Ruffling of Metalloporphyrins Bound to IsdG and IsdI, Two Heme-degrading Enzymes in Staphylococcus aureus^,

Abstract

IsdG and IsdI are paralogous proteins that are intracellular components of a complex heme uptake system in Staphylococcus aureus. IsdG and IsdI were shown previously to reductively degrade hemin. Crystal structures of the apoproteins show that these proteins belong to a newly identified heme degradation family distinct from canonical eukaryotic and prokaryotic heme oxygenases. Here we report the crystal structures of an inactive N7A variant of IsdG in complex with Fe³⁺-protoporphyrin IX (IsdG-hemin) and of IsdI in complex with cobalt protoporphyrin IX (IsdI-CoPPIX) to 1.8 Å or better resolution. These structures show that the metalloporphyrins are buried into similar deep clefts such that the propionic acids form salt bridges to two Arg residues. His⁷⁷ (IsdG) or His⁷⁶ (IsdI), a critical residue required for activity, is coordinated to the Fe³⁺ or Co³⁺ atoms, respectively. The bound porphyrin rings form extensive steric interactions in the binding cleft such that the rings are highly distorted from the plane. This distortion is best described as ruffled and places the β- and δ-meso carbons proximal to the distal oxygen-binding site. In the IsdG-hemin structure, Fe³⁺ is pentacoordinate, and the distal side is occluded by the side chain of Ile⁵⁵. However, in the structure of IsdI-CoPPIX, the distal side of the CoPPIX accommodates a chloride ion in a cavity formed through a conformational change in Ile⁵⁵. The chloride ion participates in a hydrogen bond to the side chain amide of Asn⁶. Together the structures suggest a reaction mechanism in which a reactive peroxide intermediate proceeds with nucleophilic oxidation at the β- or δ-meso carbon of the hemin.

Staphylococcus aureus is a leading cause of hospital-acquired bacterial infections (1). The establishment of methicillin-resistant strains of S. aureus is a concern in both the clinic and, more recently, within the community (2, 3). Iron uptake pathways have received significant attention because of the requirement of iron for the growth of most organisms (4). For human pathogens, iron concentrations are limited by host storage, transport, and innate immune mechanisms (2, 5). 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 ∼75% of the total iron (6). This heme-iron is predominantly found within hemoglobin in circulating red blood cells and myoglobin of muscle cells. Because of its abundance, an ability to acquire heme-iron from host sources represents a significant advantage for bacterial pathogens (7-9).

S. aureus acquires heme-iron predominantly through the Isd (iron-regulated surface determinant) system. IsdA, IsdB, IsdC, and IsdH/HarA are cell wall-anchored proteins (10) that contain heme-binding NEAT domains (11, 12). The host hemoprotein hemoglobin and its carrier protein haptoglobin are bound by IsdB and IsdH at the cell surface. Heme is proposed to be removed from hemoglobin and transferred to IsdA and subsequently to IsdC for passage through the cell wall (13). Subsequently, heme is bound by IsdE (14), an Isd transporter-binding protein, and delivered into the cytoplasm by the membrane transporter IsdDEF (10, 15). Once in the cytoplasm, IsdG and IsdI liberate iron by heme degradation (16, 17).

