Staphylococcus aureus IsdG and IsdI, Heme-degrading Enzymes with Structural Similarity to Monooxygenases

Abstract

Heme-degrading enzymes are involved in human diseases ranging from stroke, cancer, and multiple sclerosis to infectious diseases such as malaria, diphtheria, and meningitis. All mammalian and microbial enzymes identified to date are members of the heme oxygenase superfamily and assume similar monomeric structures with an all α-helical fold. Here we describe the crystal structures of IsdG and IsdI, two heme-degrading enzymes from Staphylococcus aureus. The structures of both enzymes resemble the ferredoxin-like fold and form a β-barrel at the dimer interface. Two large pockets found on the outside of the barrel contain the putative active sites. Sequence homologs of IsdG and IsdI were identified in multiple Gram-positive pathogens. Substitution of conserved IsdG amino acid residues either reduced or abolished heme degradation, suggesting a common catalytic mechanism. This mechanism of IsdG-mediated heme degradation may be similar to that of the structurally related monooxygenases, enzymes involved in the synthesis of antibiotics in Streptomyces. Our results imply the evolutionary adaptation of microbial enzymes to unique environments.

Staphylococcus aureus acquires iron, an essential nutrient required for infection, by binding host heme-carrying proteins and extracting and transporting heme across the bacterial cell wall and plasma membrane envelope (1–3). Once inside the staphylococcal cytoplasm, heme is either incorporated into bacterial heme proteins or degraded to release iron for subsequent incorporation into polypeptides and cofactors (4–6). One heme acquisition system, encoded by the isd (iron-regulated surface determinants) gene cluster, encodes the heme-degrading enzyme IsdG (1, 5). IsdI, a homolog of IsdG, is encoded elsewhere on the staphylococcal chromosome (5). IsdG and IsdI show no significant sequence similarity to known heme oxygenases (7–10) and do not contain the conserved N-terminal histidine or the GXXXG motif characteristic for these enzymes. Nonetheless, purified IsdG and IsdI cleave heme tetrapyrrole in the presence of suitable electron donors (2, 5). This activity can functionally substitute for the classical heme oxygenase activity and permits growth of Corynebacterium ulcerans (5) lacking the HmuO heme oxygenase on media with hemin as a sole source of iron (8). Here we address the question of whether IsdG and IsdI have structures and catalytic mechanisms similar to those of members of the heme oxygenase superfamily (11), enzymes degrading heme in the cytoplasm of eukaryotic cells and in some bacterial species such as Corynebacterium diphtheriae, Pseudomonas aeruginosa, and Neisseria spp (9).

Experimental Procedures

Preparation of Proteins for Crystallization

The cells were grown at 37 °C in Luria-Bertani broth in the presence of 100 μg/ml ampicillin and 30 μg/ml kanamycin, respectively. Expression of His-tagged fusion proteins in Escherichia coli strain BL21(DE3) (12) was induced with 1 mm d-isopropyl-β-thiogalactoside when the optical density at 600 nm reached ∼0.6 and incubated at 20 °C overnight. The cells were harvested by centrifugation and suspended in five volumes of lysis buffer containing 50 mm HEPES, pH 8.0, 300 mm NaCl, 10 mm imidazole, 10 mm β-mercaptoethanol, 5% glycerol, and inhibitors of proteases (P8849; Sigma-Aldrich). For IsdG cells obtained from 2 liters of culture (∼5.0 g) were incubated for 30 min on ice with 1 mg/ml lysozyme (Sigma-Aldrich) followed by sonication (6 × 30 s, on ice) (13). The sample was centrifuged at 30,000 × g (RC5C-Plus centrifuge; Sorvall) for 20 min, and supernatants were filtered through 0.4- and 0.22-μm in-line membranes (Gelman Sciences Inc., Ann Arbor, MI). His₆-tagged IsdG was purified by affinity chromatography using Ni-NTA¹ Superflow resin (Qiagen, Valencia, CA). Eluted IsdG (total yield, 119 mg) was treated with the His₇-tagged tobacco etch virus protease to remove the His₆ tag for 16–24 h at 4 °C following basic protocol (13). The cleavage was monitored by SDS-PAGE and Coomassie Brilliant Blue R (Amersham Biosciences) staining. After the cleavage, the reaction mixture was applied to a 1-ml Ni-NTA column, and the column was washed with 3 column volumes of buffer A. All of the chromatographic steps were performed at 22 °C. The column flow-through and wash were collected and then applied to a desalting column Sephadex G-25 fine XK 26/20 (Amersham Biosciences) and equilibrated with storage buffer containing 20 mm Tris-HCl, pH 7.5, 500 mm NaCl, 2 mm dithiothreitol. The protein was analyzed by SDS-PAGE stained with Coomassie Brilliant Blue R. The IsdG was concentrated with simultaneous buffer exchange using Centriplus-3 (Amicon Bioseparations, Bedford, MA) (3-kDa cutoff). The final IsdG yield was 90 mg.

