IruO Is a Reductase for Heme Degradation by IsdI and IsdG Proteins in Staphylococcus aureus*

Background: Staphylococcus aureus utilizes heme as an iron source during an infection. Results: An oxidoreductase, IruO, can supply electrons to IsdI and IsdG for heme degradation and iron extraction. Conclusion: IruO is likely the in vivo reductant for heme degradation to the staphylobilins. Significance: Heme degradation is a potential target for anti-S. aureus therapeutics. Staphylococcus aureus is a common hospital- and community-acquired bacterium that can cause devastating infections and is often multidrug-resistant. Iron acquisition is required by S. aureus during an infection, and iron acquisition pathways are potential targets for therapies. The gene NWMN2274 in S. aureus strain Newman is annotated as an oxidoreductase of the diverse pyridine nucleotide-disulfide oxidoreductase (PNDO) family. We show that NWMN2274 is an electron donor to IsdG and IsdI catalyzing the degradation of heme, and we have renamed this protein IruO. Recombinant IruO is a FAD-containing NADPH-dependent reductase. In the presence of NADPH and IruO, either IsdI or IsdG degraded bound heme 10-fold more rapidly than with the chemical reductant ascorbic acid. Varying IsdI-heme substrate and monitoring loss of the heme Soret band gave a Km of 15 ± 4 μm, a kcat of 5.2 ± 0.7 min−1, and a kcat/Km of 5.8 × 103 m−1 s−1. From HPLC and electronic spectra, the major heme degradation products are 5-oxo-δ-bilirubin and 15-oxo-β-bilirubin (staphylobilins), as observed with ascorbic acid. Although heme degradation by IsdI or IsdG can occur in the presence of H2O2, the addition of catalase and superoxide dismutase did not disrupt NADPH/IruO heme degradation reactions. The degree of electron coupling between IruO and IsdI or IsdG remains to be determined. Homologs of IruO were identified by sequence similarity in the genomes of Gram-positive bacteria that possess IsdG-family heme oxygenases. A phylogeny of these homologs identifies a distinct clade of pyridine nucleotide-disulfide oxidoreductases likely involved in iron uptake systems. IruO is the likely in vivo reductant required for heme degradation by S. aureus.

used as the electron donor (22). IsdG and IsdI cleave the porphyrin ring at either the ␦-meso or ␤-meso carbons, resulting in two different products, 5-oxo-␦-bilirubin and 15-oxo-␤-bilirubin, that are known as the staphylobilins. They are similar to but distinct from biliverdin, the product of heme degradation by conventional heme oxygenases such as human heme oxygenase (HO-1), suggesting that the reaction mechanism is different (27). Unlike HO-1, which generates CO during heme degradation, IsdG and IsdI generate formaldehyde (28). Heme bound to IsdG and IsdI is significantly distorted from planarity in a fashion described as ruffling (29,30); IsdI amino acid variants with decreased heme ruffling capability have decreased heme degradation rates (31). Outstanding questions about heme degradation in S. aureus, include how the reaction differs from other heme degrading enzymes to generate these novel products, what is the intracellular fate of the staphylobilins, and what is the in vivo electron donor for the reaction?
Here, we show that a protein encoded by NWMN2274 in S. aureus strain Newman can act as a source of electrons for heme degradation by IsdG and IsdI in the presence of NADPH. In vitro heme degradation in the presence of this protein yields the same products as reactions with ascorbic acid as an electron donor. From the specificity of the reaction, NWMN2274 is proposed to be the biological reductase associated with IsdG or IsdI heme degradation within the cytoplasm of S. aureus, and we have named this protein the iron utilization oxidoreductase, or IruO.

EXPERIMENTAL PROCEDURES
Chemicals-All chemicals were obtained from Fisher unless noted below.
IsdI and IsdG were expressed in E. coli BL21 (DE3) cells from the plasmid pET15b, purified by His-tag affinity chromatography, and digested with the tobacco etch virus protease to remove the His tag as previously described (22). Tobacco etch virus protease was purified as previously described (33).
