The AdcACB/AdcAII system is essential for zinc homeostasis and an important contributor of Enterococcus faecalis virulence

ABSTRACT Bacterial pathogens require a variety of micronutrients for growth, including trace metals such as iron, manganese, and zinc (Zn). Despite their relative abundance in host environments, access to these metals is severely restricted during infection due to host-mediated defense mechanisms collectively known as nutritional immunity. Despite a growing appreciation of the importance of Zn in host-pathogen interactions, the mechanisms of Zn homeostasis and the significance of Zn to the pathophysiology of E. faecalis, a major pathogen of nosocomial and community-associated infections, have not been thoroughly investigated. Here, we show that E. faecalis encoded ABC-type transporter AdcACB and an orphan substrate-binding lipoprotein AdcAII that work cooperatively to maintain Zn homeostasis. Simultaneous inactivation of adcA and adcAII or the entire adcACB operon led to a significant reduction in intracellular Zn under Zn-restricted conditions and heightened sensitivity to Zn-chelating agents including human calprotectin, aberrant cell morphology, and impaired fitness in serum ex vivo. Additionally, inactivation of adcACB and adcAII significantly reduced bacterial tolerance toward cell envelope-targeting antibiotics. Finally, we showed that the AdcACB/AdcAII system contributes to E. faecalis virulence in a Galleria mellonella invertebrate infection model and in two catheter-associated mouse infection models that recapitulate many of the host conditions associated with enterococcal human infections. Collectively, this report reveals that high-affinity Zn import is important for the pathogenesis of E. faecalis establishing the surface-associated AdcA and AdcAII lipoproteins as potential therapeutic targets.


Introduction
Iron (Fe), manganese (Mn), and zinc (Zn) are essential trace metals to all forms of life. They serve structural, catalytic, and regulatory functions to metalloproteins involved in a variety of biological processes [1][2][3][4]. As a result of this essentiality, hosts deploy a variety of strategies to deprive access of invading pathogens to trace metals, an active process termed nutritional immunity [2,[5][6][7][8]. To date, the best characterized nutritional immunity strategy is based on mobilization of metal-chelating proteins to the infection site by host immune cells [2]. Among them, calprotectin, a member of the S100 protein family produced by neutrophils and other types of immune cells that is secreted in large quantities during infection and inflammatory processes, is the main host protein responsible for Mn 2+ and Zn 2+ sequestration [9][10][11]. To overcome trace metal limitation, microbial pathogens evolved effective metalscavenging systems that include expression of surface-associated high-affinity metal uptake systems and, in some bacterial species, synthesis and trafficking of organic extracellular molecules known as metallophores [8].
Although the importance of Fe in host-pathogen interactions has been extensively examined [6], the role of Mn and Zn in host-pathogen interactions and the mechanisms utilized by bacteria to maintain their cellular levels and ratios properly balanced are less understood [7,[12][13][14][15][16][17]. The second most abundant trace metal in vertebrates, Zn, is estimated to be incorporated into approximately 5% of the bacterial proteome and plays structural and catalytic roles in multiple biological processes [18,19]. In bacteria, Zn acquisition under severe Zn-restricted conditions such as those that can be encountered in host environments depends on the activity of surface-associated Zn uptake systems from the ATP-binding cassette (ABC) transporter family (reviewed in Refs. [20][21][22]. Moreover, major human pathogens such as Pseudomonas aeruginosa and Staphylococcus aureus produce Zn-binding metallophores, also known as zincophores [23]. To date, the contributions of Zn uptake systems to virulence have been demonstrated in a number of bacterial species, including several Gram-positive pathogens that are phylogenetically related to Enterococcus faecalis, the subject organism of the present study. In S. aureus, inactivation of either the ABC-type transporter AdcABC, the staphylopine (Stp) zincophore, or its cognate multi-metal transporter CntABCDF was sufficient to impair bacterial growth under Zn-restricted conditions in vitro [24]. Loss of both AdcABC and Stp/CntABCDF systems resulted in further growth impairment under Zn-restricted conditions and attenuated virulence in a mouse retro-orbital infection model [24]. In streptococci, which to date reportedly do not synthesize zincophores, Zn acquisition is mediated by the ABC-type transporter AdcABC. In addition to AdcABC, streptococcal species encode an additional adcA homologue, known as adcAII (reviewed in [25]), coding for a second Zn-binding lipoprotein. In Streptococcus pyogenes, strains lacking adcC, adcA, or adcAII grew poorly in the presence of purified human calprotectin and displayed attenuated virulence in a necrotizing fasciitis mouse model [26] and in a humanized-plasminogen skin infection mouse model [27]. Moreover, S. pyogenes Δadc strains retained wildtype strain levels of virulence in calprotectin-negative (S100a9 −/− ) mice [26], which validates the central role of calprotectin in host-mediated Zn sequestration and protection against bacterial infection. Similarly, virulence of Streptococcus pneumoniae ΔadcAΔadcAII and Streptococcus agalactiae ΔadcAΔadcAIIΔlmb (lmb encodes for a 3 rd Zn-binding lipoprotein) strains was significantly attenuated in mouse models of systemic and nasopharyngeal colonization [16,28]. Finally, in the oral pathogen S. mutans, one of the few streptococci that do not encode the orphan adcAII gene, inactivation of the adcABC system significantly impaired bacterial growth under Zn-restricted conditions and reduced bacterial colonization of the dental biofilm in a rat model [12,29].
A commensal of the gastrointestinal (GI) tract, E. faecalis, is also a prevalent opportunistic pathogen of localized and systemic infections, including but not limited to infective endocarditis, catheter-associated urinary tract infections (CAUTI), and wound infections [30][31][32][33]. A major virulence trait of E. faecalis is its remarkable capacity to adapt to adverse conditions in the GI tract (their natural host environment) and several other host tissues, and to survive exposure to hospital-grade disinfectants and antibiotic treatments [34,35]. Because very little is known about the mechanisms of Zn homeostasis in enterococci, we sought to characterize the Zn acquisition systems of E. faecalis in this study. Similar to streptococci, the core genome of E. faecalis encodes for a conserved AdcACB system (originally annotated as znuACB) and an orphan substrate-binding lipoprotein AdcA-II that is annotated as adcA. In this report, we isolated a panel of E. faecalis ∆adc strains, including strains lacking every adc gene (∆adcACB∆adcAII) or both genes coding for the substrate-binding lipoproteins (∆adcA∆adcAII), and then used these mutants to define the role of AdcACB and AdcAII in E. faecalis pathophysiology. Our results revealed that simultaneous inactivation of adcA and adcAII or of the entire adcACB operon yielded the most impactful phenotypes, which included severe growth/survival defects in the presence of calprotectin or in human serum, and attenuated virulence in both invertebrate and vertebrate infection models. We also discovered that the inability to maintain Zn homeostasis diminished the recognized high tolerance of E. faecalis to antibiotics that target the cell envelope. Collectively, this study reveals that AdcACB and AdcAII work cooperatively to maintain E. faecalis Zn homeostasis during infection such that the surface-associated AdcA and AdcAII lipoproteins can be considered potential targets for the development of antimicrobial interventions.

