Characterization of a Hemophore-like Protein from Porphyromonas gingivalis*

The porphyrin auxotrophic pathogen Porphyromonas gingivalis obtains the majority of essential iron and porphyrin from host hemoproteins. To achieve this, the organism expresses outer membrane gingipains containing cysteine proteinase domains linked to hemagglutinin domains. Heme mobilized in this way is taken up by P. gingivalis through a variety of potential portals where HmuY/HmuR of the hmu locus are best described. These receptors have relatively low binding affinities for heme. In this report, we describe a novel P. gingivalis protein, HusA, the product of PG2227, which rapidly bound heme with a high binding constant at equilibrium of 7 × 10−10 m. HusA is both expressed on the outer membrane and released from the organism. Spectral analysis indicated an unusual pattern of binding where heme was ligated preferentially as a dimer. Further, the presence of dimeric heme induced protein dimer formation. Deletional inactivation of husA showed that expression of this moiety was essential for growth of P. gingivalis under conditions of heme limitation. This finding was in accord with the pronounced increase in gene expression levels for husA with progressive reduction of heme supplementation. Antibodies reactive against HusA were detected in patients with chronic periodontitis, suggesting that the protein is expressed during the course of infection by P. gingivalis. It is predicted that HusA efficiently sequesters heme from gingipains and fulfills the function of a high affinity hemophore-like protein to meet the heme requirement for growth of P. gingivalis during establishment of infection.

Microorganisms rely on iron for a wide range of metabolic and signaling functions. A significant amount of iron in the host is coordinately bound to heme (Fe 2ϩ PPIX) or hemin (Fe 3ϩ PPIX), a prosthetic group of many biologically active proteins such as hemoglobin, myoglobin, and cytochromes. Although formal definitions distinguish "heme" and "hemin," the first term is widely used to indicate iron protoporphyrin IX in any oxidation state. As a major biological iron source for microbes, heme is also an important porphyrin source for certain bacterial species such as Hemophilus influenzae and Porphyromonas gingivalis that are unable to synthesize the tetrapyrrole ring de novo (1,2). Bacteria have developed two general systems to scavenge heme from their environs. The first involves synthesis of specific outer membrane receptors enabling direct contact between the organism and exogenous heme sources; for instance, HmuR, an outer membrane protein with relatively low heme binding affinity in P. gingivalis (3). The second strategy depends on secretion of heme-capturing molecules, hemophores, which scavenge free heme or heme from various carriers. There are currently only three characterized hemophores in the bacterial kingdom: two in Gram-negative bacteria including HasA in Serratia marcescens (4) and HxuA in H. influenzae (5) and one in a Gram-positive organism being the IsdX1 in Bacillus anthracis (6).
P. gingivalis, a Gram-negative anaerobe, is a leading pathogen in chronic periodontitis, a disease process involving progressive destruction of teeth-supporting tissues, including bone (7). Convincing data also exist from experimental models and clinical investigations revealing that P. gingivalis contributes to atheromatous plaque formation that predisposes to heart disease and stroke (8). Within the periodontopathic microbiota involved in periodontal disease, P. gingivalis is reported as one of the early colonizers of dental plaque with other bacterial species in the development of dental plaque biofilms (9,10). For successful colonization of the gingival crevice, P. gingivalis must acquire heme from limited quantities of host hemoproteins in the healthy gingival crevice, as well as compete with other heme/iron requiring microorganisms to scavenge essential heme (11). Because of the absolute requirement of heme for growth of this organism, it has been speculated that P. gingivalis may utilize a hemophore as a heme scavenger, although no candidate has been detected.
Here, we report the characterization of the product of PG2227 as a novel hemophore-like heme-binding protein HusA (heme uptake system protein A) detected in P. gingivalis growing under continuous culture in heme-limited conditions. HusA was found to be a high affinity heme-binding protein that preferably binds the ligand as -oxo dimeric heme. Further, our finding indicates that P. gingivalis responds to heme limitation by producing HusA as an outer membraneassociated protein as well as releasing it into the culture medium to act as a heme scavenger.

Bacterial Strains and Growth Conditions
P. gingivalis wild-type strain W83 and mutant derivatives were grown in enriched Tryptic Soy Broth (eTSB; per liter: 30 g of Trypticase soy broth, 5 g of yeast extract, 5 mg of hemin, pH 7.5, supplemented with 0.5 g of L-cysteine and 2 mg of menadione) 3 or blood eTSB agar (eTSB medium plus 15 g/liter agar and supplemented with 3% defibrinated sheep blood) at 37°C in an anaerobic chamber (Don Whitley Scientific Limited, UK) with an atmosphere of 90% N 2 , 5% CO 2 , and 5% H 2 . Escherichia coli strain DH5␣ was used for all plasmid construction work and was grown in Luria-Bertani broth medium and agar. For antibiotic selection in E. coli, ampicillin was used at 100 g/ml and erythromycin used at 300 g/ml. For P. gingivalis growth selection on solid medium, erythromycin was used at 5 g/ml and doubled in liquid culture. The phenotype and genotype of all strains and plasmids are listed in supplemental Table S1.
For continuous culture, P. gingivalis W83 was grown in a custom-designed chemostat system with a 70-ml working volume. Overnight P. gingivalis start culture was inoculated at 1:25 into modified basal medium (per liter: 10 g of proteose peptone, 5 g of yeast extract, 5 g of tryptone, 2.5 g of KCl, 0.5 g of L-cysteine, and 2 mg of menadione, pH 7.5) supplemented with hemin at various concentrations. The dilution rate was 0.05 h Ϫ1 , giving a mean generation time of 13.9 h; the pH was maintained at 7.5 Ϯ 0.1. Once steady state growth was established, cultures were harvested at 4°C. The biomass of the culture was monitored by optical density (Beckman DU640; Beckman Coulter), and culture purity was checked by Gram staining.

