Iron-Regulated Surface Determinant (Isd) Proteins of Staphylococcus lugdunensis

Abstract

Staphylococcus lugdunensis is the only coagulase-negative Staphylococcus species with a locus encoding iron-regulated surface determinant (Isd) proteins. In Staphylococcus aureus, the Isd proteins capture heme from hemoglobin and transfer it across the wall to a membrane-bound transporter, which delivers it into the cytoplasm, where heme oxygenases release iron. The Isd proteins of S. lugdunensis are expressed under iron-restricted conditions. We propose that S. lugdunensis IsdB and IsdC proteins perform the same functions as those of S. aureus. S. lugdunensis IsdB is the only hemoglobin receptor within the isd locus. It specifically binds human hemoglobin with a dissociation constant (K_d) of 23 nM and transfers heme on IsdC. IsdB expression promotes bacterial growth in an iron-limited medium containing human hemoglobin but not mouse hemoglobin. This correlates with weak binding of IsdB to mouse hemoglobin in vitro. Unlike IsdB and IsdC, the proteins IsdJ and IsdK are not sorted to the cell wall in S. lugdunensis. In contrast, IsdJ expressed in S. aureus and Lactococcus lactis is anchored to peptidoglycan, suggesting that S. lugdunensis sortases may differ in signal recognition or could be defective. IsdJ and IsdK are present in the culture supernatant, suggesting that they could acquire heme from the external milieu. The IsdA protein of S. aureus protects bacteria from bactericidal lipids due to its hydrophilic C-terminal domain. IsdJ has a similar region and protected S. aureus and L. lactis as efficiently as IsdA but, possibly due to its location, was less effective in its natural host.

INTRODUCTION

Staphylococcus lugdunensis is a species of coagulase-negative staphylococcus (CoNS) first described by Freney et al. in 1988 (23). The bacterium is part of the normal flora of skin of humans. It colonizes several distinct niches primarily in the lower part of the body (7). Standard identification procedures used in clinical microbiology laboratories can allow misidentification of S. lugdunensis as Staphylococcus aureus (58). The bacteria have similar colony morphology, and many strains of S. lugdunensis are hemolytic and express clumping factor activity (38). S. lugdunensis can cause infections similar to those caused by S. aureus ranging from localized to systemic, including skin and soft tissue infections, breast abscesses, osteomyelitis, prosthetic joint infections, pneumonia, or meningitis (4, 8, 33, 35, 73). Most notably S. lugdunensis can cause acute infective endocarditis, which has a high mortality rate of ca. 50% (2, 41, 68). S. lugdunensis infections occur at much lower frequency than those caused by S. aureus, although the numbers may be underestimated. The majority of infections caused by S. lugdunensis are community acquired, occurring in healthy adults, which contrasts with the opportunistic character of other CoNS infections (3, 10, 75, 77).

Understanding the molecular determinants of virulence has been facilitated by the availability of the genome sequences of two strains, N920143 and HKU09-01 (30, 67). S. lugdunensis N920143 coding sequences have 77.8% matches with S. aureus MRSA252, 74.7% matches with S. epidermidis RP62a, and 95.4% matches with HKU09-01 (30). Uniquely for CoNS, S. lugdunensis possesses a cluster of genes with similarity in terms of both sequence and organization to the iron-regulated surface determinant (isd) locus of S. aureus (see Fig. 1A). The isd locus occurs in both sequenced and annotated genomes of S. lugdunensis and is tandemly duplicated in the HKU09-01 (30). In S. aureus the isd genes are induced by iron limitation with promoters that are controlled by the ferric uptake regulator (Fur). The S. aureus Isd system is upregulated in vivo during infection due to iron sequestration in the host (1, 6, 29). The system comprises nine proteins (IsdA-IsdI) whose main task is to bind hemoglobin and to extract heme, which is transported into the cytoplasm to provide a source of iron. Four of the proteins are anchored to cell wall peptidoglycan by either sortase A or sortase B. IsdA, IsdB, and IsdH each contain sortase A recognition sequences (LPXTG). IsdC has a distinct recognition sequence, NPQTN, and is sorted by a type B sortase encoded by srtB, which is itself part of the isd regulon (44). These proteins each contain one or more near iron transporter (NEAT) motifs, which bind to hemoglobin or heme. The NEAT domains are conserved in several Gram-positive bacteria. They consist of ∼125 amino acids forming an IgG-like beta sandwich fold of seven or more β-strands in two β-sheets (20, 25–27, 36).

Fig 1.

Schematic representations of the isd locus and Isd proteins of S. lugdunensis with their comparison to conserved Isd proteins of S. aureus. (A) Comparison of the isd loci of S. aureus and S. lugdunensis. Black lines connect orthologous genes. NEAT domains are indicated by gray boxes. The isdH and isdI genes of S. aureus are located outside the locus, as indicated by circles. (B) Schematic representation of the putative cell wall-anchored proteins IsdB, IsdC, IsdJ, and IsdK of S. lugdunensis and IsdA, IsdB, IsdC, and IsdH of S. aureus. The positions of the NEAT domains were determined from their predicted structures based on homology to available crystal structures of Isd proteins of S. aureus. Signal sequences (S) and putative cell wall-anchoring motifs (W T) are indicated. (C) Scheme representing conservation within specified regions of S. aureus IsdA and S. lugdunensis IsdJ. (D) Table comparison of the amino acid conservation between NEAT domains of (vertical) S. aureus and (horizontal) S. lugdunensis Isd proteins (as indicated). Fields highlighted in gray represent Hb-binding NEAT domains, while white fields show heme-binding NEAT domains. –, Homology of <25% identity between the domain's amino acid chains. The most significant levels of homology are underlined.

The N-terminal NEAT motifs of IsdH (NEAT1 and NEAT2) and IsdB (NEAT1) bind to the haptoglobin-hemoglobin complex and to hemoglobin but cannot bind heme (19, 55, 66). They share 47 to 65% identity and bind to a site on the α-chain of human Hb using conserved aromatic residues within loop 1 (β1-β2) (36). Heme is then removed from Hb and transported to the heme-binding NEAT domains that coordinate heme via a conserved Tyr in β8 (25). The heme-binding NEAT domains of IsdH NEAT3 and IsdB NEAT2 transfer heme to those of IsdA and IsdC and then to the membrane located heme transporter (IsdEF) (26, 40, 49, 76). In the cytoplasm the porphyrin ring is broken by heme oxygenases (IsdG and IsdI) and free iron is released (57).

There is evidence that surface-exposed Isd proteins may have a broader role in S. aureus colonization and pathogenesis (9, 14, 55, 66). The IsdA protein is multifunctional and, in addition to transporting heme, it can promote adhesion to the extracellular matrix and to the surface of desquamated nasal epithelial cells, possibly by binding loricrin and cytokeratin 10 in the cornified envelope of squames (16). It promotes nasal colonization in the cotton rat, survival on human skin by conferring resistance to bactericidal lipids in sebum and survival within phagocytes by contributing to resistance to the oxidative burst (11–15, 17). IsdH was shown to facilitate enhanced conversion of complement opsonin protein C3b to iC3b and C3d enabling S. aureus to avoid neutrophil uptake (70). The IsdB protein promotes S. aureus adhesion to platelets resulting in platelet aggregation due to its ability to bind to the platelet integrin (45).

The objective of the present study was to investigate four putative cell wall anchored Isd proteins of S. lugdunensis—IsdB, IsdC, IsdJ, and IsdK—with respect to their expression, localization, hemoglobin- and heme-binding capabilities, and role in bacterial growth under iron limitation. Finally, the IsdJ protein, which has similarity to IsdA of S. aureus, was tested for its ability to bind to fibrinogen, cytokeratin 10, and loricrin and to confer resistance to killing by linoleic acid.

MATERIALS AND METHODS

Growth conditions, bacterial strains, and plasmids.