AdcACB and AdcAII work in concert to promote growth under Zn-restricted conditions
Using the NCBI BLASTn tool, we identified the genes coding for the highly conserved ABC-type transporter AdcACB (OG1RF_RS00260-RS00270), the orphan substrate-binding AdcAII lipoprotein (OG1RF_RS12625) and the transcriptional repressor Zur (OG1RF_RS09465) in the Enterococcus faecalis OG1RF genome (GenBank: CP002621.1) (Figure 1a). The translated gene products of OG1RF_RS00260 (AdcA; Accession ID: AEA92738.1, protein ID: WP_002367576.1) and OG1RF_RS12625 (AdcAII; Accession ID: AEA95159.1, protein ID: WP_002392710.1) display respectively, 57% and 64% amino acid similarity to the S. pneumoniae AdcA and 42% and 39% similarity to S. pneumoniae AdcAII [16] ( Figure S1). Pairwise alignment between E. faecalis AdcA and AdcAII also revealed 53% similarity, indicative of functional redundancy ( Figure S1). AdcAII, the larger of the two Zn-binding lipoproteins of E. faecalis, contains a ZinT-like domain at the C-terminus that is also observed in the S. pneumonia AdcA and was shown to mediate Zn binding via the so-called trap door Growth characteristics of E. faecalis and its Zn-deficient mutants under Zn-restricted conditions. (a) Schematic of the gene locus of the Zn transport system in E. faecalis OG1RF core genome. Growth curves of E. faecalis wild type (WT) and its isogenic mutants in BHI (b), BHI supplemented with 100 µmZnSo 4 (c), 10 µm TPEN (d), combination of TPEN and ZnSo 4 (e), TPEN and MnSo 4 (f) or TPEN and Fe SO 4 (g). In (b) and (d-f), data points represent the average of nine biological replicates. Finally, the growth curve of E. faecalis wild type (WT) and genetically complemented ∆adc mutants in BHI supplemented with 10 µm TPEN (h). In (c) and (g), data points represent the average of six biological replicates. Error bar represents the standard error of margin (SEM). Statistical analysis was performed using simple linear regression ofthe exponential growth phase, and slope of each mutant's growth kinetics was compared with that of the parent strain.
To probe the role of AdcACB and AdcAII in Zn acquisition and, most importantly, determine the significance of these systems in E. faecalis pathophysiology, we used a markerless in-frame deletion strategy [42] to generate strains lacking one or both substratebinding lipoproteins (∆adcA, ∆adcAII, and ∆adcA∆adcAII) and the entire adcACB operon alone or in combination with adcAII (∆adcACB and ∆adcACB∆adcAII). In addition, we isolated a strain lacking the transcriptional repressor Zur (∆zur). Next, we compared the ability of E. faecalis OG1RF (wildtype strain) and mutant derivatives to grow in BHI, a complex media that contains ~10 µM Zn [43], or in BHI supplemented with TPEN [44]. In BHI, inactivation of either adcA or adcAII alone did not impact growth kinetics or growth rates, whereas simultaneous inactivation of adcA and adcAII (∆adcA∆adcAII) or the entire adcACB operon alone (∆adcACB) or in combination with adcAII (∆adcACB∆adcAII) resulted in slower growth rates without affecting final growth yields ( Figure 1b). Inactivation of the zur regulator did not affect growth rates but led to a slight increase in the final growth yield in BHI (Figure 1b). BHI supplementation with 100 µM ZnSO 4 (10-fold in excess of labile Zn pools in BHI media) restored the growth defects of ∆adcA∆adcAII, ∆adcACB, and ∆adcACB∆adcAII strains (Figure 1c). Addition of 10 µM TPEN to BHI (BHI+TPEN) minimally impacted the growth of the parent OG1RF strain (OG1RF growth is severely impaired at TPEN concentrations ≥20 µM, Figure S3). On the other hand, the ∆adcAII single mutant grew poorly in BHI+TPEN, while growth of the ∆adcACB and ∆adcACB∆adcAII strains was completely inhibited by 10 µM TPEN (Figure 1d). Finally, the addition of 10 µM ZnSO 4 to the BHI +TPEN media restored growth of the ∆adcAII, ∆adcACB, and ∆adcACB∆adcAII strains (Figure 1e). To verify if the inhibitory effect of TPEN on growth of ∆adc strains was indeed Zn-specific, we tested if addition of Mn (10 µM MnSO 4 ) or Fe (10 µM FeSO 4 ) could also restore cell growth. With the exception of ∆adcAII that was able to grow in BHI+TPEN after Mn supplementation, Fe or Mn supplementation did not restore growth of the other mutants in BHI+TPEN (Figure 1f-g, compared to Figure 1d). In trans complementation of ∆adcACB and of ∆adcAII fully or partially rescued their growth defects in BHI+TPEN ( Figure 1h). The reasons for Mn rescuing growth of ∆adcAII in BHI + TPEN and the partial complementation of ∆adcAII are at present unknown.
Next, we sought to determine the ability of our panel of ∆adc strains to grow in the presence of human calprotectin, a potent Mn and Zn chelator. We carried out growth kinetic assays to compare the ability of wild-type (WT) and ∆adc mutants to grow in BHI supplemented with purified human calprotectin (hCP) or recombinant calprotectin (hCP∆ Mn-tail ) defective in Mn sequestration [10]. In the presence of hCP, growth of ∆adcA was not significantly different when compared to WT, whereas ∆adcAII, ∆adcA∆adcAII, ∆adcACB, and ∆adcACB∆adcAII mutants displayed reduced growth or were fully inhibited by the native version of calprotectin (Figure 2a-b). While inactivation of the Mn-binding residue in hCP ∆Mn-tail improved growth of WT, ∆adcA and ∆adcAII strains, the other mutants remained highly sensitive to hCP ∆Mn-tail (Figure 2c). In trans complementation fully rescued the growth defects of ∆adcACB mutant in the presence of both versions of calprotectin, whereas growth of the complemented ∆adcAII was partially rescued ( Figure S4). The reason for the partial complementation of ∆adcAII in the presence of hCP is at present unknown. Taken together, these findings reveal that AdcABC and AdcAII work independently but cooperatively to mediate E. faecalis growth under Zn-restricted conditions. Based on the identical phenotypes of ∆adcACB and ∆adcA∆adcAII strains, these results also indicate that both AdcA and AdcAII associate with AdcB (inner membrane permease) and AdcC (cytoplasmic ATPase) to form tripartite Zn transporters. Finally, growth kinetics in the presence of the Zn-chelating agents TPEN and calprotectin hint that AdcAII might be a more effective Zn scavenger than AdcA, at least under the more severe Zn-restricted conditions.
Next, we used inductively coupled optical emission spectrometry (ICP-OES) to determine intracellular Zn pools in mid-log grown cultures of WT and derivative ∆adc strains grown in BHI or in BHI supplemented with 7.5 µM TPEN ( Figure 3a). In BHI, the ∆adcACB accumulated less Zn when compared to the WT strain albeit this difference was rather small (~12%) and not supported by a similar or greater decrease in Zn pools in the ∆adcA∆adcAII and ∆adcACB∆adcAII strains  ( Figure 3a). In agreement with the predicted role of Zur as a transcriptional repressor of adcABC and adcAII, the ∆zur strain accumulated two times more Zn than the WT strain when grown in BHI. The addition of TPEN to the growth media led to an unexpected increase in intracellular Zn pools in the WT strain when compared to cells grown in BHI (~50% increase). Nonetheless, all mutants accumulated less Zn when compared to the WT strain when grown in BHI +TPEN. We suspected that the higher intracellular levels of Zn in the WT strain grown in BHI+TPEN compared to BHI only correlated with increased transcription of the adcACB and adcAII genes and that this response was controlled by Zur. To verify this possibility, we used quantitative RT-PCR to determine mRNA levels of adcA and adcAII in the WT and ∆zur strains grown to mid-log phase in BHI and then treated with either 30 µM TPEN or 4 mM ZnSO 4 for 1 h. In line with previous transcriptional studies [40,41], TPEN treatment significantly induced adcA (~1-log) and adcAII (~2-log) transcription, whereas Zn supplementation reduced adcA levels by ~2-log and adcAII by ~1-log when compared to the BHI control ( Figure 3b). As expected, inactivation of zur resulted in increased transcription of adcA and adcAII grown in BHI (~1-log adcA, ~2-log adcAII) or BHI+Zn (~2-log both genes) when compared to the WT strain grown under the same conditions. Taken together, these results confirm that Zn deprivation (TPEN-treated cells) alleviates Zur repression triggering a strong induction of adcACB and adcAII expression that allows E. faecalis overcome Zn starvation.

