Heme redox potentials hold the key to reactivity differences between nitric oxide reductase and heme-copper oxidase

Significance

Nitric oxide reductases (NORs) catalyze the reduction of NO as part of the denitrification cycle, while structurally related heme-copper oxidases (HCOs) catalyze oxygen reduction during respiration. Despite decades of investigation into the structure/function of HCOs and NORs, factors governing their reaction specificity remain unknown. By tuning E°′ of a biosynthetic model of NOR in myoglobin and using electrochemical, spectroscopic, and computational methods to understand the impact of such tuning, this work reveals heme E°′ as a key to determining the specificity of these two enzymes and explains their cross-reactivity. Beyond resolving such a long-standing issue, this work provides guidelines for the design and synthesis of artificial metalloenzymes that can catalyze reactions relevant to global nitrogen cycles and biological respiration that is important for oxygen reduction reaction in fuel cells.

Abstract

Despite high structural homology between NO reductases (NORs) and heme-copper oxidases (HCOs), factors governing their reaction specificity remain to be understood. Using a myoglobin-based model of NOR (Fe_BMb) and tuning its heme redox potentials (E°′) to cover the native NOR range, through manipulating hydrogen bonding to the proximal histidine ligand and replacing heme b with monoformyl (MF-) or diformyl (DF-) hemes, we herein demonstrate that the E°′ holds the key to reactivity differences between NOR and HCO. Detailed electrochemical, kinetic, and vibrational spectroscopic studies, in tandem with density functional theory calculations, demonstrate a strong influence of heme E°′ on NO reduction. Decreasing E°′ from +148 to −130 mV significantly impacts electronic properties of the NOR mimics, resulting in 180- and 633-fold enhancements in NO association and heme-nitrosyl decay rates, respectively. Our results indicate that NORs exhibit finely tuned E°′ that maximizes their enzymatic efficiency and helps achieve a balance between opposite factors: fast NO binding and decay of dinitrosyl species facilitated by low E°′ and fast electron transfer facilitated by high E°′. Only when E°′ is optimally tuned in Fe_BMb(MF-heme) for NO binding, heme-nitrosyl decay, and electron transfer does the protein achieve multiple (>35) turnovers, previously not achieved by synthetic or enzyme-based NOR models. This also explains a long-standing question in bioenergetics of selective cross-reactivity in HCOs. Only HCOs with heme E°′ in a similar range as NORs (between −59 and 200 mV) exhibit NOR reactivity. Thus, our work demonstrates efficient tuning of E°′ in various metalloproteins for their optimal functionality.

Nitric oxide reductases (NORs) from denitrifying bacteria catalyze the two-electron reduction of NO to nitrous oxide (N₂O) as part of the global denitrification cycle that converts nitrite and nitrate to nitrogen (1). Specifically, the cytochrome c-dependent NORs (cNORs) are integral membrane proteins and are evolutionarily related to the respiratory heme-copper oxidases (HCOs). They possess a catalytic subunit that anchors a heme-nonheme iron active site (heme b₃/nonheme Fe_B) that is analogous to the heme-copper center of HCO (2, 3). These two classes of enzymes are not only homologous in sequence and structure (4) but exhibit cross-reactivity as well. Some NORs can reduce O₂ to H₂O while certain HCOs can reduce NO to N₂O, albeit the catalytic rate and efficiency for the cross-reactivity of these enzymes are much lower than their native reactivity (4, 5). A crucial difference between HCOs and NORs, however, is in the reduction potential (E°′) of their catalytic heme iron (Fe³⁺/Fe²⁺). While heme E°′ of different HCO types varies over a wide range from −59 mV (6) to +450 mV (7) (SI Appendix, Table S2), NORs are found to exhibit a narrower range of heme E°′ between −168 mV (8) and +86 mV (9), generally lower than that of HCOs (SI Appendix, Table S3). The reason behind these differences in heme E°′ for two enzyme classes that are otherwise very similar and that potentially evolved from a common ancestral protein (10) remains unknown. Moreover, the impact that such variation in heme E°′ has on the reactivity of NORs is not understood. Addressing these issues requires systematic variation of heme E°′ in NORs within physiological E°′ range. However, the large size and membranous nature of HCO/NOR make it difficult to systematically modulate redox potential in the native proteins and study its associated impact on the enzymatic activity. As an example, HCO and NOR assemble and fold in the presence of heme cofactors and it is difficult to individually replace these cofactors. Thus, a systematic study on redox control of enzymatic activities is challenging in multimetal center-containing enzymes, such as HCOs and NORs.