The bacterial strains and plasmids used in the present study are listed in Table 1. S. lugdunensis and S. aureus strains were grown in tryptic soy broth (TSB) or RPMI 1640 medium (Sigma) to create iron-restricted conditions (47). Escherichia coli strains XL1-Blue, DC10B, and TOPP10 used for cloning were grown in Luria-Bertani (LB) broth. E. coli TOPP10 used for recombinant protein expression was grown in LB broth or RPMI to purify proteins with low heme content. L. lactis was grown in M17 medium supplemented with 0.5% glucose (GM17) (63). The following antibiotics (Sigma) were added to the media as required: ampicillin (Amp) at 100 μg ml⁻¹, chloramphenicol (Chm) at 10 μg ml⁻¹, tetracycline (Tet) at 2 μg ml⁻¹, and erythromycin (Erm) at 10 μg ml⁻¹.

Table 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristicsa	Source or reference
Strains
Staphylococcus lugdunensis
N920143	Human breast abscess isolate	30
N910319	Human endocarditis isolate	F. Vandenesch (unpublished data)
N910320	Human endocarditis isolate	F. Vandenesch (unpublished data)
N930432	Human endocarditis isolate	F. Vandenesch (unpublished data)
N940025	Human endocarditis isolate	F. Vandenesch (unpublished data)
ΔisdB mutant	isdB-null mutation in N920143	This study
ΔisdC mutant	isdC-null mutation in N920143	This study
ΔisdJ mutant	isdJ-null mutation in N920143	This study
ΔisdK mutant	isdK-null mutation in N920143	This study
Δisd mutant	isd-null mutation in N920143	This study
Δfbl mutant	fbl-null mutation in N920143	This study
Staphylococcus aureus
Newman	Widely used laboratory strain
Newman isdA	Human clinical isolate, Newman isdA::Tn917; Ermr	14
Newman isdA clfA clfB	Frameshift mutations in clfA, clfB::Tetr, and isdA	45
Lactococcus lactisNZ9000	L. lactis subsp. cremoris MG1363 carrying nisin resistance cassette	37
Escherichia coli
XL1-Blue	E. coli cloning host; Tetr	Stratagene
DC10B	A dcm-deficient E. coli DH10B strain	46
TOPP10	E. coli cloning and protein purification host	Invitrogen
Plasmids
pQE30	E. coli cloning and expression vector; Ampr	Stratagene
pQE30isdB	pQE30 encoding residues 45 to 655 of IsdB	This study
pQE30isdC	pQE30 encoding residues 30 to 190 of IsdC	This study
pQE30isdJ(45-610)	pQE30 encoding residues 45 to 610 of IsdJ	This study
pQE30isdJ(45-295)	pQE30 encoding residues 45 to 295 of IsdJ	This study
pQE30isdJ(290-610)	pQE30 encoding residues 290 to 610 of IsdJ	This study
pQE30isdK	pQE30 encoding residues 35 to 426 of IsdK	This study
pGEX-4t-2	E. coli cloning and expression vector; Ampr	Amersham
pGEX-4t-2isdC	pGEX4t-2 encoding residues 30 to 190 of IsdC	This study
pRMC2	Chmr vector allowing aTc-inducible expression	16
pRMC2isdA	Chmr; plasmid for inducible expression of IsdA	This study
pRMC2isdJ	Chmr; plasmid for inducible expression of IsdJ	This study
pNZ8048	L. lactis shuttle vector containing the PnisA promoter and start codon in NcoI site, Chmr; allowing nisin-inducible expression of insert	37
pNZ8048isdJ	pNZ8048 encoding isdJ, for controlled expression of IsdJ in L. lactis	This study
pNZ8048isdA	pNZ8048 encoding isdA insert, for controlled expression of IsdJ in L. lactis	This study
pKOR1	E. coli/Staphylococcus shuttle vector allowing allelic replacement in staphylococci	5
pKOR1ΔisdC	pKORI plasmid for in-frame deletion of isdC	This study
pKOR1ΔisdK	pKORI plasmid for in-frame deletion of isdK	This study
pIMAY	E. coli/Staphylococcus shuttle vector allowing allelic replacement in staphylococci	46
pIMAYΔisdJ	pIMAY plasmid for in-frame deletion of isdJ	This study
pIMAYΔisdB	pIMAY plasmid for in-frame deletion of isdB	This study
pIMAYΔisd	pIMAY plasmid for in-frame deletion of the isd locus	This study

^aChm^r, chloramphenicol resistance; Tet^r, tetracycline resistance; Amp^r, ampicillin resistance; Erm^r, erythromycin resistance.

Expression and purification of recombinant proteins.

DNA fragments encoding Isd proteins were amplified using specific primers (see Table S1 in the supplemental material) and cloned into pQE30 or pGE4t-2 vectors (Table 1). His-tagged and glutathione S-transferase (GST)-tagged proteins were purified by nickel-affinity chromatography using HisTrap HP (GE Healthcare) columns or glutathione-affinity chromatography using GSTrap FF columns (GE Healthcare), as described previously (51). Proteins were dialyzed against phosphate-buffered saline (PBS), concentrated by ultrafiltration (Amicon Ultra; EMD Millipore), and stored at −70°C.

Anti-Isd sera.

Antibodies were raised in rabbits to recombinant IsdB, IsdC, IsdJ(45-610), and IsdK (see Fig. S1 in the supplemental material), and the immunoglobulin fraction was purified. The potency and specificity of the sera was determined by enzyme-linked immunosorbent assay (ELISA). Antisera were used to probe whole cells by dot immunoblotting and cell fractions by Western immunoblotting.

Recombinant IsdB mutants.

Mutations creating amino acid substitutions were introduced in pQE30isdB by sequential primer overlap mutagenesis with Phusion high-fidelity polymerase. Primers used for each step are listed as 13 to 18 in Table S1 in the supplemental material. Complementary forward and reverse primers incorporating a mutation were extended by PCR to produce a mutated plasmid. PCR products were digested with DpnI for 1.5 h at 37°C to eliminate methylated parental DNA and then transformed into E. coli strain XL1-Blue.

Inducible expression of IsdJ and IsdA.

The isdA and isdJ genes were amplified by PCR (see Table S1 in the supplemental material) and cloned into pRMC2 using the primers listed as 19 to 22 in Table S1 in the supplemental material. Plasmids were cloned in E. coli DC10B and subsequently electroporated into S. aureus Newman clfA clfB isdA and S. aureus Newman spa (Table 1) using protocol of Lofblom et al. (42). The plasmids pRMC2isdA and pRMCisdJ allow anhydrotetracycline (aTc)-inducible expression of the full-length IsdJ and IsdA proteins in S. aureus. To induce the expression of the proteins, aTc (1 μg ml⁻¹) was added to exponentially growing cultures.

The isdJ gene was amplified (primers 23 to 24; see Table S1 in the supplemental material) and cloned into pNZ8048, a nisin-inducible expression vector. Plasmid pNZ8048isdJ was cloned into electrocompetent L. lactis NZ9000. Induction of IsdJ expression was carried out by adding nisin (300 μg ml⁻¹) to exponentially growing cultures, followed by incubation for 4 h.

Isolation of isd mutants by allelic replacement.

DNA fragments comprising 500 to 1,000 bp upstream from the start codon of each gene and 500 to 1,000 bp downstream from the stop codon (see the primers in Table S1 in the supplemental material) were cloned into either pKOR1 or pIMAY in E. coli DC10B (Table 1) (5, 46). Plasmids were electroporated into S. lugdunensis N920143 (42). Allelic exchange was performed by homologous recombination initiated by switching the bacterial growth temperature from that optimal for plasmid replication (28°C) to the nonpermissive temperature of 37°C (pIMAY) or 42°C (pKOR1). Detailed protocols were described previously by Bae et al. (pKOR1) and Monk et al. (pIMAY) (5, 46).

Cell fractionation.

