Hemin binding by Porphyromonas gingivalis strains is dependent on the presence of A‐LPS

Porphyromonas gingivalis is a Gram‐negative black pigmenting anaerobe that is unable to synthesize heme [Fe(II)‐protoporphyrin IX] or hemin [Fe(III)‐protoporphyrin IX‐Cl], which are important growth/virulence factors, and must therefore derive them from the host. Porphyromonas gingivalis expresses several proteinaceous hemin‐binding sites, which are important in the binding/transport of heme/hemin from the host. It also synthesizes several virulence factors, namely cysteine‐proteases Arg‐ and Lys‐gingipains and two lipopolysaccharides (LPS), O‐LPS and A‐LPS. The gingipains are required for the production of the black pigment, μ‐oxo‐bisheme {[Fe(III)PPIX]2 O}, which is derived from hemoglobin and deposited on the bacterial cell‐surface leading to the characteristic black colonies when grown on blood agar. In this study we investigated the role of LPS in the deposition of μ‐oxo‐bisheme on the cell‐surface. A P. gingivalis mutant defective in the biosynthesis of Arg‐gingipains, namely rgpA/rgpB, produces brown colonies on blood agar and mutants defective in Lys‐gingipain (kgp) and LPS biosynthesis namely porR, waaL, wzy, and pg0129 (α‐1, 3‐mannosyltransferase) produce non‐pigmented colonies. However, only those mutants lacking A‐LPS showed reduced hemin‐binding when cells in suspension were incubated with hemin. Using native, de‐O‐phosphorylated and de‐lipidated LPS from P. gingivalis W50 and porR strains, we demonstrated that hemin‐binding to O‐polysaccharide (PS) and to the lipid A moiety of LPS was reduced compared with hemin‐binding to A‐PS. We conclude that A‐LPS in the outer‐membrane of P. gingivalis serves as a scaffold/anchor for the retention of μ‐oxo‐bisheme on the cell surface and pigmentation is dependent on the presence of A‐LPS.

Black-pigmenting Bacteroides species have been shown to degrade plasma proteins involved in the transport and conservation of body iron, namely albumin, hemopexin, haptoglobin, and transferrin, to varying degrees, with Bacteroides gingivalis (P. gingivalis) being the most effective. 9 The cysteine protease Lys-gingipain (Kgp) of P. gingivalis can cleave free hemoglobin, 10 haptoglobin, hemopexin, and transferrin in human serum but was not able to degrade hemoglobin, or the β-chain of haptoglobin when these were present in a haptoglobin-hemoglobin complex in serum. 11 Porphyromonas gingivalis possesses additional outer membrane proteins that are important in the binding and transport of heme and which form part of the hmu hemin-uptake locus, 12 namely HmuY, 13 and HBP35 protein has been described as an important heminbinding protein. 14 Several putative TonB-dependent outer-membrane receptors have been described including Tlr, 15 IhtA (iron heme transport), 16 HmuR (hemin utilization receptor), 6,17,18 and HemR (heminregulated receptor). 19 HmuR exhibited amino acid sequence homology to TonB-dependent receptors involved in heme, vitamin B12 or iron-siderophore transport in other bacteria. 17 A P. gingivalis hmuR isogenic mutant strain was shown to have impaired growth on hemin and hemoglobin as sole source of iron and showed decreased ability to bind hemin and hemoglobin. Escherichia coli cells overexpressing P. gingivalis HmuR as well as purified recombinant HmuR were able to bind hemin, hemoglobin and serum albumin-hemin complex. 17 Porphyromonas gingivalis W50 produces several virulence factors including gingipain proteases and two lipopolysaccharides (LPSs), namely O-LPS 20 and A-LPS. 21 In this study, we addressed the question whether the high abundance low-affinity hemin-binding site described by Tompkins et al. 22 may be one of the LPS of P. gingivalis. To test this hypothesis, we examined a variety of isogenic mutant strains of P. gingivalis lacking Arggingipains, Lys-gingipain and defective in the biosynthesis of O-LPS and A-LPS for their ability to pigment and to bind hemin not only to whole cells but also to LPS, de-phosphorylated LPS and de-lipidated LPS. Mutant strains of P. gingivalis: porR (PG1138), which is defective in A-LPS synthesis, and galE (PG0347), which synthesizes a truncated O-PS repeating unit of O-LPS, are described in greater detail here in this manuscript. Shoji et al. 23 described the porR mutant strain in P. gingivalis ATCC33277 isolated by transposon and targeted mutagenesis and Gallagher et al. 24 have referred to a porR mutant strain isolated by inactivation of PG1138 in P. gingivalis W50. The results of hemin-binding to the mutant strains of P. gingivalis exhibit a consistent pattern, which suggests that the deposition of μ-oxo-bisheme on the cell surface of the P. gingivalis strains appears to be related to the synthesis/presence of A-LPS in the outer leaflet of the outer membrane. We propose that the presence of A-LPS serves as a matrix for the deposition of μ-oxo-bisheme on the P. gingivalis cell surface.
Ltd (Poole, UK) and were the purest grades available. Nα-acetyl-Lys-p-nitroanilide was obtained from Bachem Feinchemikalein AG (Bubendorf, Switzerland). Hemin was obtained from Roche (Burgess Hill, UK). Restriction and modification enzymes were purchased from New England BioLabs (Ipswich, MA, USA), and DNA purification reagents were obtained from Qiagen (Hilden, Germany).

