Ruffling is Essential for Staphylococcus aureus IsdG-catalyzed Degradation of Heme to Staphylobilin

Abstract

Non-canonical heme oxygenases are enzymes that degrade heme to non-biliverdin products within bacterial heme iron acquisition pathways. These enzymes all contain a conserved second-sphere Trp residue that is essential for enzymatic turnover. Here, UV/Vis absorption (Abs) and circular dichroism (CD) spectroscopies were employed to show that the W67F variant of IsdG perturbs the heme substrate conformation. In general, a dynamic equilibrium between “planar” and “ruffled” substrate conformations exists within non-canonical heme oxygenases, and that the second-sphere Trp favors population of the “ruffled” substrate conformation. ¹H nuclear magnetic resonance and magnetic CD spectroscopies were used to characterize the electronic structures of IsdG and IsdI variants with different substrate conformational distributions. These data revealed that the “ruffled” substrate conformation promotes partial porphyrin-to-iron electron transfer, which makes the meso carbons of the porphyrin ring susceptible to radical attack. Finally, UV/Vis Abs spectroscopy was utilized to quantify the enzymatic rates, and electrospray ionization mass spectrometry was used to identify the product distributions, for variants of IsdG with altered substrate conformational distributions. In general, the rate of heme oxygenation by non-canonical heme oxygenases depends upon the population of the “ruffled” substrate conformation. Also, the production of staphylobilin or mycobilin by these enzymes is correlated with the population of the “ruffled” substrate conformation, since variants that favor population of the “planar” substrate conformation yield significant amounts of biliverdin. These data can be understood within the framework of a concerted rearrangement mechanism for the monooxygenation of heme to meso-hydroxyheme by non-canonical heme oxygenases.

Graphical Abstract

Synopsis: Staphylococcus aureus IsdG has access to two substrate conformations, “planar” and “ruffled”, which differ based upon the magnitude of their b_1u out-of-plane deformation from idealized D_4h symmetry. This article correlates staphylobilin production with population of the “ruffled” conformation, biliverdin production with the “planar” conformation, and provides an explanatory mechanistic model.

1. Introduction

Non-canonical heme oxygenases exhibit remarkable relationships between substrate structure, electronic structure, and enzyme function. When the X-ray crystal structure of cyanide-inhibited Staphylococcus aureus IsdI (IsdI–heme–CN) was reported in 2011 [1], it represented the most ruffled protein-bound heme reported. It was later shown that this ruffling, a b_1u out-of-plane distortion from idealized D_4h symmetry [2], depended upon a second-sphere interaction between Trp66 and the heme substrate [3]. A few years later, X-ray crystallography revealed that a second non-canonical heme oxygenase, Mycobacterium tuberculosis MhuD, also induces a significant ruffling deformation of its heme substrate, albeit to a lesser extent [4]. The heme substrate ruffling by MhuD also depends upon a steric interaction with a second-sphere Trp residue, Trp66 [5]. Furthermore, detailed spectroscopic and computational studies of several MhuD variants demonstrated that the heme distortion detected by X-ray crystallography is the weighted average of a dynamic equilibrium between two distinct heme substrate conformations: “planar” and “ruffled” [6]. Most recently, spectroscopically-validated hybrid quantum mechanics/molecular mechanics (QM/MM) calculations suggested that S. aureus IsdG, a third non-canonical heme oxygenase, induces an intermediate degree of heme ruffling compared to IsdI and MhuD (Fig. 1) [7]. The observation of two distinct heme conformations within the MhuD active site is fairly unique within the heme protein literature, and the possibility that IsdG and IsdI also stabilize two distinct heme substrate conformations had not been evaluated prior to this work.

Fig. 1.

WT IsdG induces an out-of-plane ruffling distortion of its heme substrate based upon spectroscopically-validated QM/MM calculations (top) [7]. The W67F substitution was hypothesized to decrease heme ruffling by reducing steric interactions between residue 67 and the substrate (middle). This heme structural change would be expected to change the ground state electronic configuration from ²B_2g to ²E_g by decreasing mixing of the Fe 3d_xy and porphyrin a_2u orbitals (bottom) resulting in altered heme oxygenation activity [5, 26].

The highly ruffled heme substrate of non-canonical heme oxygenases is responsible for unique electronic structures that promote novel biochemical reactivity. Characterization of IsdI–heme–CN with ¹H nuclear magnetic resonance (NMR) spectroscopy revealed that this species has an unusual ²B_2g electronic ground state with an Fe (3d_xz,yz)⁴(3d_xy)¹ electron configuration [1], which had previously only been observed for highly ruffled model complexes [8]. Preparation and ¹H NMR characterization of W66Y IsdI–heme–CN, which houses a less ruffled heme substrate, clarified that the unusual ²B_2g electronic ground state of WT IsdI–heme–CN is derived from the large ruffling deformation of heme [3]. Subsequent characterization of cyanide-inhibited MhuD (MhuD–heme–CN) using magnetic circular dichroism (MCD) spectroscopy demonstrated that this species also has the unusual ²B_2g electronic ground state, and hinted at underlying complexity [4]. Indeed, spectroscopic characterization of additional MhuD variants, coupled with multi-reference perturbation theory calculations, revealed that MhuD–heme–CN has a ²B_2g electronic ground state corresponding to the “ruffled” conformation with a low-lying ²E_g excited state associated with the “planar” substrate conformation [6]. The ²E_g electronic state has an Fe (3d_xy)²(3dxz,yz)³ electron configuration, and is the most typical electronic ground state for low-spin ferric heme [9]. Spectroscopic characterization of cyanide-inhibited IsdG (IsdG–heme–CN) has also provided evidence for a thermal mixture of ²B_2g and ²E_g electronic states (Fig. 1) [10, 11]. These electronic structure changes assume additional significance in light of the fact that the electronic structure change induced by the “ruffled” conformation, partial porphyrin to iron electron transfer, promotes porphyrin hydroxylation [12].

The unusual electronic structures of MhuD–heme–CN and IsdG–heme–CN do merit further comment. To state matters concisely, this vibronic structure is a manifestation of a hidden pseudo Jahn-Teller effect [13, 14]. The Jahn-Teller theorem states that “all non-linear nuclear configurations are therefore unstable for an orbitally degenerate electronic state”, which can be proven using group theory [15]. The ²E_g and ²B_2g electronic states described above are not orbitally-degenerate, but they are near degenerate and hence subject to a pseudo Jahn-Teller distortion. In this case, the structure distortion to alleviate the near degeneracy of the ²E_g and ²B_2g electronic states is ruffling, which is a b_1u distortion from idealized D_4h symmetry. The b_1u distortion couples the ²B_2g and ²A_2u excited states to the point of driving the energy of the ²B_2g state below the ²E_g state for large ruffling distortions. This rather complex scenario is termed a hidden pseudo Jahn-Teller effect because the distortion relieves near degeneracy by stabilizing an electronic excited state. Heme ruffling is an out-of-plane vibration with a frequency of 40–60 cm⁻¹ [16, 17], and thermal energy (220 cm⁻¹ at 37 °C) is more than sufficient to trigger a dynamic equilibrium between a “planar” substrate conformation with a ²E_g ground state and a “ruffled” conformation with a ²B_2g ground state at physiologically-relevant temperatures.

