Short-lived intermediate in N₂O generation by P450 NO reductase captured by time-resolved IR spectroscopy and XFEL crystallography

Significance

The short-lived intermediate formed during the reduction of nitric oxide (NO) to nitrous oxide (N₂O) in denitrification, microbial anaerobic respiration, is a key state for understanding the generation mechanism of N₂O, known not only as a greenhouse gas but also as an ozone-depleting substance on the global level. This paper combined state-of-the-art, time-resolved techniques, such as flow-flash infrared spectroscopy and X-ray free electron laser-based crystallography, and captured the intermediate of a P450-type NO reductase at the atomic and electronic levels. The intermediate was identified as a singly protonated Fe³⁺–NHO^•− radical, offering insights into a radical–radical coupling mechanism for the N–N bond formation in N₂O generation.

Abstract

Nitric oxide (NO) reductase from the fungus Fusarium oxysporum is a P450-type enzyme (P450nor) that catalyzes the reduction of NO to nitrous oxide (N₂O) in the global nitrogen cycle. In this enzymatic reaction, the heme-bound NO is activated by the direct hydride transfer from NADH to generate a short-lived intermediate (I), a key state to promote N–N bond formation and N–O bond cleavage. This study applied time-resolved (TR) techniques in conjunction with photolabile-caged NO to gain direct experimental results for the characterization of the coordination and electronic structures of I. TR freeze-trap crystallography using an X-ray free electron laser (XFEL) reveals highly bent Fe–NO coordination in I, with an elongated Fe–NO bond length (Fe–NO = 1.91 Å, Fe–N–O = 138°) in the absence of NAD⁺. TR-infrared (IR) spectroscopy detects the formation of I with an N–O stretching frequency of 1,290 cm⁻¹ upon hydride transfer from NADH to the Fe³⁺–NO enzyme via the dissociation of NAD⁺ from a transient state, with an N–O stretching of 1,330 cm⁻¹ and a lifetime of ca. 16 ms. Quantum mechanics/molecular mechanics calculations, based on these crystallographic and IR spectroscopic results, demonstrate that the electronic structure of I is characterized by a singly protonated Fe³⁺–NHO^•− radical. The current findings provide conclusive evidence for the N₂O generation mechanism via a radical–radical coupling of the heme nitroxyl complex with the second NO molecule.

Nitric oxide (NO) is a radical gas molecule exhibiting high reactivity, but it is synthesized in some cellular systems. We should know how biological systems manage the highly cytotoxic NO without damaging biological compounds, such as proteins, lipids, and nucleic acids. The nitrogen oxide–based anaerobic respiration of microorganisms, so-called denitrification, is one such biological NO-generating system, in which NO is generated from nitrite (NO₂⁻) with one-electron reduction catalyzed by NO₂⁻ reductase, but is immediately decomposed into nitrous oxide (N₂O) to avoid its cytotoxicity before diffusing into the cell. This NO decomposition is catalyzed by the iron-containing enzyme (i.e., NO reductases [NORs]) by following the reaction: 2 NO + 2 e⁻ + 2 H⁺ → N₂O + H₂O. Because the chemistry of this NOR reaction contains N–N bond formation and N–O bond cleavage steps (1), its mechanism has garnered broad interest to be established at the atomic and electronic levels. In environmental science, it is also believed that the molecular mechanism of NO decomposition in biological systems should be established, as the product N₂O of the NOR-catalyzed reaction acts as a greenhouse gas and an ozone-depleting substance on the global level (2), and ∼70% of the global N₂O emission is attributable to NO reduction by NORs in microorganisms in the soil. From such chemical, biological, and environmental points of view, the NOR reaction is highly attractive (3).

NOR in a fungal denitrification system is a cytochrome P450-type enzyme (P450nor), which contains one Cys-coordinated heme at the active site (4). This protein catalyzes the reduction of NO using NAD(P)H as an electron donor: 2 NO + NAD(P)H + H⁺ → N₂O + H₂O + NAD(P)⁺. In the first step of this reaction, NO and NAD(P)H bind to P450nor in the resting ferric state to generate a ternary complex. Although the order of NO and NAD(P)H binding to P450nor is yet unknown, it has been presumed that NO binding precedes NAD(P)H binding because of the low affinity of NAD(P)H (5). The Fe³⁺–NO heme in the complex subsequently reacts with NAD(P)H to form a short-lived intermediate with a ∼100-ms half-life, designated as I (6). The resulting NAD(P)⁺ is proposed to then be released from the protein as the next step (7), and finally, I decays back to the resting state, whose rate depends on the NO concentration, suggesting that the process involves an attack of a second NO to produce N₂O. Thus, it is realized that I is a key state for N–N bond formation and N–O bond cleavage in the P450nor enzymatic reaction. Hence, the coordination and electronic structures of I in this reaction have been a central subject of the NO reduction mechanism in the last few decades.

