Heme Coordination by Staphylococcus aureus IsdE*

Staphylococcus aureus is a Gram-positive bacterial pathogen and a leading cause of hospital acquired infections. Because the free iron concentration in the human body is too low to support growth, S. aureus must acquire iron from host sources. Heme iron is the most prevalent iron reservoir in the human body and a predominant source of iron for S. aureus. The iron-regulated surface determinant (Isd) system removes heme from host heme proteins and transfers it to IsdE, the cognate substrate-binding lipoprotein of an ATP-binding cassette transporter, for import and subsequent degradation. Herein, we report the crystal structure of the soluble portion of the IsdE lipoprotein in complex with heme. The structure reveals a bi-lobed topology formed by an N- and C-terminal domain bridged by a single α-helix. The structure places IsdE as a member of the helical backbone metal receptor superfamily. A six-coordinate heme molecule is bound in the groove established at the domain interface, and the heme iron is coordinated in a novel fashion for heme transporters by Met78 and His229. Both heme propionate groups are secured by H-bonds to IsdE main chain and side chain groups. Of these residues, His229 is essential for IsdE-mediated heme uptake by S. aureus when growth on heme as a sole iron source is measured. Multiple sequence alignments of homologues from several other Gram-positive bacteria, including the human pathogens pyogenes, Bacillus anthracis, and Listeria monocytogenes, suggest that these other systems function equivalently to S. aureus IsdE with respect to heme binding and transport.

Staphylococcus aureus is a leading cause of hospital acquired bacterial infections (1). It is rapidly being recognized as an emerging pathogen because of relentless increases in drug resistance (2). The establishment of two strains is problematic in the clinic, methicillin-resistant S. aureus and vancomycin-resistant S. aureus. Methicillin resistant S. aureus strains are increasing in prevalence both in hospitals and, more recently, within the community (3,4). This drug resistance has highlighted the need to better understand S. aureus pathogenesis and physiology.
Iron uptake pathways have received significant attention because of the essential requirement of iron for the growth of most organisms (5). In aerobic environments, free iron concentrations are generally low. For human pathogens, iron concentrations are further limited by host storage, transport, and innate immune mechanisms (6,7). Many bacterial pathogens have sophisticated systems to directly utilize host iron sources to satisfy their physiological requirements. Heme iron represents the most abundant iron source in the human body, accounting for ϳ75% of the total iron (8). This heme iron can be found within hemoglobin in circulating red blood cells, myoglobin, and many other heme proteins. Because of its abundance, an ability to acquire heme iron from host sources represents a significant advantage for bacterial pathogens (9 -11).
Several heme uptake pathways have been characterized in Gram-negative bacteria (5,7). In contrast, uptake systems in Gram-positive bacteria are not as well understood. The Grampositive pathogen S. aureus utilizes heme iron through the iron-regulated surface determinant (Isd) 3 pathway (see below). More recently, a transposon mutant that preferentially acquired iron from transferrin versus heme implicated the heme transport system (Hts) in heme uptake as well (12).
The Isd system in S. aureus consists of nine members. IsdA, IsdB, IsdC, and IsdH/HarA are cell wall anchored surface proteins (13). These four proteins contain the conserved NEAT domain in one to three copies (14). IsdB and IsdH have been shown to bind hemoglobin and hemoglobin-haptoglobin, respectively (13,(15)(16)(17). Recently, for IsdA and IsdC, the heme binding properties and crystal structures of the NEAT domains in complex with heme were determined, demonstrating similar mechanisms of heme coordination (18 -21). Based on the predicted localization of these four proteins within the cell wall, Skaar and Schneewind (22) proposed a model for Isd heme transport. In this model, heme is removed from host heme proteins bound by IsdB and IsdH at the cell surface, transferred to IsdA and subsequently to IsdC. From there, heme moves to the ABC transporter-binding protein, IsdE (a lipoprotein), and subsequently through the transporter (13,19). Once in the cytoplasm, IsdG and IsdI liberate iron by heme degradation (23,24).