Investigations into simple, well-characterized heme proteins like myoglobins and cytochromes have revealed numerous structural features that proteins employ to achieve fine control over redox potentials such as different heme ligands and heme types, variation of secondary coordination sphere hydrogen-bonding network, and distortion in heme planarity (11–13). A few of these factors have also been shown to play important roles in HCO/NOR enzymatic systems. For instance, HCO and NOR from different species contain different heme types: bovine cytochrome c oxidase (CcO) contains heme a at its catalytic center with an electron-withdrawing formyl group and exhibits the highest heme E°′ value of approximately +480 mV (7). On the other hand, Escherichia coli bo₃ oxidase contains heme o with no formyl groups and exhibits a comparatively low E°′ value of approximately +180 mV (14). Moreover, H-bond variation of proximal histidine (6) and the presence of nonheme metal cofactors also vary redox properties of catalytic heme iron.

To meet challenges associated with the study of large, complicated metalloproteins like native HCOs and NORs, models of native enzymes based on small molecules, peptides, and proteins have been designed (15–22), but no synthetic model of NOR, to our knowledge, has been used to investigate the impact of heme E°′ on its functionality. Moreover, no synthetic model of NOR exhibits more than a single turnover of NO reduction. In this work, we utilize a myoglobin (Mb)-based structural and functional model of NOR (23–28) to investigate the impact of tuning heme E°′ on NOR functionality. The structural simplicity of Mb makes it amenable to a variety of modulations for redox tuning: the H-bonding interaction to the proximal ligand of the heme iron can be readily tuned in Mb through site-directed mutagenesis of nearby amino acid residues (29). Moreover, the heme cofactor can be easily replaced in the protein with modified hemes that are analogous to different heme types in HCOs/NORs (30). Overall, Mb models of HCO/NORs allow for a systematic and isolated study of the effect of redox potential on enzymatic activities, without convoluting the influence of other factors.

The active site of NORs consists of a histidine residue ligated to a high-spin heme iron coupled to a high-spin nonheme iron (Fe_B). The Fe_B center is coordinated to three His (H211, H258, and H259) and one Glu (E211) in a distorted trigonal bipyramidal geometry (Fig. 1A) (3, 31). The Mb-based NOR model was named Fe_BMb as it could bind nonheme iron (Fe^II) at the designed Fe_B center (23, 28). The Fe^II-bound complex was called Fe^II-Fe_BMb(Fe^II), where the former Fe^II refers to the nonheme iron in the Fe_B center and the latter Fe^II in parenthesis refers to the heme iron in the ferrous state. The Fe^II-Fe_BMb(Fe^II) was shown to bind nonheme Fe^II via three His ligands (H29, H43, and H64) and one Glu ligand (E68) in a distorted trigonal bipyramidal geometry like native NORs (Fig. 1B) and selectively reduced NO to N₂O analogous to the NO reduction reactivity of NORs (23, 32). Moreover, spectroelectrochemical experiments on Fe^II-Fe_BMb revealed a heme E°′ value of −59 ± 8 mV that was in-between the two extremes of E°′ values reported for cNORs from Marinobacter hydrocarbonoclasticus (−168 mV) (8) and Paracoccus denitrificans (+86 mV) (33).

Fig. 1.

Design of NOR mimics with tuned heme E°′ values. (A) Crystal structure of catalytic active site of cNOR from Pseudomonas aeruginosa (PDB ID code 3O0R). (B) Crystal structure of the heme center in Fe^II-Fe_BMb(Fe^II) (PDB ID code 3K9Z). (C) Overlay of the crystal structures of L89S-Fe_BMb (orange; PDB ID code 6D45) with Fe_BMb (green; PDB ID code 3K9Z). (D) Nernst fit of the spectroelectrochemical measurements for Fe^II-Fe_BMb and its variants. The dotted line represents the point at which the fraction of protein reduced is equal to that oxidized and the corresponding potential value is the heme E°′. (E) Variation of heme E°′ for HCOs (in blue) and NORs (in green). Experimental values are shown in bold, while the range is shown in lighter colors. Lines on the far right represent the heme E°′ of Fe^II-Fe_BMb variants used in this study.