| Bacterial strains and growth conditions
Porphyromonas gingivalis W50 and mutant strains were grown either on blood agar plates containing 5% defibrinated horse blood or in brain-heart infusion broth (Oxoid, Basingstoke, UK) supplemented with hemin (5 μg/mL) in an anaerobic atmosphere consisting of 80% N 2 , 10% H 2 and 10% CO 2 . 25 Clindamycin HCl and tetracycline HCl were added to 5 μg/mL and 1 μg/mL respectively, for selection of ermF and tet Q in P. gingivalis. Ampicillin (Na + salt; 100 μg/mL) or erythromycin (300 μg/mL) was added to the growth medium to select for pUC-derived or ermAM-containing plasmids respectively, in Escherichia coli.

| Generation of P. gingivalis mutants
Purification and general manipulation of DNA, restriction mapping of plasmids and transformation of E. coli were as described previously. 25,26 A list of P. gingivalis strains used in this study is shown in the Table   S1.
For the generation of P. gingivalis mutant strains porR and galE, chromosomal DNA from P. gingivalis W50 was used as the template for amplification/cloning purposes. The nomenclature originally used by TIGR is used throughout the manuscript. The genes encoding UDPglucose-4-epimerase galE and porR 23,24,27,28 in P. gingivalis W50 were insertionally inactivated with ermF-ermAM by allelic exchange following electro-transformation and are described in detail in the Figure S1.
The primers used in this study are listed in detail in Supplemental Methods.

| Measurement of enzyme activity
Arg-gingipain and Lys-gingipain activities in whole cultures and culture supernatants of P. gingivalis and isogenic mutant strains were measured using N-benzoyldl-arginine-p-nitroanilide (dl-BRpNA) and Nα-acetyll-lysine-p-nitroanilide (l-AcKpNA) respectively as substrates, in spectrophotometric assays, as previously described. 29 Units of enzyme activity are expressed as change in absorbance at 405 nm/min per optical density at 600 nm (OD 600 ) at 30°C. Enzyme activities were usually measured in triplicate using batches of bacterial cultures grown on different days. Student's t test for paired samples was used and the data were considered to be significant at a P value <.05.

| SDS-PAGE and SDS-Urea-PAGE
Sodium dodecyl sulfate (SDS)-urea-polyacrylamide gel electrophoresis (PAGE) of LPS was performed according to Inzana and Apicella. 30 Samples were transferred onto nitrocellulose membranes and probed with MAb1B5, which recognizes the epitope Manα1-2-Manα1-phosphate fragment in A-PS of A-LPS, as described previously. 27 Silver staining of gels was performed using the Silver staining kit (Sigma-Aldrich Co. Ltd.) according to the manufacturer's instructions.