Perhaps not surprisingly, the unusual geometric and electronic structures of the heme substrate in non-canonical heme oxygenases have often been invoked to explain the novel mechanisms of these enzymes. Prior to the characterizations of the geometric and electronic structures of S. aureus IsdG and IsdI described above, it was discovered that these enzymes degrade heme to novel staphylobilin products that are distinct from the biliverdin products of canonical heme oxygenases [18]. IsdG- and IsdI-catalyzed heme oxygenation can use either ascorbate or IruO as an external reductant, and both reactions yield identical staphylobilin products [19]. Subsequent studies identified formaldehyde as the C1 product [20], and identified two reaction intermediates: meso-hydroxyheme and formyloxobilin [21]. A concerted rearrangement mechanism has been proposed for the initial monooxygenation of heme to meso-hydroxyheme [10, 22], but the remaining mechanistic steps from meso-hydroxyheme to staphylobilin remain unknown with minimal chemical precedent to inform reasonable hypotheses. It turns out that MhuD degrades heme to a third product, mycobilin [23], via sequential monooxygenation and dioxygenation steps [24]. Recently, it has been reported that both the R26S and W66F second-sphere substitutions can change the primary MhuD product from mycobilin to biliverdin [25, 26]. These observations have been attributed to a mechanism where the structural and electronic dynamics of heme-bound MhuD are critical: MhuD converts “ruffled” heme to meso-hydroxyheme and “planar” meso-hydroxyheme to mycobilin. The proposed monooxygenation mechanism is consistent with a recent QM/MM study of MhuD-catalyzed heme degradation [27]. In this work, we sought to determine whether the structural and electronic dynamics observed for MhuD are also present in IsdG and IsdI, as well as whether they can be exploited to alter the enzymatic product of IsdG.

This article elucidates the influence of the Trp67 residue on the structure, electronic structure, and mechanism of S. aureus IsdG. The W67F variant was designed with the goal of isolating the contribution of the second-sphere Trp residue to the structure and electronic structure of the IsdG-bound heme substrate. In order to separate the contributions of the amino acid substitution to the substrate and protein structures, circular dichroism (CD) spectroscopy was used to compare the secondary structures of WT and W67F IsdG. A recently established UV/Vis absorption (Abs) spectroscopy-based assay was employed to quantify heme ruffling instead of the more laborious crystallography-based approach previously utilized for IsdI [1, 3, 5]. Ruffling-induced heme electronic structure changes have been invoked as a key component in non-canonical heme oxygenase mechanisms [10, 20, 22, 24, 26, 27], so a careful assessment of the heme electronic structure of these variants using ¹H NMR and MCD spectroscopies was carried out. Finally, the rates of heme oxygenation by WT and W67F IsdG were quantified using a UV/Vis Abs-based assay and the products of these enzyme variants were identified using electrospray ionization mass spectrometry (ESI-MS) [21]. The implications of these data in terms of a general model for the role of a conserved second-sphere Trp in non-canonical heme oxygenases will be discussed.

2. Experimental

Unless noted otherwise, all materials in this work were purchased from Fisher Scientific and used without further purification.

2.1. Protein expression and purification

Single amino acid substitutions were introduced into genes encoding S. aureus IsdG and IsdI using site-directed mutagenesis. The preparations of a pET15b (Amp^r, Novagen) vector encoding a short-linker construct of WT IsdG [28], a pET15b (Amp^r, Novagen) vector encoding WT IsdI [6], and a PRK793 (Amp^r) vector encoding S219V tobacco etch virus (TEV) protease have been previously described [29]. The W67F mutation was introduced into the recombinant isdG gene, and the W66Y mutation was introduced into the isdI gene, using the QuikChange Lightning site-directed mutagenesis kit (Agilent). DNA primers for the isdG mutagenesis reaction were designed according to the QuikChange manual and obtained from Midland Certified Reagent Company (Table S1). DNA primers for isdI mutagenesis were obtained from Integrated DNA Technologies (Table S2). DNA sequencing performed by the University of Vermont DNA analysis facility confirmed successful preparations of W67F IsdG and W66Y IsdI (Tables S3–S4).

WT IsdG, W67F IsdG, and W66Y IsdI were expressed as previously reported for short-linker WT IsdG [7]. WT IsdG and W67F IsdG were purified using nickel affinity chromatography as previously described for short-linker WT IsdG, but some minor modifications to the chromatography procedure were required to obtain pure W66Y IsdI. Specifically, the nickel-charged HiTrap Chelating HP column (GE Healthcare) equilibrated with 50 mM Tris pH 7.4, 150 mM NaCl was washed with a linear gradient of imidazole from 0 to 150 mM prior to elution of W66Y IsdI with 300 mM imidazole using an Äkta Pure fast protein liquid chromatography (FPLC) system (GE Healthcare). The N-terminal His₆-tags were removed as previously described for WT IsdI [6, 30]. W67F IsdG and W66Y IsdI were obtained in >99% and >95% purity, respectively, as determined by SDS-PAGE gel electrophoresis (Figs. S1–S2). Apoprotein concentrations were determined by Bradford assay with bovine serum albumin (Pierce) as the standard.

The secondary structures of substrate-free WT and W67F IsdG were assessed using CD spectroscopy. The CD spectrum of WT IsdG was reported previously [7]. W67F IsdG was exchanged into 10 mM potassium phosphate (KP_i) pH 7.4 using a PD-10 desalting column and loaded into a 2 mm path length quartz cuvette (Starna). CD data were acquired from 260 to 190 nm with a 20 nm/min scan speed, a 1.0 nm bandwidth, an 8s digital integration time, and a 0.5 nm data pitch using a Jasco J-1700 spectrophotometer. The CD spectra were analyzed using the Dichroweb program SELCON3 with reference dataset 4 [31–35].

2.2. Spectroscopic characterization

The heme-bound and cyanide-inhibited forms of IsdG and IsdI were characterized with UV/Vis Abs spectroscopy as described previously [1, 36]. Briefly, heme-bound forms of IsdG (IsdG–heme) and IsdI (IsdI–heme) were prepared in 50 mM Tris pH 7.4, 150 mM NaCl as described previously for WT IsdG–heme [7, 36]. The extinction coefficients of W67F IsdG–heme and W66Y IsdI–heme were determined to be ε₄₁₂ = 117.1 mM⁻¹cm⁻¹ and ε₄₀₁ = 92.4 mM⁻¹ cm⁻¹, respectively, by employing the pyridine hemochrome assay [37]. In order to determine the extinction coefficients of IsdG–heme–CN and IsdI–heme–CN, a single crystal of KCN was added to a heme-bound enzyme sample of known concentration as described previously for IsdG–heme–CN [10]. The extinction coefficients of WT IsdG–heme–CN, W67F IsdG–heme–CN, and W66Y IsdI–heme–CN were determined to be ε₄₂₀ = 94.9 mM⁻¹cm⁻¹, ε₄₁₈ = 95.8 mM⁻¹ cm⁻¹, and ε₄₁₆ = 94.4 mM⁻¹ cm⁻¹, respectively. UV/Vis Abs data were acquired from 700 to 300 nm on a Cary 100 spectrophotometer with a scan rate of 600 nm/min, 2 nm bandwidth, 0.1 s digital integration time, and 1 nm data interval.

