A Novel Enterococcus faecalis Heme Transport Regulator (FhtR) Senses Host Heme To Control Its Intracellular Homeostasis

Enterococcus faecalis, a normal and harmless colonizer of the human intestinal flora can cause severe infectious diseases in immunocompromised patients, particularly those that have been heavily treated with antibiotics. Therefore, it is important to understand the factors that promote its resistance and its virulence. E. faecalis, which cannot synthesize heme, an essential but toxic metabolite, needs to scavenge this molecule from the host to respire and fight stress generated by oxidants.

as a leading cause of enterococcal infections, and it is the third most frequent source of hospital-acquired nosocomial infections (3). E. faecalis is thus considered a major public health threat due to its intrinsic resistance to antibiotics and the emergence of multidrug-resistant isolates (3). Selective outgrowth of enterococci following intestinal dysbiosis is frequent, regardless of whether it results from antibiotic treatment, intestinal inflammation, or infection (4). In addition to intrinsic and acquired antibiotic resistances, E. faecalis is resistant to other antimicrobial factors, such as bile, and tolerates a wide variety of stress factors such as temperature, pH, oxygen tension, or oxidation (1).
For most living organisms, heme (iron porphyrin) (in this report, heme refers to iron protoporphyrin IX regardless of the iron redox state, whereas hemin refers to ferric iron protoporphyrin IX) is an essential cofactor of enzymes such as cytochromes, catalases, or peroxidases (5). The importance of heme resides in the unique properties of its iron center, including the capacity to undergo electron transfer, to perform acidbase reactions and to interact with various coordinating ligands (6). Paradoxically, the high potential redox of heme iron catalyzes the production of reactive oxygen species (ROS). Oxidative stress generated by heme together with its accumulation in membranes explains its toxicity (7)(8)(9). Most bacteria carry the enzymatic machinery for endogenous heme synthesis. However, numerous bacteria lack some or all the enzymes needed for autosynthesis but still require this molecule for their metabolism (5). Interestingly, E. faecalis, like the majority of species constituting the ga`strointestinal microbiota, cannot synthesize heme (10,11). When heme is added to an aerated culture, E. faecalis activates a terminal cytochrome bd oxidase, causing a shift from fermentation to an energetically favorable respiratory metabolism (11,12). E. faecalis, unlike other Firmicutes that cannot synthesize heme, also carries a gene that encodes a heme catalase (KatA; EC 1.11.1.6), limiting hydrogen peroxide stress when heme is available (10). Both activities contribute to the virulence of several Gram-positive pathogens (13,14). Although the importance of heme as a cofactor for numerous cellular functions is established (5,15), the mechanisms governing exogenous heme internalization and secretion that contribute to heme homeostasis vary among bacteria and are poorly understood. However, heme homeostasis must be strictly regulated in all bacteria to avoid toxicity (6,8). Heme efflux is a documented defense mechanism against heme toxicity in some Firmicutes. (i) The Pef regulon comprises two multidrug resistance efflux pumps and a MarR-type heme-responsive regulator in Streptococcus agalactiae (16). (ii) The HatRT system involves a TetR family heme binding transcriptional regulator (HatR) and a major facilitator superfamily heme transporter (HatT) in Clostridium difficile (17). (iii) Heme homeostasis in several Gram-positive bacteria relies on HrtBA (heme-regulated transport) proteins, an ABC transporter, which promotes heme efflux (14,(18)(19)(20). Expression of hrtBA is controlled by hssRS genes, encoding a two-component heme sensor and response regulator in numerous Gram-positive pathogens, including S. agalactiae, Staphylococcus aureus, and Bacillus anthracis (14,(19)(20)(21)(22). In contrast, the food bacterium Lactococcus lactis regulates HrtBA expression via the TetR family heme sensor HrtR (19). To date, the mechanisms involved in E. faecalis management of environmental heme are unknown.
In this work, we describe the mechanism by which a novel E. faecalis TetR regulator, called FhtR (for faecalis heme transport regulator), induces expression of HrtBA Ef (named HrtBA Ef for the HrtBA from E. faecalis), a conserved heme efflux transporter. We show that FhtR binds intracellular heme, resulting in derepression and increased transcription of hrtBA Ef . Heme iron coordination specifies FhtR as a heme sensor, and a critical role for the tyrosine 132 was defined. Our results also establish this system as a master mechanism of control of intracellular heme availability as shown by its requirement for the expression of the heme-dependent E. faecalis KatA. Finally, the relevance of the FhtR system to E. faecalis is shown in a mouse intestine model, suggesting the importance of FhtR for E. faecalis adaptation in the GIT. Our conclusions lead to a new picture of heme homeostasis in E. faecalis.

