Identification of Multiple Iron Uptake Mechanisms in Enterococcus faecalis and Their Relationship to Virulence

ABSTRACT

Among the unfavorable conditions bacteria encounter within the host is restricted access to essential trace metals such as iron. To overcome iron deficiency, bacteria deploy multiple strategies to scavenge iron from host tissues, with abundant examples of iron acquisition systems being implicated in bacterial pathogenesis. Yet the mechanisms utilized by the major nosocomial pathogen Enterococcus faecalis to maintain intracellular iron balance are poorly understood. In this study, we conducted a systematic investigation to identify and characterize the iron acquisition mechanisms of E. faecalis and to determine their contribution to virulence. Bioinformatic analysis and literature surveys revealed that E. faecalis possesses three conserved iron uptake systems. Through transcriptomics, we discovered two novel ABC-type transporters that mediate iron uptake. While inactivation of a single transporter had minimal impact on the ability of E. faecalis to maintain iron homeostasis, inactivation of all five systems (Δ5Fe strain) disrupted intracellular iron homeostasis and considerably impaired cell growth under iron deficiency. Virulence of the Δ5Fe strain was generally impaired in different animal models but showed niche-specific variations in mouse models, leading us to suspect that heme can serve as an iron source to E. faecalis during mammalian infections. Indeed, heme supplementation restored growth of Δ5Fe under iron depletion and virulence in an invertebrate infection model. This study revealed that the collective contribution of five iron transporters promotes E. faecalis virulence and that the ability to acquire and utilize heme as an iron source is critical to the systemic dissemination of E. faecalis.

INTRODUCTION

A resident of the gastrointestinal tract of animals and humans, Enterococcus faecalis is also a major opportunistic pathogen causing, for example, central line-associated bloodstream infections (CLABSI), infective endocarditis, catheter-associated urinary tract infections (CAUTI), and wound infections (1). Over the past several decades, the haphazard prescription of antibiotics combined with the intrinsic hardy nature of E. faecalis, including natural and acquired resistance to antibiotics, has contributed to a sustained and often increased presence of enterococcal infection outbreaks in health care settings or in the community (2). E. faecalis is generally considered a low-grade pathogen due to the limited number of tissue-damaging factors encoded in its core genome; its virulence potential is thought to derive from a capacity to form robust biofilms on tissues or on indwelling devices, to thrive under a variety of adverse environmental conditions, and to subvert the immune system (3–5). Therefore, a better understanding of the mechanisms utilized by E. faecalis to survive under unfavorable conditions, especially those encountered within the human host, can potentially provide new therapeutic leads.

Among the adverse conditions pathogens encounter during infection is limited access to essential trace metals, in particular iron, manganese, and zinc, that are actively sequestered by metal-binding host proteins as part of an antimicrobial process known as nutritional immunity (6–10). Iron is of particular significance because it is the preferred metal cofactor of enzymes that carry out fundamental cellular processes such that it plays a central role in host-pathogen interactions (6, 11, 12). Despite being the most abundant trace metal in vertebrate tissues, iron is not readily available to bacterial pathogens because the vast majority of this element found in the host is complexed to heme inside red blood cells or bound to ferritin, an intracellular protein produced in hepatocytes that serves as the principal iron storage protein in mammalian cells (13). In addition, several host-produced proteins avidly bind free iron either to avoid iron toxicity to host tissues or as part of the nutritional immunity process (12, 14, 15). For instance, the liver produces and secretes transferrin (TF), which binds free Fe³⁺ in the bloodstream and at sites of infection, which is then recycled by macrophages by unloading iron to intracellular ferritin and returning apo-TF into circulation (13). At mucosal surfaces, free iron is sequestered by lactoferrin, which is also found in high concentrations in human secretions such as saliva (16). While primarily known for its role in manganese and zinc sequestration, neutrophil-secreted calprotectin has been shown to efficiently chelate Fe²⁺ in anaerobic environments in vivo (9, 17). All these factors combined with the low solubility of Fe³⁺ in serum make free iron concentrations within vertebrates several orders of magnitude below the concentration range required for microbial growth (12, 18, 19).

To overcome host-imposed iron starvation, bacterial pathogens deploy multiple strategies to scavenge free iron directly, bound to organic molecules (such as heme) within hemoproteins, or mobilized to iron-binding proteins (6, 10, 20–22). Perhaps the most effective strategy utilized by bacteria to scavenge iron is the production of siderophores (“iron carrier” from the Greek), which are low-molecular-mass organic molecules that are among the strongest metal chelators known to date (23, 24). While not all bacteria synthesize siderophores, high-affinity surface-associated iron transporters are ubiquitous in bacteria, with some of the most successful bloodborne pathogens encoding at least one dedicated heme acquisition system in addition to elemental iron transporters (19, 21, 25, 26). Not surprisingly, many of the genes associated with siderophore biosynthesis and uptake as well as iron and heme transporters have been directly implicated in bacterial virulence (6, 25, 27–30). In recent years, our group identified and characterized the manganese and zinc import systems of E. faecalis, showing that the well-coordinated activity of either manganese (EfaCBA, MntH1, and MntH2) or zinc (AdcABC and AdcAII) transporters is critical to E. faecalis fitness and virulence (31, 32). However, when it comes to the mechanisms utilized by the enterococci to maintain iron homeostasis and its relationship to enterococcal pathogenesis, current knowledge is restricted to in silico and transcriptome-based studies showing that E. faecalis encodes three highly conserved iron import systems, namely, EfaCBA, FeoAB, and FhuDCBG, that are regulated by either the DtxR-like/EfaR repressor (EfaCBA) or the Fur-like repressor (FeoAB and FhuDCBG) (33–35). While our group has shown that EfaCBA is a major dual manganese and iron transporter (31), the species of iron transported by EfaCBA is unknown. However, the FhuDCBG operon encodes a known ferrichrome transporter, while FeoAB is a highly conserved Fe⁺² transport system. To fill the many knowledge gaps on iron homeostasis mechanisms and their contribution to enterococcal pathophysiology, we sought to identify and characterize the mechanisms utilized by E. faecalis to overcome iron starvation and determine the individual and collective contributions of iron uptake systems to E. faecalis virulence. Through transcriptomics, we identified two additional and previously uncharacterized ABC-type iron transporters that appear to be restricted to enterococci and a small number of streptococcal species. We named the novel iron transporters FitABCD and EmtABC and generated strains lacking one or both transporters using the ΔfitAB ΔemtB double mutant as the background to generate a quintuple mutant also lacking efaCBA, feoAB, and fhuDCBG (Δ5Fe strain). Characterization of these mutant strains revealed that E. faecalis indeed utilizes multiple iron transporters to acquire iron under iron-depleted conditions and that their collective activity is important for enterococcal pathogenesis in a niche-dependent manner. In addition, evidence that E. faecalis can utilize heme as an alternative iron source and that an unidentified heme transporter(s) might be critical for systemic dissemination and disease outcome is also provided.

RESULTS

Two uncharacterized ABC-type transporters are the most upregulated genes in E. faecalis OG1RF grown under iron-depleted conditions.

