewed in [6]). At the same time, the host is capable of harnessing metal ions to intoxicate invading pathogens. For example, Cu and Zn are released in large quantities within phagocytic cells to kill invading pathogens [3,7].
Among the most biologically relevant trace metals, Zn is unique because it does not undergo redox-cycling. Due to its top position in the Irving-Williams series, Zn can bind more avidly and form more stable interactions with proteinaceous ligands when compared to most other metal ions. As a result, Zn is poisonous when in excess due to adventitious binding to non-cognate metalloproteins or even to residues that should not be occupied by metals, a process known as protein mismetallation [7,8]. Moreover, Zn is critical for proper functioning of the immune system as it is required for activation of pro-inflammatory signaling pathways, neutrophil extracellular trap (NET) formation and, as indicated above, it can be mobilized into phagosomes to kill invading pathogens [3,4,9]. Because it has antimicrobial properties, low toxicity to mammalian cells, and stimulates immune cell function, Zn has long been used as a therapeutic agent to treat or prevent a variety of medical conditions, from wound healing to gingivitis, or as an adjuvant in the treatment of common colds and other viral infections.
While the significance of Zn deprivation in host-pathogen interactions and the mechanisms utilized by bacteria to overcome Zn deprivation have been studied to some depth (reviewed in [8,10], the physiological consequences of Zn poisoning and the mechanisms by which pathogens overcome Zn intoxication are less well understood. In the initial response to toxic levels of Zn, bacteria have been shown to induce the expression and activity of membrane-associated Zn efflux systems to remove Zn from their cytosol. To date, three types of Zn efflux systems have been described in bacteria, namely P-type ATPases, resistance nodulation division (RND) pumps, and cation diffusion facilitator (CDF) metal/H + antiporters [11]. In Escherichia coli (the Gram-negative paradigm), the P-type ATPase ZntA is the major Zn efflux pump, whereas Bacillus subtilis (the Gram-positive paradigm) has been shown to utilize both a P-type ATPase (known as CadA) and CDF antiporter (known as CzcD) to pump Zn out of the cell [12][13][14][15]. In all streptococci studied to date, a CDF-type antiporter, homologous to the B. subtilis CzcD, has been shown to serve as the major Zn-resistance determinant, with several reports showing that inactivation of czcD significantly diminishes bacterial virulence [9,[16][17][18][19][20][21][22].
A resident of dental biofilms, Streptococcus mutans is a keystone pathogen of dental caries, one of the most prevalent and overlooked infectious diseases in the world [23,24]. In addition to its well-recognized role as a cariogenic pathogen, S. mutans is also implicated in infective endocarditis [25], with a recent retrospective study associating transient S. mutans bloodstream infections (BSI) with the highest risk of developing into infective endocarditis when compared to other streptococcal BSI [26]. In oral health, Zn has been incorporated into mouthwash and toothpaste formulations to prevent and treat halitosis and gingivitis [27,28]. In dental caries, the usefulness of Zn as a therapeutic agent is controversial, with a number of studies concluding that Zn is not effective contrasting with studies lauding Zn as a highly effective anti-plaque or anti-caries agent [29][30][31][32][33]. Controversy aside, mechanistic studies have shown that high concentrations of Zn inhibit the activity of glycolytic enzymes of oral streptococci in vitro and acid production by dental plaque in vivo [33][34][35][36]. Moreover, Zn and fluoride were shown to act synergistically, impairing S. mutans' acid tolerance as well as its ability to synthesize extracellular polymers of glucan from dietary sucrose [37].
Despite the widespread incorporation of Zn to oral healthcare products, very little is known about the mechanisms utilized by oral bacteria, S. mutans included, to maintain Zn homeostasis. Recently, we showed that inactivation of adcABC, which codes for an ABC-type Zn-specific transporter, drastically reduced the ability of S. mutans to colonize the dental biofilm in a rat model, indicating that host-mediated mechanisms for Zn sequestration, competition with microbial commensals, or both, restrict Zn availability to S. mutans in dental plaque [38]. In this report, we sought to uncover the mechanisms utilized by S. mutans to cope with toxic levels of Zn. We discovered that S. mutans exhibits much higher tolerance to Zn than all other streptococci tested, including several commensal oral streptococci associated with oral health. Through transcriptome and bioinformatic analyses, we identified a previously uncharacterized P-type ATPase transporter and cognate MerR-type transcriptional regulator as being directly responsible for the remarkable tolerance of S. mutans against Zn intoxication. In vitro and in vivo characterizations of a strain lacking this novel Zn efflux system and the fact that it is unique to S. mutans indicates that it is a potential target for the development of Zn-based antimicrobial therapies specifically tailored to kill S. mutans.

S. mutans is inherently more tolerant to Zn intoxication when compared to other streptococci
To uncover the mechanisms utilized by S. mutans to cope with toxic levels of Zn, we first determined the minimum inhibitory concentration (MIC) of ZnSO 4 in a panel of 17 streptococcal strains from 9 different species that were part of our lab collection. For the MIC determination and most other experiments, we chose BHI growth medium because it contains low levels of Zn (~10 μM) and because it supports robust growth of most streptococci. The panel included S. mutans and S. sobrinus (caries-associated streptococci), oral commensals (e.g., S. gordonii, S. sanguinis), and pyogenic streptococci S. agalactiae and S. pyogenes. Remarkably, all S. mutans strains displayed much higher Zn tolerance (MIC = 4 mM) when compared to the other species tested (MIC ranging from 0.5 to 1 mM) ( Table 1). The intrinsically high Zn tolerance of the S. mutans strains when compared to other streptococci was further validated in disc diffusion assays and growth curve analysis (Figs 1 and S1).