IsdI was purified using this same procedure (13). ∼6.8 g of cells obtained from 2 liters of culture were treated with lysozyme (Sigma) followed by sonication (6 × 30 s, on ice) (13). The sample was centrifuged at 30,000 × g for 20 min, and the supernatants were filtered through 0.4- and 0.22-μm in-line membranes (Gelman Sciences Inc.). His₆-tagged IsdI was purified by affinity chromatography using Ni-NTA resin. The total yield of IsdI after Ni-NTA column was 110 mg. The IsdI was also treated with the His₇-tagged tobacco etch virus protease to remove the His₆ tag. The cleavage was monitored by SDS-PAGE. After the cleavage, the reaction mixture was applied to a 1-ml Ni-NTA column, and the column was washed with 3 column volumes of buffer A. The column flow-through and wash were collected and then applied to a desalting column Sephadex G-25 fine XK 26/20 (Amersham Biosciences) equilibrated with storage buffer containing 20 mm Tris-HCl, pH 7.5, 500 mm NaCl, 2 mm dithiothreitol. The protein was analyzed by SDS-PAGE. The protein was concentrated with simultaneous buffer exchange using Centriplus-3 (Amicon Bioseparations, Bedford, MA) (3-kDa cut-off). The final IsdI yield was 78 mg. Selenomethionine-labeled IsdI and IsdG proteins were prepared using regular expression strains in M9 medium and the methionine biosynthesis inhibition method (14). The protein expression and final yield were very similar to those of native IsdG and IsdI proteins.

A 2 mm protein stock solution in 10 mm Tris-HCl, pH 7.4, 20 mm NaCl, and 1 mm dithiothreitol were used for crystallization. The best crystals of IsdI were obtained at 295 K using vapor diffusion and hanging droplets in VDX plates (Hampton Research) from sodium cacodylate buffer pH 6.5 and 250 mm NaCl using 25% polyethylene glycol 4000 as a precipitating agent. The IsdI crystals reach the data collection dimensions (0.3 × 0.2 × 0.2) in 2 days. The best crystals of IsdG were obtained at 295 K using vapor diffusion and hanging droplets in VDX plates from 100 mm Bis-Tris HCl buffer, pH 6.5, 200 mm NH₄SO₄ using 30% polyethylene glycol 4000 as a precipitating agent. For cryoprotection of crystals, the polyethylene glycol 4000 concentration was increased to 33%, and the crystals were flash frozen in liquid nitrogen. The IsdG crystals reach the data collection dimensions (0.2 × 0.2 × 0.1 mm) in 3 days.

Data Collection

Diffraction data were collected at ∼100 K at the 19ID beam line of the Structural Biology Center at the Advanced Photon Source, Argonne National Laboratory. For IsdI, the three-wavelength inverse beam multi-wavelength anomalous diffraction (MAD) data set (peak, 12.6603 keV (0.9794 Å); inflection point, 12.6620 keV (0.9793 Å); high energy remote, 13.0000 keV (0.9538 Å)) was collected from a selenomethionine-labeled protein crystal. One crystal (0.3 × 0.2 × 0.2 mm) was used to collect all MAD data to 1.5 Å resolution, with 5 s of exposure/1°/frame using a 200-mm crystal to detector distance. The total oscillation range was 180 degrees as predicted using strategy module of the HKL2000 suite (15). The space group was P2₁ with cell dimensions of a = 45.23 Å, b = 38.01 Å, c = 61.55 Å, α = γ = 90°, β = 93.9°. All of the data were processed and scaled with HKL2000 (Tables I and II). For phase extension and model refinement using CNS, peak data were used.