Cultures of S. lugdunensis were harvested by centrifugation (4,000 × g, 10 min), washed in PBS, and adjusted to an optical density at 600 nm (OD₆₀₀) of 20. A 1-ml portion of the bacterial suspension was pelleted and resuspended in 200 μl of digestion buffer (50 mM Tris-HCl, 20 mM MgCl₂, 30% [wt/vol] raffinose; pH 7.5) containing complete mini-EDTA-free protease inhibitors (Roche). Cell wall proteins were solubilized by digestion with lysostaphin (500 μg ml⁻¹) at 37°C for 30 min. Protoplasts were harvested by centrifugation (5,000 × g, 15 min), and the supernatant was retained as the cell wall fraction. Protoplast pellets were washed once in digestion buffer, sedimented (5,000 × g, 15 min), and resuspended in ice-cold lysis buffer (50 mM Tris-HCl [pH 7.5]) containing protease inhibitors and DNase (80 μg ml⁻¹). Protoplasts were lysed on ice by vortexing. Complete lysis was confirmed by phase-contrast microscopy. The membrane fraction was obtained by centrifugation at 18,500 × g for 1 h at 4°C. The pellet (membrane fraction) was washed once and resuspended in ice-cold lysis buffer. Both fractions were centrifuged again under the same conditions, and the pellets were resuspended in 200 μl of lysis buffer and retained as the membrane fraction. Proportional volumes of culture supernatants were concentrated using Amicon Ultra centrifugal filter units of 10,000 K (Merck/Millipore) and resuspended at a final volume of 200 μl to maintain proportionally concentrated fractions. Fractions were mixed with sample buffer 1:1 (Laemmli) and separated by SDS-PAGE. Cellular fractionation of S. aureus and L. lactis was performed as described previously (18, 60).

Western immunoblotting and fibrinogen- affinity blotting.

Proteins or cellular fractions were separated by SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes (Roche) for 1 h at 100 V using a wet transfer cell (Bio-Rad). Membranes were incubated for 1 h at 37°C or overnight at 4°C in TS buffer (10 mM Tris-HCl [pH 7.4], 150 mM NaCl) containing 10% skimmed milk (Marvel). Primary antibodies or fibrinogen were diluted in 10% Marvel TS, followed by incubation with the membranes for 1 h at room temperature with shaking. Unbound antibody was removed by three 15-min washes with TS buffer. The horseradish peroxidase (HRP)-conjugated secondary reagents were incubated with the membranes for 1 h at room temperature with shaking. Unbound secondary reagent was removed by three washes with TS buffer, and bound HRP was developed using chemiluminescent substrate (LumiGlo; New England BioLabs) in an ImageQuant LAS imager (GE Healthcare).

Whole-cell and protein dot immunoblotting.

Bacterial cells were washed twice in PBS and adjusted to an appropriate OD. Each doubling dilution (5 μl) of the bacterial suspensions or soluble protein (5 μl) was dotted onto a nitrocellulose membrane (Protran). The membranes were blocked for 1 h with Marvel TS buffer, followed by detection performed as for the Western blotting described above.

Heme content.

The absorbance spectra of recombinant proteins were recorded on a Shimadzu UV-160PC spectrophotometer. The heme content of proteins was estimated from the reduced pyridine hemochrome spectra (61). The protein concentrations were estimated by using BCA protein assay reagent (Thermo Scientific/Pierce).

Hemin transfer from IsdB to IsdC.

Recombinant IsdB, purified from E. coli grown in RPMI, was gradually supplemented with porcine hemin (Sigma) dissolved in dimethyl sulfoxide at 10 mM. Insoluble material was removed by centrifugation. The mixture was loaded onto calibrated Sephadex G-25 (GE Healthcare), and chromatography was performed to separate the saturated IsdB from unbound hemin. IsdB (hemin donor) and IsdC-GST (hemin acceptor) were mixed at 10 and 20 μM, respectively. After 4 h of incubation at 22°C, the mixture was loaded onto a GST column to separate GST-tagged IsdC from IsdB. Separation of the two proteins was monitored by SDS-PAGE. The spectra of the separated donor and acceptor were recorded to assess the hemin transfer.

Peroxidase (TMBZ) staining.

Detection of heme-protein complexes was accomplished by staining gels with 3,3′,5,5′-tetramethylbenzidine (TMBZ; Sigma), a chromogen that turns blue in areas of heme-associated peroxidase activity (64).

ELISA.

Hemoglobin-haptoglobin (Hb-Hpt) complexes were prepared by mixing human plasma haptoglobin (Molecular Innovations) with hemoglobin (Pointe Scientific) at a 1:1 molar ratio for 1 h at 22°C. Hemoprotein binding assays were performed in two ways. (i) For the first type of assay, microtiter plates (Nunc) were coated with recombinant Isd proteins at 2 μM in 100 μl of 50 mM sodium carbonate buffer (pH 9.5) overnight at 4°C. Coated wells were washed three times with PBS and blocked with 5% Marvel in PBS for 2 h at 37°C. After three washes with PBS, various concentrations of methemoglobin (MetHb), hemoglobin (Hb), Hb-Hpt complexes, or haptoglobin (Hpt) were added to the wells in 2% Marvel PBS, followed by incubation for 2 h at 37°C. Wells were washed with PBS, and bound proteins were detected by incubation with rabbit anti-human Hb serum (Dako) or rabbit anti–human Hpt serum (Sigma) in 2% Marvel PBS for 1 h at 37°C. (ii) For the second type of assay, reciprocal binding experiments were performed by coating microtiter (Nunc) wells with MetHb, Hb, Hb-Hpt, Hpt, or myoglobin (Mb) at 1 μM in 100 μl of 50 mM sodium carbonate buffer (pH 9.5) overnight at 4°C. After blocking with 5% Marvel PBS, the wells were incubated with increasing amounts of recombinant Isd proteins in 2% skimmed milk in PBS for 2 h at 37°C. Bound proteins were detected by incubation with the appropriate polyclonal rabbit anti-Isd IgG in 2% Marvel PBS for 1 h at 37°C and washed three times with PBS. To detect specific interactions in both types of assays, HRP-conjugated goat anti-rabbit IgG (Dako) in 2% Marvel PBS was added for 1 h at 37°C. After three washes, 100 μl of TMBZ (Sigma) at 0.1 mg ml⁻¹ prepared in 0.05 M phosphate citrate buffer containing 0.006% (vol/vol) hydrogen peroxide was added, and the plates were developed for 5 min. The reaction was stopped by the addition of 50 μl of 2 M H₂SO₄ to each well. Plates were read at 450 nm using a Multiscan ELISA plate reader. Half-maximum binding values reported for all ELISA-type binding assays were predicted using GraphPad Prism software.

Human GST-tagged loricrin (GST-loricrin; 1 μM), GST-cytokeratin 10 (1 μM) (71), GST (1 μM), and human fibrinogen (10 μg ml⁻¹; Calbiochem) in 50 mM sodium carbonate buffer (pH 9.5) were coated onto microtiter wells (Nunc) overnight at 4°C. All plates were blocked with 5% bovine serum albumin (BSA) in PBS for 2 h at 37°C. After three washes with PBS, various concentrations of recombinant IsdJ in 1% Marvel PBS were added, followed by incubation at 37°C for 2 h. After three washes with PBS, bound protein was detected by incubation with rabbit anti-IsdJ IgG in 1% Marvel PBS at 37°C. After 1 h of incubation, the wells were washed three times with PBS, followed by incubation with protein A-peroxidase conjugated to HRP (Dako) in 1% Marvel. After incubation for 1 h at 37°C, the wells were washed three times with PBS, and bound HRP was detected by using TMBZ as described above. Purified human GST-loricrin and GST-cytokeratin 10 were kindly provided by Michelle Mulcahy.

Bacterial adherence to immobilized ligands.