In this study, we show that inactivation of S. aureus isdE impairs growth on heme as a sole source of iron. To gain insight into the function of IsdE in heme binding and transport, the crystal structure of the IsdE-heme complex was determined. The structure reveals His-Met heme iron coordination that is unique to heme transport proteins. Corroborating the structure, alanine substitutions in binding pocket residues showed that mutation of Met 78 and His 229 resulted in significant loss of heme binding and that IsdE H229A was incapable of supporting IsdE-mediated growth on heme as a sole source of iron in growth promotion assays. This work adds substantial mechanistic detail to the complex model of heme uptake by S. aureus and provides a framework for future studies into the mechanism of bacterial binding protein-dependent heme acquisition.

Cloning and Protein Expression for Structure Determination-
The IsdE coding region (Gly 32 -Lys 289 ), excluding the signal sequence and 11 N-terminal amino acids following the Cys lipidation site, and three C-terminal amino acids were amplified from S. aureus N315 chromosomal DNA and cloned into the expression vector, pET-28a(ϩ) (Novagen). This construct was designed to optimize crystallization based on an analysis using the DISOPRED2 disorder prediction server (25). Recombinant protein was expressed with an N-terminal His 6 tag in Escherichia coli BL21. Cultures containing the expression vector were grown at 30°C to an optical density at 600 nm of ϳ0.8 followed by induction with 0.5 mM isopropyl ␤-D-thiogalactopyranoside and growth overnight at 25°C. Cells were resuspended in 20 mM Tris (pH 8), 200 mM NaCl and lysed at 4°C using an Emulsi Flex-C5 homogenizer (Avestin). His 6 -IsdE was purified using a Ni-Sepharose high performance column (GE Healthcare) and dialyzed against 50 mM Tris (pH 8), 100 mM NaCl prior to thrombin digestion to remove the His 6 purification tag. Cleaved protein was dialyzed into 50 mM HEPES (pH 7.5) for Source S column (GE Healthcare) purification followed by dialysis into 20 mM Tris (pH 8) and reconstitution with hemin as described previously (26). Selenomethionine-labeled IsdE was produced as described previously (27) and purified similarly to native IsdE.
Cloning and Protein Expression for Spectroscopy-The majority of the isdE gene, corresponding to amino acids 21-292 (excluding the signal sequence), was cloned into the GST fusion vector pGEX-2T-TEV (28) to generate pGST-IsdE. Overexpression of GST-tagged IsdE in E. coli ER2566 (protease-deficient) was achieved by growing plasmid-containing cultures in Luria-Bertani broth (Difco), containing 100 g/ml ampicillin, at 37°C to an A 600 of ϳ0.8. Isopropyl ␤-D-thiogalactopyranoside (0.4 mM) was added, and cultures were grown for a further 20 h at room temperature. Bacterial cells were pelleted, resuspended in phosphate-buffered saline, and lysed in a French pressure cell. Insoluble material was removed by centrifugation at 100,000 ϫ g for 20 min. GST-IsdE fusions were purified by passage of cell lysates across a 20-ml GSTPrep column (GE Healthcare). GST-IsdE was eluted from the column with 10 mM reduced glutathione, 100 mM NaCl, and 50 mM Tris-Cl, pH 9.0.
UV/Visible Absorption Spectroscopy-All proteins (wild type and mutant) were purified as expressed from E. coli, and rela-tive heme binding was assessed based on the ability of the proteins, all expressed in identical fashion, to scavenge and retain association with heme derived from the cytoplasm. Proteins were adjusted to an equivalent concentration, and electronic spectra were recorded using a Cary 500 spectrophotometer (Varian) with a 1-cm path length and 1-ml quartz cuvettes. All recordings were taken at room temperature.
IsdE Structure Determination-Heme-bound IsdE crystals were grown by hanging drop vapor diffusion at room temperature. The well solution contained 50 mM MES (pH 5.5), 0.2 M ammonium acetate, and 28% polyethylene glycol 4000. Drops were made from 1 l of 30 mg/ml protein solution and 1-l well solution. Crystals were briefly immersed in well solution supplemented to 16% glycerol prior to immersion in liquid nitrogen.