Production and Purification of Recombinant HusA
The husA gene was amplified by PCR from P. gingivalis W83 genomic DNA using rPG2227NcoIF, rPG2227XhoIR primers (see supplemental Table S2), and Accuprime Pfx DNA polymerase (Invitrogen) and subsequently cloned into NcoI and XhoI sites of the T7 expression pET24d(ϩ) plasmid (Novagen; Merck) to create the plasmid pETR7. The final gene construction encodes for HusA minus the first 23 amino acid residues of a putative signal peptide sequence as predicted by SignalP 3.0. The stop codon of husA was replaced by a thrombin cleavage site followed by a C-terminal His 6 tag.
E. coli strain BL21 (DE3) (Invitrogen) carrying the plasmid pETR7 was grown in 50 ml of LB medium with 50 g/ml kanamycin at 37°C overnight before being inoculated into 1 liter of prewarmed LB/kanamycin medium with vigorous shaking at 37°C. At A 600 of 0.6, 0.5 mM isopropyl-␤-D-thiogalactopyranoside was added, and the culture was allowed to grow for an additional 3 h before harvesting. The bacteria were sedimented by centrifugation and resuspended in cold native lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, and 10 mM imidazole, pH 8.0) at ϳ5 ml/g pellet followed immediately by a freeze-thaw process. The bacteria were lysed by pulse sonication on ice, and the supernatant was passed through a nickel-chelating resin to purify the His 6 -tagged protein (Qiagen). Purified protein was dialyzed into 100 mM Tris, pH 8.0, 250 mM NaCl before the addition of 10 mM CaCl 2 for thrombin cleavage. The His 6 tag was removed from the fusion HusA using the Thrombin CleanCleave TM kit (Sigma-Aldrich) at 4°C according to the manufacturer's instructions. Cleaved His 6 fragments were removed through a nickel-chelating column, and the detagged rHusA was dialyzed and concentrated into 100 mM Tris, pH 8.0, 250 mM NaCl.

Heme Solution Preparation
All of the heme solutions were prepared immediately before use. Fresh 10 mM stock solution of hemin (Ͼ98% HPLC; Fluka) in 0.1 M NaOH was diluted into 100 mM Tris buffer, pH 8.0, for -oxo bisheme binding assays. To obtain heme solutions comprised predominantly of monomeric heme, the pH of the 10 mM stock solution was adjusted to pH 7.5 by slow dropwise addition of HCl followed by dilution into 50 mM PIPES buffer, pH 6.5 (12). Surface heme-containing pigments from P. gingivalis colonies on blood agar were extracted by incubation in 140 mM NaCl, 100 mM Tris, pH 9.8, for 10 min at 20°C (13). Heme concentration in the extracted pigment solution was calculated using the heme molar extinction coefficient ⑀ 385 ϭ 58,400 (14).