Overview of changes to the S. mutans UA159 transcriptome following exposure to excess Zn
To obtain mechanistic insight into the attributes that promote the high Zn tolerance of S. mutans, we used RNA deep sequencing (RNA-Seq) to uncover the transcriptome of S. mutans UA159 following exposure to excess ZnSO 4 . To determine the most appropriate concentration of Zn for this experiment, the growth kinetics of exponentially grown cultures (OD 600~0 .3) of S. mutans UA159 treated with increasing concentrations of ZnSO 4 were monitored (S2 Fig). As exposure of exponentially grown cultures to 4 mM ZnSO 4 resulted in a marked growth defect, but not complete growth arrest as seen at 6 mM ZnSO 4 , this concentration was chosen for the transcriptome analysis. Applying a False Discovery Rate (FDR) of 0.05 and 2-fold cutoff, we found 96 genes upregulated and 62 genes downregulated after 15 minutes (T 15 ) of exposure to 4 mM ZnSO 4 when compared to the unstressed control (S1 Table). After 90 minutes (T 90 ), the number of differently expressed genes increased slightly from 158 at T 15 to 199 genes, with 85 genes upregulated and 114 genes downregulated (S1 Table). Among the genes differently expressed, 60 genes displayed the same expression trends (up or downregulated) at both time points whereas 9 genes showed opposing trends of regulation. To illustrate these results, differentially expressed genes were grouped according to Clusters of Orthologous Groups (COG) functional categories (Fig 2) [39].
One of the most highly induced (greater than 25-fold upregulation) genes at both time points was smu2057c, coding for a putative P 1B -type ATPase transmembrane protein. Also of interest, the entire copYAZ operon, of which copA encodes another P 1B -type ATPase involved

PLOS PATHOGENS
Streptococcus mutans zinc tolerance in Cu tolerance [40][41][42] were among the most upregulated genes at T 15 (20-to 24-fold upregulated) remaining induced to a lesser extent at T 90 (3-to 4-fold upregulated). Another group of genes highly upregulated at both time points encode for components of the lactose/galactose phosphoenolpyruvate transferase (PTS) and tagatose pathways (lacXGEFDCBAR, 4-to 40-fold upregulated). While genes associated with lactose uptake and utilization were strongly induced by high Zn stress, genes coding for the main PTS enzymes responsible for the uptake of sucrose (scrA), glucose (manLMN), and glucose disaccharide (celCRB) were downregulated at both time points. In addition, at T 15 , genes involved in the uptake or metabolism of cysteine (tycABC, tcyEFGH, cysK) and glutathione (gst, gshT, gor) were upregulated by 2-to 5-fold whereas genes encoding for mutacins IV (nlmD), V (nlmC) and VI (nlmAB) were induced by 8-to 16-fold at T 90 . Finally, several stress genes were induced at T 15 , including genes coding for heat shock proteins/molecular chaperone (groES-EL and hrcA-grpE-dnaK-dnaJ operons) and oxidative stress genes (dpr, gor and tpx). In contrast, the same dpr and tpx genes induced at T 15 and additional genes classically associated with oxidative stress responses (ahpCF, nox and sodA) were downregulated at T 90 . Cultures were grown to mid-exponential phase, spread onto BHI plates, and topped with filter paper discs saturated with 1 mM ZnSO 4 . Growth inhibition zones (in mm) around the discs were measured after 24 h incubation at 37˚C in 5% CO 2 . Data represent means and standard deviations of results from 3 independent experiments. One-way ANOVA was used to compare the zones of growth inhibition of each strain to that of S. mutans UA159. A p value of <0.05 was considered significant ( � ). https://doi.org/10.1371/journal.ppat.1010477.g001

PLOS PATHOGENS
Streptococcus mutans zinc tolerance smu2057c encodes a multi-metal translocating P-type ATPase unique to S. mutans As smu2057c was one of the most strongly upregulated genes during high Zn stress and knowing that a handful of bacterial P 1B -type ATPases are implicated in metal ion transport [13][14][15]43], we predicted that Smu2057c would function as a Zn efflux system. Based on searches of public databases, Smu2057c is exclusively found in S. mutans genomes, with the exception of a single Streptococcus troglodytae genome encoding a hypothetical protein that was 94% identical to Smu2057c. A phylogenetic analysis confirmed that ZccE is evolutionarily distant from other P-type ATPases available on public databases (Fig 3). Compared to P 1B -type ATPases previously implicated in metal tolerance, the protein coded by smu2057c shares 33% amino acid identity with the B. subtilis CadA, which mediates tolerance to cadmium (Cd), cobalt (Co) and Zn [14,44], and 36% identity to E. coli ZntA that mediates tolerance to Cd, lead (Pb) and Zn [13,15] (S3 Fig). Finally, we determined the predicted structure of ZccE using AlphaFold2 and compared it to the available structure of ZntA (S4 Fig). Despite the similarities at the amino acid level, the two structures are distinctly different.
To determine if Smu2057c mediates Zn tolerance, we generated a smu2057 deletion strain in the UA159 wild-type background and then tested the capacity of the mutant to grow in chemically-defined (FMC) or complex (BHI) media containing increasing concentrations of Zn. Because the experiments described below implicate Smu2057c in Cd, Co, Cu and Zn tolerance, we named smu2057c as zccE, for zinc, cadmium, cobalt, and copper exporter. When compared to UA159, the ΔzccE strain displayed a much higher sensitivity to Zn as it was completely unable to grow in media supplemented with 1 mM ZnSO 4 (Fig 4A-4C). In agreement with the growth curves and plate titration assay, the Zn MIC of ΔzccE was 20 times lower than the MIC of the parent strain (

PLOS PATHOGENS
Streptococcus mutans zinc tolerance strain) fully rescued the Zn sensitive phenotype (Fig 4A-4C), while inactivation of zccE in the S. mutans OMZ175 background strain led to comparable and drastic reduction in Zn tolerance supporting that ZccE is the primary Zn tolerance determinant of S. mutans (S5 Fig). Because P-type ATPases implicated in Zn tolerance have been shown to confer resistance to other metal ions, especially those of similar size and charge, we also tested the capacity of ΔzccE to grow on BHI plates supplemented with Cd, Co, Cu, Fe, Mn, and Ni. As shown in Fig 4D, ZccE confers resistance to Cd, Co, and Cu but not to Fe, Mn, or Ni. The increased sensitivity of ΔzccE to Cd, Co and Cu was restored in the complemented strain.