RESULTS
The conserved heme efflux transporter HrtBA Ef is functional in E. faecalis. E. faecalis OG1RF genome encodes two adjacent open reading frames (ORFs), OG1RF_ RS02770 and OG1RF_RS02775 sharing, respectively 24% and 45% amino acid (AA) sequence identity with HrtB and HrtA from Staphylococcus aureus (18) (see Fig. S1A and S1B in the supplemental material). We thus verified the role of these ORFs, referred to as HrtB Ef and HrtA Ef , in heme efflux. Growth of an in-frame DhrtBA Ef deletion mutant was severely impaired at hemin concentrations $ 25 mM compared to the wild-type (WT) OG1RF strain (Fig. 1A). However, WT OG1RF could overcome up to 500 mM hemin, highlighting the involvement of HrtBA Ef in limiting heme toxicity in E. faecalis increases sensitivity to hemin toxicity. Overnight cultures of WT and DhrtBA Ef strains were diluted to an OD 600 of 0.01 and grown with the indicated concentrations of hemin (in micromolar) for 10 h at 37°C in a microplate Spark spectrophotometer (Tecan). OD 600 was measured every 20 min. Values are the means 6 standard deviations (error bars) from three biological replicates. (B) Heme accumulates in the DhrtBA Ef strain. WT and DhrtBA Ef strains were grown to an OD 600 of 0.5 prior to the addition of 5 mM hemin in the culture medium for an additional 1.5 h. Bacteria were pelleted by centrifugation, and heme content was determined by the pyridine hemochrome assay on cell lysates. Heme content was normalized to the protein concentration. Background from bacteria not incubated with hemin was subtracted. Results represent the means plus standard deviations (error bars) from three biological replicates. Statistical significance was determined by t test where **** = P , 0.0001. (C) Visualization of cellular heme accumulation in the DhrtBA Ef mutant. Cells, grown as described above for panel A, were incubated for 1.5 h with 5 mM hemin. The bacteria were photographed following centrifugation. The results are representative of three independent experiments. (D) HrtBA Ef reduces heme cytoplasmic concentration. WT and DhrtBA Ef strains carrying the intracellular sensor plasmid, pP hrt -hrtR-lac were grown as described above for panel B. b-Gal activity was quantified by luminescence in relative light units [RLU]) after 1.5 h of incubation with 5mM hemin. Results represent the means plus standard deviations from three biological replicates. Statistical significance was determined by t test where **** = P , 0.0001. (E) HrtBA Ef prevents hemin-induced oxidative stress. WT and DhrtBA Ef strains were grown as described above for panel B with 5 mM hemin. Cells were washed with PBS plus 0.5% glucose, and ROS generation was quantified by the fluorescence of dihydrorhodamine 123. Results represent the means plus standard deviations from three biological replicates. Fluorescence background from bacteria not incubated with hemin was subtracted. Statistical significance was determined by t test where **** = P , 0.0001. (F) Induction of hrtBA Ef operon by hemin. The WT strain transformed with the reporter plasmid pP hrtBA -lac was grown, and b-gal activity was determined as described above for panel D following incubation with the indicated concentrations of hemin. Results represent the means 6 standard deviations from three biological replicates. Statistical significance was determined by one-way analysis of variance (ANOVA) with Dunnett's multiple-comparison test comparing each concentration of hemin to no-hemin control with statistical significance indicated as follows: *, P = 0.0202; ****, P , 0.0001.
An Intracellular Heme Sensor in E. faecalis ® (Fig. S1C). The DhrtBA Ef mutant grown in 5 mM heme-containing medium accumulated about twofold more intracellular heme than the WT strain, as evaluated by the pyridine hemochrome assay (23) (Fig. 1B). This result correlated with the intense red color of culture pellets from the DhrtBA Ef mutant compared to the WT strain (Fig. 1C). Intracellular heme concentrations were also monitored using the intracytoplasmic heme sensor HrtR (19): b-galactosidase (b-gal) activity from the reporter plasmid P hrt -hrtR-lac was about 4 times higher in DhrtBA Ef compared to the WT exposed to 5 mM heme (Fig. 1D). Finally, accumulation of heme in the DhrtBA Ef mutant correlated with a more than twofold increase of cellular ROS generated by heme as shown by the fluorescence of dihydrorhodamine 123 (24) (Fig. 1E). In conclusion, E. faecalis expresses a functional HrtBA Ef heme efflux transporter that modulates intracellular heme levels, thus reducing oxidative stress.
Transcriptional regulation of hrtBA Ef by heme was then investigated using a hrtBA Ef promoter reporter, P hrtBA -lac. b-Gal expression in the WT strain was induced as a function of concentration between 0.1 and 2.5 mM hemin in the culture medium. Induction reached a maximum at concentrations below 5 mM (Fig. 1F). This concentration range is far below WT strain sensitivity to heme toxicity ($25 mM) (Fig. 1A). We conclude that HrtBA Ef expression is induced at subtoxic heme concentrations.
A new TetR regulator, FhtR, controls hrtBA Ef expression. The above findings prompted us to investigate the mechanism of hrtBA Ef induction. Several Gram-positive pathogens regulate hrtBA Ef via an adjacent two-component system HssR and HssS (14,20,21). No hssR hssS genes were identified in or near the hrtBA Ef operon in E. faecalis OG1RF or other E. faecalis genomes. However, a monocistronic gene encoding a TetR family transcriptional regulator, OG1RF_RS02765, is adjacent to hrtBA Ef ( Fig. 2A), sharing no significant AA identity with the hrtBA regulator, HrtR, in Lactococcus lactis (19). We hypothesized that OG1RF_RS02765 was the transcriptional regulator of hrtBA Ef and tentatively named it FhtR (for faecalis heme transport regulator) ( Fig. 2A).