HusA heme Binding Assays
Tetramethylbenzidine (TMBZ) Staining-Heme staining of SDS-PAGE gels by TMBZ/H 2 O 2 detects heme bound to proteins. Samples subjected to PAGE were pretreated with conventional reducing sample buffer, with or without boiling. After electrophoresis, the gels were fixed in the dark for 1 h in a prechilled solution of sodium acetate (250 mM, pH 5.0), methanol, and H 2 O at a ratio of 6:3:1 (v:v). The gels were stained with 7:2:1 (v:v) of sodium acetate (250 mM, pH 5.0), TMBZ (6.3 mM in methanol), and H 2 O for 30 min, followed by color development with 30 mM H 2 O 2 for 30 min at 4°C in the dark.
Hemin-Agarose Binding-Briefly, rHusA was mixed and incubated with prewashed hemin-agarose (Sigma-Aldrich) in 100 mM NaCl, PBS buffer, pH 7.4, for 3 h at 37°C. Nonspecifically adsorbed proteins were washed off with 1 M NaCl, PBS, and 0.5% Sarkosyl, and bound proteins were eluted with SDS-PAGE sample buffer. Maltose-binding protein (pMYB5; New England Biolabs) and unsubstituted agarose beads were used as negative controls.
UV-visible Absorption Spectroscopy-All of the absorbance spectra were recorded in a quartz cuvette (Starna Pty Ltd.) using a Beckman DU800 spectrophotometer (Beckman Coulter). The absorption spectrum of monomeric heme on binding to HusA was recorded in 250 mM NaCl, 50 mM PIPES, pH 6.5. To further evaluate the bisheme binding property, holo-HusA was prepared by incubation of apo-HusA with heme (ratio of 1:1) at 25°C for 20 min in 250 mM NaCl, 100 mM Tris, pH 8.0. The HusA-heme complex was purified over a Sephadex G25 desalting column (1 cm ϫ 10 cm) to remove free heme before the absorption spectrum of holo-HusA was recorded. Subsequently, sodium dithionite crystals (ϳ1 mg) were added, and the spectra under these reducing conditions were recorded. For kinetic absorption spectra, similar conditions as above were used with measurement carried out at predetermined time points. All of the spectra were collected at 25°C. The observed rate constant was calculated using Prism software (GraphPad Software Inc.).
Tryptophan Fluorescence Quenching Assays-The binding affinity of rHusA to heme was investigated using the perturbation of intrinsic tryptophan fluorescence with a Luminescence spectrometer LS 50B (PerkinElmer Life Sciences) using a 10-mm-path length quartz cuvette (Starna Pty Ltd.). Fluorescence intensity at 338 nm with 295-nm excitation was recorded for 1000 l of 400 nM rHusA in 100 mM Tris buffer, pH 8.0, and at 10 min after each subsequent titration of 0.5 l of 80 M fresh heme stock into the sample. The monomeric heme binding affinity assay was performed in 50 mM PIPES buffer, pH 6.5. The binding titration data were fitted to the equation below, in which a single binding site is assumed (15), where F obs is the observed fluorescence, F 0 is the initial fluorescence, F max is the maximum amplitude of fluorescence quenching, [L] t is the total ligand concentration, [E] t is the total concentration of protein, and K d is the apparent dissociation constant. The average K d values from three independent assays were reported for each condition.

Size Exclusion Chromatography
Analytical gel filtration chromatography was carried out using a Superdex 75 10/300 GL column (GE Healthcare) coupled to an Ä KTA TM purifier system with multiple wavelength sensor (GE Healthcare). The column was pre-equilibrated with 250 mM NaCl, 100 mM Tris, pH 8.0, at a flow rate of 0.8 ml/min prior to sample loading of 12.5 M rHusA with/without 12.5 M heme in a total volume of 400 l. Protein elution was monitored at 280 nm, and heme absorption was monitored at 399 nm and 385 nm. The column was calibrated with Blue Dextran 200 (void volume), conalbumin (75 kDa), carbonic anhydrase (29 kDa), ribonuclease A (13.7 kDa), and aprotinin (6.5 kDa) (GE Healthcare).

Cell Fractionations and Protein Localization Studies
Subcellular components were separated by Sarkosyl treatment and analytical centrifugation as previously described (16). Briefly, the cultures were adjusted to A 600 of 0.5 with cold PBS buffer, pH 7.4, and treated with protease inhibitors: 4 mM N ␣ -tosyl-lysine chloromethyl ketone and 1% (v/v) protease inhibitor mixture (Sigma-Aldrich) for Ն2 h at 37°C. The bacteria were pelleted by centrifugation at 6,000 ϫ g for 30 min, and the supernatant was further ultracentrifuged at 150,000 ϫ g for 1 h to separate vesicles and vesicle-free medium. Proteins in the supernatant fraction were precipitated by a pyrogallol red-molybdate-methanol method (17). The purity of the outer membrane fraction was confirmed by the exclusive presence of LPS and the purity of the inner membrane fraction by the exclusive presence of a biotin-containing oxaloacetate decarboxylase as detected by Western blotting using anti-LPS 1B5 mAb and alkaline-phosphataseconjugated streptavidin, respectively (16).

Western Blotting
Known amounts of protein antigen or whole bacterial cell cultures were standardized, separated by SDS-PAGE, and electroblotted onto 0.2-m nitrocellulose membranes (Bio-Rad). The membranes were blocked with 2% bovine serum albumin in PBS buffer for Ն2 h. Anti-HusA polyclonal antibody was produced through a subcontractor (Genscript Inc.) by immunizing rabbits with a synthetic peptide GGGKDKAL-PFAEKS present near the C terminus of HusA. Purified polyclonal antibody against HusA was used to probe the membranes at 1:5,000 dilution in TBST buffer for 3 h. Patient sera from a cohort that has been reported previously to be positive for P. gingivalis and has reactivity toward P. gingivalis proteins (18) were used at 1:40 dilution in the same conditions above. Alkaline phosphatase-conjugated goat anti-rabbit or goat anti-human IgG (Dako Corp.) antibodies were used at 1:10,000 dilution in TBST before final color development with alkaline phosphatase-conjugated conjugate substrate kit (Bio-Rad). Densitometric analysis was performed from scanned images in a linear range using ImageJ from the National Institutes of Health.