ZccE and CopA work synergistically to protect S. mutans against Cu intoxication in oxidizing environments
In several bacteria, including all streptococci, Cu tolerance is mediated by the Cu-translocating P-type ATPase encoded by the copA gene, which is organized in an operon with copY (Cusensing transcriptional repressor) and copZ (Cu chaperone) [45][46][47]. Because we found that ZccE also confers Cu tolerance and that the copYAZ operon was induced upon high Zn stress, we sought to determine to what extent ZccE contributes to Cu tolerance and to probe the possible association of CopA with Zn tolerance. To this end, we deleted the entire copYAZ operon in both the UA159 and ΔzccE background strains to generate the ΔcopYAZ and ΔzccEΔcopYAZ strains. Next, we tested the capacity of each mutant strain to grow in media containing high concentrations of CuSO 4 or ZnSO 4 . Despite strong upregulation of the copYAZ operon during the initial 15 minutes of high Zn stress (S1 Table, Fig 2), inactivation of copYAZ alone did not

PLOS PATHOGENS
Streptococcus mutans zinc tolerance affect Zn sensitivity (Fig 5A). While the ΔzccEΔcopYAZ strain phenocopied ΔzccE on plates containing 0.1 or 1 mM ZnSO 4 , the double mutant was slightly more sensitive than the ΔzccE single mutant on plates containing 0.05 mM ZnSO 4 (Fig 5A). Collectively, these results indicate that CopA plays a minor role in Zn tolerance that becomes dispensable when ZccE is present.
As shown by others [40,45], inactivation of the copYAZ operon drastically heightened Cu sensitivity albeit the ΔzccE strain was slightly more sensitive than ΔcopYAZ on plates containing 500 μM CuSO 4 (Fig 5B). Unexpectedly, inactivation of both copYAZ and zccE did not appear to further enhance Cu sensitivity when compared to the single mutant strains (Fig 5B). Though Cu should primarily be found as Cu + in the reducing bacterial cytoplasm [48], it is capable of alternating between Cu + or Cu 2+ . Thus, we wondered if ZccE could mobilize both oxidized and reduced Cu and, if so, whether there was a preference for one redox state over another. Of note, previous work has shown that the pneumococcal Cop operon can efficiently remove Cu + and Cu 2+ from the bacterial cytoplasm into the extracellular milieu [49]. In light of this, we reevaluated the Cu tolerance of ΔzccE and ΔcopYAZ single and double mutants under anaerobiosis such that all Cu ions available should be in the reduced Cu + form. In agreement with the knowledge that Cu + is more toxic than Cu 2+ , supplementation of agar plates with 1 mM CuSO 4 caused complete growth inhibition of all strains, such that experiments using an anaerobic environment were conducted with plates containing at most 0.1 mM CuSO 4 . Interestingly, we found that Cu tolerance under anaerobiosis was solely mediated by

Streptococcus mutans zinc tolerance
CopA, suggesting that ZccE cannot efficiently mobilize Cu + (S6 Fig). Taken together, these results reveal that CopA and ZccE interchangeably mediate Cu protection in aerobic environments. However, CopA is likely to have greater biological relevance to Cu tolerance than ZccE due to its capacity to export both Cu + and Cu 2+ .

ZccE is positively regulated by a MerR-type regulator
Sequence analysis revealed that zccE is located immediately upstream but transcribed in the opposite orientation of smu2058, an uncharacterized MerR-type transcriptional regulator ( Fig  6A). To explore a possible role of Smu2058 in zccE regulation, the smu2058 coding sequence was replaced by a Spec R cassette generating the ΔzccR strain (for zccE regulator). Using the same growth conditions as were used in the RNA-Seq analysis described above, quantitative real time PCR (qRT-PCR) analysis revealed that zccE transcription is almost entirely dependent on ZccR (Fig 6B). In agreement with this observation, the ΔzccR strain phenocopied the ΔzccE strain and was unable to grow in broth (S7 Fig) or on agar plates (Fig 6C) supplemented

PLOS PATHOGENS
Streptococcus mutans zinc tolerance with Zn. Genetic complementation of ΔzccR (ΔzccR Comp strain) reversed the heightened Zn sensitivity indicating that the phenotype was not due to polar effects on zccE or secondary mutations (Figs S7 and 6C).
To determine if ZccR regulation of zccE transcription is direct or indirect, we performed electron mobility shift assays (EMSAs) using recombinant ZccR purified from E. coli and a DNA probe encompassing the entire zccE-zccR intergenic region (Fig 6A). The EMSA results revealed that addition of sufficient ZccR resulted in a shift of the protein: DNA complex, indicating that ZccR binds specifically to the zccE-zccR intergenic region (Fig 6D). Formation of the ZccR:DNA complex was inhibited by the Zn-specific chelator TPEN (N,N,N 0 ,N 0 -tetrakis (2-pyridinylmethyl)-1,2-ethanediamine) and competitively reverted by the excess Zn supplementation (Fig 6D). Thus, we conclude that ZccR positively and directly regulates zccE expression, likely by sensing and responding to intracellular Zn pools.

