Spectroscopic characterization of heme iron–nitrosyl species and their role in NO reductase mechanisms in diiron proteins

Abstract

Nitric oxide (NO) plays an important role in cell signalling and in the mammalian immune response to infection. On its own, NO is a relatively inert radical, and when it is used as a signalling molecule, its concentration remains within the picomolar range. However, at infection sites, the NO concentration can reach the micromolar range, and reactions with other radical species and transition metals lead to a broad toxicity. Under aerobic conditions, microorganisms cope with this nitrosative stress by oxidizing NO to nitrate (NO₃⁻). Microbial hemoglobins play an essential role in this NO-detoxifying process. Under anaerobic conditions, detoxification occurs via a 2-electron reduction of two NO molecules to N₂O. In many bacteria and archaea, this NO-reductase reaction is catalyzed by diiron proteins. Despite the importance of this reaction in providing microorganisms with a resistance to the mammalian immune response, its mechanism remains ill-defined. Because NO is an obligatory intermediate of the denitrification pathway, respiratory NO reductases also provide resistance to toxic concentrations of NO. This family of enzymes is the focus of this review. Respiratory NO reductases are integral membrane protein complexes that contain a norB subunit evolutionarily related to subunit I of cytochrome c oxidase (CcO). NorB anchors one high-spin heme b₃ and one non-heme iron known as Fe_B, i.e., analogous to Cu_B in CcO. A second group of diiron proteins with NO-reductase activity is comprised of the large family of soluble flavoprotein A found in strict and facultative anaerobic bacteria and archaea. These soluble detoxifying NO reductases contain a non-heme diiron cluster with a Fe–Fe distance of 3.4 Å and are only briefly mentioned here as a promising field of research. This article describes possible mechanisms of NO reduction to N₂O in denitrifying NO-reductase (NOR) proteins and critically reviews recent experimental results. Relevant theoretical model calculations and spectroscopic studies of the NO-reductase reaction in heme/copper terminal oxidases are also overviewed.

1 Nitric oxide reduction in biology

Hemoproteins play a dominant role throughout the chemistry of NO in biology. The production of NO from arginine by NO synthase depends on a heme cofactor, and regardless of whether NO is utilized as a secondary messenger, the regulatory domain of its target enzyme, guanylyl cyclase, also relies on a heme cofactor. Heme cofactors are equally crucial to the microbial response to toxic concentrations of NO by flavohemoglobins and other hemoglobin-like proteins under aerobic conditions. Here, we focus on a third branch of NO chemistry in biology: the reduction of NO to N₂O by hemoproteins (eqn 1).

$$2\mathrm{NO}+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\to {\mathrm{N}}_2\mathrm{O}+{\mathrm{H}}_2\mathrm{O}$$ (1)

The reactions by which NO is produced by reduction of nitrite (NO₂⁻) and consumed by reduction to nitrous oxide (N₂O) are essential steps of the denitrification pathway. Respiratory NO reductases in true denitrifying organisms and in partial denitrifiers like the facultative anaerobes Neisseria gonorrhoeae and Neisseria meningitidis, provide these organisms with added resistance to toxic concentrations of NO.^1-3

Denitrifying NO reductases (NOR) are integral membrane protein complexes for which there are no crystal structures yet available. The major subunit common to all denitrifying NORs, norB, is evolutionarily related to subunit I of cytochrome c oxidase (CcO). Specifically, the hydropathy pattern of CcO and the six histidine ligands of heme a, heme a₃, and Cu_B appear to be conserved in NORs.^4-6 Purification of NO reductases from a variety of denitrifying bacteria have confirmed that the NorB subunit (53 kDa) anchors one high-spin heme b₃ and one non-heme iron now known as Fe_B, in analogy with Cu_B in CcO.^5-9 Similar to the terminal oxidase superfamily, the NOR family exhibits variability in the electron-entry portion of the enzyme (Fig. 1).

Fig. 1.

Schematic representations of 3 subfamilies of NO reductases, and the X-ray structure of T. thermophilus cytochrome ba₃ (based on coordinates from ref. 83).

The most extensively characterized subfamily of denitrifying NO reductases contains an additional NorC subunit (17 kDa), which binds a heme c; its members are referred to as cNORs.^8,10-12 NO reductases from Ralstonia eutropha and from the archaeon Pyrobaculum aerophilum are purified as single-component enzymes, which include quinol binding sites fused to their NorB subunits (84 kDa),^13,14 and are thus referred to as qNORs. Most strikingly, in Bacillus azotoformans (B.a. qCu_ANOR) two branches of electron acceptors coexist, a menaquinol binding site on NorB and a dinuclear copper site, Cu_A, on subunit II.^15-17

In contrast to terminal oxidases, which couple the reduction of O₂ with translocation of protons across the membrane,¹⁸ the catalytic cycle of cNOR is non-electrogenic.^19,20 Consequently, protons are supplied on the periplasmic side of the membrane, as are the electrons that originate from soluble and membrane-bound cytochrome c. Fully-reduced cNORs can complete two NO reductase catalytic cycles and have also been shown to reduce O₂ to H₂O.^21-23 UV-vis monitoring of the reaction of O₂ with fully-reduced cNOR reveals two major phases: one is assigned to O₂-binding to heme b₃ (k = 2 × 10⁷ M⁻¹ s⁻¹), and the second corresponds to the oxidation of the low-spin heme b and c (k = 40 s⁻¹), a process which is coupled with proton uptake.^21,23

By analogy with the mechanism of O₂ reduction in CcO and NOR, it is certainly reasonable to assume that NO binds at the dinuclear site composed by heme b₃ and Fe_B, and evidence to support this view continues to accumulate. Furthermore, it is possible to envisage several possible catalytic mechanisms. After a brief description of spectroscopic signatures of iron–nitrosyl species and of putative models for the NO reductase catalytic cycle, the most pertinent spectroscopic information obtained to date and its interpretation in terms of catalytic relevance will be reviewed.