X-ray diffraction data were collected at the Stanford Synchrotron Radiation Laboratory at 100 K on beam lines 11-1 and 9-2 for the selenomethionine and native crystals, respectively. Single wavelength anomalous diffraction data were collected at a wavelength of 0.978894 Å. Native crystal data were collected at a wavelength of 1.0 Å. Data were processed using HKL2000 (29). Crystals grew in the space group P4 3 2 1 2 with one IsdE molecule in the asymmetric unit. The programs Solve (30) and Resolve (31,32) were used to obtain phases from the nine identified selenium sites and to build a preliminary model (supplemental Fig. S1). The phase solution had an initial figure of merit of 0.37 that was improved to 0.62 by density modification. The structure was manually constructed using Coot (33) and refined using translation libration screw parameters (34,35) with Refmac5 (36) from the CCP4 program suite (37). All analysis and figures were generated from the native IsdE structure. The refined structure contains residues Gly 32 -Lys 289 and 246 water molecules. The Ramachandran plot reveals 92% of residues are in the most favored conformation, and none are in the disallowed regions. Data collection and refinement statistics are shown in Table 1. Structure figures were generated in PYMOL (DeLano Scientific, San Carlos, CA).
Construction of S. aureus isdE::K m -Allelic replacement, using methodologies described previously (38), was used to generate a non-polar insertion mutation in the chromosomal copy of isdE; the kanamycin cassette, from plasmid pDG782 (39), was inserted into a unique EcoRI site present within the isdE coding region. The Newman isdE mutant was named strain H834. To complement the chromosomal isdE mutation, a DNA fragment containing isdE was PCR-amplified from the S. aureus Newman chromosome and cloned into pAW8 to generate pCLVEc. Plasmid pCLVEc was introduced into H834, via RN4220, by standard methodologies.
Site-directed Mutagenesis of isdE-Site-directed mutagenesis was used to alter residues in the IsdE-heme binding pocket. Specifically, site-directed mutagenesis was performed using the QuikChange PCR kit (Invitrogen), with Pfu Turbo polymerase and pGST-IsdE or pCLVEc as a template. The PCR products were incubated with DpnI (Roche) for 45 min to degrade template DNA and transformed into E. coli ER2566. Mutations were confirmed by sequencing. Constructs generated using pCLVEc as the template were introduced, via electroporation, into S. aureus RN4220 (40) and subsequently transduced to H834 (isdE Ϫ ) using phage 80␣. The constructs were isolated from H834 and sequenced to ensure that mutations had not been introduced during the transformation and transduction procedures.
Heme-dependent Bacterial Growth Promotion Studies-S. aureus strains were pre-grown, from single colony, overnight in tris-minimal succinate growth media (41). The cells were washed with saline, and 10 7 CFU of each strain were inoculated into tris-minimal succinate containing 5 M ethylenediaminedi-(o-hydroxyphenyl)acetic acid with or without, either 50 M FeCl 3 or 5 M hemin. Cultures (300 l) were incubated at 37°C with continuous shaking, and bacterial growth was monitored every 30 min over 20 h using a Bioscreen C (MTX Lab Systems, Inc.). Growth curves were plotted using Sigma Plot 2000.

RESULTS
Overall Protein Structure-The selenomethionine and native IsdE structures were solved to 1.95 and 2.15 Å resolution, respectively. The structures reveal a protein with bi-lobed conformation with two domains, each composed of a central-most parallel ␤-sheet surrounded by ␣-helices (Fig. 1A). The N-terminal domain (Gly 32 -Arg 138 ) ␤-sheet contains only two ␤-strands, whereas the C-terminal domain (Asn 163 -Lys 289 ) contains a 5-stranded ␤-sheet. The two domains are connected by a single long ␣-helix (Lys 139 -Lys 162 ) that spans the length of the molecule. A large interface is formed between the two domains and is contributed to by a mix of small hydrophobic and several hydrophilic residues. Although a Dali search (42) did not reveal any structurally similar heme-binding proteins, it did identify that the IsdE structure was distantly related to BtuF, a cobalamin transporter, and to the siderophore transporters CeuE and FhuD. For each structure, similarity is characterized by a Z-value Ͼ20, a root mean square deviation of 2.8 -3.3 Å for 234 -244 C␣ atoms and a sequence alignment identity Ͻ18%. These proteins are all members of the helical backbone metal receptor superfamily.