Table I. Summary of IsdI crystal data.

Unit cell	a = 45.23 Å, b = 38.01 Å, c = 61.55 Å, α = γ = 90°, β = 93.9°
Space group	P21
Molecular mass (residues)	12,420 Da (108)
Mol	2 AU
Selenomethionine	2 AU

Table II. Summary of IsdI MAD data.

	MAD data collection
	Edge	Peak	Remote
Wavelength (Å)	0.9794	0.9793	0.9538
Resolution range (Å)	1.50	1.50	1.50
No. of unique reflections	31,477	31,888	32,197
Completeness (%)	93.2	94.7	95.4
Rmerge (%)	9.3	8.7	6.3

For the IsdG the three-wavelength inverse beam MAD data set (peak,: 12.6603 keV (0.9794 Å); inflection point, 12.6620 keV (0.9793 Å); high energy remote, 13.0000 keV (0.9538 Å)) was collected from a selenomethionine-labeled protein crystal at 100 K. One crystal (0.2 × 0.2 × 0.1 mm) was used to collect all MAD data to 1.9 Å resolution, with 5 s of exposure/1°/frame using a 200-mm crystal to detector distance. The total oscillation range was 180 degrees as predicted using the strategy module of HKL2000. The space group was P2₁ with cell dimensions of a = 51.74 Å, b = 66.78 Å, c = 67.68 Å, α = γ = 90°, β = 105.2°. All of the data was processed and scaled with HKL2000 (Tables III and IV). For phase extension and model refinement using CNS, peak data were used.

Table III. Summary of IsdG crystal data.

Unit cell	a = 51.74 Å, b = 66.78 Å, c = 67.68 Å, α = γ = 90°, β = 105.2°
Space group	P21
Molecular mass (residues)	12,305 Da (107)
Mol	4 AU
Selenomethionine	4 AU

Table IV. Summary of IsdG MAD data.

	MAD data collection
	Edge	Peak	Remote
Wavelength (Å)	0.9794	0.9793	0.9538
Resolution range (Å)	1.9	2.0	1.85
No. of unique reflections	32,894	28,270	36,563
Completeness (%)	93.2	92.8	96.7
Rmerge (%)	7.6	10.2	7.1

Structure Determination and Refinement

The structures of IsdI and IsdG were determined independently by MAD phasing using CNS (16) and refined initially to 1.5 and 1.9 Å, respectively, using CNS against the averaged peak data. The initial models were built manually using QUANTA 2000 (Molecular Simulations Inc., San Diego, CA). The models were further refined against MAD peak data. For IsdI, the final R was 20.3%, and the free R was 24.5% with zero σ cut-off (Tables V and VI). For IsdG, the final R was 23.0%, and the free R was 27.5% with zero σ cut-off (Tables VII and VIII). The stereochemistry of the structures was examined with PROCHECK (17) and the Ramachandran plot. For IsdI, the main chain torsion angles for all residues are in allowed regions, and the five residues are in additional allowed regions. There is one additional residue from the construct visible on the N terminus. For IsdG, the main chain torsion angles for all residues are in allowed regions, five residues are in the additional allowed regions. There are two additional residues from the construct visible at the N terminus.

Table V. IsdI crystallographic statistics.

Resolution range (Å)	Phasing
	Centric		Acentric		All
	FOM	Phasing power	FOM	Phasing power	No.	FOM	Phasing power
50.0–1.5	0.73	2.24	0.60	2.04	28,607	0.61	2.05
Density modification						0.855

Table VI. IsdI refinement statistics.

Resolution range (Å)	30–1.5
No. of reflections	65,616
σ cut-off	0
R value (%)	20.3
Free R value (%)	24.5 (2922)
Bond length (1–2) (Å)	0.008
Angle (°)	1.3
Dihedral (°)	25.0
Improper (°)	0.64
No. of atoms
Protein	1731
Water	235
Mean B-factor (all atoms; Å2)	21.5
Ramachandran plot statistics (%)
Residues in most favored regions	95.97
Residues in additional allowed regions	4.03
Residues in disallowed region	0.0

Table VII. IsdG crystallographic statistics.