To identify the genes and pathways utilized by E. faecalis to grow under iron starvation, we used RNA deep sequencing (RNA-seq) to compare the transcriptome of the parent strain OG1RF grown to mid-log phase in the chemically defined FMC medium (31, 36) with and without the addition of FeSO₄ as an iron source (see Table S1 in the supplemental material). Despite the ~1,600-fold difference in iron contents of the two medium formulations (~80 μM total iron in FMC[+Fe] compared to ~0.05 μM total iron in FMC[−Fe] [Table 1]), the abilities of different E. faecalis and Enterococcus faecium strains to grow under iron-replete or iron-depleted conditions were remarkably similar (Fig. 1). Moreover, quantification of intracellular elemental iron in the E. faecalis OG1RF strain grown to mid-log phase in FMC[+Fe] and FMC[−Fe] revealed a small and not statistically significant difference between the two conditions (0.410 ± 0.122 μM intracellular iron in FMC[+Fe] versus 0.322 ± 0.127 μM iron in FMC[−Fe]). These results strongly indicate that the enterococci are well equipped to scavenge iron and maintain iron homeostasis under extremely adverse conditions. To facilitate interpretation of the RNA-seq study, we used a false-discovery rate (FDR) of 0.01 and applied a 2-fold cutoff to generate a list of differently expressed genes (Table S2). For illustration purposes, the 200 differentially expressed genes (92 upregulated and 108 downregulated) were grouped according to Clusters of Orthologous Groups (COG) functional categories, with genes coding for membrane-associated transporters (22%), metabolism (31%), and hypothetical proteins (53%) comprising the majority of genes identified in the comparison (Fig. 2). Compared to cells grown in FMC[+Fe], the most upregulated genes (varying from 2.6- to 7.7-fold induction) in cells grown in FMC[−Fe] coded for proteins that belong to two uncharacterized ABC-type transport operons (OG1RF_RS12045 to OG1RF_12060 and OG1RF_RS12585 to OG1RF_12595) (Fig. 3 and Table S2). While there is no previous experimental evidence that these transporters are involved in metal uptake, the operon from OG1RF_RS12045 to OG1RF_12060 was previously shown to be a member of the Fur (ferric uptake regulator) regulon (33) and is annotated as a putative ABC-type iron transporter (35). Herein, we refer to the operon from OG1RF_RS12045 to OG1RF_12060 as fitABCD for Fur regulated iron transporter and the operon from OG1RF_RS12585 to OG1RF_12595 as emtABC for enterococcal metal transporter. Based on searches of public databases and phylogenetic tree analyses with the substrate binding proteins FitD or EmtC, the proteins encoded by fitABCD and emtABC are highly conserved among the enterococci (Fig. 3 and 4). Beyond enterococci, FitD shares ~48% amino acid identity and ~65% similarity with Bacillus subtilis YclQ and Streptococcus pneumoniae SPD_RS08810, whereas the nonenterococcal protein most closely related to EmtC is Streptococcus pyogenes RS01525, which shares 29% identity and 47% similarity with EmtC. Notably, Bacillus subtilis YclNOPQ has been implicated in the uptake of the petrobactin siderophore (33, 37) such that it is possible that FitABCD is involved in the uptake of siderophores. Other than the upregulation of the fitABCD and emtABC operons, a few other notable alterations in the iron starvation transcriptome were upregulation of genes from the mannose phosphotransferase system (PTS) and pyrimidine biosynthesis operons and downregulation of two P-type ATPases annotated as magnesium import transporters and the tellurite (toxic anion) resistance protein (Table S2). While studies to understand the significance of these other notable transcriptional changes to growth under iron starvation were not pursued in this investigation, these changes are suggestive of adaptation to iron starvation triggering changes in carbon and nucleic acid metabolism as well as metal resistance profiles.

TABLE 1.

Iron and manganese quantifications in the media used in this study

Medium	Concn (μM) of:
	Fe	Mn
FMC	80.02 (±10.123)	118.401 (±16.265)
FMC[−Fe]	0.052 (±0.008)	109.881 (±6.870)
FMC-LM	10.538 (±0.829)	10.414 (±0.807)
FMC-LM[−Fe]	0.002 (±0.0048)	14.853 (±3.362)
FMC[−Fe and −Mn]	0.0376 (±0.103)	0.0413 (±0.0215)
FMC[−Fe and +heme]	4.584 (±0.182)	11.992 (±3.811)
BHI	6.581 (± 0.950)	0.570 (±0.0208)

FIG 1.

Growth of E. faecalis and E. faecium strains in FMC[+Fe] or FMC[−Fe]. Growth was monitored by measuring OD₆₀₀ every 30 min using an automated growth reader. Error bars indicate standard deviations from three independent biological replicates.

FIG 2.

Summary of RNA-Seq analysis comparing E. faecalis grown under iron-depleted versus iron-replete conditions. Shown is a dot plot of genes differently expressed, via RNA sequencing, under conditions of iron depletion as determined by Degust (http://degust.erc.monash.edu/). The y axis indicates the fold change in expression compared to control cultures, while the x axis indicates position within the genome.

FIG 3.

Genetic organization of FitABCD (A) and EmtABC (B) and homologues found in selected Gram-positive bacteria. Percentages of amino acid identity and positive identity to OG1RF for substrate binding protein, permease, and ATPase are indicated.

FIG 4.

Phylogenetic tree analysis of the substrate binding proteins FitD (A) and EmtC (B). BLASTP searches against FitD and EmtC were used to identify homologues across species of enterococci, streptococci, bacilli, and other Gram-positive bacteria. Phylogenetic trees were constructed using multiple-sequence alignments of representative species using Clustal Omega and iTOL.

Temporal expression of iron transporters in response to iron starvation.

Previous studies revealed that the conserved iron transporter genes feoAB and fhuDCBG are regulated by the iron-sensing Fur regulator (33), whereas transcription of the dual iron/manganese transporter genes efaCBA is controlled by the manganese-sensing EfaR regulator (38). While none of the genes from the feoAB, fhuDCBG, and efaCBA operons were differently expressed in our RNA-seq analysis, we suspected that their transcriptional activation in response to iron starvation may occur immediately after cells undergo iron starvation, returning to basal expression levels after cells have adapted to the new (low-iron) environment. To investigate this possibility, we monitored (via reverse transcriptase quantitative PCR [RT-qPCR]) the transcriptional expression profile of efaCBA, feoAB, and fhuDCBG as well as fitABC and emtABC within the first hour after cultures were switched from iron-replete to iron-depleted condition. Using one representative gene for each operon as a proxy, we found that all transcriptional units were upregulated in response to iron starvation (Fig. 5). Notably, this induction occurred in two distinctly separated surges. In the first surge, emtB and efaA were strongly induced 10 min after cells were starved for iron but returned to near basal levels of expression after 60 min. In the second surge, fitA and fhuB were much more strongly induced at the later (60-min) time point. Finally, transcription of feoB was not altered during the initial 30 min but displayed a modest (yet significant) upregulation at 60 min such that feoAB was considered part of the second surge. These results strongly suggest that E. faecalis encodes, at the minimum, five bona fide iron import systems that can be separated into early and late responders.