Inactivation of zccE or zccR drastically alters Zn:Mn ratios in high Zn conditions
Next, we used inductively coupled plasma mass spectrometry (ICP-MS) to quantify the intracellular trace metal pools in the UA159 (parent strain), ΔzccE and ΔzccR strains exposed to high Zn stress. Briefly, cultures were grown in BHI broth (~10 μM Zn,~0.5 μM Mn [50]) to mid-log phase and treated with a ZnSO 4 solution to reach a final concentration of 1 mM of Zn. For ICP-MS analysis, samples were obtained immediately before (control) and 90 minutes

PLOS PATHOGENS
Streptococcus mutans zinc tolerance after addition of Zn. Despite the high concentration of Zn added to the culture, parent and complemented mutant strains maintained the same intracellular amounts of Zn before and after the Zn challenge (Fig 7A). While intracellular levels of Zn in the ΔzccE and ΔzccR strains before addition of Zn did not differ from parent and complemented mutants, intracellular Zn pools rose by~10-fold in the mutants after Zn challenge (Fig 7A). Moreover, accumulation of Zn in the ΔzccE and ΔzccR strains inversely correlated with a substantial decrease (~10-fold) in intracellular Mn pools (Fig 7B). The intracellular levels of Cu or Fe did not alter upon Zn stress either in the parent strain UA159 and ΔzccE (Fig 7C & 7D), while other d-block elements such as Co and Cd were below the detection limit. To determine whether the other 3 metals (Cd, Co, and Cu) accumulate in ΔzccE when provided in excess, we expanded our ICP-MS analysis to include cells grown in BHI supplemented with a combination of the 4 metals (Zn, Co, Cu, and Cd) associated with ZccE activity. As shown in Fig 8, the ΔzccE strain accumulated significantly more Cd, Co, Cu and Zn compared to parent or complemented strains.

Disruption of Zn homeostasis diminishes oxidative stress tolerance
As both Zn and Mn are linked to oxidative stress responses [50][51][52], we sought to determine if simultaneous perturbations of Zn and Mn homeostasis (and therefore of Zn:Mn ratios) in

Streptococcus mutans zinc tolerance
ΔzccE affected the ability of this strain to cope with peroxide stress. In disc diffusion assays, both UA159 and ΔzccE strains displayed similar sensitivity to H 2 O 2 when grown in BHI alone (~10 μM Zn) while supplementation of sub-inhibitory concentrations of Zn (50 to 100 μM ZnSO 4 ) resulted in significantly greater sensitivity of ΔzccE when exposed to H 2 O 2 (Fig 9A). Next, we tested the ability of S. gordonii DL1, a peroxigenic oral commensal, to inhibit the

PLOS PATHOGENS
Streptococcus mutans zinc tolerance growth of the S. mutans parent and ΔzccE strains using an antagonism plate assay. Like the disc diffusion assay, both strains showed similar sensitivities to S. gordonii when grown on BHI plates, while supplementation of the media with as little as 50 μM ZnSO 4 virtually abolished growth of ΔzccE (Fig 9B). To confirm that the growth defect was due to sensitivity to the H 2 O 2 produced by S. gordonii, a control assay was performed in which catalase was added atop the S. gordonii prior to spotting of the S. mutans cultures. When the H 2 O 2 was neutralized by catalase, no growth sensitivity by the S. mutans strains was observed (Fig 9B). To quantify the effect of Zn in S. gordonii competition we also measured CFUs after 24 hours of incubation of S. mutans UA159, ΔzccE and ΔzccE comp in BHI containing increasing concentrations of Zn (0, 50 and 100 μM Zn) either in monoculture or co-culture with S. gordonii (1:2 ratio of S. mutans:S. gordonii). As expected, only the ΔzccE strain became more susceptible to S. gordonii killing after the addition of excess Zn (S8 Fig).

Topical Zn treatment impairs oral colonization efficiency of the ΔzccE strain
To determine the impact of Zn homeostasis in an in-vivo model, we assessed the ability of the parent and ΔzccE strains to colonize the oral cavity of rats fed a cariogenic diet, while simultaneously testing whether topical Zn treatment could inhibit bacterial colonization (Fig 10). To simulate oral hygiene such as a mouthwash, rats that had been orally infected with S. mutans UA159 or ΔzccE were treated twice daily with an oral application of a ZnSO 4 solution. When comparing the control treatment groups (saline), similar numbers of colony-forming units (CFU) were recovered from animals infected with either UA159 or ΔzccE, indicating that the multi-metal tolerance conferred by ZccE was dispensable for oral colonization in the absence of excess Zn. However, daily treatment with 60 mM or 150 mM ZnSO 4 solutions, a concentration range found in commercially available oral healthcare products, resulted in significant reduction (~1-log) of the CFU recovered from rats that had been infected with the ΔzccE strain. These same Zn treatments showed only modest, and not statistically significant, inhibitory effects against the parent strain. Importantly, Zn treatment did not affect the abundance of total flora recovered from the S. mutans-infected animals, suggesting that the inhibitory effect was specific to the ΔzccE strain. Taken together, these results provide the proof of

PLOS PATHOGENS
Streptococcus mutans zinc tolerance concept that ZccE is a suitable target for future development of an antimicrobial therapy specific to treat S. mutans infections.