Because the Soret absorption peak of I is significantly different from those of the Fe³⁺–NO and Fe²⁺–NO complexes of P450nor, we proposed its electronic structure as a two-electron reduced form of the Fe³⁺–NO species, described as {Fe–NO}⁸ in the Enemark–Feltham notation, resulting from the direct hydride (H⁻) transfer from NAD(P)H (6). After our proposal, spectroscopic and theoretical studies have been conducted for the P450nor enzyme and model systems, and some possibilities for the detailed electronic structure of I have been proposed. A time-resolved (TR) resonance Raman study revealed that the N–O bond of the first NO is not cleaved in I and proposed transient hyponitrite (HON = NO⁻) formation by an attack of the second NO on the N atom of I (8). The pulse radiolysis experiment suggested that I could be a Fe³⁺–NHOH^• species (or an equilibrium between Fe³⁺–NHOH^• and Fe⁴⁺–NHOH⁻), because the generation of a hydroxylamine radical (NHOH^•) in the presence of ferric P450nor gave a Soret peak similar to that of I (5). Theoretically, two pathways for hyponitrite formation were first calculated via Fe²⁺–NHO or Fe⁴⁺–NHOH⁻, the latter of which was proposed to be I (9). By contrast, recent quantum mechanics/molecular mechanics (QM/MM) calculations combined with magnetic circular dichroism and Mössbauer spectroscopies suggested that I is more likely to be either Fe³⁺–NHO^•− or Fe³⁺–NHOH^• (10, 11). In parallel, model compound studies have also progressed. For example, the possibility of Fe³⁺–NHOH^• was supported by a study synthesizing Fe³⁺–NHOMe^• (12), whereas another study suggested the possibility of Fe²⁺–NHO by analyzing the reaction of ferric nitrosyl heme complexes with a H⁻ reagent (13, 14). As summarized, despite extensive efforts over the decades by multiple approaches, the electronic structure of I remains elusive, in that possible models (i.e., Fe²⁺–NHO, Fe³⁺–NHO^•−, Fe³⁺–NHOH^•, or Fe⁴⁺–NHOH⁻ state) have been proposed. To establish the NO reduction mechanism of P450nor, there is a need for conclusive experimental evidence for the coordination and electronic structures of the Fe–NO moiety of I. To address this issue, we have been trying to directly observe the intermediates in the P450nor reaction using the TR techniques.

Most recently, we successfully characterized the coordination and electronic structures of the Fe³⁺–NO complex of P450nor with TR, serial femtosecond crystallography (TR-SFX) using an X-ray free electron laser (XFEL) (15). In TR-SFX, caged NO (16) was a useful tool to supply the substrate NO, as it can generate NO molecules quantitatively in the microsecond time domain upon ultraviolet (UV) pulse illumination, and it was possible to track the NO reduction reaction in a TR manner with a light trigger. Additionally, we also developed a new measurement system for TR-infrared (IR) spectroscopy. While conventional Fourier transform IR (FTIR) requires the enzyme sample in the form of a thin layer (<10 μm) of concentrated solution (>millimolar) to suppress the interference by bulk water absorption (17), our TR-IR measurements use a femtosecond laser-based TR-IR spectrometer (18, 19) with an improved design for microspectroscopy, allowing the use of our “microflow flash” system with a thick flow channel (≥15 µm) and relatively low-enzyme concentration (0.4 mM). The details of the TR-IR spectrometer are described in SI Appendix, Fig. S1. In this study, these two TR experimental techniques were applied to elucidate the electronic and coordination structures of the short-lived intermediate I in the P450nor enzymatic reaction, which would provide direct experimental evidence to establish the molecular mechanism.

Results and Discussion

TR-Visible Analysis on the Reaction Kinetics.