S. aureus IsdG and IsdI are dimeric, paralogous proteins that share 64% sequence identity (supplemental Fig. S2). Both enzymes contain two predicted active sites that degrade hemin to release iron in the presence of NADPH-cytochrome P450 oxidoreductase or ascorbate (16). An in vivo role for these proteins in heme degradation is supported by the observation that expression of IsdI in a Corynebacterium diphtheria heme oxygenase-deficient mutant rescues its ability to use hemin as an iron source (16). Crystal structures of the apo-forms of IsdG and IsdI revealed that both enzymes exhibit similarity to the ActVA-Orf6 monooxygenase from Streptomyces coelicolor, and belong to the ferredoxin fold superfamily in which a β-barrel is formed at a dimeric interface (17). This fold is distinct from that of the eukaryotic heme oxygenases (HOs)2 such as HO-1. The putative active sites of IsdG and IsdI both contain conserved His, Trp, and Asn residues, in which mutation nullifies heme degradation activity (17). Studies focused on the Bacillus anthracis IsdG homolog, also named IsdG, have revealed that this protein binds and degrades hemin with concomitant release of iron. IsdG is required for B. anthracis utilization of hemin as a sole iron source, and it is also necessary for bacterial protection against heme-mediated toxicity (18). More recently, a homolog of IsdG from Bradyrhizobium japonicum, named HmuQ, has also been shown to function as a heme oxygenase, and the reaction product of HmuQ-mediated heme catalysis is believed to be biliverdin (19).

Mechanistic understanding of how IsdG and IsdI function in heme metabolism is limited by the absence of information on enzyme-substrate interactions. In this study, we present the crystal structures of an inactive variant of IsdG in complex with ferriheme (IsdG-hemin) and IsdI in complex with cobalt-protoporphyrin IX (IsdI-CoPPIX). These structures reveal that the porphyrin rings are highly ruffled, and a single conserved His residue coordinates to the metal. Furthermore, a chloride ion coordinated to Co³⁺ in the IsdI-CoPPIX structure is hydrogen bonded to Asn⁶. The homologous Asn is replaced in the inactive IsdG variant suggesting a direct role for this residue in catalysis.

EXPERIMENTAL PROCEDURES

Degradation of Metalloporphyrins—IsdG and IsdI were expressed in Escherichia coli BL21 (DE3) and purified as described previously (16). Metalloporphyrins (Frontier Scientific) were dissolved in 0.1 n NaOH (10 mm), and spectra were obtained with a Varian Cary 50-BIO UV-visible spectrophotometer from 300 to 700 nm. For degradation studies purified protein (10 μm) was combined with dissolved metalloproporphyrin (10 μm) in Tris-buffered saline (50 mm Tris, pH 7.5, 150 mm NaCl) and allowed to bind for 1 h at 4 °C. Initial spectra were obtained, and ascorbic acid (1 mm) was then added as an electron donor, and spectra were obtained every 5 min for 90 min. For spectra of metalloporphyrins alone, the same procedure was used without the addition of purified protein.

To measure the oxygen dependence of the heme degradation reaction, oxygen was removed from solutions of hemin-reconstituted IsdG and IsdI (20 mm Tris, pH 7.5) by purging with argon in cuvettes sealed with a septum. Argon-purged ascorbate was added using a syringe, and UV-visible spectra were collected.

Protein Expression and Crystallization—IsdG-N7A and IsdI were expressed in E. coli BL21 (DE3) and purified as reported previously (17). As the overexpressed proteins are isolated in the apo-form, IsdG and IsdI at 10 mg/ml were reconstituted with metalloporphyrins (20). Hemin and CoPPIX (Sigma) were dissolved in DMSO (10 mg/ml). Metalloporphyrin complexes were prepared by adding ∼2:1 in molar ratios of hemin and CoPPIX to IsdG-N7A and IsdI, respectively. DMSO and excess metalloporphyrins were removed using an Econo-Pac 10DG desalting column (Bio-Rad) equilibrated with 200 mm NaCl and 20 mm Tris-HCl, pH 8. The protein was concentrated by Amicon centrifugal filters (molecular weight cut-off 10,000, Millipore) to a final concentration of ∼15 mg/ml, as determined by the Bradford method using bovine serum albumin as a standard (Sigma).

Deeply colored crystals were obtained by mixing equal volumes of protein solution with crystallization buffers in a hanging drop vapor diffusion setup. IsdG-hemin crystals were obtained with a reservoir containing 24% (w/v) PEG3350 (Fluka) and 0.1 m BisTris, pH 5.3. For IsdI-CoPPIX, the reservoir contained 15% (w/v) PEG3350 and 0.1 m sodium citrate, pH 3.2. Crystals were transferred to crystallization solutions supplemented to 25% with ethylene glycol as a cryoprotectant and immersed in liquid nitrogen.