To investigate the role of FhtR in heme-dependent transcription of hrtBA Ef , an fhtR in-frame deletion in strain OG1RF (DfhtR) was constructed and transformed either with pP hrtBA -lac or pfhtR encompassing both (pP hrtBA -lac and P fhtR -fhtR) expression cassettes (Fig. 2B). In contrast to the WT strain, b-gal was expressed independently of heme in DfhtR(pP hrtBA -lac) (Fig. 2B). Transformation of pfhtR with DfhtR led to overcomplementation compared to the WT(pP hrtBA -lac) strain (Fig. 2B). Moreover, on solid medium, DfhtR exhibited a complete absence of sensitivity to heme compared to the WT and complemented DfhtR(pfhtR) strains (Fig. 2C). Similar results were obtained in liquid culture (Fig. S2). These results are in line with the observation that heme accumulation is reduced in the DfhtR mutant strain compared to the WT or DfhtR(pfhtR) strain (Fig. 2D) and that HrtBA Ef may be constitutively expressed in the DfhtR mutant (Fig. 2B). Indeed, P fhtR was constitutively transcribed, with no effects of heme, nor of FhtR expression as shown using P fhtR -lac as the reporter (Fig. 2E), and by Western blotting (WB) using FhtR-hemagglutinin (HA) tagged fusion expressed from P fhtR (Fig. 2F). We conclude that E. faecalis uses a constitutively expressed, unique intracellular heme sensor, FhtR, to control hrtBA Ef expression.
FhtR is a heme binding protein. Members of the TetR family of transcriptional regulators act as chemical sensors (25,26). Ligand binding alleviates TetR protein interactions with their respective operators, leading to promoter induction (25,26). To verify that heme was the signal that relieves FhtR-mediated hrtBA Ef repression, recombinant FhtR was purified as a fusion to the maltose binding protein (MBP-FhtR) from Escherichia coli. MBP-FhtR appeared green (Fig. 3A, inset), and its UV-visible spectrum exhibited a strong Soret band, suggesting that FhtR scavenges endogenously produced heme (Fig. 3A). To purify an apoFhtR, MBP-FhtR was expressed from a heme synthesis-deficient E. coli strain (hemA::kan) (Fig. 3B, dashed line). The purified protein bound hemin b in vitro (i.e., noncovalently) with a similar UV-visible spectrum as observed above for in vivo-bound heme: a Soret band at 407 nm and Q bands at 491 nm, 528 nm, and 608 nm (Fig. 3B, holoFhtR). Size-exclusion chromatography profiles showed that both apo-and holo-MBP-FhtR eluted as a single peak corresponding (B) FhtR controls hrtBA Ef expression. WT and DfhtR strains carrying the reporter plasmid pP hrtBA -lac and a DfhtR strain carrying a plasmid, pfhtR, combining both P hrtBA -lac and P fhtR -fhtR were grown to an OD 600 of 0.5, and b-gal expression was quantified by luminescence as reported in the legend to Fig. 1 with the indicated concentrations of hemin. Results represent the means plus standard deviations (error bars) from three biological replicates. Statistical significance was determined by t test as follows: ns, not significant (P . 0.5); ****, P , 0.0001. (C) fhtR deletion abrogates heme toxicity. Stationary-phase cultures of WT, DfhtR, and DfhtR(pfhtR) strains were plated on solid medium. Hemin (10ml of a 1mM stock solution) was pipetted directly onto plates, which were incubated for 24 h. Inhibition zones appear as a circular clearing in the center of each panel. No inhibition zone was observed for the DfhtR strain. The results are representative of three independent experiments. (D) Visualization of the impact of FhtR on heme accumulation. WT, DfhtR, and DfhtR(pfhtR) strains were grown and incubated with 5 mM hemin as described above for panel A. The bacteria were pelleted by centrifugation and photographed. The results are representative of three independent experiments. (E) FhtR expression is constitutively induced. b-Gal expression upon hemin addition to the culture medium in WT and DfhtR strains transformed with the pP fhtR -lac reporter plasmid was determined by luminescence as described in the legend to Fig. 1. Results represent the means plus standard deviations from three biological replicates. (F) fhtR transcription is not mediated by hemin. The DfhtR strain transformed with pP fhtR -fhtR-HA or carrying an empty vector (control) was used to monitor FhtR expression by Western blotting (WB) using an antihemagglutinin (anti-HA) antibody (a-HA). Bacteria were grown to an OD 600 of 0.5 and incubated with 2.5mM hemin for 1.  (25). The 608-nm charge transfer band and Soret at 407 nm are indicative of a ferric high-spin tyrosinate-ligated heme where heme is anchored through a proximal tyrosinate side chain (27,28). Hemin pentacoordinate high-spin ligation to FhtR was further confirmed by electron paramagnetic resonance (EPR) spectroscopy (see below). The heme dissociation coefficient (K d ) was 310 nM as determined by MBP-FhtR fluorescence quenching over increasing concentrations of hemin (Fig. 3D). Heme titration by differential absorption spectroscopy at 407 nm showed that the saturation point corresponded to the binding of one molecule of hemin per MBP-FhtR monomer (Fig. 3D, inset). Altogether, these data demonstrate that FhtR is a heme binding protein, suggesting that heme interaction is the primary event leading to activation of hrtBA Ef transcription.