To spectrophotometrically follow the NO reduction reaction of P450nor, an original microflow flash system was utilized in combination with caged NO photolysis (3). In this experimental system, the sample containing P450nor (0.4 mM), the caged NO that can provide ∼2 equivalents of NO, and NADH (5 mM) were continuously flowed for sample exchange, and the enzymatic reaction could be tracked up to 10 ms. Using this system, visible absorption spectral changes of P450nor were followed upon the caged NO photolysis at pH 6 and 20 °C. Fig. 1A shows the TR difference spectra thus observed. The spectra exhibited a positive Soret difference at 435 nm on the submillisecond time scale, which was shifted to 446 nm on the millisecond time scale. The global-fitting analysis of these spectral changes showed the formation/decay of two intermediates; the time constant (τ) of formation of the 435-nm species was 60 µs, and τ of its decay and the formation of the 446-nm species was 1 ms (Fig. 1B). The 435- and 446-nm species were already observed in our previous stopped-flow analysis and assigned to the Fe³⁺–NO state and I, respectively (6), indicating that this original experimental system is feasible for detecting I of P450nor.

Fig. 1.

TR-visible absorption spectroscopy of NO reduction reaction by P450nor. (A) TR difference spectra of P450nor after caged NO photolysis. The difference was calculated by subtracting the spectrum of resting P450nor recorded before the photolysis. (B) Amplitude spectra obtained by global-fitting analysis of the TR difference spectra. The 435- and 446-nm species are assigned to the Fe³⁺–NO complex and I, respectively. From the 435-nm peak intensity in the amplitude spectrum, the reaction yield was estimated to be ∼90%.

TR-IR Analysis on the Short-Lived Intermediate(s).

To characterize the chemical structure of the short-lived intermediate(s) of P450nor, TR-IR spectroscopy was performed using the same microflow flash system and conditions as in the TR-visible spectroscopy. Fig. 2 shows the TR-IR difference spectra of P450nor upon caged NO photolysis. Here, to measure the N–O stretching frequency (ν_NO) of the heme-bound NO, caged ¹⁴NO and caged ¹⁵NO compounds were used, and the net difference that highlights only ¹⁵N-isotope–sensitive bands arising from the ligand NO moiety was calculated; that is, spectral differences were calculated in three steps: 1) a UV-illuminated TR spectrum minus a UV-unilluminated dark spectrum (TR difference calculation), 2) an ¹⁴N-spectrum minus an ¹⁵N-spectrum (isotopic shift detection), and 3) a spectrum using the sample containing P450nor with caged NO and NADH minus a spectrum using the sample containing caged NO and NADH only (subtraction of caged NO signals). A detailed procedure is described in SI Appendix, Fig. S2. In Fig. 2 A and B, the TR-IR spectra in the high-frequency (1,760 to 1,890 cm⁻¹) and low-frequency (1,220 to 1,380 cm⁻¹) regions are illustrated, respectively. There is no additional ¹⁵N-isotope–sensitive signal within the 1,060- to 1,220-cm⁻¹ spectral window monitored in the present experiments (SI Appendix, Fig. S3).

Fig. 2.

TR-IR difference spectra of P450nor after caged NO photolysis in high-frequency (A) and low-frequency (B) regions. (C) Gaussian-fitting analysis of the spectra in the low-frequency region. Red and black traces are the measured and fitted spectra, respectively. The spectra contain two sets of the peak (¹⁴NO)/trough (¹⁵NO) components, as shown in blue and green traces. (D) Time traces of the band intensities at 1,330 cm⁻¹ (red circles) and 1,290 cm⁻¹ (blue circles). Black lines represent the global fitting of the time traces. (E) TR-IR difference spectra in the N₂O band region. (F) Static FTIR spectrum for the final state of P450nor after the reaction.

In the high-frequency region (Fig. 2A), the peak (¹⁴NO)/trough (¹⁵NO) arising from the Fe³⁺–NO complex was observed at 1,852/1,814 cm⁻¹ (20) at 0.25 ms and decayed with τ of 0.94 ms, consistent with the TR-visible result (1 ms). Corresponding to the decay of the signal from Fe³⁺–NO species, new ¹⁵NO isotope–sensitive bands around 1,300 cm⁻¹ concomitantly appeared and increased in their intensities (Fig. 2B). The Gaussian-fitting analysis revealed that the spectra can be comprised of two sets of the peak (¹⁴NO)/trough (¹⁵NO) components: 1,330/1,307 and 1,290/1,267 cm⁻¹ (Fig. 2C). The time traces of the band intensities at 1,330 and 1,290 cm⁻¹ are shown in Fig. 2D. The 1,330-cm⁻¹ band appeared with τ of 0.93 ms, comparable with the τ of the decay of the Fe³⁺–NO complex obtained from the TR-visible and IR-spectral changes (Figs. 1 and 2A). Then, the 1,290-cm⁻¹ band appeared with τ of 16 ms, concomitant with the disappearance of the 1,330-cm⁻¹ band. As this τ was outside the detectable time window of our TR-IR system (10 ms), only an early phase of the conversion from 1,330- to 1,290-cm⁻¹ species was detected here. For further characterization of the 1,330- and 1,290-cm⁻¹ species, a TR-IR experiment was conducted with a low NO concentration (one equivalent per P450nor), and the result is provided in SI Appendix, Text S1 and Fig. S4A.