Heme Binding-Previous studies demonstrated that IsdE is a heme-binding protein (13,19). We demonstrate that a single heme molecule is bound to IsdE along the groove formed between the two lobes ( Fig. 1, A and B). Heme is oriented within the pocket at approximately a 45°angle in rela-tion to the longest axis of the protein such that the propionates interact primarily with the N-terminal domain. Approximately 160 Å 2 (19%) of the total heme surface area is solvent exposed (as determined with AREAIMOL (37)). Several hydrophobic residues line the interior of the pocket and interact with the largely hydrophobic porphyrin ring. The N-terminal domain contributes Pro 77 , Val 96 , and Ile 99 to this hydrophobic environment, and the C-terminal domain contributes Val 175 , Pro 17 6, Leu 180 , Tyr 208 , and Ile 270 (Fig. 2C).  IsdE-bound heme iron is six-coordinate with axial coordination by the thioether of Met 78 (2.28 Å) from the N-terminal domain and the imidazolate of His 229 (2.00 Å) from the C-terminal domain (Figs. 1A and 2A). The angles formed between the tetrapyrrole nitrogen plane and the axial ligands are both ϳ90°. The tetrapyrrole ring is close to planar, and the iron is displaced from the plane formed by the tetrapyrrole nitrogens by less than 0.04 Å (Fig. 2A). His 229 participates in a complex H-bond network that includes Glu 265 from the C-terminal domain and residues Tyr 61 and Lys 62 from the N-terminal domain (Fig. 2B). His 229 N␦1 forms an H-bond with HOH15 (2.9 Å), which, in turn, forms an H-bond to Glu 265 O⑀2 (2.9 Å). Glu 265 also forms H-bonds directly to Lys 62 N (3.4 Å) and Tyr 61 O H (2.7 Å) through carboxylate atoms O⑀2 and O⑀1, respectively.
The heme propionates are oriented approximately parallel to the binding groove and are essentially buried (Fig. 2D). They are well ordered in the electron density map and form extensive interactions with IsdE (Fig. 2B). Lys 62 N forms a H-bond to HOH13 (2.9 Å) that, in turn, forms H-bonds with both heme propionates (2.6 and 2.9 Å). One of the heme propionates forms additional direct H-bonds with the main chain amides of Val 41 (2.7 Å) and Ala 42 (3.0 Å). Val 41 and Ala 42 are located at the N terminus of ␣-helix 1 that is oriented such that the positive helix dipole is directed toward the propionate carboxylate group (Fig. 1A). An additional HOH16 bridged (2.9 Å) interaction is formed with Thr 40 OH (2.8 Å) and Thr 271 OH (3.1 Å). The other propionate forms direct H-bonds to Ser 60 O␥ (2.7 Å) and the main chain amide of Tyr 61 (3.2 Å) (Fig. 2B).
Multiple Sequence Alignments-A sequence search using BLAST (43) revealed several homologous proteins (E-value Ͻ 4 ϫ 10 Ϫ37 ) in other Gram-positive organisms, namely species of Bacillus, Listeria, Clostridia, Streptococcus, and Lactobacillus. Previously, the IsdE homologues from Listeria monocytogenes, Clostridium tetani, and Bacillus anthracis are shown to be associated with related Isd uptake systems (22). The sequences were aligned with ClustalX (44), and the alignments were manually edited in BioEdit (45). Each sequence shares greater than 28% identity over the 292 residues of S. aureus IsdE (Fig. 3 and supplemental Fig. S2). In the identified sequences, the predicted secretion signal and ϳ15 N-terminal residues after the Cys lipidation site are poorly conserved. The alignments reveal several conserved residues within the heme pocket of the IsdE structure. Both of the iron axial ligands (i.e. Met 78 and His 229 ) are completely conserved in the homologous sequences (Fig. 3). Residues forming the H-bonding network with His 229 are also generally conserved. Lys 62 is conserved in the bacilli and listerial proteins, but it is replaced by Tyr in the streptococcal and clostridial species. Glu 265 is generally conserved or replaced by Asn. Residues interacting with the propionates are also conserved as are the hydrophobic heme pocket residues; where different, they are substituted for residues with similar hydrophobic properties (Fig. 3).