The ground state electronic structure of IsdG–heme–CN was characterized using ¹H NMR spectroscopy in a similar manner to that previously described for IsdI–heme–CN [1]. WT and W67F IsdG–heme–CN were prepared as described above for UV/Vis Abs characterization, exchanged into sodium phosphate (NaP_i) pH 7.4, concentrated using an Amicon stirred cell (Millipore), and diluted with D₂O (Cambrdige Isotope Laboratories). This procedure yielded 0.84 and 1.58 mM samples of WT and W67F IsdG–heme–CN, respectively, in 10 mM NaP_i pH 7.4, 10% D₂O (v/v). ¹H NMR data was acquired on a Varian Unity Inova 500 MHz spectrometer equipped with an inverse triple-resonance probe. 256 scans were acquired with a 1.5 s relaxation delay, presaturation of the ¹H₂O resonance during this delay, a 1 s acquisition time, and a 30 kHz sweep width at 25 °C. The ¹H NMR spectra were indirectly referenced to sodium 2,2-dimethyl-2-silapentane-5-sulfonate (DSS, Cambridge Isotope Laboratories) via the residual ¹H₂O resonance and processed using the ACD/Labs NMR processor with 10 Hz exponential line broadening and zero filling to 32786 points.

The excited state electronic structures of IsdG–heme–CN and IsdI–heme–CN were interrogated using MCD spectroscopy. Samples of W67F IsdG–heme–CN and W66Y IsdI–heme–CN in 50 mM KP_i pH 7.4, 60 % glycerol (v/v) were prepared as described previously for IsdI–heme–CN [6]. MCD spectral data were acquired using a home-built setup consisting of either a Jasco J-815 (IsdG–heme–CN) or Jasco J-1700 (IsdI–heme–CN) spectrometer in conjunction with an Oxford SM4000-8T Spectromag. MCD data was collected with a scanning speed of 200 nm/min, a bandwidth of 1 nm, a digital integration time of 0.25 s, and a data pitch of 0.5 nm. Data was acquired for temperatures ranging from 2 to 10 K and magnetic fields ranging from −7 to +7 T. In addition, variable-temperature, variable-field (VTVH) saturation magnetization curves were measured ranging from 2 to 20 K with a magnetic field ramp of 0.7 T/min between ±7 T for the 424 and 422 nm bands of W67F IsdG–heme–CN and W66Y IsdI–heme–CN, respectively. For all MCD data, the CD contribution was eliminated by subtracting the negative field data from the positive field data and dividing by 2.

2.3. Enzyme activity assays

The enzymatic activities of WT and W67F IsdG were assayed in a manner similar to that previously described for WT enzyme [18, 36]. Briefly, WT and W67F IsdG–heme were prepared as described above for UV/Vis Abs spectroscopic characterization. 10 μM IsdG–heme was mixed with 10 mM sodium ascorbate, 200 units of bovine liver catalase (Sigma Aldrich), and 1 mM ethylenediaminetetraacetic acid in 50 mM KP_i pH 7.4. Ascorbate was employed as an external reductant instead of IruO because this redox partner will not interfere with the spectral characterizations employed in this work. UV/Vis Abs spectra of the reaction mixture were acquired every 5 min over a course of 60 min on a Cary 100 spectrophotometer using the instrumental parameters listed above. An additional assay was completed for W67F IsdG with 30 μM IsdG–heme, 1 mM sodium ascorbate, 200 units of bovine liver catalase (Sigma Aldrich), and 1 mM ethylenediaminetetraacetic acid in 50 mM KP_i pH 7.4. This additional assay was monitored by UV/Vis Abs spectroscopy for 120 minutes using the parameters noted above.

The products of WT and W67F IsdG-catalyzed heme degradation were identified using in proteo mass spectrometery (MS) according to a procedure similar to that previously described for WT enzyme [21]. After 60 min, the IsdG–heme reaction mixtures described above were loaded onto a C₁₈ column equilibrated with 2% acetonitrile (MeCN) with 0.1% formic acid (v/v) in water (v/v) using a Shimadzu Prominence high-performance liquid chromatography system. Tetrapyrroles were eluted with a linear gradient from 2% to 98% MeCN with 0.1% formic acid (v/v) in water (v/v) over 50 min. ESI-MS data for m/z 600–1500 was acquired in positive ion mode using an ABI Sciex 4000 QTrap Pro hybrid triple-quadrupole/linear ion trap.

3. Results

3.1. The W67F substitution does not perturb the secondary structure of IsdG

Analyses of X-ray crystal structure and UV CD data validated usage of the SELCON3 algorithm to extract secondary structure information from the CD spectra of IsdG variants. UV CD spectra of proteins contain valuable information regarding secondary structure since this region of the spectrum is dominated by electronic transitions within the amide backbone groups, but a suitable analysis algorithm is needed to extract this information from the raw data [31]. In general, these analyses rely upon reference datasets for spectra with known secondary structure, and it is necessary to benchmark a UV CD analysis method and reference set for a protein with a similar fold in order to obtain accurate results. Here, substrate-free WT IsdG was used to benchmark the UV CD analysis because an X-ray crystal structure of this species is available (PDB ID 1XBW) [38]. The X-ray crystal structure of WT IsdG was analyzed using the standard DSSP algorithm [39, 40], resulting in an α-helix content of 24% and a β-sheet content of 41% (Table S5). The UV CD spectrum of WT IsdG is consistent with the mixed α/β secondary structure since the minima at 225 and 210 nm associated with α-helices can be observed as a trough and shoulder, respectively, superimposed on a broad negatively-signed feature associated with a β-sheet (Fig. 2). The SELCON3 algorithm with reference dataset 4 was previously used to successfully extract the secondary structure of MhuD [5, 33–35], an enzyme with a similar fold to IsdG [41], so that approach was tested here. Based upon SELCON3 analysis of the UV CD spectrum, WT IsdG has 20 ± 7% α-helix and 41 ± 7% β-sheet content in aqueous solution. The differences between the X-ray crystallography and UV CD-derived secondary structure of WT IsdG are within the reported error for the algorithm and reference dataset. Thus, the combination of SELCON3 and reference dataset 4 is a reliable tool to extract secondary structure information from UV CD spectra of IsdG variants.

Fig. 2.

The W67F substitution does not significantly alter the secondary structure of IsdG. UV CD spectra were acquired in 10 mM KP_i pH 7.4 and analyzed using the SELCON3 algorithm [7, 33–35].

Analysis of the UV CD data for W67F IsdG using this validated approach does not yield evidence for a secondary structure induced by this amino acid substitution. The UV CD spectrum of W67F IsdG is also consistent with a mixed α/β secondary structure (Fig. 2). SELCON3 analysis of these data yields an estimated secondary structure of 20 ± 7% α-helix and 41 ± 7% β-sheet (Table S5) [33–35]. Compared to WT enzyme, this is a 1% decrease in α-helical content and 6% increase in β-sheet content. However, it is important to note that both of these changes are within algorithmic error. Thus, the W67F substitution induces a minimal secondary structural change. Furthermore, considering that IsdG is a 12.5 kDa single-domain protein [36, 38], the W67F substitution likely does not significantly alter the protein tertiary structure either. Consequently, structural, electronic, and functional changes induced by the W67F substitution will be interpreted to arise from altered second-sphere interactions between IsdG and the heme substrate.