We began our studies by modulating hydrogen bonding (H-bonding) interactions to the proximal histidine (H93) that coordinates to the heme iron (29, 34). Our previous work had suggested that removing the weak H-bond between N_δ of proximal histidine (H93) and S92 increases heme E°′ by 27 mV (Fig. 1C) (30). We hypothesized that doing the reverse (i.e., increasing H-bond donation to H93) will decrease heme E°′. The L89S mutation shown in Fig. 1C was thus engineered to H-bond to H93 and increase hydrophilic character of the heme proximal pocket that has previously been shown to decrease heme E°′. (29) The crystal structure of L89S-Fe_BMb, obtained at 1.8 Å resolution, showed that S89 was, in fact, directed away from H93. The L89S residue was held in this orientation through H-bonding to two water molecules in the proximal pocket. The addition of hydrophilic L89S residue resulted in lowering the heme E°′ of Fe_BMb, with Fe^II-L89S-Fe_BMb exhibiting an E°′ value of −130 ± 2 mV (Fig. 1D, gray curve). To increase the heme E°′ of the NOR mimic, we employed an alternative strategy of using nonnative heme types with increased heme E°′ values. Monoformyl (MF-) and diformyl (DF-)hemes when incorporated in apo-Fe_BMb resulted in Fe_BMb(MF-heme) and Fe_BMb(DF-heme). The ferric forms of the nonnative heme containing Fe_BMb variants displayed a Soret band at a range of ∼400 nm followed by α and β bands in the visible region, typical of a well-folded myoglobin (Fig. 2 and Table 1). The Fe^II-Fe_BMb(MF-heme) and Fe^II-Fe_BMb(DF-heme) displayed increased heme E°′ values of +53 ± 7 mV and +148 ± 5 mV, respectively (Fig. 1E) (30, 35). Thus, by tuning hydrogen-bonding interactions and by employing modified heme cofactors, we can systematically tune the E°′ of an NOR mimic by ∼280 mV, covering most of the range of NOR heme E°′ and even beyond.

Fig. 2.

UV-visible spectra of the ferric (dark) and ferrous (light) forms of L89S-Fe_BMb are shown in gray (A), Fe_BMb in green (B), Fe_BMb(MF-heme) in orange (C), and Fe_BMb(DF-heme) in red (D), after reducing the proteins with dithionite. Insets show the visible portion of the spectra magnified ∼5 times for each protein.

Table 1.

UV-visible spectral parameters of Fe_BMb variants

	Ferric form			Deoxy form
FeBMb variant	Soret	Visible Prominent peak	Visible Shoulder peak	Soret	Visible
FeBMb	406	501	618	433	556
L89S-FeBMb	406	502	619	433	556
FeBMb(MF-heme)	422	507	656	443	575
FeBMb(DF-heme)	430	509	662	462	585

The wavelength maxima above are reported in nanometers.

This designed set of NOR models with tuned heme E°′ provides us with an opportunity to investigate the impact of tuning E°′ on the chemical, functional, and electronic character of NO, and the resulting mechanistic implications on the NOR reaction. We began with FTIR spectroscopic studies on NO-treated Fe^II-Fe_BMb(Fe^II) variants to investigate if E°′ tuning modulates vibrational characteristics of heme-nitrosyl species. Fe^II-Fe_BMb(Fe^II), when reacted with one equivalent of NO, displayed a strong FTIR vibration at 1,549 cm⁻¹ that shifted to 1,527 cm⁻¹, with ¹⁵NO corresponding to the N-O stretch of a 6-coordinate low-spin (6c-LS) heme-nitrosyl (Fig. 3A) (25, 32). This N-O stretch frequency systematically upshifted to 1,560 (Δ¹⁵N −28) cm⁻¹ and 1,570 (Δ¹⁵N −31) cm⁻¹ in Fe^II-Fe_BMb(MF-heme)(Fe^II) and Fe^II-Fe_BMb(DF-heme)(Fe^II), respectively. This trend suggests the weakening of the N-O bond with decreased heme E°′ in NOR models, which can be attributed to the higher electron density of heme iron at low E°′ values. The increased electron density results in increased dπ-pπ* back bonding (36, 37), leading to the strengthening of the Fe-N bond and, in turn, weakening of the N-O bond.