Data Collection and Structure Determination—The diffraction datasets of IsdG-hemin and IsdI-CoPPIX were collected at the Stanford Synchrotron Radiation Laboratory on beamlines 9-2 and 7-1, respectively. Images were integrated and scaled using the program Mosflm and Scala (21). The overall Wilson B-factors were 24.3 and 15.5 Ų for the IsdG-hemin and IsdI-CoPPIX datasets, respectively.

Initial phase estimates for each structure were obtained by molecular replacement as implemented by the program Mol-Rep (22), using the respective apo-structures (PDB entries 1XBW and 1SQE) as search models after removal of the solvent atoms. In each case, the top scoring solution was clearly separated from the alternatives. For each structure, electron densities for the bound metalloporphyrins were clearly visible in the maps derived from the molecular replacement solution. Initial model building was performed by the Arp/Warp program (23). Complete crystallographic refinement was achieved using the XtalView and Refmac5 programs (24, 25). The metal ligand distances were not restrained. Data collection and refinement statistics are presented in Table 1.

TABLE 1.

Data collection and refinement statistics for the IsdG and IsdI complexes

	IsdG-hemin	IsdI-CoPPIX
Data collectiona
Resolution range	57-1.8 Å	56-1.5 Å
Space group	P1	P21
Unit cell dimensions (Å)	a = 46.59, b = 58.63, c = 59.69, α = 77.6°, β = 74.4°, γ = 75.1°	a = 37.32, b = 65.67, c = 58.66, β = 107.8°
Unique reflections	48,756	41,366
Completeness (%)	90.5 (90.5)	95.9 (80.1)
Average I/σI	6.9 (2.7)	10.9 (2.3)
Redundancy	5.3	3.5
Rmerge	0.061 (0.225)	0.045 (0.323)
Refinement
Rwork (Rfree)	0.200 (0.234)	0.182 (0.201)
B-factors
All atoms	27.3 A2	15.9 Å
Protein	26.7 A2	14.7 Å
Metal PPIX	25.0 A2	14.1 Å
Water	36.1 A2	24.9 Å
r.m.s.d. bond length	0.012 Å	0.011 Å

^aValues for the highest resolution shell are shown in parentheses.

Distortion of the bound metalloporphyrins in the complexes was analyzed by the normal-coordinate structural decomposition (NSD) program (version 2) (26). All graphic renditions were prepared using the program PyMol (27).

RESULTS

Degradation of Metalloporphyrins by IsdG and IsdI—IsdG and IsdI-mediated degradation of various metalloporphyrins in the presence of the electron donor ascorbate was analyzed by UV-visible spectroscopy (Fig. 1). As has been shown previously (16), we found that IsdG and IsdI rapidly degrade hemin (Fe-PPIX) in the presence of electron donor (Fig. 1A). In contrast, neither enzyme was able to degrade Co-PPIX, Ga-PPIX, Mn-PPIX, or Zn-PPIX using similar reaction conditions (Fig. 1, B-E). This lack of activity was despite the fact that all of these non-iron metalloporphyrins are bound by IsdG and IsdI as judged by the alterations in the visible absorption spectra upon co-incubation of enzyme with metalloporphyrin. Notably, Co-PPIX, Ga-PPIX, and Zn-PPIX undergo spectral changes upon exposure to ascorbate, and these changes are inhibited by complex formation with IsdG or IsdI. Taken together, these results demonstrate that, despite the fact that IsdG and IsdI complex non-iron metalloporphyrins in vitro, the catabolic activity of IsdG and IsdI is specific for iron-containing metalloporphyrin. Addition of ascorbate to anaerobically prepared solutions of IsdG-hemin and IsdI-hemin resulted in initial minor reductions in the Soret maxima likely because of the presence of trace oxygen (supplemental Fig. S1). After exposure to air, heme degradation proceeded as observed previously (16), supporting the assignment of IsdG and IsdI as heme oxygenases.

