Molecular Understanding of Heteronuclear Active Sites in Heme-Copper Oxidases, Nitric Oxide Reductases, and Sulfite Reductases through Biomimetic Modelling

Abstract

Heme-copper oxidases (HCO), nitric oxide reductases (NOR), and sulfite reductases (SiR) catalyze the multi-electron and multi-proton reductions of O₂, NO, and SO₃²⁻, respectively. Each of these reactions is important to drive cellular energy production through respiratory metabolism and HCO, NOR, and SiR evolved to contain heteronuclear active sites containing heme/copper, heme/nonheme iron, and heme-[4Fe-4S] centers, respectively. The complexity of the structures and reactions of these native enzymes, along with their large sizes and/or membrane associations, make it challenging to fully understand the crucial structural features responsible for the catalytic properties of these active sites. In this review, we summarize progress that has been made to better understand these heteronuclear metalloenzymes at the molecular level though study of the native enzymes along with insights gained from biomimetic models comprising either small molecules or proteins. Further understanding the reaction selectivity of these enzymes is discussed through comparisons of their similar heteronuclear active sites, and we offer outlook for further investigations and areas of ongoing study.

1. Introduction

Respiratory metabolism is the cornerstone of cellular energy processes.^1–3 To drive cellular energy production, organisms utilize a wide variety of so-called “terminal electron acceptors”^4–8 that are often simple small molecules or ions that are prevalent throughout the organism’s immediate environment and possess thermodynamically favorable reduction potentials. For example, anaerobic respiration using sulfate (SO₄²⁻) as the terminal electron acceptor in oxidative phosphorylation can be traced to some of the earliest lineages of bacterial life.^9,10 These organisms, and extant sulfate-reducing organisms, release sulfide (H₂S) as a byproduct and play a crucial part in the biogeochemical cycle of sulfur.^1,10 Similarly, the globally-relevant process of respiratory denitrification uses nitrate (NO₃⁻) for this purpose, ultimately yielding gaseous nitrogen (N₂) and water.^11,12 Eventually, molecular oxygen (O₂), an energetically potent terminal electron acceptor, became sufficiently abundant on the planet to allow for the developing prevalence of cellular aerobic respiration—reducing O₂ to H₂O.^13,14 The greater amounts of energy that cellular respiration can provide to the organism is considered to be an underlying cause for the development of higher organisms.^15,16

While these small molecules are good terminal electron acceptors for cellular energy processes, efficiently reducing them can be challenging due to high energetic barriers for their activations. To address this issue, nature has recruited metal ions as cofactors because they possess tuneable thermodynamic reduction potentials and electron configurations to catalyze the activation of these small molecules that cannot be achieved easily with nonmetals. Therefore, metalloproteins are ubiquitous components of cellular respiratory pathways and serve roles both in electron transfer (ET) processes and as catalysts for bond-breaking/bond-forming chemical transformations. These metalloenzymes have evolved to efficiently catalyze small molecule redox transformations, in many cases with high degrees of energy conservation and selectivity while avoiding the release of toxic or reactive intermediates. As a result, many different metal ions and coordination environments have been discovered in these proteins to perform a variety of reactions involved in cellular respiration. Even though the building-blocks for the construction of these metalloproteins are limited to 20 natural amino acids, a handful of naturally-derived organic cofactors (e.g. porphyrin), and Earth abundant metal ions—considerably less diverse than the vast number of building-blocks which can be used to prepare synthetic inorganic and organometallic compounds—metalloenzymes display remarkably high activity and efficiency while operating under relatively mild reaction conditions. The protein structures that produce these remarkable metal catalysts, each specialized for a particular reaction, are an inspiration to chemists interested in the development of effective catalysts relevant to the current needs of society.

A type of reaction central to nearly all cellular respiratory pathways, including sulfate reduction, denitrification, and aerobic respiration, is multi-electron, multi-proton small molecule transformation. Most metalloenzymes that catalyze these reactions use multiple redox-active metal centers in close proximity within the enzyme active site (Figure 1). The precise reasons why these complex heteronuclear catalysts evolved for these reactions and how they carry out these complex reactions efficiently and selectively pose some of the greatest unknowns in biochemistry. Recent advances have begun to fill in many of the blanks through increasingly versatile and inexpensive molecular biology tools available today, along with the development and advancement of multiple computational, spectroscopic and crystallographic techniques, which have allowed more detailed studies of complex native enzymes than ever before. In parallel, numerous bio-inspired synthetic, small molecular model complexes, including functional catalysts, have been created to reconstruct the core features of these metalloenzymes. Their ability to go beyond the relatively confined space of biological coordination chemistry has allowed us to begin to understand the deeper underlying chemical principles at work within these complex metalloenzymes. In a complementary approach to studying native enzymes and their synthetic models, our lab and others have used small and robust proteins as scaffolds to design “biosynthetic” models, or artificial metalloenzymes (ArMs), to recreate the complex catalytic centers of native metalloenzymes from the bottom-up while preserving the advantages of facile design and stability inherent to biomimetic modelling.^17–21 Biosynthetic modelling of metalloproteins has also been pursued through new-to-nature de novo designed protein scaffolds, further expanding the structural diversity of protein-based models.^17,22–26 Both synthetic and biosynthetic models provide several opportunities to complement the study of native enzymes: targeted construction of minimal catalytic components have led to models that helped identify specific, local factors that lead to efficient bond activation and reactivity and these models have allowed for the isolation and characterization of catalytic intermediates whose presence in the enzyme mechanism is implied but has otherwise not been possible to directly interrogate in the native system. The variety of available functional groups, many of which are non-biological, has led to a number of probing, systematic studies. The approach to model native enzymes by redesigning smaller proteins has further provided experimental support for the catalytic roles of complex protein structures beyond the metal primary coordination sphere.

Figure 1.

Active site structures of heteronuclear metalloenzymes (A) NOR (PDB ID: 3O0R), (B) HCO (PDB ID: 5B1B), (C) Heme/Cu SiR (SiRA; PDB ID: 4RKM), and (D) siroheme-[4Fe-4S] SiR (PDB ID: 1AOP).

In this review we will focus on recent advances in our molecular understanding of the heteronuclear metallocofactor-containing nitric oxide reductase (NOR), terminal oxidase (heme-copper oxidase, HCO), and sulfite reductase (SiR), along with insights gained from the study of biomimetic transition metal complexes. Each of these enzymes catalyzes a key multi-electron reduction reaction relevant to their respective respiratory pathways in the cell: the 2 e⁻ reduction of nitric oxide (NO) to nitrous oxide (N₂O) by NOR, the 4 e⁻ reduction of O₂ to water by HCO, and the 6 e⁻ reduction of sulfite (SO₃²⁻) to hydrogen sulfide (H₂S) in SiR. Notably, these metalloenzymes share a number of structural similarities within their active sites and highlight the way similar building-blocks and catalyst design strategies can accomplish distinct chemical transformations. All three classes of proteins use a heme cofactor to bind and activate the substrate molecule. In the case of NOR and HCO, which are closely related proteins that fall within the same enzyme superfamily, their active sites display a neighboring nonheme metal center that participates in the reduction reaction. A recently determined structure of a multiheme class of SiR has shown a similar heme/nonheme active site structure, although little is currently known about the functional role of this nonheme metal. Regardless, another group of SiRs also utilize a heme cofactor, which is part of a different heteronuclear active site structure, by virtue of a covalent link to a [4Fe-4S] cluster. These native heteronuclear metallocofactor active sites and their biomimetic models have proved to be a challenging but fruitful area of study for structural and computational biologists, biochemists, spectroscopists, and synthetic chemists.

2. Heme-Copper Oxidase (HCO) and Related Biomimetic Models

2.1. The Active Site of Heme-Copper Oxidases.

Aerobic cellular respiration utilizes terminal oxidases, which couple the 4e⁻ reduction of O₂ to water with proton translocation, to generate potential energy in the form of a proton gradient that leads to ATP production by ATP synthase (eq 1).^14,27

$${\text{O}}_2+{8{\text{H}}^{+}}_{(\text{in})}+4{\text{e}}^{-}\to 2{\text{H}}_2\text{O}+{4{\text{H}}^{+}}_{(\text{out})}$$ (1)

One of the most prevalent groups of terminal oxidases are integral membrane proteins present across all kingdoms of life collectively known as heme-copper oxidases (HCO). They belong to the heme-copper oxidase superfamily, which also include the heme/Fe_B-containing NOR enzymes (section 3). During O₂ reduction, electron and proton transfer steps are highly regulated, which is achieved through precise interactions between the metallocofactors and residues in the proton transfer (PT) channel(s) located across different protein subunits. This synergy helps ensure that harmful, partially reduced oxygen species (PROS)–1e⁻-, 2e⁻-, and 3e⁻-reduced superoxide, peroxide, and hydroxyl radical, respectively–are not released while maintaining energy conservation necessary for proton translocation. HCOs are divided into three main types (A, B, and C), based on their subunit compositions and proton-pumping functions (Figure 2).^14,28,29 HCOs are also divided into two distinct classes: one is quinol oxidase (QO) which obtain electrons from quinol, and the other is cytochrome c oxidase (CcO), which obtains reducing equivalents from cytochrome c.¹⁴ Due to the wide diversity of their protein structures, electron donors, proton pathways, and metallocofactors, the evolutionary lineage between members in the HCO superfamily remains challenging to fully elucidate.^14,30,31

Figure 2.

Sub-classes of HCO protein superfamily. Figure adapted from ref³². Copyright American Chemical Society 2014.

All HCOs share a similar subunit I, which contains the catalytic binuclear center (BNC) that is composed of a high spin heme and copper ion (Cu_B). Arguably, the most well-studied is the A1 subfamily, particularly aa₃-type HCO that can be found in mitochondria, Rhodobacter sphaeroides, and Paracoccus denitrificans.¹⁴ HCOs from the B subfamily have been found mainly in archaea but some bacterial B-type HCOs have also been identified and characterized, including ba₃ oxidases from Thermus thermophilus and Rhodothermus marinus.³³ In A- and B-type CcO, electrons are transferred from cytochrome c to a Cu_A site in subunit II on the P side of the membrane (Figure 3). These electrons are then passed to a six-coordinate heme a in aa₃ oxidase or heme b in ba₃ oxidase, which are transferred to the BNC composed of the high-spin heme a₃ or b₃ and Cu_B. Other versions of A- and B-type HCOs contain a high-spin heme o₃ cofactor in the BNC (Scheme 1).³⁴ The C type HCO is comprised only of cbb₃ oxidases, which have been purified from several species, including R. sphaeroides, Vibrio cholerae, Paracoccus denitrificans and Pseudomonas stutzeri.^35–37 These enzymes are commonly expressed under conditions of low O₂. Among the different types of HCOs, cbb₃ oxidase has more in common structurally with bacterial NOR.³¹ Unlike aa₃ and ba₃ oxidase, cbb₃ oxidase contains heme b₃ at the BNC.²⁷ These oxidases contain a subunit N, which is the central subunit containing the BNC, and either an O subunit which has one heme c or both O and P subunits where subunit P contains two heme c cofactors.³⁸

Figure 3.

ET and PT Pathways in subunit I and II of bovine aa₃ CcO (PDB ID 5B1B). The schematic diagram depicts chemical protons transferred to the BNC through the D and K pathways while pumped protons are translocated from N- to P-side of the membrane via the H pathway. HCO types without the H pathway are proposed to translocate pumped protons via spatial analogues of the D and K pathways near the BNC, and then to the P-side through a poorly understood PT pathway.

Scheme 1.

Observed metallocofactors in HCO and ET Pathway

Besides differences in their ET pathways, there is also a wide variation between the PT channel(s) among HCO families. Protons that are transferred from the N-side of the membrane to the BNC active site to form water from O₂ are called “chemical” protons, while the remainder of protons are separately translocated to the P-side of the membrane to generate the so-called electrochemical gradient for ATP synthase, known as the “pumped” protons.^35,36,39 Type A oxidases such as aa₃ oxidase can have three PT pathways named D, K, and H pathways (Figure 3). The D and K pathways are named for the residues, Asp and Lys, respectively, which define these channels.^28,33 The H pathway only exists in A type CcOs (not QOs), and a standing hypothesis is that this pathway serves as “dielectric well” to help compensate for the thermodynamically unfavorable ET from the P side of the membrane.⁴⁰ Several articles and reviews have discussed the structure and potential roles of the H pathway.^34,40–42 It has been suggested that the H pathway is responsible for translocating all the pumped protons in mammalian A type CcO, while the D and K channels transfer the chemical protons to the BNC.^34,43–45 Bacterial A-type CcOs do not rely on the H pathway and it has been established that the K pathway transfers two chemical protons while D pathway is thought to transfer the remaining chemical and pumped protons.⁴³ A recent mutagenesis study has also shown that S. cerevisiae mitochondrial CcO potentially use a similar D pathway for translocation of pumped protons.⁴⁴ On the other hand, in the B and C type HCOs, including ba₃ and cbb₃ oxidase, only a single PT pathway is present, which spatially aligns with the A-type K pathway, and is responsible for transferring all of the chemical and pumped protons.^28,38,46,47 It has been speculated that the single PT pathway in type B and C oxidases contributes to their lower proton pump stoichiometry, along with differences in ligand binding (O₂, NO, CO, etc.) since the D pathway in aa₃ oxidase overlaps with hydrophobic residues essential for O₂ transfer to the BNC.^33,35,48 Regardless of the type of HCO, pumped protons are transferred to a hypothesized “proton loading site” close to the BNC; the exact identity of this loading site is unknown, although it is speculated to be propionate A of heme a₃ which receives protons from E242 (Figure 3); therefore, the precise PT pathway for pumped protons to enter the P-side is not currently well understood.^49,50 Since this review focuses on insights gained from biomimetic models of native enzyme active sites, which is principally concerned with the efficient and selective O–O activation in HCO (see section 2.4), detailed discussion concerning the proposed mechanisms of the proton pumping activity of HCOs will not be included. Other recent reviews have extensively covered the proton pumping aspect of HCO.^49–51

There are many reported structures of type A and B HCO.^46,52–54 Currently, there is only one structure of a type C oxidase reported.³⁸ All three types of HCO enzymes display a highly conserved BNC active site structure buried within the transmembrane domain of the catalytic subunit. The heme cofactor is bound by a single His residue, and the distal pocket contains the Cu_B center ~ 5 Å away, which is coordinated by three conserved His residues. Diffusion of O₂ to the BNC has been studied crystallographically with bovine aa₃ and Tt ba₃ oxidase.^55–57 By collecting X-ray diffraction (XRD) data of Xe-pressurized crystals of HCO, these studies demonstrate that O₂ migrates through a hydrophobic channel in HCO enzymes (Figure 4) to reach the active site.

Figure 4.

XRD structure of Xe-pressurized cyrstals of HCO, showing the hydrophobic channel for O₂ transfer. Figure adapted from ref. ⁵⁵ Copyright American Chemical Society 2012.

Structures of bovine CcO with various substrate analogues (NO, CO, CN⁻) suggest there may be functionally significant variation in Cu_B binding modes (Figure 5).^58,59 Largely, the Cu_B center adopts a trigonal planar geometry in the absence of any ligand, or a pseudo-tetrahedral geometry when a ligand bridging to heme is present. Binding of NO to the heme a₃ of bovine CcO has little effect on the Cu geometry; however, in one instance, CN⁻ binding leads to dissociation of a His ligand with Cu_B and a pseudo trigonal planar geometry by binding to the cyanide nitrogen (PDB ID: 3AG4). Changes to the coordination of CuB have been implicated in modulating the reduction potential of the metal for ET during O–O cleavage or maintaining energy conservation during particular steps of catalytic turnover.^58,60

Figure 5.

Structures of (A) O₂- (PDB ID: 2Y69)-, (B) CN⁻- (PDB ID: 3AG4), (C) CO- (PDB ID: 5X1F), and (D) NO-bound (PDB ID 3AG3) forms of bovine HCO. The BNC in (A) is in the oxidized state while (B), (C) and (D) are in the fully reduced state.

Another notable feature in HCO is a His-Tyr crosslink–a covalent bond between one of the His residues that binds Cu_B and a functionally crucial Tyr residue. This Tyr residue serves as a proton and electron donor during catalysis. Small organic models of the phenol-imidazole crosslink suggest it raises the pK_a and reduction potential of the Tyr phenol sidechain in HCO.^61,62 Implications of the structural and electron donating effect of the cross-linked His to Cu_B coordination have also been considered.⁶³ This residue is also part of the K PT channel (vide supra), and has H-bonding interactions with a water network in the BNC that interacts with the substrate O₂, suggesting it has an important role in PT during turnover.^33,58 The position of this Tyr residue in the polypeptide sequence is not conserved, however; C-type cbb₃ oxidase displays a His-Tyr crosslink at a different position, where the Tyr is further away in sequence and located on a neighboring helix.^38,63 A phylogeny study of HCO sequences has revealed groups of oxidases that lack Tyr at either of these two positions.⁶⁴ Homology modelling of these sequences reveal two other possible sites near the BNC that can have Tyr residues, which may or may not be covalently linked to one of the Cu_B His residues. The observation of functionally conserved, but sequentially dissimilar Tyr residues in HCO is interesting from the perspective of understanding the evolution of HCOs.³¹

2.2. Other Oxidases that Reduce O₂ to Water

There are a multitude of metalloenzymes that bind and reduce O₂. Along with respiratory enzymes that utilize O₂ as an electron acceptor, there are numerous enzymes that couple O₂ reduction to the oxidation of organic substrates. Similar to HCOs, the large majority of these metalloproteins use Fe or Cu centers to activate O₂. Covering the vast literature concerning these proteins is outside the scope of this review. Instead, we refer the reader to recent reviews that have focused on these other O₂ activating metalloenzymes containing heme,^65–67 nonheme Fe,^68–70 and Cu^32,66,71 active sites. This section will focus on metalloenzymes that selectively catalyze the complete 4 e⁻ reduction of O₂ to two molecules of water as their native reaction.

2.2.1. bd Oxidase.

These integral membrane proteins exist in a number of prokaryotes and are involved in O₂ scavenging and respiratory pathways.^72,73 Specifically, all known bd oxidases are quinol oxidases which belong to a family distinct from HCOs.⁷² These enzymes do not pump protons across the membrane, but are able to generate a proton gradient during O₂ reduction by consuming protons from the N-side of the membrane, while oxidizing quinol reducing equivalents near the P-side of the membrane. The bd oxidase contains three heme cofactors, a six-coordinate heme b₅₅₈ involved in ET, along with a high-spin heme b₅₉₅ and a heme d (Figure 6A). The heme d cofactor binds to the protein through a weakly-coordinating Glu and displays high O₂ affinity.⁷⁴ Heme d is typically invoked as the site of O₂ reduction; however, the high-spin heme b₅₉₅ appears to be capable of reacting with O₂ and other various ligands, leading some to propose it also plays a role in O₂ activation.^72,75 Flow-flash kinetics experiments of O₂ reduction by bd oxidase are consistent with a mechanism similar to HCO (see section 2.3), with a notable difference that there is evidence for an observable Fe^III-OOH (hydroperoxide) intermediate.⁷⁶

Figure 6.

Active site structures of non-HCO oxidase enzymes (A) bd oxidase from Geobacillus thermodenitrificons (PDB ID 5IR6), (B) alternative oxidase (AOX) from Trypanosoma brucei (PDB ID 5ZDP) with ferulenol bound, (C) a flavodiiron oxidase (FDP) from Giardia intestinalis showing its diiron binding site (PDB ID 2Q9U), and (D) laccase, a multi-copper oxidase (MCO) from Trametes versicolor containing a T1 and a trinuclear copper center (T2 and T3 Cu) (PDB ID 1GYC).

2.2.2. Alternative Oxidase (AOX).

These membrane-associated enzymes are found in all domains of life and are closely linked to aerobic respiration, although these oxidoreductases are distinct in being incapable of generating any electrochemical gradient. These enzymes are thought to have various regulatory roles: for example, heat generation in tissue, metabolic homeostasis, defense against oxidative stress, and regulation of cellular signaling pathways.^77,78 AOX contain a diiron active site (Figure 6B), which has been unambiguously confirmed recently by the first reported crystal structure.⁷⁹ The proposed O₂-activating mechanism is similar to other carboxylate-bridged diiron proteins, and also shares some features reminiscent to HCO, like a universally conserved Tyr residue near the binuclear active site, which may serve as an electron donor during turnover.⁸⁰

2.2.3. Flavodiiron Protein (FDP).

FDPs, like the HCO superfamily, are a large family of metalloproteins that catalyze either O₂ or NO reduction.^81–83 Many FDPs are capable of catalyzing both of these reactions with nearly equivalent activities, while some appear to be selective for one reaction. FDP are globular proteins typically found in anaerobic organisms and are thought to play a role in O₂ detoxification. Some FDPs have also been identified in photosynthetic cyanobacteria and are involved in regulating the pool of photosynthetic reducing equivalents under changing light conditions.^81,84 Like AOX, the FDP active site comprises a carboxylate bridged diiron center (Figure 6C). Crystal structures of A-type FDPs shows that the diiron metal center is on the opposite end of the flavin mononucleotide cofactor. In order to maintain close proximity for efficient electron transfer, the monomers can form dimer pairs in which the C-terminus of one monomer interacts with the N-terminus of the other.^83,85

2.2.4. Multicopper Oxidase (MCO).

MCOs are a large family of enzymes that couple the reduction of dioxygen to water with the oxidation of organic molecules or metal ions.^32,86 The site of substrate oxidation is distinct from the O₂-reducing site, making MCOs unique from most enzymes that couple O₂ reduction to substrate oxidation. MCOs have a mononuclear type 1 (T1) Cu center, which serves as the primary electron acceptor from the substrate molecule, along with the O₂-reducing trinuclear copper center (TNC), comprised of a binuclear type 3 (T3) site and neighboring type 2 (T2) Cu (Figure 6D). Complete O₂ reduction occurs from the fully reduced form (four Cu^I), which cleaves the O–O bond in two steps, forming a transient peroxy intermediate (PI), followed by the so-called native intermediate (NI).⁸⁷ In the presence of substrate, the NI form is re-reduced and H₂O is released. One turnover of the O₂ reduction catalytic cycle leads to multiple turnovers of substrate oxidation.

