Crystal structures of nitric oxide reductases provide key insights into functional conversion of respiratory enzymes



Summary: Respiration is an essential biological process to get bioenergy, ATP, for all kingdoms of life. Cytochrome c oxidase (COX) plays central role in aerobic respiration, catalyzing the reduction of O2 coupled with pumping proton across the biological membrane. Nitric oxide reductase (NOR) involved in anaerobic nitrate respiration is suggested to be evolutionary related to COX and share the same progenitor with COX, on the basis of the amino acid sequence homology. Contrary to COX, NOR catalyzes the reduction of nitric oxide and shows no proton pumping ability. Thus, the respiratory enzyme acquires (or loses) proton pumping ability in addition to the conversion of the catalytic property along with the environmental change on earth. Recently, we solved the structures of two types of NORs, which provides novel insights into the functional conversion of the respiratory enzymes. In this review, we focus on the structural similarities and differences between COXs and NORs and discuss possible mechanism for the functional conversion of these enzymes during molecular evolution. © 2013 IUBMB Life, 65(3):217–226, 2013


Microbial denitrification is an example of anaerobic respiration, in which nitrate (NOmath image) is used as a terminal electron acceptor and is sequentially reduced to nitrogen gas (NOmath image → NOmath image → NO → N2O → N2) by several metallo-enzymes. Membrane-integrated nitric oxide reductase (NOR) is an iron-containing enzyme involved in the denitrification process and catalyzes the reduction of nitric oxide (NO) to nitrous oxide (N2O) with two protons and two electrons (2NO + 2H+ + 2e → N2O + H2O) via cleavage of the N[BOND]O bond and concomitant formation of the N[BOND]N bond.

From the fact that the product N2O from NOR-catalyzed reaction is a greenhouse gas 310 times as powerful as CO2 and is an ozone depleting substance (1), NOR fascinates environmental scientists. The main source of N2O emission into the atmosphere is bacterial breakdown of nitrogen compounds such as chemical fertilizer in soils. In other words, NOR is one of the main contributors to produce N2O and global warming. Furthermore, given that pathogenic bacteria uses NOR as a defense against NO produced from macrophage in host's immune system (2), the study in NOR is an area of interest for the pharmacologist and medical scientists. Recently, the analysis of the biofilm formed by pathogenic Neisseria gonorrhoeae indicated that NOR is an essential element for the biofilm formation (3), raising the biological importance of NOR.

The evolutional relationship with terminal oxidase (cytochrome c oxidase, COX) in aerobic respiration is another interesting aspect of NOR. COX, which catalyzes the reduction of O2 to water with four protons and electrons (O2 + 4H+ + 4e → 2H2O), shows sequence similarities to NOR (4, 5). Phylogenetic analysis suggested that NOR and COX are classified as members of the heme-copper oxidase superfamily and share the same ancestor protein (5–7). NORs and some COXs have cross-reactivity with respective substrate (8–10), further supporting the view that NOR is evolutionary related to COX. However, NOR shows no proton pumping activity (11, 12), whereas COX can generate the proton gradient by pumping protons in a process coupled with O2 reduction. The produced proton gradient is essential for efficient ATP synthesis. Therefore, elucidating in the structures of NORs and COXs has been desired to understand the functional conversion, for example, the catalytic reaction and proton pumping activity, of the respiratory enzymes during the molecular evolution.

The structural and functional properties on COX have been extensively studied since the crystal structure was solved in 1990. However, the elucidation of the structural factors contributing the functional conversion in the respiratory enzymes has been hampered by the lack of the structural information on NOR. Very recently, we solved the crystal structures of two different types of NORs; one is cytochrome c dependent NOR (cNOR) from Pseudomonas aeruginosa (13) and the other one is quinol dependent NOR (qNOR) from Geobacillus stearothermophilus (14). cNOR, a first characterized NOR from denitrifying bacterium, receives electrons from soluble proteins such as cytochrome c, and consists of two subunits called NorB and NorC. The NorC subunit has hydrophilic domain toward periplasm with heme c which accepts electrons from electron donor proteins. Larger NorB subunit contains heme b and heme b3/nonheme FeB binuclear active center in the transmembrane region (TM). qNOR is a related single subunit NOR that was observed in pathogenic nondenitrifyer as well as denitrifyer (5, 15). qNOR lacks heme c, but the overall structure is quite similar to that of cNOR. The structure of binuclear active center is well conserved between cNOR and qNOR. Two NOR structures provide us with new insights into common and different structural features in NORs.

