Abstract
In all kingdoms of life, the regulation of membrane-bound enzyme function via oligomerization is a fundamental aspect of cell physiology. Often, the mechanistic role of oligomerization is unclear, due to a lack of structure-function comparisons between constituent forms of the enzyme. Here, we elucidate the structural underpinnings of enzyme regulation and oligomerization in the quinol-dependent nitric oxide reductase (qNOR) from Neisseria meningitidis, by high-resolution structural analyses of the less active monomeric form (2.25 Å) and the highly active dimeric form (1.89 Å). The comparison revealed that broad helical flexibility near the dimer interface of the monomer causes a conformational change in a critical amino acid near the active site, located apart from the dimer interface. We demonstrate that the crosstalk between the dimer interface and catalytic site in qNOR allows enhanced activation of the enzyme via dimerization. Given Neisseria meningitidis’ dependence on qNOR to detoxify the host’s immune response of nitric oxide, our results pave a way for new strategies to combat bacterial infections, via the inactivation of qNOR by monomerization. More broadly, this provides new insights into the role of membrane protein oligomerization and its influence on regulating activity.
Similar content being viewed by others
Introduction
Respiratory enzymes that belong to the heme-copper oxidase (HCO) superfamily play a vital role in cellular respiration. In aerobic respiration (oxygen used as a terminal electron acceptor), cytochrome c oxidase (CcO) or quinol oxidase catalyzes the reduction of oxygen (O2 + 4H+ + 4e− → 2H2O), at a heme/Cu binuclear active center1. Conversely, other members of the HCO, cytochrome c-dependent nitric oxide reductase (cNOR) and the closely related quinol-dependent enzyme (qNOR), whose active sites consist of heme and non-heme iron (FeB), are crucial for one kind of anaerobic respiration, so-called denitrification2. In addition to the critical roles of NORs in reducing non-oxygen electron acceptors (nitrogen oxides), catalytic nitric oxide (NO) reduction by NORs (2NO + 2e− + 2H+ → N2O + H2O) is responsible for the elimination of cytotoxic NO produced by hosts’ immune systems in some human pathogens, including Pseudomonas aeruginosa3,4, Neisseria meningitidis (Nm)5,6, and Staphylococcus aureus7. Given that cellular respiration is essential in many organisms and that NORs of pathogens are critical for survival in their hosts, HCO enzymes are a key target for the development of antimicrobial drugs. Indeed, recent work on screening for an allosteric inhibitor against the HCO enzymes demonstrated great potential for obtaining new antimicrobial reagents8. Thus, gathering detailed structural and functional information on the HCO enzymes is imperative for future drug design efforts.
Despite a lower abundance of structural information on NORs compared to the extensively studied CcOs, we have conducted several structural studies on NORs and found that qNOR could exist as both a monomer and dimer9,10,11. The biochemical data on qNOR showed that the dimer exhibits 2 ~ 4-fold higher catalytic activity than the monomer in qNOR from Achromobacter xylosoxidans (Ax) and Neisseria meningitidis10,12. These observations suggest that the dimer represents the active form in qNOR. Structural analysis with single-particle cryogenic electron microscopy (cryoEM) revealed that transmembrane helix 2 (TM2), an extra helix that is absent in cNOR, assisted the formation of the dimer in qNORs9,10,11,13. However, the distinct reason why the dimer showed higher activity than the monomer in qNOR is not clear, due to a lack of a precise structural comparison of the monomer and the dimer.
Even in the case of other members of the HCO family, like CcOs, the relationship of the monomer-dimer states to enzymatic function is a principal topic. Recent advances in structural biology provided the fact that CcOs usually form supermolecular complexes in the biological membrane14,15,16,17,18,19,20,21. For example, human CcO is involved in a supermolecular complex consisting of two monomers of complex I (CI2), two monomers of complex III (CIII2), and two monomers of complex IV (cytochrome c oxidase) (CIV2)22, wholly denoted as CI2CIII2CIV2. However, a supermolecular complex in yeast contains two monomers of complex III and two monomers of CcO18,23 (CIII2CIV2) or x1 CcO24 (CII2CIV). It is interesting to note that CcO participates in the supermolecular complex exclusively as a monomer, although the crystal structure of the broadly studied bovine CcO is a dimer25. Furthermore, recent work on the monomer-dimer topic on bovine CcO revealed that the oligomeric state is linked to the function of CcO, showing that the monomeric form exhibited 2–5-fold higher enzymatic activity than that of the dimer, implying that the monomeric form would be the active form for bovine CcO26. The structural analysis of each oligomeric state of CcOs enriched our understanding of the structure-function relationship in CcOs. Therefore, the structural information on the monomer and dimer forms of qNOR will shed light on the molecular mechanism for effective NO reduction in qNOR, leading to further elucidation of the functional importance of constituent oligomeric states of the HCO superfamily.
To that end, we determined the NmqNOR monomer (~85 kDa) structure to 2.25 Å resolution and re-analyzed the NmqNOR dimer to 1.89 Å resolution by single-particle cryoEM. The data shed light on how qNOR monomerization deforms a key dimer stabilizing intra-helical arrangement involved in proton transfer and alters the stability of a functionally critical glutamate. These help us further comprehend the structural elements required for an effective catalytic reaction in qNOR, explaining the structural basis for superior enzymatic activity in the dimer.