Resolution range (Å)	Phasing
	Centric		Acentric		All
	FOM	Phasing Power	FOM	Phasing power	No.	FOM	Phasing power
50.0–1.90	0.71	2.90	0.529	2.22	56,450	0.532	2.23
Density modification						0.895

Table VIII. IsdG refinement statistics.

Resolution range (Å)	30–1.90
No. of reflections	63,987
σ cut-off	0
R value (%)	23.0
Free R value (%)	27.5 (3151)
Root mean square deviations from ideal geometry
Bond length (1–2) (Å)	0.009
Angle (°)	1.3
Dihedral (°)	23.1
Improper (°)	0.73
Protein	3,162
Water	164
Mean B-factor (all atoms; Å2)	27.0
Ramachandran plot statistics (%)
Residues in most favored regions	95.4
Residues in additional allowed regions	4.6
Residues in disallowed region	0.0

Bacterial Strains and Construction of Expression Vectors for Enzymatic Assays

E. coli strain DH5α (F⁻ ara D(lac-proAB) rpsL φ80dlac-ZDM15 hsd R17) was used for DNA manipulation (18), and E. coli strain BL21 (DE3) (F⁻ ompT hsdS_B(r_B⁻m_B⁻) gal dcm (DE3)) was used for the expression of isdG (12). Plasmid constructs containing the isdG coding sequence for expression of wild-type IsdG (pET15BisdG) were described previously (5).

Expression and Purification of IsdG for Biochemical Analysis

E. coli BL21(DE3) strains carrying pET15BisdG were grown overnight at 37 °C in Luria-Bertani broth containing 100 μg/ml ampicilin. The cells were subcultured into fresh medium and grown at 37 °C to mid-log phase. At this time, the expression of the vectors was induced using 1 mm isopropyl-β-d-thiogalactopyranoside. Cell growth was continued for 3 h at 30 °C, and the cells were harvested by centrifugation (10,000 × g for 15 min). The cells were lysed at 14,000 p.s.i. using a French press after suspension in 50 mm Tris-HCl, pH 7.5, 150 mm NaCl containing 100 μm phenylmethylsulfonyl fluoride. The cell suspension was then centrifuged at 100,000 × g for 60 min. After centrifugation, the soluble supernatant was applied to the Ni-NTA column, equilibrated with 50 mm Tris-HCl, pH 7.5, 150 mm NaCl. The column was washed with 20 volumes of 50 mm Tris-HCl, pH 7.5, 150 mm NaCl followed by a second washing with 30 volumes of 50 mm Tris-HCl, pH 7.5, 150 mm NaCl containing 10% glycerol and 10 mm imidazole. The protein was then eluted in 50 mm Tris-HCl, pH 7.5, 150 mm NaCl containing 500 mm imidazole, and the fractions were dialyzed against 50 mm Tris-HCl, pH 7.5, 150 mm NaCl. The purified proteins were stored at −20 °C. The protein yields were similar to those described previously for IsdG and IsdI (5).

Generation of Mutant IsdG Proteins

IsdG proteins containing amino acid substitutions were created using the Pfu mutagenesis technique (Stratagene, La Jolla, CA). Briefly, complimentary oligonucleotides containing the desired alanine mutation were synthesized (IDT Technologies, Coralville, IA; oligonucleotide sequence available upon request) and used in polymerase chain reactions with pET15BisdG as template. PCR conditions followed recommendations by the manufacturer (Stratagene) including 1× reaction buffer, 10 ng of template DNA, 22 nm oligonucleotide primers, 200 μm equivalently mixed dNTPs, and water to a final volume of 50 μl. The cycling parameters included 16 cycles comprised of 30 s at 94 °C, 1 min at 55 °C, and 12 min at 68 °C. Following PCR, 1 μl of the restriction enzyme DpnI was added directly to each amplification reaction to digest methylated DNA, and the reactions were incubated overnight at 37 °C. Following digestion, DNA was transformed into DH5α, and successful transformants were selected on ampicillin agar. Successful incorporation of single amino acid changes were verified by DNA sequencing, and mutant proteins were purified as described above.

Hemin Degradation Assay

All of the absorption spectra were obtained using a Cary 50BIO (Varian, Walnut Creek, CA). Ascorbic acid-dependent degradation of hemin was monitored spectrophotometrically as previously described (7) with the listed changes. IsdG-hemin (10 μm) in 50 mm Tris-HCl, pH 8.0, was incubated at 37 °C with ascorbic acid at a final concentration of 10 mm. The spectral changes between 300 and 800 nm were recorded every 2 min for the duration of the experiment.