Now, the structural information on phylogenetically distinct several types of COXs (aerobic A- and B-type COXs and micro-aerobic C-type COXs) and anaerobic NORs, included in heme-copper oxidase superfamily, is available, which enables us to discuss how nature controls the function of the respiratory enzymes along with the environmental changes from anaerobic to aerobic condition. In this review, we focus on the structural properties related to the functional differences between NORs and COXs and discuss possible mechanism of the functional conversion of the respiratory enzymes.


New findings from the structures of NORs provide novel information on the evolutionary relationship between NORs and COXs. We identified a Ca ion bridging heme propionates of heme b and heme b3 both in cNOR and qNOR (13, 14). The Ca ligands, glutamate and tyrosine (Glu429 and Tyr93 in G. stearothermophilus qNOR, and Glu135 of NorB and Tyr73 of NorC in P. aeruginosa cNOR), are highly conserved in NORs. The structure-based mutagenesis confirmed that the Ca ion is essential for the NO reduction activity in G. stearothermophilus qNOR (14). The Ca ion was also identified at the equivalent position in C-type cbb3 COX, and the one of the Ca ligands is conserved glutamate residue which corresponds to the Ca ligand in NORs (16). Most recently, Gennis and coworkers reported the functional importance of the Ca ion in C-type cbb3 COX (17), indicating that structural and functional aspects of the Ca ion are conserved between NORs and C-type COX. In A- and B-type COXs, on the other hand, positively charged groups from two conserved arginine residues locate at the equivalent position to the Ca ion (18–21). It is therefore likely that NORs and C-type COX are closely related to one another and are evolutionary distinct from A- and B-type COXs, which is consistent with the view from the phylogenetic analysis (7, 22).

The difference in the structure of hydrophilic domain also supports the evolutionary relationship of NORs and C-type COX. The hydrophilic domain exhibits cytochrome c fold in C-type COX, cNOR, and even in qNOR which does not have heme c (13, 14, 16), whereas corresponding hydrophilic domain shows cupredoxin fold in A- and B-type COXs (18–21). Analysis on the structural homology of cytochrome c domains suggests possible evolutional link in NORs and C-type COX. Figure 1 shows the superimposed structures of the cytochrome c domains based on the structural homology. The cytochrome c domains of both qNOR and C-type cbb3 COX have extra α-helices compared to the NorC subunit of cNOR, although the positions of the extra helices are different between qNOR and C-type cbb3 COX. Therefore, there is no substantial structural homology directly between hydrophilic cytochrome c domain in G. stearothermophilus qNOR and cytochrome c domain of P. stutzeri cbb3 COX. It could be likely that cNOR has evolutionary links with both qNOR and C-type COX, but that qNOR and C-type COX are not directly related each other.

Figure 1.

Superposition of the soluble cytochrome c domains of the respiratory enzymes based on the structural alignment. The structural data of G. stearothermophilus qNOR, P. aeruginosa cNOR, and P. stutzeri cbb3 COX were used for the alignment. (A) NorC subunit of cNOR (yellow) and qNOR (green). (B) NorC subunit of cNOR (yellow) and cytochrome c domain of C-type cbb3 COX (magenta). (C) qNOR (green) and cytochrome c domain of C-type cbb3 COX (magenta). The marked structural differences were highlighted by dotted circles.