FAD Identification-Electronic spectra from 250 to 800 nm were measured with a Cary 50 Bio UV-visible spectrophotometer. Samples were NWMN2274 (as purified), NWMN2274 after heat denaturation and protein removal, and FAD, FMN, and riboflavin standards (Sigma). All samples were at 30 M in a buffer of 50 mM Tris-HCl (pH 8.0), 100 mM NaCl. To heatdenature and remove NWMN2274 from the flavin molecule, the protein solution was boiled for 10 min and centrifuged at 21,000 ϫ g for 3 min, and the supernatant was centrifuged through a Nanosep centrifugal device (PALL Life Sciences) with a molecular mass cut-off of 3 kDa as described previously (34).
Flavin removed from NWMN2274 as well as FAD, FMN, and riboflavin standards (all at 20 M) were separated using an Infinity 1260 Quaternary high performance liquid chromatography (HPLC) system (Agilent) equipped with an Aqua 5-mm C 18 column (Phenomenex). A previously established procedure was utilized (34) with modifications. A flow rate of 1 ml/min and a column temperature of 20°C were maintained during the entire analysis. The column was equilibrated with 85% solvent A (10 mM ammonium acetate (pH 6.5)) and 15% solvent B (methanol) at sample injection. After a 5-min post-injection period, a linear gradient was developed over 20 min to 75% solvent B followed by a second linear gradient over 5 min to 100% solvent B. Flavins were detected by absorption at 264 nm using an Infinity 1260 multiple wavelength detector (Agilent).
Heme Degradation Assays-In vitro heme degradation assays were conducted as previously described (29). 10 mM stock solutions of porcine hemin (Sigma) in 0.1 M NaOH were prepared and kept at Ϫ20°C. 10 M IsdI and 10 M porcine hemin (Sigma) were mixed in buffer containing 50 mM Tris-HCl (pH 7.5) and 150 mM NaCl and incubated at 4°C for 1 h. Spectra from 300 to 800 nm were measured with a Cary 50 Bio UVvisible spectrophotometer. As noted below, some reactions were also supplemented with 200 M NADPH (EMD Biosciences), 200 M NADH (Roche), 5 M bovine liver catalase (Sigma), or 4 units/ml bovine erythrocyte superoxide dismutase (Sigma). To initiate degradation reactions, 1 mM ascorbic acid, 1 M NWMN2274, or NWMN0732, 20 M to 2 mM H 2 O 2 , or 5 units/ml Aspergillus niger glucose oxidase (Sigma), and 1 mM glucose (Sigma) were added to cuvettes. Spectra were recorded either every minute or every 10 min for up to 90 min depending on the rate at which the reaction progressed.
For kinetic analysis, 600 M NADPH and 0.1 M NWMN2274 were added to reactions containing 1-25 M IsdG-heme or IsdI-heme, and the decrease in the Soret peak at 412 nm was monitored every 0.1 s for 180 s. The concentration of IsdI-heme was determined from Soret absorbance measurements for 30 s starting at 10 s after the addition of the reductase using the reported extinction coefficients for IsdG-heme (131 mM Ϫ1 cm Ϫ1 ) and IsdI-heme (126 mM Ϫ1 cm Ϫ1 ) (22). Slopes determined by linear regressions of these data were taken as initial reaction rates. Initial steady-state rates were plotted against the concentration of IsdG-heme or IsdI-heme, and a non-linear regression of the data were calculated to fit Michaelis-Menten kinetics. Linear and non-linear regressions were calculated using GraphPad Prism 6.
Heme Degradation Product Analyses-For these experiments, 1-ml reactions containing 100 M IsdI-heme were prepared, and 2 mM ascorbic acid, 2 mM H 2 O 2 , 2 units/ml glucose oxidase, and 5 mM glucose or 1 mM NADPH and 5 M NWMN2274 or NWMN0732 were added to initiate reactions. Reactions were monitored, and once complete, heme degradation products were purified as previously described (27) with one important modification. Purification includes sample filtration through a Nanosep centrifugal device. The presence of the larger NWMN2274 protein blocked the pores of spin columns with a molecular mass cutoff of 3 kDa and prevented samples from passing easily through the column. Pore size was increased to a cutoff of 10 kDa, which improved product purification significantly, but lower yields were typically obtained from the NWMN2274/NADPH reactions than from the ascorbic acid reactions.