3.2. The W67F substitution decreases heme ruffling induced by IsdG

UV/Vis Abs spectroscopy was employed to identify the distal ligand to heme in W67F IsdG–heme. The UV/Vis Abs spectra of water- and hydroxide-ligated IsdG–heme have been reported in the literature. Water-bound IsdG–heme is a high-spin species with an intense Soret band at 403 nm, a moderately intense Q-band at 541 nm, and a shoulder at 637 nm [10]. On the other hand, hydroxide-ligated IsdG–heme is a low-spin species with a Soret band at 412 nm and a vibronically-split Q band. At pH 7.4, the UV/Vis Abs spectra of WT and W67F IsdG–heme exhibit Soret bands at 411 and 412 nm, respectively, with vibronically-split Q bands (Fig. 3). These data are within experimental error of one another, and consistent with hydroxide-ligated forms of the heme substrate being predominant at pH 7.4. Thus, the W67F substitution does not alter the protonation state of the distal ligand to iron.

Fig. 3.

UV/Vis Abs spectra of WT IsdG–heme (panel A), W67F IsdG–heme (panel B), WT IsdG–heme–CN (panel C), and W67F IsdG–heme–CN (panel D). IsdG–heme data was acquired in 50 mM KP_i pH 7.4, and IsdG–heme–CN data was acquired in 50 mM Tris pH 7.4, 150 mM NaCl. At pH 7.4, both WT and W67F IsdG–heme are low-spin complexes with distal hydroxide ligands [10]. The W67F substitution induces a 100 cm⁻¹ blue-shift of the heme Soret band, which can be attributed to decreased heme substrate ruffling [5].

UV/Vis Abs spectroscopy was also employed to assess the influence of the W67F substitution on heme substrate ruffling by IsdG. In order to evaluate heme ruffling within the W67F IsdG active site, an established UV/Vis Abs-based assay was used to estimate heme ruffling changes arising from second-sphere substitutions for cyanide-bound heme [5]. An X-ray crystal structure for WT IsdG–heme–CN is not available, but heme ruffling in WT IsdG–heme–CN has been estimated to be 2.0 Å based upon spectroscopically-validated QM/MM calculations [7]. The UV/Vis Abs spectra of WT and W67F IsdG–heme–CN are quite similar (Fig. 3), but closer inspection reveals that the W67F substitution blue-shifts both the Q and Soret bands by 100 cm⁻¹. Based upon the QM/MM structure of WT IsdG–heme–CN, the UV/Vis Abs spectra of WT and W67F IsdG–heme–CN, and the established ruffling model noted above, the magnitude of heme ruffling in W67F IsdG–heme–CN was estimated to be 1.7 Å. A similar decrease in heme ruffling was previously reported for IsdI–heme–CN upon introduction of the W66Y substitution based upon X-ray crystallography (PDB IDs 3QGP and 4FNI) [1, 3], but it is important to note that IsdI induces more heme ruffling than IsdG. Altogether, these data reveal that the W67F substitution decreases heme ruffling by IsdG.

3.3. The W67F substitution perturbs the heme electronic structure

The electronic structure of IsdG–heme–CN was investigated using ¹H NMR spectroscopy. ¹H NMR has been used successfully in the past to characterize the unusual electronic structures of IsdI–heme–CN and MhuD–heme–CN [1, 4], and it was expected to provide similar insight into IsdG–heme–CN. The ¹H NMR spectrum of WT IsdG–heme–CN contains several resonances outside the typical 0–10 ppm region (Fig. 4), which arise from hyperfine coupling between nuclear and electron spins [9]. Typically, several far downfield resonances arising from the heme methyl groups are observed in low-spin ferric hemes, but no ¹H resonances larger than 15 ppm were observed for WT IsdG–heme–CN. Instead, several upfield hyperfine-shifted resonances are observed that have been assigned to the heme meso groups in low-spin, ferric heme species with ²B_2g electronic ground states. Resonance assignments for IsdG–heme–CN were not made here, since the IsdG–heme–CN electronic structure will be further characterized using VTVH MCD below, but it is noteworthy that this ¹H NMR spectrum of IsdG–heme–CN is similar to that reported previously for IsdI–heme–CN. Based upon the similarities between the IsdG–heme–CN and IsdI–heme–CN ¹H NMR spectra, the IsdG–heme–CN ground state is tentatively assigned to be ²B_2g, but further investigation using VTVH MCD is warranted.

Fig. 4.

¹H NMR spectra of WT IsdG–heme–CN (top) and W67F IsdG–heme–CN (bottom). ¹H NMR data was acquired in 10 mM NaP_i pH 7.4, 10% D₂O (v/v) at 25 °C. Downfield hyperfine-shifted resonances for low-spin ferric heme have been associated with a ²E_g electronic state whereas upfield hyperfine-shifted resonances have been associated with a ²B_2g electronic state [9]. Thus, The W67F substitution increases the population of a ²E_g electronic state.

Next, the influence of the W67F substitution on the IsdG–heme–CN electronic structure was evaluated using ¹H NMR. Once again, the observation of hyperfine-shifted ¹H NMR resonances is clear evidence for a paramagnetic species (Fig. 4). Compared to WT enzyme, several additional downfield hyperfine-shifted resonances are observed in the ¹H NMR spectrum of W67F IsdG–heme–CN. The furthest downfield-shifted resonance is located at 17.8 ppm, which is similar to what was observed previously for W66Y IsdI–heme–CN [3]. On the upfield side of the W67F IsdG–heme–CN ¹H NMR spectrum, the resonances observed at −15.0, −10.4, and −8.5 ppm for WT enzyme are no longer seen in the variant. Based upon a comparison of the WT IsdG–heme–CN and IsdI–heme–CN ¹H NMR spectra [1], these three upfield resonances likely arise from the heme meso groups of a low-spin, ferric heme in the ²B_2g electronic state. The failure to observe these ¹H resonances in W67F IsdG–heme–CN suggests that population of a ²B_2g electronic state at room temperature is decreased in the W67F variant. W67F IsdG–heme–CN is most likely a thermal mixture of a ²E_g electronic ground state with a thermally-accessible ²B_2g excited state because the downfield-shifted ¹H resonances have significantly smaller chemical shifts than those reported for cyanide-ligated myoglobin [42], which has a pure ²E_g ground state. However, line broadening arising from chemical exchange and paramagnetic relaxation can make ¹H NMR resonances difficult to detect, so MCD spectroscopy was employed to further investigate the electronic structure of IsdG–heme–CN.