Results
High-resolution single-particle cryoEM structure of NmqNOR dimer
After purifying and isolating dimeric and monomeric NmqNOR from the same bacterial culture, we measured the enzymatic activity of each oligomeric form and found ~4-fold lower activity in the monomer compared with the dimer (Table 1 and Supplementary Fig. 1). The difference optical spectra (reduced minus oxidized UV-Visible absorption spectra) of each respective form are identical (Supplementary Fig. 1e), in both the Soret region (~407 nm) and Q-band region (500–600 nm), the latter of which is sensitive to changes in local environment (ligands, solvent, oxidation state). To understand the structural origin of the lower enzymatic activity of the monomer, we solved the structures of both the monomer and dimer states of NmqNOR using the same cryoEM instrument, a CRYO ARMTM 30027,28 (JEOL Ltd.). Owing to the optimization of the conditions for the grid preparation, even though the sample preparation for NmqNOR was fused with apocytochrome b562 (BRIL) at the C-terminus (details of the sample preparation are in the Methods section), the resolution of NmqNOR dimer was improved to 1.89 Å from 3.1 Å10 in this study (Supplementary Figs. 2–4 and Supplementary Table 1). The high-resolution structure of the NmqNOR dimer was essentially the same as our previously determined structure10, in which each protomer contains 14 TM helices with heme b and the heme b3/non-heme FeB binuclear active center, except for the identification of water molecules. The high-resolution density map for the NmqNOR dimer allowed us to newly model several water molecules (Fig. 1a). One water cluster is located at the proposed quinol binding site near heme b, Asn628, Ser686 and Arg724. Another chain of water molecules is observed around the propionates of heme b3 and Ca ion, as observed in the other qNORs (Ax and Geobacillus stearothermophilus (Gs)) and even in cNOR from P. aeruginosa13 (Supplementary Figs. 5 and 6). Notably, based on our structure, only a few water molecules are observed in the presumed water channel from the cytoplasmic side to the active center in the high-resolution NmqNOR dimer (Fig. 1b), unlike the similarly high-resolution (2.25 Å) cryoEM structure of the AxqNOR dimer29 and X-ray crystal structure of GsqNOR11. This observation suggests that the apparent lack of ordered water molecules in the putative water channel of the NmqNOR dimer could be due to a disordered nature of the water molecules in the channel. Whether this is directly correlated with the increased catalytic activity of NmqNOR in general is not clear, and the assignment method of water molecules across different qNOR structures may not be consistent, due to variations in attained resolution and the techniques used for structure determination.
A Cylinder representation of the NmqNOR dimer, with individual protomers colored in slate blue and hot pink. Water molecules are shown as red spheres surrounded by transparent red density map, with heme b shown as orange sticks and heme b3 as cyan sticks. The lipid bilayer is depicted as a beige rectangle. B View of the potential proton transfer channel in one protomer from the cytoplasmic end toward the active site heme b3 and non-heme iron (brown sphere), with selected hydrophilic side chain residues shown as blue sticks with surrounding density in transparent pink. Glu494 density is not shown for clarity purposes (see Fig. 3A).
Single particle cryoEM structure of the NmqNOR monomer
The NmqNOR monomer is ~85 kDa, asymmetric, and lacks large, extracellular domains, making it a relatively challenging target for sub-3 Å cryoEM reconstruction. Nevertheless, we obtained a structure with a resolution of 2.25 Å from 274,346 particles using data collected on a CRYO ARMTM 300 (JEOL Ltd.) and cryoSPARC v4.1.1 for image processing (Supplementary Figs. 7–9 and Supplementary Table 1). The density for the BRIL portion is diffuse, so no model could be built for it, as was the case with the dimer. The overall structure of monomer NmqNOR is almost identical to a protomer of the NmqNOR dimer (Root mean square deviation = 1.35 Å for Cα atoms). Akin to the high-resolution dimer structure, the EM map of the monomer revealed the location of water molecules near the possible quinol binding site and the propionates of heme b3, yet not in the putative water channel from the cytoplasmic side (Fig. 2a, b). We used CAVER in an attempt to “model” potential channels that originate from the cytoplasm to the active site in NmqNOR. A possible pathway originates close to the dimer interface (between TM2 of one monomer and TM10 of the other monomer), and travels through a pathway sandwiched between TM10 and TM11 (Supplementary Fig. 10a). The pathway is lined or, supported by several hydrophilic residues (R572, E573, E576, H577, E498, and E563) that are located on TM10 and TM11. R572, H577, and H257 (TM2) were selected as mutation candidates (yet to have been mutated in previous studies), yet, upon mutation to alanine, they all retained identical activity and gel filtration separation compared to wildtype (Table 1). This suggests that, as isolated variants, they are not critical to the catalytic performance of NmqNOR.
A Cylinder representation of the NmqNOR monomer (gold) structure, with water molecules shown as red spheres surrounded by transparent density, heme b as red sticks and heme b3 as teal sticks. The lipid bilayer is depicted as a gray rectangle. B View of the potential proton transfer channel in ribbon representation, from the cytoplasmic end to the active site, with selected hydrophilic side chain residues shown as gold sticks. Non-heme iron and calcium ions are shown as brown and green spheres, respectively.
CAVER analysis revealed that in contrast to the dimer, the pathway originates just in front of the monomer TM2 loop and extends toward the active site (Supplementary Fig. 10b). The channel width is also larger compared to the dimer, suggesting that the hydrophilic channel in the monomer is more diffuse as a result of the TM10 flexibility, which may impact the ability to maintain an apparently more direct channel, like in the dimer.
Structure comparison of the NmqNOR monomer and dimer
The reconstructions at comparable resolutions for the monomer (2.25 Å) and the dimer (1.89 Å) of NmqNOR allow us to directly compare the structure of the monomer with that of the dimer to get insights into the enzymatic activity difference between the two forms. In particular, the significance of higher resolution reconstructions can divulge changes in water networks, side chains, and/or heme propionate conformation. These components are all pivotal to enzyme functionality, with even subtle alterations potentially causing large differences in enzyme reactivity.