HPLC separation of the degradation products was completed as previously described (27) using either a Waters 2695 separation module with a Waters 2996 photodiode array detector or an Agilent Infinity 1260 multi-wavelength detector. With either setup the flow rate was 0.5 ml/min, and a Waters XTerra C18 column was used.
Bioinformatics Analyses-Homologs of NWMN2274 were identified with BLASTP (35) searches of the annotated genomes of S. aureus, Listeria monocytogenes, Bacillus subtilis, Bacillus anthracis, and Mycobacterium tuberculosis. Only hits with E-values equal to or less than 0.001 and covering 40% or more of the query sequence were considered significant. Multiple sequence alignments were generated with T-Coffee (36,37). Maximum likelihood phylogenetic trees were built with the PhyML method in SeaView Version 4 (38) using the LG model and bootstrapping with 100 replicates.

Identification of NWMN2274 as a Putative Electron Donor to
IsdG and IsdI-The heme-degrading proteins IsdG and IsdI require a source of electrons for porphyrin cleavage and iron release (22). We sought to identify a protein that could act as the in vivo source of electrons when IsdI and IsdG degrade heme within the cytoplasm of S. aureus. No candidate reductases are located within the two operons that encode the Isd system of S. aureus or in the regions immediately upstream or downstream. We hypothesized that the candidate reductase would be annotated in the genome as a reductase and that mRNA expression would be co-regulated with other Isd genes.
Data from several published microarray studies demonstrating expression changes in the Isd genes (39 -46) were analyzed for changes in expression of uncharacterized genes encoding predicted reductases. Using a combination of cell culture and infection models with a bovine mastitis S. aureus isolate (strain SHY97-3906), Allard et al. (39) found that SACOL2369 was up-regulated under both iron-restricted growth conditions in cell culture and in tissue cages embedded in mice abdomens. SACOL2369 is annotated as a pyridine nucleotide-disulfide oxidoreductase (PNDO), and the authors of this study confirmed the microarray results by real time PCR. Identified upstream of the homologous gene (SA2162) in S. aureus strain N315 is a putative Fur box, the operator sequence to which the ferricuptake regulator (Fur) binds in the promoter of iron-regulated genes (47,48). Most other Isd genes were also up-regulated under the conditions of this study. Furthermore, Malachowa et al. (42) showed that isdA-I were all highly up-regulated (in some cases up to 200-fold) in S. aureus strain USA300 when grown in human blood or serum compared with standard media and that SAUSA300_2319 (the USA300 gene orthologous to SACOL2369) was also up-regulated in serum and blood, although not nearly to the same extent.
The Isd system has been extensively characterized in S. aureus strain Newman, and the orthologous gene of SACOL2369 is NWMN2274 (Fig. 1A). The genomic context of NWMN2274 (49) is similar to that of SACOL2369 (50) and SAUSA300_2319 (51), and all three genes encode nearly identical proteins (Ͼ97% amino acid identity). NWMN2274 is in a predicted two-gene operon with NWMN2273 that encodes a putative GCN5-like N-acetyltransferase. The 231-base pair intergenic region between NWMN2274 and NWMN2275 includes probable Ϫ10 and Ϫ35 sites for RNA polymerase binding and a predicted Fur box identical to that identified by Allard et al. (39) for SA2162 (Fig. 1B). The gene encodes a 344amino acid protein (Fig. 1C) with a predicted Rossman fold domain for NAD(P) binding and a predicted PNDO domain (49).
NWMN2274 Binds FAD-Purified NWMN2274 in solution is a dark yellow color consistent with binding of a flavin group.
To determine the identity of the unknown flavin, we used HPLC to compare it to FAD, FMN, and riboflavin standards. The unknown flavin isolated from NWMN2274 had the same retention time by HPLC analysis as FAD and the retention time differed significantly from FMN and riboflavin (Fig. 2). In addition, the unknown flavin had absorption maxima at 376 and 450 nm and a spectrum nearly identical to that of a FAD standard and differed from that of FMN or riboflavin standards that both have absorption maxima at 375 and 447 nm (data not shown). The absorption maxima of FAD at 450 nm and FMN at 447 nm are consistent with previous studies and can be used to distinguish between the two molecules (34). Together these data indicate that NWMN2274 is an FAD-binding protein.