Quantitative RT-PCR
Chemostat cultures were diluted to A 600 of 0.6 with medium and stabilized with RNAprotect bacteria reagent (Qiagen) before RNA was extracted with the RNAqueous Micro kit (Ambion, Australia). Reverse transcription was carried out on 2 l of total RNA using the AffinityScript QPCR cDNA synthesis kit (Stratagene) in a volume of 10 l as per the manufacturer's instructions. Quantitative PCR was carried out in 25-l singleplex reactions using 5 l of 1:5 dilution of the cDNA along with TaqMan probe/primers against husA and 5 l of 1:200 dilution of cDNA for the calibrator gene 16S (see primers and probes in supplemental Table S2) on a Mx3005P real time PCR system using the Brilliant II QPCR Master Mix (Stratagene). Normalization and calibration methods were the same as in our previous publication (16).

Construction of P. gingivalis husA Mutant Strain
Creation of the deletional inactivation husA mutant was carried out as described previously (16). Briefly, a 1.2-kb upstream fragment (primers 2227frAEcoRF and 2227frASacR, with SacI and EcoRI sites) and a 1.3-kb downstream fragment (primers 2227frBPstIF and 2227frBSphIR, with PstI and SphI sites) to the husA gene (GenBank TM accession number PG2227) were amplified from chromosomal DNA of P. gingivalis W83 by PCR and inserted into pUC19. A 2.2-kb ermF/ ermAM cassette (primers ermSacF and ermPstIR, with PstI and SacI sites) was amplified from the plasmid pVA2198 (19) and inserted into the modified plasmid above to create pWD7. All of the primers used in this study are listed in supplemental Table S2. The plasmids were verified by DNA se-quencing prior to being electroporated into P. gingivalis W83 for homologous recombination into the genome to create a deletion of the gene. The husA-positive control mutant was created the same way except that the husA gene itself was amplified (primers 2227frCEcoRIF and 2227frCSacIR) and inserted into the plasmid rather than the upstream fragment to the husA gene to create pWDC7. Transformed cells were selected on erythromycin eTSB agar, and resistant clones were further confirmed by PCR and Southern blot using the ermF/ ermAM DIG-labeled probe (Roche).

Growth Experiments
Fresh colonies of P. gingivalis wild-type W83, deletion mutant WD7, and control mutant WDC7 were inoculated into eTSB medium without heme supplementation as the starter cultures for growth curve analysis. Heme stores were depleted by daily passage of the starter culture into medium without heme supplementation at 1:10 inoculum until the biomass of the following passage could no longer attain an A 600 of 0.3 after 24 h of incubation. The penultimate culture was adjusted to A 600 of 0.5, and a 1:10 inoculum was transferred anaerobically into 5 ml of prewarmed, prereduced eTSB medium supplemented with 50 nM or 5 M heme in individual Teflon screw cap test tubes (13 ϫ 100 mm). The tubes were capped tightly, removed from the chamber, and incubated at 37°C in a water bath. At predetermined time points, the tubes were vortexed, and absorbance at 600 nm was recorded using a Beckman DU640 spectrophotometer. Three independent experiments were performed in triplicate.

Statistical Analysis
Prism v3.03 software (GraphPad Software Inc.) was used for all statistical analyses. Real time PCR data and protein expression integrated densities from Western blotting were tested for normal distribution, and differences were compared using one-way analysis of variance, with Bonferroni's correction and 95% confidence intervals. Growth curve data from three independent experiments were expressed as the means and standard deviations, and the differences were analyzed by two-way analysis of variance. p values of less than 0.05 were considered significant.