By contrast, there was no N–N stretching band of the N₂O product in the TR-IR spectra within the 10-ms time frame (Fig. 2E), whereas it was observed at 2,230 cm⁻¹ in a static FTIR spectrum for the final state of P450nor after the reaction upon continuous UV illumination (Fig. 2F). These observations suggest that both the 1,330- and 1,290-cm⁻¹ species would be intermediates formed after the reaction of the Fe³⁺–NO complex with NADH and before N₂O production. The two species were indistinguishable in the TR-visible spectrum of I (Fig. 1), and thus, most probably, the same chemical species with respect to the electronic structure of the heme Fe–NO moiety. Here, the two species are tentatively designated as I₁ (ν_NO: 1,330 cm⁻¹) and I₂ (ν_NO: 1,290 cm⁻¹). Because the decay of the Fe³⁺–NO complex is not concomitant with the rise of I₂, direct transition from the Fe³⁺–NO complex to I₂ is inconceivable. Noting the appearance and disappearance of I₁ and I₂, it seems reasonable to describe that I₁ is an initial product upon the H⁻ transfer from NADH, and then, it is converted into I₂. Previously, it was proposed that NAD⁺ is released from the protein as a subsequent step of the H⁻ transfer from NADH to the Fe³⁺–NO complex (7). Thus, it is assumed that NAD⁺ might be present in the active site of I₁, and the dissociation of NAD⁺ would give rise to I₂. To examine this possibility, a TR-IR experiment was performed under excess NAD⁺. Although the τ of I₁ formation was not changed by the presence of NAD⁺, I₂ formation was apparently delayed (SI Appendix, Fig. S4B), which strongly suggests the involvement of NAD⁺ release in the transition from I₁ to I₂. Moreover, the dissociation time constant of NAD⁺ from P450nor in the resting form, analyzed by the stopped-flow technique, was estimated to be 26 ± 14 ms (SI Appendix, Fig. S5), which is comparable with the conversion time from I₁ to I₂ (∼16 ms). Therefore, it is most likely that I₁ and I₂ are short-lived intermediates generated from the Fe³⁺–NO of P450nor via the H⁻ transfer from NADH, but NAD⁺ is present in I₁ and absent in I₂.

To further confirm the above-stated assignment, the deuterium effect on the TR-IR spectra was investigated using deuterated [4,4-D,D]-NADH in H₂O and D₂O (SI Appendix, Figs. S6 and S7). By the deuteration of NADH, I₁ formation at 1,330 cm⁻¹ was slowed down by a factor of ∼4, and I₂ was observed after I₁ was accumulated. For both species, an apparent deuterium peak shift was not observed within the current frequency resolution (6 cm⁻¹).

XFEL-Crystallographic Analysis on the Short-Lived Intermediate.

We then tried to determine the coordination structure of the short-lived intermediate(s) in the radiation damage-free form (21) by XFEL-based crystallography using P450nor microcrystals. Before the XFEL experiment, the reaction kinetics in the microcrystals was analyzed by TR-visible microspectroscopy. As reported in our previous paper (15), the appearance of the 450-nm Soret band (corresponding to the 446-nm band in solution [i.e., formation of the short-lived intermediate I]) was significantly delayed in crystallo and observed 1 s after the photolysis of caged NO (SI Appendix, Fig. S8A), because of crystal-packing contacts at the entrance of the NADH channel. The formation of the short-lived intermediate(s) in the microcrystals was also followed using rapid-scan TR-FTIR microspectroscopy with a frequency resolution of 8 cm⁻¹ on the second time scale after caged NO photolysis. Only the ¹⁵NO isotope–sensitive band at 1,298 cm⁻¹ was observed in the net difference spectrum (SI Appendix, Fig. S8B), indicating that I₂, the NAD⁺-free intermediate, accumulated in the microcrystals, although the difference of this ν_NO value from that in solution (1,290 cm⁻¹) would probably be due to the crystal-packing effect.