I) Heme/non-heme Fe binuclear active center
Figure 3 represents the catalytic site of NmqNOR, consisting of a high-spin heme b3 and a non-heme iron, FeB, in both the monomer and dimer states. The coordination structure and the geometry of the heme b3/FeB active center are comparable in the monomer and the dimer. For example, in addition to three conserved histidine residues (His490, His541, and His542), the conserved Glu494 coordinates FeB in a monodentate manner both in the monomer and the dimer. The distances between heme b3 iron and FeB in the monomer and the dimer are 3.64 and 3.9 Å, respectively, both of which are reasonable distances to accommodate the presence of an μ-oxo bridging ligand30,31,32. We modeled water as a bridging ligand in the dimer active site; however, due to the lower resolution and relatively weaker density bridging the irons in the monomer, we did not assign a bridging ligand in the map. However, the dynamic property of conserved Glu494 could likely be altered between the monomer and dimer. In the dimer, the density assignable to Glu494 was only clear at the low sigma level (Fig. 3a), indicating the flexible nature of this residue. However, this property of Glu494 was not observed in the monomer active site (Fig. 3b), despite similar reconstruction resolutions and total electron dose applied during data collection.
Active site of the dimer (slate blue) (A) and monomer (gold) (B) with model and map representation. Metal-ligand bonding is shown with gold dashes, whilst water (red spheres)-ligand bonding is shown with black dashes. Numbers indicate the distance in Å. View of the upper half of TM10 and associated Glu563 residue, in the context of the active site for both dimer (C) and monomer (D) structures. Heme b3 is shown as gray and teal sticks for the dimer and monomer structures. Non-heme iron and calcium ions are depicted as brown and green spheres, respectively.
In addition to possible changes in the dynamics of Glu494, the orientation of conserved Glu563, in the second coordination sphere for FeB, is different between the monomer and the dimer of NmqNOR (Fig. 3c, d). The carboxylate group of Glu563 is facing towards FeB and is within hydrogen-bonding distance (3.3 Å) of the carboxylate group of Glu494 in the dimer, whereas the side-chain of Glu563 is flipped away (or, not in a metastable conformation) from FeB in the monomer. Since Glu563 is suggested to work as a proton donor for catalytic NO reduction at the end of the water channel11,33, it is highly plausible that the dynamic, conformational difference of Glu563 is associated with the lowered enzymatic activity of the monomer than that of the dimer.
II) TM helices involved in dimerization
Given that Glu563 is located on TM10, a helix responsible for the dimerization of NmqNOR, it is of great interest to compare the structural properties of TM10 in the monomer and the dimer. The current high-resolution structure of the NmqNOR dimer indicated that TM10 is responsible for dimerization (Fig. 4a). Ala574 and Trp578 of TM10 and Val598, Phe602, and Ile606 of TM11 (chain A) effectively sandwich the lower portion of TM2 via interactions with Leu246, Trp249, and Phe253 from TM2 of the other monomer (chain B) enabling dimerization (Fig. 4a). Additionally, Leu569 in TM10 and Ile244 in TM2 of each monomer forms a hydrophobic core to make a four-helix bundle (TM2A, B and TM10A, B) like structure (Fig. 4b).
A TM2 of one protomer (slate blue) forming intrahelical interactions with TM10 and TM11 of the other protomer (hot pink), mediated by W249, F253 on TM2, F602 on TM11, and W578, and Y575 on TM10. Residues are contoured with the sharpened map densities. B View from the periplasmic side showing the qNOR dimerization site; the four-helix bundle-like structure (TM2 x2, TM10 x2) and several hydrophobic residues (Ile244, Met248 and Leu569) contributing to the interaction. Residues are contoured with the sharpened map densities. C The monomer structure’s dimerization site shows an absence of TM2 from the other protomer. Unsharpened density is shown for the whole region, with TM10 displaying broader, smeared density. D View from the periplasmic side of (C), showing selected residues from TM2 and TM10 which normally contribute to dimerization. Leu569 conformational change is displayed by the curved arrow, with the residue contoured with unsharpened density. The vacant pocket of the dimerization site in the monomer is displayed with a gray oval. E Gel filtration profile of the Ile244Phe/Leu569Phe mutant (black trace), with purple and gold traces corresponding to wildtype dimer and monomer elution profiles, respectively. All traces are showing 280 nm absorbance after being run down a Superdex 200 10/300 column.
In the wildtype monomer form, TM10 is notably free from any other interactions (Fig. 4c). The lower portion contains residues Tyr575 and Trp578, which help form the dimer interface, and the accompanying density is much broader and less defined than the surrounding density (Fig. 4c). In the absence of the four-helix bundle-like structure, the monomer Leu569 (on TM10) is rotated away from this pocket (Fig. 4d). From the observed differences in the dimerization interface between the dimer and monomer, we wanted to assess the importance of I244 and L569 in maintaining the dimeric form by constructing a double mutant (Ile244Phe/Leu596Phe). After purification and gel filtration, we observed a reversal of the dimer: monomer ratio compared to wildtype (Fig. 4e). Thus, the introduction of the bulkier phenylalanine perturbs the dimer interaction at the four-helix bundle site and can induce monomerization. Furthermore, the ensuing activity measurements of both dimer and monomer fractions of the double mutant exhibited roughly ~25% specific activity compared to that of the wild-type dimer (Supplementary Fig. 16). This could be due to an altered structure around the potential proton transfer pathway, affecting overall activity, whilst this dimer variant may also have a higher propensity to dissociate into monomer- thus exhibiting monomer like loss of activity.
The above observations suggested that the monomer NmqNOR TM10 may have increased flexibility compared to the dimer TM10. To improve the clarity in these regions, we performed focused 3D classification (with no alignment) and masked refinements. However, this did not yield reconstructions with discrete conformations of the full region of TM10 or TM2. In an attempt to describe the potential, continuous motion of TM10 and 2 within the monomer particle images, we performed 3D Variability Analysis (3DVA) in cryoSPARC v4.4.134, which has been successfully employed in studies of G-protein coupled receptor dynamics35,36,37. As proteins often exhibit some degree of flexibility for their mechanism, there are potentially many possible 3D structures that it can adopt. Instead of reconstructing one single 3D structure from particle images, i.e., assuming one rigid conformation, this algorithm reconstructs a family of related 3D structures. These can help describe the various flexible conformations (subunit rotation/hinge-like movement, helical bending and swaying, etc.) that are present in the molecule. This provides a visual insight into the conformational space of a macromolecule using 3D movies to record the movements (frames).