2.3. Mechanism of O₂ Reduction by HCOs.

Although certain aspects of the O₂ reduction mechanism by HCO remain unclear, a general consensus of the basic steps has emerged through extensive structural and spectroscopic studies of bovine aa₃ oxidase and other HCOs (Scheme 2). In this section, we will summarize the mechanistic steps of O₂ reduction by bovine aa₃ oxidase, along with discussion of some remaining mechanistic uncertainties. For more detailed information of the spectroscopic and kinetics data pertaining to our understanding of the mechanism of O₂ reduction and proton translocation in HCO, we refer the reader to other recent reviews.^34,49,88 In brief, much of the mechanistic insight into O₂ reduction at the BNC comes from time-resolved spectroscopies, including resonance Raman (rR) and UV-Vis absorption. One particular method that has been key to our understanding of the mechanism is flow-flash spectroscopy.^34,49,89,90 This technique starts with a reduced, or partially reduced, CO-bound HCO in O₂ saturated buffer. Time-resolved spectra of O₂ reduction are obtained upon photolysis of the CO ligand from the BNC. A drawback of this technique is that it relies on fast dissipation of CO from the active site, which happens to be the case for bovine aa₃ oxidase but not all HCOs.⁹¹ More recently, analogous experiments have been performed without CO, using a photolabile O₂-carrier, which have determined an overall identical O₂-reduction process.⁹²

Scheme 2.

Proposed Mechanism of O₂ Reduction by HCO.

2.3.1. O₂ binding in HCO.

In flow-flash experiments with fully reduced bovine aa₃ oxidase (R), the photolabile CO leaves with an apparent lifetime of ~1 μs, and the process of O₂ binding can be observed with a lifetime about an order of magnitude greater, corresponding to a rate constant of ~10⁸ M⁻¹ s⁻¹.⁹⁰ In the absence of O₂, CO still dissociates from the BNC relatively quickly but heme a₃ will rebind exogenous CO after tens of milliseconds.⁹³ Time-resolved vibrational spectroscopic studies reveal formation of a transient Cu_B–CO adduct, concomitant with protein conformational changes that affect the ligand affinity of heme a₃, along with solvent access to a nearby water channel.⁹¹ Based on these observations, it is proposed that Cu_B serves as the initial site of O₂ binding, which is coupled to protein conformational changes that ensure that there is no back-flow of four proton equivalents, before O₂ is transferred and reduced at heme a₃ (Figure 7).⁹³

Figure 7.

Proposed mechanism of gated O₂ binding to BNC with protein conformational changes. (A) O₂ binds initially to Cu_B since the conformation of helix X lowers the O₂-affinity of Fe_a3. The neighboring water channel is based on the helix X conformation. (B) Ser382 is proposed to act as a sensor of H₂O and, once it hydrogen bonds with the water channel, it leads to a change in protein structure which increases Fe_a3 O₂-affinity through changes to helix X and the Fe-His376 bond (gray arrow). (C) In this manner, the O₂-affinity of Fe_a3 is highest when enough H₂O is present in the water channel. Ligand binding to Fe_a3 leads to further changes to helix X (gray arrow): a ‘bulge’ conformation on Val380 shifts to Ser382 ‘closing’ the water channel (D).Figure adapted from ref. ⁹³ Copyright Elsevier 2015.

2.3.2. O–O Cleavage Steps (Oxidative Phase^‡).

After O₂ binding, the next stage of the HCO mechanism is complete O–O bond cleavage, which is known as the oxidative phase of the catalytic cycle.^66,94 Coordination of O₂ to reduced heme a₃ leads to a species with an O₂-sensitive rR signal at 571 cm⁻¹, which is characteristic of the so-called A state.³⁴ This rR peak is indicative of the Fe–O stretching vibration (ν_Fe–O) of an end-on Fe–O₂ species, but the structure can be Fe_a3^II–O₂ (oxy), Fe_a3^III–O₂⁻ (superoxo), Fe_a3^III–O–O–Cu^II (peroxo), or Fe_a3^II–OOH (hydroperoxo) (Scheme 3). The oxy or superoxo assignments for the A state have been favored, due to its similar ν_Fe–O to oxymyoglobin (569 cm⁻¹)—an end-on Fe^II–O₂ or Fe^III–O₂⁻ hemoprotein;³⁴ however, vibrational studies of heme model complexes are consistent with the peroxo or hydroperoxo forms being equally plausible.⁶⁶

Scheme 3.

Possible Structures of A and I_P Intermediates.

The intermediate of the A state decays forming a so-called P state intermediate with a process lifetime of 32 – 40 μs. It was named the P state because it was initially thought to contain a peroxide; however, rR with isotopic labeling ruled this possibility out. In fact, spectroscopic data is consistent with complete O–O bond cleavage at this stage, to yield an Fe_a3^IV=O/Cu^II–OH/Tyr-O^• species (P_M state).³⁴ The exact formulation of the P state depends on whether O₂ is introduced at the fully reduced (R) state, or the two-electron reduced state (where Cu_A and heme a have no reducing equivalents to transfer to the BNC). When heme a is reduced, Tyr-O^• is not observed in the P state, instead heme a is oxidized and Tyr is anionic (P_R state). Therefore, there is uncertainty whether the P_M state is a relevant catalytic intermediate, but the answer likely depends largely on the specific physiological concentrations of O₂ and reducing equivalents in vivo.⁹⁵ Evidence for the relevance of the Tyr-O^• in the HCO mechanism has only been indirectly supported through an iodine radical trapping experiment.⁹⁶ The P_M state is EPR silent;⁴⁹ however, treatment of oxidized bovine aa₃ oxidase with H₂O₂ produces an intermediate that bears weak EPR signals consistent with a delocalized radical on Tyr244, Trp236, and Tyr129, supporting the notion that a Tyr-O^• is a feasible intermediate in HCO during turnover.⁹⁷

No other intermediates have been observed during the A to P state transition, although often a transient peroxo-containing state, I_P, is invoked as a logical species between these intermediates.^49,66,98 The exact nature of this peroxo intermediate is debated, specifically whether Cu_B binds to the distal oxygen, and whether this oxygen is protonated (Scheme 3). Computational studies suggest proton transfer is necessary to lower the barrier to O–O bond cleavage.^99,100 Numerous crystallographic studies of HCO enzymes have purportedly isolated peroxo complexes within the BNC, which would correspond to this I_P state, initially casting doubt on its catalytic relevance; however, analysis of these structures by Adam et al. have identified inconsistencies of the proposed peroxo assignments from a general coordination chemistry perspective in all but one of these structures, warranting caution in taking these HCO structure assignments at face-value.⁶⁶ Numerous models of HCO display observable bridging peroxo-intermediates, Fe^III–O₂²⁻Cu^II, and their relevance to O–O cleavage processes are further discussed in section 2.4.

2.3.3. Reductive Phase.

The remaining steps of the catalytic cycle involve transfer of protons and electrons to regenerate the R state. Each of these steps in the ‘reductive phase’ are coupled to a proton translocation (for the A type oxidases), which is discussed more extensively elsewhere.^34,49–51 Single electron injection experiments have been useful for studying the processes of this phase and these studies conclude that the intermediates in the reductive phase proceed through mostly the same steps: reduction of heme a, followed by proton transfer to BNC, which raises the potential of heme a₃ to allow for reduction of the active site.^34,101 Studies of the mechanism of proton pumping by type A oxidase suggests it is triggered by the oxidation of heme a, which occurs four times during the reductive cycle, and based on the general principle of Coulombic balance of the electron injection into the BNC at a nearby “proton loading site” (Figure 3).^49,50,102 This process is separate from the transfer of protons to the BNC to produce water. The protons that are used to form H₂O are denoted “chemical protons”, and, for the remainder of this discussion, only the chemical protons will be considered.

In the P_M state, the BNC is fully oxidized, which, in the presence of a reducing equivalent from heme a, will instead rapidly form the previously discussed P_R state. Transition from the P_R to the F state occurs through transfer of a proton, leading to formation of a BNC structure of Fe_a33^IV=O/Cu_B^II-OH₂/TyrO⁻. Another series of electron and proton transfers produce the fully oxidized (O_H) state of HCO. This state is poorly understood due to the lack of direct experimental data for it. It is known that the O_H state structurally differs from the as-isolated fully oxidized HCO state that can be structurally and spectroscopically characterized (O) since the O state has a reduction potential too low for the energy conservation function of HCO to be catalytically relevant.^103,104 Computational studies of HCO suggest potential structures of O_H are Fe_a3^III–OH---Cu_B^II/TyrO⁻, Fe_a3^III–OH/Cu_B^I/TyrO•, or Fe_a3^III–μ₂–O–Cu_B^II/TyrOH (Scheme 4).^105,106 It is hypothesized that the O_H state readily converts to O under conditions where reducing equivalents are limiting. Theoretical studies suggest either proton transfer^105,107 or some other structural rearrangement, like change in metal coordination number,^99,106 is responsible for the lower reduction potential in O. For further discussion of the O/O_H states, we refer the reader to recent comprehensive reviews of HCO.^49,66

Scheme 4.

Possible Structures of the High Potential O_H and Lower Potential O States in HCO.

Transition from O_H to the partially reduced (E) state has been studied through electron injection experiments with P. denitrificans aa₃ oxidase, with reduction and proton translocation occurring on the sub-millisecond timescale.^104,108 The exact structure of the E state is unknown, but proposed to be Fe_a3^III–OH/Cu^I/TyrOH, based on the reduction potentials of each redox active center in the BNC.⁶⁶ The final E to R state conversion has also been examined with Pd aa₃ oxidase, through two-electron reduction of the F state with CO, showing a similar series of steps as the other protonation/reduction/proton translocation steps in the reductive phase.^34,109

2.4. Insights Gained from Biomimetic Models of HCO.

HCOs have been studied structurally and mechanistically for several decades and these studies have helped establish an overall view of the H-bonding network at the BNC, the ET pathway that carries electrons to the active site, and the PT channel(s) that transfer protons from the N-side of the membrane towards the BNC, and across the membrane to the P-side. Despite this progress, many aspects of the structural features responsible for such an efficient and selective reduction of O₂ to water are still not fully understood (see section 2.3) because it is quite challenging to purify and study these megadalton-sized membrane proteins with multi-domain structures. There are structural features of HCO that are hypothesized to be essential but are difficult to directly interrogate due to the limitations of site-directed mutagenesis using just 20 natural amino acids. In addition to these substitutional limitations, mechanistic studies are also limited to available spectroscopy or crystallographic techniques compatible with these membrane-bound proteins. Furthermore, it can be difficult to capture and study intermediates in HCO due to presence of spectroscopic features of other cofactors that may interfere or even dominate those of BNC where the O₂ reduction occurs. These challenges motivate researchers to prepare biomimetic models of HCO as a means to gain deeper insight into the O₂ reduction process by this enzyme. These models are much smaller, and can be easier to study because they are free of other cofactors. In addition, it is possible to introduce non-native metal ions or ligands in these models, to probe the roles of each functional group more systematically.

A recent review comprehensively covers studies of synthetic Fe/Cu complexes relevant to HCO and other O₂-reducing metalloenzymes.⁶⁶ The goal of this section is to provide an overview of how these models have helped our understanding of the underlying structural features of native HCO relevant for efficient and selective O₂ reduction. Specifically, we will focus on biomimetic studies relevant towards O₂ binding, O–O cleavage, and catalytic dioxygen reduction.

Many approaches to mimicking the heme/Cu active site structure of HCO have been employed.^{19,66,110,111} One is to chemically synthesize heme and Cu complexes separately and then combine them in the presence of O₂ or H₂O₂. This approach has been useful for understanding how the coordination of Fe and Cu affect the structure and stability of reduced O₂ intermediates. A notable recent example is the preparation of a “naked” copper-heme peroxo complex, which can serve as a starting point to generate a variety of heme-Cu assemblies.¹¹² A second approach is the synthesis and study of tethered heme/Cu complexes. While most heme/Cu complexes are highly reactive towards O₂, and require low temperatures for spectroscopic study, one tethered heme/Cu-peroxo complex has been synthesized which is stable enough to be structurally characterized.¹¹³ These systems have also been applied to electrocatalytic O₂ reduction reaction (ORR), which has been a useful way to interrogate O₂ activation.^114–116 Lastly, another approach to model the HCO active site is to use a smaller and robust protein, such as sperm whale myoglobin (swMb), as a scaffold to engineer structural features which mimic those of HCOs. This approach, called biosynthetic modelling, can take advantage of the well-defined protein scaffolds to introduce amino acids in the secondary coordination to probe the roles of weak non-covalent interactions, such as hydrogen bonding networks involving water, more precisely which facilitates systematic study of their structure-property relationships.^117–119

2.4.1. O₂ Binding of Biomimetic HCO Models.

Studies of simple synthetic Fe porphyrin complexes show that reactions with O₂ are commonly encountered in the reduced Fe^II state.⁶⁶ The greater propensity of small molecule complexes towards intermolecular reactions often leads to formation of unreactive bridging-oxo complexes with sterically unencumbered porphyrins ((porph)Fe^III–μ₂–O–Fe^III(porph)). Reactivity with O₂ at a single Fe site has been examined through the use of sterically demanding porphyrins, to disfavor dimerization, along with the use of cryogenic temperatures to stabilize the early Fe–O₂ intermediates. These studies reveal that O₂ readily reacts with Fe^II and typically yields a side-on or end-on Fe^III–O₂⁻, depending on whether a trans axial ligand is present. In the presence of Cu^I, further activation of O₂ is often observed, leading to bridging-peroxo intermediates (Fe^III-(O₂²⁻)-Cu^II) (see section 2.4.2) (Scheme 5).

Scheme 5.

Characterized Synthetic Fe and Fe/Cu O₂ Species Relevant to HCO.

Tethered heme/Cu ligand scaffolds have been useful tools to compare the effect of Cu on O₂ binding, since the Fe porphyrin can be prepared with or without the nonheme metal. For example, Collman et al. have reported a study of (L1)Fe^II and [(L1)Fe^II/Cu^I]⁺, which both form an Fe^III–O₂⁻ upon reaction with O₂ (Scheme 6).¹²⁰ With Cu^I, irreversible binding of O₂ occurs at room temperature, and the superoxo complex is stable to multiple degassing freeze-pump-thaw cycles. On the other hand, the Fe-only complex only binds O₂ once cooled −60 °C. Related model complexes that are able to electrocatalytically reduce O₂, [(L2)Fe^II/Cu^I]⁺ and (L2)Fe^II, display rate limiting O₂ binding.¹²¹ Further study of the L2 system showed that Zn^II has a similar effect as Cu^I on the rate constant of O₂ binding (k_on). Interestingly, the corresponding k_on for the Fe-only complex was higher, but also highly sensitive to the presence of H₂O, which inhibits O₂ binding.¹²² Halime et al. reported a HCO model system, comprised of a tethered high-spin Fe porphyrin and appended Cu, [(L3)Fe^II/Cu^I]⁺, and its Cu-free form, (L3)Fe^II, are competent for O₂ reduction catalysts (Figure 8).¹²³ The rate limiting step of O₂ reduction of both complexes is O₂ binding at room temperature, and the ORR rate of [(L3)Fe^II/Cu^I]⁺ is higher than (L3)Fe^II under these conditions. At low temperature, O–O cleavage is the rate limiting step and there is no difference in rate between [(L3)Fe^II/Cu^I]⁺ and (L3)Fe^II. These results are consistent with Cu principally promoting O₂ binding in this system, but having a negligible role in the subsequent O–O bond cleavage.

Scheme 6.

Tethered Heme/Cu Complexes Which Display Effect of Nonheme Metal on O₂ Binding Reported by Collman et al. ^120,121

Figure 8.

Proposed mechanisms of homogenous catalytic O₂ reduction by (A) [(L3)Fe^II/Cu^I]⁺ and (B) (L3)Fe^II. Mechanism reproduced from ref. ¹²³ Copyright National Academy of Sciences 2011.

The Lu group has used a rationally-designed mutant of sperm-whale myoglobin (Leu29His, Phe43His swMb; Cu_BMb) to model aspects of the BNC structure in HCO (Figure 9). By incorporating different heme cofactors into Cu_BMb, the influence of heme reduction potential on O₂ binding was investigated.¹¹⁹ Both the k_on and k_off O₂ binding rate constants increase with higher heme reduction potential, but k_off displays a steeper dependence on potential, thus making the O₂ affinity decrease overall. The presence of a nonheme metal also leads to higher O₂ affinity, based on studies using Ag^I as a redox-inactive analogue of Cu^I.¹²⁴ A related biosynthetic protein (Leu29His, Phe43His, Val68Glu swMb; Fe_BMb), that differs by the presence of a coordinating Glu residue to the nonheme metal, demonstrate that the Lewis acidity of the nonheme metal impacts O₂ binding. The rate of formation and stability of oxyheme species was shown to depend on the identity of the nonheme metal (Fe^II, Cu^I, Co^II, Mn^II, and Zn^II), and could be directly compared when the nonheme metal was not able to be oxidized.^117,118

Figure 9.

Overlay of structure of Phe33Tyr Cu_BMb (magenta) with HCO active site (white).

In native HCO, insight into the effect of Cu_B on O₂ binding are informed by CO photolysis studies (see section 2.3.1), which reveal that when CO dissociates from heme a₃ it transiently binds to Cu_B. This is thought to model, in reverse, the steps of O₂ binding to the BNC active site. Similar studies have been carried out with biomimetic model complexes, which generally display this same property as the native enzyme.^125–128 Other examples of model substrate binding studies include the [(L4)Fe^III/Cu^II]³⁺ complex reported by Dallacosta et al., which displays a fivefold increase in azide binding affinity relative to [(L4)Fe^III]⁺ (Scheme 7).¹²⁹

Scheme 7.

Protoporphyrin IX-Derived Tethered Heme HCO Model Complex and Catalytic H₂O₂-mediated para-Cresol Oxidation.¹²⁹

2.4.2. O–O Activation and Cleavage.

Direct studies of the effect of Cu^I on the properties of the initial heme-O₂ intermediates are challenging, due to their propensity for oxidation of Cu^I and formation of bridging-peroxo complexes. As previously mentioned, examination of redox-inactive metals in place of Cu^I has been useful for probing the early stages of O₂ binding and activation. In Fe_BMb, increased Lewis acidity of the nonheme metal increases the observed ν_Fe–O of the oxyheme, which indicates a strengthened Fe–O interaction, and subsequent weakening of the O–O bond.¹¹⁸ Only a few stable Fe^III-O₂⁻/Cu^I HCO model complexes have been characterized, although short-lived Fe-superoxo species have been observed in stopped-flow kinetic studies of Fe^III–(O₂²⁻)–Cu^II formation.¹³⁰ The previously discussed picket-fence porphyrin complex L1 with Fe and Cu forms a stable Fe^III-O₂⁻.¹²⁰ Studies of the corresponding Co-porphyrin complex with an empty distal pocket, or with nonheme Cu^I, Co^II, or Zn^II, show that these nonheme metals have only a small effect on the ν_O–O of the Co^III–O₂⁻.¹³¹ In an intriguing example of a Fe^III-O₂⁻/Cu^I complex, Liu et al. reported that this species forms in [(L5)Fe^II/Cu^I]⁺ through a transient bridging-peroxo intermediate, seemingly the reverse of the formation pathways of a majority of Fe^III–(O₂²⁻)–Cu^II complexes (Scheme 8).¹³² Two factors were proposed to be crucial for the shift in equilibrium towards Fe^III-O₂⁻/Cu^I with L5: (i) the ‘push’ effect of the axial imidazolyl weakens the binding interactions between Cu and the peroxide and (ii) the relative order of reduction potentials of Fe and Cu in this system thermodynamically favor Cu^I. This conclusion is corroborated by a report of a stable peroxide in a related system that bears no heme axial ligand, which would both raise the heme potential and remove any ‘push’ effect.¹³³

Scheme 8.

Formation of Fe^III-O₂⁻ Through a Transient Fe^III–(O₂²⁻)–Cu^II. ¹³²

Numerous studies of biomimetic Fe^III–(O₂²⁻)–Cu^II intermediates have been performed to gain a deeper understanding of how the coordination environment of the metals influence the binding and reactivity of the peroxide moiety.^66,130 For example, the properties of a series of (L)(L6)Fe^III–(O₂²⁻)–Cu^II(L’) complexes were investigated, where the presence of an axial ligand (L = DCHIm [1,5-dicyclohexylimidazole] or nothing) and chelating ligand to Cu (L’ = TMPA [tris(2-pyridylmethyl)amine] or AN [bis(3-[dimethylamino]propyl)amine]) had significant impact on the peroxo moiety (Scheme 9).^{112,134–136} The HS-TMPA (no L, L’ = TMPA) complex displays a side-on binding between Fe and O₂²⁻,^134,137 which is confirmed by characterization of a tethered complex that has a highly similar ligand architecture and spectroscopic properties (Figure 10).¹¹³ The related HS-AN complex displays a greater degree of O–O activation, a consequence of the Cu center shifting from binding end-on to the peroxide in HS-TMPA to side-on in HS-AN due to the change in coordination number about Cu.^135,137 Addition of an axial ligand to Fe leads to a change in its spin state, and loss of one Fe–O interaction, resulting in end-on binding in LS-TMPA and LS-AN for both Fe and Cu centers.^112,136,138 The consequences of the end-on/end-on binding mode include decreased activation of O₂²⁻ (based on vibrational spectroscopy) and an increased Fe–Cu distance.

Scheme 9.

Influence of Fe and Cu Coordination Environment on Peroxo-Moiety.