FIGURE 1.

Metalloporphyrin degradation by IsdG and IsdI. In vitro enzymatic degradation of several metalloporphyrins was analyzed via UV-visible spectrophotometry. Degradation of hemin (A), Co-PPIX (B), Ga-PPIX (C), Mn-PPIX (D), and Zn-PPIX (E) using ascorbic acid as an electron donor was analyzed in the presence of IsdG (2nd column), IsdI (3rd column), or without protein (1st column). Red lines indicate absorbance (Abs) before addition of electron donor. Blue lines indicate absorbance after 90 min in the presence of ascorbate.

Structure of IsdG-Hemin—Crystallization screens of native IsdG or IsdI in complex with hemin did not yield positive results. The N7A variant of IsdG was shown to bind hemin but is not capable of catalytic degradation (17). Reconstitution of this variant with hemin did yield diffraction quality crystals of space group P1 with four protomers or two IsdG dimers in the asymmetric unit. Residues 0-107 of each peptide chain (residue 0, a threonine, is a remnant of the tobacco etch virus protease cleavage site) as well as 266 water molecules were fit to the electron density map. The structures of the two dimers are not noticeably different (r.m.s.d. of 0.34 Å for 107 C-α atoms). Similarly, the core fold of apo-IsdG is retained upon hemin binding (Fig. 2A). The C-α atoms of the apo-IsdG and IsdG-hemin can be superimposed with an r.m.s.d. of 0.36 Å (95 C-α atoms).

FIGURE 2.

The structure of IsdG-hemin. A, superposition of apo-IsdG (cyan) and IsdG-hemin (yellow). The loop of IsdG-hemin structure that is disordered in the apo-IsdG structure is colored red. B, active site of IsdG viewed from the distal side with a F_o - F_c omit map contoured at 3 σ. C, active site of IsdG-hemin with residues that interact with porphyrin ring depicted in stick model and labeled. In each panel, residues in the active site are depicted in the stick model and labeled. Oxygen atoms are red, and nitrogen atoms are blue. Hemin carbons are gray, and amino acid residue carbons are colored according to the main chain of each structure.

Each monomer in the asymmetric unit binds a heme molecule that is readily modeled into an F_o - F_c omit map (Fig. 2B and supplemental Fig. S3). The heme molecules are bound in deep clefts as predicted by modeling into the apo-structure; however, in the apo-structure residues 81-88 were not included because of disorder in the absence of substrate (17). The binding of heme stabilizes these residues, and thus a complete model of IsdG-hemin structure could be determined. These nine residues form part of a long loop in the IsdG-hemin structure (Fig. 2A).

The ferric iron atom of the heme group is pentacoordinate with the N-ε2 atom of His⁷⁷ at the fifth axial position (ligand bond length of 2.2 Å on average). His⁷⁷ is located at the carboxyl end of helix 3, and the imidazole ring stacks with phenyl group of Phe⁷³ from the same helix (Fig. 2C). Also, the N-δ1 atom of His⁷⁷ is hydrogen bonded to the main chain carbonyl of Phe⁷³. The iron atom is displaced ∼0.2 Å out of the pyrrole nitrogen plane toward the axial ligand. The distal side of the iron center is in direct van der Waals contact (3.9 Å) with the C-δ1 atom of Ile⁵⁴ (Fig. 2B).