Based on these in crystallo spectroscopic analyses, the coordination structure of the Fe–N–O moiety of I₂ was characterized crystallographically. The ferric P450nor crystals soaked with caged NO and NADH were illuminated by UV light and then flash-frozen 5 s after illumination to trap I₂. Structural data were collected with the fixed-target SFX technique (22). Fig. 3A shows the crystal structure of I₂ at 1.8 Å resolution. The occupancy of I₂ was 40%, whereas the resting state remained at 60% occupancy. As expected, no electron density assignable to NAD⁺ was observed close to the Fe–NO moiety of I₂. The structure of the Fe³⁺–NO complex determined previously by TR-SFX is also shown for comparison (Fig. 3B). The slightly bent Fe–N–O coordination (Fe–NO = 1.67 Å, N–O = 1.15 Å, Fe–N–O = 158°) in the Fe³⁺–NO state (15) is more significantly bent in I₂ (Fe–NO = 1.91 Å, N–O = 1.27 Å, Fe–N–O = 138°). We also tried to determine the structure of the Fe²⁺–NO complex, but the crystals of this complex were too unstable to be prepared with dithionite reduction. Therefore, a crystal structure of the radiation-damaged Fe³⁺–NO complex, which could be in a photo-reduced ferrous NO state (15), is shown in Fig. 3C. The NO coordination geometry in this state (Fe–NO = 1.68 Å, N–O = 1.19 Å, Fe–N–O = 147°) is in between those of the Fe³⁺–NO state and I₂. Taken together, it is likely that the reduction of the Fe³⁺–NO moiety is accompanied by the bending of the Fe–N–O angle, with slight lengthening of the Fe–N and N–O bonds. In the protein moiety, there was no significant structural difference between the Fe³⁺–NO state and I₂.

Fig. 3.

Crystal structures of P450nor. (A) Freeze-trapped I₂ obtained by fixed-target SFX. (B) Fe³⁺–NO complex obtained by TR-SFX (PDB No.: 5Y5H). (C) Radiation-damaged Fe³⁺–NO complex obtained using SPring-8 with an X-ray dose of 0.72 MGy (PDB No.: 5Y5F). The F_o–F_c maps are shown in turquoise and contoured at 3.5σ in A and 7.5σ in B and C. The 2F_o–F_c maps are shown in gray and contoured at 1.4σ.

QM/MM Analysis on the Short-Lived Intermediate(s).

As we successfully obtained the experimental data of the Fe–N–O characters in I₂ [d(Fe–NO), ∠Fe–N–O, d(N–O), and ν_NO], it is possible to discuss its electronic structure by directly (explicitly) comparing them to theoretically calculated data. Concerning I₁, it is presently impossible to obtain its crystallographic structural data because of the very slow formation of the short-lived intermediates in the crystal form. Although the presence or absence of NAD⁺ might affect the Fe–N–O coordination sterically or electronically, we could expect that the electronic structure of the Fe–N–O moiety is basically similar between I₁ and I₂, as their spectroscopic data were not drastically different from each other. One possible explanation for the difference of ν_NO between I₁ and I₂ is provided in SI Appendix, Text S2.

For the theoretical calculation of the singly protonated models (Fe²⁺–NHO and Fe³⁺–NHO^•−) and doubly protonated ones (Fe³⁺–NHOH^• and Fe⁴⁺–NHOH⁻), which have been proposed as possible short-lived intermediates of P450nor so far, we first constructed the initial model structure of I using the crystal structure of I₂, in which a heme, the nitroxyl ligand, and the side chain of Cys352 were included in the QM region, whereas the remaining protein portions were in the MM region (SI Appendix, Fig. S9). After geometry optimization was performed for the QM region, the ν_NO values were calculated for the four possible electronic structures (SI Appendix, Table S1). As shown in SI Appendix, Table S1, the observed ν_NO values for I₁ (1,330 cm⁻¹) and I₂ (1,290 cm⁻¹) are similar to the calculated value (1,324 cm⁻¹) for Fe³⁺–NHO^•– rather than those for Fe²⁺–NHO (1,398 cm⁻¹) and doubly protonated states (1,127 and 1,042 cm⁻¹). Thus, the short-lived intermediate I in the reaction of the Fe³⁺–NO of P450nor with NADH, including both I₁ and I₂, is in a singly protonated state and possibly the Fe³⁺–NHO^•− species. For a further discussion on this possibility, the proton affinity of the Fe³⁺–NHO^•− species was estimated and is provided in SI Appendix, Text S3.