Figure 10.

XRD structure of stable Fe/Cu-peroxo complex reported by Chishiro et al.¹¹³.

The characterization of heme/Cu-superoxo and -peroxo complexes discussed here have important implications for understanding the cooperativity between Fe_a3 and Cu_B in HCO during O₂ activation. A major conclusion from these HCO model studies is that the precise role of Cu_B will be highly dependent on its coordination geometry, distance from the heme, and protonation state of the reduced O₂ species. Whether the early intermediates in HCO involve end-on bridging peroxo species, like LS-AN, or if the protein environment shifts the equilibrium towards an Fe^III–O₂⁻/Cu^I, similar to what has been observed for other tethered heme/Cu assemblies, would effect our perspective of a number of aspects of HCO catalysis, such as the role of protons in formation of the P state, and the selectivity the native enzyme displays for complete O₂ reduction to water.

We can consider most of the discussed Fe^III–O₂⁻/Cu^I and Fe^III–(O₂²⁻)–Cu^II complexes as models of HCO active site in the early A or I_P states. Further activation to completely cleave the O–O bond in synthetic biomimetic complexes nearly always requires the addition of some form of reducing equivalent. Interestingly, this observation is despite the fact that there are available electrons from the metal centers, or porphyrin ligand. An exception of complete O–O bond cleavage of H₂O₂ with an Fe^III/Cu^II complex has been reported by the Casella group: the tethered protoporphyrin IX/Cu complex demonstrates peroxidase-like activity with (L4)Fe^III/Cu^II for the oxidation of para-cresol (Scheme 7).¹²⁹ Further study of the effect of nonheme metal on this peroxidase-like activity confirm the major role of the metal is to enhance H₂O₂ binding to heme, and promote heterolytic cleavage to form the compound I-like active oxidant, (porph^•+)Fe^IV=O.¹³⁹ Although this reaction is not totally relevant to O₂ reduction in HCO, it demonstrates that obtaining additional reducing equivalents from the ligand is a viable way to cleave the O–O bond. In contrast, in initial attempts to promote O–O cleavage in HS-AN and HS-TMPA, the addition of strong acid to these Fe^III–(O₂²⁻)–Cu^II intermediates simply leads to release of H₂O₂ (Scheme 10).¹³⁵ This is a different outcome from related mononuclear Fe^III–O₂²⁻ complexes, which can form analogues to compound I upon addition of acid.^66,140 Complete O–O cleavage from HS-AN and HS-TMPA requires addition of at least two equivalents of reductant, yielding the corresponding bridging-oxo product [(L6)Fe^III–μ₂-O–Cu^IIL’]⁺.¹³⁵

Scheme 10.

Reactivity of HS-TMPA and HS-AN Towards Acid and Reductant to Yield H₂O₂ and H₂O, Respectively.¹³⁵

There are examples of phenols being used to promote complete O–O cleavage in heme/Cu assemblies, which can support the important role of the active site Tyr as a proton and electron donor in HCO. Collman et al. reported the reaction of [(L1)Fe^III–O₂⁻/Cu^I]⁺ with sterically-hindered phenols generates a phenoxyl radical along with [(L1)Fe^Iv=O/Cu^II–OH]⁺, analogous to the P_M state of HCO (Scheme 11).¹⁴¹ In a later study, they showed that a modified tethered ligand (L2), bearing a mimic of the His-Tyr crosslink could similarly achieve complete O–O cleavage of O₂ with intramolecular PCET from the appended phenol.¹⁴² More recently, a detailed computational and kinetic isotope effect study of the reaction between LS-AN with phenol has provided insight to the mechanism of the phenol-promoted O–O cleavage.¹⁴³ A mechanism involving hydrogen-bond assisted homolytic cleavage of the bridging peroxo was favored over an alternative that involved initial proton transfer from phenol to generate a transient hydroperoxo (Scheme 12). The potential relevance of this mechanism to HCO was discussed, and it was concluded that proton transfer from the His-Tyr crosslink is an essential prerequisite for its capacity to serve as an electron donor for O–O cleavage.¹⁴³ On the other hand, many other reported heme/Cu assemblies do not display complete O–O cleavage in the presence of phenols, or appended phenol moieties. It is thought that the low spin ferric heme of [(L1)Fe^III–O₂⁻/Cu^I]⁺ is crucial for its difference in observed reactivity.¹³⁷ It is surprising, then, that a similar phenol-appended ferric-superoxo complex reported by Liu et al. is unreactive (Scheme 8).¹³² Furthermore, Adam et al. have recently demonstrated that two low spin Fe^III–(O₂²⁻)–Cu^II complexes (LS-4DCHIm and LS-3DCHIm) display different reactivity towards weak phenolic acid and reductant, based on differences in Cu coordination (Scheme 13).¹⁴⁴ There are likely other important structural and electronic features of heme/Cu assemblies which determine their propensity to completely cleave the double-bond of O₂ which remain to be elucidated.

Scheme 11.

O–O Cleavage Induced by Inter- and Intramolecular Proton Coupled Electron Transfer from Phenol.^141,142

Scheme 12.

Hydrogen Bond-Assisted Mechanism of O–O Cleavage in LS-AN.¹⁴³

Scheme 13.

Reactivities of Low-Spin Fe^III–(O₂²⁻)–Cu^II Complexes with Different Cu Coordination Number.¹⁴⁴

2.4.3. Catalytic Dioxygen Reduction by HCO Models.

Study of the catalytic performance of biomimetic heme/Cu assemblies has been complementary to low-temperature spectroscopic studies of heme/Cu O₂ intermediates relevant to HCO, as discussed in the previous sections. In particular, the application of biomimetic heme/Cu complexes to electrocatalysis has been particularly useful in understanding the importance of reduction potentials and electron flux on efficient and selective O₂ reduction. More extensive reviews covering the breadth of (electro)catalytic O₂ reduction by synthetic complexes related to HCO have been published.^66,145–147 Here we will highlight insights gained relevant to our understanding of the mechanism of the native enzyme. Catalytic studies are useful for investigating key structural features that are relevant for high ORR activity, with high selectivity (low PROS formation), and, in some cases, can also be studied using in situ spectroscopic techniques to directly examine the nature of reactive, catalytically-relevant intermediates.

Many relatively complex synthetic porphyrin structures have been examined for catalytic ORR activity. Among these that are relevant to HCO, interesting dependences (or lack thereof) of the Cu center on the extent of PROS formation have been observed. For example, a series of ‘capped’ porphyrin complexes reported by Collman et al. are able to electrocatalytically reduce O₂ to form 2 or 4 e⁻ reduced products (Scheme 14).¹⁴⁸ The complexes where the Cu is easier to reduce than the Fe porphyrin display mixtures of 2 and 4 e⁻ O₂ reduction, where the 2 e⁻ reduced products are attributed to reaction pathways that only involve Cu. Substitution of Fe for Co, which is easier to reduce than the Cu center, leads to exclusive formation of H₂O.¹⁴⁹ Interestingly, many other heme/Cu ORR catalysts display no increase in selectivity with Cu, likely due to the catalytic conditions that employ a large excess of reducing equivalents (Scheme 15).^{114,123,150,151} In other examples of electrocatalytic ORR by tethered and supramolecular heme/Cu assemblies, Cu has been shown to slightly increase the average electrons transferred under electrocatalytic conditions, suggesting the Cu is playing a role in storing electrons (Scheme 15; (L11)Fe^III/Cu^II(TerpyCD2) and [(L12)Fe^II/Cu^I]⁺);^116,152,153 however, no dramatic rate enhancement comes from the presence of Cu.

Scheme 14.

Selectivity of Electrocatalytic ORR By Heme/Cu Complexes.^148,149

Scheme 15.

Effect of Cu on Catalytic ORR Selectivity for Various HCO Model Complexes. ^{114,116,123,150–153}

Another structural feature of HCO that has been mimicked in a number of ORR catalysts is the active site Tyr residue. Electrocatalytic ORR studies of synthetic tethered heme/Cu complexes containing a pendant phenol demonstrate that, in some cases, the phenol moiety has no effect on ORR,¹⁵⁴ while, in slightly different metal coordination environments under different catalytic conditions, the phenol can help reduce PROS formation.¹¹⁵ In the latter case, the phenol is postulated to reduce PROS through mimicking the role of the His-Tyr in HCO, which is supported through single-turnover studies (Scheme 11).¹⁴²

The effect of the active site Tyr, and the His-Tyr crosslink, on catalytic ORR has been thoroughly investigated through the biosynthetic Cu_BMb system. Miner et al. demonstrated that introducing a Tyr residue near the active site (Phe33Tyr or Gly65Tyr) dramatically increases the rate and turnover number (Figure 11).¹⁵⁰ The Gly65Tyr Cu_BMb mutant displays more than double the activity relative to Phe33Tyr Cu_BMb, demonstrating the importance of this residue’s positioning relative to the heme active site. Furthermore, these Tyr-containing mutants display greater selectivity for complete O₂ reduction, which was directly confirmed by measuring H₂¹⁷O formation from ¹⁷O₂. Remarkably, further engineering of Gly65Tyr Cu_BMb through introducing positively charged residues on the protein surface, in order to facilitate faster ET, produces an ORR enzyme with activity comparable to that of a native HCO.¹⁵⁵ An XRD structure of reduced Phe33Tyr Cu_BMb with O₂ shows that the introduced Tyr residue participates in hydrogen-bonding interactions with a water network leading to the bound O₂ (Figure 12).¹⁵⁶ This proton-delivery pathway was proposed to be a major reason for the higher activity of Phe33Tyr Cu_BMb, relative to WT swMb. Further investigation of this structural feature in Phe33Tyr was done through replacing the Tyr with unnatural amino acid analogues of Tyr bearing different electron-withdrawing groups.¹⁵⁷ Decrease of the pK_a of the phenol sidechain leads to an increase in ORR activity, consistent with this residue being involved in PT during turnover. Evidence for Tyr residue also serving as an electron donor, and forming a Tyr-O^•, similar to the mechanistic proposals of HCO, were observed by treating oxidized Phe33Tyr Cu_BMb (or the unnatural Tyr analogues) with H₂O₂.^157,158 An EPR signal consistent with a phenoxyl radical is observed, which is not present in Cu_BMb without this Phe33Tyr mutation. Lower amounts of this radical are also observed when the reduced biosynthetic protein is incubated with O₂, supporting its relevance to the catalytic mechanism.¹⁵⁸ Finally, to understand the potential importance of the covalent link between His and Tyr in native HCO, an unnatural amino acid mimicking this cross-link (imidazolyl-tyrosine; imiTyr) was incorporated into Cu_BMb (imiTyr Cu_BMb).¹⁵⁹ This biosynthetic protein was unique for displaying higher Cu^II affinity, and higher ORR activity than Phe33Tyr Cu_BMb (12 μM O₂ min⁻¹ for imiTyr Cu_BMb+Cu^II versus ~5 μM O₂ min⁻¹ for Phe33Tyr Cu_BMb, under identical conditions). Furthermore, addition of Cu^II was shown to significantly decrease PROS formation from 30% to 6%, suggesting this crosslink may have a role in retaining and tuning the Cu center for more selective O₂ reduction.

Figure 11.

ORR activity of swMb and biosynthetic HCO models. Figure adapted from ref. ¹⁵⁰ Copyright Wiley-VHC 2012

Figure 12.

H-bonding network in O₂-bound Phe33Tyr Cu_BMb crystal structure. Figure adapted from ref. ¹⁵⁶ Copyright American Chemical Society 2016.

As mentioned previously, the majority of direct electrocatalytic or homogenous ORR studies of heme/Cu systems show little benefit from Cu or, in some cases, display greater propensity for PROS production when Cu is present. Collman et al. hypothesized that to observe HCO-like ORR catalysis, the nature of ET rate and active site isolation should be closer to the conditions of the native enzyme.¹⁶⁰ Indeed, depositing the [(L1)Fe^II/Cu^I]⁺ complex into a phosphtidylcholine film on the electrode surface ensures site-isolation of the catalyst and reduces the rate of ET to being diffusion limited.¹⁶⁰ Under these conditions, [(L1)Fe^II/Cu^I]⁺ displays selective 4 e-reduction of O₂. Further control of the ET rate from electrode to catalyst could be accomplished by covalent attachment of heme/Cu complexes onto self-assembled monolayer (SAM) films on Au electrodes.^114,115 These studies use functionalized SAM components that facilitate fast (~10³ - 10⁴ e⁻ s⁻¹) or slow (4 - 6 e⁻ s⁻¹) ET to various HCO mimics. Comparison studies of related picket-fence porphyin complexes, (L1)Fe^II,[(L1)Fe^II/Cu^I]⁺, and [(L2)Fe^II/Cu^I]⁺, demonstrate that the catalyst with all 3 redox active centers, [(L2)Fe^II/Cu^I]⁺ (FeCuArOH), produces the least amount of PROS under fast and slow ET regimes, however the difference between the three catalysts is much more pronounced when ET is slow (Figure 13).¹¹⁵ Analogous studies with (L3)Fe^II, [(L3)Fe^II/Cu^I]⁺, and [(Im)(L3)Fe^II/Cu^I]⁺ show similar effects: Cu decreases PROS formation, which is further reduced when an addition axial heme ligand is present, but only under low ET rates with C₁₆SH (Figure 14).¹¹⁴ Together, these results demonstrate that, under conditions where reducing equivalents are transferred slowly, structural features that more faithfully model the features of HCO (nonheme Cu^I, nearby phenol, and low-spin ferric heme) produce a more selective ORR catalyst.

Figure 13.

Effect of Cu and phenol on electrocatalytic ORR selectivity under ‘fast’ and ‘slow’ ET regimes, through covalent tethering of catalyst to different Au SAM electrodes. Figure adapted from ref. ¹¹⁵ Copyright the American Association for the Advancement of Science 2007.

Figure 14.

Effect of Cu and axial ligand on PROS formation on electrocatalytic ORR with (L3)Fe^III (⁶L-Fe) and related complexes. Electron flux is controlled by measuring catalysis with different electrodes: fast ET with edge-plane graphite (EPG), slow ET with 1-octanethiol SAM on Au (C₈SH), and very slow ET with 1-1-hexadecanethiol SAM on Au (C₁₆SH). Figure adapted from ref. ¹¹⁴ Copyright American Chemical Society 2015.

An advanced spectroscopic technique, known as surface-enhanced resonance Raman spectroscopy (SERRS) has been used by the Dey group for the study of catalytic intermediates of some of these biomimetic ORR catalysts.¹⁴⁷ Using SERRS to examine the mechanism of ORR by (L3)Fe^II, [(L3)Fe^II/Cu^I]⁺, and [(Im)(L3)Fe^II/Cu^I]⁺ the authors observed [(Im)(L3)Fe^II/Cu^I]⁺ produces a six-coordinate low-spin (l.s.) Fe^II species upon reduction, with or without O₂ present (Scheme 16).¹¹⁴ The sluggish substitution of the sixth ligand (presumably H₂O) for O₂ was attributed to this complex’s higher PROS at high electron flux, due to side reactions of Cu^II-O₂⁻ species. Furthermore, spectroscopic evidence for significant formation of the (L3)Fe^III–O₂²⁻–Cu^II intermediate is observed, with and without an imidazole axial ligand. Build-up of Fe^IV=O is observed with (L3)Fe^II, but not the other two complexes, which the authors attributed to a high-potential compound I analogue forming upon heterolytic cleavage of the peroxo intermediates.^114,147 A similar mechanistic study of ORR by Gly65TyrCu_BMb yielded SERRS spectra of numerous oxidized Fe signals under catalytic turnover, including l.s. Fe^III—possibly superoxo or peroxo intermediates—and Fe^IV=O.¹⁶¹ By attaching Gly65Tyr Cu_BMb to this electrode, this biosynthetic protein is able to display ORR activity over 100 times faster than other synthetic models, and 10 times faster than electrode-immobilized native HCO.

Scheme 16.

Proposed Mechanism of ORR by [(Im)(L3)Fe^II/Cu^I]⁺ Based on intermediates Observed Via SERRS. Figure Adapted from Ref. ¹⁴⁷. Copyright American Chemical Society 2016.

2.5. Summary and Outlook of HCO and Biomimetic Models.

Through decades of studying native HCO and biomimetic models, we have improved our understanding of key structural features responsible for efficient and selective reduction of O₂ to H₂O through the proposed mechanism that has been illustrated in Scheme 2. Probing the O₂ reactivity of heme/Cu assemblies has demonstrated that both the primary coordination sphere (i.e., the heme with axial ligand and tridentate coordination of Cu_B) and the surrounding secondary coordination sphere (including hydrogen-bonding interactions with water molecules anchored by the phenol of the highly conserved His-Tyr moiety) is crucial for various aspects of catalytic O₂ reduction, including O₂ binding and O–O cleavage. Specifically, while the precise role of Cu is highly dependent on its coordination geometry, distance from the heme, and protonation state of the reduced O₂ species, it has a functional role in promoting O₂ binding through its Lewis acidity, along with increasing the binding affinity of the heme to various other ligands. In addition, coordination of an axial ligand to heme can lead to a change in the Fe spin state, affecting the binding properties towards O₂; this axial ligand also has an effect on the binding interactions between Cu and O₂. Synthetic heme/Cu systems have been observed with a range of O₂ binding modes, from close side-on/side-on binding (μ-η²,η²) to negligible Cu-O₂ interactions, favoring a Fe^III–O₂⁻ complex. Furthermore, the active site Tyr residue in HCO is an essential proton and electron donor for O–O cleavage and its covalent crosslink to the Cu-coordinating His residue tunes the reactivity of the phenol and Cu_B that further promote O₂ activate. Interestingly, while many catalytic ORR studies of heme/Cu systems show little benefit from Cu, conditions where reducing equivalents are transferred slowly reveal that models with structural features that more faithfully model the core components of HCO (nonheme Cu, nearby phenol, and low-spin ferric heme) produce a more selective ORR catalyst. The application of biomimetic HCO models towards catalytic ORR has not only improved our understanding of these enzymes, but has shown that, by incorporating these crucial structural features from the native active site, catalytic activity that matches, or even surpasses, native enzymes can be accomplished. One crucial characteristic of the native HCO mechanism that is relatively underdeveloped in biomimetic complexes, however, is that it is capable of ORR with considerable energy conservation, to generate potential energy though proton translocation. This functional feature of HCO is derived from precise control of the reduction potentials and pK_a values of the BNC during turnover. Learning how to mimic these aspects in simpler models would further improve our molecular understanding of HCOs.

While significant progress has been made concerning our mechanistic understanding of HCO, there remain a number of key questions:

1. What is the nature of the (transient) peroxo intermediate, I_P? Is this intermediate protonated prior to O–O cleavage? Does Cu bind to this intermediate prior to transition to P state?

2. What structural features in the O_H state lead to its high reduction potential?

3. How is the His-Tyr cross-link formed?

While it is challenging to trap these intermediates in the native system, synthetic models have been an invaluable means to interrogate these proposed intermediates. Further study of these model systems may inform how the structural aspects of these proposed intermediates impact their reactivity and potential catalytic relevance.

3. Bacterial Nitric Oxide Reductase (NOR) and Related Biomimetic Models

3.1. The Heme/Fe_B Active Site of Bacterial NORs.

Certain bacteria are able to reduce NO to nitrous oxide (N₂O) during a metabolic process known as denitrification, which converts nitrate (NO₃⁻) and nitrite (NO₂⁻) to N₂.^162,163 Denitrification can be considered an anaerobic version of respiration—the corresponding aerobic respiration process performs 4 e⁻ reduction of O₂ to H₂O by heme-Cu oxidase enzymes (HCO), which generates chemical potential energy through proton translocation (see section 2). HCO and bacterial NORs belong to the same enzyme superfamily known as the heme-Cu oxidase superfamily, which are a diverse group of integral transmembrane proteins that share a relatively similar catalytic subunit, with a diverse array of secondary subunits involved in proton and electron transfer (Figure 2).^164,165 The NOR enzyme class has been sub-divided principally on the initial electron donors (Scheme 17): cytochrome c for cNOR and quinol for qNOR. A third sub-class of NOR has been discovered which contains a Cu_A cofactor and obtains its electrons from either cytochrome c or quinol, denoted Cu_ANOR. The Cu_ANOR subclass is further subdivided into bNOR, eNOR, and sNOR, based on homology models of predicted genes.³⁰ XRD and cryogenic electron microscopy (cryo-EM) structures of cNORs and qNORs have been obtained from various bacteria.^166–172 To date, no structures of Cu_ANORs have been reported, but biochemical studies from a Cu_ANOR from Bacillus azotoformans have been described.^173,174

Scheme 17.

Various Cofactors Observed in NORs and ET Pathway to the BNC Active Site.