| Isolation of LPS and Lipid A
Lipopolysaccharide from P. gingivalis W50 and mutant strains for use in SDS-urea-PAGE experiments was prepared using an LPS extraction kit from Intron Biotechnology (South Korea).
Lipopolysaccharides used in hemin-binding studies was prepared as described previously. 31 De-O-phosphorylated LPS samples used in hemin-binding experiments were prepared by dissolving LPS (10-15 mg) in 0.5 mL of 48% aqueous hydrofluoric acid at 4°C and incubating at 4°C for 16 hours. Excess hydrofluoric acid was removed by dialysis against distilled water (6000-8000 MWCO tubing) at 4°C followed by freeze-drying.
De-lipidation of LPS samples was carried out by treatment with 1.5% aqueous acetic acid at 100°C for 2-4 hours in a heating block.
Insoluble lipid A and traces of undegraded LPS were removed by ultracentrifugation at 30,000 g for 30 minutes at 10°C. The water-soluble supernatant was lyophilized twice to remove all traces of acetic acid.

| Hemin binding to whole cells of P. gingivalis
Porphyromonas gingivalis W50 and mutant strains were grown for 48 hours and cells were harvested by centrifugation (13 300 g) for 20 minutes at 4°C in Eppendorf tubes. The cells were washed with ice-cold sterile PBS (3×1 mL) and stored at −70°C until required. for each set of experiments. The reaction mixture was centrifuged at 13,300 g for 20 minutes at 4°C, the supernatant was transferred to 1-mL plastic disposable cuvettes and the OD 400 was measured.
Concentration of hemin in the supernatant was calculated from standard curves for hemin. The hemin bound (μg/OD 600 of cells) was equal to the difference between the values for the control samples (hemin solution with no added cells, zero binding) and the supernatant from the experimental samples (bound hemin). The standard deviation was calculated.
For statistical analysis, a Student's t test for paired values was used, and data were considered to be significant at a P value <.05.

| RESULTS
The P. gingivalis mutant strains used in this study have been described elsewhere (see Table S1). These include strains in which the genes encoding the proteases Rgps (rgpA/rgpB) and Kgp (kgp) have been inactivated, leading to loss of Arg-gingipains and Lys-gingipain, respectively, 33
Also shown is the P. gingivalis mutant strain wbpB, which has been described in detail elsewhere 28,37 and gives non-pigmented colonies on blood agar plates.

| Cross-streaking experiments
Porphyromonas gingivalis W50 was initially streaked on a blood agar plate and following the formation of a zone of hemolysis (3 days), the cells were removed with a swab containing clindamycin to suppress regrowth of the wild-type strain and the plates were cross-streaked with P. gingivalis mutant strains ( Figure 3). Although rgpA/rgpB and kgp give brown and non-pigmenting colonies when grown on blood agar plates because of the lack of Rgps and Kgp, respectively, they do pigment when cross-streaked on plates on which P. gingivalis W50 has been previously grown and caused hemolysis ( Figure 3). This suggests that rgpA/rgpB and kgp have the ability to pigment if supplied with externally added hemin. However, cross-streaking of P. gingivalis porR, waaL, wzy, and pg0129 strains on blood agar plates as above did not cause the deposition of hemin/black pigment on the surfaces of these cells ( Figure 3). This indicates that the mutant strains are unable to harness any available hemin in the environment and retain it on their cell surface.