With respect to Fe³⁺–NHO^•− species, we also calculated the deuterium shift of ν_NO, which exhibited a 3-cm⁻¹ upshift upon deuteration of water and/or NADH (SI Appendix, Fig. S10). The calculation indicated that the N–O stretching is accompanied by a motion of the nitroxyl hydrogen for Fe³⁺–NHO^•−, while not so for Fe³⁺–NDO^•−. Such a deuteration effect consequently caused a slight increase in the force constant of the N–O stretching mode, but the resulting upshift of ν_NO is predicted to be within the resolution of the TR-IR measurements (6 cm⁻¹). Thus, the observation on the deuteration effect in the TR-IR experiment is consistent with the assignment of Fe³⁺–NHO^•−.

Reaction Scheme of P450nor.

Based on the present results, the reaction scheme of P450nor can be described as shown in Fig. 4. The NO-bound state (Fe³⁺–NO) is reduced with two electrons by NADH to transiently generate short-lived I₁ with NAD⁺ present near the heme pocket. I₁ is then converted to I₂ upon NAD⁺ dissociation. The electronic structures of I₁ (ν_NO = 1,330 cm⁻¹) and I₂ (ν_NO = 1,290 cm⁻¹) are both characterized by the Fe³⁺–NHO^•− state. Importantly, the nitroxyl moiety in the Fe³⁺–NHO^•− state has a radical character with the spin densities as depicted in SI Appendix, Fig. S11. It is known that the electronic structures of heme NO species can be well characterized by the ν_NO values, correlated with the N–O bond lengths (SI Appendix, Fig. S12 and Table S4) (23), and for the ferric heme nitroxyl, the ν_NO value of NHO moiety reflects the spin population in the antibonding NHO π* orbital (24). The ν_NO value is lower, as the NHO π* orbital is more populated. Thus, the lower ν_NO of I₂, relative to I₁, indicates a stronger π-back bonding from the Fe d_π to NHO π* orbitals in I₂. Thus, I₂ generated upon NAD⁺ dissociation is more reactive with a larger spin density in the NHO π* orbital by a stronger π-back bonding (lower ν_NO).

Fig. 4.

Reaction scheme of P450nor.

This radical character of I₂ would be important to facilitate N–N bond formation with a second NO in a radical–radical coupling manner (10, 12). The generated hyponitrite (O = N–NHO⁻) is then decomposed, via tautomerization (⁻ON = NOH) and protonation processes, to produce N₂O (10), in which the central N and O atoms are suggested to derive from second NO molecule (14). Here, it should be stressed that I₂ is a singly protonated form but not a doubly one. This implies that a proton would be involved in the decomposition of the hyponitrite (i.e., N–O bond cleavage) after the binding of the second NO. Alternatively, it is also conceivable that N–N bond formation by the second NO binding could be coupled with proton transfer, as we proposed for bacterial NOR (3).

In conclusion, this study applied the TR-spectroscopic and -crystallographic techniques with caged NO to analyze the enzymatic reaction of P450nor and successfully characterized the coordination and electronic structures of the key intermediate I, which has been a long-standing subject of the N₂O generation mechanism by this enzyme. The QM/MM analysis combined with the spectroscopic and crystallographic results highlighted the radical nature of I as the singly protonated Fe³⁺–NHO^•− species, demonstrating the radical–radical coupling mechanism for N–N bond formation. The importance of the radical nature of I has been predicted by theoretical and model compound work (10, 12), but this study provided its experimental basis with the direct observation of I. It is known that the nitroxyl adduct of myoglobin with His ligation adopts a very stable Fe²⁺–NHO state (25). By contrast, I of P450nor is short lived and has clearly a different electronic structure from it. The formation of the Fe³⁺–NHO^•− radical in P450nor is attributable to the push effect of the Cys ligand, which may be a reason why a P450-type enzyme can catalyze the NO reduction reaction, besides the ability of NADH accommodation.

Materials and Methods

TR-visible and TR-IR spectroscopic measurements were performed using the microflow flash system. X-ray diffraction experiments were conducted using SPring-8 Angstrom Compact free-electron LAser (SACLA). The details, including sample preparation and QM/MM calculations, are provided in SI Appendix, SI Materials and Methods.

Data Availability

The atomic coordinates and structure factors have been deposited in the Protein Data Bank (PDB) (http://www.wwpdb.org, PDB ID code 7DVO). X-ray diffraction images have been deposited in the Coherent X-Ray Imaging Data Bank (https://cxidb.org/id-179.html). All other data are included in the article and/or supporting information.