RESULTS
HusA Is a High Affinity Heme-binding Protein-To determine functional characteristics, purified and untagged recombinant HusA (rHusA) (supplemental Fig. S1) was incubated with heme, resolved on SDS-PAGE, and initially stained with TMBZ to detect the presence of heme, followed by Coomassie Blue staining for protein detection. Heme was found to be associated with rHusA under reducing SDS-PAGE without boiling, but complete denaturation of the protein-heme complex by boiling of the sample resulted in the loss of staining for heme (Fig. 1A). This suggested that heme was strongly associated with rHusA but is not covalently bound to the polypeptide. The strength of heme binding by rHusA was confirmed by its binding to hemin-agarose despite a wash step with 1 M NaCl and 0.5% sarkosyl (supplemental Fig. S2).
Heme binding by rHusA was also detected by changes in the Soret region of the UV-visible spectrum of heme. Heme in solution exists as a monomer-dimer equilibrium (20). At pH 6.5, monomeric heme predominates, giving an absorption peak at 365 nm (Fig. 1B). During incubation of HusA with monomeric heme, a shift in the Soret band peaking at 399 nm emerged that is accompanied by reduction in absorption at 365 nm. The emergence of -oxo heme was indicated by the shift of the charge transfer band (Q band) to 600 nm with reduction of the monomeric heme band at 630 nm (Fig. 1B, inset) (12). This spectral pattern of heme absorption following ligation to HusA is also observed when heme monomer is incubated with gingipains that have been reported to mediate heme -oxo dimer formation in P. gingivalis (12).
To further evaluate the bisheme binding property of rHusA, the rHusA-heme complex at pH 8.0 was purified by chromatography over a desalting column. Under these conditions, free heme demonstrated a doublet Soret band at 365 and 385 nm, corresponding to monomeric and dimeric ferric heme, respectively (Fig. 1C). This is in agreement with evidence that alkaline solutions favor the formation of dimeric species such as -oxoor -*-bisheme (20,21). Binding of heme to rHusA at pH 8.0 was associated with a shift in peak absorbance of dimeric heme from 385 to 399 nm with a slight shoulder at 375 nm. This was supported by an emergence of a charge transfer band at 600 nm (Fig. 1C), characteristic of a hemoprotein with ligated -oxo bisheme (12,22). Taken together, the spectrum indicated that the primary binding mode of rHusA is to bisheme and the slight shoulder of the spectrum at 375 nm may indicate minor binding to other heme species. Similar binding characteristics have been reported for the surface heme-binding protein IsdH of Staphylococcus aureus (23). To investigate the nature of heme coordination by HusA, the rHusA-heme complex was reduced with excess dithionite. Following reduction of Fe 3ϩ to Fe 2ϩ , the Soret band underwent a red shift with the appearance of two well resolved Q bands at 539 and 569 nm, indicating the rapid formation of a low spin ferrous heme (24 -26) in the rHusAheme complex (Fig. 1C). Binding of low spin ferrous heme suggested HusA coordinates only to the fifth axis of the heme iron, with the sixth coordination site being available for occupation by oxygen (27,28). This is analogous to binding by oxymyoglobin (29). Identical shifts in spectra were observed following incubation of rHusA with black pigment composed of -oxo dimers extracted from P. gingivalis (supplemental Fig. S3).
Kinetic interaction of rHusA with heme in solution at pH 8.0 was monitored by measuring time-dependent changes in the Soret absorption spectrum (Fig. 1D). Within 30 s of mixing into rHusA solution, there was a pronounced shift of the heme Soret band from 385 to 399 nm, which progressively increased to reach equilibrium within 10 min. One-phase exponential association curve fitting was employed to describe the kinetic behavior of the absorbance signal. The observed rate constant K ob was calculated to be (4.3 Ϯ 0.4) ϫ10 Ϫ3 s Ϫ1 at 25°C. This indicated a fast association rate constant (k on ) for the kinetic interaction of heme with rHusA.