NWMN2274 and NADPH Are an Electron Source for Heme
Degradation by IsdI or IsdG-As previously demonstrated (29), upon the addition of ascorbic acid to IsdI-heme or IsdG-heme, the Soret peak at 412 nm decreases over the course of 90 min, indicative of heme degradation (Fig. 3, A and B). When a 20-fold excess of NADPH was added to IsdI-heme or IsdGheme and purified NWMN2274 (Fig. 1D) was added to initiate the reaction, there were rapid decreases in both absorption at 340 nm, indicating that NADPH was being oxidized to NADP ϩ , and at 412 nm, indicating that heme was being degraded (Fig. 3, C and D). Heme degradation occurred more rapidly under these conditions with the reactions nearly complete by 10 min. The data in Fig. 3, C and D, suggest that, at least qualitatively, the reaction with IsdG-heme is slower than with IsdI-heme. NWMN2274 or NADPH added alone to either IsdI-heme or IsdG-heme did not initiate heme degradation (data not shown). Similar experiments were also conducted with NADH substituted for NADPH, and in the presence of NWMN2274 these reactions progressed but at rates slower than the control reactions in the presence of ascorbic acid (data not shown). Taken together, these experiments indicate that NWMN2274 can act as a reducing agent in the presence of NADPH and provides electrons for heme degradation by IsdI and IsdG.
The stoichiometry of heme degradation by IsdG and IsdI is unknown, including the number of electrons required. Some uncoupled oxidation of NADPH to yield H 2 O 2 and superoxide is likely to occur in these in vitro reactions, and thus the equivalents of NADPH consumed does not necessarily equate to reduction equivalents for heme degradation.
NWMN2274 Kinetic Parameters-To determine steadystate kinetic parameters for NWMN2274 with IsdI-heme or IsdG-heme as the substrate, initial velocities were plotted against the concentration of IsdG-heme or IsdI-heme, and data were fit by nonlinear regression to the Michaelis-Menten equation (Fig. 4). From these data the k cat of the overall reaction in the presence of IsdI-heme was determined to be 0.09 Ϯ 0.01 s Ϫ1 . The K m of NWMN2274 for IsdI-heme was calculated to be 15 Ϯ 4 M. The specificity constant (k cat /K m ) for NWMN2274 catalyzed heme degradation was 5.8 ϫ 10 3 M Ϫ1 s Ϫ1 . Our data from reactions at concentrations below the K m fit the theoretical curve well, whereas above the K m there is greater error. This may be due to heme degradation by IsdI being a multistep pro-  (49). B, the intergenic region between NWMN2274 and NWMN2275 contains predicted Ϫ10 and Ϫ35 sites for RNA polymerase binding, a Fur box, and a ribosome binding site (RBS) that were manually identified. C, NWMN2274 encodes a 344-amino acid protein with a Rossman fold (including the consensus GXGXXG motif beginning at the eighth amino acid) and a PNDO domain predicted by Pfam (57). D, NWMN2274 was purified to Ͼ95% homogeneity and is close to its predicted size (38 kDa) when separated by SDS-PAGE and stained with Coomassie. cess. At high concentrations of IsdI-heme some heme may only be partially degraded and not contribute to decreased Soret peak height. As mentioned above, some uncoupling of NWMN2274 from IsdI may be present and could alter the observed kinetics. The potential effect of uncoupling is mitigated by the presence of excess of IsdI-heme to NWMN2274 (10 -250-fold) and the monitoring of the heme Soret peak to define the rate. Additionally, as shown below (see Fig. 7), the addition of catalase and superoxide dismutase did not significantly alter heme degradation reactions.
Reactions with IsdG-heme were slower and, at the concentrations of IsdG-heme tested, increases in enzyme velocity were still in the linear range (Fig. 4). We were, therefore, unable to calculate reliable kinetic parameters for this reaction. This observation leaves open the possibility that IsdG may preferentially use an alternative reductase or interact with some other component in vivo.