Discussion
The second most abundant trace metal in the human body, Zn is naturally found in saliva and dental plaque at levels that, in theory, should not be limiting or toxic to bacteria. Yet, our recent characterization of the AdcABC Zn import system suggested that Zn might be a growth-limiting factor in dental biofilms, as inactivation of the adcABC metal transport system greatly reduced S. mutans colonization in a rat oral infection model [38]. Whether this apparent Zn restriction was caused by host-mediated Zn sequestration mechanisms, increased competition for environmental Zn with microbial commensals, or the combination of both remains to be determined. At the other end of the spectrum, Zn is required for host immune function and shows antimicrobial properties at elevated concentrations, such that it is harnessed by the host to intoxicate invading pathogens within phagocytic cells and, for centuries, has been used as a therapeutic agent to treat a variety of human conditions due to its antimicrobial and anti-inflammatory properties. In oral health, Zn is incorporated into oral healthcare products to treat gingivitis and to control halitosis [27,28] though its usefulness to caries treatment or prevention has proven to be controversial [27,33,53,54]. In this study, we used transcriptomic and molecular genetic approaches to uncover the mechanisms that allow S. mutans to overcome Zn toxicity. Through RNA-Seq analysis, we identified a previously uncharacterized P-type ATPase transporter, encoded by smu2057c (zccE), as involved in Zn tolerance. Importantly, we discovered that zccE is part of the core genome and unique to S. mutans. By comparing Zn tolerance levels of several streptococcal species that do not encode ZccE with those of the wild-type and ΔzccE strains of S. mutans, it became clear that ZccE was directly responsible for the inherently high Zn tolerance of S. mutans. Like other bacterial metal-translocating P-type ATPases that have the capacity to translocate multiple metals, loss of zccE also led to increased sensitivity to Cd, Co, and Cu. In addition, we identified zccR (smu2058), a member of the MerR family of transcriptional factors, as the major transcription factor controlling zccE expression in a Zn-dependent manner.
In addition to identification of zccE, the RNA-Seq analysis provided additional clues on how S. mutans might overcome Zn intoxication. For example, transcriptional changes in the expression of sugar transport and sugar metabolism genes suggest that S. mutans activates the tagatose-6-phosphate pathway to bypass the initial steps of the glycolytic pathway. These steps have been shown to be inhibited by millimolar concentrations of Zn in the closely related species S. pyogenes . Specifically, our transcriptome results revealed that elevated Zn concentrations repressed transcription of PTS genes involved in glucose and glucose disaccharide metabolism whereas genes of the lactose PTS and the tagatose pathway were strongly induced (S1 Table and Fig 2). A well-known mechanism utilized by bacteria to overcome Zn toxicity is based on the accumulation of Zn-buffering molecules including cysteine-containing molecules such as glutathione and free cysteine [8,55]. Our RNA-Seq analysis also revealed that S. mutans upregulates expression of genes involved in glutathione transport (tcyABC) and regeneration (gst), cystine transport (tcyDEFG) and metabolism (cysK) in response to high Zn stress. Future studies are necessary to test the hypothesis that S. mutans activates lactose metabolism to bypass a possible metabolic bottleneck created by Zn-mediated inhibition of the initial steps of glycolysis, and to whether glutathione and cysteine are indeed used as Zn-buffering systems to protect S. mutans against Zn intoxication.
Despite the multiple strategies bacteria utilize to mitigate Zn poisoning, it is the capacity to efficiently pump Zn out of their cytosol that seems to primarily define the levels of Zn

PLOS PATHOGENS
Streptococcus mutans zinc tolerance tolerance. Until now, the only Zn efflux system known in streptococci was encoded by czcD, a CDF-type efflux system that has been characterized in major human pathogens including S. pneumoniae, S. pyogenes, and S. agalactiae [9,18,19,22]. However, the genome of S. mutans does not encode a czcD homologue; the only S. mutans gene coding for a CDF-type transporter is smu1176 (mntE), which was recently shown to alleviate Mn toxicity [56]. Of interest, several bacteria, including the Gram-positive paradigm B. subtilis, have been shown to utilize both a CDF-type transporter and a P-type ATPase transporter to maintain Zn homeostasis during high Zn stress [12,14]. P-type ATPases are a large family of transmembrane proteins that use the energy generated by ATP hydrolysis to transport cations and other substrates across membranes. P 1B -type ATPases are primarily involved in metal ion transport and are comprised of six to eight transmembrane helices that form the ionic channel across the cell membrane with two soluble cytoplasmic domains. In these metal translocating P-type ATPases, one of the eight helices bears a tripeptide 'CPC' metal-binding signature motif with a conserved center proline flanked by either cysteine/serine or histidine at either side [(C/S/T)P(C/H/P)] [57,58]. While the proline residue is essential, the amino acids surrounding the proline have been shown to define metal specificity of these transmembrane proteins. For example, the canonical E. coli ZntA has a fully conserved 'CPC' motif that confers high selectivity to Pb followed by Zn and Cd, and low selectivity to Co, Cu and Ni -substitution of one of the surrounding cysteines by histidine or serine resulted in loss of binding to Pb but not to Zn or Cd [59]. Based on previous studies, P 1B -type ATPases with 'CPC' motifs are selective for silver (Ag), Cd, Cu, Pb, and Zn whereas proteins with SPC or CPH motifs are associated with Co and Cu export [59]. In addition to the 'CPC' motif, metal specificity is determined by the presence of GxxCxxC motif at the N-terminal end and the presence of a preceding aspartate residue [59]. Although ZccE possesses an 'SPC' motif, the N-terminal GxxCxxC motif is absent in ZccE such that the coordination chemistry and key residues of ZccE that determine metal specificity or promiscuity remain to be fully determined.
One of the best described mechanisms by which Zn exerts its toxicity in bacteria is by impairing Mn homeostasis [18,60]. In S. pneumoniae, Zn binds to the Mn transporter PsaABC when present at high concentrations, competitively affecting Mn uptake and rendering S. pneumoniae hypersensitive to oxidative stress and immune cell killing [61]. Of note, Zn outcompetes Mn for binding to PsaABC without being internalized [61]. Follow up studies with S. pneumoniae and S. agalactiae indicated that a drastic alteration in Zn:Mn ratios, rather than Zn concentration itself, was the main cause for Zn toxicity; Mn supplementation alone restored the equilibrium of Zn:Mn ratios and alleviated the growth defect phenotypes associated with high Zn stress [22,60,61]. Here, we found that the~10-fold increase in intracellular Zn levels in ΔzccE was accompanied by a similar decrease in intracellular Mn pools (Fig 7), such that Zn:Mn ratios changed from~1:1 before addition of Zn to~100:1 after the challenge (S9 Fig). While Zn has been shown to compete with Mn for binding to the DtxR family transcriptional repressor SloR [62], our RNA-Seq analysis indicates that the decrease in total Mn pools was not caused by altered transcription of the SloR-regulated genes responsible for either Mn import (sloABC and mntH) or export (mntE) [50,56]. Mn plays multiple and important roles in oxidative stress tolerance, so this notable decrease in intracellular Mn pools provides one possible explanation, albeit likely not the only one, for the hypersensitivity of ΔzccE to H 2 O 2 when simultaneously exposed to sub-inhibitory concentrations of Zn (Fig 9). To check whether the Zn:Mn ratio is also affected at the sub-inhibitory concentration of Zn that was used in the H 2 O 2 sensitivity experiments, we also determined the intracellular Zn and Mn concentrations and Zn:Mn ratios in parent and ΔzccE cells grown in 150 μM Zn. While not as robust as the differences observed in media containing 1 mM Zn, Zn:Mn ratios were significantly altered when the ΔzccE strain was grown in media containing 150 μM Zn (S9 Fig). Of interest, the parent strain was capable of maintaining steady levels of both Zn and Mn (and therefore balanced Zn:Mn ratios) when challenged with either 150 or 1 mM Zn, suggesting that intracellular and not extracellular Zn levels are affecting Mn homeostasis. To further explore the association of Mn with Zn toxicity effects, we asked if Mn supplementation could rescue phenotypes of the ΔzccE strain when exposed to excess Zn. When ΔzccE was exposed to high Zn stress, the addition of Mn offered only minimal growth restoration (S10 Fig). However, the restorative effect of Mn supplementation was clear in the context of tolerance of ΔzccE to H 2 O 2 (S10 Fig). Taken together, these results support previous investigations that demonstrate that Zn and Mn homeostasis are intertwined, and that this relationship is critical to bacterial pathophysiology. For this reason, mechanistic studies to understand how intracellular Zn perturbs Mn flux in S. mutans warrants further investigation.
To explore the potential of ZccE as an antimicrobial target, we performed an oral colonization study that included topical application of a ZnSO 4 treatment to animals that had been infected with S. mutans UA159 or ΔzccE. As anticipated, the ΔzccE strain was susceptible to Zn treatment whereas the parent strain was not (Fig 10). These results suggest that targeted inhibition of ZccE expression or activity can be combined with Zn to develop an anti-caries therapy by specifically killing S. mutans.
To summarize, this report sheds new light onto the molecular mechanisms utilized by S. mutans to overcome Zn toxicity. We discovered that S. mutans is inherently more tolerant to the toxic effects of Zn than other streptococci, including several non-cariogenic oral streptococci that are associated with oral health. We also identified the major effectors of high Zn tolerance in S. mutans, the metal translocating P 1B -type ATPase ZccE and the Zn-responsive transcriptional factor ZccR. As described in Fig 11, while Zn import in streptococci is mediated mainly by a highly conserved ABC type transporter AdcABC, the Zn export in S. mutans differs significantly from other streptococcal species by utilizing a unique Zn exporter ZccE   Fig 11. Model figure showing that ZccE (a P-type ATPase transporter) and CzcD (a CDF-type transporter) mediate high Zn tolerance in streptococci. Both czcD and zccE encode for Zn efflux systems and each system is positively regulated by cognate transcriptional regulators (SzcA for czcD and ZccR for zccE) that are activated under high Zn conditions. The CzcD/SzcA pair is found in most streptococci while ZccE/ZccR is unique to S. mutans and, based on Zn sensitivity assays and mutational analysis, is directly responsible for the exceptionally high Zn tolerance of S. mutans when compared to other streptococcal species. AdcABC is an ABC-type transporter conserved among streptococci and primarily responsible for Zn import. This figure was created with Biorender. https://doi.org/10.1371/journal.ppat.1010477.g011