Fig. 3.

Vibrational spectroscopic, kinetic, and enzymatic characterization of NOR mimics. (A) Light-induced FTIR difference spectra of ¹⁴NO (black)- and ¹⁵NO (red)-bound Fe_BMb. The (¹⁴NO–¹⁵NO) difference spectra (purple) identify the N-O stretching frequencies. (B) The rates of NO binding to heme (in black), decay of 6c-LS heme-nitrosyl (in purple) and ET (in red) for Fe_BMb variants. (C) Enzymatic NO reduction assays for Fe^II-Fe_BMb(Fe^II) variants. Inset shows the percentage of N₂O produced with respect to NO for Fe_BMb(MF-heme) in GC-MS assays.

To further elucidate the modulation in NO-bond character with changes in heme E°′, we performed density functional theory (DFT)-based calculations on 6c-LS heme-nitrosyl structures of native heme, MF-heme, and DF-heme. Even though the overall structures of heme-nitrosyl species are very similar (SI Appendix, Fig. S7), there are significant differences in their electronic and chemical properties. Consistent with the results of vibrational spectroscopy, the N-O bond length increases with a decrease in heme E°′ (DF-heme: 1.188 Å; MF-heme: 1.189 Å; heme b: 1.190 Å), suggesting weakening of the N-O bond with increased electron density on heme iron. Moreover, the negative charge of heme-bound NO also increases with a decrease in heme E°′ values [−0.281 e (DF-heme) < −0.298 e (MF-heme) < −0.315 e (heme b)], suggesting enhanced activation of the N-O bond with a decrease in E°′. Overall, FTIR studies together with DFT calculations reveal that lower values of heme E°′, compared with HCOs, may be preferred by NORs to provide sufficient electron density to heme-bound NO to facilitate N-N bond formation for N₂O production.

Having established that heme E°′ can significantly modify chemical properties of heme-bound NO, we then investigated the impact of heme E°′ on NO reduction kinetics in the NOR mimics. Fe^II-Fe_BMb(Fe^II), on reacting with excess NO, converts to a 6c-LS heme-nitrosyl species with an observed rate (k_binding) of 0.87 s⁻¹ wherein the heme iron binds an NO molecule. The heme-nitrosyl species then decays with an observed rate (k_decay) of 0.49 s⁻¹ (see SI Appendix, Figs. S8 and S9 for more information on NO reduction kinetics of Fe_BMb variants). Mechanistically, such decay may result from radical coupling of the NO bound to Fe_B (through a “trans” pathway, SI Appendix, Scheme S1) or electrophilic attack of a free NO (through a “cis heme b₃” pathway) to heme nitrosyl to obtain N₂O (38). Fe^II-Fe_BMb(Fe^II) variants with different heme E°′ also showed similar spectral transitions (SI Appendix, Figs. S8 and S9). However, the rate of formation of 6c-LS heme-nitrosyl species and the rate of its decay varied significantly. Fe_BMb(DF-heme), Fe_BMb(MF-heme), and L89S-Fe_BMb displayed k_binding values of 0.015, 0.18, and 2.7 s⁻¹, respectively (Fig. 3B, black dashed line). These results indicate that decreasing heme E°′ from +148 ± 5 mV in Fe_BMb(DF-heme) to −130 ± 2 mV in L89S-Fe_BMb has resulted in a 180-fold enhancement in NO association rates. This increase in NO-binding rates can be attributed to the radical nature of the NO molecule, which would prefer binding to hemes with higher electron density and lower E°′ values. Additionally, the rate at which heme-nitrosyl species decayed also increased by 633-fold with decreasing heme E°′: Fe_BMb(DF-heme), Fe_BMb(MF-heme), and L89S-Fe_BMb displayed k_decay values of 0.003, 0.37, and 1.9 s⁻¹, respectively (Fig. 3B, purple dashed line). This trend suggests that the reactivity of heme-nitrosyl species increases significantly with decreasing heme E°′, probably due to the increased electron density of heme-bound NO, which would facilitate its reaction with a second NO molecule. Given that heme electronics play such significant roles in determining NO reduction kinetics in NOR models, these results point to heme-based “trans” and “cis heme b₃” as the biological NO reduction reaction mechanism. These results also explain why bovine CcO, with a high heme E°′ value of +460 mV, forms a highly stable heme-nitrosyl species upon its reaction with NO and is inhibited in this state (10). Overall, the variation in NO reduction kinetics with heme E°′ shows the functional implications of E°′ tuning on NO reactivity and explains the reason behind NORs exhibiting relatively lower heme E°′ values than their aerobic counterparts, HCOs.