2 Spectroscopic signatures of heme iron–nitrosyl complexes

In contrast to CO and O₂, NO coordinates both iron(ii) and iron(iii) species. In Enemark and Feltham formulism,²⁴ which counts metal d-electrons plus the lone π* electron from the nitrosyl ligand, iron(ii)–NO and iron(iii)–NO complexes are represented as {FeNO}⁷ and {FeNO}⁶ species, respectively. Crystal structures of metal–nitrosyl porphyrin complexes have shown that {MNO}⁶ species adopt a linear M–N–O structure, while {MNO}⁷ species have bent M–N–O configurations.^24-26 Ferrous hemoproteins typically exhibit a very high affinity for NO while the ferric forms bind NO with K_d values in the micromolar range.^27-30 Because the concentration of NO during denitrification remains well below the micromolar range, it is assumed that the catalytic cycle of NO reductases is initiated by the binding of NO to an iron(ii) in the reduced form of the enzyme.

Heme {FeNO}⁷ species are S = 1/2 species that can be reliably characterized by a combination of UV-vis, EPR and vibrational (resonance Raman (RR) and FTIR) spectroscopies. Specifically, 6-coordinate heme {FeNO}⁷ species display Soret absorbance at ca. 420 nm while 5-coordinate species exhibit blue-shifted Soret absorbance near 400 nm.^31-36 EPR signals for both coordination numbers are centered around g = 2.0, but show distinct superhyperfine splitting: a 3-line hyperfine structure is typically associated with a 5-coordinate heme–nitrosyl, whereas the added N-ligand of a proximal histidine in the 6-coordinated species induces further splitting to produce a 9-line hyperfine structure.^37,38 However, defining the coordination number of a {FeNO}⁷ species solely from EPR spectra may not be entirely foolproof. Indeed, in a recent study of NO binding to reduced heme/copper oxidase aa₃ from Paracoccus denitrificans, the characteristic 3-line hyperfine structure was not assigned to a 5-coordinate heme a₃ {FeNO}⁷ species, but rather to a 6-coordinate complex. In this case, the orientation of the Fe–N–O plane, with respect to the trans-imidazole plane, limits the involvement of the imidazole nitrogen p-orbital in the π-molecular orbital occupied by the unpaired electron, thus preventing further splitting of the 3-line hyperfine structure.³⁹

In RR spectra, 6-coordinate {FeNO}⁷ species display ν₃ and ν₁₀ porphyrin core modes at 1500 and 1630 cm⁻¹, respectively, while, in 5-coordinated species, these same modes are observed 5 to 10 cm⁻¹ higher.^36,40-42 The ν(Fe–NO) and ν(NO) modes can be identified with isotopic labeling and provide further confirmation of the coordination number of the heme iron. Specifically, the proximal ligand trans to the nitrosyl has been shown to intensify the back-donation of Fe d_π electrons into the nitrosyl π* orbital, thus resulting in higher ν(Fe–NO) and lower ν(NO) for 6-coordinate species.⁴³ The distal pocket environment further modulates the extent of π-backbonding within the FeNO unit. While there are only a few fully characterized 6-coordinate {FeNO} models,^25,26 extensive studies of carbonyl synthetic models have led to a convincing description of distal perturbations on π-backbonding in heme Fe–CO complexes.^44-46 In hemoproteins, the control of the distal pocket environment on the extent of backbonding is equivalent in 6-coordinate {FeNO}⁷ species and Fe–CO complexes, as demonstrated by linear correlations between ν(NO)s and ν(CO)s, with slopes near unity for a series of distal pocket variants in myoglobin.^43,47,48 As is known for carbonyl complexes, polar environments and hydrogen-bond donation to the nitrosyl group have been shown to promote greater backbonding by stabilizing electron density on NO, thus leading to lower ν(NO) frequencies. The analysis of low-frequency Fe–N–O vibrational modes are more elusive since the intrinsically bent geometry of the {FeNO}⁷ unit provides greater flexibility to the nitrosyl group and results in extensive mixing of the ν(Fe–NO) and δ(Fe–N–O) modes.⁴⁹