While we suggested the possible evolutionary links among NORs and C-type COX, the order of the appearance of C-type COX and NORs has not been established yet. Recent taxonomic study on COXs and NORs showed the congruence of A-type COX and 16S rRNA tree at the Archaea/Bacteria domain and suggested the presence of A-type COX in common ancestor (22). On the other hand, based on the assumption that NO was present in the atmosphere prior to O2, Castresana et al. (23) suggested that NORs were the first respiratory enzymes to appear, and C-type COXs was evolved from NORs. Our structural data on NORs cannot clarify the order of the appearance of the respiratory enzymes. However, the structural comparison of COXs and NORs may propose that the drastic functional changes occurred between cNOR and C-type COX.


Structural similarities are observed in the binuclear catalytic center in COX and NOR as shown in Fig. 2. For example, all histidine ligands for the binuclear center are conserved in these respiratory enzymes. In spite of the conservation of the ligands, however, the nonheme metal is different between COX and NOR. The catalytic site of NOR is composed of heme and FeB, whereas the O2 reduction sites in all known A-, B-, and C-type COXs are the heme/CuB binuclear center. This means that non-heme metal is a crucial factor controlling the catalytic reactivity of the respiratory enzyme. To understand the roles of the nonheme metal in the catalysis, Lu and coworkers created an engineered myoglobin (Mb) mimicking the binuclear active center of NOR and examined the effect of nonheme metals including Fe2+, Cu+, and Zn2+ on the function and structure (24). The functional analysis indicated that the heme/Zn2+ system showed no activity due to redox inactive property of Zn, but both Fe2+ and Cu+ bound samples reduced NO to produce N2O in the engineered Mb (24). Subsequent spectroscopic analysis showed that Fe2+ induced remarkable change in the electronic structure of heme-bound NO, which was evident from 50 cm−1 of low frequency shift of νNO upon Fe2+ binding (25). In contrast, Cu+ binding did not affect the heme–NO moiety (25). Nonheme metal may modulate the electronic structure of the heme-bound substrate and could optimize the catalytic reactivity in the respiratory enzymes.

Figure 2.

Comparison of the binuclear catalytic centers of respiratory enzymes. (A) P. aeruginosa cNOR (PDB code: 3O0R). (B) G. stearothermophilus qNOR (PDB code: 3AYF). (C) Bacterial B-type ba3 COX (1XME). (D) P. stutzeri C-type cbb3 COX (3MK7).

The coordination structure for the nonheme metal is quite similar in cNOR and qNOR, albeit unexpected replacement of nonheme FeB with Zn for the crystallization in G. stearothermophilus qNOR (Figs. 2A and 2B). The position of one of the histidine ligands in COXs is, however, changed by the characteristic histidine-tyrosine covalent linkage, creating more planar coordination geometry around the nonheme metal, which is favorable for the Cu coordination (Figs. 2C and 2D). It is interesting to note that His-Tyr crosslinkage and the planar geometry of the CuB-ligand set is a conserved structural motif in all COXs even though that recruited Tyr residues are not conserved in C-type COX compared to A- and B-type COXs (16). Thus, in the respiratory enzymes, the amino acid residues around the binuclear center can select the type of nonheme metal to construct an active site according to the catalytic function.

Comparison of the critical residues at the binuclear center of COXs and NORs suggests that at least one substitution of the active site residue in NORs is required for the construction of COX-type heme/CuB center. The substitution of essential glutamate (Glu211 of cNOR, Glu512 of qNOR) with tyrosine is needed to create the binuclear catalytic center of A- and B-type COXs. On the other hand, the introduction of tyrosine into the corresponding position, in addition to the removal of the active site glutamate, is essential for mimicking the active site of C-type COX. These simple mutations of the active site residues, however, are supposed to be not enough for the complete construction of COX type heme/CuB site in NOR, because a maturation factor, a copper chaperone, is responsible for the formation of the CuB center. Although it is not simple to convert the active site structure of the respiratory enzymes, the amino acid substitutions at the binuclear active center could be a crucial step for the functional conversion of the respiratory enzymes.