FIG 5.

Quantitative real-time PCR analysis of iron transport genes transferred from iron-replete (FMC[+Fe]) to iron-depleted (FMC[−Fe]) conditions. Data shown represent four independent cultures with two technical replicates each. Line is set at 1 to indicate expression equivalent at T₀. Significance was determined by one-way analysis of variance (ANOVA) using Dunnett’s post hoc test to compare mRNA levels at T₀ with those at T₁₀, T₃₀ and T₆₀. ***, P ≤ 0.001; ****, P ≤ 0.0001.

FitABCD and EmtABC are important but not critical for growth under low-iron conditions.

To determine the contributions of FitABCD and EmtABC to growth of E. faecalis under iron-replete or iron-depleted conditions, each system was inactivated alone or in combination and the ability of ΔfitAB, ΔemtB, and ΔfitAB ΔemtB strains to grow in media containing different concentrations of iron and manganese was assessed. In brain heart infusion (BHI), a complex medium with ~6.5 μM iron, all mutants grew as well as the parent strain, OG1RF (Fig. 6A and Tables S3 and S4). In (chemically defined) FMC, which contains high concentrations of iron (75 μM FeSO₄) and manganese (100 μM MnSO₄) in the original recipe (36), all strains grew well, although the ΔfitAB and ΔfitAB ΔemtB strains attained slightly lower final growth yields (Fig. 6B and Tables S3 and S4). The omission of FeSO₄ from FMC slightly impacted growth rates and further lowered final growth yields of the ΔfitAB and ΔfitAB ΔemtB strains as well as the ΔemtB strain (Fig. 6C and Table S3 and S4). Because iron and manganese may function as interchangeable cofactors and E. faecalis is deemed a “manganese-centric” organism (31), we prepared a modified low-metal FMC (LM-FMC) formulation containing 1/10 of the original concentrations of iron and manganese for subsequent studies (Table 1). Like with the original FMC recipe, the ΔfitAB and ΔfitAB ΔemtB strains exhibited growth rate defects and reached significantly lower final growth yields in complete LM-FMC, with all mutants growing significantly more poorly in LM-FMC[−Fe] (Fig. 6D and E and Tables S3 and S4). Finally, all strains (parent strain included) grew slower and reached lower final growth yields in LM-FMC lacking both iron and manganese (Fig. 6F). Collectively, these results indicate that FitABCD and EmtABC contribute but are not essential to growth under iron-depleted conditions.

FIG 6.

Growth of OG1RF, ΔfitAB, ΔemtB, ΔfitAB ΔemtB, and Δ5Fe strains in BHI (A), FMC[+Fe] (B), FMC[−Fe] (C), LM-FMC[+Fe] (D), LM-FMC[−Fe] (E), and FMC[−Fe/−Mn] (F). Growth was monitored by measuring OD₆₀₀ every 30 min using an automated growth reader. Error bars indicate standard deviations from three biological replicates.

Simultaneous inactivation of efaCBA, feoAB, fhuDCBG, fitABCD, and emtABC further impairs growth under iron-depleted conditions.

To probe the individual and collective contributions of EfaCBA, FeoAB, and FhuDCBG to iron homeostasis, we took advantage of the ΔefaCBA strain that was already available in the lab (31) and isolated two new deletion mutants lacking FeoAB (ΔfeoB) and FhuDCBG (ΔfhuB). In BHI, LM-FMC, LM-FMC[−Fe], or LM-FMC[−Fe and −Mn], the ΔfeoB and ΔfhuB single mutants phenocopied the parent strain (Fig. S1). The ΔefaCBA strain also phenocopied growth of the parent strain in BHI, LM-FMC, or LM-FMC[-Fe] but could barely grow in LM-FMC[−Fe and −Mn] (Fig. S1), a phenotype that can be attributed to the role of EfaCBA in the uptake of both iron and manganese (31). Because the dual role of EfaCBA in iron and manganese acquisition creates a confounding factor (impaired manganese uptake), we next isolated a ΔfeoB ΔfhuB ΔfitAB ΔemtB quadruple mutant strain by sequentially inactivating feoB and fhuB in the ΔfitAB ΔemtB background such that a functional EfaCBA was retained in this mutant. However, this quadruple mutant grew exactly like the ΔfitAB ΔemtB double mutant in either LM-FMC or LM-FMC[−Fe] (Fig. S2). For this reason, our next step was to introduce the efaCBA deletion in the quadruple mutant background, yielding the ΔefaCBA ΔfeoB ΔfhuB ΔfitAB ΔemtB quintuple mutant strain, which we call Δ5Fe strain onwards. In BHI, FMC[±Fe], and LM-FMC, growth of the Δ5Fe strain was comparable to the growth rates and yields obtained for all single, double (ΔfitAB ΔemtB), and quadruple (ΔfeoB ΔfhuB ΔfitAB ΔemtB) mutants (Fig. 6A to D, Fig. S1 and S2, and Tables S4 and S5). However, the Δ5Fe strain grew slower and had lower growth yields than the ΔfitAB, ΔemtB, and ΔfitAB ΔemtB strains grown in LM-FMC[−Fe] and LM-FMC[−Fe and −Mn] (Fig. 6E and F and Tables S3 and S4). We next used the expression vector pTG001 (31) to express each individual iron transporter in the Δ5Fe background one at a time and then assess their ability to restore or alleviate the growth defect of the Δ5Fe strain in LM-FMC[−Fe] and FMC[−Fe and −Mn]. As shown in Fig. S3, expression of any one of the five transporters partially rescued final growth yields of the Δ5Fe strain in LM-FMC[-Fe] whereas (as expected) only the pTG-efaCBA vector (dual iron and manganese transporter) rescued final growth yields of Δ5Fe in FMC[−Fe and −Mn]. These results confirm the collective contribution of each transport system to iron homeostasis and further highlight the importance of the dual iron/manganese transporter EfaCBA to E. faecalis trace metal homeostasis.

To better understand the specific contributions of FitABCD and EmtABC and the collective contribution of the five transporters to iron homeostasis, we used inductively coupled plasma optical-emission spectrometry (ICP-OES) to determine the intracellular iron concentrations in the parent, ΔfitAB, ΔemtB, ΔfitAB ΔemtB, and Δ5Fe strains grown to mid-log phase in either LM-FMC or LM-FMC[−Fe]. In agreement with results showing that all strains grew well in iron-replete media (Fig. 6), no significant differences in intracellular iron content were observed between parent and mutant strains when grown in LM-FMC (Fig. 7A). On the other hand, intracellular iron pools were significantly lower in the ΔemtB (P ≤ 0.05) and Δ5Fe (P ≤ 0.001) strains when grown in LM-FMC[−Fe]. While the ~45% reduction in iron pools in the ΔemtB strain is apparently at odds with the results obtained with the ΔfitAB and double mutant strains, the ~90% reduction observed for the quintuple mutant bodes well with the marked growth defect of this strain in LM-FMC[−Fe]. To complement these observations, we determined iron (⁵⁵Fe) uptake kinetics in cultures of the parent, ΔfitAB ΔemtB, and Δ5Fe strains grown to mid-log phase in LM-FMC[−Fe]. Time course monitoring of ⁵⁵Fe uptake revealed a linear increase in iron uptake for the parent and ΔfitAB ΔemtB strains, while the Δ5Fe strain displayed a nonlinear and significantly (P ≤ 0.01) reduced capacity to take up ⁵⁵Fe over time (Fig. 7B).