Microtiter plates were coated with doubling dilutions of fibrinogen, GST-cytokeratin 10, and or GST alone in PBS overnight at 4°C. Control wells contained PBS only. After washing (with PBS), the plates were blocked for 2 h at 37°C with 5% (wt/vol) BSA in PBS. After three washes, 100 μl of a cell suspension (an OD₆₀₀ of 1 in PBS) was added to each well, and the plates were incubated for 2 h at 37°C. After three washes, adherent cells were fixed by adding 100 μl of 25% formaldehyde, followed by incubation at room temperature for at least 20 min. The plates were then washed and stained with crystal violet for 2 min. After several washes, 100 μl of 5% (vol/vol) acetic acid was added to the wells to solubilize the crystal violet. Absorbances were read in using an Multiscan ELISA plate reader at 570 nm.

SPR.

Surface plasmon resonance (SPR) analysis of IsdB-hemoglobin binding was performed using the BIAcore X100 system (GE Healthcare). Recombinant IsdB was captured and amine coupled to monoclonal anti-His antibodies (R&D Systems) on a flow cell. The antibodies at 50 μg ml⁻¹ in sodium acetate buffer (pH 5.0) had been previously covalently immobilized on both flow cells of a CM5 chip, using amine coupling. This was performed according to the manufacturer's protocol, using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride, followed by N-hydroxysuccinimide and ethanolamine. Increasing concentrations of analytes, such as methemoglobin (34 kDa) or hemoglobins (68 kDa) in running buffer (PBS [pH 7.4], 0.005% P20) were flowed over immobilized IsdB and a reference flow cell (coated only with anti-His antibody) at a rate of 5 μl min⁻¹. The sensorgram data presented were subtracted from the corresponding data on the reference flow cell and from the response generated by injection of buffer over the chip. We used affinity analysis (BIAevaluation Software) to determine the dissociation constants (K_d). Purified human and mouse hemoglobins were kindly provided by Gleb Pishchany and Eric Skaar.

Growth in modified RPMI supplemented with Hb.

The protocol by Pishchany et al. was used, with minor modifications (55). A single colony of S. lugdunensis was inoculated into RPMI and grown overnight. The concentration of cells was normalized to an OD₆₀₀ of 3.0, and bacteria were sedimented (4,000 × g, 10 min) and resuspended in 1 ml of NRPMI containing 0.25 mM the iron chelator ethylenediamine-di(o-hydroxyphenylacetic acid (EDDHA; LGC Standards, GmbH). NRPMI is RPMI containing 1% Casamino Acids, treated with Chelex 100 (Sigma) at 20 g liter⁻¹, and supplemented with 25 μM ZnCl₂, 25 μM MnCl₂, 100 μM CaCl₂, and 1 mM MgCl₂. The S. lugdunensis suspension was subcultured (1:50) into 5 ml of NRPMI + 0.25 mM EDDHA with hemoglobin at 10 μg ml⁻¹. Cultures were incubated at 37°C in 15-ml conical tubes with shaking at 40 rpm. OD₆₀₀ measurements were taken at the indicated time points.

Bacterial killing by linoleic acid.

Bacteria were harvested by centrifugation and washed twice in sterile distilled H₂O (dH₂O). Cell suspensions were adjusted to 10⁸ CFU ml⁻¹ in dH₂O, followed by incubation at 37°C with linoleic acid (Sigma) at 10 μg ml⁻¹. The CFU was determined at 0, 30, and 60 min by viable counting.

In silico structural modeling.

Structural models were obtained using the PHYRE website (34) (Imperial College London, Structural Bioinformatics Group [http://www.sbg.bio.ic.ac.uk/∼phyre/]) and analyzed using the molecular modeling UCSF Chimera software (University of California, Resource for Biocomputing Visualization and Informatics [http://www.cgl.ucsf.edu/chimera/]).

RESULTS

Sequence analysis of S. lugdunensis proteins with NEAT domains.

The predicted amino acid sequences of putative cell wall-anchored Isd proteins of S. lugdunensis N920143 and HKU01-09 were compared to homologous Isd proteins of S. aureus. The IsdB and IsdJ proteins of S. lugdunensis are each predicted to have two NEAT domains, whereas IsdC and IsdK have one (Fig. 1B). IsdB and IsdJ have sortase A recognition sequences (LPATG and LPNTG, respectively), while IsdC and IsdK have sortase B-like recognition sequences (NPQTS and NKQPN, respectively) (Fig. 1B). S. lugdunensis IsdB and IsdC proteins are orthologues of S. aureus proteins with 37 and 57% identities, respectively (Fig. 1A). The identities occur mostly within the NEAT domains. The S. lugdunensis IsdB NEAT1 domain is 60% identical, whereas the NEAT2 domain is 56% identical to S. aureus IsdB NEAT1 and NEAT2 domains, respectively (Fig. 1D). Both IsdJ and IsdK have segments with similarities to IsdA of S. aureus (Fig. 1C and D). The NEAT1 and NEAT2 domains of IsdJ have 48 and 55% identities, respectively, to the single NEAT domain of IsdA, whereas the IsdJ NEAT2 and C-terminal residues 457 to 646 have 39% identities to IsdA 63-350 (Fig. 1CD). The IsdK NEAT domain is 36% identical and 60% similar to the IsdA NEAT domain (Fig. 1D).

The NEAT domains of S. lugdunensis Isd proteins were modeled in three dimensions based on known X-ray crystal structures of the highly conserved NEAT domains of S. aureus. IsdA (PDB ID 2ITE) was used to model the IsdJ NEAT1 and NEAT2 domains and the IsdK NEAT domain (27). S. aureus IsdH NEAT3 (PDB ID 2E7D) and IsdB NEAT2 (PDB ID: 3RTL) were used to model S. lugdunensis IsdB NEAT2 (24, 72). The NEAT domain of IsdC was modeled on its orthologue from S. aureus (PDB ID 2O6P) (59, 69).

Each of the putative heme-binding NEAT domains was predicted to be composed of an IgG-like β-sandwich fold with eight anti-parallel β-strands. Each has conserved residues within the predicted hydrophobic heme-binding pocket (Fig. 2A). A key metal coordinating Tyr is conserved in β8. Conserved Lys, Ser, and Met residues, which have been shown to form hydrogen bonds with the propionate groups of bound heme, occur in a predicted helix linking β1b and β2 (20, 25, 27, 54). Hydrophobic residues involved in capturing the porphyrin occur also in β1-β2 (Tyr) and β7 (Val, Ile, Val, or similar) (Fig. 2A) (24, 25). S. aureus IsdC is unique in using Tyr and Ile in β4 to make van der Waal's contacts with heme. These residues are present in the S. lugdunensis orthologue (Fig. 2A), along with a five-residue insertion in loop β7-β8 that is identical to that in IsdC of S. aureus (Fig. 2A) (59, 69).

Fig 2.

Multiple sequence alignments of the NEAT domains of S. lugdunensis and S. aureus. (A) Alignment of the putative heme-binding NEAT domains of S. lugdunensis (IsdB NEAT2, IsdC, IsdJ NEAT1, IsdJ NEAT2, and IsdK) and their structural predictions based on the homology to available crystal structures of S. aureus NEAT domains of IsdA, IsdC, and IsdH. Underlined residues represent fragments predicted to form β-strands, as indicated above the alignment. (B) Multiple sequence alignment of the putative hemoglobin-binding NEAT domain S. lugdunensis IsdB NEAT1 with hemoglobin-binding NEAT domains of S. aureus (IsdB NEAT1, IsdH NEAT1, and NEAT2). Residues predicted to form β-strands are underlined. Conserved residues are indicated by asterisks (*), Colons (:) mark at least two identical residues across the alignment. Residues predicted to be involved in heme capture are displayed in boldface.

The NEAT1 domain of IsdB from S. lugdunensis is 56% identical to the IsdH NEAT1 domain of S. aureus (Fig. 1D). Therefore, the available crystal structures of IsdH NEAT1 (PDB ID 3SZK, 3OVU, and 2H3K) were used to model the domain, allowing the prediction that it binds to hemoglobin (36, 54). The NEAT1 domain of IsdB from S. lugdunensis is 60% identical to the NEAT1 domain of S. aureus IsdB and 59% identical to the NEAT 2 of IsdH (Fig. 1D). The structural model presented a nine-stranded distorted β-barrel with an immunoglobulin-like fold. A characteristic helix-containing loop occurs between β1 and β2 which contains a string of aromatic residues that are known to be crucial for binding to α-chain of hemoglobin (Fig. 2B).