AdcACB and AdcAII contribute to growth in serum but not in urine ex vivo
To determine the contribution of Adc-mediated Zn uptake to E. faecalis virulence, we first monitored the ability of the ∆adc strains to grow and survive in pooled human serum or human urine ex vivo. When incubated in serum, growth of ∆adcA and ∆adcAII strains did not significantly differ from WT, whereas ∆adcA∆adcAII, ∆adcACB and ∆adcACB∆adcAII grew poorly and, most relevantly, displayed sharp decreases in survival after 8 hours and onward, ultimately showing a ~3-log reduction in colony-forming unit (CFU) recovered after 48 hours of incubation in serum (Figure 4a). These growth and survival defects were fully reversed by the addition of 500 µM ZnSO 4 (Figure 4b) or in trans complementation (Figure 4c). On the other hand, the ability of all ∆adc strains to grow/survive in urine was not found to differ from WT (Figure 4d), suggesting that Zn is not a growth-limiting factor in urine (at least ex vivo).
Finally, the inactivation of zur did not impact growth nor survival in serum or urine (Figure 4a-b,d).
While Zn levels and bioavailability in serum or in urine were not determined, the strong phenotype of the ∆adcA∆adcAII, ∆adcACB and ∆adcACB∆adcA strains in serum was expected as Zn levels in blood circulation are low with most Zn sources bound or sequestered by host cells and proteins [45,46]. On the other hand, Zn is abundant in the bladder environment as any excess Zn, typically from dietary sources, is excreted through urine via the gastrointestinal route [47][48][49][50].