FIG 7.

The Δ5Fe strain displays a major defect in iron acquisition. (A) ICP-OES analysis of intracellular iron content in OG1RF, ΔfitAB, ΔemtB, ΔfitAB ΔemtB, and Δ5Fe strains grown in LM-FMC[+Fe] or LM-FMC[−Fe]. (B) ⁵⁵Fe uptake kinetics of OG1RF, ΔfitAB ΔemtB, and Δ5Fe strains. Cells were grown in LM-FMC[−Fe] to mid-log phase and iron uptake was monitored over time after addition of 10 μM ⁵⁵Fe. The results shown represent the averages and standard deviations of five biological replicates for each data point. Significance was determined by two-way ANOVA followed by Dunnett’s post hoc comparison test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Next, we asked if loss of FitABCD, EmtABC, or all five iron transporters affected the pathogenic potential of E. faecalis by testing the ability of the ΔfitAB, ΔemtB, ΔfitAB ΔemtB, and Δ5Fe strains to grow and remain viable in human sera ex vivo as well as their virulence potential in the Galleria mellonella invertebrate model and in two mouse infection models. We found that in comparison with the parent strain, the Δ5Fe strain but not the ΔfitAB, ΔemtB, or ΔfitAB ΔemtB strain was recovered in significantly lower numbers after 24 h of incubation in pooled human sera at 37°C (Fig. S4). We expanded the serum growth/survival analysis by comparing the abilities of parent and Δ5Fe strains to grow and then remain viable in sera for up to 48 h. Similar to previous studies showing that mutants with defects in manganese or zinc uptake grow poorly in sera (31, 32), the Δ5Fe strain displayed a marked and significant growth defect in sera, growing less than 1 log during the initial 12 h of incubation, compared to the parent strain, which grew nearly 2 logs over the same period of time (Fig. 8A).

FIG 8.

The Δ5Fe strain displays defective growth/survival in human serum ex vivo and attenuated virulence in animal infection models. (A) Serum was obtained from blood pooled from 3 healthy donors, bacteria were inoculated into serum at 1.5 × 10⁶ CFU and incubated at 37°C, and growth/survival was monitored for 48 h. The experiment was repeated on four independent occasions, with three bacterial biological replicates on each occasion. Error bars indicate SEM, and significance was determined using the Mann-Whitney U test. (B) Percent survival of Galleria mellonella infected with OG1RF, heat-killed OG1RF, or the ΔfitAB, ΔemtB, ΔfitAB ΔemtB, or Δ5Fe mutant. Twenty larvae were infected with 5 × 10⁵ CFU of designated strains and incubated in the dark at 37°C to monitor survival over time. The Kaplan-Meier plot is a representative of three independent experiments. Significance was determined using the Mantel-Cox log rank test. (C) Seven-week-old C57BL6/J mice were infected via intraperitoneal injection with 2 × 10⁸ CFU of the designated strain. At 48 h postinfection, mice were euthanized, and peritoneal washes and spleens were collected for CFU determination. Mann-Whitney U test was used to determine significance. (D) Seven-week-old C57BL/6 mice were wounded with a 6-mm biopsy punch and infected with 2 × 10⁸ CFU of designated strains. At 3 days postinfection, mice were euthanized, and wounds were extracted and homogenized for CFU determination. (C and D) Data points shown are a result of the Robust regression and Outlier removal (ROUT method); bars indicate median values. Statistical analyses were performed using the Mann-Whitney test. *, P ≤ 0.05; **, P ≤ 0.01; ****, P ≤ 0.0001.

Because trace metal sequestration is an evolutionarily conserved defense mechanism present in both vertebrates and invertebrates (39–41) and previous studies conducted by our group revealed that virulence of manganese or zinc transport mutants in Galleria mellonella was severely compromised (31, 32), we assessed the ability of these mutants to kill G. mellonella. While the trends of the Kaplan-Meier curves shown in Fig. 8B are indicative that virulence may have been compromised in all the mutants tested, statistical significance (P ≤ 0.01) was noted only when comparing parent and Δ5Fe strains.

Our next step was to expand the in vivo studies to two mouse infection models: a peritonitis model in which infection becomes systemic within 12 to 24 h (42–44) and an incision wound infection model (45). In the peritonitis model, the Δ5Fe strain showed an ~1-log reduction (P ≤ 0.0001) in the number of total bacteria recovered from the peritoneal cavity 48 h postinfection compared to the parent, ΔfitAB, ΔemtB, and ΔfitAB ΔemtB strains (Fig. 8C). However, the parent and all mutants, including the Δ5Fe strain, were recovered in similar numbers from spleens (Fig. 8C). On the other hand, with the exception of the ΔemtB strain, all mutants were recovered from wounds in significantly lower numbers (P ≤ 0.05) than the parent strain (Fig. 8D).

E. faecalis can utilize heme as an iron source.

To this point, the results obtained indicate that E. faecalis relies on the cooperative activity of at least five iron uptake systems to overcome iron deficiency. However, the in vivo results also suggest that E. faecalis can deploy additional strategies to quench its need for iron during host infection. Because the most abundant source of iron in mammals is in the form of heme, whereby an iron ion is coordinated to a porphyrin molecule, and considering that some of the most successful invasive pathogens encode at least one dedicated heme import system (12, 26–29, 46–49), we suspected that E. faecalis can also use heme as an iron source. In fact, E. faecalis has at least two heme-dependent enzymes, catalase (KatA) and cytochrome oxidase (CydAB) (50–52), and a heme exporter (HrtAB) and heme-sensing regulator (FhtR) that are necessary to overcome heme intoxication (53). Yet none of the hundreds of E. faecalis genomes available encode the machinery for heme biosynthesis or systems homologous to any of the more conserved heme uptake systems, such that it remains elusive how E. faecalis acquires extracellular heme. Next, we asked if supplementation of the growth media with 10 μM heme could restore growth of the Δ5Fe strain in LM-FMC[−Fe]. As suspected, addition of heme greatly increased growth rates and yields of the Δ5Fe strain in iron-depleted media, although it also enhanced the final growth yield of the parent strain (Fig. 9A and Tables S5 and S6). Most likely, the beneficial effects of heme on cell growth are due to heme serving as the enzymatic cofactor for cytochrome oxidase and an iron source. For this reason, we grew cells in decreasing amounts of heme and found that as little as 10 nM heme was sufficient to rescue the growth defect of the Δ5Fe strain without enhancing growth of the parental strain (Fig. 9B and Tables S5 and S6). Conversely, when parental and Δ5Fe strains were grown in fresh human sera supplemented with heme, addition of 10 μM heme but not 10 nM heme partially rescued the growth defect phenotype of the Δ5Fe strain without providing a noticeable growth advantage to the parent strain (Fig. 9C). This result was expected, as serum contains heme-sequestering proteins such as hemopexin that should be capable to sequester free heme away from the bacteria before reaching saturation. To further probe the role of heme in iron homeostasis, we compared intracellular levels of heme and iron in parent and Δ5Fe strains grown in LM-FMC[−Fe] plus or minus 10 μM heme. As expected, heme was undetectable unless it was added to the growth media, with the two strains accumulating comparable levels of heme when grown in heme-supplemented LM-FMC (Fig. 9D). Importantly, heme supplementation more than doubled intracellular iron levels in the parent strain and restored intracellular iron homeostasis in the Δ5Fe strain (Fig. 9E). Collectively, these results reveal that E. faecalis can internalize and then degrade heme to release the iron ion. On a separate note, the differences in intracellular iron levels in the Δ5Fe strain grown in LM-FMC[−Fe] that were significantly lower but quantifiable in Fig. 7A but below the limit of detection in Fig. 9C are a faithful representation of the variations that we observe between different batches of media.