Cellular localization of Isd proteins in S. lugdunensis.

Specific antibodies were used to detect surface-exposed proteins of S. lugdunensis using a whole-cell dot blot method. S. lugdunensis N920143 was grown to stationary phase in TSB containing iron (48) or iron-restricted RPMI with or without added FeCl_3. Whole cells were dotted onto membranes beginning at an OD₆₀₀ of 1 for probing with anti-IsdC and anti-IsdB sera or an OD₆₀₀ of 50 for IsdK and IsdJ. IsdB, IsdC, and IsdK were detected on the surfaces of cells grown in RPMI but not on TSB-grown cells, indicating that they were only expressed in iron limited conditions (Fig. 3A to C). The addition of increasing concentrations of FeCl₃ to RPMI resulted in decreasing levels of the proteins being expressed. FeCl₃ at 1 mM completely abolished the expression of IsdC and IsdK, whereas 2 mM FeCl₃ inhibited the production of IsdB (Fig. 3A to C).

Fig 3.

Expression of the IsdB, IsdC, IsdJ, and IsdK proteins by S. lugdunensis under iron-restricted and iron-rich conditions. (A, B, and C) Whole-cell dot immunoblot to detect surface displayed of Isd proteins of S. lugdunensis grown in either RPMI, RPMI supplemented with various concentrations of FeCl₃ (as indicated), or TSB. (D) Western immunoblotting was performed to detect the expression of Isd proteins in cellular fractions of S. lugdunensis N920143 grown in either TSB or RPMI. Lanes marked “CW” contain solubilized cell wall fraction. Lanes marked “M” contain membrane fractions. Dot and Western immunoblotting was performed with rabbit serum recognizing IsdB (anti-IsdB) (A and D), IsdC (anti-IsdC) (B and D), IsdJ (anti-IsdJ) (D), and IsdK (anti-IsdK) (C and D). Bound IgG was detected by HRP-conjugated protein A.

Unexpectedly, IsdJ was not detected on the cell surface. In order to determine whether the antisera recognized the N-terminal region (likely to be surface exposed), rIsdJ(45-295) and rIsdJ(295-610) truncates were expressed and purified. Dot immunoblotting and ELISA indicated that both regions were equally immunogenic (see Fig. S2 in the supplemental material), so it can be concluded that the N-terminal region is not exposed on the surface. Since the anti-IsdK serum was as potent in recognizing the recombinant protein as anti-IsdB and anti-IsdC serum (data not shown), it can also be concluded that IsdK is poorly expressed on the surface, whereas IsdB is well exposed (Fig. 3A to C).

Cells grown in TSB and RPMI were suspended in a buffer containing 30% raffinose and protoplasts formed by treatment with lysostaphin. The protoplasts were removed by centrifugation, leaving solubilized cell wall proteins in the supernatant (wall fraction). The protoplasts were lysed and the membranes separated by centrifugation to yield the membrane fraction. Wall and membrane fractions were analyzed by Western immunoblotting (Fig. 3B). Each protein was expressed in RPMI with only very slight expression of IsdB seen in TSB. IsdB and IsdC were predominantly found in the wall fraction, but a significant amount was always found in the membrane fraction. Interestingly, IsdC from the membrane fraction is larger than that from the wall fraction, which indicates that a proportion of IsdC has not been sorted and is retained in the membrane with its hydrophobic and positively charged C terminus (Fig. 3D and 4B). Surprisingly, IsdJ and IsdK were only detected in the membrane fraction and appeared not to be sorted at all (Fig. 3D and 4B). The same pattern of expression of Isd proteins was seen in four other isolates of S. lugdunensis (see Fig. S3 in the supplemental material). This suggests that sortases are defective in anchoring the IsdJ and IsdK, perhaps due to their inability to recognize the proteins' sorting signals.

Fig 4.

Validation of protein expression by S. lugdunensis isd deletion mutants. (A) Expression of surface-exposed IsdB and IsdC by S. lugdunensis deletion mutants demonstrated by whole-cell dot immunoblotting of RPMI-grown cells. (B) Western immunoblotting was performed to detect expression of Isd proteins in the cell wall (CW) and membrane (M) fractions of S. lugdunensis N920143 grown in RPMI. Immunoblotting was performed with rabbit anti-Isd IgG (as indicated), followed by HRP-conjugated protein A.

We investigated the possibility that IsdJ and IsdK are secreted by S. lugdunensis. Wild-type N920143 and the isdK and isdJ mutants were grown in RPMI, culture supernatants were concentrated, and proportional amounts of each fraction were probed with anti-Isd sera. Similar amounts of IsdJ and IsdK were detected in the membrane and supernatant fractions and were missing from cultures of appropriate mutants (see Fig. S4 in the supplemental material). However, the cell wall-anchored IsdB protein was also detected in the supernatants, possibly resulting from autolysis during growth. It is therefore difficult to determine whether IsdJ and IsdK occur in the supernatant due to direct secretion because all three Isd proteins found in the supernatants were smaller, suggesting that proteolytic cleavage had occurred (see Fig. S4 in the supplemental material).

Isd mutants.

Mutations were isolated to delete the isdC, isdJ, isdK, and isdB genes and the entire isd locus (15.174 kb) by allelic exchange using the S. aureus vectors of pKOR1 or pIMAY (Table 1). Each of the mutants was grown in RPMI, and the expression of Isd proteins was measured by whole-cell dot immunoblotting with anti-IsdC and anti-IsdB sera. The data show that the proteins were expressed on the surface at the same level as those on the surface of the wild type and were absent from the appropriate mutants (Fig. 4A). Western immunoblotting of fractionated cells revealed the same pattern of expression as in Fig. 3D with IsdB and IsdC partially sorted to the wall and with IsdJ and IsdK being found only in the membrane fraction (Fig. 4B). The levels of expression of IsdB were the same in the wild type and in the isdC, isdJ, and isdK mutants, and IsdB expression was missing in the isdB mutant and the complete isd deletion mutant. Similar patterns were obtained with IsdC, IsdJ, and IsdK, indicating an absence of polarity (Fig. 4B).

In order to investigate the expression of IsdJ in other Gram-positive bacterial hosts, the isdJ gene was cloned into the nisin-inducible Lactococcus lactis expression vector pNZ8048 and into the anhydrotetracycline-inducible staphylococcal expression vector pRMC2 (Table 1). After induction, expression of IsdJ was detected on the surface of L. lactis and on the surface of S. aureus Newman spa by whole-cell dot immunoblotting (Fig. 5A). Interestingly, the IsdJ protein was found in the cell wall fractions of L. lactis and S. aureus, indicating that it had been sorted in the heterologous host (Fig. 5B). This further suggests that sortase A is defective in S. lugdunensis.

Fig 5.

Cellular localization of IsdJ expressed in S. aureus and L. lactis from the inducible plasmids pRMC2isdJ and pNZ8048isdJ, respectively. (A) Whole-cell immunoblot to detect expression of surface-exposed IsdJ by S. aureus Newman spa and L. lactis NZ9000 and S. lugdunensis. (B) Western immunoblotting was performed to identify the cellular localization of IsdJ expressed by S. aureus and L. lactis. Cell wall (CW) and membrane (M) fractions were separated by SDS-PAGE. Immunoblotting was performed with rabbit anti-IsdJ serum, followed by HRP-conjugated goat anti-rabbit IgG. Experiments were performed three times with similar results.

Absorption spectra of Isd proteins.