In addition to studying NO reduction kinetics, we also measured rates of electron transfer (ET) to heme iron as E°′ tuning is expected to impact ET rates. The ET rates were determined by measuring the rate at which ferric Fe_BMb variants were reduced to their ferrous forms. The measurements were conducted on Zn^II-added Fe_BMb variants using 300-equivalent N,N,N′,N′-tetramethyl-p-phenylenediamine (TMPD) as an electron donor and 3,000-equivalent ascorbate as a sacrificial reductant. The redox inactive Zn^II metal was incorporated at the Fe_B center to model structural aspects of nonheme iron without interfering with the ET process. The transformation of ferric Fe_BMb to its ferrous form was followed by monitoring the decay of the Soret band centered at 406 nm (typical of ferric heme) accompanied by an increase in the Soret band centered at 433 nm (typical of ferrous heme). The absorbance change at 406 nm was fitted to first exponential decay kinetics with a rate of k_ET = 2.03 × 10⁻⁴/s (SI Appendix, Fig. S10). The Fe_BMb variants with higher E°′ values also showed similar transformations, displaying ET rates of 2.3 × 10⁻³/s for Fe_BMb(MF-heme) and 1.9 × 10⁻²/s for Fe_BMb(DF-heme). The L89S-Fe_BMb variant, on the other hand, showed a biphasic decay with rate constants of 3.7 × 10⁻³/s and 7.4 × 10⁻⁵/s. Nonetheless, we observed an increase of >85-fold in ET rates upon increasing the heme E°′ from Fe_BMb to Fe_BMb(DF-heme), consistent with the Marcus theory of ET in a regime where ΔE°′ is less than reorganization energy (11).

Thus, rate-determining factors—namely, NO-binding/heme-nitrosyl decay rates and ET rates—follow opposite trends with heme E°′ values. We reasoned that by controlling these opposing factors, we can potentially obtain an NOR mimic that performs NO-reduction turnovers. Using an NO electrode system previously used for native NORs (31), we measured the NO-reduction turnovers of different Fe_BMb variants (Fig. 3C). Interestingly, only Fe_BMb(MF-heme) with appropriately tuned NO binding/heme-nitrosyl decay and ET rates performed ∼35 turnovers of NO reduction, generating N₂O as the product as determined in a separate set of GC-MS experiments (SI Appendix, Fig. S12). Therefore, only when E°′ is optimally tuned in Fe_BMb(MF-heme) for NO binding, heme-nitrosyl decay, and ET does the protein achieve multiple turnovers. These results suggest that E°′ tuning is an effective strategy for modulating functional properties of enzymes and small molecule catalysts. To our knowledge, this is the first NOR model that performs multiple turnovers of an NO reduction reaction. These results also suggest that NORs exhibit appropriately tuned E°′ values that maximize their enzymatic efficiency and help achieve a balance between opposing factors: fast NO binding and decay of heme-nitrosyl species facilitated by low E°′ and fast ET facilitated by high E°′. In all, our experiments reveal heme E°′ tuning as an efficient method for NORs and other heme enzymes to control their chemical reactivity, ET rates, and enzymatic turnovers.