The study of bacterial NORs have been challenging due to being integral membrane proteins that contain multiple redox active transition metal cofactors. The first in vitro study of NO-reducing activity of membrane fractions of extracts from Alcaligenes faecalis IAM 1015 was reported in 1971.^164,175 However, it wasn’t until 1989 that a preparation of purified, active NOR, cNOR from Pseudomonas stutzeri (PsNOR) was established by Heiss et aI., using Triton X-100 detergent as a stabilizing agent.¹⁷⁶ Biochemical studies of PsNOR, and later purified NORs from other bacteria, confirmed that these enzymes selectively catalyze the 2 e⁻ reductive coupling of two NO molecules to form N₂O (eq 2),^176–178 avoiding alternative reactions, including reductive NO disproportionation (eq 3) which is considered a more common reaction that occurs between NO and transition metal complexes.¹⁷⁹

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

$$3\mathrm{NO}+{\text{e}}^{-}\to {\text{N}}_2\text{O}+{\text{NO}}_{2^{-}}$$ (3)

Later efforts established that cNOR contains three heme cofactors (heme c, b, and b₃, where b₃ denotes a high spin heme b cofactor), along with a nonheme Fe (Fe_B).^180,181 In 1994, analysis of the available sequences from bacterial NORs and multiple HCOs established for the first time an evolutionary link between these two enzyme classes.¹⁸² A suitably high resolution XRD structure of a bacterial NOR enzyme was not obtained until 2010,¹⁶⁶ but by 1998 the overall protein structure of NOR—and the positioning of its metallocofactors—had been inferred via homology modelling to a published HCO crystal structure.¹⁸³ By this time, it was widely accepted that bacterial NORs had an architecture similar to the catalytic subunit of HCO, including an analogous bimetallic active site, denoted the binuclear center (BNC), where a histidine-coordinated high-spin heme cofactor is positioned next to a nonheme metal that is bound to three conserved His residues.¹⁶³ A distinguishing feature of NOR sequences, compared to HCO, is the presence of multiple conserved glutamate residues near the BNC, including Glu211 (the numbering from Pseudomonas aeruginosa cNOR; PaNOR) which was proposed to be an additional ligand to Fe_B, to satisfy a preferred octahedral geometry.¹⁸⁴ Biochemical studies by Butland et al. in the early 2000’s demonstrated that some of the conserved Glu residues in P. denitirificans cNOR (PdNOR) are essential for enzymatic activity (Glu198 and Glu125, PdNOR numbering), while not significantly disrupting the assembly and coordination of the metal cofactors (including Fe_B).¹⁸⁵ Glu198 of PdNOR corresponds to Glu211 in PaNOR, the potential Fe_B binding residue.

When the first XRD structure of a bacterial NOR, PaNOR, was disclosed by Hino et al. in 2010, it confirmed much of the prevailing ideas concerning the general organization of the polypeptide backbone, and the molecular arrangement of the BNC active site (PDB ID: 3O0R).¹⁶⁶ PaNOR is composed of two subunits, NorB (56 kDa), which displays 12 transmembrane α-helices that contain the BNC and the heme b cofactor, coordinated by two His residues, and NorC (17 kDa), a single transmembrane α-helix attached to a periplasmic-facing domain that contains the heme c center coordinated by His and Met residues (Figure 15). An ET pathway from the periplasmic surface of PaNOR can be traced from the heme c, b, and b₃ centers, with a Ca^II ion bridging the propionates of hemes b and b₃, which is thought to aid in arranging these cofactors for efficient ET. The enzyme was crystallized in its oxidized state, leading the BNC to contain a μ₂-O ligand between heme b₃ and Fe_B (Figure 16A). The Fe–Fe distance of the BNC is relatively short, 3.9 Å, compared to the Fe–Cu distance of 4.4 Å in Tt ba₃ oxidase (an HCO).¹⁶⁶ Fe_B is 5- or 6-coordinate, with either k¹-O or k²-O,O’ binding to the carboxylate side chain of Glu211 (Fe–O distances of 2.04 and 2.47 Å), along with bonds to the μ₂-O and the three conserved His residues (His207, 258, and 259). Multiple Glu residues near Fe_B (Glu 211, 215, and 280) lead to a relatively electronegative distal pocket for heme b₃, significantly lowering its reduction potential in PdNOR to ~ 60 mV (vs NHE), compared to the potentials of heme c and heme b, (310 mV and 345 mV, respectively).¹⁸⁶ These Glu residues are also thought to be part of a putative PT pathway from the periplasm to the buried active site (Figure 17). The distance between one of the heme b₃ propionates and the next hydrogen-bonding residue (Thr330) is 8.0 Å in the crystal structure, suggesting conformational changes in PaNOR are necessary to form a complete PT pathway.¹⁶⁶ Molecular dynamics simulations of PaNOFt suggest two possible PT pathways, based on protein structural rearrangements over time.¹⁸⁷

Figure 15.

XRD structure of PaNOR (PDB ID: 3O0R), comprised of NorB (green) and NorC (magenta) subunits.

Figure 16.

Different XRD structures of BNC of NORs. (A) Resting state PaNOR (PDB ID 300R) with μ₂-O, (B) reduced PaNOR (PDB ID 3WFB) with μ₂-Cl⁻, (C) reduced PaNOR (PDB ID 3WFC) with CO bound, (D) reduced PaNOR (PDB ID 3WFD) with acetaldoxime, (E) GsNOR (PDB ID 3AYF) with Zn^II at the Fe_B site, (F) NmNOR (PDB ID 6LIX) with Zn^II inhibitor coordinating 3 Glu residues, (G) cryo-EM structure of NmNOR (PDB ID 6L3H), and (H) cryo-EM structure of AxNOR (PBD ID 6QQ5).

Figure 17.

Structure of PaNOR showing the putative PT channels (1 and 2), along with the hydrophobic channel for NO transfer (yellow). Figure adapted from ref. ¹⁶⁶ Copyright the American Association for the Advancement of Science 2010.

PaNOR has also been crystallized in reduced and ligand-bound forms, to gain further insight into the possible BNC structures during turnover.¹⁶⁹ The reduced structure (PDB ID: 3WFB) contains a bridging Cl⁻ between heme b₃ and Fe_B and displays a slightly increased distance between the Fe centers (Fe–Fe distance of 4.2 Å, compared to 3.9 Å of oxidized form) (Figure 16B). This greater distance is still significantly shorter than the Fe–Cu distances in structures of most HCO enzymes (see Table 3 in section 5),¹⁸⁸ leading to speculation that two diatomic molecules may not be able to bind the metals simultaneously within the BNC. On the other hand, the CO-bound structure of reduced PaNOR contains an ovoid electron density between heme b₃ and Fe_B, which is difficult to model (PDB ID: 3WFC), but detailed analysis by Sato et al. concluded that, while ambiguous, the electron density is only suitably accounted for with four non-hydrogen atoms in the active site (a CO and two H₂O, or two CO ligands) (Figure 16C).¹⁶⁹ Similarly, acetaldoxime (CH₃CHNOH) binds to the BNC in PaNOR, and it can be considered an analogue to hyponitrite (ONNO²⁻), a key intermediate in all major NOR mechanistic proposals (see section 3.3). The acetaldoxime-bound structure (PDB ID: 3WFD) more clearly demonstrates that four non-hydrogen atoms can fit within the BNC of NOR, although the binding mode displays a relatively short Fe_b3–N(R)–O(H)–Fe_B bridging motif, which doesn’t necessarily support an interpretation where each metal can bind NO simultaneously (Figure 16D). It has been proposed that Glu211 may dissociate from Fe_B to increase the Fe–Fe distance, and facilitate binding of a second NO molecule during catalysis;^185,189 despite this proposal, all three of these reduced PaNOR XRD structures show Glu211 bound to Fe_B, with relatively short Fe–Fe distances of 4.2 – 4.4 Å.¹⁶⁹ Interestingly, flash photolysis vibrational spectroscopy of CO-bound Cu_ANOR from B. azotoformans clearly demonstrates that two CO molecules bind to heme b₃ and Fe_B simultaneously (see section 3.3.3);¹⁹⁰ although it should be noted that, based on the sequence and homology structure of this Cu_ANOR, there may be no fifth residue to coordinate Fe_B.¹⁷⁴

Table 3.

Representative NOR and HCO Heme-Nonheme Distances.

Class	PDB Code	Form	Nonheme metal	Distance (Å)
cNOR (Pa)	3O0R	oxidized	Fe	3.8
cNOR (Pa)	3WFB	reduced	Fe	4.2
cNOR (Pa)	3WFC	reduced, CO-bound	Fe	4.4
qNOR (Ax)	6QQE	oxidized	Fe	4.1
qNOR (Nm)	6L1X	oxidized, Zn(II)-inhibited	Fe	3.8
qNOR (Gs)	3AYF	oxidized, inactive	Zn	4.6
A-type CcO (Rs)	1M56	oxidized	Cu	4.8
A-type CcO (Rs)	3FYE	reduced	Cu	5.3
B-type CcO (Tt)	1EHK	oxidized	Cu	4.4
B-type CcO (Tt)	3EH3	reduced	Cu	5.1
C-type CcO (Ps)	3MK7	oxidized	Cu	4.6

Variations in the Fe_B coordination geometry have been observed in crystal structures of qNORs from Geobacillus stearothermophilus (GsNOR), Neisseria meningitidis (NmNOR), and Acaligenes xylosoxidans (AxNOR), that may be relevant to our mechanistic understanding of NOR.^{167,170–172} qNOR proteins are single subunit (~85 kDa), with 14 transmembrane α-helices arranged in a pattern similar to the NorB/NorC architecture of cNORs. The Zn^II-inhibited crystal structure of NmNOR (PDB ID: 6L1X) displays Fe_B coordinated by only three His residues and a μ₂-O to heme b₃, where conserved Glu residues near the BNC bind to a Zn^II ion instead (Glu494, 498, and 563 with NmNOR numbering) (Figure 16F).¹⁷² While this structure is not catalytically relevant due to the presence of Zn^II interacting with residues in the PT pathway to the BNC, it demonstrates that Fe_B remains present in NOR without Glu coordination. A recent cryo-EM structure of active NmNOR without Zn^II, although at a very low resolution of 9 Å, appears to display a relatively weak monodentate Glu494–Fe_B interaction (Fe–O ~ 2.4 Å), while also having a relatively large Fe–Fe distance (~4.5 Å), which supports the hypothesis that Zn^II interferes with Glu494 binding to Fe_B in the previous NmNOR structure (Figure 16G).¹⁷² A separate cryo-EM structure of AxNOR (3.2 Å resolution) further demonstrates that Glu coordination is not necessary for Fe_B binding (Figure 16H).¹⁷¹ The structure of the oxidized enzyme also displays a larger Fe–Fe distance of 4.1 Å, compared to 3.9 Å in PaNOR, supporting the notion that Glu dissociation could play a role in increasing the distance between the Fe centers in the BNC. Importantly, this structure is a catalytically active form of qNOR. It is known that mutation of the nearby Glu490 residue in AxNOR (analogous to Glu211 in PaNOR) greatly diminishes catalytic activity, consistent with a crucial role of this residue during catalysis, either in stabilizing Fe_B and/or facilitating PT.^185,189 while the structural studies of NOR have revealed a range of coordination geometries that Fe_B can adopt, it is currently unclear whether this possible lability in Glu coordination has any functional significance.

3.2. Other Enzymes that Catalyze NO Reduction.

Bacterial NORs are not the only metalloenzymes that have evolved to achieve the selective reductive coupling of NO to N₂O. The active sites of the other NO-reducing enzymes display a diversity of structures, which highlights the versatility within living organisms towards developing transition metal catalysts (Scheme 18).¹⁹¹ The scope of this review, concerning NO and O₂ reduction reactions, is focused on the heme/nonheme active sites of the HCO superfamily and the following section will provide only a brief overview of the other NO reducing enzymes that have been described. More extensive reviews detailing their structure and mechanisms have been published.^{82,165,192–194}

Scheme 18.

Metalloenzyme Active Sites of Native and Promiscuous NO Reductases.

Two well-studied examples of NO reducing enzymes are flavodiiron NO reductases (FNORs), which contain a bimetallic nonheme Fe active site,^82,83,193 and fungal NOR enzymes called P450nor, which contain a single cysteine-coordinated heme b.^165,195 A common feature of these enzymes is that they belong to protein superfamilies that catalyze either NO or O₂ reduction (NOR and HCO, FNOR and FDP, P450nor and P450s).^82,165,196 This is particularly intriguing since there is no a priori reason to expect the 2 e⁻ bimolecular coupling of NO and 4 e⁻ reduction of O₂ would share sets of similar active sites in nature. More recently, NO reductase activity has also been described in a protein bearing a unique tetranuclear FeS cluster, denoted the hybrid cluster protein (Hcp);¹⁹⁷ however, biologically-relevant in vivo NOR activity by Hcp is currently inconclusive.¹⁹⁸ There are other examples of metalloproteins that appear to have promiscuous NO reductase-like activity in vitro, although the biological relevance of these reactions is unclear. Nevertheless, these metalloenzymes also add to the diversity of biological active site structures that can achieve NO coupling. These enzymes include close relatives to native NOR enzymes (i.e. HCO, or O₂-reducing flavodiiron proteins),^82,199,200 and certain nitrite reductase hemoproteins, like cytochrome c nitrite reductase.²⁰¹

3.2.1. Flavodiiron Nitric Oxide Reductase (FNOR).

FNOR enzymes are expressed by certain bacteria under conditions of high concentrations of NO, as a means of detoxification. NO production is part of mammalian immune response to pathogenic organisms; therefore, FNOR are part of the defense strategy of many pathogenic bacteria.¹⁹³ Instead of a heteronuclear heme-nonheme BNC as in bacterial NOR, the FNOR active site is comprised of two nonheme Fe centers, which are each coordinated by one or two His, a Glu/Asp carboxylate, and a bridging Asp residue, similar to many binuclear nonheme Fe proteins.²⁰² NO is proposed to bind to each Fe prior to N–N bond formation, similar to the trans mechanism in bacterial NORs (see section 3.3),²⁰³ although alternative mechanisms have been considered.^165,204 It is unknown whether electron donation from the FMN cofactor to the putative Fe–NO species is required prior to N₂O formation, or if it regenerates the active diferrous state after N–N bond formation occurs.¹⁹³ Efforts towards synthetic models of FNOR have demonstrated the feasibility of many of these mechanistic proposals, with diiron dinitrosyl and diiron mononitrosyl complexes producing N₂O upon 1 or 2 e⁻ reduction.^205–207

3.2.2. Cytochrome P450 Nitric Oxide Reductase (P450nor).

P450nor catalyzes NO reduction in denitrifying fungi using a single heme cofactor without the need for another nonheme Fe.^195,208 These enzymes belong to the cytochrome P450 family of enzymes, which are widely known for coupling O₂ reduction to the oxidation of organic substrates.^65,209 Like other P450 enzymes, the active site of P450nor is a cysteine-coordinated heme b cofactor. P450nor is distinguished from other P450s by a more open distal heme pocket, which facilitates binding of NAD(P)H next to the metal center.^210,211 Detailed mechanistic studies of P450nor favor a unique direct hydride transfer mechanism, from the NAD(P)H cofactor to the NO ligand that binds to the initial Fe^III heme b.^212,165,213 The resulting HNO(H) (nitroxide or hydroxylamide) species is proposed to remain bound to Fe, and produce N₂O upon reaction with an incoming NO molecule.^214,215 Recent efforts to mimic P540nor with synthetic models by the Lehnert group have demonstrated the importance of axial thiolate ligation on Fe–NO binding, and characterization of an analogue to the putative hydroxylamide intermediate.^216,217

3.2.3. (Putative) Nitric Oxide Reductase Activity of Hybrid Cluster Protein (Hcp).

Hcp was first isolated in 1992 from a Desulfovibrio vulgaris substrain.²¹⁸ Since then, decades of study of Hcp has led to extensive spectroscopic and structural characterization of its metal cofactors; however, determination of the biological role of this protein remains elusive.¹⁹⁷ It is known that this protein bears a unique FeS cluster that contains bridging-oxido ligands, best formulated as [Fe₄S₂O₃] (with a unique persulfido cysteine) in its oxidized states, and [Fe₄S₃] composition in its reduced form.^219,220 In vitro catalytic activity towards various small molecules, including NO,²²¹ NH₂OH,²²² and H₂O₂,²²³ have been reported, along with auto-S-nitrosothiol formation.²²⁴ It is unclear if any of these reactions are relevant to its biological function; however, studies over the past 5 years support Hcp playing a role in protecting the cell against nitrosative stress under anaerobic conditions.¹⁹⁸ A recent study of the molecular mechanism of NO reduction by Hcp confirms it is competent to catalyze reductive NO coupling, and also identified a putative dinitrosyl intermediate by EPR spectroscopy.²²¹

3.3. Mechanism of NO Reduction by Bacterial NORs.

The precise catalytic mechanism of the heme/Fe_B active site in bacterial NOR enzymes remain elusive. One complication to our mechanistic understanding of NORs is the substrate inhibition it displays, which makes it challenging to interpret which reactions with NO are catalytically relevant.^225–227 Extensive spectroscopic and computational studies have focused on three major mechanisms of N₂O formation (Scheme 19). These are the so-called (i) cis-Fe_B, (ii) cis-heme b₃, and (iii) trans mechanisms, which are named based on the structure of the putative dinitrosyl intermediate that forms prior to N–N bond formation. The section will describe each mechanism, along with supporting data obtained from studies of native NORs.

Scheme 19.

Proposed Mechanisms of Bacterial NOR.

Common to all mechanistic proposals is the initial 2 e⁻ reduction of the NOR resting state (Fe_b3^III–μ₂-O–Fe_B^III) with the transfer of 2 H⁺, to arrive at the reduced Fe_b3^II/Fe_B^II state. Binding of NO to the BNC forms an Fe–NO adduct at one or both of the Fe centers. Due to the nature of NO acting as a redox ‘non-innocent’ ligand to transition metal complexes, further discussion of Fe–NO species will use Enemark-Feltham notation, which distinguishes M–NO species based on the sum of electrons occupying the metal d and N–O π* orbitals.²²⁸ For example, binding of gaseous NO (which contains a single unpaired electron in the N–O π* orbital) to Fe^II ion (d⁶) is described as an {FeNO}⁷ species, which can be subdivided into the formal metal–ligand oxidation state assignments of Fe^I–NO⁺, Fe^II–NO^•, or Fe^III–NO⁻—although in many cases, the precise description will be somewhere between these extremes.²²⁹

3.3.1. Cis-Fe_B Mechanism.

In this mechanism, two NO molecules bind to the nonheme Fe_B center forming a dinitrosyl species, Fe_B(NO)₂. From this intermediate, formation of a cis-hyponitrite (ONNO²⁻) occurs after electron transfer from heme b₃. Proton transfer to this species results in N–O cleavage and N₂O formation. This was an early proposed mechanism of NOR, based on the postulate that heme {FeNO}⁷ species are unreactive catalytic ‘dead-ends’ and was further supported by existing precedent of NO coupling with synthetic nonheme organometallic complexes,^12,230,231 even though later studies have demonstrated that heme {FeNO}⁷ species can be reactive under certain conditions.^232–234 Furthermore, detailed characterization of the bonding and reactivity of a synthetic mononuclear dinitrosyl Fe complex suggest there is a high inherent kinetic barrier to N–N coupling from Fe(NO)₂ species, due to the parallel alignment of spin on NO from coupling to the unpaired electrons on Fe.²³⁵ Later, a modified cis-Fe_B mechanism was proposed, which suggest free NO attacks a nonheme {FeNO}⁷.²²⁵ Supporting data for this mechanistic proposal include the observation of a significantly lower reduction potential for heme b₃ in PdNOR than Fe_B or the other heme cofactors.¹⁸⁶ This has led some to suggest that semi-reduction of the BNC, and binding of NO at Fe_B^II while heme b₃ remains Fe^III, is catalyticaIly relevant;²³⁶ however, recent studies of electrocatalytic NO reduction by PaNOR shows that heme b₃ must be reduced for catalysis.²³⁷ Furthermore, electrochemical studies with PdNOR and CO suggest substrate binding can raise the potential of heme b₃, and may serve as a way to control its activity.²³⁸ Much of the uncertainty about the functionally relevant redox states of the NOR BNC stems from the inconsistent reduction potential values observed for Fe_B. A number of studies have measured a very low potential for Fe_B (−0.37 V, below the −0.16 V potential of heme b₃) by cyclic voltammetry²³⁹ and supported by Mössbauer spectroscopy of Pseudomonas nautica NOR (PnNOR).²⁴⁰ Despite these findings of the low Fe_B reduction potential, relatively high potentials have been observed with spectroelectrochemical experiments of PdNOR and PaNOR^186,237 which happen to be more consistent with potentials observed in biosynthetic and computational model studies.^117,241 Further studies are necessary to establish any mechanistic relevance of partially reduced states of the BNC, based on the relative order of reduction potentials of Fe_B and heme b₃. Based on the measured potentials of hemes b and c in the ET pathway, structural changes to the BNC are likely necessary for ET to take place for NO reduction, which will have significant implications for the reduction potentials of the BNC metal centers.

There is spectroscopic evidence that heme b₃ remains six-coordinate upon reduction, binding water or hydroxide, and may participate principally as an electron donor to Fe_B.²⁴⁰ This proposal was based on study of PnNOR, which displays differences in the properties of its heme b₃ compared to other NORs, including a low spin resting state,²⁴⁰ and a higher reduction potential than Fe_Bthat is also insensitive to NO.²³⁹ It is conceivable that sufficient differences in primary sequence among NOR enzymes can lead to a difference in mechanism, which can possibly explain these disparate observations between PnNOR and other NORs. Recent computational modelling of the BNC active site during turnover determined that the cis-Fe_B mechanism, specifically binding of NO to Fe_B and the resulting hyponitrite geometry upon N–N coupling, is energetically less favorable than other mechanisms.²⁴² Differences in the coordination number or spin state of Fe_B and heme b₃ would be expected to alter the potential energy landscape of turnover, possibly in ways that could favor the cis-Fe_B mechanism.