3.4. W67F IsdG–heme–CN has a ²B_2g electronic ground state

MCD spectroscopy was employed as a complementary approach to assess the influence of the W67F substitution on the electronic structure of IsdG–heme–CN. The 5 K, 7 T MCD spectrum of W67F IsdG–heme–CN was acquired and compared to published data for WT enzyme (Fig. 5) [10]. Several changes are apparent in the 13,000–21,000 cm⁻¹ region, similar to those previously associated with an electronic state change from ²B_2g to ²E_g upon introduction of the W66F substitution into MhuD–heme–CN [5]. In addition, a 100 cm⁻¹ blue-shift from 23,500 to 23,600 cm⁻¹ of the intense, negatively-signed component of the Soret band was observed upon introduction of the W67F substitution. This spectral shift is consistent with what was observed using UV/Vis Abs spectroscopy (Fig. 3), and is further evidence that the heme substrate of W67F IsdG is less ruffled than the heme substrate of WT IsdG. Finally, the MCD-detected Soret band of W67F IsdG–heme–CN is four times more intense than that reported for WT enzyme. This intensity change has previously been attributed to increased population of a ²E_g state in the W66F variant of MhuD–heme–CN. The additional MCD intensity arises from enhanced spin-orbit coupling between the two components of a ²E_g state [4]. Thus, both MCD and UV/Vis Abs data are consistent with a less ruffled heme, and MCD and ¹H NMR data are consistent with increased population of a ²E_g state, for W67F IsdG compared to WT enzyme.

Fig. 5.

5 K, 7 T MCD spectra of WT and W67F IsdG–heme–CN (top), plus saturation magnetization curves for the 424 nm MCD band of W67F IsdG–heme–CN (bottom) [10]. All data was acquired for samples is 50 mM KP_i pH 7.4, 150 mM NaCl, 60% glycerol (v/v). The W67F substitution increases the population of a ²E_g electronic state, but retains a ²B_2g electronic ground state.

The temperature and field dependence of the W67F IsdG–heme–CN MCD spectrum was analyzed to gain further insight into the electronic structure of this species. VTVH MCD has proven to be an exquisitely sensitive probe of transition metal electronic structure [43], and a framework has been established to readily interpret data for low-spin ferric heme [4]. It has been well-established since the advent of the so-called “Gouterman four-orbital model” that the heme Soret band is xy-polarized [44], which means that the magnetic field dependence of this band depends upon g_z. It has also been shown that different low-spin ferric heme electronic states have different g_z values [9], and it is arguably easier to measure the values indirectly using VTVH MCD as compared to directly via EPR. VTVH MCD does not suffer from saturation effects that may make some EPR signals undetectable. The VTVH MCD curves for W67F IsdG–heme–CN are clearly consistent with an S = ½ ground state based upon their similarities to simulated curves for two low-spin heme model complexes (Fig. 5) [45]. It should be noted that, based upon recently published EPR data for cyanide-ligated myoglobin [11], the simulated curves for the ²E_g model complex may significantly underestimate the slope for the ²E_g state of IsdG–heme–CN. The differences between the slopes of the 2, 5, and 10 K curves are slight, but similar data has been reported previously for MhuD–heme–CN [4], IsdG–heme–CN [10], and IsdI–heme–CN [6]. The “nesting” of the 2, 5, and 10 K curves arises from thermal population of low-lying electronic excited states at cryogenic temperatures [6]. The slope of the VTVH MCD saturation magnetization curves increases with increasing temperature, indicating that W67F IsdG–heme–CN has a ²B_2g ground state and a ²E_g excited state. This is a surprising result since the signals arising from the ²E_g state dominate the ¹H NMR spectrum of W67F IsdG–heme–CN due to exchange broadening (Fig. 4), and illustrates the advantage of VTVH MCD as a heme electronic structure probe. Consequently, the electronic ground state of W67F IsdG–heme–CN is ²B₂g and this species possesses a low energy ²E_g excited state that can be thermally populated.

3.5. W66Y IsdI–heme–CN also has a ²B_2g electronic ground state

As the W66Y IsdI–heme–CN variant has a very similar ¹H NMR spectrum to that reported for W67F IsdG–heme–CN, the electronic structure of this species was investigated using MCD spectroscopy. X-ray crystal structures of both WT and W66Y IsdI–heme–CN are available (PDB IDs 3QGP and 4FNI) [1, 3]; this second-sphere substitution does not alter the secondary or tertiary structure of the enzyme. As noted above, the ¹H NMR spectrum of W66Y IsdI–heme–CN has downfield resonances previously assigned to the heme methyl groups of a ²E_g electronic state [3], but this study has shown that exchange broadening can mask signals arising from a ²B_2g electronic state. There are several differences in the 13,000–21,000 cm⁻¹ regions of the WT and W66Y IsdI–heme–CN MCD spectra (Fig. 6) [6], and these differences are very similar to those observed between WT and W67F IsdG–heme–CN (Fig. 5). Also similar to the observations for WT and W67F IsdG–heme–CN, the Soret band MCD intensity increases four-fold from WT to W66Y IsdI–heme–CN. As noted above, these spectral changes can all be attributed to increased population of a ²E_g state in the variants. These data are consistent with a ²E_g ground state for W66Y IsdI–heme–CN, but they could also be explained by a subtler electronic structure change. If both WT and W66Y IsdI–heme–CN are thermal mixtures of ²E_g and ²B_2g electronic states, then the ¹H NMR and MCD observations could be explained by a perturbation of the relative populations of the ²E_g and ²B_2g electronic states. As ¹H NMR and MCD spectral data could not resolve the ambiguity regarding the electronic ground state of W66Y IsdI–heme–CN, the temperature and field dependence of the MCD data were carefully examined.

Fig. 6.

5 K, 7 T MCD spectra of WT and W66Y IsdI–heme–CN (top), plus saturation magnetization curves for the 422 nm MCD band of W66Y IsdI–heme–CN (bottom) [6]. All data was acquired for samples is 50 mM KP_i pH 7.4, 150 mM NaCl, 60% glycerol (v/v). The W66Y substitution increases the population of a ²E_g electronic state, but retains a ²B_2g electronic ground state.

VTVH MCD was employed to definitively assign the electronic ground state of W66Y IsdI–heme–CN. As was the case for W67F IsdG–heme–CN, the MCD saturation magnetization curves of W66Y IsdI–heme–CN are clearly consistent with an S = ½ ground state (Fig. 6). Also similar to observations for W67F IsdG–heme–CN, the MCD saturation magnetization curves for W66Y IsdI–heme–CN are “nested” indicating thermal population of more than one S = ½ electronic state at cryogenic temperatures. Notably, the MCD saturation magnetization curves previously reported for WT IsdI–heme–CN are also “nested” [6], which reveals that both WT and W66Y IsdI–heme–CN are thermal mixtures of ²E_g and ²B_2g electronic states. Finally, the slope of the MCD saturation magnetization curves for W66Y IsdI–heme–CN increases with increasing temperature, revealing that the electronic ground state for this species is ²B_2g. Thus, both WT and W66Y IsdI–heme–CN have ²B_2g electronic ground states, and both WT and W66Y IsdI–heme–CN have thermally-accessible ²E_g electronic excited states. However, the energy difference between the ²B_2g and ²E_g states is smaller for W66Y IsdI–heme–CN resulting in greater population of the ²E_g state at physiologically-relevant temperatures.