Heme {FeNO}⁶ species can be viewed as S = 0 low-spin iron(ii)–NO⁺ complexes that are isoelectronic with iron(ii)–CO complexes. Accordingly, these species are EPR silent, and the Fe–N–O unit is usually linear with ν(NO) > 1900 cm⁻¹. Both 6- and 5-coordinate {FeNO}⁶ porphyrinate models have been characterized, but in proteins, only 6-coordinate species are known.^25,26 Electronic absorption spectra of {FeNO}⁶ in hemoproteins can be difficult to differentiate from those of their {FeNO}⁷ counterparts, and are sometimes strikingly similar.⁵⁰ Porphyrin skeletal vibrations observed in the high-frequency range of RR spectra, obtained with Soret excitations support a low-spin description of {FeNO}⁶ complexes.^30,50-54 The ν₃ and ν₁₀ porphyrin core modes are observed at higher frequencies in {FeNO}⁶ than in 6-coordinate {FeNO}⁷ species.^30,50-54 Importantly, vibrational modes of the FeNO unit in {FeNO}⁶ species do not seem to correlate with those of {FeNO}⁷ and carbonyl complexes. In fact, ferric–nitrosyl complexes in P450cam, with various substrates bound in the distal pocket, reveal a direct correlation between ν(Fe–NO) and ν(NO) rather than the inverse correlation observed between ν(Fe–XO) and ν(X–O) (i.e., where X = N or C) in {FeNO}⁷ and carbonyl complexes.^49,53 Recently, Rodgers and coworkers extended these studies of P450-like enzymes to include other porphyrin {FeNO}⁶ systems using density function theory calculations.^55,56 Their results provide descriptions of higher-occupied molecular orbitals and their bonding characters, with respect to the Fe–NO and N–O bonds, in an effort to define relationships between observed vibrational frequencies, electronic structures, and reactivity.

3 Putative reaction mechanisms

The catalytic mechanism of NO reduction in dinuclear proteins can be divided into four steps: the initial coordination of NO at the diiron active site, the formation of an N–N bond, the cleavage of an N–O bond and the release of N₂O. In terms of the first two steps, three possible mechanisms have been proposed (Fig. 2).

Fig. 2.

Putative mechanisms of NO reductase.

3.1 trans-Mechanism

The catalytic cycle of this mechanistic model is initiated by the binding of two NO molecules at the diiron(ii) cluster to form an iron–nitrosyl dimer [{FeNO}⁷]₂ intermediate. The close proximity of the two metal–nitrosyl species promotes N–N bond formation either via an electrophilic attack by one nitrosyl at the nitrogen lone pair of the second nitrosyl, or via a radical coupling process, where the two nitrosyls combine to form a metal bound hyponitrite species (N₂O₂ ²⁻). Several experimental observations support a trans-mechanism, including the characterization of an (FeCO)₂ heme/non-heme dicarbonyl complex at the active site of B.a. qCu_ANOR.⁵⁷ Also, rapid-freeze quench samples from single turnover reactions in cNOR have revealed EPR signals consistent with the trapping of S = 1/2 low-spin heme {FeNO}⁷ and S = 3/2 non-heme {FeNO}⁷ species.⁵⁸

3.2 cis–Fe_B mechanism

This mechanism is characterized by the binding of two NO molecules at the non-heme iron center to form a dinitrosyl complex.⁵⁹ Dinitrosyl iron complexes exhibit isotropic g = 2.0 EPR signals and can be described as {Fe(NO)₂}⁹ species.^60,61 In this mechanistic model, the role of the heme iron would be limited to electron transfer, although interactions with the putative hyponitrite complex may participate in the N–O bond cleavage. The cis mechanism has been strongly favored by Thomson and coworkers because it does not involve the formation of a heme b {FeNO}⁷ species, which they view as a potential dead-end product.^62-65 Indeed, in many hemoproteins, ferrous–nitrosyl complexes are unreactive species.^66-70 Redox titrations of cNOR by Thomson and coworkers have indicated that the redox potential of heme b₃ is 200 mV more negative than that of the non-heme iron Fe_B, suggesting that a three-electron reduced state, where heme b₃ remains oxidized, may be relevant to catalysis.^64,65

3.3 cis-Heme b₃ mechanism

In this mechanism, a heme b₃ {FeNO}⁷ species forms before reacting with a second NO molecule. The lone electron pair of the metal-bound nitrosyl is then the target of an electrophilic attack by a free NO to generate a hyponitrite radical. The reduction of the N₂O₂⁻ group to hyponitrite may occur subsequently to the formation of an asymmetrically-bridged diiron cluster. This reaction mechanism is exemplified in the Pd/Cu-catalyzed reduction of NO in aqueous solution where, in a rate limiting step, cuprous salt reduces the dinitrogen dioxide PdCl₃(N₂O₂)²⁻ to give PbCl₄⁻, N₂O, and water.⁷¹ The cis-heme b₃ mechanism is also reminiscent of flavohemoglobin, if the initial {FeNO}⁷ species in NOR is seen as analogous to the initial superoxo complex {FeO₂}⁸ in flavohemoglobin. This mechanism has been suggested for the NO reductase activity in terminal oxidases where a heme {FeNO}⁷ species accumulates in pseudo-steady state conditions.^72,73

The timing of the N–O bond cleavage and protonation events that follow the N–N bond formation remain to be defined. Isomerization of putative transient hyponitrites to an O-coordinated species to promote the formation of a bridging oxo group and release of N₂O may be energetically unfavorable.⁷³ However, an O-coordination of the hyponitrite dianion to Fe_B could lead to N–O bond cleavage and formation of a Fe_B(VI)=O.⁷⁴ Alternatively, protonation of the hyponitrite anion (or a hyponitrite radical precursor) by a catalytic acid could favor the heterolytic cleavage of the N–O bond.