By employing a set of biosynthetic models of NOR with tuned E°′ we show that heme E°′ is a critical factor in guiding NO reduction reactivity of NORs. The heme E°′ in NORs is evolutionarily set in a lower range than HCOs to facilitate N-O bond activation, NO binding, and heme-nitrosyl decay. These results also explain the selective NO reduction cross-reactivity of HCOs: HCOs with low heme E°′ values such as caa₃ oxidase (E°′ = 180 mV), ba₃ oxidase (E°′ = 199 mV) (39) from Thermus thermophilus, bo₃ oxidase from E. coli (E°′ = 160–200 mV) (40), and cbb₃ oxidase (E°′ = −59 mV) from Pseudomonas stutzeri perform NOR reactivity (41). Specifically, the cbb₃ oxidases are C-type HCOs that exhibit unusually low heme redox potential (approximately −50 mV) due to H-bond donation by their proximal His to nearby carboxylate like the Asp-His-Fe triad of peroxidases (42). These C-type HCOs exhibit the highest NO reduction cross-reactivity, consistent with the conclusions of this study that low values of heme E°′ are preferred for NO reduction reactivity. On the other hand, bovine CcO with E°′ values at a higher range of +460 mV does not reduce NO and is instead inhibited in the presence of nanomolar NO (10). Building upon these results, we conducted a survey of E°′ in different heme enzymes, and our results revealed significant variation in E°′ values in enzymes for their optimal function (SI Appendix, Fig. S13). For instance, lower E°′ values in oxygen-activating cytochrome P450 (−250 to −350 mV) (43) compared with oxygen-transporting globins (+40 to +180 mV) (11) are to facilitate O-O bond cleavage in the former. Overall, this work explains the biochemical and mechanistic reason behind E°′ variation in different heme proteins. Heme E°′ is efficiently tuned in different heme enzymes to control their chemical reactivity, functionality, and turnovers.

Materials and Methods

Preparation of Fe^II-Fe_BMb(Fe^II) Variants.

The Fe^II-Fe_BMb(Fe^II) variants were prepared as described previously (6) with slight modifications. Briefly, E-Fe_BMb(Fe^II) was degassed and transferred in an anaerobic glovebag, where it was reduced using dithionite and exchanged with a PD10 column equilibrated with 50 mM pH 7.3 Bis-Tris buffer. FeCl₂ was then titrated in the E-Fe_BMb(Fe^II) to obtain Fe^II-Fe_BMb(Fe^II) variants.

Spectroelectrochemical Measurements.

The reduction potential was measured by a spectroelectrochemical method using an optically transparent thin-layer cell (OTTLE) as described previously (19). The working electrode was made from a piece of 52-mesh platinum gauze. A 1-mm-diameter platinum wire was used as the auxiliary electrode, while a piece of Pasteur pipette filled with agar gel containing 0.2 M KCl was used as a salt bridge to connect the Ag/AgCl (3 M KCl) reference electrode to the bulk solution containing the working and auxiliary electrodes. Generally, the redox titration was performed using ∼0.6 mL of working solution containing 0.2 mM protein, 40 mM phenazine methosulfate, 40 mM anthraquinone-2-sulfonate, and 20 mM TMPD as redox mediators in 50 mM Bis-Tris buffer, pH 7.3, 100 mM KCl. The working solution was purged gently with argon before transferring to the OTTLE cell. A model 362 potentiostat from Princeton Applied Research was used to control the potential of working electrode. After the potential was applied (typically with 25-mV increments), the UV-visible spectra were recorded using a Cary 3E spectrophotometer until no further spectral changes occurred. The Ag/AgCl (3 M KCl) reference electrode was calibrated using ferrocene carboxylic acid as the standard and was found to be 220 mV versus standard hydrogen electrode (SHE). All reduction potentials mentioned in this work are reported against SHE.

X-Ray Crystallography.

Crystals of L89S-Fe_BMb were set up and grown at 4 °C on hanging drops containing 0.1 M MES, pH 6.57, 0.2 M NaOAc.3H₂O, 30% PEG 6K and the well buffer. Drops contained an equal volume of 1 mM L89S-Fe_BMb in 20 mM Tris.H₂SO_4, pH 8, and the well buffer. Before mounting, the L89S-Fe_BMb crystals were soaked in buffer containing 0.1 M Mes pH 6.0, 0.2 M NaOAC.3H2O, and 30% PEG 6K for ∼30 min and then frozen in a cryoprotectant solution of 50 mM Mes pH 6.0, and 30% PEG 400. Diffraction data were collected at the Brookhaven National Lab Synchrotron Light Source X29 beamline. The crystal structure was solved using the same method as for S92A-F33Y-Cu_BMb (20).