3.3.2. Cis-Heme b₃ Mechanism.

Similar to the modified cis-Fe_B mechanism, the cis-heme b₃ mechanism proposes NO binds only to the heme b₃ cofactor, with a subsequent electrophilic attack of incoming NO to the resulting heme {FeNO}⁷. Experimental data distinctly supporting this mechanism over the trans mechanism (vide infra) is sparse.¹⁶⁴ NO reduction by other metalloproteins at a single heme site, such as P450nor or nitrite reductases, certainly provide general precedent for this mechanism, although in some cases hydroxylamine (NH₂OH) is formed instead of N₂O.^165,195,243 Recent characterization of PdNOR isolated in the absence of NorD and NorQ—two proteins necessary for insertion of the Fe_B cofactor—unambiguously demonstrate the essential role of nonheme Fe for NOR activity.²⁴⁴ Possible roles of Fe_B in the cis-heme b₃ mechanism, then, include: promoting N–O activation at heme {FeNO}⁷, stabilizing the hyponitrite intermediate, and electron donation at some step after N–N bond formation. Time-resolved spectroscopy of NOR under single turnover conditions support that NO binds preferentially to heme b₃.^245,246 Lack of NO binding at Fe_B could be a consequence of the relatively small space between Fe in the BNC that has been observed in structures of NORs. Reduced and ligand-bound crystal structures showed evidence that four atoms may fit in the reduced BNC (see section 3.1), but modeled distances between these four atoms are unreasonably short (~1-2 Å) to be consistent with two different diatomic molecules.¹⁶⁹ Notably, nearly every computational study of the NOR catalytic mechanism has found that the cis-heme b₃ mechanism is the most energetically favorable.^{241,242,247–249} Furthermore, calculated vibrational frequencies for a cis-hyponitrite originating from this mechanism better match with an assigned N–N stretching vibration from a spectroscopically observed hyponitrite species in NOR (υ_N–N: 1332 cm⁻¹);²⁵⁰ the calculated frequency is closer to the experimental value by about 400 cm⁻¹, compared to the calculated trans-hyponitrite in the trans mechanism.²⁴² It should be noted, however, that this assigned hyponitrite signal is in very low abundance and difficult to confirm its position or identity with confidence; similarly, vibrational frequency calculations are highly dependent on the model and basis set used and, while they can reflect relative changes in vibrational frequencies, do not typically predict experimental IR stretches with absolute accuracy.

3.3.3. Trans Mechanism.

The trans mechanism hypothesizes that NO binds to both heme b₃ and Fe_B before N–N coupling. Spectroscopic evidence for this trans dinitrosyl intermediate was reported in 2004: rapid freeze quench EPR of reduced PaNOR in the presence of NO displays new signals at g ~ 4 and g ~ 2.01, which are consistent with nonheme and heme {FeNO}⁷ EPR signals, respectively.²⁵¹ These signals disappear upon annealing at higher temperatures, suggesting they are intermediates relevant to N₂O formation. However, it was noted in the original report that these signals represent only ~30% of the total BNC, and later re-evaluation of these spectra has cast doubt on their relevance to the NOR catalytic cycle.¹⁸⁹ This is due, in part, to synthetic and computational modelling of NOR, which predicts an EPR silent trans dinitrosyl intermediate, because of exchange between the two {FeNO}⁷ centers. The unambiguous identification of a trans-dicarbonyl species in a Cu_ANOR strongly supports the proposal that a trans-dinitrosyl intermediate is achievable in the BNC.¹⁸⁹ Confirmation that vibrational signals in the dicarbonyl of Cu_ANOR arise from the heme and non-heme carbonyl species within a single enzyme, as opposed to a mixture or equilibrium of monocarbonyl species, was obtained by signal integration relative to total enzyme concentration, along with low-temperature identification of a semi-bridging carbonyl form originating from the heme carbonyl, which has a measurable effect on the nonheme Fe carbonyl υ_C–O.¹⁹⁰ Notably, the analogous experiments with PdNOR is more consistent with a mixture of a single CO binding to each metal center,¹⁸⁹ possibly indicating substrate binding variation among the NOR classes. As mentioned previously in section 3.1, the Cu_ANOR BNC site may be uniquely more open as a consequence of differences in the coordination of Fe_B.

3.3.4. Steps After N–N Bond Formation.

After formation of the hyponitrite intermediate, through any of the proposed pathways (cis-Fe_B, cis-heme b₃, or trans), N–O bond cleavage is required to complete the catalytic cycle. Three possibilities for this process include: (i) reformation of the oxidized resting state of NOR (Fe_b3^III-μ₂-O-Fe_B^III), followed by subsequent proton/electron transfer to regenerate the reduced state, (ii) proton transfer steps occurring at the hyponitrite intermediate, and concomitant loss of H₂O and N₂O to obtain a vacant oxidized state (Fe_b3^III and Fe_B^III), which is later reduced to turn over the enzyme, or (iii) an intermediate mechanism, where one proton is transferred to the hyponitrite, inducing N–O cleavage and forming an Fe^III–OH (bridging or terminal) and N₂O, following protonation/reduction in later steps (Scheme 20). Recent, stopped-flow time resolved spectroscopy of NOR supports either proton-dependent pathway ii or iii, based on an observed pH effect of N₂O formation.²⁴⁶ A separate study concluded, however, that no protons are transferred from bulk solvent during the first turnover, suggesting pathway i may be operative under certain conditions.²⁵² Freeze-quench EPR of NOR under single turnover shows an increase in g ~ 6 and g ~ 4 signals upon warming to room temperature, which are interpreted as formation of the uncoupled oxidized heme b₃ and Fe_B after NO reduction.²⁵¹ These signals are also seen in EPR spectra of the NOR resting state in small amounts and have also been attributed to inactive states of the enzyme or impurities.^164,251,253 It should be noted that formation of oxidized BNC with a μ₂-O ligand is unobservable by EPR. Computational models of NOR show it is energetically feasible to cleave the N–O bond of hyponitrite without any proton transfer. In fact, this mechanism is favored, because the vacant oxidized state of the BNC is very high in energy.^241,242,249 Multiple PT pathways have been proposed in NORs, with major differences between nonelectrogenic cNOR and proposed electrogenic PT pathways in qNOR from the cytoplasmic and not periplasmic side, which introduce certain thermodynamic constraints involved in energy conservation.^166,170,172 Therefore, it is possible that the precise mechanism(s) of N–O cleavage that occur during turnover may vary between NOR subclasses, or reaction conditions.

Scheme 20.

Proposed N–O Bond Cleavage Pathways in Bacterial NORs.

3.4. Insights Gained from Biomimetic Models of Bacterial NORs.

Like HCOs, bacterial NORs are large, membrane bound enzymes containing multiple cofactors, making it difficult to carry out biochemical and biophysical studies due to the challenge in purifying NORs with high homogeneity and yields, along with studying them spectroscopically without interference from other cofactors. Therefore, biomimetic modelling through simpler and smaller compounds or proteins that are free of other cofactors have played key roles in understanding structural features responsible for the activity and mechanism, similarly to the success demonstrated in biometric modelling of HCOs. Since there remain many uncertainties concerning the molecular mechanism of NOR (see section 3.3), it is tempting to look to these simpler models for answers to these mechanistic questions. However, as explained by Kim et al. in their review of synthetic HCO model complexes, models, by their very nature, cannot conclusively prove a mechanism of a native protein, but instead their goal is to “sharpen or focus relevant questions” for further study of the native system.¹¹¹

In this review, we will provide an overview of models that mimic bacterial NORs both structurally and functionally, because these models provide insight into reactivity and mechanisms concerning NO binding, N–N bond forming, and N–O bond cleaving chemistry relevant to NOR. Literature concerning the properties of heme and nonheme Fe–NO complexes is vast, and many extensive reviews surveying NOR-relevant {FeNO}⁷ complexes have also been reported.^{165,179,192,194,229,254–256}

3.4.1. Structural and Functional NOR Models.

An early reported synthetic heme/nonheme complex with putative NOR-like activity was disclosed in 2000 by Ju et al (Scheme 21).²⁵⁷ A 5-coordinate Fe-porphyrin complex tethered to a nonheme Fe is reduced in the presence of dithionite to afford a vacant diferrous complex, [(L14)Fe^II/Fe^II(Cl)]⁺. Based on UV-Vis spectroscopy, at low concentrations, this species converts back to the oxidized μ₂-O complex upon addition of excess NO. Performing the same experiment at higher concentrations, for identification of gaseous products by GC-MS, led to both N₂O and NO₂ formation along with a new Fe-containing product with a diferrous oxidation state.²⁵⁷ The change in overall reaction products based on experimental conditions complicates validation that reductive NO coupling occurs with this complex at low concentrations, or whether an alternative mechanism is operative, such as NO disproportionation.

Scheme 21.

Functional Heme/Fe NOR model reported by Ju et al.²⁵⁷

Mimicking the structure of the BNC was also pursued through ‘biosynthetic’ modelling using small and robust heme proteins, such as myoglobin, as modelling scaffolds based on the hypothesis that introducing a nonheme metal site into such a protein—one that is easier to prepare and study than native NOR—would be able to suitably model the structure and function of the native enzyme. In 2006, Zhao et al. reported that swMb which has been engineered to contain a nonheme binding pocket similar to the Cu_B binding site of HCO enzymes (Cu_BMb) could catalytically reduce NO to N₂O.²⁵⁸ The production of N₂O requires the presence of Cu^I; the hemoprotein-only or Zn^II-bound forms are catalytically inactive. Later in 2009, another engineered mutant of myoglobin (Leu29His, Phe43His, Val68Glu swMb; Fe_BMb) was reported to be able to bind Fe^II and subsequently react with NO to form N₂O.²⁵⁹ Comparison of the crystal structure of Fe_BMb (PDB ID: 3K9Z) with the structure of PaNOR reported a year later (PDB ID: 3O0R) showed close agreement between the placement and coordination geometry of the nonheme Fe and the native Fe_B cofactor (Figure 18).²⁶⁰ It should be recognized that similar to the functional model by Ju et al., it remains unclear whether N₂O forms via a 2 e⁻ reductive coupling of NO in Fe_BMb. One e⁻ oxidation of the heme, based on the shift in its Soret peak, has been confirmed in single-turnover reactions, but no spectroscopic evidence suggesting oxidation of the nonheme Fe has been reported.^259,261 Later catalytic studies demonstrate that the absence of nonheme Fe^II does not halt N₂O formation, and Fe^II can even be substituted for redox-inactive Zn^II and still produce similar amounts of N₂O.²⁶² These data suggest that Fe_BMb, is capable of performing 1 e⁻ reductive disproportionation of NO (eq 3), while its potential to follow a 2e⁻ reduction mechanism (eq 2) under other reaction conditions has not been established.

Figure 18.

Overlay of Fe_BMb structure (cyan; PDB ID: 3K9Z) and PaNOR (yellow; PDB ID: 3O0R). Figure adapted from ref. ²⁶⁰ Copyright American Chemical Society 2015.

A NOR model system which unambiguously performs 2 e⁻ reduction of 2 NO to N₂O has been reported by Collman et al.²³³ It is a synthetic diferrous heme/nonheme complex which forms N₂O quantitatively upon addition of excess NO (Scheme 22). The resulting metal-containing species is assigned to a [(L15){FeNO}⁶/Fe^III–OH]²⁺ complex, based on UV-Vis and EPR spectroscopy, consistent with the oxidation of the initial complex by 2 e⁻. NO reduction does not occur with the mixed-valent complex ([(L15)Fe^III/Fe^II(Cl)]⁺), or the complex without nonheme Fe ((L15)Fe^M), demonstrating the important role of the two redox active metals in this model complex.²³³

Scheme 22.

Structural and Functional NOR Model Reported by Collman et al. and Mechanism of N2O Formation.²³³

3.4.2. NO Binding at Biomimetic NOR Models.

Mechanistic studies of NO reduction by structural and functional NOR model systems are aided by the fact that metalation of the porphyrin and/or nonheme binding sites can be controlled in a systematic way. For example, Collman et al. demonstrated that Zn^II or Fe^II can be selectively inserted into the porphyrin of their tethered heme/nonheme ligand L15.²³³ Further metalation of the nonheme binding site with Zn^II or Fe^II allows for characterization of stable NO adducts of the heme and nonheme Fe, separately. The (L15){FeNO}⁷ complex displays a υ_N–O of 1630 cm⁻¹ and a υ_Fe–N at 581 cm⁻¹ consistent with a 6-coordinate heme S = 1/2 {FeNO}⁷. These IR stretches change little in the presence of Zn^II in the nonheme site (1635 cm⁻¹ and 583 cm⁻¹, respectively), suggesting no significant interaction between the {FeNO}⁷ and the nonheme metal.²³⁴ Similarly, it is possible to characterize NO binding in the [(L15)Zn^II/{FeNO}⁷]²⁺ complex; the υ_N–O is 1810 cm⁻¹. An EPR signal for the nonheme {FeNO}⁷ is observed at g ~ 4, consistent with an S = 3/2 complex. Characterization of the separate heme/nonheme {FeNO}⁷ complexes allowed for a meaningful analysis of the low-temperature intermediates that form when the diferrous complex reacts with NO. At −80 °C in dichloromethane, the first intermediate has EPR and resonance Raman (rR) spectral features nearly identical to the [(L15)Zn^II/{FeNO}⁷]²⁺ complex. Warming to −40 °C leads to disappearance of the EPR signal, and a rR band at 587 cm⁻¹ which is close to the Fe–N stretch of the heme {FeNO}⁷. The authors interpret this data as consistent with a trans dinitrosyl intermediate, which is EPR silent due to exchange between the heme and nonheme {FeNO}⁷ spin systems. Warming this intermediate to room temperature produces the previously mentioned oxidized complex ([(L15){FeNO}⁶/Fe^III–OH]²⁺) (Scheme 22). These data support a mechanism similar to the proposed trans NOR mechanism (see section 3.3.3) for this functional model.

Similar heme/nonheme substitution studies were conducted on the engineered Fe_BMb protein by Chakraborty et al.²⁶³ Substitution of heme b for Zn(protoporphyrin IX), followed by binding of nonheme Fe^II and incubation with excess NO, produces an EPR signal at g ~ 4, consistent with a S = 3/2 {FeNO}⁷. QM/MM studies of the Fe-NO species led the authors to favor a description of the {FeNO}⁷ as an anti-ferromagnetically coupled high-spin Fe^II–NO^• species; however, it should be noted that the spectral properties (namely, the υ_N–O, ⁵⁷Fe Mössbauer, and EPR spectra) of this species conforms to nearly all other characterized nonheme S = 3/2 {FeNO}⁷ complexes, which have also been described as high-spin Fe^III–NO⁻ by DFT.^194,255 Careful addition of 1 equivalent NO to Fe_BMb, or its more reactive mutant Fe_BMb Ile107Glu,²⁶⁴ leads to NO binding exclusively to the heme cofactor. Fe_BMb is able to bind a variety of nonheme metals, including Fe^II, Cu^I, or Zn^II, which have different effects on the degree of N–O bond activation by the heme cofactor.²⁶⁵ With no nonheme metal, the heme {FeNO}⁷ of Fe_BMb displays a υ_N–O of 1601 cm⁻¹, similar to other six-coordinate S = 1/2 {FeNO}⁷ heme complexes.^229,254 The presence of Fe^II in the nonheme site significantly decreases the υ_N–O to 1549 cm⁻¹. A similar effect is observed for Zn^II in the Ile107Glu mutant (υ_N–O of 1550/1577 cm⁻¹), but not Cu^I (υ_N–O of 1601 cm⁻¹). Recently, Abucayon et al. have demonstrated a similar effect of exogenous Lewis acids on {FeNO}⁷ heme complexes.²³² Addition of BF₃•OEt₂ to (OEP){FeNO}⁷ (OEP = octaethylporphyin) leads to weakening of the N–O bond, as evidenced by a dramatic decrease in the υ_N–O by ~ 200 cm⁻¹ (Scheme 23). Interestingly, this shift in υ_N–O is roughly equivalent to related anionic {FeNO}⁸ heme complexes, suggesting a suitable Lewis acid can be as activating as reducing the Fe-NO moiety by one e⁻.²³² The activating effect of this Lewis acid leads to N₂O formation via reductive disproportionation, from an initially stable {FeNO}⁷ complex. A weaker Lewis acid, [K(2.2.2-cryptand)]⁺, can similarly induce N–N coupling, albeit giving a lower yield of N₂O.^232,266 These studies suggest that one of the potential roles of Fe_B in NOR may be to promote the reactivity of the heme nitrosyl by acting as a Lewis acid, which would be relevant to the proposed cis-heme b₃ pathway.

Scheme 23.

Lewis acid-assisted NO Reduction by (OEP){FeNO}⁷.²³²

Mechanistic study of the NOR-like activity of Fe_BMb Ile107Glu with excess NO shows that, similar to the [(L15)Fe^II/Fe^II]²⁺, NO binds initially to the nonheme Fe.²⁶¹ This assignment is based on rapid freeze quench rR spectra of Fe_BMb Ile107Glu, which show a υ_N–O band at 1755 cm⁻¹ at the earliest time point (6 ms) upon NO addition. This intermediate leads to the formation of a trans dinitrosyl, with six-coordinate heme {FeNO}⁷ and nonheme {FeNO}⁷, identified by growth of the heme υ_Fe–N at 568 cm⁻¹. From the trans dinitrosyl intermediate, Fe_BMb Ile107Glu either decays to an oxidized heme species with a rate of ~0.7 s⁻¹ at 4 °C, or shifts to a ‘dead-end’ state containing a five-coordinate heme {FeNO}⁷. The main source of this deactivation appears to be due to attack of NO on the proximal face of the heme.²⁶⁷

Lastly, while reductive NO-coupling could not be unambiguously confirmed from the tethered [(L14)Fe^II/Fe^II(Cl)]⁺ complex, initial attempts to characterize the reaction intermediate at low temperature by UV-Vis were consistent with a mono or dinitrosyl heme species. It is unclear whether the nonheme Fe binds NO (UV-Vis absorbance is not very sensitive to binding of NO by nonheme Fe). Surprisingly, when these experiments were conducted on a compositional isomer of L14 (where the tether attaches to the 6-position of the pendant pyridyl moiety, instead of the 5-position; [(L3)Fe^II/Fe^II(Cl)]⁺), no NOR-like activity was observed.²⁶⁸ Instead, a stable dinitrosyl species forms, based on the presence of two υ_N–O bands in the FTIR at 1689 and 1798 cm⁻¹ for heme and nonheme {FeNO}⁷ moieties, respectively (Scheme 24). Characterization of the heme {FeNO}⁷ in the absence of the nonheme Fe shows almost no difference in the υ_N–O (1683 cm⁻¹), consistent with little effect of the nonheme Fe on NO activation in this model system. Since the heme {FeNO}⁷ complexes are 5-coordinate, one possible explanation for its inactivity is that NO binds at the opposite face of the tethered nonheme, preventing N–N bond formation.²⁶⁸ Similarly, Berto et al. have reported that combining a six-coordinate heme {FeNO}⁷ complex with a nonheme {FeNO}⁷ model compound of the Fe_B center displays no intermolecular N₂O formation (Scheme 25).²⁶⁹ Overall, these studies suggest that there are certain geometric or electronic requirements that must be met before bimetallic N–N coupling occurs from a putative trans dinitrosyl intermediate in NOR, or its functional models.

Scheme 24.

Proposed structures of Stable Dinitrosyl Complex [(L3){FeNO}⁷/{FeNO}⁷(CI)]⁺²⁶⁸

Scheme 25.

Attempted intermolecular Reactivity of Synthetic Fe_B Model {FeNO}⁷ and Heme {FeNO}⁷.²⁶⁹

3.4.3. N–N Bond Formation.

Direct studies of N–N bond formation in functional NOR models are challenging due to the relatively fast kinetics of the steps after binding of the second NO.^234,261 For Fe_BMb, Bhagi-Damodaran et al. have recently shown that the rate of decay of the heme-nitrosyl intermediate during N₂O formation is highly dependent on the electronics of the heme cofactor.²⁷⁰ By substituting different heme cofactors (heme b substituted for mono-formyl, and di-formyl analogues) and introducing a mutation to create a hydrogen bond to the δ1 NH of the proximal His (Leu89Ser), the reduction potential of the heme cofactor could be tuned from −130 mV to +148 mV (vs SHE) (Figure 19A). The rate of heme {FeNO}⁷ decay varies over 2 orders magnitude based on these differences in heme reduction potential (Figure 19B). The degree of N–O bond activation is also impacted, with the most reducing {FeNO}⁷ displaying a υ_N–O of 1527 cm⁻¹, which increases up to 1570 cm⁻¹ in the highest reduction potential cofactor.²⁷⁰ These results demonstrate the crucial role of the heme reduction potential on the nature of the {FeNO}⁷ and the rate of N–N coupling. For further discussion of the importance of heme reduction potential on NOR activity, see section 5.1.2.

Figure 19.

(A) Heme substitution and mutagenesis of Fe_BMb creates a range of heme reduction potentials that span the range that is observed with the HCO superfamily. (B) Rates of NO binding, heme-NO decay, and electron transfer based on heme reduction potential. (C) NO reduction activity of FeBMb with range of heme reduction potentials. Figure adapted from ref. ²⁷⁰ Copyright the National Academy of Sciences 2018.

More general reviews of N–N bond formation by transition metal-nitrosyl complexes have been published.^165,256,271 Among these complexes, a relevant example is the Tp₂{Ru₂(NO)₂}¹²(μ-Cl)(μ-pz) complex (Tp = trispyrazolylborate; pz = pyrazolate), which can achieve reversible interconversion of dinitrosyl to O=N–N=O complexes upon 2 e⁻ reduction/oxidation (Scheme 26).²⁷² This is a rare example of well-defined radical NO coupling from two {M-NO}⁷ species, and is related to proposed N–N bond forming steps in the trans mechanism of bacterial NORs, along with some proposed dinitrosyl mechanisms of nonheme diiron FNORs (see section 3.2.2). The bridging O=N–N=O complex, (TpRu)₂(μ-Cl)(μ-pz)(μ-N(=O)–N(=O)-k²-N,N′), is capable of N₂O formation, and produces a μ₂-O dimer upon addition of strong acid.²⁷² it is interesting that protons are required, despite the balanced reaction having no protons being transferred. Theoretical investigations on the role of protons on N₂O formation suggest they are important for the second N–N bond forming step, facilitating the conversion of the O=N–N=O species to hyponitrite (⁻O–N=N–O⁻), which is a prerequisite to evolve N₂O.²⁷³

Scheme 26.