Coordinates

The coordinates and the structures factors have been deposited in the Protein Data Bank (accession codes 1SQE and 1XBW).

Results

Crystal Structure of IsdG and IsdI

The high resolution crystal structures of S. aureus IsdG and IsdI in their native forms were determined at 1.9 and 1.5 Å resolution, respectively (Tables I–IV). Crystallographic phases were determined by MAD from selenomethionine-containing enzymes. IsdG and IsdI crystallized as homodimers in the monoclinic space group but in two different crystal packing environments: IsdI with one dimer in asymmetric unit and IsdG with two dimers in asymmetric unit. Both proteins assume a virtually identical ferredoxin-like fold consistent with their 64% sequence identity (78% similarity) (Fig. 1). The IsdI and IsdG structures can be superimposed with root mean square deviation 1.03 Å over 100 residues. The distance matrix alignment analysis placed these two proteins in the α+β sandwich superfold family, defined by the ferredoxin-like α+β sandwich with an anti-parallel β-sheet. Each monomer is assembled into a very tight dimer (Fig. 2) similar to the ActVA-Orf6 monooxygenase from Streptomyces coelicolor (Protein Data Bank entry 1LQ9) (21) (Fig. 3), a protein with little sequence similarity to IsdG and IsdI. The structures can be superimposed with root mean square deviation 1.92 Å over 112 residues. IsdG and IsdI also show structural similarity to several other proteins in the Protein Data Bank. The closest structural homolog is TT1380, a hypothetical protein from Thermus thermophilus (root mean square deviation 1.89 Å, 102 residues; Protein Data Bank entry 1IUJ). Other structural homologs include proteins with ferredoxin-like folds such as the hypothetical protein YjcS from Bacillus subtilis (Protein Data Bank entry 1Q8B). IsdG amino acid numbering is used throughout the text to describe the structures and results of functional analysis.

Fig. 1. The crystal structure of IsdG and IsdI.

Residues fully conserved in all members of the IsdG family are shown in green, and other strongly conserved residues are shown in gray. A and B, IsdG dimer with subunit A is labeled in shades of blue, and subunit B is labeled in shades of orange. In B, the homodimer of IsdG is presented with the 2-fold axis of symmetry perpendicular to the plane of the picture in A. C and D, IsdI dimer with subunit A is labeled in shades of aqua, and subunit B is labeled in shades of pink. In D, the homodimer of IsdI is presented with the 2-fold axis of symmetry perpendicular to the plane of the picture in C. E, IsdG homodimer in the same orientation as B. F, IsdI homodimer in the same orientation as D, both showing predicted heme binding pockets. C (light gray) and N (dark gray) termini are shown along with the predicted binding sites of the heme substrate (red).

Fig. 2. Stereo image of the IsdI dimer.

Both subunits are shown with the secondary structure elements. The N and C termini are labeled.

Fig. 3. Comparative structural analysis of the IsdG and ActVA families of monooxygenases.

A, ActVA-Orf6 homodimer with subunits shaded in blue and red. B, IsdG homodimer with subunits shaded in orange and green. C, IsdI homodimer with subunits shaded in aqua and pink. D, residues predicted to be involved in binding and oxidation of the substrate in ActVA-Orf6 (blue) that are conserved in IsdG (green) and IsdI (pink). E, residues shown to be required for oxidation of heme in IsdG (green) and the equivalent residues in IsdI (pink). ActVA-Orf6 residues that occupy a similar space as the catalytic residues of IsdG are shown in blue.

The β-sheets of two monomers form a 10-strand, anti-parallel β-barrel 17 × 24 Å in size. The barrel is built of two smaller sheets that are connected by long C-terminal strands crossing over from one monomer to the other providing important interactions within the dimer. The core of the barrel is mainly hydrophobic with several residues presented by β-sheets and engaged in van der Waals' contacts (Met¹, Met⁴, Phe³⁹, Leu⁵⁵, Leu⁵⁷, Ile⁹⁷, Ile¹⁰¹, Met¹⁰⁶, and their symmetry mates). Interestingly, the hydrophobic core of the structural homolog ActVA-Orf6 is composed of different side chains that are engaged in different hydrophobic interactions. The sum of these interactions, however, amounts to a highly similar structure, even though ActVA-Orf6 displays only 15% sequence identity and 28% sequence similarity with IsdG (Fig. 4).