The alignments reveal a conserved patch of residues located at the surface of the N-and C-terminal lobes traversing the heme pocket (Fig. 2D). Several conserved, charged residues are evident within this conserved patch. Notably, Glu 83 and Glu 214 are completely conserved, whereas Glu 242 differs only in the Clostridium perfringens sequence (Fig. 3). Lys 237 and Lys 241 are also conserved as large charged residues at the lobe surface in all aligned sequences (Fig. 3).
Another striking feature revealed by mapping of amino acid conservation onto the structure is the maintenance of nine Pro residues between residues 32-105 in the N-terminal lobe of the homologues (supplemental Fig. S3). Pro 77 occurs in the heme Contribution of Residues to Heme Binding and IsdE-mediated Heme Transport in Vivo-As shown in Fig. 4A, spectroscopic analysis of GST-IsdE shows strong absorption in the Soret region as well as signals in the visible region around 650 nm; these signals are characteristic of heme binding. As expected, GST alone showed none of these signals. To validate the crystal structure of the IsdE-heme complex, Ala point mutations were constructed in several of the conserved heme binding pocket residues of IsdE. Mutation of the conserved heme iron-coordinating His 229 and Met 78 , individually, resulted in a significant reduction in heme binding by IsdE. Moreover, mutation of both His 229 and Met 78 to Ala in the same protein completely abolished IsdE heme binding activity (Fig. 4A). Notably, we also observed altered IsdE heme binding properties upon mutation of other residues whose side chains coordinated directly or indirectly (via waters) to the heme structure (supplemental Fig. S4).
We and others have shown previously that S. aureus is capable of growing on hemin as a sole source of iron (12,13,17,18). Also we demonstrated that IsdA, localized to the cell wall, contributes to this process (18). In the present study, we characterized Isd-mediated heme iron acquisition by defining the involvement of IsdE in this process at the cell membrane. In Fig.  4B, we show that a S. aureus isdE knock-out mutant, H834, although still able to grow on hemin as a sole source of iron, grew slower than wild type. Notably, complementation of this mutant with pCLVEc (expresses isdE) restored growth on hemin to greater than wild type levels. These data confirm that IsdE contributes to heme iron acquisition in S. aureus and also suggest that other heme transport systems function in this bacterium. Ala substitutions were constructed in plasmid pCLVEc and used to show that although a M78A mutation in pCLVEc (pCLVEc-M78A) had no impact on the ability of IsdE to participate in S. aureus growth on hemin, a H229A mutation in pCLVEc completely abolished the ability of the plasmid to complement the isdE knock-out phenotype. These results indicate that although Met 78 is dispensable in vivo, His 229 is required for the biological activity of IsdE.

DISCUSSION
In this study, we demonstrate that S. aureus growth on heme as a sole iron source is impaired by inactivation of isdE. Moreover, IsdE-mediated heme transport in S. aureus is reliant on the key heme iron axial ligand, His 229 (Fig. 4). Our work implicates IsdE as a central component for heme iron uptake by the Isd system.
The crystal structure of IsdE provides the first view of heme binding by a substrate-binding protein associated with an ABC transporter. Heme is bound within a deep groove of IsdE formed at the interface of the two domains. The heme is nearly completely buried with only a single edge of the porphyrin ring exposed (19% of the total heme surface). In contrast, heme protrudes from the surface of the NEAT domains of the cell wall anchored components IsdA (18) and IsdC (20). Heme bound to these NEAT domains is about 35% solvent exposed. The small amount of heme exposure in holo-IsdE is comparable with that of the transporter HemS (18%) from Yersinia enterocolitica (46).