Disruption of AdcABC-AdcAII lowers tolerance toward cell envelope-targeting antibiotics
Because the ∆adcA∆adcAII, ∆adcACB and ∆adcACB∆adcAII strains displayed a reduced growth rate under Zn-restricted conditions, like S. pneumoniae ΔadcAΔadcAII [16], we wondered if this was due to altered cell division. To investigate this, we observed bacterial morphology using a light microscope. Indeed, these mutants formed longer chains when compared to the WT, ∆adcA and ∆adcAII strains that primarily formed only short chains or diplococcus ( Figure 5). This observation and the fact that S. pneumoniae ΔadcAΔadcAII mutant displayed aberrant cell septation [16] led us to wonder if expression of virulence traits that occur at the cell surface interface were similarly affected in the mutant strains. First, we compared the capacity of WT and mutants to form biofilms after 24 hours of incubation in BHI supplemented with 10 mM glucose. The total biofilm biomass of ∆adcA and ∆adcAII single mutants was significantly reduced when compared to WT, but the very small differences observed (5 to 10% reduction) are unlikely to have major biological implications (Figure 6a). On the other hand, the ∆adcA∆adcAII, ∆adcACB and ∆adcACB∆adcAII strains formed more robust biofilms with an ~30 to 50% increase in biofilm biomass ( Figure 6a).
Next, we tested the capacity of WT and mutants to grow in BHI supplemented with antibiotics that target different steps of cell wall biosynthesis (ampicillin, bacitracin, and vancomycin) or disrupt membrane integrity (daptomycin) by determining the minimal inhibitory concentration (MIC) for each antibiotic. Once again, relevant phenotypes were restricted to the ∆adcA∆adcAII, ∆adcACB, and ∆adcACB∆adcAII strains. Specifically, ∆adcA∆adcAII, ∆adcACB, and ∆adcACB∆adcAII showed lower MICs for ampicillin, bacitracin, and daptomycin (Figure 6b-d). However, WT and all mutants showed the same MIC for vancomycin (Figure 6e).