The molecular mechanism of the NO reduction by NOR still remains to be solved, while such knowledge is invaluable to understand the structural reason for the functional difference between NOR and COX. There are three proposed mechanisms on the basis of the structural, spectroscopic, and theoretical studies (26–29). Using a rapid freeze quench method combined with electron paramagnetic resonance spectroscopy, we found that simultaneous formation of ferrous heme-NO and ferrous FeB-NO at 0.5 ms after the introduction of NO into fully reduced cNOR (26). From this observation, we suggested trans-mechanism; two NO molecules bind to heme b3 iron and FeB, respectively and produce trans-hyponitrite intermediate upon the N[BOND]N bond formation. Trans-mechanism is also supported by the spectroscopic study on a synthetic model of the heme/nonheme binuclear active site of NOR (30). However, Blomberg and Siegbahn (28) concluded from recent theoretical study, using our cNOR structure that their previously proposed cis-heme b3 mechanism is energetically feasible, and that trans mechanism is energetically unfavorable. In cis-heme b3 mechanism, first NO molecule coordinates to heme b3, followed by the cis-hyponitrite formation upon the direct binding of second NO to heme-bound NO. There is also alternative suggestion of cis-FeB mechanism in which two NO molecules bind to nonheme FeB (29), because ferrous heme–NO species would be highly stable dead end product.

In any proposed mechanisms, two NO molecules are needed to accommodate to binuclear center of NOR. The active site structure of P. aeruginosa cNOR indicates that the coordination sphere is crowded due to the close distance between heme iron and nonheme FeB (3.8 Å) and shows no space for accommodating two NO molecules. On the other hand, G. stearothermophilus qNOR has larger space at the binuclear active center, which is confirmed by elongated heme iron–nonheme ZnB (4.6 Å). In addition, Glu512 in qNOR, corresponding to the FeB ligand Glu211 in cNOR, dissociates from nonheme ZnB, increasing the active site space. These observations suggest that some conformational changes at the binuclear active center in NOR would be involved in the NOR catalysis (31, 32). Dissociation of the glutamate ligand from FeB is a possible structural change for the accommodation of two NO molecules to the active site (31).

It is noteworthy that, even though only one O2 molecule binds to the active site during the O2 reduction by COX, the distance between heme and nonheme CuB is longer than those of NORs. Table 1 summarizes the distances between heme iron and nonheme metal in the NOR and COX structures. Although it is not simple to compare the heme–nonheme metal distances of NORs and COXs because of the different oxidation and/or ligand-bound states and X-ray induced reduction during the data collection, the heme iron–FeB distance in cNOR (3.8 Å) is significantly shorter than the heme–CuB distance in COXs (the average ± standard deviation calculated from the values in Table 1 is 4.9 ± 0.3 Å). In qNOR, relatively short heme iron–ZnB distance (4.6 Å) is observed, as compared with those of COXs. Such short distance of heme iron and nonheme FeB could allow two NO molecules to be closer for the effective N[BOND]N coupling, facilitating the N2O generation. For more detailed discussion on the functional meaning of the heme–nonheme metal distances in the respiratory enzymes, we need to solve the NOR structures with different oxidation and/or ligand-bound states.

Table 1. Heme Fe–nonheme metal distances in the respiratory enzymes
inline image

As shown in Figure 2, the positional difference of the His ligand for nonheme metal between NOR and COX likely controls the heme/nonheme metal distance. In NORs, one histidine ligand for nonheme FeB, His207 in P. aeruginosa cNOR and His508 in G. stearothermophilus qNOR, locates on the axis connecting heme iron and FeB. Therefore, this histidine residue works as a cap to suppress the movement of nonheme FeB away from the heme. On the other hand, the position of the corresponding histidine residue in COX is in the plane consisting of CuB and the other two histidine ligands and largely deviates from the heme iron–CuB axis. Nonheme CuB could easily move along the axis perpendicular to the heme plane, resulting in longer heme iron–CuB distance. To get more insights into the NO reduction mechanism, we are currently working on the X-ray structural analysis on different oxidation and ligand-bound states of NOR and attempt to apply new technique using X-ray free electron laser for structural characterization of the short-lived reaction intermediates.