Tyrosine 132 is a crucial heme axial ligand in FhtR. According to UV-visible and EPR spectra, the likely candidate for axial ligand of oxidized heme is a tyrosine (Y) (see above). Several Y residues present in FhtR were substituted to phenylalanine (F) (Fig. S3A, in blue). F and Y both have phenyl ring structures, so that F substitution minimizes an impact on FhtR conformation. Although F lacks the hydroxyl group that u.]) was recorded and plotted against hemin concentration. The experiment was repeated three times, fitted using the nonlinear regression function of GraphPad Prism 4 software, and gave a K d of 310 nM. The inset depicts the absorbance at 407 nm of ApoMBP-FhtR plotted against hemin concentration. The curve is representative of 10 independent experiments and was fitted using the nonlinear regression function of GraphPad Prism 4 software, which determined that the stoichiometry of the FhtR-hemin complex was 1:1.
coordinates heme, FhtR heme binding was not modified for several mutants tested individually (Fig. S3A). Only FhtR Y132F was purified from E. coli with a strong decrease in heme content compared to WT MBP-FhtR, indicating a loss of heme affinity in vivo (Fig. 4A). Surprisingly, apoMBP-FhtR Y132F purified from hemA::kanA E. coli exhibited similar UV-visible spectra (Fig. 4B) and K d upon hemin addition (data not shown), questioning the implication of this tyrosine in heme binding. The role of Y132 was further analyzed by EPR spectroscopy (Fig. 4C). HoloMBP-FhtR exhibited an axial high-spin (S = 5/ 2) heme signal with two well-defined resonances at around g ; 6 (with a crossing point at 1,190 G) (Fig. 4C, inset) and a resonance at g ; 2 (;3,390 G), indicative of a 5coordinated Fe III structure. Although the UV-visible spectra of FhtR and FhtR Y132F supplemented with hemin do not differ to a detectable level, the EPR spectra of FhtR Y132F was significantly modified, thus showing that either the ligand of the iron has been exchanged for another one or more likely, the interaction of the axial ligand with its environment has been modified. To conciliate these results, it is possible that while Y132 is the primary ligand, another distal ligand can take over ligation in the Y132F mutant to become the dominant ligand in vitro (meanwhile hydrophobic contacts would ensure retention of the binding affinity).
We then compared FhtR and FhtR Y132F activities in vivo. The DfhtR mutant was complemented either with pfhtR (pP hrtBA -lac; P fhtR -fhtR) or pfhtR Y132F (pP hrtBA -lac; P fhtR - b-gal activity from the DfhtR strain transformed either with pP hrtBA -lac, P fhtR -fhtR or pP hrtBA -lac, or P fhtR -fhtR Y132F was determined as described in the legend to Fig. 1C following incubation with the indicated concentrations of hemin. Results represent the means plus standard deviations from three biological replicates. Statistical significance was determined by one-way analysis of variance (ANOVA) with Dunnett's multiplecomparison test comparing each concentration of hemin to pfhtR (0 mM) control with statistical significance indicated as follows: ns, not significant (P . 0.5); ****, P , 0.0001.
An Intracellular Heme Sensor in E. faecalis ® fhtR Y132F ), and b-gal expression was monitored upon hemin addition to medium (Fig. 4D). WT FhtR and FhtR Y132F were expressed to similar levels as confirmed on WB (Fig. S3B). Expression of both proteins prevented hrtBA Ef transcription in the absence of heme, in contrast to full expression in DfhtR (Fig. 4D). WT FhtR and FhtR Y132F were expressed to similar levels as confirmed on WB (Fig. S3B). However, hemin addition led to P hrtBA -lac expression in the strain carrying WT FhtR, but not FhtR Y132F , suggesting that heme derepression was impaired (Fig. 4D). Altogether, these data specify FhtR Y132 as a critical residue in the coordination of heme with FhtR, which enables hrtBA Ef transcription.
FhtR controls hrtBA Ef transcription by binding two distinct 14-nt palindromic repeat sequences. TetR family operators usually comprise a 10-to 30-nucleotide (nt) inverted repeat sequence with internal palindromic symmetry (25). Two such 14-ntlong palindromes were identified in the 210/235 region of the hrtBA Ef promoter (called P1 and P2; Fig. 5A). An electrophoretic mobility shift assay (EMSA) was performed with apoMBP-FhtR, using a 325-bp DNA segment comprising the hrtBA Ef promoter (Fig. 5B) or a segment covering the internal hrtB region as a control (Fig. S4). FhtR-specific interaction with the P hrtBA DNA segment confirmed FhtR binding specificity. The shifted DNA migrated as two distinct bands (C1 and C2), in proportions that depended on the MBP-FhtR: DNA ratio (Fig. 5B), plausibly revealing that FhtR complexes with either one or two palindromes (Fig. 5B). To test this, we performed random substitutions of P1 and/or P2 nucleotides (P1* and P2*) and analyzed DNA shifts by EMSA. Replacement of both distal and proximal operators (P hrtBA P1*, P2* ) abolished the FhtR-induced DNA shift, confirming the role of palindromes in the interaction of FhtR with P hrtBA (Fig. 5C). Single replacement of P1 (P hrtBA P1* ) or P2 (P hrtBA P2* ) resulted in complete DNA shifts that migrated faster in the gel (C1) than seen with the native nucleotide sequence (Fig. 5C). We conclude that both P1 and P2 are FhtR binding sites.