Zn is critical for E. faecalis virulence during infection
In the last series of experiments, we used three in vivo models to probe the contributions of AdcABC and AdcAII to virulence. In conformity with in vitro and ex vivo phenotypes, virulence of ∆adcA∆adcAII, ∆adcACB and ∆adcACB∆adcAII but not the single mutants (∆adcA, ∆adcAII and ∆zur) was highly attenuated in the Galleria mellonella model (Figure 7a). Next, we used two catheterassociated mouse infection models that recapitulate some of the environmental and immunological conditions that promote enterococcal infections in human. From this point, we did not include the ∆adcA∆adcAII and ∆adcACB strains as these mutants consistently phenocopied the ∆adcACB∆adcAII mutant. In the catheterassociated peritonitis model, the ∆adcA and ∆adcAII single mutants colonized the peritoneal cavity, catheter and spleen (infection becomes systemic after 12 to 24 h) in significantly fewer numbers than the WT strain ( Figure 7b). Not surprisingly, colonization defects of single mutants were significantly more pronounced in the ∆adcACB∆adcAII strain. Finally, in a catheter-associated urinary tract infection (CAUTI) model, virulence of ∆adcA and ∆adcACB∆adcAII strains was attenuated, but bacterial burden recovered from bladders of retrieved catheter of animals infected with WT or ∆adcAII strains were nearly identical (Figure 7c). Moreover, ∆adcAII was recovered from kidneys and spleen in higher bacterial titers, albeit the trend of the latter was not considered significant (Figure 7c).