3.6. W67F IsdG has a diminished rate of heme oxygenation

Next, established UV/Vis Abs assays were employed to ascertain the impact of the heme substrate geometric and electronic structure changes induced by the W67F substitution on IsdG activity. WT IsdG degrades heme upon the addition of ascorbate in the presence of catalase (Fig. 7) [36], to prevent non-enzymatic coupled oxidation [46]. After 60 minutes, the UV/Vis Abs spectrum of the reaction mixture is very similar to that reported previously for the IsdG product [21], which strongly suggests that WT IsdG degrades heme to staphylobilin under these conditions. When this assay was repeated for W67F IsdG, a distinct Soret band at 412 nm was still observable after 60 minutes. This suggests that heme degradation by W67F is slower than for WT enzyme. In addition, a broad band centered at 680 nm gains intensity for the first 30 minutes following ascorbate addition and then decreases to approximately half its maximal intensity after an additional 30 minutes. We reasoned that this 680 nm band could be better resolved by repeating the assay with a higher enzyme concentration and a lower sodium ascorbate concentration. Indeed, for 30 μM W67F IsdG and 1 mM sodium ascorbate, a clear 680 nm band grows for the first 115 minutes of the assay before beginning to diminish. Absorbance at this wavelength is likely derived from production of an unstable biliverdin product [47], but further characterization is needed to reach a definitive conclusion. Nevertheless, based upon the UV/Vis Abs data presented here, it is clear that the W67F substitution alters the rate and/or mechanism of heme degradation by IsdG.

Fig. 7.

Ascorbate-mediated, aerobic degradation of heme by WT IsdG and W67F IsdG (panels A-C), and fits of the time-course for the Soret band to a pseudo-first order kinetic model (panel D). IsdG-catalyzed heme degradation by 10 μM enzyme was monitored for 60 min in 50 mM KP_i pH 7.4 (panels A-B). In addition, heme degradation by 30 μM W67F IsdG was monitored for 120 min (panel C). The rate of heme degradation by 10 μM WT enzyme (0.111 ± 0.004 min⁻¹) is more than double that of the 10 μM W67F variant (0.049 ± 0.006 min⁻¹, panel D).

The pseudo-first order rates for heme oxygenation were extracted from the UV/Vis Abs assays. The Soret band arises from a π → π* transition of the heme substrate [44], so the IsdG-catalyzed oxygenation of heme to meso-hydroxyheme can be monitored by following the time-course of this band [36]. Kinetic analysis is complicated by the fact that IsdG intermediates and products absorb light at this wavelength [21], but these issues have been addressed in a recently reported analytical expression [26]. This analytical expression corrects for non-zero absorbance at infinite time, which is especially important when comparing variants that yield spectroscopically distinct products. Based upon the data reported here (Fig. 7), and the kinetic model reported previously, WT IsdG oxygenates heme with a pseudo-first order rate constant of 0.111 ± 0.004 min⁻¹. Upon introduction of the W67F substitution, the rate decreases to 0.049 ± 0.006 min⁻¹. Thus, IsdG-catalyzed heme oxygenation shows the same substrate ruffling-dependence that was previously reported for two other non-canonical heme oxygenases: IsdI and MhuD [3, 5]. Furthermore, heme oxygenation by IsdG is markedly faster than oxygenation by MhuD, which may be derived from the observation that the IsdG active site induces more heme substrate ruffling than the MhuD active site. Clearly, the W67F substitution decreases the rate of heme oxygenation by IsdG and the likely origin of this observation is reduced heme ruffling.

3.7. W67F IsdG converts heme to biliverdin

ESI-MS was employed to identify the product(s) of IsdG-catalyzed heme degradation. Previous work has identified staphylobilins as the major product of IsdG based upon product isolation and characterization [18]. However, a recent in proteo MS study of the non-canonical HO MhuD revealed that the product isolation approach can fail to detect alternate enzyme products [26]. Indeed, ESI-MS of the WT IsdG reaction mixture 60 min after the addition of ascorbate reveals a mixture of m/z 583, 599, and 611 species (Fig. 8). The 599 and 611 Da species were previously observed for IsdG-catalyzed heme degradation [21], and attributed to the staphylobilin product and formyloxobilin intermediate, respectively. However, the 583 Da species was not reported previously. This molecular weight corresponds to biliverdin, which is the major product of canonical HOs and minor product of MhuD [25, 48]. The in proteo MS data is not rigorously quantitative, but biliverdin appears to represent ~10% of the IsdG product distribution. This level of biliverdin production by IsdG is consistent with the previous observation that CO represents 14 ± 2% of the C1 products for this enzyme [20]. Furthermore, this minor biliverdin product is not likely derived from coupled oxidation since the addition of hydrogen peroxide to IsdG–heme does not yield staphylobilin or biliverdin products [19]. Thus, IsdG degrades heme to a mixture of staphylobilin and biliverdin products.

Fig. 8.

Ascorbate-mediated, aerobic degradation of heme by WT IsdG (top) and W67F IsdG (bottom). The product distributions were assessed using ESI-MS after 60 min of reaction time in 50 mM KP_i pH 7.4. Prior to ESI-MS, samples were eluted from a C₁₈ column with a MeCN gradient. Data were acquired for m/z 600–1500, but only the m/z 500–700 region is shown here for clarity. The major products of WT IsdG are staphylobilin (599 Da) and formyloxobilin (611 Da). On the other hand, the major products of W67F IsdG are biliverdin (583 Da) and formyloxobilin (611 Da).

Finally, the products of W67F IsdG-catalyzed heme degradation were analyzed with ESI-MS. After 60 min, the W67F IsdG reaction intermediate is dominated by a mixture of m/z 583, 611, and 616 species (Fig. 8). These ions correspond to biliverdin [25], formyloxobilin [21], and heme, respectively. The observation of a significant fraction of unreacted heme substrate is consistent with the decreased oxygenation rate determined above (Fig. 7). More surprisingly, a major component of the W67F IsdG reaction mixture is formyloxobilin, but no staphylobilin is observed. Based upon these data, the W67F substitution either alters the regiospecificity of heme monooxygenation and/or inhibits the deformylation of formyloxobilin. Finally, it is particularly noteworthy that biliverdin produced by W67F IsdG is increased compared to WT enzyme. This result is consistent with the observation of a broad band centered at 680 nm in the UV/Vis Abs-monitored assay. Based on these data, it is clear that the W67F substitution alters the IsdG mechanism resulting in a novel product distribution.

4. Discussion

4.1. Heme substrate structure in non-canonical heme oxygenases

A relatively complete picture for the structural role of the conserved second-sphere Trp of non-canonical heme oxygenases is now available. The X-ray crystal structures of IsdI–heme–CN and MhuD–heme–CN (PDB IDs 3QGP and 4NL5) revealed that these enzymes induce a b_1u out-of-plane ruffling distortion of their heme substrates from planarity [1, 4]. More recently, a spectroscopically-validated model of IsdG–heme–CN showed that the IsdG active site generates an intermediate amount of heme ruffling (2.0 Å) compared to IsdI (2.3 Å) and MhuD (1.4 Å, Fig. 9) [3, 7]. Previous studies have shown that substituting a Tyr or Phe for the conserved second-sphere Trp residue decreases the average amount of heme substrate ruffling by 0.4 – 0.8 Å in IsdI and MhuD [3, 5], and the data presented here shows that this same substitution decreases heme ruffling by 0.3 Å in IsdG (Fig. 3). Thus, the structural role of the second-sphere Trp is conserved between IsdG, IsdI, and MhuD; steric interactions between a second-sphere Trp and the heme substrate induce a b_1u out-of-plane ruffling deformation of heme.

Fig. 9.