PLOS PATHOGENS
Streptococcus mutans zinc tolerance that provides S. mutans a clear survival advantage over other streptococci in high Zn environment. Because ZccE is unique to S. mutans, we propose that small molecules that specifically inhibit ZccE activity (or expression) can be combined with a Zn source to kill S. mutans. Thus, the combination of a ZccE-specific inhibitor with Zn can be exploited as a new antimicrobial therapy to prevent dental caries or to treat systemic S. mutans infections.

Ethics statement
Animal procedure for rat oral colorizations was approved by the University of Florida Institutional Animal Care and Use Committee (protocol # 201810421). All animal care was consistent with the Guide for the Care and Use of Laboratory Animals from the National Research Council and the USDA Animal Care Resource Guide.

Bacterial strains and growth conditions
The strains used in this study are listed in Table 2. All strains were routinely grown in brain heart infusion (BHI) at 37˚C in a 5% CO 2 atmosphere. When appropriate, antibiotics were added to cultures at the following concentrations: spectinomycin (1 mg ml -1 ), kanamycin (1 mg ml -1 ), erythromycin (10 μg ml -1 ). BHI and the chemically-defined medium FMC [63] were used to generate growth curves using an automated growth reader (Bioscreen C; Oy Growth Curves Ab, Ltd.). Briefly, overnight cultures were sub-cultured 1:20 in fresh media, grown to mid-exponential phase (OD 600 of 0.4), and diluted 1:50 into the appropriate medium with or without metal supplementation in the wells of a microtiter plate. An overlay of sterile mineral oil was added to each well to minimize generation of reactive oxygen species. For RNA-Seq and qPCR analyses, bacterial cultures were grown in BHI to mid-log phase (OD 600 of 0.4, untreated control) and then exposed to a final Zn concentration of 4 mM (using a concentrated ZnSO 4 solution) for 15 or 90 minutes.