We then tested the effects of heme on FhtR binding by EMSA. Addition of hemin to MBP-FhtR abolished the formation of the DNA-FhtR complex, as seen by the progressive disappearance of band shifts with increasing hemin concentrations (Fig. 5D). Complete release of FhtR from P hrtBA was obtained when hemin was in 10-fold molar excess over FhtR (Fig. 5D). Both C1 and C2 complexes were revealed when intermediate amounts of heme were added (0.1 and 1 mM; Fig. 5D). This suggests the release of MBP-FhtR from only one operator depending on the saturation level of FhtR with hemin.
The role of the two operators in the control of hrtBA Ef was investigated in vivo, using P hrtBA or a P hrtBA P1*, P2* promoter variant to control lac gene (Fig. 5E). In contrast to pP hrtBA -lac, which was strongly induced with 1 mM hemin, P hrtBA P1*, P2* -lac was constitutively expressed (Fig. 5E). Finally, the role of each operator was investigated (Fig. 5F). In the absence of heme, either P1 or P2 is sufficient for full P hrtBA repression by FhtR. Release of the promoter was facilitated in the presence of only P1 or P2 as shown with increased transcriptional activities of P hrtBA P1* or P hrtBA P2* in the presence of hemin compared to native P hrtBA (Fig. 5F). We propose that the presence of two operators provides strong repression of the hrtBA Ef promoter, thus preventing transcriptional leakage and allowing for fine tuning of HrtBA Ef expression. Taken together, these results demonstrate that FhtR is a heme sensor that directly controls heme homeostasis by regulating hrtBA Ef transcription.
FhtR controls HrtBA Ef , the gatekeeper of intracellular heme availability. Our observation that FhtR regulates intracellular heme pools even at low heme concentrations led us to hypothesize that FhtR controls intracellular heme availability in E. faecalis. We tested this possible role of FhtR on the E. faecalis endogenous heme-dependent catalase (KatA). While katA transcription is not susceptible to heme induction, KatA protein stability relies on the presence of heme (10,12). KatA-mediated H 2 O 2 catalysis was measured in WT, DfhtR, and DfhtR(pfhtR) strains (Fig. 6A). In the absence of hemin, H 2 O 2 consumption was at a basal level (Fig. S5A), thus excluding major contributions of other enzymes in our conditions. In the presence of 1 mM hemin (Fig. 6A), the DfhtR mutant exhibited about 30% catalase activity compared to WT and complemented DfhtR(pfhtR) strains, as evaluated by the percentage of catabolized H 2 O 2 (Fig. 6A). This was further confirmed by comparing the amounts of KatA (holoKatA) by WB, using anti-KatA antibody (kindly provided by L. Hederstedt). In the absence of hemin, KatA was expressed at low levels in WT, DfhtR, and DfhtR(pfhtR) strains (Fig. 6B). Comparatively, addition of hemin strongly increased the amounts of KatA in WT and complemented DfhtR(pfhtR) strains, but not in the DfhtR mutant (Fig. 6B). Low KatA availability in the DfhtR mutant is readily explained by constitutive heme efflux (via HrtBA Ef ), and consequently depleted intracellular heme pools in this mutant. We then evaluated the survival capacity of E. faecalis OG1RF WT, DfhtR, and DfhtR(pfhtR) strains when challenged with 2.5 mM H 2 O 2 . In the absence of hemin, all strains grew equivalently without H 2 O 2 (Fig. S5B). In contrast, while hemin addition rescued the survival of both the WT and DfhtR(pfhtR) strains, the DfhtR strain remained hypersensitive to H 2 O 2 (E) Substitution of the two palindromic nucleotide sequences, P1 and P2, in P hrtBA abrogates FhtR-mediated control of hrtBA Ef transcription. The WT strain was transformed either with the reporter plasmid pP hrtBA -lac or pP hrtBA P1*P2* -lac. b-Gal activity was determined as described in the legend to Fig. 1 following incubation with 2.5 mM hemin. Results represent the means plus standard deviations from three biological replicates. Statistical significance was determined by t test with statistical significance indicated as follows: ns, not significant (P . 0.5); ****, P , 0.0001. (F) Substitution of either P1 or P2 nucleotide sequences in P hrtBA enhances its transcriptional activation by hemin. The WT strain was transformed either with the reporter plasmid pP hrtBA -lac, pP hrtBA P1* -lac, or pP hrtBA P2* -lac. b-Gal activity was determined as described in the legend to Fig. 1 following incubation with hemin. Results represent the means plus standard deviations from three biological replicates. Statistical significance was determined by one-way ANOVA with Tukey's multiple-comparison test with significance indicated as follows: ns, not significant (P . 0.5); *, P = 0.0140; ****, P , 0.0001.