During refinement, severe distortion from planarity of the porphyrin ring of the heme group became apparent. To accurately model the heme conformation, the stereochemical library for heme in Refmac5 required modification to allow for greater torsional rotation of bonds within the porphyrin ring. When viewed down the β, δ-meso carbon axis, the porphyrin ring appears kinked at the meso carbons such that the iron atom is displaced toward the His ligand (Fig. 2B). Also, the imidazole ring of His⁷⁷ is aligned approximately with the βδ-meso carbon axis (Fig. 2C and supplemental Fig. S4A). When viewed along the α, γ-meso carbon axis, a more severe kink divides the ring into two planes that bend away from the axial ligand, forming an angle of ∼140° (Fig. 2C and supplemental Fig. S4B). The heme group makes extensive hydrophobic interactions with residues Leu⁹, Phe²³, Phe⁶⁴, Trp⁶⁷, Leu⁶⁸, and Val⁸⁰. The side chains of Ala⁷ and Phe²³ from the distal side and Trp⁶⁷ and Val⁸⁰ from the proximal side are pressed against meso carbons to outline the ruffling in the porphyrin ring (Fig. 2C). Specifically, atoms from these four residues make contacts 3.5 to 3.7 Å to the porphyrin ring (supplemental Fig. S5A).

The propionic acid moieties are buried deep in the cleft and are anchored by the guanidine groups of Arg²² and Arg²⁶. In addition, the side chain of His²⁷ and main chain amides of His²⁷ and Ile²⁹ interact with the propionic groups. Only the porphyrin ring edge that contains the α-meso carbon is exposed to the solvent. The solvent-accessible surface of the heme group is 73 Ų on average (∼9% of the total heme surface area), as calculated by the program AreaIMol (28). The configuration of hemin binding along the α, γ-meso carbon axis could not be determined unequivocally due to disorder in electron density maps at the vinyl groups, one of which is solvent-exposed.

Structure of IsdI-CoPPIX—Crystals were obtained of IsdI in complex with CoPPIX, which is structurally similar to hemin but is not degraded (Fig. 1). A single IsdI dimer occupies the asymmetric unit. The monomers are nearly identical and can be superposed with an r.m.s.d. of 0.29 Å for 108 C-α atoms. The complete chain from -1 to 107 (residues Ala^-1 and His⁰ are derived from the tobacco etch virus protease cleavage site) and 186 waters were modeled into the electron density map. The structures of apo-IsdI (PDB code 1SQE) and IsdI-CoPPIX can be superposed with a r.m.s.d. of 0.77 Å (95 C-α atoms). As observed for IsdG, a single large loop (residues 80-87) is ordered upon binding of CoPPIX to IsdI (Fig. 3A).

FIGURE 3.

The structure of IsdI-CoPPIX. A, superposition of apo-IsdI (cyan) and IsdI-CoPPIX (green). The loop of IsdI-CoPPIX structure that is disordered in the apo-IsdI structure is colored red. B, active site of IsdI-CoPPIX viewed from the distal side with a F_o - F_c omit map contoured at 3 σ. C, active site of IsdI-CoPPIX with residues that interact with porphyrin ring depicted in stick model and labeled. In each panel, residues in the active site are depicted in the stick model and labeled. Oxygen atoms are red, nitrogen atoms are blue, and the chloride ion is a magenta sphere. CoPPIX carbons are gray, and amino acid residue carbons are colored according to the main chain of each structure.

In a superposition of a monomer of IsdI-CoPPIX and IsdG-hemin (r.m.s.d. of 0.64 Å over 105 C-α atoms), conserved active site residues generally overlap (supplemental Fig. S5). However, the porphyrin ring of IsdG-hemin is displaced as much as ∼1.1 Å in the direction of residue Ala⁷, which is an asparagine in the IsdI-CoPPIX structure (supplemental Fig. S5). Nonetheless, the N-ε2 atom of the homologous His⁷⁶ in IsdI-CoPPIX coordinates to the cobalt atom (∼2.1 Å). As in IsdG-hemin, the side chain rings of His⁷⁶ and Phe⁷² are stacked, and the imidazole N-δ1 atom forms a hydrogen bond to the main chain carbonyl group of Phe⁷² (Fig. 3C). In contrast, the cobalt atom lies closer to the pyrrole nitrogen plane (<0.1 Å displacement).