IsdI-Heme Reactions with NWMN2274 and NADPH Generate the Staphylobilins-Heme degradation products derived from IsdI with either ascorbic acid or NWMN2274/NADPH were analyzed by HPLC. As previously reported (27), reactions with ascorbic acid resulted in two major products that absorb at 465 nm (Fig. 5A). Two products with similar HPLC retention times (Fig. 5B) and electronic spectra (Fig. 5, C and D) were also obtained from reactions with NWMN2274 and NADPH. Spectra for these peaks are highly similar to those previously published for the staphylobilins (27). These spectra differ from the IsdI/heme degradation spectra (Fig. 3) due to alterations in the molecular environment. Heme and products are removed from the protein to a solution containing 50% acetonitrile and 0.1% trifluoroacetic acid. Note that the products are also at least 10 times more concentrated than in the degradation reaction spectra (Fig. 3). These data confirm that the same reaction products are derived when IsdI cleaves heme with either ascorbic acid or NWMN2274/NADPH as a reductant. Thus, the staphylobilins identified in in vitro degradation assays are likely produced by IsdI cleavage of heme within an S. aureus cell as well.
The Presence of H 2 O 2 Also Leads to Heme Degradation but with an Altered Product-Based on previous attempts to crystallize IsdI or IsdG with heme for structural determination, it was observed that red crystals containing protein bound to heme were only obtained with either inactive variants of IsdG (29) or for wild-type IsdI crystallized at 4°C (27). We hypothesized that at room temperature and with active protein, a component in the crystallization solution leads to heme degradation  and a resulting failure to crystallize protein bound to intact heme. In fact when enzyme assays similar to those above are conducted in conditions that mimic the crystallization conditions (0.1 M Bis-tris (pH 5.5), 0.2 M MgCl 2 , 25% polyethylene glycol-3350) there is slow heme degradation at room temperature but not at 4°C (data not shown), and we believe that peroxides generated by the polyethylene glycol-3350 may be the cause. Under our standard reaction conditions, but with H 2 O 2 added to initiate the reaction in place of ascorbic acid or NWMN2274/NADPH, heme was degraded by IsdI (Fig. 6A).
The product of the IsdI-heme reaction with H 2 O 2 was analyzed by HPLC and consisted of one major peak that absorbs at 405 nm but not 465 nm (Fig. 6B). The retention time of the peak is very similar to that of heme extracted from untreated IsdIheme and analyzed by HPLC (Fig. 6B). Spectra of both are similar with an absorbance maximum at 398 nm (Fig. 6C). Similar results were obtained for heme degradation reactions by IsdG in the presence of H 2 O 2 (data not shown). These data indicate that reaction of IsdI-heme or IsdG-heme with H 2 O 2 leads to loss of the Soret peak, a heme degradation product that is similar to heme and, importantly, no production of the staphylobilins.
Catalase and Superoxide Dismutase Do Not Inhibit Heme Degradation by NWMN2274/NADPH-Based on these observations, we thought that degradation of heme by IsdI or IsdG in the presence of NWMN2274/NADPH could occur via two pathways. One, NWMN2274 could oxidize NADPH to NADP ϩ and provide electrons directly to IsdI or IsdG for heme degradation. Or two, NWMN2274 could oxidize NADPH to NADP ϩ and transfer electrons to mediators in the reaction mixture, such as dissolved dioxygen, forming superoxide or hydrogen peroxide that may cause heme degradation by IsdI or IsdG.
To address these issues we conducted a series of experiments in which we either uncoupled the reaction, premixing NWMN2274 and NADPH for a period of time before the addition of IsdI-heme and/or added catalase and superoxide dismutase to reactions to see if they had a significant impact on heme degradation. First, we mixed NWMN2274 and NADPH in our standard reaction buffer but without IsdI-heme and followed the spectrum of the reaction for 30 min (Fig. 7A). In the absence of IsdI-heme, the NADPH absorption peak decreased rapidly, indicating that NWMN2274 still oxidized NADPH to NADP ϩ in the absence of IsdI-heme. Then IsdI-heme was added, and the reaction was monitored for an additional 20 min. In this uncoupled reaction, heme degradation still occurs despite the absence of NADPH. When this uncoupled reaction was conducted in the presence of catalase and superoxide dismutase, NADPH was oxidized to NADP ϩ over the first 30 min, but upon the addition of IsdI-heme to the reaction, very little heme degradation was observed (Fig. 7B). Thus, free superoxide or hydrogen peroxide, generated when NWMN2274 oxidizes NADPH in the absence of IsdI-heme, can serve as a mediator to transfer electrons from NWMN2274 to IsdI-heme.