Discussion
While there has been increasing appreciation of the multiple contributions of trace metals other than Fe to bacterial fitness and virulence [2,5,20], the mechanisms utilized by E. faecalis to maintain Zn homeostasis and their specific contributions to pathogenesis were, until now, poorly understood. Previously, our group showed that E. faecalis encodes three high-affinity Mn transporters and that while it was necessary to inactivate all three Mn transport systems (efaCBA, mntH1 and mnthH2) to severely impair Mn uptake in vitro, the inactivation of only two of them (efaCBA and mntH2) was sufficient to abolish E. faecalis virulence [51]. In this report, we showed that both AdcACB and the orphan AdcAII predicted to mediate Zn import in a Zur-regulated manner and are critical for the ability of E. faecalis to grow and survive Zn-restricted conditions. Although the differences were not as striking as those seen with the Mn transport mutants [51], virulence of strains lacking the AdcACB/AdcAII system was significantly attenuated in both invertebrate and vertebrate infection models.
During characterization of the adc mutants, we noted that simultaneous deletion of adcACB and adcAII led to readily discernible morphological and biophysical alterations, that is increased cell chaining and cell-cell aggregation, which led us to wonder if the inability to maintain Zn homeostasis affected cell envelope homeostasis. Indeed, the ∆adcA∆adcAII, ∆adcACB, and ∆adcACB∆adcAII strains showed heightened sensitivity to ampicillin, bacitracin, and daptomycin, with the latter showing a striking 16-fold lower MIC than the WT strain MIC (≤16 μg ml −1 ml, compared to 256 μg ml −1 ). Previous studies have shown that exposure to bacitracin resulted in downregulation of adcA (0.2-fold) in E. faecalis V583, whereas adcAII transcription increased (8-fold) after vancomycin treatment [52]. In addition, exposure to chlorhexidine, a cationic antimicrobial agent that targets the cell membrane, resulted in upregulation of adcCB (~4-fold) [53]. While the roles played by Zn in E. faecalis envelope homeostasis are unknown, previous studies have associated loss of Zn transporters or Zn-dependent enzymes to surfaceassociated defective phenotypes. For example, in S. pneumoniae, deletion of adcACB and adcAII resulted in asymmetrical septa formation, abnormal cell division patterns, and emergence of small, aborted cells when S. pneumoniae was forced to grown under Zn-restricted conditions [16]. In the distantly related Gram-negative pathogen Acinetobacter baumannii, inactivation of a Zn-dependent peptidase, ZrlA, increased cell permeability and susceptibility to the β-lactam antibiotic carbenicillin [54].
Similar to streptococci, enterococcal genomes do not encode the machinery to synthesize opine-like zincophores and their cognate transporters. In Grampositive cocci, Zn acquisition is primarily mediated by adcABC (also known as znuABC) and adcAII, both of which are under AdcR (ZuR) negative control [16,22,26,29,55]. Similar to our findings showing the additive contribution of E. faecalis adcA and adcAII genes to Zn homeostasis, simultaneous inactivation of both adcA and adcAII is necessary to (nearly) abolish Zn import and, as a result, drastically impairs virulence of major human pathogens such as S. pyogenes and S. pneumoniae [16,26,27].
Although E. faecalis AdcA and AdcAII share 52% identity, they do not have a similar domain organization. Biochemical and biophysical characterizations of S. pneumoniae AdcA and AdcAII proteins indicated that these functionally redundant proteins employ distinct Zn acquisition mechanisms [17]. AdcAII Spn has two Zn-binding domains, an amino terminal cluster A-I domain typical of solute-binding proteins and a C-terminal domain that is structurally related to ZinT, a periplasmic Zn chaperone of Gram-negative bacteria, whereas AdcA Spn has only the terminal cluster A-I domain [17]. Moreover, AdcAII Spn -mediated Zn uptake in vivo has been shown to depend on proteins  (c) Total CFU recovered after 24 hours from bladder, kidney, spleen, and catheter of mice infected with 1 × 10 7 CFU of WT or indicated mutants. In (b and c), ten mice were infected with two biological replicates and data points shown were a result of using the ROUT outlier test. The black line represents the median. Statistical analysis was performed using the Mann-Whitney test. * p ≤0.05, ** p ≤0.01, *** p ≤0.001, and **** p ≤0.0001. The dashed line represents the limit of detection (LOD = 50 CFUs).
with poly-histidine (Pht) triad (HxxHxH) motifs that scavenge extracellular Zn and then transfer it to AdcAII Spn for internalization [56]. While streptococcal species have been shown to encode as many as four Pht protein homologues (e.g., S. pneumoniae PhtA, PhtB, PhtD and PhtE), E. faecalis genomes do not contain genes with HxxHxH motifs. Of interest, a small openreading frame (OG1RF_RS12620) (40 amino acids) coding for a putative uncharacterized protein is located upstream and separated by only 21-bp from the adcAII Ef start codon. While the predicted amino acid sequence of the OG1RF_RS12620 contains only a single histidine residue and is not enriched for other amino acids (such as cysteine and methionine) that typically coordinate Zn, it will be interesting to explore the possible role of OG1RF_RS12620 in Zn acquisition in future studies.
Because an E. faecalis strain lacking the entire adcACB operon is phenotypically similar to the ∆adcA∆adcAII double mutant, an important observation from this study is that, most likely, AdcA and AdcAII can only form functional complexes with AdcB (inner membrane permease) and AdcC (cytoplasmic ATPase). Also of interest was the distinct importance of AdcA and AdcAII to colonization and systemic dissemination in the two mouse models. While attenuated virulence of ∆adcA and ∆adcAII single mutants was comparable in the peritonitis model, adcAII was dispensable for bladder and catheter colonization in the CAUTI model. In addition, the recovery of viable bacteria from kidney and spleen of mice infected with ∆adcAII indicates, at first glance, that the loss of AdcAII promotes bacterial dissemination to kidneys and spleen. While speculative at this point, we believe that these phenotypes are due to differences in the expression levels or activity of AdcA and AdcAII in the bladder environment. One possibility is that adcA responds more strongly than adcAII to environmental cues encountered in the bladder, such that adcAII becomes dispensable for E. faecalis proliferation in urine. To test this possibility, studies to compare the adc transcriptional profiles of ∆adcA and ∆adcAII in the bladder, peritoneal cavity, and bloodstream environments will soon be underway. Alternatively, the large and constant fluctuations in the bladder environment in solute concentrations and of other important biophysical and biochemical parameters such as pH caused by intermittent cycles of urination can somehow compromise Zn-binding capacity of AdcAII or ability to interact with the AdcBC partner proteins.
In summary, our findings reveal that the AdcACB/ AdcAII system is a bona fide Zn acquisition system of E. faecalis contributing additively to the maintenance of Zn homeostasis. More importantly, this report demonstrates that the AdcACB/AdcAII system mediates E. faecalis virulence. Therefore, both the substratebinding and surface-associated AdcA and AdcAII proteins can be viewed as suitable targets for the development of antimicrobial therapies to treat or prevent enterococcal infections.