After analyzing the NmqNOR monomer with a solvent mask excluding the detergent micelle (to focus only on the protein dynamics), significant swaying and bending of TM10 were observed, with additional minor, lateral movements of TM2 (Supplementary Movie 1–3). Aside from this, no other portion of the molecule displayed such dynamic motions. Across three different trajectories (variability components) in the space of 3D structures where the molecule exhibits variability, all of them resolved the swaying of TM10 back and forth into the dimer interface pocket, with a 7 Å maximal positional difference between the first and last frame of the 3D movie series (Fig. 5a). The region from Pro566 to Gly571 begins to display greater variability of helical angle and direction (Fig. 5a). Variability component 2 resolved intermediate movements compared to others, with more straightening of the bottom portion of TM10 (Fig. 5a). The conformational flexibility of TM10 in the dimer form of NmqNOR was also analyzed using 3DVA, yet, no such characteristic motion was observed in TM10 of either protomer (Supplementary Movie 4–6 and Fig. 5b), with only minimal, rigid body motion (slight rotation) of the protomers observed. These results indicated that only the monomer TM10 conformational landscape is less restricted (e.g., significant conformational variability), due to the lack of the neighboring protomers TM2 and TM10 interaction that stabilizes its conformation.
3DVA results from the NmqNOR monomer and dimer. A Representative snapshots from the 3DVA of the NmqNOR monomer. Density maps from superposed frames 1 and 20 from components 1 and 2 are displayed in the left and right columns, respectively. Large differences in TM10 conformation (shown by double-headed arrows) are evident in both component analyses. B Representative snapshots from the 3DVA of the NmqNOR dimer. Superposed frames 1 and 20 from component 1 and 2 are displayed in the left and right columns, respectively, in the same viewing orientation as shown in (A). Frames are almost indistinguishable from each other (comparatively little motion) compared to the monomer TM10 movements.
The difference in the dynamics of TM10 in the monomer and the dimer is illustrated by the modeled conformational and structural change in TM10 compared to the dimer (Fig. 6a). The TM10 helix in the dimer has a kink formed by highly conserved Pro566 and Gly571 residues (Fig. 6a, Supplementary Fig. 11), which creates a 35° bend in the helix. The dissociation of the dimer causes the outward swinging motion (~6 Å) that occurs from Gly571 onwards in TM10. The lower portion of monomer TM10 partially occupies the space where the opposing protomer TM2 exists, in the dimer form and induces a positional shift of TM2 by ~3.0 Å (Fig. 6a). Although the area around Glu563 in TM10 shows a relatively small positional shift in the monomer as compared with the dimer (Fig. 6b), the broader structural differences of TM10 involved in dimerization of NmqNOR may be accompanied by increased flexibility of the conserved Glu563 between the monomer and the dimer forms of NmqNOR (Fig. 6b).
A Comparison of TM10 between monomer (gold) and dimer forms (blue). Ribbon model representation showing the helical swing (black curved arrow) in the cytoplasmic region, after G571. The black box is shown in detail in (B). B Superposition of the active sites of NmqNOR monomer (gold) and NmqNOR dimer (slate blue), showing differing conformations of Glu563, which is located on TM10. Heme b3 is shown as transparent gray and teal sticks for the dimer and monomer structures, respectively.
III) Glu563 and the impact on activity and dimer stability
The substitutions of conserved Glu563 with Ala and Leu, both of which cannot form a hydrogen bond with Glu494, confirmed the functional importance of Glu563 and postulate a possible role in maintaining the dimer assembly. As summarized in Table 1, the substitutions of Glu563 lowered the NO reduction activity to less than 10% of the wild-type dimer, even in their dimeric states (Fig.7a). Difference optical spectra of the variants showed no major changes compared to the wildtype, with the Glu563Leu dimer showing a slight shift in the Soret peak (423 to 425 nm) compared to wildtype. Another minor difference was the less pronounced peak at 552 nm upon reduction in the Glu563Leu dimer sample (Fig. 7b and Supplementary Fig. 12). Based on this data, this indicates that the heme b environment in the variants was largely unperturbed. In addition to the crucial role of Glu563 in the enzymatic function, we found that the mutations also affected the dimerization ratio as evident from the size exclusion chromatography profile (Fig. 7c). An increase in the apparent void peak (V0) intensity of these variants may be attributed to general destabilization of the enzyme, possibly due to loss of the non-heme iron in the catalytic site. We performed iron quantification using the nitroso-PSAP method38 to assess the amount of free iron (non-heme iron) in the samples. This revealed that Glu563 variants contained ~0.2–0.3 Fe/qNOR, compared to ~0.8–0.9 Fe/qNOR in the wild type (Table 1). Analysis of the oligomeric content by blue native PAGE showed that both mutations of Glu563 increased the population of the monomer as compared with wild-type (Fig. 7d). We note that Glu563Leu dimer bands show weaker intensity compared to the wildtype, which is possibly to differences between the dimer-monomer equilibrium and/or timescale of dimer stability in Glu563Leu. Overall, Glu563 is suggested to be essential for enzymatic function and to be important for dimer formation via interaction with Glu494.