The His-tagged recombinant Isd proteins purified from E. coli grown in LB medium were red, indicating that heme was bound. When the absorption spectra of each protein were measured, Soret peaks at 400 nm (IsdB), 406 nm (IsdC), 404 nm (IsdK), and 403 (IsdJ) were detected (Fig. 6A). Both NEAT domains of IsdJ bound heme, with each having Soret peaks at 403 nm (data not shown). It was determined from the spectra of pyridine hemochromes that the occupancy level was 10 to 30%. In contrast, proteins purified from E. coli cells grown in RPMI were not red in color and had only 1 to 3% heme occupancy.

Fig 6.

Absorption spectra of Isd proteins and spectral shifts following coincubation of hemin-saturated IsdB with GST-IsdC. (A) Recombinant His-tagged IsdB, IsdC, IsdJ, and IsdK proteins (at 10 μM) purified from E. coli grown in LB broth. (B) Recombinant IsdB and the Y467A mutant purified from E. coli grown in LB broth. (C) Visible spectra of His-tagged IsdB and GST-tagged IsdC (5 μM) purified from E. coli grown in RPMI and their spectral shifts following IsdB saturation with hemin and hemin transfer to GST-IsdC. (D) Coomassie blue- and TMBZ-stained SDS-PAGE gels with equal amounts of following proteins: apo-IsdB, IsdB saturated with hemin, IsdB after hemin transfer to IsdC, apo-IsdC, and IsdC upon hemin transfer from IsdB. These are representative results of three independent experiments.

Heme transfer.

A conserved Tyr in β8 of the NEAT motif of heme-binding Isd proteins is essential for ligand binding. A substitution Y467A was introduced into NEAT2 of rIsdB of S. lugdunensis which, when purified from E. coli, resulted in a protein that lacked a Soret peak (Fig. 6B) and did not gain it after incubation with porcine hemin (data not shown). This strongly suggests that IsdB uses the same conserved heme-binding mechanism as S. aureus. In S. aureus, heme is stripped from hemoglobin bound to NEAT1 and NEAT2 of IsdH and/or NEAT1 of IsdB and transferred (via IsdA) to IsdC and then to the membrane transporters (40, 49, 76). S. lugdunensis lacks an orthologue of IsdH and IsdB NEAT1 is the only putative hemoglobin-binding NEAT domain among its Isd proteins. Since IsdB and IsdC are the only wall-anchored surface-exposed Isd proteins, we propose that IsdB NEAT1 captures hemoglobin, and the stripped-out heme is then transferred from NEAT2 to IsdC. To test part of this hypothesis, we grew E. coli pQE30isdB in RPMI, purified the apo form of the protein, and saturated it with hemin. This was then incubated with GST-tagged IsdC. The latter was separated from the mixture by glutathione-Sepharose chromatography. The eluted GST-IsdC had an enhanced Soret peak at 406 nm, indicating that the protein had accepted heme transferred from IsdB (Fig. 6C). Equal amounts of the proteins were separated by SDS-PAGE and stained with either Coomassie blue or TMBZ peroxidase, a reagent that turns blue as a result of the hemin-generated peroxides (Fig. 6D).

Isd proteins binding to hemoglobins.

We used solid-phase binding assays and SPR to investigate the binding of recombinant Isd proteins to methemoglobin (MetHb; occurs mostly as an αβ dimer with heme groups that contain oxidated ferric iron that are unable to bind oxygen), human hemoglobin (hHb; an α2β2 tetramer), the hemoglobin-haptoglobin complex (Hb-Hpt), human haptoglobin (Hpt), and human myoglobin (hMb). IsdB is the only Isd protein of S. lugdunensis that bound to soluble hemoproteins (Fig. 7A and B). Human MetHb, Hb, and Hp-Hpt bound to immobilized IsdB dose dependently and saturably, with half-maxima at 8 ± 1.6 nM, 12 ± 2 nM, and 2 ± 0.5 nM, respectively (Fig. 7A and B, data not shown). Binding to haptoglobin or myoglobin was not detected (data not shown). Aromatic residues in the NEAT1 and NEAT2 domains of IsdH (β1-β2 loop) were shown to be critical for binding to α-chains of hHb (36, 54). A mutant of S. lugdunensis recombinant rIsdB (Y189A/H190E/Y191A) was generated and found not to bind to hHb (Fig. 7C) or to Hb-Hpt complexes (data not shown). This demonstrates a conserved mechanism of binding of hemoglobin by IsdB of S. lugdunensis and IsdH of S. aureus. Solid phase assays measured specific binding of rIsdB to hHb, MetHb, and Hb-Hpt with half-maximal binding concentrations in the low nanomolar range. To measure the affinities more accurately, SPR binding analysis was performed. rIsdB was immobilized onto a dextran chip, and the hHb was passed over the surface in concentrations ranging from 9.766 nM to 6 μM (Fig. 8A). Human hemoglobin immediately associated with recombinant IsdB, reaching an equilibrium response for every concentration (Fig. 8A). The steady-state responses against concentrations were fitted with a two-binding-site affinity model, since there are two predicted binding sites (two α-chains) on tetrameric Hb. An accurate fitting was judged from the low χ² value of 9, giving an affinity K_d of 23.13 ± 4 nM. A similar affinity was obtained for human MetHb (data not shown).

Fig 7.

IsdB binding to human hemoglobin (Hb) and hemoglobin-haptoglobin (Hb-Hpt) complexes. Plates (96 well) were coated with recombinant IsdB, IsdC, IsdJ, and IsdK (A and B) or recombinant wild-type IsdB and IsdB Y189A/H190E/Y191A (C) and then incubated with various concentrations of Hb (A and C) or Hb-Hpt complexes (B). Bound proteins were detected with polyclonal rabbit anti-human hemoglobin serum, followed by HRP-conjugated goat anti-rabbit IgG. Values represent the means for triplicate wells. Binding assays were performed three times with similar results.

Fig 8.

Surface plasmon resonance (SPR) binding analysis of human Hb (hHb) (A) and mouse Hb (mHb) (B) to recombinant IsdB. The data were generated using Biacore X-100. IsdB was immobilized on the surface of a CM5 sensor chip, followed by passing increasing concentrations of hHb or mHb over the surface. Injections began at 0 s and ended at 180 s. The data are presented as real-time graphs of response units (RU) against time (left) and as SPR signals against ligand concentrations (right) in the late dissociation phase.

As demonstrated by Pishchany et al., S. aureus IsdB has a higher affinity for hHb than mouse hemoglobin (mHb) (55). We measured S. lugdunensis IsdB binding by mHb using SPR. Purified mHb was passed over the surface in concentrations ranging from 156.25 nM to 40 μM. In contrast to hHb, the mouse Hb binding levels were too low to be evaluated for an accurate affinity K_d since the response levels did not reach the required steady-state equilibrium (Fig. 8B). This demonstrated a low affinity of S. lugdunensis IsdB for mHb, with a K_d predicted at ∼177 μM.

Growth of iron-starved S. lugdunensis is enhanced by hemoglobin.

To test the hypothesis that the Isd system of S. lugdunensis is important for iron acquisition from hemoglobin and that IsdB in particular has a crucial role, bacteria were cultured in a modified RPMI medium that was pretreated with Chelex 100 and contained a divalent metal chelator. S. lugdunensis growth was significantly impaired when cultured in iron-depleted medium by comparison with rich medium (TSB) (data not shown). The addition of hemoglobin resulted in significant growth enhancement starting at 16 h postinoculation and continuing until 48 h postinoculation. Wild-type S. lugdunensis grew to an OD₆₀₀ of 0.86 in 48 h in modified RPMI containing human hemoglobin, while in the absence of hHb the OD₆₀₀ was <0.4 (Fig. 9A and E). Mutants of S. lugdunensis lacking IsdB or the entire Isd locus grew to an OD₆₀₀ of ca 0.6 in the presence of hHb and to <0.4 in its absence (Fig. 9BCD). This demonstrates that the growth of S. lugdunensis is enhanced by hHb in an IsdB-dependent fashion. Nevertheless, the presence of hHb did support growth to a limited extent, perhaps due to similar unspecific binding of bacteria to hemoglobin that was previously observed for S. aureus (66). To confirm the specificity of the role of hHb and IsdB, wild-type S. lugdunensis was grown in modified RPMI supplemented with mouse Hb, which in in vitro studies was showed to bind very weakly to IsdB (Fig. 8B). Wild-type S. lugdunensis grew better when supplemented with human than mouse Hb (Fig. 9E and F), demonstrating the IsdB-dependent preference toward the human hemoprotein. However, the addition of mouse Hb resulted in enhanced growth compared to growth in the medium without added Hb. This growth enhancement initiated by mHb resulted in ODs similar to those of the isdB mutant supplemented with hHb, suggesting that it is caused by the same IsdB-independent mechanism (Fig. 9E and F). This indicates a role for IsdB in binding hemoglobin to support bacterial growth when iron is strictly limited.