4 Spectroscopic characterization of denitrifying NO reductases

4.1 The diiron cluster can accommodate an Fe–Fe distance ≤ 3.5 Å

As is often the case with heme-containing proteins, RR spectroscopy was instrumental in providing unequivocal support for the similarities between NO reductases and CcO and for the presence of a heme/non-heme dinuclear cluster at the active site of NO reductase. Similar to the heme a₃ of CcO, RR spectra of fully reduced P.d. cNOR defined the heme b₃ as a 5-coordinate high-spin heme. Moreover, the detection of an iron(ii)–N_His stretching mode at 218 cm⁻¹ identified the heme b₃ axial ligand as a neutral histidine side chain.¹¹ In the oxidized enzyme, the presence of a μ-oxo-group bridging the two iron(iii) centers was revealed by isotopic exchange with ¹⁸O-labeled water.⁷⁵ Comparison of the iron(iii)–O–iron(iii) stretching frequency observed in P.d. cNOR with other ν(Fe–O–Fe) vibrations in non-heme diiron models⁷⁶ and a synthetic heme/non-heme synthetic model ([(⁵L)Feⁱⁱⁱ–O–Feⁱⁱⁱ–Cl]⁺)⁷⁷ provided the first direct evidence of a diiron cluster with an Fe–Fe distance ≤ 3.5 Å (Fig. 3).⁷⁵ Because the heme iron is 5-coordinate in both oxidized and reduced cNOR, and because a ν(Fe–N_His) was identified only in the reduced enzyme,¹¹ we concluded that the μ-oxo bridge must be hydrolyzed upon reduction, allowing for the coordination of the proximal histidine to heme b₃ and leaving an open coordination site facing Fe_B.⁷⁵

Fig. 3.

Resonance Raman characterization of heme/non-heme μ-oxo bridged diiron(iii) cluster in oxidized cNOR (in ¹⁶OH₂ (blue), ¹⁸OH₂ (green) and difference spectrum (red)), and in a synthetic model in CH₃CN with 20% water (in ¹⁶OH₂ (blue), ¹⁸OH₂ (green) and difference spectrum (red)). The ν_as(Fe–O–Fe) modes are identified by their 40 cm⁻¹ downshift upon exchange of the O-atom with ¹⁸O-labeled water. The schematic structures of the diiron clusters in cNOR and the model compound are also shown (adapted from ref. 75).

The μ-oxo bridged diiron(iii) cluster observed in P.d. cNOR is associated with a ligand-to-metal charge transfer absorption band at 595 nm. A lack of variation in both this absorption band and spectral features in the RR spectra suggest that the cluster is insensitive to pH changes between 6.0 and 9.0.^75,78 Mediated redox potentiometry has shown that the heme b₃ presents an unexpected low midpoint potential (E_m = 60 mV), which allows for the formation of a three-electron reduced form of the enzyme where the heme b₃ remains as the sole iron(iii) species.⁶⁴ Room temperature absorption and MCD analyses of the three-electron reduced enzyme at different pH values suggest that, upon reduction of Fe_B, the μ-oxo bridge dissociates from Fe_B(ii) and the heme b₃ iron(iii) rebinds the proximal histidine to form a 6-coordinate aquo/hydroxo complex with charge transfer bands at 635 and 605 nm, respectively.⁷⁸ Not surprisingly, cyanide was shown to bind much more readily to heme b₃ in the three-electron reduced form of the enzyme, where the displacement of the sixth aquo/hydroxo ligand is facile (e.g., as in metmyoglobin), than in the fully oxidized enzyme, where the μ-oxo group bridge between the two iron(iii) is quite stable.⁶⁵ From these observations, Thomson and coworkers proposed that the μ-oxo bridged diiron(iii) cluster might represent a “closed” resting state of the enzyme, while the catalytic cycle may, in fact, be initiated by the binding of NO to the mixed valence heme b₃(iii)–Fe_B(ii) cluster.^64,65,78

Regardless of whether the diiron(iii) μ-oxo bridge cluster is part of the catalytic cycle of NO reductases, this structure reveals a metal–metal distance that is 1.5 Å shorter that that measured in the crystal structures of terminal oxidases between the iron of heme a₃ and Cu_B.^79-83 This significant distinction between the active sites of P.d. cNOR and CcO may explain their different reactivity toward NO. The iron–iron distance in NOR is more comparable with those measured for the carboxylate-bridged non-heme diiron clusters of the R2 subunit of E. coli ribonucleotide reductase, methane monooxygenase,^84-86 and flavoprotein A.⁸⁷ Interestingly, NO-reductase activity has been reported in all three of these diiron proteins.^88-90 The characterization of fully-oxidized and fully-reduced P.d. cNOR makes clear that the proximal histidine is labile and suggests that the catalytic intermediate might include a pentacoordinated heme {FeNO}⁷ species. Structural characterizations of iron(ii)–nitrosyl porphyrin complexes by Scheidt and coworkers have shown that, as the proximal ligand is released, the iron is pulled further out of the porphyrin plane by 0.2 Å, while the Fe–N and N–O distances and the Fe–N–O angle decrease only slightly.^25,26,91 The latter changes are consistent with a modest strengthening of both σ and π bonding in the FeNO unit. A recent crystal structure of nitrosyl–hemoglobin in the presence of the effector inositol hexaphosphate concurs with the iron displacement measured in model complexes.⁹² Specifically, in β-subunits where 6-coordinated {FeNO}⁷ species form, the iron remains 0.12 Å out of the porphyrin plane on the proximal side, however, in α-subunits where 5-coordinated {FeNO}⁷ species form, the iron is located 0.25 Å out of the porphyrin plane toward the distal side. At the dinuclear site of NO reductases, a 5-coordinate {FeNO}⁷ structure at the heme b₃ would decrease the distance between the nitrosyl group and a second {FeNO}⁷ species at Fe_B, a change that may favor radical coupling and N–N bond formation.