A NO reduction activity measurements (representative traces) of wildtype (WT) dimer (purple trace), Glu563Ala (black trace), and Glu563Leu (red trace) dimer samples. The addition of substrate NO and qNOR enzyme is indicated by arrows. B UV–Visible absorption of as-isolated Glu563Ala and Glu563Leu dimer samples after SEC analysis. The Soret band at 410 nm is indicated with a solid black line, which is identical to the WT Soret band position. C Size exclusion chromatograms of Glu563Ala (top) and Glu563Leu (bottom). V0, P1, and P2 refer to column void volume, peak 1 and peak 2 elution positions. Blue and red traces correspond to absorbance values at 280 and 410 nm, respectively. D Blue native-PAGE results of wildtype peak 1 (dimer) samples against Glu563Ala and Glu563Leu peak 1 sample. 0.4, 0.6, and 1.2 indicated the final, loaded concentrations in mg/mL for each lane. The molecular weight ladder is shown on the left with labels.
III) Structures relating to the electron transfer and NO binding pathway
The presumed quinol binding site, which is composed of conserved His303 and Asp728, is not significantly affected by the dimer formation in NmqNOR (Supplementary Fig. 13b, c). This suggests that the property of quinol binding is not a factor affected by dimerization. Furthermore, the redox potentials of hemes, which are related to the rate of electron transfer, are unlikely altered by the dimerization, since the water network around the active site and heme propionates, determinants of the redox potential39,40,41, are almost identical between the monomer and the dimer (Supplementary Fig. 13a). It is, therefore, less plausible that the electron transfer process is altered by the dimerization of NmqNOR, especially since the quinol binding site does not overlap with the dimer interface.
The possible NO binding and transport channel, a Y-shaped hydrophobic channel in the TM region42, is almost indistinguishable between the monomer and the dimer of NmqNOR (Supplementary Fig. 13d–f), suggesting that the substrate transport process could be independent of the oligomeric state of the enzyme.
Discussion
qNOR, a member of the HCO superfamily, was shown to have higher enzymatic activity in the dimeric state. However, the oxygen-reducing mitochondrial CcO was shown to have higher catalytic activity in the monomeric state26, as opposed to the extensively characterized dimer43. These observations suggest that the origin of higher catalytic activities between the two systems may differ. In our current study, upon obtaining and comparing high-resolution cryoEM structures of the monomeric (2.25 Å resolution) and dimeric (1.89 Å resolution) states of NmqNOR, we suggest that activity enhancement results from possible improved provision of protons via favorable changes in the global environment near the putative proton entry site, as was the case for monomeric CcO.
Our attained resolution for the NmqNOR dimer is one of the highest resolution single particle reconstructions for a macromolecule of its molecular weight range (< 200 kDa). Having structures of a similar resolution range determined using the same cryoEM instrument allowed us to perform a detailed structural comparison of the monomer and dimer, to help reveal structural insights into the origin of superior enzymatic activity in the latter. Amongst the three elemental processes required for the NO reduction reaction, substrate NO binding, electron transfer from quinol, and proton transfer, the lowered enzymatic activity in the monomer form is most likely attributed to suppression of the proton transfer process. We postulated that pH-dependent activity assay measurements may indicate if there is a perturbation in the proton transfer process, since an increase in pH may affect the amount of protons in solution and protonation states of key amino acids involved in catalysis/proton transfer, etc. After conducting measurements at pH 6, 8, and 9, we found that the specific activity of both dimer and monomer NmqNOR was essentially unchanged at pH 6 and 8, with ~25–40% decrease at pH 9 for both monomer and dimer (Supplementary Fig. 14). This showed that even in the highly active dimer, the rate-limiting step of the reaction could be linked to proton transfer dynamics under the current assay conditions. Thus, since the monomer already exhibits lower specific activity and has intact hemes and correct metallation, we postulate that TM10 flexibility and Glu563 disorder may be one factor that contributes to this extra reduction in activity in the monomer compared to the dimer, possibly via perturbation of proton transfer dynamics. However, further investigation, such as by hydrogen: deuterium exchange mass spectrometry (HDX-MS) and flow-flash technique combined with time-resolved UV-Visible spectroscopy, would be required to address the proton transfer dynamics and reaction rate limitation, as demonstrated by studies on bacterial CcO44 and P.aeruginosa cNOR45, respectively.
It was proposed from several structural and structure-based mutational studies that protons required for the catalytic reaction are supplied from the cytoplasm through a putative water channel in qNOR11,12,46. There is a hydrophilic cavity connecting the cytoplasm and the active site in the current cryoEM structures of both monomer and dimer, indicating that a possible water channel could be conserved in both states (Supplementary Fig. 10). The major differences from CAVER analysis were that the monomer hydrophilic channel was more diffuse (wider) and originated further inside the protein, away from the dimer interface region. This could originate from conformational changes in TM10 and TM2, which increase flexibility and widen the proposed channel, thereby affecting proton transfer dynamics.
We note that in the GsqNOR X-ray crystal structure, several water molecules were resolved in the hydrophilic channel. This may be due to the relatively ordered nature of them within the crystal environment, perhaps also influenced by the fact that the enzyme was inactive, due to zinc substitution in the active site. However, the region around the active site portion of the hydrophilic channel (i.e., the termini of said channel) exhibits a marked difference between the monomer and the dimer. The side-chain of Glu563, one of the conserved Glu residues near the binuclear active center, is facing a FeB ligand, Glu494, in the dimer. Conversely, in the monomer, the side-chain of Glu563 is flipped away from the active site (Fig. 3). Amongst a selection of NOR structures, only the X-ray crystal structure of NmqNOR with zinc bound (to Glu563, Glu498, and Glu494) displays a similar Glu563 conformation10 as we observed in the current cryoEM monomer NmqNOR structure (Supplementary Fig. 15). The Zn-bound structure likely represented an inactive state, based on in-solution activity measurements10. It is noted that the GsqNOR structure had a zinc ion bound in the binuclear active site (instead of iron), thereby not influencing the Glu563 (Glu581 in GsqNOR) conformation (Table 2). From our current study, we further strengthen the notion that Glu563 is essential for the enzymatic activity of qNOR (Table 1). The orientation and dynamic variation of the side chain could alter the efficiency of essential proton transfer, thereby leading to lowered NO reduction activity in the monomer.