In addition to being more buried, the binding mode of heme to IsdE differs in many respects from that of the NEAT domains of IsdA and IsdC. Heme bound to the NEAT domains is fivecoordinate with the single axial ligand provided by the phenolate of a Tyr residue (18,20,21). The phenolate oxygen in turn forms a H-bond to a second Tyr residue. In contrast, the structure of IsdE shows an extensive H-bond network attached to iron ligand His 229 , involving residues from both domains of IsdE (Fig. 2B). The same H-bond network forms interactions with both propionates through water molecules. This H-bond network, in combination with the dipole of ␣-helix-1, assists in neutralizing the negative charge on the buried propionates. The NEAT domains bind heme such that the propionates are largely solvent exposed (18,20) negating the need for charge neutralization. The more extensive interactions of IsdE with the heme iron and porphyrin group suggest greater specificity and tighter binding than by the NEAT domains of the cell wall anchored components of the Isd system. Indeed, previous spectroscopic studies demonstrated that IsdC is able to bind heme and protoporphyrin IX from the E. coli cytoplasm, whereas IsdE is associated predominantly with heme (19).
The His-Met iron coordination observed in the IsdE-heme complex structure is unique to known heme transporters. This axial ligand combination is best known in several c-type cytochromes (47). His-Met coordination is in agreement with previous magnetic circular dichroism spectroscopic data for IsdE. Mack et al. (19) previously demonstrated that IsdE heme iron is six-coordinate, and though the axial ligand in addition to His could not be determined conclusively, His-Met was proposed as a potential configuration. In contrast, EPR studies of the close homologue, SpHtsA from S. pyogenes, suggested heme iron in this system was mediated through two axial N ligands because of similarities in the EPR spectrum from b-type cytochromes (48). However, molecular modeling of HtsA using the structure of IsdE as a template and the lack of a His in the N-terminal lobe of SpHtsA suggest Met, as in IsdE, is the second axial ligand in SpHtsA.
Typically ABC transporter-binding proteins form a bi-lobed structure with the ligand bound in a groove formed between the lobes. Three classes of these proteins are described based on the structure of their interdomain connection. The majority of the proteins of known structure belong to either class I or II that contain three or two flexible domain-bridging ␤-strands, respectively. The strands impart significant flexibility allowing for large interdomain conformational change upon ligand binding (49,50). However, IsdE belongs to the third class of bacterial-binding proteins that are characterized by a single domain spanning ␣-helix. The DALI search identified several class III mem- bers as structural homologues of IsdE with functions in metal (ZnuA), siderophore (FhuD, CeuE), and vitamin B12 (BtuF) transport (49,(51)(52)(53). The general architecture of members of the class three binding proteins seems to be very conserved. The known members of this family all display a similar bi-lobed architecture with two lobes made up of central ␤-strands surrounded by several ␣-helices bridged by a single long ␣-helix. Because of the domain-spanning ␣-helix, members of class III are relatively inflexible and undergo minimal conformational change upon ligand binding and release (41,49,54). As is typical of class III-binding proteins, the interdomain interface of IsdE is much larger than those observed in class I and class II proteins. These large interaction surfaces are proposed to contribute to the inflexibility of these proteins (52). The IsdE interface is formed from several hydrophilic residues. However, in the structural homologue, FhuD, the interdomain interface is predominantly composed of hydrophobic residues, which Clarke et al. (52) suggest results in a greater restriction of domain movement during ligand binding. CeuE, lacking a discernable hydrophobic interface (53), is more similar to IsdE.