4.2 The diiron site can accommodate two CO molecules

Carbon monoxide (CO) is an ideal probe to investigate the binding of exogenous diatomic ligands at the dinuclear site of NO reductases. Heme iron–carbonyl complexes are stable and exhibit a characteristic ν(C–O) between 1940 and 1980 cm⁻¹ in RR and FTIR spectra.⁹³ Moreover, heme iron–carbonyls are photolabile and can be used to probe the heme distal pocket. In myoglobin, for example, the photolyzed CO can be trapped in several docking sites within the distal pocket.^94-96 These photolyzed CO molecules in proteinaceous docking sites display weak, but well-resolved, ν(C–O) near 2130 cm⁻¹ in low-temperature FTIR spectra.^97-99 In terminal oxidases, CO dissociates from heme a₃ and coordinates to Cu_B. The Cu_B–CO complex can be trapped at cryogenic temperatures, and ‘dark’ minus ‘illuminated’ FTIR difference spectra exhibit a positive ν(CO) from the a₃–CO complex and a negative ν(CO) between 2040–2070 cm⁻¹ characteristic of Cu_B–CO.^100-103

In 1998, Saraste and coworkers attempted to use this approach with P.d. cNOR and reported that FTIR difference spectra could only be observed between dark and continuous illumination at 234 K.¹⁰⁴ The asymmetric absorption changes, centered around 1970 cm⁻¹, were interpreted in terms of a ‘dark’ signal at 1977 cm⁻¹, assigned to heme b₃–CO, and a ‘light’ signal at 1963 cm⁻¹, assigned to Fe_B–CO. However, these absorption changes may, in fact, originate from variations in populations of heme a₃–CO conformers under continuous illumination rather than CO-coordination to Fe_B. FTIR difference spectra between P.d. cNOR–¹²CO and cNOR–¹³CO obtained at Oregon Health & Science University confirmed that only one CO molecule binds per diiron site.¹⁰⁵ ‘Dark’ minus ‘illuminated’ FTIR difference spectra at 15 K exhibit a ‘dark’ ν(CO) at 1972 cm⁻¹, assigned to heme b₃–CO, and an ‘illuminated’ ν(CO) signal at 2120 cm⁻¹ that is characteristic of non-coordinated CO docking at the hydrophobic site near the distal pocket.¹⁰⁵ Thus, from these FTIR experiments, it remains unclear whether CO can bind Fe_B following photolysis from heme b₃ in P.d. cNOR.

The CO photolysis in P.d. cNOR was also investigated at room temperature using time-resolved electronic absorption. These experiments revealed a very fast recombination rate (k_on = 1.7 × 10⁸ M⁻¹ s⁻¹), approximately three orders of magnitude faster than in terminal oxidases.¹⁰⁶ In contrast, when Watmough and collaborators measured the rate constant for the initial binding of CO by stopped-flow spectroscopy, they observed a much slower rate (k_on = 1.2 × 10⁵ M⁻¹ s⁻¹).¹⁰⁶ The authors suggested that binding of CO at the heme a₃ requires the initial displacement of a distal heme ligand. The slow rebinding of this ligand, compared to CO, would explain the photolysis kinetics.¹⁰⁶ However, RR studies identify the ferrous heme b₃ as a 5-coordinate high-spin species,^11,107 thus, the kinetic data may be consistent with the presence of a weakly interacting group that can adopt variable configurations rather than with the displacement of a true distal ligand of heme b₃ (vide infra).

In sharp contrast with P.d. cNOR–CO, recent experiments with the CO complex of qCu_ANOR show that two CO molecules bind concomitantly at the diiron site of qCu_ANOR.⁵⁷ Specifically, room temperature (¹²CO minus ¹³CO) FTIR difference spectra of qCu_ANOR–CO show two ν(CO) modes at 1972 and 2068 cm⁻¹ (Fig. 4). The band at 1972 cm⁻¹ (Δ¹³CO=−44 cm⁻¹) corresponds to a heme–CO and is nearly identical to that observed in the heme–CO complex of P.d. cNOR.^11,105,108 The ν(CO) at 2068 cm⁻¹ (Δ¹³CO = −47 cm⁻¹) is consistent with a non-heme iron–CO where the back-bonding donation from the iron d_π orbitals to the carbonyl π* orbitals is weakened compared to its heme–CO counterparts. The comparable integrated area of these two bands suggests that these two species are present in equivalent concentrations. These room temperature data may be interpreted in two ways: i) two CO molecules occupy the dinuclear active site, or ii) the active site contains only one CO molecule in a binding equilibrium between the heme and the non-heme irons. However, the comparison of the integrated areas of the room temperature signals with the total enzyme concentration favors the first alternative, and low-temperature photolysis experiments confirm this conclusion.⁵⁷

Fig. 4.

FTIR difference spectra of qCu_ANOR–CO. At room temperature (top traces), the ν(CO) are isolated using ¹²CO–¹³CO spectra (blue trace: no chloride added, green trace: 3 M chloride). At 15 K (bottom traces), ‘dark’ minus ‘illuminated’ spectra reveal photosensitive ν(CO) modes (black trace: ¹²CO; red trace: ¹³CO; and green trace: ¹²CO in the presence of 3 M NaCl). A schematic interpretation of the binding of CO in qCu_A NOR is also shown (adapted from ref. 57).