An Intracellular Heme Sensor in E. faecalis ( Fig. 6C and D). Deletion of hrtBA Ef in the DfhtR strain (DfhtR DhrtBA Ef ) restored the survival capacity in the presence of hemin (Fig. S5C). Thus, poor survival of DfhtR reflects the lack of heme needed to stabilize KatA (Fig. 6D). Finally, the OG1RF mutant (katA:: tetR) was hypersensitive to hemin toxicity, showing that KatA was required for controlling oxidative stress generated by heme (Fig. 1E and Fig. 4E). Taken together, these results identify FhtR as the direct and indirect regulator of HrtBA Ef -mediated heme efflux and KatA activity, respectively, with both mechanisms lowering heme stress in E. faecalis OG1RF. FhtR is thus a key mediator of heme homeostasis, and consequently, of oxidative stress response in E. faecalis generated by H 2 O 2 .
Heme sensing in the gastrointestinal tract. E. faecalis is a normal resident of the GIT of vertebrates, an ecosystem where heme is available (29)(30)(31)(32). We therefore investigated whether hrtBA Ef -mediated heme management is required by E. faecalis in the GIT in a murine gastrointestinal model. We generated E. faecalis OG1RF strains expressing the luxABCDE (lux) operon from Photorhabdus luminescens driven by the following: (i) P hrtBA (pP hrtBA -lux), which emits light specifically in the presence of hemin (Fig. 7A); (ii) a constitutive promoter P23 (plux), constitutively emitting light for bacterial tracking (14); or (iii) a control promoterless vector, pP 1 -lux. Cultures of these strains were  orally inoculated in the digestive tracts of mice, and light emission from whole live animals was measured in an in vivo imaging system IVIS200, 6 h postinoculation (Fig. 7B). This time delay corresponded to the maximum light emission from the tracking strain OG1RF(plux) (Fig. S6A). Luminescence signaling from the ingested E. faecalis pP hrtBA -lux heme sensor strain also localized in the abdomen, similar to the tracking strain (Fig. 7B). Examination of dissected organs revealed that the heme sensor-associated luminescence was mainly detected in the cecum (Fig. 7C), correlating with the high bacterial load of this organ [WT(plux); Fig. 7C]. A significant signal was also detected in the feces from inoculated animals, further highlighting that E. faecalis was able to scavenge and internalize heme within the digestive tract to induce hrtBA Ef expression (Fig. 7D). Finally, mice and human fecal samples (as well as fecal waters [ Fig. S6B and S6C]) from healthy individuals were able to induce luminescence from WT(pP hrtBA -lux) in vitro, excluding the possibility that induction of P hrtBA in vivo could result from the inoculation procedure (Fig. 7E). Therefore, representative animals corresponding to a total of 15 animals for each condition in three independent experiments. (C) Ceca exhibit high heme sensing signal. Animals as described above for panel B were euthanized and immediately dissected. Isolated GITs were imaged as described above for panel B. Ceca that exhibited most of the luminescence are shown (acquisition time, 5 min; binning 8). Bar = 1 cm. (D) Visualization of heme sensing in feces collected from mice following ingestion of WT(pP Ø -lux), WT(pP hrtBA -lux), or WT(plux). WT(pP hrtBA -lux) as described above for panel B were collected 6 to 9 h after gavage. Feces were imaged as described above for panel B (acquisition time, 20 min; binning 16). Bar = 1 cm. Results are representative of three independent experiments. (E) Human and mouse fecal samples activate heme sensing. Human feces from three healthy human laboratory volunteers and mouse feces from 6-month-old female BALB/c mice were deposited on M17G agar plates layered with soft agar containing WT OG1RF (pP hrtBA -lux). Plates were incubated at 37°C for 16 h and imaged in the IVIS 200 system (acquisition time, 10 min; binning 8). The figure shows representative results of a total of three independent experiments. An Intracellular Heme Sensor in E. faecalis ® FhtR heme sensor activity is active and relevant to E. faecalis heme management in the lumen of the GIT.