Proton transfer mechanism is also important to elucidate the functional conversion of the respiratory enzymes. The proton transfer (pumping) pathways are composed of hydrophilic protonatable residues and water molecules in the hydrophobic TM region, allowing protons to reach the active site in the TM region and to move across the hydrophobic membrane. The mechanism of proton transfer and pumping was most extensively studied in A-type COXs. A-type COXs have at least two proton transfer pathways called K- and D-pathways from the inside of the cellular membrane to the active center (39, 42). Some protons are used for the catalytic O2 reduction at the active center, and the other protons are further transferred to the outside of the membrane through proton pumping pathway(s) which is not clearly identified yet. B-type ba3 COX only has well-defined one proton transfer pathway analogous to K-pathway in A-type COX for the O2 reduction and proton pumping (20). Similar single proton transfer pathway (K-pathway) is also suggested in C-type cbb3 COX (16).

Contrary to COXs, the proton transfer pathway in NOR was less established, but biochemical and biophysical studies unambiguously showed no proton pumping activity and the catalytic proton transfer from the outside of the membrane in cNOR (11, 12). The structure of P. aeruginosa cNOR demonstrated that there is no proton transfer pathway from the inside of the membrane and suggested two possible proton transfer pathways, channels 1 and 2, from the outside of the membrane (13) (Fig. 3). Structure-based molecular dynamics (MD) simulation further supports this view (45). By contrast, we could not find the obvious proton transfer pathway from the outside of the membrane in G. stearothermophilus qNOR (14). Conserved key aspartate, Asp198 in P. aeruginosa cNOR, in channel 1 of cNOR is substituted with the other type of residue in qNOR (Ala499 in G. stearothermophilus qNOR) (Fig. 3), which collapses channel 1 in qNOR. Channel 2 is located at the C-terminal region of the NorC subunit in cNOR. Because the C-terminus of hydrophilic region that corresponds to the NorC subunit is fused to the catalytic core domain in qNOR, the structural characteristics of channel 2 region is largely different in between cNOR and qNOR (Fig. 3). Thus, both channels 1 and 2 are nonoperational owing to the structural properties unique in qNORs.

Figure 3.

Comparison of the regions of proposed proton transfer pathways in cNOR and qNOR. (A) P. aeruginosa cNOR (PDB code: 3O0R). (B) G. stearothermophilus qNOR (PDB code: 3AYF). Computation with Caver program (44) found two hydrophilic cavities from the outside of the membrane in cNOR (channels 1 (cyan) and 2 (green)) as potential proton transfer pathways, whereas no such cavity from the outside of the membrane was identified in qNOR. Instead, there is a water-containing hydrophilic cavity, designated as a water channel, from the inside of the membrane in qNOR (magenta). Enlarged panels are for the comparison of the proposed proton transfer pathways and corresponding regions in cNOR and qNOR.

We unexpectedly found a water containing hydrophilic channel designated as a water channel from the inside of the membrane to the active site in G. stearothermophilus qNOR (Fig. 3) (14). The mutagenesis and MD simulation indicate that the water channel serves as the proton transfer pathway for the NO reduction reaction in G. stearothermophilus qNOR. The amino acid residues along with the water channel are not fully conserved, but the type of the residues (hydrophilic and/or small residues) are highly conserved in qNORs. Taken together with the fact that the absence of the proton transfer pathway from the outside of the membrane would be general aspect in qNORs, we can propose that the presence of the water channel from the inside of the membrane as the catalytic proton transfer pathway is common to qNORs.