JOURNAL OF BIOLOGICAL CHEMISTRY 40031
The heme binding capacity of rHusA was determined by quenching of intrinsic tryptophan fluorescence (15). Sequential titration of 40 nM heme into 400 nM rHusA solution resulted in progressive quenching of rHusA fluorescence as shown in Fig. 2. Control titrations were obtained using Nacetyltryptophanamide as the fluorophore. Heme quenching data from the N-acetyltryptophanamide solution were used to correct for inner filter effects of the titrant. The binding constant at equilibrium at pH 8.0 was calculated to be 7.0 Ϯ 2.5 ϫ 10 Ϫ10 M, demonstrating high affinity binding of heme by rHusA. Moreover, binding saturation occurred at 400 nM heme titrated into 400 nM HusA, suggesting that HusA bound heme at a stoichiometry of 1:1. Binding affinity of rHusA for monomeric heme at pH 6.5 was calculated as 1.0 Ϯ 2.2 ϫ 10 Ϫ8 M, which was lower than for the dimeric form. The binding affinity for the extracted black pigment was 2.0 Ϯ 1.7 ϫ 10 Ϫ9 M.
Dimerization of HusA by Binding with Heme-Recombinant apo-HusA is a stable monomer under the experimental conditions used. However, the optical spectrum of HusA The inset shows the change in the Q bands. C, steady state UV-visible spectra of hemin, purified rHusA-heme complex, and rHusA-heme complex with sodium dithionite at pH 8.0. The Soret region of heme shifted Abs max from 385 to 399 nm on binding to rHusA. Following reduction of the ferric iron with the dithionite, the Abs max of the Soret region shifted further to 416 nm with better resolved Q bands. Increased absorbance at 370 nm under reducing conditions was attributed to sodium dithionite. The inset is an enlargement of the far-visible spectrum to show the Q bands. D, timeresolved spectra showing increases in absorbance over time in the Soret region of the mixtures of rHusA and heme at stoichiometry ratio 1:1 at pH 8.0. The inset shows the increase in absorbance at 399 nm versus time. The change in absorption was modeled to a one-phase exponential association equation to determine the K ob . binding with heme indicated the presence of bisheme species. To probe the apparent anomaly of monomeric protein-binding dimeric heme with 1:1 stoichiometry, molecular sieve chromatography was performed with UV-visible light spectroscopic analysis. As shown in Fig. 3, in the absence of heme, rHusA migrated as a monomeric form with an apparent molecular mass of 21 kDa, estimated by the elution volume relative to that of molecular weight markers. In the presence of heme at a molar ratio of 1:1, a rHusA dimer peak in front of the monomer peak was observed in the chromatogram, associated with absorption at 399 and 385 nm used to monitor heme in the holo-HusA complex. The apparent molecular mass of the dimer complex was ϳ43 kDa (including the heme dimer of 1.233 kDa). This clearly illustrated the dependence of protein dimerization on the binding of heme. These data indicated that rHusA dimerized upon binding to bisheme. Similar observations also have been reported for heme-binding proteins of the Gram-negative species Pseudomonas aeruginosa and Campylobacter jejuni and the Gram-positive species Streptococcal pyogenes and S. aureus (23, 30 -32).
Heme Limitation Up-regulates Transcription and Expression of husA-Expression of known heme uptake systems in P. gingivalis is often up-regulated in response to low levels of iron in the microenvironment (33). To investigate the role of heme concentration in the regulation of HusA expression, husA transcription and translation were studied in continuous cultures of P. gingivalis W83. Steady state continuous culture is of value as mimicry of growth in vivo by enabling bacteria to be cultured under conditions of nutrient limitation. Con-tinuous cultures of P gingivalis W83 at different heme concentrations were performed to probe husA transcriptional and translational responses by quantitative RT-PCR and Western blots, respectively. Transcription of husA was found to be inversely correlated to the concentration of heme supplementation with significant up-regulation in response to progressive reduction in heme from 2 to 0.05 M (Fig. 4A). The level of husA transcription under 0.05 M heme supplementation was ϳ3-fold higher than that under 0.1 M heme and 20-fold higher than under heme replete conditions of 1-2 M. Using Western blot with anti-HusA polyclonal antibody, expression of HusA in outer membrane fractions extracted from the same continuous cultures was found to correlate  similar analysis by immunoblot using anti-HusA polyclonal antibody to quantify the production of HusA in the outer membrane fraction. The bacterial cultures were adjusted to A 600 of 0.5 before outer membrane extraction as per "Experimental Procedures." C, densitometry of bands was measured as integrated density values (band area ϫ relative intensity) to determine the quantity of expressed proteins using National Institutes of Health ImageJ software. The data were calculated from two separate experiments. Significant differences from the heme replete condition of 2 M were analyzed by one-way analysis of variance. **, p Ͻ 0.01; ***, p Ͻ 0.001.
well to the level of husA transcription; the highest production of HusA was at 0.05 M heme supplementation with minimal production under heme replete conditions (Fig. 4, B and C). These findings indicated that expression of husA was tightly controlled by environmental heme concentration.
HusA Is Critical for P. gingivalis Growth under Heme Limitation-To examine whether the biochemical properties ascribed to HusA correlate with biological function, growth of wild-type P. gingivalis strain W83, husA-deleted mutant (WD7), and husA ϩ control mutant (WDC7) was investigated under heme excess (5 M) and heme-limited (0.05 M) conditions in batch culture (Fig. 5). In the absence of heme, none of the strains could grow after heme starvation (data not shown). Under a heme-limited environment of 0.05 M heme supplementation, the WD7 strain did not grow, whereas the wild type and control mutant recovered (Fig. 5B). Under hemereplete conditions, all of the strains recovered from heme depletion (Fig. 5C). The data confirmed that husA expression is essential for growth of P. gingivalis under heme-limited conditions.
HusA Demonstrated Hemophore-like Function-The presence of an N-terminal signal peptide and the absence of a membrane anchoring signal indicated that HusA could be secreted. To test this, P. gingivalis W83 was grown under conditions of heme limitation, and cellular fractionation was employed to separate proteins secreted into the medium from those tethered to the outer/inner membrane or located in periplasmic and cytoplasmic lysate. When analyzed by Western blot using anti-HusA antibodies, HusA was found to be distributed on the outer membrane, secreted in extracellular vesicles and in the vesicle-free media fraction, indicating that HusA could be secreted into the extracellular milieu to function as a heme scavenger (Fig. 6A). A slightly higher level of HusA could be seen in the vesicle fraction. This observation is reminiscent of the hemophore HxuA of H. influenzae, which has also been reported to partition predominantly to the culture supernatant and outer membrane fractions (5,34). Thus, HusA is the first hemophore-like protein reported in P. gingi-  A, continuous culture of P. gingivalis grown under 0.05 M heme supplementation was fractionated into equi-volume subcellular fractions in the presence of protease inhibitor cocktails as per "Experimental Procedures." Western blot using anti-HusA shows the presence of the protein in extracellular fractions and extracts from the outer membrane of the organism. B, outer membrane fractions extracted from P. gingivalis W83 grown under heme-limited condition were separated on two-dimensional PAGE gel followed by immunoblotting with anti-HusA. A series of spots ranging from isoelectric points of 5.4 to 6.2 and at apparent molecular mass of 26 kDa are recognized.
valis. Of note, the secreted form of HusA was readily detected in nutrient-rich eTSB medium supplemented with 0.05 M heme but not in basal medium (35) with 0.05 M heme (data not shown). The mechanism for different partitioning behavior in differing media is unknown, but it has precedence in the heme-binding protein HtaA from the Gram-positive pathogen Corynebacterium diphtheriae. This protein is an ironregulated heme-binding protein that has been shown to be secreted during growth in nutrient-rich medium but is predominantly membrane-associated during growth in a semidefined minimal medium (36). To further assess membraneassociated HusA, outer membrane fraction containing HusA was further separated by two-dimensional PAGE and immunoblotting. Six major isoforms in four clusters with isoelectric points ranging from 5.4 to 6.2 were detected on Western blot by anti-HusA antibody (Fig. 6B). Multiple isoforms of HusA may reflect the consequence of glycosylation or other posttranslational modifications to the expressed protein (37).
Humoral Immune Response to P. gingivalis HusA Protein-To test whether HusA is expressed during infection by P. gingivalis, Western blot analysis of rHusA was performed against sera from periodontitis patients determined by PCR to be infected with P. gingivalis. All six sera from infected periodontitis patients reacted with rHusA. This result was interpreted to indicate that HusA was expressed during the course of infection and induced a specific immune response (Fig. 7).