Reversible Radical Coupling of {(RuNO)₂}¹² Upon 2 e⁻ Reduction and Proton-Induced N–O Cleavage.²⁷²

3.4.4. N–O Cleavage.

Formation of a hyponitrite intermediate is a ubiquitous step of NO coupling mechanisms. Biomimetic models of NOR display both proton-dependent and -independent mechanisms of cleaving one of the N–O bonds in hyponitrite to release N₂O. Both of the tethered heme/nonheme NOR functional models accomplish NO reduction in organic solvent, without the addition of exogenous protons.^233,257 By UV-Vis absorbance spectroscopy, Ju et al. observed formation of a [(L14)Fe^III–μ₂-O–Fe^III(Cl)]⁺ species, a model of the NOR enzyme resting state, after reduction of NO.²⁵⁷ Similarly in the [(L15)Fe^II/Fe^II]²⁺ functional NOR model, Collman et al. observed that one of the oxygen atoms remains bound to the metal complex, forming a nonheme Fe^III–OH, which is confirmed by its comparable spectroscopic features to an independently synthesized analogue, (L15){FeNO}⁶/Fe^III–OMe.²³³ In contrast to these examples, Wang et al. showed that protons are necessary to promote NO reductive coupling, over competing reductive disproportionation, with both synthetic tethered and bimolecular heme/Cu model systems (Scheme 27).^274,275 The authors propose that protons are involved in the N–O cleaving step of N₂O formation, because a [(L3)Fe^III-μ₂-O-Cu^II]⁺ species is never observed in these reactions, despite multiple examples of such complexes being stable and isolable.

Scheme 27.

Formation of N₂O Upon Addition of Cu^I and H+ to a {Fe(NO}2}8 Complex.²⁷⁴

A number of well-defined binuclear metal-hyponitrite complexes require protons for N–O bond cleavage. Xu et al. have synthesized a stable binuclear heme hyponitrite species by addition of hyponitrous acid (HO–N=N–OH) to ((OEP)Fe^III)₂(μ₂-O) (Scheme 28).²⁷⁶ Addition of strong acid would induce N–O cleavage, with corresponding N₂O and H₂O formation. Surprisingly, this hyponitrite species has also been observed to perform the microscopic reverse of N–N coupling to form two equivalents of (OEP){FeNO}⁷ upon slow thermal decomposition.²⁷⁷ Xu et al. more recently reported a modified bridging heme complex (OEP–CH₂–CH₂–OEP)(Fe^III)₂(μ₂–O) which forms another bridging hyponitrite complex in the presence of hyponitrous acid (as observed by IR spectroscopy).²⁷⁸ This complex, however, is unstable in the absence of excess hyponitrous acid, and releases N₂O to re-form the initial μ₂-O complex. The authors suggest a consequence of the ethylene bridge is moving the Fe centers closer together to destabilize the hyponitrite intermediate, an observation which may be a relevant to the close Fe–Fe distances observed in NOR structures.

Scheme 28.

Stable and Reactive Heme Hyponitrite Species Reported by Xu et al.^276,278

3.5. Summary and Outlook of NOR and Biomimetic Models.

Native bacterial NOR enzymes have proven to be challenging proteins to study. Like many other integral membrane enzymes, purification of stable, active NOR was an initial difficulty. Despite decades of study, it is only within the last 10 years that a structure of NOR has been reported. Since then, researchers have been able to obtain a handful of other NOR crystal structures, many of which display unique structural variations that may be relevant to catalysis. Another complicating factor includes the presence of multiple auxiliary heme and other redox-active cofactors necessary for electron transfer to the interior BNC, which can convolute spectroscopic study of the heme/nonheme active site. While the detailed molecular mechanism of NOR is not completely understood, much insight has been gained through the combination of spectroscopic, structural, and theoretical studies of the native enzymes. Concurrent study of simpler biomimetic models has allowed for further exploration of the mechanistic possibilities. Specifically, the model studies support the possible role of Fe_B to be promoting activation of heme-NO intermediate through acting as a Lewis acid, or by binding a second equivalent of NO and forming the N–N bond through a trans dinitrosyl. Further cleavage of the N–O bond can occur through proton dependent or independent pathways.

The following are examples of questions concerning the mechanism of NOR that are crucial to our full understanding of NOR:

1. What is the role of Fe_B in the catalytic cycle? Is it necessary for binding of NO and coupling of NO, via the proposed cis-Fe_B or trans mechanisms, or does it have a role in activating the heme {FeNO}⁷ for electrophilic attack by free NO, via the cis-heme b₃ mechanism? While the heme reduction potential has been shown to play a role in NOR reaction, what about the role of Fe_B potential?

2. Is the conserved Glu residue that is observed to coordinate Fe_B in the BNC always bound during the catalytic cycle? In addition to binding Fe_B in the resting state, does this Glu and other conserved Glu around the BNC have other roles in the catalytic mechanism, such as gating NO binding, PT, etc.?

3. Is the resting state of NOR (Fe^III–μ₂-O–Fe^III) relevant to the catalytic cycle? In what ways does the BNC depend on protons to cleave the N–O bond?

4. Is a single mechanism for N–N formation (cis-Fe_B, cis-heme b₃ or trans) and N–O cleavage operative in the majority of NORs or can the mechanism change between different enzyme sub-classes or catalytic conditions?

Biomimetic models have helped greatly in refining these questions and further studies stand to improve them moving forward.

4. Sulfite Reductases (SiR) and Related Biomimetic Models

4.1. Sulfite Reduction in Cellular Metabolism.

Sulfate-reducing bacteria are widespread among many of Earth’s environments, and more than 250 prokaryotic species with sulfate-reducing genes have been isolated and characterized.^10,279 These organisms can couple the reduction of sulfate and sulfite to energy production in the form of a proton motive force, analogous to the respiratory mechanisms of denitrification and aerobic respiration. A notable difference in respiratory sulfate reduction compared to aerobic respiration and denitrification is that the sulfate and sulfite reducing enzymes are globular proteins rather than integral membrane proteins. The mechanism of how these proteins produce a proton gradient for energy production is not completely understood, but the prevalent understanding is that specific respiratory membrane complexes are crucial for generating the proton motive force. These membrane complexes transfer electrons from the periplasm to sulfite reductase, concomitant with the release of protons into the periplasm via oxidation of membrane-associated quinols.²⁸⁰ Multiple respiratory membrane complexes have been observed to be associated with sulfate reducing pathways.²⁷⁹ The cellular pathway for sulfate respiration relies on three enzymes: ATP sulfurylase, 5′-adenylylsulfate reductase (APS reductase), and sulfite reductase (SiR). The initial steps of sulfate uptake require ATP sulfurylase, which generates 5′-adenylylsulfate (adenosine-5′-phosphosulfate, APS) from ATP and sulfate, followed by APS reductase, which transfers 2 e⁻ to APS, generating sulfite and AMP (Figure 20).²⁸¹ Sulfite reductase then performs the remaining 6 e⁻ and 7 H⁺ reduction of SO₃²⁻ to HS⁻ and water in a series of reductive dehydration steps.

Figure 20.

Enzymes involved in the sulfate reduction respiratory pathway. Figure adapted from ref. ²⁷⁹ Copyright Elsevier 2015.

Two different forms of sulfite reductase have been described in sulfate respiratory pathways. The most prevalent is known as dissimilatory sulfite reductase (dSiR), which contains a unique heme-like cofactor (siroheme) and a [4Fe-4S] cluster in its active site (Figure 21).^282,283 The proximal ligand to siroheme is also one of four cysteine ligands that coordinate the [4Fe-4S] cluster, creating a covalent bridge between the two iron cofactors. Another dSiR that has been identified is a member of the octaheme nitrite reductase family (ONR) known as multicytochrome c sulfite reductase A (MccA), or simply SiRA, which contains a heme c/Cu active site. The arrangement of heme and Cu is somewhat similar to HCO; however, the coordination of Cu is drastically different, using two Cys residues in a linear binding mode, instead of the His₂His-Tyr of HCO.^284,285

Figure 21.

Active site structures of siroheme-[4Fe-4S] SiRs (A) Dvir (PDB ID: 2V4J), (B) SiRHP (PDB ID: 1AOP), and (C) dSiR (3MM5).

Sulfite reducing enzymes with siroheme-[4Fe-4S] cofactors are also involved in the production of bio-available sulfur from sulfate/sulfite and are known as assimilatory sulfite reductases (aSiR). Sulfide generated from aSiR is the major sulfur source for biological sulfur accumulation in these organisms by the conversion of L-serine to L-cysteine via O-acetylserine (thiol)-lyase,²⁸⁶ and disruption of this pathway typically substantially inhibits growth. These enzymes are found across all domains of life but have been principally studied in bacteria and plants. The siroheme-[4Fe-4S] active site of aSiR has been studied extensively and is a major source of our spectroscopic and mechanistic understanding of this unique cofactor. aSiRs may be broadly grouped by whether they are NADPH-dependent (prokaryotic) or ferredoxin-dependent (Fd-SiR, eukaryotic). The NADPH-dependent enzymes are oligomeric and comprise electron acceptor subunits with FAD/FMN cofactors and catalytic hemoprotein (HP) subunits that bind the siroheme-[4Fe-4S] cofactors. The most well-studied NADPH-dependent SiR is the aSiR expressed in E. coli, which has an α₈ϐ₄ composition (a 780 kDa holoenzyme) with either one FAD or FMN in each NADPH-binding flavoprotein α subunit and one siroheme-[4Fe-4S] site per HP ϐ subunit; both subunits are ~60 kDa each.^287,288 NADPH binds and delivers electrons to the complex via an FAD flavoprotein subunit which subsequently reduces FMN. Electrons are transferred from the FMN subunit to the HP subunit where the siroheme-[4Fe-4S] cofactor is reduced by electron uptake first through the [4Fe-4S] cluster.²⁸⁸ Fd-SiR isolated from plants are monomeric (~65 kDa) and bear little sequence similarity to the catalytic subunits of the NADPH-dependent aSiR outside of the siroheme-binding domain.^289–291 A type of low molecular weight, monomeric aSiR (alSiR) has also been identified, such as the a ISiR from the Hildenborough strain of D. vulgaris. These much smaller enzymes (~27 kDa) are soluble monomers that exhibit spectroscopic features that are similar yet distinct from the oligomeric aSiRs, but they retain the same catalytic site architecture.^292,293

4.2. Structures of Sulfite Reductases.

4.2.1. Siroheme-[4Fe-4S] SiRs.

Dissimilatory enzymes (dSiR) like those identified in the sulfate reducing genera Desulfovibrio (desulfoviridin, Dvir),^{292,294–297} Desulfotomaculum,^298,299 and Archeoglobus^300,301 are crucial components of the extensive anaerobic sulfate respiratory apparatus of sulfate reducing organisms. These enzymes were initially called bisulfite reductase due to the prevalence of HSO₃⁻ at their slightly acidic pH optima (pH 6.0). They were also variably denoted by their siroheme pigmentation as desulfoviridin,^{292,295,297,302} desulforubidin,³⁰³ P-582,^298,299 or desulfofuscidin.^304–306 The active form of dSiR was first thought to be a tetramer with α₂ϐ₂ composition (DsrAB),²⁹² and each subunit was found to contain one HP and one ferredoxin-like domain, the latter of which is not present in aSiRs. Quantitative amino acid analysis later suggested that the enzyme was actually a hexamer of α₂ϐ₂γ₂ composition with the DsrAB subunits being ~50 kDa and ~40 kDa, respectively, with a much smaller (~11 kDa) unidentified γ subunit (DsrC).²⁹⁷ Furthermore, metal analyses of dSiRs often gave conflicting degrees of Fe content and siroheme metalation^{290,300,303,305,307} as well as distinct EPR and Mössbauer spectra that conflicted with the established cofactor structure of aSiR (discussed further in section 4.3). Sulfite reduction assays of purified dSiR in vitro typically resulted in lower activity than aSiR with incomplete sulfite reduction, generating primarily thiosulfate (S₂O₃²⁻) and trithionate (S₃O₆²⁻).^281,292,302 A thiosulfate reducing pathway has been partially identified,^308,309 and it was thought that a possible function of dSiRs was to reduce sulfite by only 2 or 4 e⁻, unlike the complete 6 e⁻ reduction achieved by all purified aSiR HP subunits, but the function of this pathway in live cells has been disputed, perhaps most effectively by sulfur radiolabeling experiments that demonstrated thiosulfate forms in a process that is inconsistent with a sulfite reduction reaction intermediate.³¹⁰ Publication of the high resolution X-ray crystal structure of Dvir resolved many of these long-standing issues: the structure elucidated the DsrAB complex with DsrC to form an active α₂ϐ₂γ₂ complex, and each ferredoxin domain was found to bind one [4Fe-4S] cluster;³¹¹ homologous structures were also found for other dSiRs.^301,312,313 There are four total siroheme binding sites, but only two of these sites are metalated in Dvir, giving rise to its characteristic 628 nm absorption peak of sirohydrochlorin that is present in desulfoviridin dSiRs but not desulforubidin dSiRs, whose absorption spectra are essentially the same as aSiRs.³⁰³ The involvement of DsrC in soluble dSiR leads to mechanistic differences that distinguish dSiR from aSiR that will be discussed further in section 4.4. While most dSiR identified have been cytosolic, a membrane-bound DsrABC dSiR (mSiR) was identified in the D. desuifuricans Essex 6 strain that exhibited hydrogenase-coupled activity and accepted electrons from cytochrome c₃.^314–316 Regardless of the large differences in holoenzyme structure, the siroheme binding domains of all aSiRs, dSiRs, siroheme-binding nitrite reductases (NiRs), and also the so-called “reverse” dSiRs (rSiR) expressed in Thiobacillis, which catalyze the oxidation of sulfane sulfur to sulfite,³¹⁷ are highly conserved. This conserved structural motif for the active site is termed the sulfite or nitrite reductase repeat (SNiRR), and siroheme has been identified as a cofactor only in sulfite and nitrite reducing enzymes.²⁹⁰

4.2.2. Comparison of aSiR and dSiR.

The major differences between aSiR and dSiR mechanisms are tied to differences of their structures. While siroheme in the aSiR is saddle-shaped and in edge contact with the [4Fe-4S] (Figure 21B), the dSiR siroheme is essentially planar (Figure 21C).³⁰¹ The saddling of the siroheme in aSiR is induced by the protein scaffold related to sidechains that are not conserved between the different active sites: Asn121 and Thr439 in the hemoprotein subunit of the E. coli enzyme (SiRHP) are replaced by Arg80α in dSiR; dSiR have additional Thr and Tyr residues that contact the siroheme propionates but are absent in aSiR.^301,311 Furthermore, the siroheme in NADPH-dependent aSiR has fewer basic residue contacts than in dSiRs.³⁰¹ While these differences in siroheme-protein interactions are likely responsible for siroheme saddling, the consequences for catalysis are not as clear. It is possible that the saddling in aSiR siroheme may even further enhance the electron “push” effect from siroheme π orbitals while allowing easier π-cation radical formation in siroheme, but despite the frequent observation that it is possible for the siroheme-[4Fe-4S] cofactor to donate up to 3 e⁻ in one step, this process has never been directly implicated in catalysis. Another possible purpose of siroheme ruffling is to tune the reduction potential of both siroheme and the [4Fe-4S] cluster through their strong coupling. Reported reduction potentials vary broadly between SiRs (SiRHP: E_H°′ = −340 mV, E_c°′ = −405 mV; alSiR: E_H°′ = −21 mV, E_c°′ = −303 mV;^318,319 dSiR: E_H+C°′ = −298 mV, E_H+C°′ = −620 mV),³¹⁹ where E_H°′ and E_C°′ denote the midpoint potentials of the siroheme and FeS cluster, respectively. Cofactor reduction in dSiR—for which FeS cluster contact with the siroheme edge is less than in aSiR—is apparently concerted and has not been accomplished in the clean, stepwise manner as in SiRHP; furthermore, the cofactor redox potentials in SiRHP were determined by redox titration rather than the direct electrochemical measurements used for dSiR.²⁹⁰ While differences in redox potential could certainly be related to siroheme coordination, saddling, and siroheme-edge/[4Fe-4S] cluster distance, the relationship is clearly complex.

4.2.3. Heme c/Cu SiR.

A member of the ONR family from W. succinogenes (multiheme cytochrome c sulfite reductase A, SiRA) has been recently characterized that displays high selectivity for sulfite reduction, with catalytic activity (k_cat = 200 electrons s⁻¹ heme⁻¹) that far exceeds reported activity for most aSiRs.^284,285 SiRA is a soluble homotrimeric enzyme with 7 electron transfer heme c cofactors and one catalytic heme c per monomer (Figure 23). The catalytic site contains a high number of basic residues reminiscent of the conserved Lys/Arg residues conserved in SiRs, along with a unique, linearly coordinated Cys₂-Cu^I site (Figure 24).²⁸⁵ Two Lys and one Arg residue (Lys208, Arg366, and Lys393) are present in the active site and make direct hydrogen bonding contacts with sulfite oxygens, and three Tyr residues (Tyr123, Tyr285, and Tyr301) coordinate multiple ordered waters that may be involved in proton transfer, removal of water from dehydration, or both. The enzyme isolated in a Cu-depleted form (0.2-0.5 Cu per monomer) displayed approximately half the activity of the fully metalated form (0.7-1.0 Cu per monomer), demonstrating the crucial role of Cu^I in its catalytic activity. Furthermore, the Cu^I was not observed to undergo spectroscopic changes indicative of redox cycling at any point in the reaction, and exposure to oxygen led to irreversible copper loss, leading to the hypothesis that Cu remains reduced during turnover. Interestingly, reoxidation titration of the fully reduced monomer indicated that the substrate obtained only 4 e⁻ from the 8 total hemes and that the homotrimer (24 total hemes) could fully reduce only 2 sulfite anions at a time, indicating either incomplete sulfite reduction or that the total charge of the homotrimer plays a role in its activation.²⁸⁵

Figure 23.

Overall structure of SiRA with its heme cofactors (PDB ID: 4RKM).

Figure 24.

Active site structure of SiRA (PDB ID: 4RKM).

4.3. Characteristic Spectroscopic Features of SiRs.

Spectroscopy of the siroheme-[4Fe-4S] active site in aSiR and dSiR enzymes is relatively complex, due to the combination of the unique siroheme cofactor and its interactions with the adjacent [4Fe-4S] cluster. Interpreting UV-Vis, EPR, and Mössbauer spectra of these enzymes has been crucial to our growing understanding of their catalytically-relevant structural features. Before an XRD structure had been obtained, enzymes capable of reducing sulfite, nitrite, and hydroxylamine were noted in bacteria, yeast, and plants in the early 1960s, but it was first concluded from studies of the E. coli enzyme that the nitrite and sulfite reducing enzymes were one and the same and that the primary function of this enzyme was sulfite reduction.³²⁰ The various SiRs share a common chromophore with absorption peak maxima in both the 380 nm and 580 nm regions, and they could receive electrons from NADPH, reduced methyl viologen (MV⁺), orferredoxin (Fd). Studies of yeast aSiRs established that there was likely a catalytic subunit common to all of them that contained this unique chromophore.^321,322 The chromophore extracted from SiRHP was called “siroheme” and identified as a tetrapyrrole macrocycle of the isobacteriochlorin type (reductively methylated at two adjacent pyrroline rings) with eight carboxylate moieties.^323–326 The additional 628 nm absorption peak in Dvir is due to demetalated siroheme (sirohydrochlorin).³²⁴ In addition to serving as the catalytic center in SiRs, siroheme is capable of efficiently reducing nitrite and sulfite even outside of the protein matrix.^327,328

The property of SiRs that has been the subject of the most study and of the greatest dispute is the interaction of the siroheme and [4Fe-4S] cofactors. The siroheme and [4Fe-4S] cofactors are magnetically exchange coupled in all physiological oxidation states. The oxidized enzyme (SiR⁰) has a single S = 5/2 spin arising from the high-spin Fe^III siroheme and Mossbauer parameters for a [4Fe-4S]²⁺ cluster (S = 0) but also exhibits field dependence associated with a paramagnetically active iron-sulfur cluster. ³²⁹ The 1 e⁻ reduced state (SiR¹⁻) comprises a high-spin Fe^II siroheme (S = 1 or 2) and oxidized iron-sulfur cluster, meaning the first electron localizes to siroheme, but the total number of electrons for both cofactors becomes a non-Kramer’s system; thus, the two cofactors behave like a single 5-Fe system with a single shared spin. The observed behavior is possible only when considering exchange interactions between the siroheme Fe and individual Fe sites of the [4Fe-4S] cluster and not a dipolar interaction: the two Fe sites of the cluster are strongly exchange-coupled and result in the classical diamagnetic state of the [4Fe-4S]²⁺ cluster, but they have weaker exchange interactions with the siroheme Fe so that a paramagnetic contribution is always mixed with the cluster ground state.^329,330 Novel EPR features arising from exchange interactions between integer spin siroheme and the [4Fe-4S]⁺ cluster in the 2 e⁻ reduced enzyme (SiR²⁻) demonstrates that the cofactors remain coupled in the reduced state.³³¹ Some aSiR have been isolated with low-spin siroheme, but the [4Fe-4S]²⁺ cluster Mössbauer parameters are identical to the high-spin siroheme enzymes.³³² The [4Fe-4S] cluster does not contribute significantly to the electronic absorption spectrum, and the S = 1/2 paramagnetic spectrum of the 1 e⁻ reduced [4Fe-4S]⁺ cluster does not appear in X-band EPR spectra of SiR²⁻; rather, the classical g = 1.94 ferredoxin-like features predicted by Mössbauer are significantly perturbed into novel features with g = 2.53, 2.29, and 2.07 (“S = 1/2”-like) and characteristic low-field features between g = 4.7-5.4 (“S = 3/2”-like) (Figure 25).³³¹ Reduction of SiRHP in the presence of an inhibitory strong-field ligand such as CN⁻ or CO results in S = 0 siroheme, revealing the “classical” g = 1.94 spectrum of a cubane [4Fe-4S]⁺ cluster. The perturbed [4Fe-4S] features resulting from exchange coupling with the paramagnetic siroheme could be further modulated with siroheme weak-field ligands or completely silenced by other strong-field ligands such as AsO₂⁻, S²⁻, and chaotropes (guanidinium, urea, DMSO).^{318,333–335} Electron-nuclear double resonance (ENDOR) analysis of the ¹⁴N hyperfine coupling constants in SiRHP definitively ruled out a histidine proximal ligand to siroheme and noted ϐ-CH proton coupling constants (1.88 MHz) consistent with a cysteine or serine residue as the proximal ligand.³³⁶ A low resolution (3.0 Å) X-ray crystal structure of E. coli SiRHP provided evidence that the coupling between the siroheme and [4Fe-4S] cluster prosthetic groups was due to a shared ligand and that the [4Fe-4S] cluster was in near van der Waals contact with the siroheme macrocycle edge,²⁸² and comparison of the sequences of spinach assimilatory nitrite reductase (aNiR) and Salmonella aSiR led to the creation of a model of the siroheme-[4Fe-4S] active site for which one of the Cys residues that coordinate the [4Fe-4S] cluster is the siroheme proximal ligand.³³⁷ This model was confirmed by a 1.6 Å X-ray diffraction structure of SiRHP by published by Crane, Siegel, and Getzoff in 1995, which clearly showed the bridging cysteinyl ligand and a saddle-shaped siroheme bent toward the [4Fe-4S] cluster as postulated by the ENDOR results.²⁸³ Early characterization of Dvir reported unusually low-field S = 9/2 EPR features that were attributed to one or more cubane [4Fe-4S] clusters, and it was proposed that the siroheme and FeS cluster were not actually coupled and that the signals might even arise from a 6Fe “super cluster.”^307,338 However, at the time of these studies, there was significant disagreement regarding protein Fe content, and the presence of these features was subject to purification conditions.^290,297,316 The dispute was conclusively resolved with the report of the Dvir structure, which contained two metalated sirohemes coupled to two cubane [4Fe-4S] clusters, two demetalated, uncoupled sirohydrochlorins and corresponding [4Fe-4S] clusters, and two electron transfer [4Fe-4S] clusters near the Fd docking site (Figure 22).³¹¹

Figure 25.