Fig. 4. Multiple sequence alignment of IsdG, IsdI, and members of the ActVA family of monooxygenases.

Residues involved in substrate binding and catalysis in the IsdG family are shown in red. Residues involved in substrate binding and catalysis in the ActVA family are shown in blue. Regions of strong sequence similarity are shown in gray.

One side of the IsdG and IsdI β-barrels is capped by strong salt bridges formed by four conserved acidic residues (Glu⁶ and Glu⁵¹ and their symmetry mates) and four basic residues (Arg⁸ and Lys⁵³ and their symmetry mates). This interaction is strengthened by hydrogen bonding of Arg⁸ with Tyr⁹⁹ residues. The other side of the β-barrel is very different in IsdG and IsdI. In IsdG, the β-barrel is capped by a ring of four methionines (Met¹, Met¹⁰⁶, and their symmetry mates) and a salt bridge between Asp³⁶ and Lys¹⁰⁷. These residues are not conserved between IsdI and IsdG. In IsdI, the β-barrel is capped by a rather unusual patch of six Gln residues (Gln³⁵, Gln³⁶, Gln¹⁰⁷, and their symmetry mates). These glutamines are engaged in a very extensive H-bond network. Met¹ provides a hydrophobic patch on this surface of IsdI. Phe⁴⁹ assumes the corresponding role in IsdG. Three tyrosine residues on the β-barrel surface (Tyr²⁴ from monomer A and Tyr¹⁰³ and Tyr¹⁰⁵ in the C-terminal β-strand of monomer B), conserved in both IsdG and IsdI, form stacking interactions and H-bonds between the monomers. The elaborate and extensive dimerization interface of the β-barrels of IsdG and IsdI provides a compact, stable scaffold in both proteins.

Putative Active Site of IsdG and IsdI

Four short α-helices of each subunit decorate the outside of the IsdG and IsdI β-barrel. Two large, symmetry-related cavities are formed between these α-helices and the β-sheets of the β-barrel. All of the α-helices contribute residues to the cavity. The cavity includes the β-sheet residues 2–12 (β1) and residues 71–79 from α-helix 3 (Figs. 1, E and F, and 3D). A similar cavity was shown to nest the ActVA-Orf6 monooxygenase active site (21). In IsdG and IsdI, the surface of this cavity is lined mainly with hydrophobic side chains. The edges of the cavity are hydrophilic with several positively and negatively charged residues resulting in a positively charged surface near the cavities and negatively charged surface further away. It appears that IsdG and IsdI dimers may present two separate cavities for heme binding and oxygenation. Interestingly, there is no obvious site for metal cofactor binding. IsdI and IsdG seem to carry out oxygenation of the heme without the assistance of any of the prosthetic groups or cofactors normally associated with activation of molecular oxygen. However, enzyme cofactors in the form of reductant or reductase are required for the reaction to proceed.

The majority of residues lining the putative active site cavities are similar in IsdG and IsdI; however, conformation of some side chains in the cavity is different (for example Trp⁶⁷ and His⁷⁷), suggesting some flexibility and structural adjustments of the active site upon substrate (heme) binding. In fact, a loop located near the active site is disordered in both structures (residues 81–88 in IsdG and residues 80–87 in IsdI). The majority of the amino acid substitutions between IsdG and IsdI occur on the protein surface, whereas the electrostatic potential near the active site cavities is conserved.

Conserved Residues in the Putative Active Site of IsdG Are Required for Heme Degradation

The dimeric β-barrel type structure of IsdG and IsdI is in stark contrast with the monomeric α-helical folding pattern of the heme oxygenase family, suggesting that they represent a novel family of heme-degrading enzymes. IsdG and IsdI homologs can be identified in numerous Gram-positive pathogens including Bacillus anthracis, the causative agent of anthrax, and Listeria monocytogenes, an important food-borne pathogen (Fig. 5A). Amino acid residues in the putative heme binding cavity, as well as at the interface between the monomers are highly conserved among members of the family (Figs. 1 and 4). The precise location of the active site could not be determined directly from the structures of free enzymes (without bound heme).