Heme sources for E. faecalis in the GIT. The results described above imply that E. faecalis internalizes heme in the intestinal environment to activate FhtR. Thus, an interesting question remains as to the identities of heme sources that are accessible to E. faecalis in the GIT. Normal bleeding (occult blood), exacerbated in intestinal pathologies, as well as food (as meat) are considered main sources of heme within the GIT (the second being excluded in mice) (29)(30)(31)(32). We thus visualized the ability of hemoglobin (Hb) or blood deposited on plates as schematized (Fig. 8A) to induce P hrtBA from the heme sensor strain WT(pP hrtBA -lux) as shown with hemin (Fig. 8B). Similarly, luminescence was induced in proximity of Hb and blood deposits as heme sources (Fig. 8C  and D). This result suggests that heme from physiologically available sources is internalized by E. faecalis. Crossfeeding of metabolites, including heme between bacteria, has been reported (29,33). The possibility that E. faecalis could scavenge heme from intestinal resident heme-synthesizing bacteria, such as E. coli-a phylum that becomes prevalent together with E. faecalis throughout dysbiosis-was evaluated. The WT (pP hrtBA -lux) heme sensor strain was grown in contact with E. coli (as the heme source) as illustrated in Fig. 8A. Strikingly, induction of luminescence was localized to areas of The results are representative of three independent experiments. (B) Visualization of heme sensing from hemin deposits. Hemin (1 mM) in PBS was used as described above for panel A. (C and E) Heme from blood (bovine) and hemoglobin (human) are heme sources for E. faecalis. Heparinized bovine blood (Thermo Fisher) and freshly dissolved human hemoglobin (1 mM) in PBS were used as described above for panel A. (F and G) E. coli is a heme donor for E. faecalis. E. coli (NEB10; New England Biolabs) (F) or a mutant strain that cannot synthesize heme (hemA::kan) (G) at an OD 600 of 0.1 were deposited on M17G plates as described above for panel A. Only the heme-synthesizing strain was able to crossfeed heme to E. faecalis. Panels B to F show representative results of three independent experiments. overlap between the two bacteria (Fig. 8E) and required heme synthesis by E. coli, as no sensing could be detected with a heme-defective hemA::kan mutant (Fig. 8F). This result suggests that heme synthesized by E. coli is internalized by E. faecalis. Thus, heme crossfeeding between bacterial symbionts in the gut might provide a heme source for E. faecalis. We conclude that the E. faecalis heme sensor is activated by the heme sources available in the GIT.

DISCUSSION
E. faecalis is a core member of the microbiome, and it is also the cause of a variety of severe infections (34). The central role of heme in reprogramming E. faecalis metabolism and fitness led us to investigate how heme homeostasis is controlled. A novel heme sensor, FhtR, is shown here to regulate heme intracellular homeostasis in E. faecalis. FhtRheme complexes derepress the hrtBA Ef operon, leading to HrtBA-mediated management of intracellular heme pools. While expression of HrtBA is a conserved strategy in multiple Gram-positive organisms, E. faecalis appears to be the first example of an opportunistic pathogen where HrtBA is not controlled by the two-component system HssRS. BLAST analysis of FhtR homologs in several Gram-positive bacteria showed that the regulator is present only in enterococci, vagococci, and carnobacteria (see Fig. S7A and S7B in the supplemental material for FhtR alignments and phylogenic tree). FhtR shares no homology with HrtR, a TetR regulator of hrtRBA in Lactococcus lactis (19). In contrast to HrtR which autoregulates its own expression, fhtR is monocistronic and expressed constitutively, implying that only HrtBA Ef expression is controlled by heme (19).
We characterized FhtR as a heme binding protein through pentacoordinated ligation of the heme iron, implying a tyrosine. This state of coordination is mostly found in heme receptors that transiently bind heme, such as IsdA, IsdC, and IsdH in S. aureus or HmA in Escherichia coli (35). FhtR blocks hrtBA Ef transcription by binding to two distinct 14-nt inverted repeats sequences in its promoter region. Alleviation of repression occurs when the heme-FhtR complex loses its affinity for its DNA binding sites. Conformational changes upon ligand binding is a shared mechanism among TetR regulators, leading to uncoupling from DNA (26). We thus hypothesize that these events, which we verified in vitro, explain FhtR regulation of the hrtBA Ef efflux pump in E. faecalis. The unique features of FhtR in E. faecalis compared to other regulators of hrtBA genes encoding efflux pumps support the idea that control of HrtBA-mediated heme homeostasis may vary among bacteria as a function of their lifestyle. It is thus tempting to speculate that differences in host niches, and in heme utilization and metabolism, might explain disparities in bacterial heme sensing mechanisms.
Heme efflux by HrtBA is reported as a bacterial detoxification mechanism that prevents intracellular heme overload (8,14,16,17,19). We showed here that HrtBA induction is required for E. faecalis survival when heme concentrations reached toxic levels (.25 mM). Yet, hrtBA Ef was induced at heme concentrations as low as 0.1 mM, suggesting that heme efflux is also needed at nontoxic levels. Interestingly, E. faecalis carries a gene that encodes the heme-dependent catalase whose activity relies on the amount of heme in the cytoplasm that is indirectly regulated by FhtR. This enzyme not only binds heme and thus lowers free heme levels, but it also actively lowers oxidative stress generated by heme. It will be of interest to determine the hierarchy of heme binding between FhtR and catalase in vivo.
To date, no heme import function has been identified in E. faecalis or in other tested Gram-positive bacteria that cannot synthesize heme (13,18). BLAST analysis of these bacteria failed to identify genes of the isd heme import system described in Staphylococcus aureus (36,37). In S. aureus, heme receptors and transporters are induced in iron-depleted growth media, and imported heme is used as an iron source (36). Thus, our findings led us to question the need for a dedicated transport system to internalize exogenous heme in E. faecalis and to propose an alternative hypothesis. We noted that HrtBA Ef is a member of the MacB family of efflux pumps that is distinct from other structurally characterized ABC transporters (38). A model based on MacB transport of antibiotics and antimicrobial peptides in Streptococcus pneumoniae proposed that transmembrane conformational changes promote lateral entry of substrates in the membrane before they reach the cytoplasm (39). On the basis of the previous and present data (23), we propose that HrtB Ef has the integral role as the heme "gatekeeper" in the cell. Like MacB antibiotics and antimicrobial substrates (40), membrane-bound heme could either enter passively into the intracellular compartment and or be effluxed by HrtB before this step. Altogether, our results place HrtBA Ef at the forefront of heme homeostasis in E. faecalis that is dependent on the key role of FhtR to adapt to the dichotomy between toxicity and benefits of heme which may be crucial in the host.