A strong spherical density was observed on the distal side of CoPPIX at a distance of ∼2.6 Å from the metal (Fig. 3B). A water-Co³⁺ coordination is expected at a much shorter distance of ∼2.2 Å as observed in a CoPPIX myoglobin acuomet structure (29), suggesting that an alternative sixth ligand is bound in the IsdI-CoPPIX structure. Based on the presence of 200 mm of sodium chloride in the protein solution used for crystallization, the spherical density was refined successfully as a coordinating chloride ion. The resulting B-factor of the ion is ∼20 Ų with no noticeable residual peaks in an F_o - F_c map. To accommodate the chloride ion in the active site, the χ² angle of Ile⁵³ in the IsdI-CoPPIX structure is rotated ∼60° relative to the conformation of the equivalent Ile in the IsdG-hemin structure. The chloride ion is positioned to form a hydrogen bond (3.1 Å) to the N-δ2 atom of Asn⁶ (Fig. 3B).

As in IsdG-hemin, the guanidine moieties of Arg²¹ and Arg²⁵ form salt bridges with the propionic groups of IsdI-CoPPIX. The side chain of Gln²⁷ and main chain nitrogen atoms of Gln²⁷ and Ile²⁸ also interact with the propionic groups. The α-meso carbon is exposed to the surface, and the solvent-accessible surface of the CoPPIX group is ∼93 Ų.

The porphyrin ring of IsdI-CoPPIX is more distorted from planarity than that of the IsdG-hemin structure (Fig. 3C). In addition to the difference as to the metal present, the bulkier side chain of Asn⁶ instead of Ala⁷ in the IsdG variant may explain the increased distortion of the porphyrin ring. Furthermore, the hydrophobic residues that interact with the porphyrin ring moiety are largely conserved, although some of the contact points to the porphyrin ring of CoPPIX differ some-what (supplemental Fig. S4). The four residues implicated in porphyrin ring distortion, Asn⁶, Phe²², Trp⁶⁶, and Val⁷⁹, all make contact with a meso-carbon at distances that range from 3.5 to 3.7 Å (supplemental Fig. S4).

DISCUSSION

The most striking features of the IsdG-hemin and IsdI-CoPPIX crystal structures are the highly distorted porphyrin rings. This structural similarity suggests that the porphyrin ring distortions are not artifacts of either the N7A substitution in IsdG-hemin or the presence of a nondegraded heme analog in IsdI-CoPPIX. Instead, a consistent pattern of amino acid residue contacts to the porphyrin ring appears to force the distortion from planarity. The IsdI-CoPPIX structure shows that ring distortion alone is insufficient for reductive degradation. Moreover, as shown by assays using a panel of metalloporphyrins, the nature of the metal is critical, and only hemin is degraded by either enzyme (Fig. 1).

Porphyrin ring distortions have been observed to varying extents in other heme protein structures (30). Analysis of the out-of-plane distortions of porphyrins bound to IsdI and the IsdG variant was made by the normal-coordinate structural decomposition (NSD) method (26). In both the IsdG-hemin and IsdI-CoPPIX structures, the porphyrin ring distortion is best described as ruffled (Fig. 4). The out-of-plane distortions were found to be 1.9 and 2.3 Å for the IsdG-hemin and IsdI-CoPPIX structures, respectively. Notably, these distortions are far greater than observed previously for a heme group bound to a protein. In comparison, the next most distorted porphyrin ring (∼1 Å) found in a literature search is that bound to the heme domain of a bacterial homolog of soluble guanylyl cyclase (31). Also, similar hydrophobic contacts to those at the meso-carbons in IsdI-CoPPIX are observed in the heme ruffling in ferrocytochrome c₃ (32).

FIGURE 4.

Porphyrin ring distortion analysis. NSD out-of-plane (Å) analysis (26) of the porphyrin rings from structures as follows: ChuS, E. coli HO (PDB ID 2Hq2); N-HO, Neisseria meningitidis HO (PDB ID 1J77); H-NOX, oxygen binding H-NOX domain (PDB ID 1U4H); CytC, horse heart cytochrome c (PDB ID 1HRC). Pro, wav, dom, ruf, and sad correspond to porphyrin distortions described as propeller, wave, dome, ruffle, and saddle, respectively.