EPR Spectra of SiRHP displaying g ~ 2.29 features characteristic of the exchange-coupled FeS cluster. Figure adapted from ref. ³³¹ Copyright American Chemical Society 1982.

Figure 22.

Structure of α₂ϐ₂γ₂ complex of dSiR with DsrA (geen), DsrB (gray), and DsrC (cyan) (PDB ID: 2V4J).

4.4. Other Sulfite Reducing Enzymes.

4.4.1. Siroheme-containing nitrite reductase (NiR).

The conserved SNiRR domain for binding the siroheme-[4Fe-4S] cofactor is present both in SiRs and NiRs—enzymes responsible for the 6 e⁻ reduction of nitrite (NO₂) to ammonium (NH₄⁺). Since the active sites are nearly identical, it is not surprising that proteins bearing the SNiRR domain often display both SiR and NiR activity.^318,339,340 All SiRs studied to date are capable of nitrite reduction, usually with k_cat that is 10-10³ fold faster than sulfite as a substrate, but with overall catalytic efficiency several orders of magnitude lower (except alSiR) due to higher nitrite K_M.²⁹⁰ NiRs are distinguished by their relatively higher affinity for NO₂⁻ over SO₃²⁻. Differences in the amino acids forming the distal pocket of the siroheme between NiR and SiR have been implicated as a crucial determinant for the enzyme’s selectivity. Alignment of the siroheme-[4Fe-4S] cofactors from spinach aNiR and SiRHP show that three of the catalytic residues important in SiRHP (Arg83, Arg153, and Lys215) are conserved with analogous residues in aNiR, but two of the residues (His123 and Lys217) are replaced in aNiR by Arg149 and Asn226, respectively. Studies of tobacco NiR have demonstrated that mutating Asn226 to Lys increased sulfite reduction activity by an order of magnitude.³⁴¹

A ferredoxin-dependent aSiR (called NirA) isolated from Mycobacterium tuberculosis was found to crystallize with an unusual cross-link between the siroheme-contacting residue Tyr69 and Cys161.³⁴² This type of thioether cross-link is present in galactose oxidase where it lowers the reduction potential of the catalytic Cu^II and facilitates the 2 e⁻ oxidation of alcohols to aldehydes.³⁴³ In NADPH- and Fd-dependent aSiR, the Tyr residue is conserved, but Cys is not.³⁴² Mutation of either Tyr or Cys significantly reduced activity in NiRA, and the authors interpreted this to mean that the thioether cross-link is important for NiRA activity. They also noted that the Tyr residue is conserved in Fd-dependent aSiRs but not in Fd-dependent aNiRs, and they proposed that the presence or absence of this Tyr could be used as a genetic marker of substrate preference for sulfite or nitrite.³⁴² These substitutions paint a gradually resolving picture of the structural bases underpinning the selectivity for nitrite over sulfite in aNiRs, but few of the proposed effects of these structural differences have been directly tested to-date.³⁴⁴

4.4.2. Pentaheme Cytochrome c NiR.

An important ancestral structure in bacterial nitrate metabolism is pentaheme cytochrome c nitrite reductase (ccNiR), which phylogenetic analysis has identified to be a likely precursor to hydroxylamine oxidoreductases and octaheme nitrite reductases (ONR).³⁴⁵ ccNiRs vary in sequence but have tightly conserved arrangements of heme c cofactors for which a centrally located Lys-coordinated heme serves as the active site (Figure 26). A ccNiR known as NrfA has been observed to reduce sulfite as well as nitrite.³⁴⁶ Conserved His, Tyr, and Arg residues make hydrogen bonding interactions with bound sulfite. The mutation Tyr218Phe significantly impairs nitrite reduction, but sulfite reduction by NrfA is unaffected, demonstrating a potentially intriguing difference between nitrite and sulfite selectivity.³⁴⁶ However, due to the nature of the NH₄⁺ and S²⁻ quantification assays used, it is unclear whether these mutations truly reduce the rate of nitrite to ammonium reduction or whether they alter the number of electrons transferred per substrate.

Figure 26.

Active site structure of ccNiR (PDB ID: 3BNF).

4.5. Mechanism of Sulfite Reduction in SiRs.

4.5.1. Siroheme-[4Fe-4S] SiR.

The basic mechanistic proposal of sulfite reduction in SiR follows a “push-pull” strategy that is similar to the strategy adopted by O–O bond cleaving enzymes.^290,347,348 The electron-rich siroheme-[4Fe-4S] cofactor donates charge to the π-accepting sulfur (or nitrogen) atom of the substrate. The partially saturated siroheme (isobacteriochlorin) is easier to oxidize and a superior π-donor relative to porphyrins (Scheme 29),^349–351 and it has been proposed to transfer electrons to the sulfur π-antibonding orbitals to weaken the short, stable S–O bonds; although, this proposal has never been experimentally verified. The high density of conserved positively charged residues in the distal siroheme pocket further promote S–O activation through protonation of the substrate oxygens.^283,290. In-depth computational studies of the SiR mechanism have been limited due to the complexities of the coupled siroheme-[4Fe-4S] cofactor;³⁵² therefore, a majority of mechanistic insight has been gained through crystallographic studies of SiRs with their native substrate and substrate analogues.

Scheme 29.

Structural Difference Between Heme and Siroheme Cofactors.

For aSiR, the catalytic cycle for the full reduction to S²⁻ begins with reduction of high-spin Fe^III siroheme to high-spin Fe^II, allowing sulfite to displace a phosphate anion bound in the oxidized resting state (Scheme 30). Substrate binding is accompanied by the rearrangement of a loop region that encloses the anion binding pocket,²⁸³ and substrate binding likely affects protein-siroheme contacts, activating the site through redox-gating.^318,353 Following substrate binding, electrons are transferred to siroheme through the coupled [4Fe-4S], accompanied by protonation from either ordered waters or from active site Lys/Arg residues for three successive dehydrations (Scheme 30). Typically, these steps are depicted as a transfer of 2 e⁻/2 H⁺ to the active site to form S(IV), S(II), S(0), and S(2−) intermediates, but the presence of radical intermediates from single ET steps has not been ruled out. Co-crystallization with inhibitors (CN⁻ and CO) and substrate anions revealed that the active site adopts several conformations to accommodate the substrate as it is deoxygenated (Figure 27). The sidechain of conserved Arg153 rotates from its phosphate-bound state to a sulfite-bound/closed loop state with sulfite (but not nitrite) binding, and Lys215 rotates to interact directly with sulfite. Arg153 then returns to its original conformation after the first dehydration.³⁵⁴ Single mutant experiments with SiRHP revealed that the first two protons likely come from a protonated substrate and an ordered water. Lys215 is essential for anion binding throughout the cycle, but Lys217 and Arg153 are involved in late-stage proton transfer as Ser mutants of either of these residues reduced total electrons transferred to the substrate to 4 e⁻ and 5 e⁻, respectively, but still allowed efficient 2 e⁻ reduction of hydroxylamine.^354,355

Scheme 30.

Proposed Catalytic Mechanisms and the Observed Intermediates of aSiR and dSiR.

Figure 27.

Crystal structures of SiRHP bound to substrate and inhibitors where siroheme Fe is either in its reduced (orange) or oxidized (red) state: (A) PO₄³⁻ (PDB ID: 1AOP), (B) SO₃²⁻ (PDB ID: 2GEP), (C) NO₂⁻ (PDB ID: 3GEO), (D) NO (PDB ID: 6GEP), (E) CN⁻ (PDB ID: 4GEP), and (F) NO₃⁻ (PDB ID: 8GEP).

As mentioned in section 4.2, the heterotrimeric structure of dSiRs and, in particular, the role played by the DsrC subunit suggest that the dSiR mechanism differs substantially from the aSiR mechanism to better facilitate the physiological role of dSiRs. In contrast to aSiR, the siroheme sites of dSiRs are located at the interior interface of the A/B subunits and are more deeply buried. The [4Fe-4S] and proximal face of siroheme are coordinated by residues in DsrB, and strictly conserved positively charged Lys and Arg residues homologous to the distal residues of aSiR are provided by DsrA.³¹¹ This active site opens to a substrate channel that is blocked by Tyr-334B in Dvir and Trp-119β in the A. fulgidus dSiR so that only half of the siroheme sites are active.^301,311 The C-terminal arm of DsrC reaches into the active siroheme distal pocket and positions a highly conserved cysteine, Cys104, within reach of bound substrate. Recently, Santos et al. have demonstrated that the role of DsrC in the dSiR mechanism is to attack partially reduced sulfur species bound to siroheme, and release from DsrAB as a trisulfide (Cys93–S–Cys104) (Scheme 30).³⁵⁶ Effectively, this means dSiR natively only performs a 2 e⁻ reduction of SO₃²⁻, and the remaining 4 e⁻ reduction of the trisulfide DsrC would be accomplished by the DsrK subunit of the associated respiratory membrane complex (DsrMKJOP).^311,356 This proposed mechanism essentially reconciles the observation of incomplete sulfite reduction by Dvir purified from the cytosolic fraction while sulfide is obtained as the only product of the enzyme obtained from the membrane fraction³¹⁵ and finds further support from its co-purification with the sulfite reducing products of the dsr genes and DsrKJO membrane components of sulfur oxidizing A. vinosum.³⁵⁷ However, DsrC is not expressed constitutively with DsrAB, and its association with DsrAB appears to be relatively low;³⁵⁸ therefore, some questions remain concerning the exact cellular role(s) of DsrC in sulfate reducing bacteria. An alternative—though more convoluted—mechanism was proposed from the observation that in some structures of DsrABC the Sγ of Cys104 is covalently linked to a siroheme meso carbon (Figure 21A). By this mechanism, a persulfide forms from reaction with 6 e⁻ reduced S²⁻ before protonation can occur, a second equivalent of SO₃²⁻ binds to siroheme, and after 2 e⁻ transfer to an incoming substrate anion it reacts with the nearby persulfide to form and then release S₂O₃²⁻ or repeats the process to yield S₃O₆²⁻.³¹² However, this mechanism is not as well supported as the transient involvement of DsrC, and there is a greater body of evidence that suggests thiosulfate and trithionate are not produced physiologically.

In nature, siroheme is universally found covalently linked to a [4Fe-4S] cluster, and functional implications for this conserved structural feature have been hypothesized but remain unconfirmed. Structures of the SiRHP subunit in aSiR and the multi-subunit complexes of dSiR show that the [4Fe-4S] cluster likely serves a role in ET between the siroheme active site and the other ET protein subunits.^283,311,359 Spectroscopic studies have shown that electronic coupling between siroheme and the [4Fe-4S] cluster cause siroheme to exhibit paramagnetic character in all of its physiological oxidation states (see section 4.3); however, the functional implications of this exchange coupling remain unclear. Reduction of the [4Fe-4S] cluster in CO-bound SiRHP lowers the υ_C–O from 1920 cm⁻¹ to 1904 cm⁻¹, demonstrating the coupled cluster has some impact on the degree of small molecule activation by siroheme.³⁶⁰ Recent mutation studies of residues near the [4Fe-4S] cluster in SiRHP demonstrate efficient ET to the [4Fe-4S] cluster is an important factor in catalytic activity.³⁶¹ In these studies, Cepeda et al. modified the bulky surface residues that block solvent access to the [4Fe-4S] to less bulky Ala residues. Phe437Ala and Thr477Ala mutations increase sulfite reduction activity of both the SiRHP subunit and the holo-aSiR complex (dodecamer of SiRHP and flavoprotein subunits), consistent with improved access of reductant to the [4Fe-4S] cluster (Figure 28). Computational analysis of the ET pathways between siroheme and [4Fe-4S] consider the covalent Cys bridge between the Fe in siroheme and Fe in [4Fe-4S] is the most favorable ET pathway.³⁶² This analysis further suggests a functional role of the isobacteriochlorin ring structure of siroheme, which reduces the propensity of the [4Fe-4S] cluster to directly transfer electrons to the π-system. The authors speculate that avoiding radical character on the porphyrin ring may avoid potential side-reactions between the partially reduced sulfur species bound to Fe during turnover.

Figure 28.

Position of non-coordinating residues in SiRHP (PDB ID: 1AOP) Phe437, Met444, and Thr477 that control ET rate to [4Fe-4S] cluster for SiR activity.

4.5.2. Heme c/Cu (SiRA).

Considerably less is known about the mechanism of sulfite reduction by the heme c/Cu active site of SiRA, which has been studied less than other SiRs.²⁸⁴ The recent XRD structures of different forms of SiRA in the presence of sulfite has led to an initial mechanistic proposal reliant on the nonheme Cu^I ion that differs from aSiRs and dSiRs (Figure 29D).²⁸⁵ In the Cu-depleted form, sulfite binds to the active site heme in a manner reminiscent of SiRHP (Figure 29C); however, incubation of sulfite with the Cu-containing form leads to a mixture of two ligands observed bound to heme c. The first is SO₂, which the authors propose to be the product of dehydration without ET promoted by the [(Cys)₂Cu]⁻ center to yield S^IVO₂ (Figure 29A). A second ligand, SO, was observed bound to heme c, which the authors attribute to a product of partial turnover from photoreduction during data collection (Figure 29B). In both cases, the Cu remains reduced, and its coordination state is unaltered. Thereafter, the SiRA mechanism is proposed to follow three successive 2 e⁻ reductions and two dehydrations, reminiscent of the aSiR mechanism. Further study is necessary to confirm this proposed mechanism and whether the essential Cu center does indeed play no role in ET or the binding of any intermediates.

Figure 29.

XRD structures of SiRA: (A) dehydrated form of SO₃²⁻ substrate (SO₂) (PDB ID: 4RKM), (B) photoreduced structure with SO bound to heme (PDB ID: 4RKM), (C) Cu-depleted form with SO₃²⁻ bound to heme (PDB ID: 4RKN). (D) Proposed mechanism of sulfite reduction, based on these structures. Figure adapted from ref. ²⁸⁵ Copyright Springer Nature 2015.

4.6. Biomimetic and Biosynthetic Models of SiR.

Comparatively few structural and functional models of SiR have been developed, and those that have been reported focus mainly on the siroheme-[4Fe-4S] cofactor. The Holm group has developed a series of structural SiR models, [(4Fe-4S)(L16)–μ₂–S–(L)Fe], using a strategy of cluster sub-site differentiation that could undergo regiospecific substitution in the presence H₂S to form a bridge between the FeS and Fe complex (Scheme 31).^363,364 The [4Fe-4S] cluster was formed within a tridentate ligand (L16) and was successfully bridged by a variety of routes to either (salen)Fe^III or (OEP)Fe^III (OEP = octaethylporphyrin). The resulting bridged assemblies demonstrate spin-delocalization from high-spin Fe^III to the cluster, and both complexes exhibit two redox transitions in DMF at low potential; though, the order of cofactor reduction could not be determined: −0.71, −0.96 V (salen) and −1.05, −1.32 V (OEP).³⁶³ This lab subsequently produced a bridged assembly that was a closer analogue of the SiR catalyst from a [4Fe-4S] cluster and an isobacteriochlorin-type complex (OEiBC) by the same method ([(4Fe-4S)(L16)-μ₂-S-Fe^III(OEiBC)]²⁻) that also demonstrated spin delocalization;³⁶⁴ however, this complex was not electrochemically characterized. An alternative strategy to mimic the [4Fe-4S]-heme structure in SiRs has been reported by Gerlach et al., who utilized pyridyl- and imidazoylthiolate linkers between a site-differentiated [4Fe-4S] cluster and Zn(porph) complexes.³⁶⁵ These studies clearly demonstrate an electronic effect of tethering a metal porphyrin to an [4Fe-4S] on the cluster redox potentials, which shift from −440 mV ([4Fe-4S]^2+/1+ vs NHE) to −780 mV. None of these structural models have been reported to show any sulfite reductase-like catalytic activity.

Scheme 31.

Reported Structural Models of Siroheme-[4Fe-4S] SiRs. ^363,365

It was found that siroheme is a competent sulfite reduction catalyst using MV⁺ as an electron donor without the protein scaffold, though not as efficiently as isolated HP subunits of native SiRs.³²⁷ A study of {FeNO}⁷ complexes ligated by model porphyrin (OEP), chlorin (OEC = 7′,8′-dihydro-octaethylporphyrin), and isobacteriochlorin (OEiBC: 2′,3′,7′,8′-tetrahydro-octaethylrphyrin) ligands found that there was no significant difference between the isobacteriochlorin versus the porphyrin or chlorin for single electron reduction, but it was notable that OEiBC preferentially formed a macrocycle cation radical over oxidation to {FeNO}⁶ (Scheme 32)³⁶⁶, which typically display much higher reduction potentials in related Fe-porphyrin complexes.³⁶⁷ Examples of electrocatalytic sulfite reduction by other heme complexes have been reported. The six electron reduction of sulfite to sulfide has been reported for the synthetic water-soluble Fe-porphyrin complex, (L11)Fe (L11 = meso-tetrakis(p-sulfanatophenyl)porphyrin).³⁶⁸ Electrocatalytic sulfite reduction has also been described with surfactant-based films of myoglobin.³⁶⁹ These studies determined that the protein likely releases its heme cofactor. Both of these examples observe the catalytic wave at or near the Fe^II/Fe^I couple, with potentials much lower than what is relevant for the native enzyme (< −0.4 V and < −0.9 V vs SHE for the model complexes, respectively) (Scheme 33). Nevertheless, these examples demonstrate sulfite reduction can be accomplished by an Fe center with various porphyrin ligands.

Scheme 32.

Reduction Potentials and υ_N–O of Model Porphyrin (OEP), Chlorin (OEC), and isobacterochlorin (OEiBC) Complexes of Fe.³⁶⁶

Scheme 33.

Reported Electrocatalytic SO₃²⁻ Reducing Heme Complexes and Catalytic Properties.^368,369

An early biosynthetic modification of the E. coli aSiR to create a minimal functional holoenzyme with only a single flavoprotein and hemoprotein subunit was achieved by Zeghouf et al.³⁷⁰ They found that they could crystallize the aSiR flavoprotein subunit (SiR-FP, the α subunit of the α₈ϐ₄ holoenzyme) by truncation of the first 51 amino acids to yield a functional monomeric form of SiR-FP (α′) that they called SiR-FP60.³⁷¹ The truncated SiR-FP60 formed a functional dimer with purified SiRHP of the form α′ ϐ and was competent to perform NADPH-dependent catalysis at 20% the rate of the native holoenzyme. Given the stability of this complex, they disputed the accepted α₈ϐ₄ oligomeric state of the holoenzyme, proposing that, based on titration experiments and in the absence of a holoenzyme structure, α₈ϐ₈ was also plausible.³⁷⁰ The first true structural and functional biosynthetic model of sulfite reductase was developed by Mirts et al., by the redesign of the heme proximal binding site in cytochrome c peroxidase (CcP) to bind a [4Fe-4S] cluster adjacent to the native heme b prosthetic group (Figure 30).³⁷² Like native SiR this biosynthetic model incorporated a bridging Cys residue to act as the heme proximal ligand and a ligand to one of the Fe sites in the [4Fe-4S], building off an earlier attempt to create a cytochrome P450-like Cys-heme active site in CcP.³⁷³ The designed enzyme, called SiRCcP, explored several combinations of SCS mutations around the designed [4Fe-4S] cluster site and substrate coordinating distal site. Mutations were introduced that mimicked the conserved Arg153, Lys215, and Lys217 residues in SiRHP, while the native CcP residue Arg48 served an analogous role to SiRHP residue Arg83 as both a (siro)heme and substrate contacting residue. The most active mutant (called KRK-SiRCcP.3) could catalyze the reduction of sulfite to sulfide at ~18% the rate of the M. tuberculosis aSiR.³⁷²

Figure 30.

Modelled structure of KRK-SiRCcP.3 (gray) overlaid with SirHP (pink).