Bacterial strains and growth conditions
The bacterial strains and vectors used in this study are listed in Table 1. Bacteria were routinely grown in brain heart infusion (BHI broth (BD Difco TM , for E. faecalis) and Luria-bertani (BD Difco TM , for E. coli) at 37°C under static conditions. Strains possessing the pGCP123 plasmid [59] were grown in the presence of kanamycin (300 µg ml −1 for E. coli and 500 µg ml −1 for E. faecalis). For growth kinetics assays, overnight cultures were normalized by cell density to an OD 600 of 0.25 and inoculated into BHI media at a 1:50 ratio, with the OD 600 monitored in an automated growth reader (Bioscreen c, Oy Growth Curves AB). Native calprotectin (hCP) and Mn-deficient (Cp ∆Mn-tail ) calprotectin were gifts from Dr. Walter Chazin (Vanderbilt University, USA) [10]. Experiments using purified calprotectin were performed in BHI supplemented with 20% (v/v) CP buffer (40 mM NaCl, 0.5 mM βmercapethanol, 1.2 mM CaCl 2 , 8 mM Tris-HCl, pH 7.5). Normalization of starter cultures for most experiments included centrifugation or overnight cultures at 4000 rpm for 10 mins to remove spent media, two cell  Primers for qRT-PCR  adcA_F'  TCGTGGAAGCCAGTCAATCC  -adcA_R'  TTTGCGCTAACACGGGATCT  -adcA-II_F'  TTCTGGTCGATGGCAAACCA  -adcA-II_R'  GTTGGTGCGATGTTGTGGTC  -Primers for cloning in pCJK47 vector  adcA_Arm1 F'  TCGCTAGTTCTAGAGCGGCCGCCTTATTAAAGCGTACCTATAATATT  NotI  adcA_Arm1 R'  TGTTGCCAGAGTTCTGTCCTCTTTTCGATTT  -adcA_Arm2 F'  AGGACAGAACTCTGGCAACAGAAAGGAGCCAA  -adcA_Arm2 R' CTTAGCATGCCATGGTACCCGGGTTTTTTAAATTTTTCCCACATTTCT CGCTCTAGAACTAGTGGATCTTAGTTTCTTTGCATTTTTTGTTTT a Restriction sites are underlined and bold in the primer sequence.  [60,61], and image files (PDB) were constructed using ChimeraX1.3 [62].

General cloning techniques
Bacterial genomic DNA (gDNA) was isolated using a Wizard Genomic DNA purification kit (Promega). Plasmid purification was performed using the Monarch plasmid miniprep kit (New England BioLabs). Isolation of PCR amplified products was performed using the Monarch DNA gel extraction kit (New England BioLabs). Typical cloning, either directional or non-directional depending on plasmid used, was performed using the In-Fusion HD cloning kit (TaKaRa Bio). Colony PCR was performed using PCR 2x Master Mix (Promega) with primers listed in Table 2.

Construction of deletion and genetically complemented strains
Deletion of zur, adcA, adcAII, or the entire adcACB operon was carried out using a markerless genetic exchange system based on the pCJK47 vector [42]. Briefly, nucleotide sequences flanking zur, adcA, adcCB, and adcAII were amplified using the primers listed in Table 2. These PCR-amplified products were directly cloned into pcJK47, electroporated into E. faecalis CK111 (donor strain), followed by conjugation into E. faecalis OG1RF. The deletion mutants were isolated by following the steps for markerless counterselection detailed elsewhere [42]. The ∆adcA∆adcAII and ∆adcACB∆adcAII double mutants were obtained by conjugating the pCJK-adcAII plasmid with the ∆adcA and ∆adcACB mutants, respectively. Because the adcACB operon is ~3kb, we first isolated ∆adcA and then performed conjugation using the pCJK-adcBC vector to generate the ∆adcACB strain. PCR sequencing was performed to confirm the absence of zur, adcA, adcAII, or the entire adcACB operon in these deletion mutant strains. For genetic complementation, adcACB and adcAII genes were amplified by PCR and cloned into the pGCP123 plasmid [59] using the primers listed in Table 2. The full-length adcACB and adcAII nucleotide sequences were incorporated into linearized pGCP123 plasmid using the In-Fusion HD cloning kit. Plasmids were propagated in E. coli and transformed into E. faecalis as described previously [59].

Inductively coupled plasma-optical emission spectrometry (ICP-OES)
The metal (Zn) concentration in BHI media and in bacteria was determined as previously described [51]. For quantification of metals in BHI, 9 ml of broth was digested with 1 ml of trace-metal grade 35% nitric acid (HNO 3 ) prior to analysis. For quantification of intracellular metal, overnight E. faecalis cultures were washed with PBS twice and inoculated at a ratio of 1:40 in BHI or BHI supplemented with 7.5 µM TPEN. Mid-log grown cultures (OD 600 of 0.5) were harvested by centrifugation and washed twice with PBS containing 0.5 mM EDTA. After washing, cell pellets were collected in a polyethylene scintillation vial (Fisher Scientific) and digested in 1 ml of trace-metal grade 35% nitric acid (HNO 3 ) at 90°C for 1 hour. Digested bacterial cells were diluted at a ratio of 1:10 in reagentgrade water prior to analysis. Intracellular Zn pools were then quantified using a 5300DV ICP Atomic Emission Spectrometer (Perkin Elmer) at the University of Florida Institute of Food and Agricultural Sciences (UF-IFAS) Analytical Services Laboratories. The bicinchoninic acid (BCA) assay kit (Pierce TM ) was used to calculate the total protein content for normalization of the metal concentration obtained.

MIC determinations
Overnight E. faecalis cultures were normalized to an OD 600 of 0.25 and diluted at a ratio of 1:1000. The diluted cultures were inoculated at a ratio of 1:20 into BHI containing antibiotics (ampicillin, bacitracin, daptomycin, and vancomycin). The absorbance at OD 600 after incubation at 37°C for 24 hours was measured using a Synergy H1 microplate reader (Molecular Devices).