DISCUSSION
In human tissues, iron is sequestered by lactoferrin, transferrin, and ferritin as a primary defense mechanism at the onset of infection (38). Microorganisms in turn synthesize and secrete high affinity iron chelators, siderophores, to solubilize and take up Fe 3ϩ (39). However, P. gingivalis does not produce siderophores or ferric reductases. Although P. gingivalis employs the Feo system to uptake ferrous iron in anaerobic or microaerophilic environments (40), growth studies have shown that P. gingivalis preferentially acquires iron in the form of heme rather than from other sources (9,41). For P. gingivalis, heme is also an essential nutrient source of protoporphyrin IX because the organism lacks the capacity to synthesize the porphyrin macrocycle de novo (2). To obtain exogenous heme and compete with other microorganisms, P. gingivalis expresses a number of outer membrane heme-binding proteins, particularly under heme-limiting growth conditions (42). However, no hemophore has previously been described for P. gingivalis.
HusA, as reported here, is a novel hemophore-like protein in P. gingivalis. A BLAST search revealed a lack of significant homology between HusA and any known function proteins. Our findings showed that a significant proportion of total HusA protein synthesized by P. gingivalis under heme-limited conditions is released into the culture supernatant, as concluded from Western blot analysis of subcellular fractions (Fig. 6A). This finding has a precedent in the hemophore HxuA first reported as a lipoprotein on the surface of Hib (H. influenzae type b) and where some Hib strains secrete HxuA into the culture supernatant. Both forms of HxuA have been shown to scavenge heme from hemopexin (34,51). Unlike other hemophores, HusA has an unusual mode of heme binding as indicated by spectral absorption (Fig. 1), whereby iron from -oxo bisheme is presumably coordinated at the fifth position by HusA. Using size exclusion chromatography and UV-visible spectroscopy, we showed that apo-HusA is completely monomeric with dimerization occurring only following heme binding (Fig. 3). Because fresh ferric heme exists predominantly as a dimer species in aqueous solutions at physiological pH, HusA binding to this entity may induce dimerization of the protein as observed. Interestingly, S. marcescens was reported to secrete a dimeric form of the hemophore HasA (DHasA) in response to iron deficiency conditions (14). DHasA with two heme molecules ligated at high affinity serves the function of a heme reservoir in this organism (14). Compared with other heme-binding proteins found in P. gingivalis, HusA was shown to have a much higher affinity at 7.0 Ϯ 2.5 ϫ 10 Ϫ10 M for the cofactor (more than 1,000 times higher than HmuY with K d ϭ ϳ3 ϫ 10 Ϫ6 M). Because of oxidation toxicity, in the host, free heme is rapidly sequestered by carrier proteins, including hemopexin, albumin, and lipoproteins (52,53). With high heme binding affinity and a fast heme association rate, HusA could be able to compete against certain host hemoproteins including serum albumin (K d ϭ ϳ1.0 ϫ 10 Ϫ8 M) for free heme (49).
In P. gingivalis, genes encoding proteins involved in iron/ heme transport are typically clustered together on the genome. Three multigenic clusters, the iht, htr, and hmu loci, encoding proteins contributing to putative heme acquisition pathways, have been detected in the genome of P. gingivalis W83 (42). Deletion of the husA gene resulted in failure of growth under heme limitation (Fig. 5). Therefore, whereas the expression of multiple heme receptors including HmuY, Tlr, and IhtB in response to heme limitation has been described FIGURE 7. Immunoreactivity of sera from chronic periodontitis patients with recombinant HusA by Western blot. Individual patient sera (numbered) were diluted 1:40 and hybridized to the blot as previously reported (18). The positive control band (lane ϩ) was probed with rabbit polyclonal anti-HusA antibody as the primary antibody.
for P. gingivalis, these heme uptake systems could not compensate for the deficiency of the husA deletion mutant under heme limitation. The ORF located immediately upstream of husA encodes a hypothetical protein PG2226 identified as a heme-binding protein. 4 We annotated this gene adjacent to husA as husB (Fig. 5A). husB encodes an 83-kDa product with homology to TonB-dependent outer membrane receptors from other bacterial species, including an outer membrane siderophore receptor FpvA in P. aeruginosa, sharing 33% similarity to HusB. Further, two additional ORFs are located directly upstream of husB. We designated these as husC (PG2225) and husD (PG2224). The husD gene product is a hypothetical protein with little identity to other known functional proteins. BLAST analysis of husC predicted it to be a member of the MarR family of transcriptional regulators with significant identity (63%) and similarity (84%) to the MarR/ emrR family of transcriptional regulators in Prevotella sp. Oral taxon 472. This homology extended to a number of additional MarR/emrR family transcriptional regulators and hypothetical