The ‘dark’ minus ‘illuminated’ FTIR difference spectrum of qCu_ANOR–CO obtained at 15 K (Fig. 4) shows a positive cluster of bands centered around 1972 cm⁻¹ that is assigned to comparable conformers of the photolabile heme–CO complex. A new band at 1910 cm⁻¹ (Δ¹³CO = −44 cm⁻¹), which is not observed at room temperature, is characteristic of a semi-bridging configuration between two metal ions.⁹³ Accompanying this band is an S-signal with a positive band at 2046 cm⁻¹ and a negative band at 2066 cm⁻¹ (Fig. 4). The influence of temperature on the FTIR difference spectra shows the interdependence of these two signals. Indeed, as the temperature is raised to ~100 K, the 1972 cm⁻¹ signal decreases as geminate rebinding begins to occur, but the 1910 cm⁻¹ band and the 2046/2066 cm⁻¹ S-signal are unchanged. We have assigned the S-signal to a downshift of the non-heme Fe_B–CO as it is perturbed by the photolysis of the semi-bridging heme–CO at 1910 cm⁻¹ (Fig. 4).⁵⁷ These results lend support to the trans-mechanistic model in which N–N bond formation is promoted by the binding of two NO molecules to form a [{FeNO}⁷]₂ transient species.

Upon addition of chloride to qCu_ANOR–CO, the binding of CO at the Fe_B site is inhibited (Fig. 4), and the FTIR characterization of qCu_ANOR–CO resembles that of P.d. cNOR–CO. Similarly, in the heme/non-heme synthetic model [(⁶L)–Feⁱⁱ … Feⁱⁱ–(Cl)]⁺, CO binding occurs at the heme iron, but the non-heme iron(ii) does not bind CO, at least in part because of the chloride coordination.¹⁰⁹ It remains unclear in the case of NOR, whether, chloride directly binds to Fe_B(ii) or whether its affect on Fe_B(ii) ligation is only allosteric, but control experiments suggest that different states of the diiron site cannot solely be explained by the effect of chloride.^105,110 Thus, in both NOR proteins, there is clear experimental evidence for the presence of a yet unidentified ligand(s) that can adopt variable coordination geometries. At this time, it is tempting to speculate that a glutamate side chain at the diiron site might provide coordination flexibility to this site. Several glutamate residues are conserved in norB while absent from subunit I in terminal oxidases.^4-6 In particular, conserved residues Glu198 and Glu202 (numbering from P. denitrificans norB) are located in transmembrane helix VI, which includes His194, a putative ligand to Fe_B. Moreover, the RR characterizations of the CO complexes of P.d. cNOR and qCuANOR suggest the presence of negatively charged residues in the distal pocket of heme b₃.^11,105

Richardson and coworkers have engineered expression systems from norB variants in P. denitrificans and E. coli and analyzed the NO reductase activity of membrane fragments and solubilized proteins.²² Three mutants of conserved glutamate residues were successfully expressed and purified: E198A and E202A, which are expected to be at or near the diiron active site, and E125A, which is predicted to be on the periplasmic surface. The location of Glu125, and the loss of NO reductase activity in E125A suggest that this residue may play a role in proton transfer from the periplasm to the active site. While E202A shows an NO-reductase activity comparable to the wild type protein, the E198A variant is inactive. These results suggest that E198 plays an important role at the diiron active site, either as a ligand to Fe_B or as a proton transfer group.

In non-heme diiron proteins, X-ray crystallography has revealed variations in the coordination of carboxylate groups that accompany changes in the oxidation state of the diiron cluster. For example, in the R2 subunit of E. coli ribonucleotide reductase and in methane monooxygenase, a coordinating glutamate side chain switches from bridging the two irons in the fully-reduced enzymes to terminally coordinating a single iron(iii) in the oxidized proteins.^111-116 Variations in carboxylate coordination geometry also take place without changes in iron oxidation states. For example, interactions of regulatory subunits such as MMOB with the hydroxylase protein of methane monooxygenase and acyl carrier protein with Δ⁹-desaturase affect the spectroscopic characteristics of the diiron(ii) cluster at the active site.^117-119 These observations led to the concept of carboxylate shifts controlling the number of open coordination sites and the range of Fe–Fe distances during catalytic turnover.¹²⁰ Thus, in NOR, where glutamate side chains in the vicinity of coordinating histidine residues are known to be conserved, it is conceivable that carboxylate shifts might control the coordination of diatomic molecules at the heme/non-heme diiron site.

4.3 Spectroscopic studies of NO reactions with NOR

Using rapid-freeze quenching, Shiro and coworkers have shown that NO and fully-reduced cNOR react in a sub-millisecond time scale, as evidenced by the appearance of new EPR signals within 0.5 ms after mixing.⁵⁸ One signal, at g = 4, is consistent with a non-heme S = 3/2 {FeNO}⁷ species, and another at g = 2.01, with a pronounced 3-line hyperfine structure, is characteristic of a heme S = 1/2 {FeNO}⁷ system. Shiro and coworkers estimate that these two nitrosyl signals represent 30% of the total diiron site concentration. In addition, they show that these species decay within 10 ms when their EPR signatures are replaced with those of ferric low-spin b and c hemes.⁵⁸ The data have been interpreted in terms of a trans-mechanism where a [{FeNO}⁷]₂ unit forms at the active site within the first ms after mixing.⁵⁸