FIG 9.

Heme restores growth and virulence of the E. faecalis Δ5Fe strain in iron-depleted environments. Shown is growth of OG1RF or Δ5Fe strains in LM-FMC[−Fe] with or without 10 μM (A) or 10 nM (B) heme supplementation. Growth was monitored by measuring OD₆₀₀ every 30 min using an automated growth reader. Error bars indicate standard deviations from three biological replicates. (C) Twenty-four-hour growth of OG1RF and Δ5Fe strains in fresh human serum with or without supplementation with 10 μM FeSO₄, 10 μM heme, or 10 nM heme. The experiment was performed on two separate occasions, with three bacterial biological replicates. Error bars indicate SEM; significance was obtained using 2-way ANOVA with Sidak’s multiple-comparison test. (D) Intracellular heme content determined from cultures grown to mid-log phase in LM-FMC[−Fe] with or without 10 μM heme supplementation. Absorbances of chloroform extract samples and heme standards were used in the correction equation, Ac = 2 × A₃₈₈ − (A₄₅₀ + A₃₃₀), and normalized to total protein content. (E) ICP-OES analysis of intracellular iron content of OG1RF and Δ5Fe strains grown in LM-FMC[−Fe] with or without heme supplementation. (D and E) Significance was determined by two-way ANOVA and Dunnett’s post hoc comparison test. (F) Larvae of G. mellonella were injected with 50 pmol of heme or PBS 1 h prior to infection with the OG1RF or Δ5Fe strain. The control group was injected with heat-killed OG1RF. The Kaplan-Meier curve shown is representative of six independent experiments with 20 larvae per experiment. Significance was determined using the Mantel-Cox log rank test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

Next, we asked if exogenous heme could restore virulence of the Δ5Fe strain in the G. mellonella model. Because oxygen is transported not via hemoglobin/Fe-heme complexes in insects but rather through binding to two copper ions coordinated by histidine residues in hemocyanins (54), nonhematophagous insects such as G. mellonella are considered heme free (55). Thus, in the last set of experiments, we injected the hemolymph of G. mellonella with 50 pmol of heme b (in the form of hemin) 1 h prior to infecting the larvae with the desired E. faecalis strain. While heme administration did not affect the pathogenic behavior of the parent strain, it fully restored virulence of the Δ5Fe strain (Fig. 9F). These results led us to conclude that E. faecalis can acquire heme from the environment and that host-derived heme is an important source of iron during infection.

DISCUSSION

Despite the nearly universal role of iron in host-pathogen interactions (6, 7, 14, 19–23, 56), very little is currently known about the mechanisms utilized by E. faecalis to obtain iron from the extracellular milieu and much less so about the contribution of iron import systems to enterococcal fitness and pathogenic behavior. In a series of studies that spanned through 2 decades, Lisiecki and colleagues were the first to propose that enterococci utilize multiple strategies to scavenge iron, which included production of siderophores, expression of high-affinity iron transporters, and an undefined capacity to seize iron directly from host transferrin and lactoferrin (57–59). Yet most of their observations have not been validated by others and, at least in the case of siderophore production, appear to be incorrect based on the absence of the machinery necessary for siderophore biosynthesis in E. faecalis genomes. Indeed, our multiple attempts to detect siderophore production in different strains of E. faecalis or E. faecium using the chrome azurol S (CAS) method (60) were not successful (data not shown). In addition to the work by Lisiecki and colleagues, in silico and transcriptome-based analyses using a Δfur mutant have indicated that E. faecalis possesses three highly conserved iron import systems, the ferrous iron transporter FeoAB, the ferrichrome transporter FhuDCBG, and the dual iron/manganese transporter EfaCBA (31, 33–35).

In this investigation, we validated previous studies (61) showing that either E. faecalis or E. faecium isolates can grow in media that can be considered virtually iron free (0 to 0.003 ppm of iron depending on the batch of medium). While the remarkable capacity of enterococci and of other lactic acid bacteria to grow under nearly iron-free conditions has been attributed to their “manganese-centric” nature, intracellular iron quantifications revealed that E. faecalis accumulates similar amounts of iron when grown in iron-replete and iron-depleted media. Rather than suggesting that E. faecalis does not require iron for growth as once suggested (62), we believe that iron is such an essential micronutrient to E. faecalis that the organism evolved multiple, diverse, and highly efficient systems to acquire and maintain iron homeostasis.

In addition to the conserved iron import systems EfaCBA, FeoAB, and FhuDCBG, our transcriptomic analysis identified two novel ABC-type iron transporters named FitABCD and EmtABC. While this is the first time that EmtABCD has been linked to iron uptake, FitABCD was previously shown to be a member of the Fur regulon (33). Moreover, ex vivo and in vivo transcriptome analyses have shown that except for fhuDCBG, all systems are highly expressed under physiologically relevant conditions. For example, fitABCD was upregulated ~4-fold in both human blood and human urine ex vivo and 23- to 42-fold in a subdermal abscess rabbit model (63–65). The dual iron/manganese transporter efaCBA was upregulated ~3-fold in either human blood or urine (63, 64), ~2-fold in the abscess rabbit model (65), and ~7-fold in a peritonitis mouse model (44). Finally, emtABC was upregulated ~3-fold in human blood (63) and feoAB was upregulated ~2-fold in human urine and ~5-fold in the abscess rabbit model (64, 65). In this study, we showed that the individual responses of these transcriptional units to iron depletion can be divided into early (efaCBA and emtABC) and late (fitABCD, feoAB, and fhuDCBG) responders. Moreover, all late responders have been shown to be regulated by Fur (33), while efaCBA is regulated by EfaR (38, 66). Through bioinformatic analysis, we identified a putative EfaR-binding motif (38) located 13 bp upstream from the emtABC translational start site. Therefore, it is conceivable that transcriptional induction of iron acquisition systems is distinctly controlled by Fur and EfaR (Fig. 10). The occurrence of these two distinct transcriptional surges is reminiscent of the stepwise induction of iron uptake systems in B. subtilis whereby elemental iron, ferric citrate, and petrobactin operons are induced in the first wave and bacillibactin synthesis and uptake and hydroxamate siderophore uptake are induced in the second wave (67). However, in B. subtilis, this sequential activation was dependent solely on the Fur regulator, with subsequent experiments demonstrating that the stepwise transcriptional activation correlated with Fur operator occupancy in vivo (67). More studies are needed to determine if EfaR directly regulates emtABC, to validate the working hypothesis that iron starvation responses in E. faecalis can be separated by EfaR-regulated early responders and Fur-regulated late responders, and to determine the significance of the changes to metabolic pathways that were observed during iron starvation.