Construction of mutant and complemented strains
Strains lacking zccE, zccR, or the copYAZ operon were constructed using a standard PCR ligation mutagenesis approach [65]. Briefly, DNA fragments (~1 kb) flanking the target gene were amplified by PCR, digested with restriction enzymes, and ligated to a similarly-digested nonpolar spectinomycin resistance cassette for constructing ΔzccE and ΔzccR strains and to a nonpolar erythromycin resistance cassette to obtain the ΔcopYAZ strain. The ligation mixtures were used to transform S. mutans UA159 in the presence of XIP (comX-inducing peptide) following an established protocol [66]. Mutant strains were selected on BHI agar supplemented with the appropriate antibiotic and gene deletions confirmed by PCR amplification and Sanger sequencing. The double ΔzccEΔcopYAZ mutant was obtained by transforming the ΔcopYAZ strain with the zccE::Spec R region. To generate complemented strains, the full length zccE, zccR, or copYAZ coding sequences with their native promoters were amplified and cloned into the integration vector pMC340B [67] to yield plasmid pMC340B-zccE, pMC340B-zccR, and pMC340B-copYAZ, respectively. Plasmids were propagated in E. coli DH10B and used to transform the S. mutans strains ΔzccE, ΔzccR, or ΔcopYAZ for integration at the mannitol utilization (mtl) locus. All primers used in this study are listed in S2 Table. Phylogenetic analysis P-type ATPases, either experimentally characterized or predicted as metal exporters, were chosen to include at least two representative proteins from each characterized sub-family as well

PLOS PATHOGENS
Streptococcus mutans zinc tolerance as proteins identified by NCBI's BLASTP [68] to be relatively closely related (either in a search among streptococci or in search excluding all streptococci) [69]. Sequence labels were standardized and sequences were aligned using EMBL-EBI's Clustal Omega [70] to generate a distance tree which was then modified for readability using the Interactive Tree of Life vs 6.5.8 [71].

RNA analysis
Total RNA was isolated from homogenized S. mutans cell lysates by acid-phenol chloroform extractions as previously described [50]. Total RNA isolated was treated with Ambion DNase I (ThermoFisher) for 30 min at 37˚C and further purified using an RNeasy kit (Qiagen), which included a second on-column DNase digestion. Sample quality and quantity were assessed on an Agilent 2100 Bioanalyzer at the University of Florida Interdisciplinary Center for Biotechnology Research (UF-ICBR). For RNA-Seq analysis, samples were prepared as previously

PLOS PATHOGENS
Streptococcus mutans zinc tolerance described [50]. Briefly, 5 μg of the total RNA isolated was subjected to two rounds of mRNA enrichment using a MICROBExpress bacterial mRNA purification kit (Thermo Fisher), and 100 ng of the enriched mRNA used to generate cDNA libraries with unique barcodes (Next UltraII Directional RNA Library Prep kit for Illumina, New England Biolabs). The cDNA libraries were pooled and subjected to RNA-Seq analysis at the UF-ICBR using the Illumina NextSeq 500 platform. Read mapping was performed on a Galaxy server using Map with Bowtie for Illumina and the S. mutans UA159 genome (GenBank accession no. NC_004350.2) as a reference.
The reads per open reading frame were tabulated with htseq-count. Final comparisons between control and Zn-treated conditions were performed with Degust (http://degust. erc.monash.edu/), with a false-discovery rate (FDR) of 0.05 and 2-fold change cutoff. Quantifications of zccE and gyrA mRNA were determined by qRT-PCR following an established protocol [72] using the primers listed in S2 Table. The fold change of zccE expression was performed using ΔΔC T (where C T is the threshold cycle) method, with gyrA being used as standardization control. One-way ANOVA was performed to verify significance of the qRT-PCR results.

Metal sensitivity assays
To test sensitivity to Zn in disc diffusion assays, 25 μl of exponentially grown cultures (OD 600 0.5) were spread onto agar plates using a sterile swab and topped with sterile Whatman filter paper discs (6-mm diameter) saturated with 20 μl of a 1 mM ZnSO 4 solution. The diameter of the zone of growth inhibition was measured after 24 h of incubation at 37˚C in 5% CO 2 . The minimum inhibitory concentration (MIC) of ZnSO 4 was determined by a broth microdilution method using two-fold serial dilutions of ZnSO 4 . Plates were incubated at 37˚C in 5% CO 2 for 16 h and the concentration of ZnSO 4 at which the absorbance values were 10% of the control condition was determined to be the MIC. For plate titration assays, exponentially grown BHI cultures (OD 600~0 .5) were serially diluted and 8 μl of each 10-fold dilution spotted on BHI plates supplemented with selected trace metals. Plates were photographed after 24 h incubation at 37˚C in 5% CO 2 or under anaerobiosis (GasPak jar, BD Biosciences).

Antagonism assay
The ability of S. gordonii to inhibit the growth of S. mutans via H 2 O 2 production was assessed as described previously [50]. Briefly, 8 μl of an overnight culture of S. gordonii DL1 was spotted on BHI plates with or without Zn (50 μM ZnSO 4 ) supplementation and incubated at 37˚C in 5% CO 2 . After 24 h incubation, 8 μl of an overnight culture of S. mutans UA159 or ΔzccE was spotted proximal to the S. gordonii spot. To confirm that growth inhibition was due to H 2 O 2 production, a control condition was included in which 8 μl of 1 mg ml −1 catalase solution was spotted on top of the S. gordonii spot prior to spotting S. mutans. To obtain quantitative data of the S. mutans-S. gordonii competition, S. mutans strains were either co-cultured with S. gordonii DL1 or grown alone for 24 hours in BHI media containing 0, 0.05, or 0.1 mM Zn before plating on media selective for S. mutans. BHI agar supplemented with 1 mg/ml of kanamycin was used for selection of UA159-Km, a derivative of the parent strain UA159 for which a kanamycin resistance gene has been incorporated at the gtfA locus [64]. BHI agar supplemented 1 mg/ml of spectinomycin was used for selection of the ΔzccE strain.

ICP-MS analysis
The intracellular Zn and Mn content in parent and mutant strains was determined via inductively coupled plasma mass spectrometry (ICP-MS) performed at the University of Florida Institute of Food and Agricultural Sciences (UF-IFAS) Analytical Services Laboratories.
Briefly, cultures (250 ml) were grown in BHI to mid-exponential phase (OD 600 0.4), harvested by centrifugation at 4˚C for 15 min at 4,000 rpm, washed in PBS supplemented with 0.2 mM EDTA (to chelate extracellular divalent cations) followed by a second wash in PBS only. The cell pellets were resuspended in 35% (v/v) HNO 3 , heated at 95˚C for 1 h before diluted to 3.5% (v/v) HNO 3 using metal free water and the intracellular Zn and Mn content determined using a 7900 ICP Mass Spectrometer (Agilent). Metal concentrations were normalized to total protein content determined by the bicinchoninic acid (BCA) assay (Pierce).