The observation of a g = 4 signal from a non-heme {FeNO}⁷ species and a g = 2.0 signal from a heme {FeNO}⁷ upon addition of NO to fully-reduced cNOR had been reported earlier in hand-mixing experiments,¹⁰⁴ but their relevance to the native structure have been seriously questioned as these EPR signals appeared to increase under prolonged incubation.¹²¹ In contrast with hand-mixed experiments, the millisecond kinetics of the formation and decay of the EPR signals observed by Shiro and coworkers support the involvement of these species in the catalytic cycle of NOR.⁵⁸ The interpretation of these EPR data may be revisited however, as one might expect a heme/non-heme [{FeNO}⁷]₂ complex to be magnetically coupled to form an EPR-silent spin integer species (most likely antiferromagnetically coupled, i.e., an S = 1 system). For example, the major species formed by the reaction of NO with the reduced non-heme diiron site of protein R2 is an EPR-silent, antiferromagnetically-coupled [{FeNO}⁷]₂ complex.^88,122 Shiro and coworkers favor the formation of an [{FeNO}⁷]₂ unit on the basis that the g = 4 signal of the reaction intermediate behaves differently, with respect to microwave power saturation, than in the CO-inhibited species, where the S = 0 low-spin iron(ii) carbonyl complex does not promote relaxation.⁵⁸ However, an increase in relaxation would be expected whether or not NO binds to heme b₃ since both heme {FeNO}⁷ and high-spin iron(ii) are S ≠ 0 species. An alternative to the authors’ proposal may be that within the first millisecond after exposure of fully-reduced cNOR to NO a mixture of [heme b₃ Fe–NO/Fe_B] and [heme b₃ Fe/Fe_B–NO] complexes can be produced. Indeed, if the metal centers are magnetically uncoupled upon complexation with a single NO molecule, both S = 3/2 and S = 1/2 {FeNO}⁷ species are expected to contribute to the EPR spectra. This alternative interpretation would imply that the two iron(ii) ions at the active site pocket have comparable binding affinity for NO, and that these {FeNO}⁷ complexes can be trapped by rapid-freeze quenched techniques. A [{FeNO}⁷]₂ complex, on the other hand, might not be observable by EPR, or it might not accumulate in the course of a single turnover reaction. The concept of a mixture of {FeNO} complexes has support from the evolutionarily-related cytochrome ba₃ oxidase from T. thermophilus, from which FTIR experiments have revealed an [heme b₃ Fe–CO/Cu_B] ↔ [heme b₃ Fe/Cu_B–CO] equilibrium.¹²³ In addition, theoretical analyses of the in silico reduction of NO with simplified heme/copper and heme/non-heme dinuclear clusters calculate reaction-energy profiles where hyponitrite, rather than dinitrosyl complexes, are the observable intermediates.^73,74 Thus, two CO molecules bound at the diiron site of NOR can be observed as the complex is stable in qCu_ANOR,⁵⁷ but a [{FeNO}⁷]₂ transient might not be a traceable intermediate if it does not accumulate in the course of the reaction.

NO can also react with fully-oxidized P.d. NOR,^10,124 but this reaction only occurs at pH < 6, where acidic conditions presumably favor the hydrolysis and displacement of the μ-oxo bridge to allow NO coordination to heme b₃.¹²⁵ The heme b₃ {FeNO}⁶ species displays a ν(Fe–NO) at 594 cm⁻¹ and a ν(NO) at 1904 cm⁻¹.¹²⁴ These stretching frequencies are at the low range of those reported for 6-coordinate heme {FeNO}⁶ species with a proximal histidine ligand.^{30,50,51,54,126-128} Varotsis and coworkers¹²⁴ noted that the low ν(NO) contrasts with the high ν(CO) associated with inhibition of π-backbonding in the negative polarity of the distal pocket.¹¹ As discussed earlier (see section 2; spectroscopic signatures of heme iron–nitrosyl complexes), the influence of distal pocket environment on the characteristics of heme {FeNO}⁶ species remains poorly understood.^55,56 It is also unclear whether or not the heme b₃ {FeNO}⁶/Fe_B(iii) state is relevant to the catalytic cycle of NOR.

5 NO reduction in terminal oxidases

The various reactions of NO with heme/copper terminal oxidases and their importance to the regulation process of O₂-respiration have been recently reviewed.^129-132 Briefly, the interaction of NO with the oxidized enzyme inhibits mitochondrial CcO via binding of NO to Cu_B(ii).^133,134 The resulting copper–nitrosyl complex adopts a Cu_B(i)–NO⁺ configuration and produces nitrite upon hydration of the bound nitrosonium ion.^129,135-137 In contrast, reduced terminal oxidases bind NO in a competitive fashion with O₂ to form a heme a₃ {FeNO}⁷ complex.¹³⁸ In some terminal oxidases, experimental observations corroborate the concomitant binding of a second NO molecule at the Cu_B(i) site. For example, FTIR experiments by Caughey and coworkers on reduced bovine CcO exposed to NO revealed two ν(NO)s at 1610 and 1700 cm⁻¹, which were assigned to heme a₃-NO and Cu_B-NO, respectively.¹³⁹ In addition, optical experiments monitoring rebinding dynamics in the picosecond time scale for the NO complex of reduced aa₃ from P. denitrificans showed that geminate rebinding to the heme a₃ site is predominant at high NO concentration. This behavior was assigned to an obstruction of the exit route for the photolyzed NO by a second NO bound to Cu_B.³⁹ It is interesting to note that while these metal–nitrosyl Fe/Cu dimers in CcO and aa₃ mimic the putative [{FeNO}⁷]₂ intermediate in NOR, neither of these terminal oxidases show any NO reductase activity.