| Analysis of LPS
The SDS-urea-PAGE followed by silver staining of LPS purified from P. gingivalis W50 and mutant strains rgpA/rgpB, kgp, porR, and galE show the characteristic laddering pattern ( Figure 4A). However, in porR and galE, the O-LPS shows a higher intensity of bands in the core-, core-plus one repeating unit and core-plus two repeating units ( Figure 4A). In the P. gingivalis galE mutant strain, the O-PS repeating

| Hemin binding by whole cells
Hemin binding by whole cells of P. gingivalis W50, rgpA/rgpB, kgp, galE, porR, waaL, wzy, and pg0129 was measured as described in the Methods section and the results obtained are shown in Figure 5.
Although kgp was non-pigmenting on blood agar plates due to the absence of Kgp, it shows hemin binding (6.1 μg/OD 600 ) when supplied with hemin (also observed when kgp is cross-streaked on blood agar plates on which P. gingivalis W50 was previously grown ( Figure 3). The P. gingivalis mutant strains porR, waaL, wzy, and pg0129, which were non-pigmenting on blood agar plates, were able to bind between ~2.5 and 3.7 μg of hemin/OD 600 of cells, which is ~45%-65% of hemin bound by the parent W50 strain.

Hence, P. gingivalis mutant strains that do not synthesize A-LPS
show reduced hemin binding.

| Hemin binding by LPS
Hemin binding to LPS isolated from P. gingivalis W50 and mutant strain porR grown in brain-heart infusion were measured at two dif-  oxyhemoglobin. 34 HRgpA, the dimeric isoform of RgpA, promotes the formation of methemoglobin from oxyhemoglobin, which is degraded by Kgp to form the black pigment μ-oxo-bisheme. 10 Figure 5). Therefore, we propose that the ability of the bacterial cells to bind hemin may parallel the retention of μ-oxo-bisheme on the cell surface when the strains are grown on blood agar plates. The inability of P. gingivalis mutant strains porR, waaL, wzy, and pg0129 to produce black-pigmented colonies in cross-streaking experiments is supported by the reduced binding of hemin to cells of these strains.
The major difference between the P. gingivalis mutant strains, which have the ability to acquire μ-oxo-bisheme (rgpA/rgpB and kgp) on cross-streaking and those mutant strains that lack this property (porR, waaL, wzy, and pg0129) is the production of A-LPS by pigmenting strains.
This suggests that the P. gingivalis cell surface must contain a molecule that provides a scaffold/matrix for the deposition and retention of any hemin or pigment that is produced/acquired by the organism. Grenier reported that the lipid A component of LPS mediated the binding of uncomplexed hemin by P. gingivalis. 39 As hemin is a lipophilic molecule, it would be expected to bind to Lipid A/LPS.
Escherichia coli, which does not require exogenous heme when grown in iron-replete conditions, was shown to bind as much uncomplexed hemin as Prevotella intermedia. This effect was inhibited by albumin, which indicated that when heme is provided in the free form, most of it binds to the bacterium with an affinity lower than that for albumin. However, Tompkins et al. 22 concluded that most gram-negative bacteria would exhibit similar non-specific hemin binding and that the LPS-mediated hemin binding is probably not biologically relevant because of the low affinity of the interaction and the presence of large amounts of host plasma proteins, which function to counter the lipophilic disposition of hemin. They showed that treatment of P. gingivalis cells with pronase caused a slight reduction in, but did not eliminate, hemin-binding and the authors suggested that this was probably due to the pronase-sensitive hemin-binding sites not being exposed on the surface of the cell and therefore not digested by pronase treatment. 22 However, it seems more plausible that it is the presence of A-LPS (which is not sensitive to pronase treatment) on the surface of P. gingivalis that acts as a site for the deposition/ binding of hemin.
Studies on hemin binding to whole cells of P. gingivalis W50 and mutant strains and hemin binding to native LPS and de-lipidated LPS from P. gingivalis W50 and porR strains show that absence of A-LPS causes a reduction in hemin binding. Hence, absence of A-LPS in the extracellular surface of P. gingivalis eliminates or reduces a scaffold/ anchoring mechanism not only for retention of Arg-and Lys-gingipains but also for the deposition of μ-oxo-bisheme pigment or hemin derived from the environment and highlights the importance of A-LPS in the virulence of this organism.