ZccR purification
To overexpress and purify a recombinant His-tagged ZccR, the 795-bp zccR coding region was amplified by PCR using the primers listed in S2 Table, and the amplicon cloned onto the expression vector pET30c (Novagen) using the In-fusion HD Cloning Plus kit (TaKaRa) to generate plasmid pET30c-zccR. The resultant plasmid was confirmed by Sanger sequencing and transformed into E. coli BL21 λDE3. The resulting E. coli strain harboring pET30c-zccR was grown in a shaking 37˚C incubator in Luria-Bertani broth to mid-exponential phase (OD 600 0.4). Recombinant His-tagged ZccR was overexpressed by addition of 0.2 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG) (Teknova) and the culture was incubated for an additional 3 h. Cell pellets were collected by centrifugation, resuspended in lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10 mM imidazole, pH 8) containing lysozyme (1 mg ml -1 ) and sonicated to clarity before centrifugation of cellular debris. The cleared lysate supernatant was mixed with pre-cleared Ni-NTA beads at 4˚C for 60 minutes before loaded onto a chromatography column. The packed column was washed with wash buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 20 mM imidazole), and His-ZccR was eluted (50 mM NaH 2 PO 4 , 300 mM NaCl, 250 mM imidazole) according to the QIAexpressionist protocol (Qiagen) for purification of native proteins with histidine tags. The elution fractions were analyzed by 12% SDS-PAGE, and the fractions containing highly pure ZccR pooled and dialyzed against PBS. Aliquots of purified protein were stored at -20˚C in 15% glycerol until further use.

Electrophoretic mobility shift assays
EMSAs were performed with slight modifications from established protocols [50,73]. Briefly, the intergenic region (IGR) between zccE and zccR was amplified using the primers listed in S2 Table, and the resulting 160-bp amplicon was biotin-labeled using the Biotin 3' End DNA Labeling kit (ThermoFisher). EMSA reactions were prepared in 20 μl reaction mixtures in binding buffer (10 mM Tris, 50 mM KCl, 1 μg poly dI-dC, 1 mM DTT; 5% glycerol, pH 7.5) containing 20 fmol of labeled probe and 0 to 1.4 μM of purified ZccR. Samples were loaded onto 6% non-denaturing polyacrylamide gels and resolved at 100V for 1h in cold buffer (4˚C). Gels were transferred to BrightStar-Plus positively charged nylon membranes (Thermo-Fisher) and bands visualized using the Chemiluminescent Nucleic Acid Detection Module (Thermo-Fisher) following manufacturer's protocol. To determine binding specificity, 100X molar excess of either non-labeled specific competitor DNA (zccE-zccR IGR) or non-specific competitor DNA (206-bp S. mutans mntH promoter region [50]) was added to the reaction.

Oral colonization rat model
A modification of a rat caries model that we have previously followed [38] was used to determine the ability of the ΔzccE mutant strain to colonize rats fed a cariogenic diet while also testing the effects of topical Zn treatment on S. mutans colonization efficiency. Briefly, specific pathogen-free Sprague-Dawley rat pups were purchased with their dams from Envigo Laboratories and screened upon arrival to ensure an absence of mutans streptococci by plating oral

PLOS PATHOGENS
Streptococcus mutans zinc tolerance swabbings on mitis salivarius (MS) agar. Prior to infection, pups and dams received 0.8 mg ml -1 sulfamethoxazole and 0.16 mg ml -1 trimethoprim in the drinking water for 3 days to suppress endogenous flora and facilitate colonization by S. mutans. After 4 days of washout during which antibiotic-free water was provided, dams and pups aged 18 days were orally infected for four consecutive days with actively growing S. mutans UA159 or ΔzccE cultures by means of cotton swab. At the time of infection, the regular chow diet was replaced by a 12% sucrose cariogenic powdered diet (ENVIGO diet TD.190707). During the infection period, the animals were provided with 5% (wt/vol) sterile sucrose-water ad libitum, then fresh water for the remainder of the study. On the final day of infection, pups were weaned and randomly placed into experimental groups. Topical treatment with ZnSO 4 or saline solutions commenced after the fourth and final day of bacterial infection and lasted 10 consecutive days. Test (60 mM or 150 mM ZnSO 4 ) or control (saline) treatments were topically administered to the rat teeth with a camel's hair brush twice a day with a 6 h interval between treatments. At the end of the treatment period, animals were euthanized by CO 2 asphyxiation, and the lower jaws removed for bacterial burden determination. Jaw sonicates were subjected to 10-fold serial dilutions and plated on MS agar (to count S. mutans) and 5% sheep blood agar (to count total flora). The number of S. mutans recovered from the animals was expressed as CFU ml-1 of jaw sonicate, and S. mutans colonies counted on MS agar divided by the total CFU on blood agar to determine the percentage of S. mutans colonies recovered over the total flora. This study was reviewed and approved by the University of Florida Institutional Animal Care and Use Committee (protocol # 201810421). and ΔzccE strains grown on BHI agar containing 250 μM Mn with or without Zn supplementation and exposed to filter paper discs saturated with 0.25% H 2 O 2 . Data represent means and standard deviations of results from at least 3 independent experiments. (TIF) S1 Table. RNA-Seq analysis of S. mutans UA159 genes differentially expressed when grown in FMC treated with 4 mM ZnSO 4 for either 15 or 90 minutes and compared to untreated control. The genes whose expression changes over 2-fold applying False Discovery Rate (FDR) of 0.05 were mentioned and functional categories were listed. (XLSX) S2 Table. The table lists  Kajfasz.

PLOS PATHOGENS
Streptococcus mutans zinc tolerance