Given that Glu494 is a possible terminal proton donor for the catalytic NO reduction46,47,48, increased structural dynamics of Glu494 upon dimerization would also contribute to the enhancement of the catalytic activity. The presence of the carboxylate group of Glu563 close to the carboxylate group of Glu494 might increase the flexibility of Glu494 via electrostatic repulsion, thereby heightening catalytic activity in the dimer. Thus, dimerization could manipulate the structure around the end of the active site side of the water channel for effective proton transfer in the catalytic NO reduction reaction. We explored whether Glu494 mutations could also affect the oligomerization behavior of qNOR. The Glu494Ala mutant displayed an almost equivalent ratio of dimer: monomer during gel filtration (Supplementary Fig. 16), whilst also losing catalytic activity. We also quantified the non-heme iron content to be ~0.3–0.4 Fe/qNOR in the monomer and dimer Glu494Ala, possibly leading to a loss of activity, consistent with the GsqNOR Glu512Ala mutation. The disruption of the interaction between Glu494 and Glu563 then impacts the dimerization property of qNOR, and thus catalytic activity.
The increased population of the monomer induced by mutations to Glu563 and Glu494, respectively, further supports the view that the region around Glu563 and the dimerization site are structurally linked.
It is noteworthy that similar mutation-induced monomer formation was also observed in our earlier work on AxqNOR10. The 4.5 Å cryoEM structure of the Glu494Ala mutant of AxqNOR (Glu498 in NmqNOR), a monomer, exhibited structural displacements of TM2 and TM10 compared with the wildtype dimer structures of qNOR, from both A.xylosoxidans and N.meningitidis. (Fig. 8a, b). The displacement of TM10 is broader toward the cytoplasmic end of the helix, where the dimerization site is located. When compared with our NmqNOR wildtype monomer structure, the mutant AxqNOR monomer TM10 shows a roughly similar conformation, indicating that mutation-induced monomerization also alters TM10 property (Fig. 8c). A possible explanation for this could be that Glu494 (Glu498 in NmqNOR) mutation to Ala eliminated the hydrogen bond with conserved Asn600 (Asn604 in NmqNOR, see Table 2 for sequence numbering differences amongst other qNORs). As this asparagine is located on TM11, which contributes to dimer formation, the loss of the interaction with the glutamate may cause downstream effects on dimer maintenance.
Superposition of the AxqNOR Glu494Ala monomer cryoEM structure (dark green ribbon, PDB ID: 6t6v) with A the AxqNOR wildtype cryoEM dimer structure (orange ribbon, PDB ID: 8bgw), B the NmqNOR wildtype cryoEM dimer structure (slate blue ribbon,this study) and, C the NmqNOR wildtype cryoEM monomer structure (gold ribbon, this study). TM2 and 10 are labeled in all the respective panels and shown as opaquely colored helices, whilst other helices are transparent.
Several studies support the critical role of qNOR in prolonging pathogen survival (e.g., N. meningitidis, N. gonorrhoeae49,50,51, and Staphylococcus aureus52) within host macrophages. Some of these pathogens have multi-drug resistant strains, which are listed as urgent and serious threats by the World Health Organization53. Therefore, targeting members of the HCO family in pathogens is seen as a critical junction to help combat antimicrobial resistance54,55. Work by Nishida and colleagues explored finding a conserved, allosteric inhibitory site in HCO enzymes, including bacterial qNOR, to improve the selectivity of inhibitors to reduce unwanted toxicity against host cells. A promising inhibitor, specific to bacterial HCO’s, had a proposed inhibitory mechanism whereby the substrate access channel is partially obstructed8. Our presented findings can contribute to broadening the search area (i.e., dimerization site and/or Glu563 interaction site) for novel inhibitor screening in qNOR, which is key in accelerating drug discovery efforts. A possible route could involve the use of deep learning structure-based drug design methodologies, where one can indicate a potential inhibitor binding site, in place of using pre-existing inhibitor-bound protein structures as a template for de novo compound design, as shown for the gastric proton pump56. Such a model, like DiffBP57, could generate molecules with high affinity and appropriate chemical properties that specifically target and disrupt the dimer interface of qNOR.
The structural determination of an enzyme’s various oligomeric states is essential to understanding key chemical processes. This has been critical for complex membrane-bound systems involved in oxygenic photosynthesis, like Photosystem I (PSI), where at least four oligomeric states have been demonstrated in cyanobacteria58,59. The cryoEM structure of the functional, monomeric cyanobacterial PSI was important to show the structural basis of red chlorophyll loss60, which is crucial for energy absorption in low light, explaining how trimeric PSI harvests light more efficiently under such conditions. It has become clear that such diversity in oligomeric states is necessary to adapt to varying ecological environments61. Hence, the structural studies of such systems are imperative in understanding the evolution of native enzyme function.