Cyanide-inhibited, non-canonical heme oxygenases with average heme ruffling distortions ranging from 0.7 to 2.3 Å have been prepared [1, 3–5, 7]. According to multi-reference perturbation theory calculations: the “planar” conformation of cyanide- and histidine-ligated ferric heme exhibits 0.5 Å of ruffling, the “ruffled” conformation exhibits 1.6 Å of ruffling, and the transition state between the two conformations has 1.0 Å of ruffling [6]. The electronic ground state of all six variants of cyanide-inhibited, non-canonical heme oxygenases are accurately predicted by the theoretical model, and all six variants stabilize a thermal mixture of the two heme conformations.

However, simply stating that the second-sphere Trp induces a ruffling deformation of heme is an over-simplification of the underlying structural dynamics. MCD experiments have demonstrated that the heme substrate ruffling distortions measured using X-ray crystallography represent a weighted average of two distinct substrate conformations [6]. To be specific, cyanide- and histidine-ligated ferric heme has “planar” (0.5 Å ruffling distortion) and “ruffled” (1.6 Å ruffling distortion) conformations separated by an energy barrier of 1.2 kcal/mol. Previously reported ¹H NMR, MCD, and multi-reference perturbation theory data have shown that the W66F substitution decreases the average ruffling of heme in MhuD by altering the relative populations of the “planar” and “ruffled” conformations [4, 5]. Here, new MCD data for IsdG and IsdI reveal that the average heme ruffling changes in these enzymes also arise from altered conformer populations (Figs. 5 and 6). Consequently, the most accurate description of heme within non-canonical heme oxygenase active sites is that of a dynamic substrate that rapidly samples both “planar” and “ruffled” conformations. The second-sphere interactions between the conserved Trp residue and the heme modulate the ratio of “planar” and “ruffled” substrate.

It is worth noting that the orientation of the heme substrate relative to the conserved second-sphere Trp residue is different in IsdG, IsdI, and MhuD. In the X-ray crystal structure of WT IsdI–heme–CN (PDB ID 3QGP), a single heme orientation was observed with a steric contact between the β-meso carbon of heme and the Trp66 side-chain [1]. In contrast, the heme substrate is rotated by ~90° in WT MhuD–heme–CN (PDB ID 4NL5), resulting in a steric contact between the α-meso carbon of heme and the Trp66 side-chain [4]. The exact orientation of heme in the X-ray crystal structures of N7A IsdG–heme (PDB ID 2ZDO) and W66Y IsdI–heme–CN (PDB ID 4FNI) is unclear due to disorder in the electron density maps noted by the crystallographers [3, 49], but the second-sphere Trp residue is certainly in contact with either the β- or δ-meso carbons of heme in those species. Regardless, any of these contacts induce the same ruffling deformation of the heme substrate since the b_1u out-of-plane normal mode responsible for heme ruffling induces coupled distortion of all four heme meso carbons [2]. The ~90° rotation of heme in MhuD compared to IsdG and IsdI would certainly be expected to alter the regiochemistry of heme hydroxylation. But, all three enzymes catalyze hydroxylation of the β- and δ-meso carbons [18, 24]. It appears that an altered structural role for the second-sphere Asn residue offsets the rotation of heme within the enzyme active site. In IsdG Asn7 forms a hydrogen bond with the iron-ligating α-oxygen of ferric–peroxoheme [10], whereas in MhuD Asn7 has been proposed to form a hydrogen bond with the terminal β-oxygen of ferric–peroxoheme based upon QM/MM calculations [27]. Thus, the altered orientations of heme within the non-canonical heme oxygenase active sites, and the altered hydrogen bonding interactions with Asn7, may combine to orient the distal peroxo ligand along the β/δ-meso carbon axis of heme in all three enzymes. Nevertheless, further experimental characterization of the structural role for Asn7 within the MhuD active site is warranted before this conclusion can be reached definitively.

4.2. Unusual heme electronic structure in non-canonical heme oxygenases

A relatively complete picture of how the conserved second-sphere Trp of non-canonical heme oxygenases tunes the electronic structure of the heme substrate is also now available. Several years ago MCD characterization of MhuD–heme–CN revealed that this species is a thermal mixture of two electronic states: a ²B_2g ground state with a (_3dxz,yz)⁴(3d_xy)¹ Fe electron configuration, and a low-lying ²E_g excited state with a (3d_xy)²(3d_xz,yz)³ Fe electron configuration (Fig. 9) [4]. Subsequent studies revealed that the ²B_2g and ²E_g electronic states are associated with the “ruffled” and “planar” heme conformations, respectively [6], and the same thermal mixture of a ²B_2g ground state and a ²E_g excited state is present in IsdG and IsdI [10]. It was previously shown that the W66F substitution swaps the relative energies of the “planar” and “ruffled” heme conformations resulting in a ²E_g ground state with a thermally-accessible ²B_2g excited state for W66F MhuD–heme–CN [5]. The data presented here reveals that the influence of a second-sphere Trp substitution is more subtle for IsdG and IsdI: the W67F and W66Y variants of IsdG and IsdI, respectively, retain a ²B_2g ground state but have more populated ²E_g excited states owing to a decreased energy difference between the two conformations. Thus, although the exact details vary from enzyme to enzyme, the ruffling deformation induced by the conserved second-sphere Trp of non-canonical heme oxygenases generally increases the population of a ²B_2g electronic state.

The electronic structure changes observed for cyanide-inhibited non-canonical heme oxygenases can be readily understood in terms of group theory. A b_1u ruffling distortion breaks D_4h symmetry and allows mixing of the Fe 3d_xy and porphyrin a_2u orbitals [44]. This mixing results in an electronic ground state change is observed when this ruffling-induced orbital mixing is large enough to raise the energy of the Fe 3d_xy orbital above the Fe 3d_xz and 3d_yz orbitals [9]. The ground state change is aided by the π*-acceptor character of cyanide, since an axial π*-acceptor will lower the energies of the Fe 3d_xz and 3d_yz orbitals. One might question the functional relevance of electronic structure changes for cyanide-inhibited enzyme when the ferric–peroxoheme species is believed to be the catalytically-relevant intermediate [22, 24]. However, although peroxide and azide are π donor ligands, which will raise the energies of the Fe 3d_xz and 3d_yz orbitals and inhibit an electronic ground state change, ruffling will still induce mixing of the Fe 3d_xy and porphyrin a_2u orbitals in these cases [7]. This electronic structure perturbation is unlikely to yield the wholesale ground state changes observed for cyanide-inhibited species, but it will still result in partial porphyrin-to-iron electron transfer. Consequently, the heme electronic structure change introduced in the “ruffled” substrate conformation of non-canonical heme oxygenases will favor hydroxylation of the meso carbons for the catalytically-relevant ferric–peroxoheme intermediate.