FIG 10.

Enterococcus faecalis utilizes five iron uptake systems to maintain iron homeostasis. The efaCBA and emtABC transporters are upregulated immediately following iron starvation, with efaCBA being regulated by EfaR. The exact mechanism of emtABC regulation remains unknown. The feoAB, fhuDCBG, and fitABCD operons are late responders to iron starvation and are known to be regulated by the Fur metalloregulator.

Even though systems homologous to EfaABC, FeoAB, and FhuDCBG are widespread and have been relatively well characterized in bacteria (21, 56, 68–71), predicted proteins sharing high levels (≥80%) of similarity with FitABCD or EmtABC are almost entirely restricted to species of the Enterococcaceae family, with FitD and EmtC sharing slightly lower similarity (~60 to 65%) with substrate binding proteins from selected streptococci and bacilli. Of interest, the B. subtilis YclNOPQ transporter is responsible for uptake of the petrobactin siderophore (37), raising the possibility that FitABCD mediates siderophore uptake. This might also be the case with FhuDCBG, which mediates uptake of ferric hydroxamate-type siderophores in other bacteria (72). As mentioned above, it appears that enterococci cannot synthesize their own siderophores such that these systems might be involved in xenosiderophore uptake, acquisition of siderophores produced by other bacteria (referred to as siderophore piracy), or uptake of other types of iron. Additional studies are necessary to determine the iron species specificity and affinities of FitABCD and EmtABC.

The finding that E. faecalis possesses multiple systems to acquire iron is not surprising considering their capacity to inhabit a variety of niches within the host, from the gastrointestinal tract to the skin, oral cavity, and the genitourinary tract, and to remain viable for prolonged periods when excreted into the environment. In addition, there are numerous examples in the literature describing how bacteria deploy multiple and complementary strategies to maintain iron homeostasis. As mentioned above, B. subtilis encodes transporters for the uptake of elemental iron, ferric citrate, and different types of siderophores, in addition to producing its own siderophore and cognate import system (67). Similarly, Staphylococcus aureus encodes transporters for elemental iron and iron hydroxamates and synthesizes two types of siderophores (staphyloferrins A and B) along with their cognate importers (21). In addition, S. aureus encodes the Isd system, which mediates binding, degradation, and uptake of iron-heme complexes (49, 73). Similar to what we observed in E. faecalis, Staphylococcus lugdunensis expresses multiple iron uptakes systems, covering a number of different physiologically available iron sources, as well as heme transporters that all contribute to virulence in vivo (27, 74). And finally, some of the major pathogenic species of streptococci encode a suite of elemental iron, siderophore, and heme transport systems (21, 29, 75, 76). As expected, inactivation of a single iron transport system had minimal or no impact on the ability of E. faecalis to grow under severe iron deficiency. To demonstrate this functional overlap, we generated a quintuple (Δ5Fe) mutant lacking all five systems. The Δ5Fe strain grew poorly in media without an added iron source, accumulated considerably less intracellular iron than the parental strain, and showed major deficiency in elemental iron uptake. The Δ5Fe strain also failed to grow in media depleted of both iron and manganese, likely because EfaCBA is a dual iron and manganese transporter. In fact, only expression of efaCBA in the Δ5Fe background, not any of the other iron transporters, restored growth in iron- and manganese-depleted media. In addition to our previous findings (31), these data emphasize the significance of efaCBA to enterococcal metal homeostasis. Despite these observations and considering that vertebrate hosts actively restrict both iron and manganese during infection, we found that the virulence potential of Δ5Fe varied depending on the model used and, possibly, the site of infection within the vertebrate host. While virulence of Δ5Fe was markedly attenuated in G. mellonella, and the mutant was recovered in significantly lower numbers from mouse peritoneal cavity and infected mouse wounds, parent and Δ5Fe strains were recovered in similar numbers from spleens in the peritonitis model. We suspected that the capacity to utilize heme as an iron source was behind this apparent conflicting result. To explore this possibility, we conducted a series of experiments that showed that E. faecalis is indeed capable of using heme as an iron source and that heme supplementation restores virulence of the Δ5Fe strain in G. mellonella. While E. faecalis does not possess the machinery for heme biosynthesis and does not require heme for growth (52, 77), it encodes at least two heme-dependent enzymes, cytochrome bd oxidase and catalase, such that it must have the capacity to obtain heme from the extracellular milieu. Yet systems homologous to known heme transport systems, such as the S. aureus Isd or the S. pyogenes Sia system, are absent in enterococcal genomes. During preparation of the manuscript, the Kline lab provided initial evidence that the ABC-type integral membrane proteins CydCD, previously implicated in cytochrome assembly and cysteine export (78, 79), might be involved in heme uptake (80). While additional studies are needed to confirm the role of CydCD in heme uptake, it is also apparent that CydCD does not work alone, since heme-dependent catalase activity can be still detected in cydABCD mutants (51). Unlike for other streptococci to which 10 μM heme is toxic, E. faecalis growth is enhanced at this concentration of heme and we have observed peak growth yields of OG1RF at 100 μM heme (data not shown). We suspect that this is in part due to the presence of the heme-sensing regulator FhtR and heme exporter HrtAB (53) as well as the use of heme as a cofactor in cytochromes to activate aerobic respiration (80) and to protect cells from oxidative stress through catalase activation (50). Nevertheless, our growth kinetic experiments (Fig. 9) suggest that E. faecalis possesses both high-affinity and low-affinity heme transporters whose altered expression may contribute to heme homeostasis. Studies to identify the elusive heme import systems of E. faecalis, to separate the significance of heme as a nutrient and as an iron source, and to determine how disruption of heme uptake affects the pathogenic potential of E. faecalis in different types of infection are ongoing.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains used in this study are listed in Table 2. All E. faecalis strains were routinely grown aerobically at 37°C in brain heart infusion (Difco). Strains harboring the pTG001 expression vector were grown with 25 μg mL⁻¹ of erythromycin. For controlled growth under metal-depleted conditions, we used the chemically defined FMC media originally developed for cultivation of oral streptococci (36), with minor modifications. Specifically, the base medium was prepared without any of the metal components (magnesium, calcium, iron, and manganese) and treated with Chelex (Bio-Rad) to remove contaminating metals. The pH was adjusted to 7.0 and the medium was filter sterilized. Magnesium and calcium solutions were prepared using National Exposure Research Laboratory (NERL) trace metal-grade water, filter sterilized, and then added to the media. Iron and manganese solutions were also prepared using NERL trace metal-grade water, filter sterilized, and added to the media as indicated in the text and figure legends. For RNA-seq analysis, overnight BHI cultures of E. faecalis OG1RF were diluted 1:100 in FMC[+Fe] or FMC[−Fe] and grown to an optical density at 600 nm (OD₆₀₀) of 0.5 before cells were collected for RNA isolation. For reverse transcriptase quantitative PCR (RT-qPCR) analysis, RNA was isolated from cells grown in FMC[+Fe] and then shifted to FMC[−Fe], with aliquots taken 10, 30, and 60 min after the shift. To generate growth curves, cultures were grown in BHI to an OD₆₀₀ of 0.25 (early exponential phase) and then diluted 1:200 into fresh media that were either BHI, FMC, or LM-FMC supplemented with heme, iron, and/or manganese as indicated in the text and figure legends. Cell growth was monitored using the Bioscreen growth reader (Oy Growth Curves).