In the context of qNOR, oligomerization and protein-protein interactions in other members of the HCO superfamily, such as the oxygen-reducing enzyme CcO is a necessary comparison. Despite CcO being one of the most well-characterized HCO enzymes62,63, it was only in 2019 that the novel, monomeric CcO structure revealed factors that allow for higher oxidase activity compared to the dimer26. A cholate molecule, which facilitates dimerization (alongside an array of lipids), was found to disrupt a hydrogen-bonded network amongst several water molecules in one proton transfer channel, the K-pathway, at the boundary of the dimerization interface (Supplementary Fig. 17a). In the CcO monomer, this hydrogen-bonded network is intact by virtue of additional waters located at the K-pathway entry point (Supplementary Fig. 17b). It is accompanied by a conformational change in Glu62, which is located at the start of the K-pathway (Supplementary Fig. 17b). Besides this, no other structural differences were found between the two forms (Supplementary Fig. 17c, d). This proposed that a subtle modification in proton uptake at the K-pathway entrance causes the activity differences. Additionally, the binding of cholate (and other amphipathic compounds) at the K-pathway entry point may stabilize helix II and restrict conformational changes during the reduction cycle64. This may affect the rapid proton access to the active site, which has led to speculation that flexibility in the entry point is necessary for functionality of the proton uptake pathway after reduction44. This is in stark contrast to qNOR, where large-scale helical rearrangements and flexibility of TM10 (via loss of inter-protomer interactions) contributed to Glu563 conformation changes near the active site, reducing the specific activity. For the oxygen-reducing bacterial ubiquinol oxidase (cytochrome bo3), a recent cryo-EM study solved the dimeric enzyme structure for the first time65. This revealed the dimer interface to be composed of hydrophobic interactions mediated by residues like valine and leucine, from helices IV and II of different protomers (Supplementary Fig. 17e). There were only very minor structural changes in one loop region compared to the monomer structure (Supplementary Fig. 17f). It is unclear if the bo3 dimer is even functional or physiologically relevant; nevertheless, having structural information on both forms provides a framework for further investigation. In broader terms, the protein-protein interaction relationship to enzyme regulation is also important to consider. For CcO, a regulator protein, Higd1a (hypoxia inducible domain family member 1a) was shown (via spectroscopic methods) to positively regulate CcO activity (i.e., increase oxidase activity) by affecting the local structure of the active site heme a66. It is proposed that Higd1a incorporates with CcO in the transmembrane region67, yet, there is no direct 3D structural evidence of the co-complex to deepen the understanding of how protein-protein interaction in the membrane modulates activity. Our findings for qNOR, on the other hand, provide a clear structural basis for how enzyme regulation via protein-protein interaction (oligomerization) in the membrane can modulate an active site located far away from the dimerization site. Overall, this suggests the mechanism for the enhancement of the catalytic reaction by dimerization in qNOR is distinctive compared to other HCO enzymes that form oligomers. It is important to note, however, that all studies have been conducted on qNOR enzyme heterologously expressed in an E.coli expression system, not directly extracted from the source, i.e, pathogenic and/or denitrifying bacteria. Thus, investigations of the native oligomeric state of qNOR via cell biology methodologies/fluorescent probes will be necessary. Additionally, the influence of (potential) denitrification supercomplexes on the oligomeric form of qNOR is unknown. For CcO, monomers (more active) are constituents of respiratory supercomplexes, whilst dimer CcOs are not.
In summary, we solved cryoEM structures of NmqNOR in its constituent monomeric and dimeric states at comparably high resolutions to ascertain a mechanistic role for oligomerization. These data show that the dimerization-induced structural changes at the dimer interface led to the positional shift of the essential Glu563, located near the active site, for an effective NO reduction reaction. Our findings offer an alternate strategy in combating bacterial infections by inducing qNOR monomerization in vivo to help suppress NO detoxifying activity.
Materials & methods
Expression and purification of recombinant NmqNOR
The over-expression and purification of NmqNOR fused with BRIL (in this work, NmqNOR represents the one fused with apocytochrome b562; BRIL) were performed according to previous studies10, with minor modifications as follows. The detergent concentration in the buffer (50 mM HEPES, pH 8.0, 150 mM NaCl) for the first round of size exclusion chromatography (SEC) using a Superdex 200 26/60 (Cytiva) was adjusted from 0.05% (w/v) n-decyl-β-D-thiomaltoside (DTM; Anatrace) to 0.1% (w/v) DTM. Fractions corresponding to the dimeric and monomeric forms of NmqNOR from the first round of SEC were separated, pooled, and concentrated to ~10 mg/mL (as judged by the extinction coefficient ε410 = 213 mM−1 cm−1) using Amicon 50 K MWCO centrifugal concentrators (Merck). Dimeric NmqNOR was immediately used to make grids following SEC, whilst monomeric NmqNOR was snap-frozen in liquid nitrogen and stored at −80 °C. For the cryoEM analysis of the monomer sample, ~500 µL of 10 mg/mL of sample was thawed and briefly spun down in a centrifuge to remove any large aggregates, before the second round of SEC. The sample was run down a Superdex 200 10/300 Increase (GE Healthcare) column equilibrated in 50 mM HEPES pH 8.0, 150 mM NaCl and 0.1% or 0.05% (w/v) DTM. Fractions corresponding to the monomer peak were concentrated to ~3–15.5 mg/mL (~40–170 µM) before grid freezing. Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE) analysis of samples from SEC was performed using NuPAGE 4–12% Bis-Tris gels, using NuPAGE MES running buffer (Invitrogen). Blue-Native PAGE analysis was also run in parallel according to the manufacturer’s protocol, using NativePAGE 4–16% Bis-Tris gels (Invitrogen).
Glu563Ala, Glu563Leu, Leu569Phe/Ile244Phe and Glu494Ala variant plasmid DNA was ordered from GenScript (Hong Kong) and transformed into C41 (DE3) cells (Lucigen), before overexpression and purification in a similar fashion to wildtype. The one exception was that SEC was performed in 0.05% (v/v) n-dodecyl-β-D-maltoside (Dojindo), not DTM. Gel filtration of Glu563 variants was performed using a Superdex 200 26/60 column (like wildtype), whilst all other variants were run down a Superdex 200 10/300 Increase column.
UV–Visible spectra measurements
UV-Visible spectra of all samples were measured using a U-3900 spectrophotometer (Hitachi). During Nickel-Nitrilotriacetic acid (Ni-NTA) affinity chromatography, fractions were analyzed by UV–Visible spectroscopy to determine the A410/A280 ratio (the so-called Rz value). Fractions with an Rz value ≥ of 0.5 were collected and used for subsequent purification. After size exclusion chromatography, fractions with an Rz value ≥ of 0.7 were concentrated and subsequently used for cryoEM analysis. Difference spectra were obtained by collecting as-isolated spectra, followed by collecting reduced optical spectra of NmqNOR samples, after excess Na-Dithionite addition under N2 atmosphere. After baseline correction, oxidized spectra were subtracted from the reduced spectra and plotted.