Next, we conducted coupled experiments where IsdI-heme and NADPH were mixed in the standard reaction buffer and NWMN2274 was added to initiate the reaction. In the absence of catalase and superoxide dismutase (Fig. 7C), the reaction progressed as previously demonstrated (compare with Fig. 3G). The addition of catalase and superoxide dismutase (Fig. 7D) did not significantly alter the rate of heme degradation. The same result was obtained with IsdG-heme (data not shown). These latter data support a model of direct electron transfer from NWMN2274 to IsdG-or IsdI-heme. Because NWMN2274, IsdG, and IsdI are available to interact in the cell, this model is a FIGURE 5. IsdI-heme reaction products with characteristics of the staphylobilins are produced from either the ascorbic acid or NWMN2274/NADPH reactions. Reaction products were separated from proteins and analyzed by HPLC, and the elution profiles at 465 nm are shown for the ascorbic acid (A) and NWMN2274/NADPH (B) reactions. In both cases there are two major peaks that absorb at 465 nm and are labeled 1 and 2. Optical spectra for peaks 1 (C) and peaks 2 (D) from ascorbic acid (blue lines) and NWMN2274/NADPH (red lines) reactions are shown with wavelengths of maximum absorbance shown for each spectra either above or below the lines. much more likely scenario. These results agree with those of Skaar et al. (22) who found that catalase did not alter heme degradation by IsdG and IsdI in reactions initiated with ascorbic acid.
As negative controls, we also analyzed the results of IsdG-or IsdI-heme reactions with either an alternative PNDO (NWMN0732) and NADPH or glucose oxidase and glucose (which generates H 2 O 2 upon oxidation of glucose) in the pres-ence or absence of catalase and superoxide oxidase. In the absence of catalase and superoxide dismutase, the reactions progressed, and heme was degraded; however, unlike for NWMN2274, the addition of catalase and superoxide dismutase prevented heme degradation (data not shown). Furthermore, HPLC analysis of heme degradation products from reactions initiated with either NWMN0732/NADPH or glucose oxidase/glucose gave profiles similar to H 2 O 2 -initiated reac-  tions, with no indication of staphylobilin formation (data not shown). These data further support the hypothesis that NWMN2274 specifically interacts with IsdG and IsdI for heme degradation.
Based on the whole of our experimental evidence, we believe that NWMN2274 is an in vivo electron donor for IsdG and IsdI-mediated heme degradation to the staphylobilins, and we propose naming NWMN2274 IruO, for iron utilization oxidoreductase.
IruO Is an Archetype for a Family of PNDOs-Finally, we searched for other potential IruO proteins in Gram-positive bacteria with IsdG-family proteins. The sequences of IruO and three related PNDOs that have been structurally characterized, E. coli thioredoxin reductase (TrxB) (53, 58), B. subtilis YumC, a ferredoxin reductase (52), and E. coli alkylhydroperoxide reductase (AhpF) (56), were used to search the genomes of S. aureus strain Newman (49), B. subtilis strain 168 (59), B. anthracis strain Ames "Ancestor" (60), L. monocytogenes strain EGD-e (61), and M. tuberculosis strain H37Rv (62). Between 6 and 12 homologs were retrieved per organism, subjected to multiple sequence alignment, and organized into a phylogenetic tree (Fig. 8). IruO groups with a number of PNDOs have also been implicated by microarray and proteomic studies to be iron-and/or Fur-regulated. B. subtilis BSU03270 and Lmo1961 of L. monocytogenes are both Fur-and iron-regulated (63,64), whereas GBAA0352 of B. anthracis is up-regulated during iron starvation (65) and during culture in macrophages (66). Of the genomes searched, only that of M. tuberculosis has no immediately obvious IruO ortholog. Among the other PNDOs of S. aureus identified through this analysis, NWMN0732 is most closely related to IruO; however, in vitro the staphylobilins are not generated with NWMN0732 as an electron donor.