Although the porphyrins are highly distorted upon binding, the overall folds of IsdG and IsdI are more rigid as evidenced by the superposition of the apo- and PPIX-bound complexes (less than 0.8 Å r.m.s.d. for all C-αs). The IsdG and IsdI folds belong to PG130-like protein family in which two ferredoxin-like β-sheets from each monomer come together to form a β-barrel (33). This family includes the actinorhodin biosynthesis monooxygenase (ActVa-Orf6), for which structures in complex with substrate analogs have been determined (34). A comparison of these ActVa-Orf6 structures to the apo-IsdG and IsdI crystal structures has led to the suggestion that IsdG and IsdI may also be monooxygenases (17). Indeed, the heme degradation reaction requires a reductant and oxygen (Fig. 1 and supplemental Fig. S1). However, in the absence of precise information regarding the product of heme degradation, the assignment of IsdG and IsdI as monooxygenases has yet to be proven.

The precise structure of the IsdG and IsdI reaction product remains elusive. A mass of 599 Da has been obtained by mass spectrometry; however, the product is unstable, limiting further analysis (35). The observed mass is much greater than the theoretical masses of biliverdin (583 Da) or PPIX (563 Da). The lack of mass correspondence with billiverdin is consistent with the yellow color of the product, which suggests a greater loss of conjugation of the porphyrin ring. In contrast, a mass of 583 Da was reported for the product of HmuQ (19); however, assignment of the product as biliverdin is inconsistent with reported visible spectra for the products of HmuQ, IsdG and IsdI. These spectra do not possess a peak at 680 nm characteristic of biliverdin (36). Regardless of the precise nature of the product, the release of iron as a result of heme degradation (16) strongly suggests that the porphyrin ring is cleaved during the reaction.

The inability of the N7A variant of IsdG to degrade heme supports a direct role for the side chain amide of Asn⁷ in the catalytic mechanism of IsdG as well as for the homologous residue in IsdI (17). In the IsdI-CoPPIX structure, Asn⁶ forms a hydrogen bond to the chloride ion coordinated to the cobalt atom. The presence of an amide forming a hydrogen bond to an exogenous ligand bound at the distal side is reminiscent of dioxygen in the H64Q variant of sperm whale myoglobin (37). Superposition of the porphyrin rings of the myoglobin variant and IsdI-CoPPIX shows that the relative positions of the amides of Gln⁶⁴ and Asn⁶ are remarkably similar (Fig. 5). Notably, this myoglobin variant was shown to have increased autoxidation activity, supporting a model for IsdG and IsdI whereby dioxygen is reduced and stabilized by an amide hydrogen bond. Moreover, Asn⁶² of ActVA-Orf6 is proposed to be required for the binding and stabilization of a peroxo intermediate (34). This residue is also located in the active site analogous to Asn⁶ in IsdI.

FIGURE 5.

Heme binding modes. Superposition of sperm whale myoglobin H64Q variant (PDB ID, 1MCY; cyan) and IsdI-CoPPIX (green). The porphyrin rings were superposed such that Asn⁷ from IsdI and Gln⁶⁴ from the myoglobin H64Q variant overlap. These residues, the porphyrin rings, and the proximal histidine ligands are depicted as sticks. The chloride ion is depicted as a magenta sphere. Oxygen atoms are red, nitrogen atoms are blue, and carbons are colored according to the structure. Myoglobin residues are labeled with an asterisk.