DFT Studies on NO-Bound Fe_BMb Variants.

Quantum chemical calculations were performed on three heme models—heme b, MF-heme, and DF-heme—in which all porphyrin substituents are kept the same as in the real systems, except that the propionate group was replaced by methyl to facilitate the calculations. The geometries were fully optimized by using the DFT method mPWVWN with a Watcher’s basis for Fe, 6–311++G(2d,2p) for all coordinated atoms plus the oxygen in NO, and a 6–31G(d) basis for the rest. The atomic charges and spin densities were further calculated by using the hybrid DFT method B3LYP with the same basis set. All calculations were performed using the Gaussian 09 program. See SI Appendix for more information.

FTIR Studies of NO-Bound Fe^II-Fe_BMb(Fe^II) Variants.

FTIR photolysis experiments were carried out as described previously (3). The entire procedure for the preparation of the IR film was performed in an anaerobic glovebox containing <1 ppm O₂ (Omnilab System; Vacuum Atmospheres Co.). Approximately 20 μL of 1 mM protein solution was loaded in an FTIR cell with a 15-μm path length and sealed in the glovebox before mounting the cell to a sample rod to flash-freeze in liquid N₂. The rod was inserted in a precooled closed-cycle cryogenic system (Omniplex; Advanced Research System) and cooled down to 10 K inside the sample compartment of the FTIR instrument. FTIR spectra were obtained on a Bruker Tensor 27 equipped with a liquid-N₂–cooled mercury cadmium telluride detector. Sets of 1,000-scan accumulations were acquired at 4 cm⁻¹ resolution. Photolysis of the nitrosyl complexes was performed by continuous illumination of the sample directly in the FTIR sample chamber using a 300-W arc lamp after filtering heat and near-infrared emissions.

ET Rate Measurements.

Experiments were performed on a Hewlett-Packard 8453 spectrometer in an anaerobic glovebag. A water bath, connected to the cuvette holder and set to 25 °C, provided temperature control. Briefly, 6 μM E-Fe_BMb(Fe^III) variant was mixed with a 3-molar-equivalent of ZnCl₂ to obtain Zn(II)-bound variant in 50 mM Bis-Tris, pH 7.3. The reaction was initiated by mixing 300× TMPD and 3,000× ascorbate to the protein mixture, and kinetics were recorded for ∼8 h.

NO Reduction Reaction Under Single-Turnover Conditions.

Experiments were performed on a Hewlett-Packard 8453 spectrometer in an anaerobic glovebag. A water bath, connected to the cuvette holder and set to 25 °C, provided temperature control. The reaction was initiated by mixing 10 μM Fe^II-Fe_BMb variant with a 17-molar-equivalent of proli-NONOate. The kinetics of the reaction were recorded for ∼1 h.

GC-MS Experiments for NO Reduction Under Multiturnover Conditions.

The product of the reaction of Fe^II-Fe_BMb(Fe^II) variants with NO was investigated by GC-MS through electrospray ionization mass spectra as described previously (6). Typically, ∼0.2 mM (∼0.5 mL) of protein was mixed with ∼50 equivalents of NO in the presence of 300× TMPD and 3,000× ascorbate.

NO Electrode Reactions Under Multiturnover Conditions.

The NO consumption rates were measured with an ISO-NO electrode equipped with ISO-NO Mark II (World Precision Instruments) in a 2-mL anaerobic reaction chamber at 25 °C. The data were recorded on a Duo 18 (World Precision Instruments). The assay solution contained 50 mM Bis-Tris, pH 7.3; 3 mM ascorbate; and 0.3 mM TMPD. After 20 min of argon flux, proli-NONOate was added to the reaction mix at a final concentration of ∼50–60 μM. Once the background consumption of NO was stable, NO reduction (consumption) was initiated by the addition of Fe_BMb (final concentration was 1 μM). More information on the experimental design, protocols, and data analysis are provided in the SI Appendix.