4.7. Summary and Outlook of SiR and Biomimetic Models.

In terms of multi-electron chemical transformations, the 6 e⁻ reduction of sulfite to sulfide is one of the most complex small molecule transformations performed by a single enzyme, matched by 6 e⁻ nitrite reduction, or the overall 8 e⁻ N₂ reduction process in nitrogenase to form ammonium and dihydrogen.³⁷⁴ Obtaining a clear understanding of the SiR catalytic center proved challenging due to the complex nature of the covalently linked and magnetically exchange-coupled [4Fe-4S] cluster and siroheme, a unique isobacteriochlorin that has been found in no other enzymes in living organisms. Examination of the crystal structures of dissimilatory and assimilatory SiRs combined with electrochemical and computational studies of model heme, chlorin, and (iso)bacteriochlorin compounds have revealed in greater detail how subtle differences in siroheme ruffling induced by protein contacts and coupling to the [4Fe-4S] cluster control the flow of electrons through the siroheme Fe and suppress oxidation of the siroheme ring to prevent the formation and release of partially reduced intermediates. Though few structures of each enzyme class have been obtained so far, they have provided the clearest insights into how the different holoenzyme structures of dSiRs and aSiRs contribute to their functions in respiratory metabolism and the generation of sulfur-containing biomolecules, respectively, as well as the number of successive dehydrations accomplished in each enzyme in its native role. Biomimetic synthetic complexes and proteins have further demonstrated that while many heme molecules can achieve sulfite reduction to a considerable degree, selective reduction to S²⁻ is highly dependent on the inclusion of substrate coordinating and ET-promoting residues provided by the protein scaffold. It is notable also that the two forms of sulfite reductase active sites described so far (siroheme-[4Fe-4S] SiRs and the heme c/Cu multiheme SiRA) utilize unique heteronuclear active sites with heme cofactors. While our understanding of features that promote the reduction of oxyanions such as sulfite and nitrite at an Fe catalyst is growing, many questions remain to be answered for a deeper understanding of the structural bases of the efficient catalytic properties of sulfite reductases. Some of these questions include:

1. What role does the nonheme cofactor ([4Fe-4S] in SiRs and Cu in SiRA) play in the binding and activation of substrate during turnover? With so little—and generally inconsistent—information regarding the redox potentials and the detailed process of cofactor redox cycling at specific reaction steps, there remain many unknowns of how these auxiliary cofactors may control redox activation, substrate/intermediate stabilization, or modulation of the redox properties of the (siro)heme cofactor at various stages of this multi-step reaction.

2. For siroheme-containing SiRs, does the siroheme ring ever act as an electron donor during turnover? Are there any functional reasons this ancient cofactor remains one of the most prevalent tools in living organisms for sulfite and nitrite reduction, or is siroheme merely a convenient relic of facile iron macrocycle synthesis in early anaerobes?

3. What is the nature of the partially-reduced sulfur intermediates (−SO_x), and how do their interactions with the enzyme active site influence their reactivity towards sulfide formation versus off-pathway products (S₂O₃²⁻ and S₃O₆²⁻) from reaction with additional substrate anions?

As has been demonstrated throughout this review, biomimetic models are an invaluable tool in expanding our understanding of the underlying chemical principles in these relatively complex active sites. The complementary study of native sulfite-reducing metalloenzymes and their biomimetic models will hopefully continue to deepen our understanding of the sulfite reduction mechanism down to the molecular level.

5. Comparison of Heme/Nonheme Active Sites of HCO, NOR, and SiR.

5.1. HCO and NOR

5.1.1. Cross-Reactivity of HCO Superfamily.

A handful of bacterial HCOs, representing two of the three main types (B and C; see section 2.1), are able to accomplish catalytic NO reduction.^{199,200,375,376} Similarly, oxidase activity has been observed in cNOR from P. denitrificans.^185,377,378 In cases where an enzyme within this superfamily displays both NO and O₂ reducing activity, it exhibits much higher activity for its native reaction (Table 1). A number of the previously discussed synthetic models of BNC active sites display both NO and O₂ reduction reactivity.^{117,258,274,379} Despite their similar roles in biological anaerobic and aerobic respiratory pathways, the molecular mechanisms of O₂ reduction to H₂O and NO coupling to N₂O are quite dissimilar from a chemical perspective. Therefore, it is noteworthy that similar heme/nonheme catalysts for these distinct reactions are found in nature—catalysts similar enough that enzymes in the HCO superfamily can achieve both the 2 e⁻ reduction of NO and 4 e⁻ reduction of O₂. On the other hand, if we choose to view the differences between HCO and NOR structure as the result of natural selection which has optimized for efficiency of each of their native reactions, they become a valuable case study in catalyst design for small molecule transformations. What makes the heme/Cu_B active site better for O₂ reduction, and the heme/Fe_B site for NO coupling? Based on our current knowledge of these enzymes, there is no definitive answer but studies of these native proteins and their biomimetic models provide some possible explanations. In the following sections, we will consider insights gained on the functional implications of differences between the BNC sites of HCO and NOR, in particular the characteristics of heme and nonheme cofactors, along with differences in enzyme PT pathways.

Table 1.

NO and O₂ Reduction Activity of HCO Superfamily Enzymes.

	NO red. activity (min−1)	O2 red. activity (min−1)	ref
NOR
PdNOR	~2.5 x 104	~2 x 103	185,377
Pd ATCC	~4 x 103	~4 x 102	378
35512 NOR
HCO
Ps cbb3 ox.	1.0 x 102	8.4 x 103	199
Tt ba3 ox.	3	1.5 x 104	375
Tt caa3 ox.	32	~450	375,380
Ec bo3 ox.	0.3	~1.5 x 104	376,381
Rs cbb3 ox.	~2 x 102	~6 x 104	200

5.1.2. Effect of Heme Cofactor.

A range of reduction potentials for the active site heme cofactors in NOR and HCO have been measured, and the general trend is that HCO contains higher potential heme cofactors than the heme b₃ center in NOR. The average potential of HCO heme is greater than that of NOR heme by ~200 mV (Table 2).²⁷⁰ The heme b₃ center in NOR has a relatively small variation in potential, between −170 mV (vs SHE) and 80 mV.^238,382 The range of active site heme potentials in HCO is much greater (between −120 mV and 460 mV), partly due to the occurrence of b₃, a₃, and o₃ forms.^383–385 The electron withdrawing formyl groups on heme a are one reason HCO potentials tend to be higher than NOR. Studies of native enzymes and related models suggest the neighboring conserved glutamate residues in NOR, which are absent in HCO, also play a key role in tuning the heme reduction potential.^185,264,386 There appears to be functional significance of these heme reduction potentials: HCO enzymes that are competent for NO reduction tend to contain lower heme potentials that are close to those of native NORs, and there is a general correlation of higher NO reductase activity in HCOs as heme potentials decrease (see Tables 1 and 2). One of the highest HCO active site heme potentials reported is the a₃ heme of mitochondrial CcO, which is around 460 mV.³⁸³ While initially considered to have NO reduction activity,³⁸⁷ later studies ruled out such reactivity and show that it is, instead, inhibited by binding NO.³⁸⁸ The possible implication of heme reduction potential on NO reduction by HCOs was analysed in a computational study.²⁴¹ This study hypothesized that a conserved Val residue in oxidases leads to a high energy barrier of N₂O formation by a mechanism that derives both reducing equivalents from the BNC active site, and therefore NO reduction in HCO relies on an alternative pathway, which involves proton-coupled reduction of the hyponitrite intermediate—the rate of which would be highly dependent on the heme reduction potential.

Table 2.

Measured Reduction Potentials of the Active Site Heme and Nonheme Cofactors of HCO Superfamily Enzymes.

Enzyme	heme	nonheme	ref
PdNOR	78-80 mV / 60 mV	320 mV (FeB)	185,186,238
PnNOR	−38 ±10 mV / −162 ±9 mV	−369 ± 14 mV (FeB)	239
PaNOR	−440 mV	−110 mV	237
M. hydrocarbonclasticus cNOR	−168 ±15 mV	n.r.	382
Bovine mitochondrial CcO (aa3 ox.)	460 mV	400 mV (CuB)	383
A. ambivalens aa3 oxidase	390 ±20 mV	n.r.	389
B. pseudofirmus caa3 ox.	334 mV	n.r.	390
B. japonicum cbb3 ox.	275 mV	n.r.	391
Bacillus YN-2000 aco ox.	250 mV	n.r.	392
E. coli bo3 ox.	200 mV	n.r.	385
T. thermophilus caa3 ox.	133 mV / 248 mV / 378 mV	n.r.	393
T. thermophilus ba3 ox.	199 mV	n.r.	394
R. marinus cbb3 ox.	−50 mV	120 mV (CuB)	395
P. stutzeri cbb3 ox.	−54 mV	372 mV (CuB)	384
R. sphaeroides cbb3 ox.	−59 mV / −95 mV	415 mV (CuB)	384
V. cholera cbb3 ox.	−120 mV	n.r.	384

A recent study of biomimetic heme/nonheme models of NOR supports the hypothesis that heme potential plays an important role on NO and O₂ reactivity. Fe_BMb proteins were prepared with a variety of heme cofactors, leading to a range of observed heme reduction potentials from −130 mV to 148 mV.²⁷⁰ These different heme cofactors would display faster NO binding, greater NO activation (lower υ_N–O), and faster NO decay with lower reduction potential (Figure 19B). Lower reduction potential would also, however, lead to reduced electron transfer rates, which could possibly slow down NO reduction catalysis. It was determined that the most active NO reduction catalyst of these proteins was one with an intermediate reduction potential of 53 mV (Figure 19C), suggesting efficient catalysis relies on a crucial balance between fast ET and sufficient reducing potential of the active site heme.

5.1.3. Effect of Nonheme Cofactor.

Another prominent difference between NOR and HCO enzymes is the identity and geometry of the nonheme cofactor. Reduction potentials of the nonheme cofactor in the HCO superfamily are much harder to measure, since they are not associated with a distinctive shift in a protein’s UV-Vis absorbance spectrum (unlike a heme Soret band). In the cases where they have been measured, similar to the trend observed with heme cofactor, the reduction potential of the Cu_B center in HCO is higher than Fe_B in NOR (Table 2). Based on extensive computational analysis of O₂ and NO reduction by the HCO enzyme superfamily, Blomberg proposed that the lower reduction potentials of the metal cofactors in the NOR BNC is advantageous for promoting irreversible NO binding and fast reduction of this toxic substrate, whereas the relatively higher potentials in HCO are important for energy conservation and proton pumping.^107,396,397

Another major difference between the Fe_B and Cu_B centers is their distance to the heme cofactor. Crystal structures of NORs show a heme/nonheme Fe–Fe distance that ranges between 3.8 and 4.4 Å, whereas HCO structures typically have Fe–Cu distances somewhere between 4.4 and 5.3 Å (Table 3). These differences can be considered the consequence of two major structural features: (i) Glu coordination of Fe_B in NOR tends to move the nonheme metal closer to heme, and (ii) the His-Tyr crosslink in HCO moves the third His ligand to Cu_B down (towards the heme), which results in shifting the nonheme metal away from heme, to achieve a stable trigonal planar geometry (Figure 31).⁵⁴ Functional implications of these structural differences (for promoting O₂ over NO reduction and vice versa) are unclear, however, due to an incomplete understanding of potential active site conformational changes that occur during turnovers.

Figure 31.

Active site structures of (A) PaNOR, and (B) bovine aa₃ oxidase. Coordination geometry and distance to heme of (C) Fe_B in PaNOR, and (D) Cu_B in bovine aa₃ oxidase.

One way to probe the effect of the nonheme cofactors on the enzymatic activity would be to replace Cu_B in HCO with Fe^II, or the Fe_B center in NOR with Cu^I. However, there is no report of such a study, probably because of the challenge of introducing a non-native metal ion into the native binding site. In contrast, it is quite easy to prepare biomimetic models of HCO superfamily enzymes with either nonheme Cu or Fe. These models have provided some insight into the role of the nonheme metal in promoting selective O₂ or NO reduction. HCO and NOR models reported by Collman et al., [(L14)Fe^II/Cu^I]⁺ and ([(L14)Fe^III/Fe^II(Cl)]⁺ respectively, have compared the role of the nonheme metal ion in O₂ reduction.³⁷⁹ Under slow electron flux (1 electron s⁻¹), the absence of a nonheme metal (L14)Fe^II) results in PROS production of 20%. Addition of Cu ([(L14)Fe^II/Cu^I]⁺) leads to a decrease of PROS to 6%, while the ([(L14)Fe^III/Fe^II(Cl)]⁺ complex results in 6% total PROS, which further decreases to 3% at fast electron flux, lower than nearly all similar picket-fence porphyrin complexes tested. This study demonstrates that either nonheme metal (Fe or Cu) can promote complete 4 e⁻ reduction of O₂. Similarly, in Fe_BMb, Fe^II and Cu^I forms were both able to catalyze O₂ reduction, with rates and selectivity for H₂O much higher than the metal-free or Zn^II bound forms of the biosynthetic proteins (Figure 32).¹¹⁷ Consistent with the trend observed for NOR and HCO, the reduction potential of the nonheme Fe^II was found to be lower than Cu^I (259 and 387 mV, respectively), and it was rationalized that the higher reduction potential of Cu^I improves catalytic activity by increasing the rate of electron transfer. Furthermore, DFT calculations of the peroxy-bound Fe_BMb active site with Fe^III and Cu^II showed greater degree of O–O activation with Cu^II, possibly due to its higher d-electron count.¹¹⁷ These studies suggest CuB is better for O₂ reduction by improving energy conservation (by virtue of its higher reduction potential) and leading to higher O₂ activation compared to Fe_B.

Figure 32.

ORR activity data of Fe_BMb with various nonheme metals. Figure adapted from ref. ¹¹⁷ Copyright Springer Nature 2017.

HCO is also distinguished from NOR by having another redox active cofactor in its binuclear active site. The conserved Tyr residue is understood to be a crucial feature for efficient O₂ reduction. Its presence in HCO provides the fully reduced BNC with a total of four reducing equivalents, which allows for complete O–O cleavage without exogenous ET, which provides a simple rationalization for the enzyme’s complete selectivity for H₂O over other PROS. Therefore, the absence of a corresponding Tyr in NOR BNC is likely one reason for its lower selectivity for O₂ reduction. Computational studies of O₂ reduction by NOR suggest a proton and electron are required to form an Fe^III–OOH intermediate prior to O–O bond cleavage, while such an intermediate in HCO (I_p) is considered a transient step by virtue of Tyr oxidation and/or PT.³⁹⁶

5.1.4. Effect of PT Pathways.

Unlike HCO, NOR are not able to pump protons; although recently, some quinol- and Cu_A-dependent NORs were found to be electrogenic by virtue of consuming reducing equivalents from the periplasm and protons from the cytoplasm.^170,174 Natively, HCOs produce a proton gradient during O₂ reduction through a combination of a similar electrogenic property, along with their unique proton pumping mechanism. Remarkably, catalytic NO reduction by Rs cbb₃ oxidase was found to be non-electrogenic, and consumes both protons and electrons from the periplasm, instead of utilizing the cytosolic PT pathways employed for O₂ reduction.²⁰⁰ It is proposed that O₂ reduction intermediates can be protonated by the cytosolic PT pathway though a residue in the pathway with a relatively high pK_a, which cannot protonate NO reduction intermediates that have a lower proton affinity. Kinetics studies of O₂ reduction by Rs aa₃ oxidase show that a Glu residue near the cytoplasmic surface is important for transferring protons during O₂ reduction, and acts as a proton donor with a pK_a > 9.³⁹⁸ In contrast, O₂ reduction by PdNOR displays pH dependence based on a Glu residue implicated in a periplasmic PT pathway (Glu122) with pK_a ~ 6.6.³⁹⁹ Interestingly, O₂ reduction is not diminished when this residue is mutated to Asp, which appears to have a pK_a similar to the PT pathway in HCO, which is >9. Overall, these studies support the hypothesis that HCO utilize residues in PT pathways with higher pK_a values, which are likely important for energy conservation and directional proton pumping, but inefficient for PT to intermediates formed during NO reduction that have lower proton affinity.

5.2. HCO and SiRA.

While significantly less is understood of the sulfite reduction mechanism in the heme c/Cu active site of SiRA than is understood of O₂ and NO reduction in the heme/nonheme Cu/Fe centers of HCO and NOR, it is worthwhile to briefly touch on the intriguing similarities and differences of these catalytic centers (Figure 33). While HCO displays promiscuity for reduction of other substrates, such as NO, it is a notable feature of SiRA that it is highly selective for its native reaction. Unlike other SiRs, SiRA displays negligible activity towards nitrite, or other substrates.⁴⁰⁰ Despite its apparent similarity to the heme/Cu active site of HCO, O₂ does not react in a similar way by SiRA; instead, oxidation of the active site leads to loss of Cu and disulfide bond formation.²⁸⁵ The Cu^I coordination in SiRA ([(Cys)₂Cu]⁻) leaves the Cu ion with a net negative charge, which is distinct from the net positive charge on nonheme Cu or Fe ions in HCO and NOR. This negative charge likely stabilizes the S^IV state of the heme-bound sulfur atom prior to electron transfer. This effect would persist through the two subsequent 2 e⁻ reduction steps, but it would be abolished if the Cu^I ion were to donate an electron, in contrast to the redox-active nonheme Fe/Cu ions. The evidence so far seems to agree that the Cu ion is indeed redox inactive, but this question has only barely been explored. Another common feature between SiRA and HCO active sites is the presence of Tyr that participates in active site water coordination. The crystal structure of SiRA suggests that Tyr285 plays a role in PT with the added role of stabilizing the liberated water molecule once the substrate has been dehydrated.²⁸⁵ The presence of a similarly positioned, conserved Tyr in structurally related ccNiR and the substrate-dependent role it plays in nitrite vs sulfite reduction raises intriguing questions regarding the full functional scope of the active site Tyr residues in SiRA and why SiRA demonstrates such high substrate selectivity for sulfite over nitrite compared to the siroheme-[4Fe-4S] SiRs. Obviously, given the extensive and informative work to understand the similarities and differences between HCO and NOR, it would be of great interest to further develop the functional scope of heme/nonheme Fe and Cu active sites and explore how the specific structural features of SiRA (or a biomimetic model of it) determine its reaction selectivity.

Figure 33.

(A) Active site structures of bovine aa₃ oxidase (PDB ID 2Y69) and (B) SiRA (PDB ID 4RKM) and coordination geometry and distance to heme in Cu_B of (C) aa₃ oxidase and (D) SiRA.

6. Outlook.

Multi-electron small-molecule transformations catalyzed by a single enzyme are among the most challenging in biology and are important for many biological processes from cellular respiration to the global cycles of the life-essential elements. To accomplish this difficult task, biology often employs heteronuclear metalloenzymes, such as the HCO, NOR and SiR active sites described in this review. However, most of these enzymes are less well-understood compared to mono- or homonuclear metalloenzymes, in part due to the inherent complexity of different metal ions arranged in close proximity. In this review, we have summarized progress that has been made in studies of each of these enzymes and their biomimetic models, with focus on insights gained in structural features responsible for each step of the reaction mechanisms.

Despite the progress made in the study of the individual enzymes, we still lack a holistic understanding of this class of heteronuclear metalloenzymes. Specifically, a major unresolved issue is what structural features are responsible for the differences in their functions; i.e., why is the heme/nonheme Fe in NOR effective in 2e⁻ reduction of NO to form the N–N bond, while the heme/Cu center in HCO is efficient in 4e⁻ reduction of O₂ to break the O–O bond, and why do the heme/Cu in SiRA and the even more complex heme-[4Fe-4S] in SiR perform 6e⁻ reduction of SO₃²⁻ to break the S–O bond? The answer to these questions will rely on a deep molecular-level understanding of these enzyme mechanisms, which is aided through complementary study of these native active sites and related biomimetic complexes.

List of abbreviations:

ET

electron transfer

NOR

nitric oxide reductase

HCO

heme-copper oxidase

SiR

sulfite reductase

PT

proton transfer

PROS

partially reduced oxygen species

QO

quinol oxidase

CcO

cytochrome c oxidase

BNC

binuclear center

AOX

alternative oxidase

FDP

flavodiiron protein

MCO

multicopper oxidase

PI

peroxy intermediate

NI

native intermediate

rR

resonance Raman

ORR

oxygen reduction reaction

swMb

sperm whale myoglobin

Cu_BMb

Leu29His Phe43His swMb

Fe_BMb

Leu29His Phe43His Val68Glu swMb

DCHIM

1,5-dicyclohexylimidazole

TMPA

tris(2-pyridylmethyl)amine

AN

bis(3-[dimethylamino]propyl)amine

imiTyr

imidazolyl-tyrosine

SAM

self-assembled monolayer

SERRS

surface-enhanced resonance Raman spectroscopy

l.s

low-spin

XRD

X-ray diffraction

cryo-EM

cryogenic electron microscopy

FNOR

flavodiiron NO reductase

Hcp

hybrid cluster protein

OEP

octaethylporphyrin

Tp

trispyrazolylborate

pz

pyrazolate

APS

adenosine-5’-phosphosulfate

ONR

octaheme nitrite reductase

MccA

multicytochrome c sulfite reductase A

aSiR

assimilatory sulfite reductase

Fd

ferredoxin

HP

hemoprotein

alSiR

low-molecular weight monomeric aSiR

dSiR

dissimilatory sulfite reductase

Dvir

desulfoviridin

NiR

nitrite reductase

SNiRR

sulfite or nitrite reductase repeat

MV

methyl viologen

aNiR

assimilatory nitrite reductase

ccNiR

cytochrome c nitrite reductase

OEC

7′,8′-dihydro-octaethylporphyrin

OEiBC

2′,3′,7′,8′-tetrahydro-octaethylrphyrin

SiR-FP

aSiR flavoprotein subunit