Fig 9.

Growth of S. lugdunensis N920143 strains in iron-restrictive medium supplemented with hemoglobins. (A, B, C, and D) Growth of wild-type S. lugdunensis and S. lugdunensis ΔisdB and Δisd locus mutants in modified RPMI medium either supplemented with human Hb (gray columns) or without added Hb (white columns). Changes in the OD₆₀₀ were measured over 48 h. Values are means representing three independent experiments. (D) Mean growth yields (after 48 h) of three independent experiments. (E and F) Growth of wild-type S. lugdunensis in iron-depleted medium supplemented with human Hb (dark gray) or mouse Hb (light gray) or without added Hb (white). Values are means representing three independent experiments. (F) Mean growth yields (after 48 h) of three independent experiments. Error bars represent standard deviations of the OD₆₀₀. Statistically significant differences are indicated (Student two-tailed t test; *, P < 0.05).

Binding of IsdJ to fibrinogen and cytokeratin 10.

The IsdA protein of S. aureus can bind to several ligands other than heme, including fibrinogen, loricrin, and cytokeratin 10, the last two of which are found only in the cornified envelope of desquamated epithelial cells (11, 15). Since significant similarities exist between IsdA and IsdJ (Fig. 1C) of S. lugdunensis, rIsdJ was tested for its ability to bind fibrinogen, keratin 10 and loricrin in solid phase assays. rIsdJ bound to each ligand with half-maxima of ca. 200 nM for loricrin, 250 nM for cytokeratin 10, and 350 nM for fibrinogen, 50- to 100-fold lower than the affinity of rIsdB for hHb (Fig. 10A). Ligand affinity blotting was performed with full-length rIsdJ and the N- and C-terminal truncates, probing with soluble fibrinogen (Fig. 10B). As with IsdA, the N-terminal NEAT-containing segment supported fibrinogen binding.

Fig 10.

Recombinant IsdJ binding to loricrin, cytokeratin 10, and fibrinogen. (A) Plates (96 well) were coated with GST-loricrin, GST-cytokeratin 10, GST, or fibrinogen, followed by incubation with increasing concentrations of recombinant IsdJ(45-610). Bound IsdJ was detected with rabbit anti-IsdJ serum, followed by HRP-conjugated protein A. The results are the means for triplicate wells. (B) Fibrinogen (Fg) affinity immunoblot to detect Fg-binding domain of IsdJ. Lanes contain equal amounts of recombinant IsdJ variants, as indicated. Bound Fg was detected with HRP-conjugated monoclonal anti-human Fg IgG. These results are representative of three independent experiments.

In order to determine whether IsdJ could support bacterial adherence to immobilized fibrinogen, wild-type S. lugdunensis was compared to the fbl and isdJ mutants grown in RPMI. There was no difference in binding by S. lugdunensis and the isogenic isdJ mutant. Moreover, the fbl mutant could not bind to fibrinogen at all, indicating that IsdJ does not contribute to bacterial adherence (see Fig. S5A and B in the supplemental material).

To test if heterologous expression of IsdJ in S. aureus or L. lactis, where the protein is exposed on the cell surface, could support adhesion to immobilized ligands, the IsdJ protein was expressed in S. aureus clfA clfB isdA, a mutant that is defective in all fibrinogen and cytokeratin 10 adhesins, and in L. lactis, which cannot bind these ligands. In neither case could adhesion be detected (data not shown). Either the affinity of binding is too weak to support adhesion or the protein is perhaps partially buried in the cell wall and not fully accessible to the ligand.

IsdJ promotes resistance to killing by linoleic acid.

The multifunctional IsdA protein of S. aureus promotes resistance to killing by bactericidal lipids that occur in sebum due to the C-terminal region that confers decreased hydrophobicity on the bacterial cell (14). Since the C terminus of IsdJ has similarity to IsdA, it was decided to investigate whether the S. lugdunensis protein conferred the same property. Mutants of S. aureus and S. lugdunensis that were defective in IsdA and IsdJ, respectively, were compared to the parental strains for survival in a buffer containing linoleic acid for 30 min. The isdA mutant survived worse than the S. aureus wild type, while there was a trend toward the same effect that was not statistically significant for the S. lugdunensis isdJ mutant (Fig. 11A). To test whether less effective protection by IsdJ could be caused by its cellular localization, we expressed IsdJ on the surface of the S. aureus Newman isdA strain to see whether the cell wall-anchored IsdJ could complement the isdA mutation in the killing assay. IsdA and IsdJ were expressed from the anhydrotetracycline-inducible pRMC2 vector in the S. aureus Newman isdA strain, and bacterial survival was measured after 30 min. The strain expressing IsdA and IsdJ showed increased protection compared to the isdA mutant or the mutant carrying the empty pRMC2 vector, reaching the survival levels of the Newman wild type (Fig. 11B). This suggested that the protein could protect bacterium from killing by the fatty acid when exposed on the bacterial surface. Similarly expression of IsdA or IsdJ in L. lactis promoted enhanced survival (Fig. 11C). Thus, IsdJ appears to confer resistance to linoleic acid and may help to promote the survival of S. lugdunensis in its natural habitat on the human skin.

Fig 11.

Bacterial killing by human skin fatty acid-linoleic acid. (A) S. lugdunensis N920143 wild-type, S. lugdunensis ΔisdJ mutant, S. aureus Newman wild-type, and S. aureus Newman isdA mutant strains were grown in RPMI overnight. (B) S. aureus Newman and S. aureus Newman isdA were grown in RPMI overnight. The Newman isdA mutant strains carrying pRMC2, pRMC2isdA, and pRMC2isdJ were grown to exponential phase and induced with anhydrotetracycline at 1 μg ml⁻¹ overnight. (C) L. lactis NZ9000(pNZ8048), L. lactis NZ9000(pNZ8048isdA), and L. lactis NZ9000(pNZ8048isdJ) were grown in GM17 to mid-exponential phase and induced for 4 h with nisin. For all three panels, bacterial suspensions at 10⁸ CFU ml⁻¹ in dH₂O were treated with 10 μg of linoleic acid ml⁻¹ for 30 min or as indicated. Survival was measured by viable counting. Values are the means of results from three independent experiments.

DISCUSSION

Uncharacteristically for CoNS, S. lugdunensis is capable of causing serious systemic diseases in healthy adults. To do so, it needs to overcome several challenges, among which is nutrient accessibility. Iron is a critical component for bacterial growth but is sequestered in humans due to its insolubility and as prevention from oxidative damage with ca. 90% bound to heme, mostly within hemoglobin (52, 53). Additional iron sequestration mechanisms are used by the human host during infection to counteract the ability of pathogens to obtain iron (6, 29, 31, 39). The inevitable competition with the host has led to both Gram-positive and Gram-negative bacterial pathogens developing several distinct mechanisms for accessing iron from heme (21, 43, 50, 56, 65, 74).