Removal of the chloride ion from the IsdI-CoPPIX structure and analysis by the CASTp server (38) with a probe radius of 1.4 Å reveals a cavity with a volume of ∼35 ų. The cavity is lined by residues Asn⁶, Leu⁸, Phe²², Ile⁵³, and the β- or δ-meso carbons of CoPPIX (supplemental Fig. S6). A recent crystal structure of formylglycine-generating enzyme, a non-heme oxygenase, in complex with a substrate mimic has a chloride ion bound in a small cavity proposed to bind oxygen (39). The chloride cavity in the IsdI-CoPPIX structure is large enough to accommodate an oxygen species. The binding of the chloride ion is accompanied by a conformational change in the side chain of Ile⁵³ resulting in the displacement of the C-δ1 atom. Thus, the conformational change in Ile⁵³ may play a role in preventing water from occupying the sixth iron coordination position or gating binding of ligands during the catalytic cycle. A comparison of UV-visible spectra of wild-type and N7A variant IsdG (supplemental Fig. S7) cannot exclude that the wild-type protein is six-coordinate in the resting state.

In most HOs characterized so far, the α-meso carbon is the preferred site of initial hydroxylation. The regiospecificity is proposed to be dependent on steering and shielding by the water network on the distal side (40). An exception is an HO from Pseudomonas aeruginosa, which favors the δ-meso carbon through a reorientation of the porphyrin ring in the active site (41). In P. aeruginosa, a bacterial phytochrome is implicated in HO product release (42). A homologous phytochrome has not been identified in S. aureus; however, the PPIX degradation products could be released from IsdG and IsdI by a protein with an analogous function.

The bacterial and eukaryotic HO-1 homologs have been shown to bind dioxygen to heme iron, and when activated this oxygen species is proposed to be protonated through a cluster of distal waters bound in the active site (43, 44). A recent crystal structure of ChuS, a different type of heme oxygenase from E. coli O159:H7, also has a network of water molecules on the distal side of the heme (45). Conversely, in the IsdG-hemin and IsdI-CoPPIX structures, no such water network is present on the distal side to serve as a proton donor to the dioxygen. Besides, no obvious alternative proton donor is readily identified from an inspection of the active sites.

The heme molecule bound to HO-1 is far less ruffled (<0.5 Å) and that of ChuS is intermediate by the same NSD analysis (Fig. 4). These key structural differences suggest that the mechanism of heme degradation by the IsdG and IsdI enzymes differs significantly from that of either the HO-1 or ChuS enzyme families. A catalytic role for heme ruffling is supported by the loss of activity in the W67A variant of IsdG (17). Trp⁶⁷ forms a steric contact with the β-meso carbon in the IsdG-hemin structure, and substitution to alanine would be expected to diminish the ruffle distortion. Furthermore, the heme distortion directly imposes torsional strain at the meso-carbons to an sp³-like conformation that would favor nucleophilic attack (36). The porphyrin ring ruffling may also perturb the reduction potential of hemin bound to IsdG or IsdI. Heme ruffling has been shown to lower reduction potential favoring the ferric form (46). For example, heme ruffling observed in the structure of nitrophorin (47) is proposed to favor the ferric state thereby lowering nitric oxide affinity. In the absence of an understanding of the catalytic mechanism, a role for decreasing reduction potential for IsdG and IsdI is not apparent.

Non-iron metalloporphyrins are potent inhibitors of bacterial heme oxygenases and have been proposed as novel therapeutics targeting bacterial infections (48). More specifically, GaPPIX has broad spectrum anti-bacterial activity and has been referred to as a molecular Trojan horse (35, 49). A potential mechanistic explanation for this toxicity is provided by our findings that Ga-PPIX is not degraded by either IsdG or IsdI (Fig. 1). Presumably, non-iron metalloporphyrins accumulate in the bacterium because of their inability to be degraded by microbial heme oxygenases. In turn, accumulated metalloporphyrins become substrates for hemoproteins that require the reactivity of the iron atom to carry out their biochemical function. This would result in inactivation of bacterial hemoproteins and subsequent cell death. These findings strengthen the supposition that non-iron metalloporphyrins may represent a novel class of antimicrobials with potential therapeutic implications.