The fully-reduced caa₃ and ba₃ oxidases from Thermus thermophilus quickly react with NO to form a heme a₃ {FeNO}⁷ complex, which decays to produce N₂O with a turnover rate approximately 1% and 0.1% that of denitrifying NORs, respectively.^138,140 Presumably, binding of a second NO to Cu_B(i) is slow and results in a facile turnover to form N₂O.^39,72 Thus, binding of a second NO molecule may occur in all oxidases, but the only catalytically competent dinitrosyl complexes form in T. thermophilus oxidases. All fully-reduced terminal oxidases react with NO to form a heme {FeNO}⁷ complex that can be spectroscopically characterized.^130,131,141 While both 5-coordinate and 6-coordinate {FeNO}⁷ hemes have been reported in terminal oxidases,^39,72,142 the coordination number of the heme iron does not appear to correlate with NO reductase activity.^138,140

The heme iron–nitrosyl complex in T. thermophilus ba₃ has been studied by time-resolved absorption spectroscopy and displays rebinding dynamics on the 15 ns time scale at room temperature.³⁹ This relatively slow rebinding process suggests that the photolyzed-NO group binds transiently to another site, presumably Cu_B. Time-resolved step-scan FTIR spectra of the heme a₃–NO complex in T. thermophilus ba₃ after flash photolysis show no differential signals and indicate that rebinding occurs within the 5-μs time resolution of this technique.¹⁴³ Using density function theory calculations, and within the context of a trans-mechanism, Varotsis and coworkers have proposed that the formation of a [heme a₃-NO – Cu_B-NO] dinitrosyl complex leads to an N,N′-bridging hyponitrite intermediate.¹⁴⁴ Using the same theoretical approach, Blomberg and coworkers favor a cis-heme a₃ mechanism and a bridging coordination of the hyponitrite to form a 5-member ring with two O-atoms coordinated to Cu_B.⁷³ Although only a few transition metal hyponitrite complexes have been structurally characterized to date,^145-148 these complexes already demonstrate a wide diversity in coordination geometries. There is little doubt that the characterization of a hyponitrite intermediate in the reaction of fully-reduced ba₃ with NO would constitute a major breakthrough toward understanding these catalytic reactions.

6 Conclusions and perspectives

Structural and functional similarities between NO reductases and the heme/copper terminal oxidases are now firmly established, and in view of the existing literature on terminal oxidases, the existence of different forms of NORs is not surprising. Indeed, resting and pulsed/fast forms of oxidized and reduced CcO are well documented.¹⁴⁹ The effect of chloride on the family of terminal oxidases is notoriously complex and ranges from direct binding to cuprous Cu_B(i),^150-152 to bridging the heme a₃ iron(iii) and Cu_B(ii),^149,153 to binding at heterotropic sites.^149,154,155 With its poor NO reductase activity and access to its three-dimensional structure, T. thermophilus cytochrome ba₃ provides an opportunity to gather spectroscopic information on NO reaction intermediates. Nevertheless, because of their different metal composition, what is learned from heme/copper terminal oxidase systems will not always be applicable to NORs. A clear definition of the catalytic cycle of NOR will require the combination of rapid-freeze quenching and multiple spectroscopic techniques.

Several carboxylate-bridged non-heme diiron proteins catalyze the reduction of NO to N₂O, and the flavoprotein A family of enzymes has recently been exposed as detoxifying NO reductases.^90,156,157 A recombinant form of flavoprotein A from Moorella thermoacetica, which exhibits comparable NO reductase activity to denitrifying NORs, was recently crystallized.⁸⁷ In this protein, the active site contains a non-heme diiron cluster with two histidines per iron and two bridging ligands: a bidentate carboxylate group and a solvent molecule.⁸⁷ Carboxylate groups have been shown to provide structural flexibility at the active sites of non-heme diiron proteins (for an extreme example, see recent crystal structures of nigerythrin, which reveal a 2 Å movement of one iron associated with a Glu ↔ His ligand “toggling” at one iron¹⁵⁸). In denitrifying NOR, Fe_B may likewise recruit a glutamate side chain as a ligand, which, in turn, may allow the site to control the reactivity of {FeNO}⁷ species, regulate the catalytic cycle and prevent the formation of dead-end complexes through carboxylate shifts.

Biography

Pierre Moënne-Loccoz was born in France in 1964. He received his PhD in 1989 from Université Pierre et Marie Curie (Paris, France) for his work on Photosystem I and II of higher plants, under the supervision of Marc Lutz and Bruno Robert (CE Saclay, France). He continued to study photoactive proteins, first with Warner Peticolas at the University of Oregon and then with Mike Evans and Peter Heathcote at University College London. He then joined Thomas Loehr and Joann Sanders-Loehr at the Oregon Graduate Institute, where he started investigating O₂-activation processes in heme and non-heme diiron proteins. He is now an Assistant Professor at the OGI School of Engineering at OHSU where he continues to work on O₂ and NO chemistry in metalloproteins using a variety of spectroscopic techniques.