We also searched for IruO proteins in two other staphylococcal species. Staphylococcus lugdunensis contains an Isd heme import system similar to that of S. aureus (67) and has an experimentally characterized IsdG protein (68) that shares Ͼ60% amino acid sequence identity with S. aureus IsdG and IsdI. The genome of S. lugdunensis (69) contains a PNDO (SLUG_06580) that shares 68% amino acid sequence identity with S. aureus IruO, is similarly organized in a two-gene operon with an acetyltransferase, and is the protein most similar to S. aureus IruO in our phylogenetic analysis (Fig. 8). In contrast, in the genome of Staphylococcus epidermidis strain (50), no Isd heme import system is detected, and the one IsdG-family protein present is distantly related, sharing only ϳ35% identity with S. aureus IsdG and IsdI. BLASTP analysis of the S. epidermidis proteome with S. aureus IruO returns three hits that matched our search criteria SERP0059, SERP0432, and SERP1047. However, these proteins share Ͼ80% amino acid sequence identity with NWMN0371, NWMN0732, and NWMN1388, respectively, and group with them in our phylogenetic tree (Fig. 8).
Thus, the boxed clade in Fig. 8 represents a family of PNDOs used by these organisms to provide electrons to heme degrading enzymes, allowing iron to be released for bacterial use. These proteins could also potentially be used to reduce iron for its release from bacterial siderophores, although that remains to be experimentally demonstrated. Interestingly, this clade clusters with a second clade that contains YumC, a ferredoxin reductase (70). The IsdG and IsdI homodimers possess ferredoxin-like folds (23), and the associated reductase might be expected to resemble a ferredoxin reductase.

DISCUSSION
We show here that the protein encoded by NWMN2274 of S. aureus strain Newman catalyzes the transfer of electrons from NADPH to IsdI and IsdG for heme degradation. Likely some reactive oxygen species are generated through uncoupled oxidation of NADPH by NWMN2274; however, heme degradation to the staphylobilins is a consequence of a coupled enzymatic reaction between NWMN2274 and IsdI or IsdG bound to heme. From the determined kinetic parameters, the specificity constant of NWMN2274 for IsdI-heme is 5800 M Ϫ1 s Ϫ1 . Finally, degradation products from reactions with NWMN2274 and NADPH are the same as those from reactions with ascorbic acid. Based on these data, we have named the protein IruO for iron utilization oxidoreductase. We acknowledge the possibility that other reductases could act as sources of electrons for cytoplasmic heme degradation by S. aureus and that it may be possible to resolve this issue through characterization of an S. aureus iruO mutant.
A second PNDO from S. aureus (NWMN0732) does not induce heme degradation leading to the staphylobilins, suggesting that there is specificity between IruO and IsdG and IsdI. Human heme oxygenase utilizes cytochrome P450 reductase as an electron donor for during heme degradation, and Skaar et al. (22) demonstrated that human cytochrome P450 reductase could serve as an electron donor for IsdG-and IsdI-mediated heme degradation in vitro. The possibility exists that the putative cytochrome P450 reductase of S. aureus (NWMN2518) might also be an in vivo electron donor for heme degradation, particularly by IsdG. IruO may act as a reductase for other S. aureus iron acquisition pathways, a possibility that we are currently exploring. Our data indicate that there may be redundancies in these systems as B. subtilis, B. anthracis, and L. monocytogenes all have additional PNDOs (YumC, GBAA5160, and Lmo2390, respectively) that group in a clade near that of IruO (Fig. 8).