Fig. 5. Conserved residues required for IsdG family mediated heme degradation.

A, multiple sequence alignment of the IsdG family of proteins and their secondary structure. Green and gray side chains represent absolutely or strongly conserved residues, respectively. B–I, UV spectral analysis of reactions between heme and ascorbic acid, the electron donor, catalyzed by IsdG mutants. The Soret peak at ∼ 412 nm is indicative of the intact porphyrin ring of heme, and a decrease in absorbance at this wavelength indicates heme degradation. Spectra at time 0 are shown in red, and spectra at 30 min are shown in blue, with intermediate measurements shown in gray. B, wild-type IsdG; C, N7A; D, K17A; E, F23A; F, M38A; G, W67A; H, S70A; I, H77A.

Heme-binding proteins exhibit a characteristic absorption spectrum with the Soret band characteristic for the macrocyclic conjunction at ∼410 nm. IsdG-mediated heme degradation in the presence of an electron donor (e.g. ascorbic acid) can be monitored spectophotometrically as a decrease in the intensity of the Soret band at 412 nm (Fig. 5B). This assay allowed identification of IsdG mutants with decreased catalytic activity. Alanine substitution mutagenesis was performed for residues conserved in all members of the IsdG family (Fig. 5A) and/or predicted to be associated with the heme binding cavity (Fig. 1). Alanine substitution of Asn⁷, Trp⁶⁷, and His⁷⁷ abolished IsdG-mediated heme degradation (Fig. 5, B–I), demonstrating a strict requirement for these residues for the catalytic activity of IsdG. A second class of mutation was identified, exemplified by alanine substitution of Met³⁸, that caused only slight reduction of heme degradation but shifted the peak absorbance of the Soret band to 430 nm, indicating changes in interaction between IsdG and heme. As a control, substitution of Ser⁷⁰, Phe²³, and Lys¹⁷, residues, predicted to be located outside of the catalytic pocket of IsdG, did not significantly affect heme binding or degradation (Fig. 5, B–I). These results corroborated our prediction that the heme binding cavity of IsdG is configured by the side chain residues of Asn⁷, Met³⁸, Trp⁶⁷, and His⁷⁷ (Fig. 6).

Fig. 6. The putative heme binding cavity of IsdG and IsdI.

A, IsdG and B, IsdI. Heme (red) is shown in the model proposed based on the enzymatic data of the IsdG mutants. Carbon (gray), oxygen (purple), nitrogen (blue), and sulfur (yellow) atoms are shown. The IsdG residues Asn⁷, Trp⁶⁷, and His⁷⁷ are required for heme degradation, and Met³⁸ is required for maintaining the integrity of heme binding.

Discussion

Based on the structure (Fig. 3), amino acid sequence comparisons (Figs. 4 and 5A), and molecular modeling (vide infra), we propose a three-dimensional model of heme binding to IsdG (Fig. 6). Heme-binding proteins often employ an electron bond between the imidazole moiety of a conserved histidine residue and the iron cation of heme (19). In this model, His⁷⁷, required for enzymatic activity, serves as the heme axial ligand (Fig. 6), and the polar Asn⁷ residue interacts with the hexacoordinate iron ion of the heme, on the opposite side of His⁷⁷, stabilizing the reaction intermediate. Alternatively, hydrogen bonding of Asn⁷ with the heme iron-bound O₂ facilitates self-hydroxylation of the porphyrin ring through a steric bending of the dioxygen, a mechanism recently proposed for the heme oxygenase family member HmuO (20). A possible role of Trp⁶⁷, required for IsdG catalysis, is to properly position the heme tetrapyrrole moiety with respect to His⁷⁷ and Asn⁷. Alternatively, Trp⁶⁷ could stabilize the reaction intermediate of heme oxygenation through direct hydrogen bonding to either a ring nitrogen of the tetrapyrrole or through a H₂O intermediate. In the orientation shown in Fig. 6, the hydrophobic methyl and vinyl groups of the heme are stabilized through interaction with the hydrophobic portion of its binding cavity comprised of residues Ile⁵⁴, Met³⁸, Phe⁶⁴, and Trp⁶⁷. In this orientation, the two propionate groups of heme are free to interact with the polar residue Thr²⁵ and the positively charged residues Arg²² and His²⁷ (Fig. 6). This suggestion is supported by changes in heme binding caused by substitution of Met³⁸.