In vivo bioluminescence imaging of E. faecalis using an FhtR-based sensor identified the GIT as an environment where HrtBA Ef is expressed. The gut lumen of healthy individuals contains heme, independently of the nature of ingested food or of the microbiota (29)(30)(31)(32). Heme in the GIT is reported to mainly originate from Hb from normal bleeding (occult blood) (41). Accordingly, E. faecalis was able to internalize heme from blood and Hb in vitro. In addition, a common microbiota constituent, Escherichia coli, is shown to be a heme donor, suggesting a novel basis for intestinal bacterial interactions. As several phyla composing the core microbiota are heme auxotrophs with vital heme requirements, it is tempting to hypothesize that normal or disease-associated fluctuations in host heme levels could be detected by FhtR to adjust its intracellular level and optimize bacterial fitness. Interestingly, E. faecalis causes a variety of severe infections, most often among antibiotic-treated hospitalized patients with intestinal dysbiosis favoring high E. coli and enterobacterial populations (42). It will be interesting to evaluate the impact of HrtBA and FhtR in E. faecalis fitness and virulence in in vivo models. Taken together, our results suggest that the FhtR sensor and the HrtBA Ef heme gatekeeper allow E. faecalis to optimize its adaptation to variable heme pools in the host.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Bacterial strains and plasmids used in this work are listed in Table S1 in the supplemental material. E. coli NEB10 (New England Biolabs) was grown in LB medium, and E. coli C600 hemA::kan was grown in M17 medium supplemented with 0.5% glucose (M17G). Experiments with E. faecalis were all performed using strain OG1RF and derivatives (Table S1). E. faecalis was grown in static conditions at 37°C in M17G. When needed, antibiotics were used for E. coli at 50 mg · ml 21 kanamycin and 100 mg · ml 21 ampicillin; for E. faecalis, 30mg · ml 21 erythromycin was used. Oligonucleotides used for plasmid constructions are listed in Table S2. Hemin (Fe-PPIX) (Frontier Scientific) was prepared from a stock solution of 10 mM hemin chloride in 50 mM NaOH. In this report, heme refers to iron protoporphyrin IX regardless of the iron redox state, whereas hemin refers to ferric iron protoporphyrin IX. For growth homogeneity, WT and mutant strains were transformed with the promoterless pTCV-lac plasmid compared to complemented strains. Plasmid construction and E. faecalis gene deletion are described in Text S1 in the supplemental material.
b-Galactosidase assays. Stationary-phase cultures were diluted at an optical density at 600 nm (OD 600 ) of 0.01 in M17G and grown to an OD 600 of 0.5. Hemin was added to cultures, which were further grown for 1.5 h. b-Galactosidase activity was quantified by luminescence in a Spark microplate luminometer (TECAN) using the b-glo substrate (Promega) as described previously (19).
Cellular ROS quantification. Stationary-phase cultures were diluted at an OD 600 of 0.01 in M17G and grown to an OD 600 of 0.5. Hemin was added to cultures, which were further grown for 1.5 h. Bacteria were washed twice with phosphate-buffered saline (PBS) plus 0.5% glucose by centrifugation at 4°C to remove extracellular heme. Cell pellets were resuspended in PBS plus 0.5% glucose supplemented with 25 mM dihydrorhodamine 123, a cell-permeant fluorescent ROS indicator (Invitrogen). Cell suspensions were distributed into the wells of a 96-well plate. After 15-min incubation, optical density at 600 nm and fluorescence (excitation 500 nm; emission, 536 nm) were measured in a Spark microplate spectrofluorimeter (Tecan).
Bacterial lysate preparation. Bacteria were pelleted at 3,500 Â g for 10 min, resuspended in 20 mM HEPES (pH 7.5) and 300 mM NaCl and disrupted with glass beads (Fastprep; MP Biomedicals). Cell debris was removed by centrifugation at 18,000 Â g at 4°C for 15 min from the bacterial lysate supernatant. Proteins were quantified by the Lowry method (Bio-Rad).
Heme concentration determination in bacterial lysates. Equivalent amounts of proteins (in a volume of 250 ml) were mixed with 250 ml of 0.2 M NaOH, 40% (vol/vol) pyridine, and 500 mM potassium ferricyanide or 5 ml of 0.5 M sodium dithionite (diluted in 0.5 M NaOH), and 500-to 600-nm absorption spectra were recorded in a UV-visible spectrophotometer Libra S22 (Biochrom). Dithionite-reduced minus ferricyanide-oxidized spectra of pyridine hemochromes were used to determine the amount of heme b by monitoring the value of the difference between absorbance at 557 nm and 540 nm using a difference extinction coefficient of 23.98 mM 21 · cm 21 (43).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.03 MB.