TABLE 2.

Strains of E. faecalis and E. faecium used in this study

Strain name or genotype	Relevant characteristics	Source or reference
E. faecalis
OG1RF	Rifr Fusr	Lab collection
V583	Vanr clinical isolate	Lab collection
555-05	Clinical isolate	Lab collection
ΔfitAB	fitAB deletion	This study
ΔemtB	emtB deletion	This study
ΔfeoAB	feoAB deletion	This study
ΔfhuCBG	fhuCBG deletion	This study
ΔefaCBA	efaCBA deletion	31
ΔfitAB ΔemtB	fitAB deletion; emtB deletion	This study
ΔfeoB ΔfhuB ΔfitAB ΔemtB	feoB deletion; fhuB deletion; fitAB deletion; emtB deletion	This study
Δ5Fe	efaCBA deletion; feoB deletion; fhuB deletion; fitAB deletion; emtB deletion	This study
CK111	OG1S upp4::P23repA4 Specr; conjugation donor strain	83
OG1RF + pTG001	Empty vector; Ermr	31
Δ5Fe + pTG001	Quintuple mutant, empty vector; Ermr	This study
Δ5Fe + pTG-fitABCD	Quintuple mutant; fitABCD+ Ermr	This study
Δ5Fe + pTG-feoAB	Quintuple mutant; feoAB+ Ermr	This study
Δ5Fe + pTG-fhuCBG	Quintuple mutant; fhuCBG+ Ermr	This study
Δ5Fe + pTG-emtABC	Quintuple mutant; emtABC+ Ermr	This study
Δ5Fe + pTG-efaCBA	Quintuple mutant; efaCBA+ Ermr	31
E. faecium
ATCC 19634	ATCC strain	Lab collection
791-05	Clinical isolate	Lab collection
824-05	Clinical isolate	Lab collection

Construction of mutant strains.

Markerless deletions of fitAB, emtB, feoB or fhuB in E. faecalis OG1RF were carried out using the pCJK47 genetic exchange system (31). Briefly, PCR products of ~1 kb flanking each coding sequence were amplified with the primers listed in Table S7. To avoid unanticipated polar effects, amplicons included either the first or last residues of the coding sequences. Cloning of amplicons into the pCJK47 vector, electroporation, and conjugation into E. faecalis strains and isolation of single mutant strains (ΔfitAB, ΔemtB, ΔfeoB, and ΔfhuB) were carried out as previously described (31). The ΔfitAB ΔemtB double mutant was obtained by conjugating the pCJK-emtB plasmid into the ΔfitAB mutant. Then, a triple mutant was obtained by conjugating the pCJK-fhuB plasmid into the ΔfitAB ΔemtB double mutant and a quadruple mutant was obtained by conjugation of pCJK47-feoB into the ΔfitAB ΔemtB ΔfhuB triple mutant. Finally, the quintuple mutant was isolated by conjugation of pCJK-efaCBA (31) into the quadruple mutant. All gene deletions were confirmed by PCR sequencing of the insertion site and flanking region.

Construction of complementation strains.

The shuttle vector pTG001, a modified version of the nisin-inducible pMSP3535 plasmid (31) with an optimized ribosome binding site (RBS) and additional restriction cloning sites, was used to complement the Δ5Fe mutant strain. Briefly, the coding sequence of feoAB, fhuCBG, efaCBA, fitABCD, or emtABC was amplified from OG1RF using the primers listed in Table S7 and used in the In-Fusion cloning system by TaKaRa Bio USA, Inc., with the pTG001 vector digested with BamHI and PstI to yield plasmids pTG-feoAB, pTG-fhuCBG, pTG-efaCBA, pTG-fitABCD, and pTG-emtABC. Upon propagation in Stellar competent E. coli, pTG001 (empty plasmid) and the complementation vectors were electroporated into the Δ5Fe strain using a standard protocol (81) modified such that electroporated cells were immediately recovered in BHI supplemented with 0.4 M sucrose. The presence of plasmids was confirmed via PCR amplification of the DNA insert region using plasmid-specific primers.

RNA analysis.

Total RNA was isolated from E. faecalis OG1RF cells grown to mid-log phase in FMC[+Fe] or FMC[−Fe] or grown to mid-log phase in FMC[+Fe] and transferred to FMC[−Fe] following the methods described elsewhere (82). The RNA was precipitated with ice-cold isopropanol and 3 M sodium acetate (pH 5) at 4°C before RNA pellets were suspended in nuclease-free H₂O and treated with DNase I (Ambion) for 30 min at 37°C. Then ~100 μg of RNA per sample was further purified using the RNeasy kit (Qiagen), which includes 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). mRNA (5 μg total RNA per sample) was enriched using a MICROBExpress bacterial mRNA purification kit (Thermo Fisher) and cDNA libraries containing unique barcodes generated from 100 ng of mRNA using the Next UltraII directional RNA library prep kit for Illumina (New England Biolabs). The individual cDNA libraries were assessed for quality and quantity by Qubit and diluted to 10 nM each, and equimolar amounts of cDNA were pooled. The pooled cDNA libraries were subjected to deep sequencing at the UF-ICBR using the Illumina NextSeq 500 platform. Read mapping was performed on a Galaxy server hosted by the University of Florida Research Computer using Map with Bowtie for Illumina and the E. faecalis OG1RF genome (GenBank accession no. NC_017316.1) as a reference. The reads per open reading frame were tabulated with htseq-count. Final comparisons between bacteria grown in FMC[+Fe] and FMC[−Fe] were performed with Degust (http://degust.erc.monash.edu/), with a false-discovery rate (FDR) of 0.05 and after applying a 2-fold change cutoff.

ICP-OES.