Pfam analysis assigns IsdG and IsdI to the ActVA family of monooxygenases responsible for the oxidation of aromatic polyketides (5). This enzyme family includes the ActVA-ORF6 from Streptomyces coelicolor (21), ElmH from Streptomyces olivaceus (22), and TcmH from Streptomyces glaucescens (23). The substrates of TcmH/ElmH (a tetracenomycin precursor) (22, 23), ActVA-Orf6 (6-deoxydihydrokalafungin (6-DDHK)) (21), and IsdG/IsdI (heme) exhibit similar five- or six-member aromatic ring architecture (Fig. 7). In the absence of any significant amino acid conservation (Fig. 4), the structural conservation of ActVA-Orf6 (21), IsdG, and IsdI (Fig. 3) is striking. These representatives of both monooxygenases subfamilies crystallized as homodimers with two substrate binding cavities. Trp⁶⁶ of ActVA-Orf6, Trp⁵⁸ of IsdG, and Trp⁵⁷ of IsdI occupy a structurally equivalent position in their respective substrate-binding sites. This residue is present in all members of the IsdG and ActVA family of monooxygenases. The Trp⁶⁶ residue was proposed to stabilize ActVA-Orf6–6-DDHK complex through H-bonding (21). Additionally, the carboxyl group of 6-DDHK extends from the active site pocket of ActVA-Orf6 and is exposed to solvent. This finding is consistent with our model of heme binding shown in Fig. 6.

Fig. 7. The substrates of the ActVA and IsdG family of enzymes.

A, tetra-cenomycin precursor. B, 6-DDHK. C, protoporphyrin IX (heme).

Differences in the substrate-binding pockets between the two subfamilies of monooxygenases can be explained by divergent substrate. Structurally equivalent asparagines Asn⁷ in IsdG, Asn⁶ in IsdI, and Asn⁶² in ActVA-Orf6 (21) (Fig. 3E) are positioned such that they would clash with 6-DDHK (ActVA-Orf6 substrate) binding to IsdG and IsdI. Consequently, we predict that the IsdG substrate-binding site is located at a greater distance from the β-sheet component of the α + β sandwich fold, when compared with the substrate-binding site of ActVA-Orf6. Furthermore, Tyr⁷² of ActVA-Orf6 is replaced by phenylalanine in IsdG (Phe⁶⁴) and IsdI (Phe⁶³) (Fig. 3D). The smaller size of phenylalanine may be necessary to accommodate heme, a larger ligand. The position of the catalytic histidines of IsdG (His⁷⁷) and IsdI (His⁷⁶) is occupied by Trp³⁹ in ActVA-Orf6 (Fig. 3E), further explaining the different substrate specificity. Histidine is a common heme axial ligand in many hemoproteins (19).

We propose, based on earlier models for ActVA-Orf6 (21) and heme oxygenase-mediated catalysis (20), that hydroxylation of the heme porphyrin ring by IsdG and IsdI occurs via substrate self-hydroxylation. We predict that binding of heme in the cavity shown in Fig. 6 is followed by its reduction from the ferric to the ferrous form, thus allowing Fe^II-O₂ binding. A subsequent electron transfer to heme may form an activated peroxide intermediate capable of hydroxylating the porphyrin ring.

It seems plausible, that TT1380/1IUJ, a structural homolog of IsdG and IsdI from T. thermophilus, functions as a heme monooxygenase, suggesting that this bacterium is capable of degrading heme. The amino acid sequence of TT1380 is 27% identical to IsdG and 33% identical to IsdI and contains key residues of the IsdG/IsdI active site (Asn⁷, Trp⁶⁷, and His⁷⁷).

The abundance of heme-iron provides a rich source of iron for bacterial pathogens of vertebrates. The convergent evolution of the heme oxygenase family and the IsdG heme monooxygenase family provides an opportunity to compare two distinct enzyme families involved in similar processes. Divergent evolution of the monooxygenase substrate specificity under selective pressure to acquire iron from the host seems to have lead to a new heme-degrading enzyme family.