We have characterized four proteins here from an iron acquisition system of the emerging pathogen S. lugdunensis being, to date, the sole example of a CoNS utilizing the Isd system for iron acquisition. The Isd proteins share significant structural and functional homology with the Isd receptors of S. aureus. Both systems are expressed under low-iron conditions. The S. lugdunensis proteins IsdB, IsdC, IsdJ, and IsdK were induced in cultures grown in iron-deficient medium. This is consistent with a report by Haley et al. that studied the role of S. lugdunensis heme monooxygenase (IsdG) (28). These findings suggest that transcription of the isd locus in S. lugdunensis, like S. aureus, is regulated by the iron-dependent repressor Fur. Indeed, Fur boxes were identified upstream from the start codons of isdB, isdC, and isdJ genes (Fig. 1A). Interestingly, the isd locus is duplicated in S. lugdunensis HKU09-01, which may increase the levels of transcription of the isd genes in that strain (30).

In both systems the IsdB and IsdC orthologues are localized in the cell wall. IsdB is likely to be processed by sortase A (SrtA), whereas the IsdC linkage to peptidoglycan is most probably catalyzed by sortase B (SrtB). We observed that the S. lugdunensis IsdB and IsdC proteins seem only to be partially wall anchored, since a portion of these proteins could be found in the membrane fraction. IsdK and IsdJ were detected only in the membrane fraction and were either hardly detectable on the surface of the bacterium (IsdK) or could not be detected at all (IsdJ). IsdJ and IsdK contain C-terminal sorting signals composed of sortase recognition motifs (LPXTG or NKQPN, respectively), followed by a hydrophobic domain and a tail of positively charged residues, and were therefore expected to be anchored to the cell wall and exposed on the cell surface. Despite SrtA and SrtB being capable of anchoring IsdB and IsdC, it is unclear why IsdJ and IsdK remain in the membrane. IsdJ was sorted to cell wall of the heterologous hosts S. aureus and L. lactis, indicating that differences in anchoring surface proteins to the cell wall must exist in S. lugdunensis.

The IsdJ and IsdK proteins were detected in S. lugdunensis culture supernatants, suggesting an additional role in iron acquisition. Heme acquisition by secreted proteins containing NEAT-like domains has been well characterized in Bacillus anthracis (20, 32, 62). The bacterium secretes IsdX1 and IsdX2 hemophores (acquiring heme from host hemoglobin), which deliver heme to IsdC exposed on the bacterial surface, allowing transfer of the iron-porphyrin into the cytoplasm. IsdX1 contains a single NEAT domain, whereas IsdX2 has five. NEAT1, -3, -4, and -5 bind heme, while all motifs associate with hemoglobin. NEAT1 and NEAT5 extract heme, which is then transferred by NEAT1, NEAT3, and NEAT4 to the cell wall-anchored IsdC protein. S. lugdunensis IsdJ and IsdK cannot bind Hb but perhaps could acquire heme directly from the extracellular environment during infection or from IsdB, which occurs abundantly in the external milieu. Perhaps two parallel hemoglobin-heme acquisition pathways exist in S. lugdunensis, one utilizing the cell wall-anchored IsdB and IsdC and the second using secreted proteins that acquire heme from the environment (IsdB, IsdJ, and IsdK) and pass it onto IsdC before transferring the porphyrin ring into the cytoplasm. However, more studies are required to reveal the role of IsdJ and IsdK in the heme acquisition mechanism.

The specific binding of hemoproteins by the Isd system of S. lugdunensis is restricted to a single IsdB receptor, which contrasts with S. aureus, where IsdH and IsdB express three hemoglobin-binding NEAT domains (19, 66). In S. lugdunensis, the IsdB protein uses its NEAT1 domain to bind to hemoglobin and hemoglobin-haptoglobin complexes but has no affinity for myoglobin or haptoglobin. Even though hemoglobin binding has been reported for a number of bacterial proteins containing NEAT-like domains, the characteristic aromatic loop coordinating hemoglobin is unique to S. lugdunensis and S. aureus. Furthermore, both species show selectivity in the source of hemoglobin. We show here that S. lugdunensis IsdB binds strongly to human Hb (K_d of 23 nM) and barely associates with mouse Hb. This correlates with the bacterial preference in utilizing the human Hb over mouse Hb to support growth in iron-restricted media. Therefore, it could be difficult to evaluate the importance of the Isd system in mouse models of colonization or infection. Since the nutrient availability may have a critical effect on bacterial pathogenesis, the humanized animal model (mice expressing human hemoglobin) should be strongly considered to study invasiveness and virulence of S. lugdunensis (55). A preference for hHb over mHb is characteristic of bacteria associated with humans, such as S. aureus and S. lugdunensis, but also S. simulans and Corynebacterium diphtheriae, whereas environmental bacteria that infect numerous hosts, such as Acinetobacter baumannii, Pseudomonas aeruginosa, Bacillus anthracis, and Bacillus cereus, grow at comparable levels on hHb and mHb (55, 56).

The S. lugdunensis IsdB protein was shown to be capable of binding heme by a conserved Tyr (IsdB Y467), probably in a pentacoordinate complex (a hydrogen bond is formed with the phenolate of another conserved Tyr in β8). This interaction is absolutely conserved among heme binding NEAT domains (25). The S. aureus heme uptake pathway IsdH-IsdB-(IsdA)-IsdC-IsdEF transfers heme from the surface-exposed hemoprotein receptors (IsdH and IsdB) through the cell wall-bound IsdA or directly onto the IsdC (buried in the cell wall), which passes it onto the membrane transporter (IsdEF) (40, 49, 76). The S. lugdunensis Isd system may present a reduced set of receptors, considering that IsdJ and IsdK may be redundant due to their cellular localization. We observed direct transfer of hemin from recombinant holo-IsdB to apo-IsdC, suggesting that these two abundant cell wall receptors may be sufficient for the efficient transfer of heme onto the putative membrane transporter system encoded by isdE and isdF. The likely reduction in the number of Hb and heme receptors could be the consequence of the lower hemolytic capacity of S. lugdunensis, which only expresses the delta-toxin-like hemolysin called SLUSH (22, 68). In contrast S. aureus expresses a range of hemolysins (alpha, beta, gamma, and delta) which lyse the erythrocytes and provide a rapid access to a pool of hemoglobin.

Several different ligands were identified for IsdJ. Recombinant IsdJ binds human loricrin and cytokeratin 10 (implicated in adherence to squamous epithelial cells and nasal carriage), as well as human plasma fibrinogen. Each ligand was previously shown to interact with IsdA and is therefore most probably recognized by conserved regions (11, 15). The recombinant N-terminal domain of IsdJ containing NEAT1 was shown to bind to fibrinogen. However, IsdJ expression did not promote bacterial adherence to immobilized ligands. Iron-starved S. lugdunensis adhered strongly to fibrinogen, but this was not due to IsdJ, since there was no difference in adherence between the wild-type strain and the isdJ-deficient mutant. Also, the iron-starved S. lugdunensis fbl mutant did not adhere to fibrinogen, suggesting that Fbl is the only fibrinogen receptor on S. lugdunensis. S. aureus and L. lactis strains expressing IsdJ on their surfaces also failed to adhere to cytokeratin, fibrinogen, and loricrin, which demonstrates that the protein is not sufficient to promote bacterial adherence to these ligands. IsdJ expression could confer some protection to S. lugdunensis from killing by linoleic acid, albeit not as efficiently as IsdA in S. aureus. However, IsdJ protected L. lactis and S. aureus as well as IsdA. Either IsdJ is expressed at a higher level in the surrogate hosts than in S. lugdunensis, or its location on the cell surface after sorting is responsible for the greater efficacy.

To summarize, S. lugdunensis, like S. aureus utilizes the Isd system to acquire heme-iron. IsdB is an abundant cell wall receptor that preferentially recognizes human Hb. IsdC is a heme receptor located in the cell wall that can accept heme from IsdB. Heme bound to IsdC could then be transferred onto the putative membrane transporters IsdEF. IsdJ and IsdK are heme-binding proteins that occur in the cytoplasmic membrane and external milieu despite their C-terminal motifs for anchorage to the cell wall peptidoglycan. Finally, IsdJ is multifunctional since it recognizes several host ligands and can confer resistance to skin fatty acids.