proteins belonging to the order Bacteroidales including Prevotella, Parabacteroides, and Bacteroides species. Proteins of the MarR family in bacteria have been reported to control a variety of biological functions, including oxidative potential and virulence (54). The MarR family transcriptional regulator PqrR of P. aeruginosa was demonstrated to regulate expression of downstream genes. Interestingly, the iron-containing prosthetic group in PqrR acts as a redox switch that regulates DNA binding activity and transcriptional repression (55). Recently, Lewis et al. (56) reported that expression of PG2227 (husA) and that of established heme uptake loci, was significantly down-regulated in the presence of oxygen. We also found that husA expression was significantly suppressed under high heme concentrations, indicating the presence of a heme-sensing repressor of the hus locus. We propose a model in which HusC mediates repression of husAB under aerophilic conditions. Under these conditions, the presence of oxidized heme could accelerate the rate of HusC-heme assembly, increase DNA binding activity, and reduce expression of the husABC locus. Alternatively, HusC may sense the presence of heme directly to negatively regulate expression of the hus locus. In either case, further investigation is needed to answer these questions.
In periodontally healthy individuals and during initiation of chronic periodontitis, P. gingivalis exists under conditions of heme limitation. The putative hus locus would be significantly up-regulated, and augmented husA expression would be critical for successful establishment of P. gingivalis in subgingival plaque. Antibodies against HusA were detected in the sera of periodontitis patients with P. gingivalis infection, indicating that HusA is expressed by P. gingivalis in vivo. Conservation of HusA function is indicated by alignment of HusA sequences from the annotated genomes of P. gingivalis strains W83 and ATCC33277 with 99% identity detected at DNA and protein levels.
In the host, hemoglobin is the major source of heme. Compared with other hemoproteins, hemoglobin is utilized more efficiently by P. gingivalis under heme-limiting conditions (41). The concentration of heme in healthy gingival fluid is low (57), and the ability to scavenge heme from limited hemoproteins, particularly hemoglobin, appears to be critical for the colonization of this organism in vivo. However, HusA cannot remove monomeric heme directly from hemoglobin or serum albumin (data not shown), a finding compatible with the observed preference of HusA for dimeric heme. P. gingivalis has been reported to produce -oxo bisheme from hemoglobin through the action of a family of cysteine proteases, the gingipains. These comprise arginine-specific RgpA, RgpB, and lysine-specific Kgp (58). RgpA has been reported to induce formation of methemoglobin containing Fe 3ϩ , resulting in a decreased affinity of hemoglobin for the prosthetic heme molecule (59). This enhances degradation of hemoglobin by Kgp and the subsequent release of monomeric heme for capture by the hemagglutinin domains of this proteinase as -oxo bisheme (60). Bisheme formation has been confirmed to be promoted by the HA2 hemagglutinin domain within Kgp (12). Several heme receptors have been reported in P. gingivalis; however, none of these has been shown to act as a bisheme scavenger. Here, we showed that HusA has a preference for dimeric heme with high affinity and may serve as the predominant bisheme chelating protein under low heme growth conditions. Subsequently, holo-HusA is proposed to transport heme directly to a specific receptor, HusB, an integral outer membrane protein, for uptake by the organism. A  Under heme limitation, P. gingivalis up-regulates Kgp expression to degrade host methemoglobin and release monomeric heme. The hemagglutinin domains of Kgp convert heme monomer into heme -oxo dimer. Tethered HusA captures bisheme that is delivered to HusB, a predicted TonB-dependent integral outer membrane protein (OM), for transport of heme into the periplasm. In addition, P. gingivalis may also secrete soluble HusA to compete with other heme sequestering proteins and scavenge available free heme released from other organisms or dying cells. schematic representation of this pathway is proposed in Fig.  8. The model of a hemophore with a surface-attached recipient has precedence in other hemophore-dependent heme uptake systems. For instance, S. marcescens secretes HasA, a hemophore that transfers captured heme to a TonB-dependent outer membrane protein HasR (61). Clearly, further investigations are needed for a better understanding of the mechanism of heme uptake through HusA.
In conclusion, the findings reported here describe a novel hemophore-like protein, HusA, which essentially mediates growth of P. gingivalis under conditions of limited heme availability. To our knowledge, it is the first report of a high affinity bisheme-binding protein in P. gingivalis, which indicates unique structural and functional characteristics that may offer potential targets for future therapeutic intervention strategies.