Growth and survival in serum and urine
Inoculum was prepared from overnight E. faecalis cultures adjusted to OD 600 of 0.5 and inoculated into pooled human serum or human urine (Lee Biosolutions) at a ratio of 1:1000. Aliquots were obtained at several time points and then plated on BHI agar containing fusidic acid and rifampicin to determine CFU.

Biofilm assay
Inoculum was prepared from overnight E. faecalis cultures adjusted to OD 600 of 0.5 that were diluted 1:25 in BHI supplemented with 10 mM glucose. Biofilm assay was performed using 96-well polystyrene plates (Grenier) that were incubated at 37°C for 24 hours. Post-incubation, spent media was discarded, biofilms washed twice with PBS, and the biomass stained with 0.1% crystal violet for 25 mins. A 33% acetic acid solution was used to dissolve the precipitated crystal violetstained biomass, and absorbance was determined at OD 595 .

Quantitative real-time PCR
Overnight E. faecalis cultures were normalized to an OD 600 of 0.5, inoculated at a ratio of 1:20 into fresh BHI, and incubated for 1 hour at 37°C. After incubation, cells were collected by centrifugation at 4000 rpm for 10 mins, washed with PBS, and incubated in the presence of lysozyme (20 mg ml −1 ) at 37°C for 30 mins. After treatment, cells were harvested by centrifugation and the total RNA was extracted using the PureLink RNA minikit (Invitrogen). Next, a Turbo DNA-free kit (Thermo Fisher) was used for purification of RNA and removal of contaminating gDNA. A High-capacity cDNA Reverse Transcription Kit (Applied Biosystems) was used for the synthesis of cDNA. Quantitative real-time PCR (qRT-PCR) was performed using iTaq Universal SYBR supermix (BioRad) with the primers listed in Table 2. For quantification of transcript numbers, E. faecalis OG1RF gDNA was used as a template to generate standard curves.

Galleria mellonella infection
To assess E. faecalis virulence, the larvae of G. mellonella were used as described previously [51,63]. Briefly, larvae (groups of 15) were injected with exponentially grown cells (~5 × 10 5 CFU), heat-killed E. faecalis (30 mins at 100°C negative control), or PBS (vehicle control). Post-injection, larvae were kept at 37°C and their survival were recorded over time.

Catheter-Associated peritonitis mouse model
The methods for the catheter-associated peritonitis model have been described previously [64], and hence, only a brief overview describing minor modifications are described here. Female C57BL/6J 8-weeks old mice were purchased from Jackson Laboratories and allowed to acclimate for at least 2 days prior to initiation of the study. Three hours post-catheter implantation, mice were injected intra-peritoneally with ~5 × 10 8 CFU of bacteria from an overnight culture. Forty eight hours post-injection, mice were euthanized and the bacterial burden was determined by plating serially diluted aliquots on selective BHI plates supplemented with rifampicin and fusidic acid. This procedure was approved and performed in compliance with the University of Florida Institutional Animal Care and Use Committee (protocol# 201910705).

Mouse catheter-associated urinary tract infection (CAUTI) model
The full description of the method for the CAUTI infection model has also been described in detail previously [65], and hence, a brief overview and modifications are described here. Female C57BL/6Ncr 6-weeks old mice were purchased from Charles River Laboratories and subjected to transurethral catheter implantation. Post-catheter implantation, mice were infected with ~1 × 10 7 CFU of bacteria from an overnight culture. 24 hours post-infection, mice were euthanized, and bacterial burden were determined by plating on selective BHI plates supplemented with rifampicin and fusidic acid. This procedure follows the University of Notre Dame Institutional Animal Care and Use Committee (protocol #18 August 4792MD).

Bright field microscopy
Overnight cultures grown in BHI were normalized to OD 600 0.5 (~5×10 8 CFU/ml) and washed once with 1 ml of PBS. Aliquot of 5 µl of bacterial inoculum was placed on microscope glass sides that were washed once in filtered 70% ethanol followed by milliQ water and left to dry at room temperature. Samples were covered with 5 µl of mounting media (Vectashield) and covered with a glass coverslip. Bright field microscopy was performed using a Leica DM2500 LED optical microscope fitted with a 100X/1.3 oil objective lens. Images acquired were further processed using FIJI software [66].

Statistical analysis
Data obtained from this study were analyzed using GraphPad Prism 9.0 software (GraphPad Software, San Diego, CA, USA). Data from multiple experiments conducted on nonconsecutive days were collated, and applicable statistical tests were used.

Data availability statement
The authors confirm that the data supporting the findings of this study are available within the article and/or its supplementary materials. Vectors and strains created from this study will be available from the corresponding author upon reasonable request.

Disclosure statement
No potential conflict of interest was reported by the author(s).