We have data possibly related to the different proton transfer pathway between cNOR and qNOR. The pH dependence on the reaction with O2 in G. stearothermophilus qNOR was markedly different from that in Paracoccus denitrificans cNOR (46). Previous work on Pa. denitrificans cNOR indicated that pKa of the reaction of fully reduced cNOR with O2 is ∼6.6, which can be the pKa of an internal protonatable residue on the proton transfer pathway (47). In contrast, the reaction between fully reduced G. stearothermophilus qNOR and O2 showed no pH dependence in the range from pH 5.5 to 9.5 (46), raising the possibility that qNOR has different proton transfer pathway from that of cNOR. Furthermore, our preliminary analysis on the pH dependence of the NO consumption rate indicates that G. stearothermophilus qNOR shows higher activity in alkaline condition than that in acidic condition, whereas acidic condition is optimal for P. aeruginosa cNOR. In addition, the NO reduction activity was enhanced by the reconstitution into liposome in G. stearothermophilus qNOR, although such effect was not observed in P. aeruginosa cNOR (E. Terasaka, T. Tosha, Y. Shiro, unpublished observation). The difference in the proton transfer pathway between cNOR and qNOR might explain the reasons why we observed different enzymatic properties. Further studies on qNORs from other sources will verify our proposal of the catalytic proton transfer from inside of the membrane.

Interestingly, the location of the water channel in qNOR is similar to K-pathway in COXs, while there is no evidence for the electrogenicity in G. stearothermophilus qNOR. The examination of the electrogenicity of qNOR by electrometric measurement combined with the flow-flash method (12) is under way. Comparing the water channel in G. stearothermophilus qNOR with the corresponding region in P. aeruginosa cNOR gives an idea on the creation of the proton transfer pathway from the inside of the membrane in cNOR, which could be related to the origin of the proton pumping ability. The structure of P. aeruginosa cNOR shows that three conserved glutamate residues locates at the close proximity to the active site and forms a small hydrophilic cavity. However, the residues equivalent to Gln545 and Glu591 in the water channel of G. stearothermophilus qNOR are substituted with hydrophobic Ile244 and Phe290 in P. aeruginosa cNOR, respectively (Fig. 3). These hydrophobic residues are highly conserved in cNOR, and likely form a bottleneck for the water channel from the inside of the membrane, suggesting that a few mutations in cNOR might create water channel like that in qNOR. In fact, I244Q/F290E cNOR variant, which is created by computation, has a cavity from the inside of the membrane to the active site (14).

The presence of the water channel in G. stearothermophilus qNOR can be a clue for how the respiratory enzyme acquires the proton pumping ability. However, it is not straightforward to add the proton pumping ability to NOR by simple amino acid mutations, as the proton pumping pathway in COX requires highly regulated machinery including an energy-driven gate. Extensive study on the proton pumping mechanism in A-type bovine aa3 COX by Yoshikawa and coworkers indicates the several structural changes around the heme active site and putative proton pumping pathway upon the reduction of the heme and/or the binding of the ligand to the heme (42). Similarly, Ferguson-Miller and co-workers observed the redox change- and/or ligand binding-induced structural changes in aa3 COX from Rhodobacter sphaeroides (39). As a transduction pathway for the structural changes, recent time-resolved spectroscopic study on bovine aa3 COX suggested that the vinyl group of heme a3 could be involved (48). The peripheral group equivalent to vinyl group of heme a3 in A-type COX is conserved in the other B- and C-type COXs (Figs. 2C and 2D), whereas the equivalent position is methyl group in NORs due to the flipped heme b3 (Figs. 2A and 2B). This difference in heme peripheral group between COXs and NORs might be one of the determinants for the proton pumping ability in the respiratory enzymes. Further structural studies on NORs in different oxidation and ligand-bound states as well as on B- and C-type COXs, which only have a single K-pathway for proton pumping, help us understanding the structural elements required for the proton pumping ability.


The authors thank all our coworkers involved in the studies on NORs; T. Hino (Tottori Univ.), S. Nagano (Tottori Univ.), Y. Matsumoto (Kyusyu Univ.), A. V. Pisliakov (RIKEN ASI), Y. Sugita (RIKEN ASI), H. Sugimoto (RIKEN RSC), E. Terasaka (RIKEN RSC), and Pia Ädelroth (Stockholm Univ.).