4.3. Functional role of a conserved second-sphere Trp residue

A general correlation between heme substrate conformation and heme oxygenation rate has been identified for non-canonical heme oxygenases. IsdG, IsdI, and MhuD have all been shown to convert heme to β- and δ-meso-hydroxyheme [18, 21, 24]. It was previously shown that the rate of IsdI-catalyzed heme degradation depends upon the average amount of heme ruffling induced by a particular variant [3], and this work clarifies that the rate depends upon population of the “ruffled” substrate conformation (Fig. 6). It was also previously shown that the rate of MhuD-catalyzed heme monooxygenation depends upon the population of the “ruffled” substrate conformation [26], although this conclusion relied upon the development of a new analytical expression to analyze kinetic data acquired for several MhuD variants. Here, this new analytical expression was used to analyze kinetic data for IsdG variants revealing that the rate of IsdG-catalyzed heme oxygenation also depends upon the population of a “ruffled” substrate conformation (Fig. 7) and, furthermore, the correlation between heme oxygenation rate and the population of the “ruffled” conformation extends to a comparison between WT IsdG and MhuD. However, substrate conformation is clearly not the only determinant for the heme monooxygenation reaction catalyzed by non-canonical HOs. The heme substrate of MhuD is rotated by ~90° in the enzyme active site compared to IsdI [4], yet these two enzymes have been shown to form the same β- and δ-meso-hydroxyheme intermediates. The resolution of this regiochemical conundrum is most likely the role of a conserved second-sphere Asn residue that has been shown to be essential for both reactions [23, 38].

All of these data can be understood in terms of a concerted rearrangement mechanism. This mechanism was previously proposed for IsdG-catalyzed heme monooxygenation (Fig. 10) [10, 22], and here we add the additional caveat that this mechanism is only accessible to the “ruffled” substrate conformation. There is not a general steric explanation for the correlation between ruffling and monooxygenation rate for non-canonical heme oxygenases because the distance between the terminal oxygen of ferric–peroxoheme and the β/δ-meso carbons of porphyrin is longer for the “ruffled” conformation of MhuD [4]. Instead, the most general explanation is electronic: ruffling favors mixing of the Fe 3d_xy and porphyrin a_2u orbitals [6], which triggers partial porphyrin to iron charge transfer [5], which increases the electrophilicity of the heme meso carbons [12]. The regioselectivity of this monooxygenation reaction can only be explained by attributing a role to the conserved second-sphere Asn residue. For IsdG, spectroscopic data indicates that a hydrogen bond between Asn7 and the iron-ligating α atom of the distal peroxo ligand in the ferric-peroxoheme intermediate directs the monooxygenation towards the β- and δ-meso carbons [10]. For MhuD, a computational model suggests that a hydrogen bond between Asn7 and the terminal β atom of the distal peroxo ligand in the ferric-peroxoheme intermediate directs the monooxygenation towards the β- and δ-meso carbons [27]. Importantly, this model explains why both the second-sphere Asn and the second-sphere Trp are required for non-canonical HO-catalyzed heme monooxygenation.

Fig. 10.

Two possible mechanisms for heme monooxygenation by non-canonical heme oxygenases. The concerted rearrangement mechanism is the most consistent with available data for WT IsdG, IsdI, and MhuD. An alternative, stepwise radical mechanism could explain the altered product distributions for W66F MhuD and W67F IsdG.

It is also now possible to identify a clear correlation between population of the “ruffled” substrate conformation and product identity for non-canonical heme oxygenases. It was previously shown that the major products of WT IsdG and WT MhuD are staphylobilin and mycobilin, respectively [18, 23]. Recently, it was realized that a second-sphere variant of MhuD that decreases population of the “ruffled” conformation, W66F, changes the major product of MhuD from mycobilin to biliverdin [26]. The ESI-MS data reported here reveals that decreased population of the “ruffled” conformation of IsdG induced by the W67F substitution also increases production of biliverdin (Fig. 8). These changes to product distributions could potentially be attributed to altered regioselectivity for the initial heme monooxygenation by invoking a stepwise radical mechanism (Fig. 10) [50], similar to that proposed for canonical HOs [51], where homolytic cleavage of the O–O bond in the ferric–peroxoheme form of HO yields a transient hydroxyl radical in the enzyme active site. However, without the organized water cluster of canonical HOs [52–54], porphyrin hydroxylation by a transient hydroxyl radical would be expected to be indiscriminant yielding a mixture of all four meso-hydroxyheme isomers as observed for non-enzymatic coupled oxidation [46]. Based upon he structures of the two known staphylobilin isomers [18], the β- and δ-meso-hydroxyheme isomers could be expected to proceed to formyloxobilin, whereas the α- and γ-isomers cannot. This could explain the roughly 50:50 formyloxobilin:biliverdin product distribution for W67F IsdG (Figure 8), but additional data for enzymatic reactions initiated from regiochemically-defined intermediates would be needed to parse this level of mechanistic detail [24].

It is also interesting to note that the W67F substitution fully suppresses the conversion of formyloxobilin to staphylobilin by IsdG. It was previously shown that WT IsdG deformylates ~45% of the formyloxobilin generated by this enzyme to staphylobilin, with the ~55% remainder unconverted for at least 12 h [21]. This product distribution is consistent with the data reported here for WT enzyme (Figure 8). In contrast, heme oxygenation by W67F IsdG exclusively generates formyloxobilin, with no staphylobilin observed. Thus, there is a clear correlation between staphylobilin production by IsdG and access to the “ruffled” conformation, but this correlation is not necessarily a causal relationship. It is important to note that formyloxobilin spontaneously deformylates upon removal from IsdG, and this reaction may not be an enzyme-catalyzed process. Instead, the differences between the reactivity of IsdG-bound formyloxobilin species could simply be due to different amounts of solvent access to the substrate for each variant–isomer combination. As noted above, further insight into the W67F IsdG mechanism will likely require enzyme assays initiated from regiochemically-defined intermediates.

5. Conclusion

It is now possible to make several general statements regarding the role of the conserved second-sphere Trp of non-canonical heme oxygenases. First, the steric interaction between Trp and heme favors population of a “ruffled” substrate conformation over the more typical “planar” substrate conformation. Second, these two substrate conformations are in dynamic equilibrium and the geometric changes are coupled with electronic changes, namely, the “ruffled” conformation promotes partial porphyrin-to-iron electron transfer. Finally, enzymatic rates and product distributions depend upon the relative populations of the “planar” and “ruffled” substrate conformations.

Highlights.

• The heme substrate has access to two conformations, “planar” and “ruffled”, when bound to IsdG.

• The major product of Staphylococcus aureus IsdG is staphylobilin.

• The minor product of Staphylococcus aureus IsdG is biliverdin.

• The product distribution depends upon the ratio of “planar” to “ruffled” heme.

• The heme oxygenation rate depends upon the population of the “ruffled” conformation.

Abbreviations

Abs

absorption

CD

circular dichroism

DSS

sodium 2,2-dimethyl-2-silapentane-5-sulfonate

ESI-MS

electrospray ionization mass spectrometry

FPLC

fast protein liquid chromatography

IsdG–heme

heme-bound IsdG

IsdG–heme–CN

cyanide-inhibited IsdG

IsdI–heme

heme-bound IsdI

IsdI–heme–CN

cyanide-inhibited IsdI

KP_i

potassium phosphate

MCD

magnetic circular dichroism

MeCN

acetonitrile

MhuD–heme–CN

cyanide-inhibited MhuD

MS

mass spectrometry

NaP_i

sodium phosphate

NMR

nuclear magnetic resonance

QM/MM

quantum mechanics / molecular mechanics

TEV

tobacco etch virus

VTVH

variable-temperature, variable-field