We identified iruO through examination of a number of microarray papers to identify a potential reductase protein whose gene expression was similar to other isd genes under iron limitation conditions. Other studies also show that a correlation exists between expression changes in previously identified isd genes and iruO, including treatment with peracetic acid (43), nitric oxide (45), and ortho-phenylphenol (40). However, correlation between gene expression changes in iruO and other isd genes is not absolute. IsdA-G but not iruO were up-regulated in the presence of hydrogen peroxide (41). A ⌬murF strain with defective cell wall biosynthesis decreased transcription of isdC-G but not of iruO (46). Finally, a ⌬clp mutant with impaired degradation of misfolded proteins increased expression of iruO and some other Fur-regulated genes but not the rest of the isd genes (44). These studies show that other regulatory elements may also tune the response of these genes to conditions the bacteria encounter. Our strategy of mining microarray papers for evidence of coordinated gene transcription changes could help to identify the pathways to which other IruO paralogs in S. aureus belong.
IruO is homologous to E. coli TrxB, an FAD binding enzyme that reduces thioredoxin (TrxA), which in turn reduces disulfide bonds in numerous cellular targets (71)(72)(73)(74). TrxB is thought to transfer electrons in multiple steps, first from NADPH to FAD, then from FAD to the protein disulfide active site, and finally to TrxA (75). IruO also binds FAD and favors NADPH over NADH as a source of electrons that are eventually funneled to IsdI-heme. The K m of IruO for IsdI-heme that we determined (14.9 M) is similar to the K m (3.0 M) of E. coli TrxB for TrxA; however, our measured k cat (0.09 s Ϫ1 ) is orders of magnitude lower than that of TrxB (23 s Ϫ1 ) (74). Kinetic parameters for TrxB were measured in an assay that monitored the reduction of DTNB by TrxA. Heme degradation by IsdI in the presence of IruO is a much slower, but also more complex reaction by comparison. Although low, the k cat value for heme degradation reported here is comparable to other reported values, as bovine and chicken heme oxygenases have turnover numbers of 0.058 and 0.32 s Ϫ1 , respectively (76,77).
Orthologs of IruO (Fig. 8) appear to also be present in other Gram-positive bacteria with IsdG-family heme oxygenases. B. subtilis, B. anthracis, and L. monocytogenes all possess IsdGfamily proteins and have homologs of IruO that are regulated in either iron-and/or Fur-dependent fashions (63-66). These observations and our phylogenetic analysis indicate that the function we have attributed here to IruO may be of significance to the lifestyle of other Gram-positive bacteria that obtain iron from heme.
Predicted IsdG orthologs are found across diverse classes of bacteria (68), but evidence suggests that other types of reductases may be used under some circumstances. For example, IsdG and IsdI share structural similarity to Streptomyces monooxygenases used in antibiotic synthesis, such as ActVA of Streptomyces coelicolor (23). However, the oxidoreductase ActVB oxidizes NADH to NAD ϩ , whereas it reduces FMN and then transfers reduced FMN to ActVA, which in turn re-oxidizes FMN during catalysis (78). No homolog of ActVB could be identified by a BLASTP analysis of the genome of S. aureus strain Newman. The highest ranked hit has an E-value of 2.0 with 30% amino acid sequence identity over only 24% percent of the query sequence. The heme degradation protein of M. tuberculosis, MhuD, shares sequence and structural similarity to IsdG and IsdI but has both distinct heme binding properties (79) and degradation products (80), and a candidate IruO homolog is not as evident. Finally, S. epidermidis, which has no Isd heme uptake system and a more distantly related IsdG homolog, has no IruO ortholog (Fig. 8).
Heme acquisition and utilization likely play an important role in S. aureus pathogenesis as heme iron has been proposed to be preferred by S. aureus to transferrin iron during an infection (25). A S. aureus mutant lacking both IsdG and IsdI grows poorly when heme is the only iron source (24, 68) Furthermore, S. aureus mutants lacking components of the Isd system have been shown to be defective in lung, heart, and kidney colonization during systemic murine infections (17,26). These data make the Isd system a target for the development of new antistaphylococcal drugs, and the identification of new members of this system provides additional targets for therapeutic development. Future directions include determining how the loss of IruO impacts S. aureus biology, particularly its ability to grow on heme as a sole source of iron and cause disease in experimental animal models. Also unknown is the nature of the physical interaction that occurs between IruO and IsdI or IsdG and how this interaction impacts the multistep reaction required for S. aureus heme degradation to the staphylobilins.