Trace metal content in bacteria or growth medium was determined by inductively coupled plasma-optical emission spectrometry (ICP-OES). For quantification of trace metals in the different media used, 18-mL volumes of prepared media (BHI, FMC, and LM-FMC) were digested with 2 mL of trace metal-grade 35% HNO₃ at 90°C for 1 h. For intracellular metal quantification, cell pellets from overnight BHI cultures were washed once in 0.5 mM EDTA and twice in trace metal-grade phosphate-buffered saline (PBS) to remove extracellular metals and diluted 1:50 in LM-FMC with or without iron or heme supplementation as described in Results. Cultures were grown aerobically at 37°C to an OD₆₀₀ of 0.4, and the cell pellets were collected by centrifugation and washed once in 0.5 mM EDTA and twice in trace metal-grade PBS to remove extracellular metals. A 10-mL aliquot of resuspended cell pellet was saved for total protein quantification and 40 mL of the suspension was used for metal quantification. For this, cell suspensions were digested in 2 mL of 35% HNO₃ at 90°C for 1 h, and the digested suspension was diluted 1:10 in reagent-grade H₂O. Metal content was determined using a 5300DV ICP atomic emission spectrometer (Perkin Elmer) at the University of Florida Institute of Food and Agricultural Sciences Analytical Services Laboratories, and the data were normalized to total protein content that was determined by the bicinchoninic acid (BCA) assay (Sigma).

⁵⁵Fe uptake.

For ⁵⁵Fe uptake experiments, nitrocellulose membranes were prewet in 1 M NiSO₄ solution to prevent nonspecific binding of ⁵⁵Fe (Perkin Elmer) to the membranes. Overnight cultures of E. faecalis parent and mutant strains grown in LM-FMC[−Fe] were diluted 30-fold in LM-FMC[−Fe] and grown to mid-log phase (OD₆₀₀ of ~0.5), at which point 10 μM ⁵⁵Fe was added to each culture and incubated at 37°C. At 0, 15, 30, and 60 min, 200-μL aliquots were transferred to the prewet nitrocellulose membrane placed in a slot blot apparatus. Free ⁵⁵Fe was removed by four washes with 100 mM sodium citrate buffer using vacuum filtration. The membranes were air dried, cut, and dissolved in 4 mL of scintillation counter cocktail. Radioactivity was measured by scintillation with the “wide open” window setting using a Beckman LSC6000 scintillation counter. The count per million (cpm) values from ⁵⁵Fe free cells were obtained and subtracted from the cpm of treated cells. The efficiency of the machine was ~30.8% and was used to convert cpm to disintegrations per minute (dpm), which were then converted to molarity and normalized to CFU.

Intracellular heme quantification.

Cultures were grown under the same conditions as used for trace metal quantifications by ICP-OES. After three washings in trace metal-grade PBS to remove extracellular heme, pellets were suspended in 1 mL of dimethyl sulfoxide (DMSO) and lysed using a bead beater. Centrifugation was used to separate glass beads from cell lysates and, consequently, cell membranes associated with the glass beads, leaving primarily intracellular heme in the cell lysates to be analyzed. Cellular heme was determined using the acidified chloroform extraction method following the protocols detailed elsewhere (29). Absorbances of the organic phases at 388, 450, and 330 nm were determined using a GENESYS 30 visible spectrophotometer (Thermo Scientific). Heme content was determined by plugging absorbance values of samples and heme standards into the correction equation Ac = 2 × A₃₈₈ − (A₄₅₀ + A₃₃₀) where Ac signifies the corrected absorbance value, and were normalized by total protein content.

Growth in human serum.

Blood from healthy donors of the same blood type was obtained from LifeSouth Community Blood Centers in Gainesville, FL (IRB 202100899). Each experiment was performed with pooled serum isolated from blood of at least 3 individual donors. Where indicated, serum was supplemented with 10 μM FeSO₄, 10 nM heme, and 10 μM heme. After overnight incubation in BHI at 37°C, cell pellets were collected, washed once in 0.5 mM EDTA in trace metal-grade PBS and twice in trace metal-grade PBS, and subcultured into serum at ~1.5 × 10⁶ CFU mL⁻¹ with constant rotation at 37°C. Total CFU at selected intervals were determined by serial dilution and plated on tryptic soy agar (TSA) containing 200 μg mL ⁻¹ of rifampicin and 10 μg mL⁻¹ of fusidic acid.

Galleria mellonella infection.

Larvae of G. mellonella were used to assess virulence of parent and selected mutants as previously described (31). Briefly, groups of 20 larvae (200 to 300 mg) were injected with 5 μL of bacterial inoculum containing ~5 × 10⁵ CFU. To investigate the impact of exogenous heme supplementation, larvae were injected with either trace metal-grade PBS or 50 pmol of heme 1 h prior to infection. Larvae injected with heat-inactivated E. faecalis OG1RF (30 min at 100°C), 50 pmol of heme, or trace metal-grade PBS were used as controls. After infection, larvae were kept at 37°C and their survival was monitored for up to 96 h.

Mouse intraperitoneal infection.

Mouse intraperitoneal infection experiments were performed under protocol 202200000241 approved by the University of Florida Institutional Animal Care and Use Committee (IACUC). The mouse peritonitis infection model has been described previously (43), so only a brief overview of the model is provided below. To prepare the bacterial inoculum, bacteria were grown in BHI to an OD₆₀₀ of 0.5, and the cells pellets were collected, washed once in 0.5 mM EDTA and twice in trace metal-grade PBS, and suspended in PBS at ~2 × 10⁸ CFU mL⁻¹. Seven-week-old C57BL6/J mice purchased from Jackson laboratories were intraperitoneally injected with 1 mL of bacterial suspension and euthanized by CO₂ asphyxiation 48 h postinfection. The abdomen was opened to expose the peritoneal lining and 5 mL of cold PBS was injected into the peritoneal cavity, with 4 mL retrieved as the peritoneal wash. Quantification of bacteria within the peritoneal wash was performed by plating serial dilutions on TSA containing 200 μg mL ⁻¹ of rifampicin and 10 μg mL⁻¹ of fusidic acid. For bacterial enumeration inside spleens, spleens were surgically removed, briefly washed in 70% ethanol, rinsed in sterile PBS, homogenized in 1 mL of PBS, serially diluted, and plated on selective TSA plates.

Mouse wound infection.

Mouse wound infection experiments were performed under protocol 202011154 approved by the University of Florida IACUC. The bacterial inoculum was prepared as described for the peritonitis model, but cell pellets were concentrated to 1 × 10¹⁰ CFU mL⁻¹ and stored on ice until infection. Seven-week-old C57BL6/J mice purchased from Jackson laboratories were anesthetized using isoflurane, their backs were shaved, and the incision wound was created using a 6-mm biopsy punch. Wounds were infected with 10 μL of culture and covered with Tegaderm dressing. At 72 h postinfection, mice were euthanized by CO₂ asphyxiation, the wounds were excised, and the wounds were homogenized in 1 mL of PBS. The wound homogenates were serially diluted and plated on selective TSA plates.

Data availability.

Gene expression data have been deposited in the NCBI Gene Expression Omnibus (GEO) database under accession number GSE221057.