Because of the extensive interaction of heme with IsdE for a transport protein, heme release is likely a function of the small structural changes induced by interaction with the transmembrane permease component of the ABC transporter. The transmembrane and ATPase components of the Isd system are not well characterized, however, the crystal structures of two bacterial substrate-binding proteindependent ATPase transporters, BtuCD and HI1470/1, have been described (49,55). Analysis of sequence conservation between IsdE sequence homologues revealed a highly conserved strip of residues spanning the width of the face involved in transporter interactions (Fig. 2D). In particular, several charged residues are conserved that could form saltbridged interactions with the permease. BtuF is the most similar structural homologue of IsdE as revealed by the DALI search. In BtuF, Glu 72 and Glu 202 are located on the surface of the N-and C-terminal lobes and are positioned to interact with complementary Arg residues when docked to the BtuC transmembrane component (49). In support of their importance to transport function, Sebulsky et al. (28) demonstrated that mutation to Ala of the corresponding Glu residues in S. aureus FhuD2 does not affect ligand (hydroxamate siderophore) binding but abrogated transport. In IsdE, Glu 83 and Glu 214 are highly conserved in IsdE sequence homologues (Fig. 3) and may serve a similar role. In fact, Glu 83 is in an equivalent position as Glu 72 in BtuF. Glu 202 (BtuF) and Glu 214 (IsdE) are in topologically similar locations; however, this region is an ␣-helix in BtuF and a loop in IsdE (supplemental Fig. S5). The electron density for Glu 214 is weak, and this loop is involved in a crystal contact suggesting that an alternative conformation may exist in the presence of the permease. Nonetheless, the conservation and surface localization of these Glu residues suggest that the proteins dock via structurally conserved salt-bridge formation, similarly to BtuF (49). Thus, we would anticipate that mutation of the conserved Glu residues in IsdE would similarly disrupt heme transport whereas not affecting heme binding.
The IsdE structure is unique in this superfamily of proteins because of the lack of well defined ␤-strands in the N-terminal domain. This is because of the large prevalence of Pro residues that are unable to participate in ␤-sheet H-bonding. Many of these Pro residues are conserved in putative IsdE orthologues (Fig. 3). Pro 77 plays a necessary role in forming a tight turn that orients Met 78 ; the conservation of additional Pro residues within domain suggests that they contribute rigidity to the backbone. Furthermore, in SpHtsA, position 58 (a Pro in IsdE) is a Cys, and a second Cys occurs at position 42, such that a disulfide bond could form between these two residues. Even though the Pro is not conserved in SpHtsA or SeHtsA (Fig. 3), the disulfide would also add rigidity to the structure.
Collectively, our data are consistent with a role for IsdE in relaying heme from the cell wall anchored surface receptors to the permease (13,22). A heme relay function for IsdE is supported by studies on the two IsdE orthologues, SpHtsA and SeHtsA, from Streptococcus equi subspecies equi (each share ϳ40% amino acid sequence identity with IsdE; Fig. 3). In streptococci, Shp is a cell surface protein that has been shown to relay heme to the HtsA lipoprotein (56,57). Those studies demonstrated that streptococcal apo-HtsA can bind and receive heme from a soluble domain of Shp. Transfer between streptococcal Shp and HtsA is believed to be driven by greater heme affinity of HtsA as compared with Shp (48,58). Based on the high sequence identity between IsdE and the two HtsA orthologues, including the conservation of residues in the heme binding pocket (Fig. 3 and supplemental Fig. S2), both the structures and heme binding properties of these proteins are likely to be similar. Furthermore, parallels in the mechanism of heme transfer to IsdE from the cell wall anchored IsdC may exist to those of the Shp-HtsA system.
As shown in Fig. 4, mutation of isdE in S. aureus does not abrogate growth on heme as a sole source of iron, suggesting other means of heme iron acquisition in this bacterium. A second ABC transporter is recently identified and named HtsABC based on the evidence that a transposon mutation in the hts operon yields a strain demonstrating a decrease in the ratio of heme to transferrin iron uptake, relative to wild type (12), leading to the suggestion that this operon expresses an alternate heme uptake system in S. aureus. In this system, HtsA is believed to be the binding protein (lipoprotein) that would be analogous to IsdE. The S. aureus HtsA (SaHtsA), however, shows only a low degree of sequence similarity to either IsdE or S. pyogenes HtsA (Ͻ15% amino acid identity), and the protein lacks the key residues that are involved in IsdE-mediated heme and heme iron coordination (see above), indicating that if the SaHtsA binds heme, it does so using entirely different coordination.
The results presented in this study add to the rapidly expanding body of knowledge about the Isd-mediated heme acquisition system in Gram-positive bacteria, providing important insight into the mechanism of heme binding and transport via the IsdE component. These results provide a template for more mechanistic studies that will determine the details of heme transfer to the IsdE lipoprotein, presumably by cell wall anchored proteins.