NO reduction activity assay measurement
The NO reduction activity assays were carried out similarly to a previous study12. NO reduction rates were calculated from seven independent purifications (five technical replicates) for the dimer and from four independent purifications (five technical replicates) for the monomer samples. NO reduction rates were calculated using the Igor Pro software package (www.wavemetrics.com). For pH dependence activity assays, measurements were carried out in pH 6 (MES), 8 (Tris) or 9 (Tris) buffer, respectively.
Non-heme iron quantification
qNOR samples (1 ~ 10 μL) and 1 μL of 1 M HCL were mixed and incubated for 5 min at room temperature to completely denature the sample. Buffer containing nitroso-PSAP was added to the sample, immediately followed by the addition of reducing agent (to convert ferric iron to ferrous, the latter of which nitroso-PSAP chelates), ascorbate (final concentration of 1 mM, final reaction volume 100 μL). After 5 min of incubation at room temperature, the denatured protein was pelleted by centrifugation. The supernatant was then used for subsequent UV-visible absorption measurements. The estimation of free iron (Fe2+) concentration was performed using the molar extinction coefficient of 45,000 cm−1 M−1 at ~750 nm.
CryoEM sample preparation and data acquisition
NmqNORBRIL dimer
R1.2/1.3 Cu 300 mesh holey carbon grids (Quantifoil) were glow discharged using a JEC-3000FC Auto Fine Coater (JEOL Ltd.) at 7 Pa, 10 mA for 30 s. Grids were then plunge-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific) operating at 4 °C and 100% humidity. Three microliters of NmqNORBRIL (~500 µM) were applied to each grid before blotting for 6 s, using a blot force of 6. Grids were first screened using EPU (Thermo Fisher Scientific) on a Glacios TEM (Thermo Fisher Scientific) and subsequently transferred to a CRYO ARMTM 300 (JEM-Z300FSC; JEOL Ltd.) operating at 300 kV, equipped with an in-column Omega energy filter (20 eV slit width) and K3 direct electron detector (Gatan). SerialEM68,69 was used for automated data collection using 5 × 5 beam-image shift patterns (coma vs image shift calibration was performed before data acquisition). A total of 7525 movies were collected using a nominal magnification of ×60,000 at a pixel size of 0.752 Å/pixel. Each movie was divided into 50 frames using a total dose of 51.2 e−/Å2 (dose per frame of 1.02 e−/Å2), with the K3 detector operating in correlated double sampling mode.
NmqNORBRIL monomer
R1.2/1.3 Cu 300 mesh holey carbon grids (Quantifoil) were glow discharged using a JEC-3000FC Auto Fine Coater at 7 Pa, 10 mA for 10 s. Grids were then plunge-frozen in liquid ethane using a Vitrobot Mark IV operating at 8 °C and 100% humidity. Three microliters of NmqNOR-BRIL (initially ranging from 40–60 µM) were applied to each grid before blotting for 3 s, using a blot force of 6. Grid-making conditions were optimized over several screening sessions using the Glacios TEM and EPU (Thermo Fisher Scientific). Finally, a grid consisting of ~170 µM NmqNOR-BRIL (in 0.05% DTM, as opposed to 0.1% DTM), blotted for 3 s using a blot force of 3, exhibited increased particle density with suitable ice thickness for high-resolution data collection. The pre-screened grid was then loaded into a CRYO ARMTM 300, and a total of 8100 movies were collected using a nominal magnification of ×60,000 at a pixel size of 0.752 Å/pixel. Data were collected using 5 × 5 beam-image shift patterns in SerialEM (coma vs image shift calibration was performed before data acquisition). Each movie (3.66 s exposure time) was divided into 50 frames, using a total dose of 50 e−/Å2 (dose per frame of 1 e−/Å2), with the K3 detector operating in correlated double sampling mode.
For both datasets collected on the CRYO ARMTM 300, fully automated hole centering at medium magnification was performed with the yoneoLocr software70, integrated as a SerialEM macro.
Single particle image processing and 3D reconstruction
NmqNORBRIL dimer
Image processing was performed with RELION-4.0-beta-2-commit-ce2e9371, with the workflow shown in Supplementary Fig. 2. 7525 movies were motion-corrected using RELION’s implementation of MotionCor2 with CTF Estimation performed using CTFFIND-4.1.14. Automated particle picking was performed in crYOLO72 and co-ordinate star files were subsequently imported into RELION. 2,574,970 particles were extracted at a box size of 368 pixels and rescaled to 92 pixels (pixel size of 3.008 Å/px). Particles were split into 10 equal subsets of 257,497 particles. After performing 2D classification (VDAM algorithm, K = 150, mask diameter = 185 Å, Limit resolution E− step = 10 Å) on one subset to confirm the initial quality of the images (e.g., no severe orientation bias or degraded particles), 3D classification using a low pass filtered 3D reconstruction from a small dataset (not shown) on a Glacios TEM (Thermo Fisher Scientific) as a reference, was performed on 3 subsets (low pass filter 50 Å, K = 4, mask diameter = 190 Å, Limit resolution E− step = 8 Å). ~365,000 particles were selected and subjected to another round of 3D classification (similar to previous, but with Limit resolution E− step = 5 Å). 221,752 particles were then selected and subject to 3D auto-refinement, which reached Nyquist limit (6.14 Å). Particles were extracted to 2.23 Å/px, subject to 3D classification, of which 112,678 particles were 3D-auto-refined to the Nyquist limit (4.5 Å). This cycle of extraction, 3D classification and 3D refinement was continued until 1.09 Å/px pixel size, before CTF Refinement (individual rounds of anisotropic magnification, beamtilt & three-fold astigmatism (trefoil), per-particle defocus & per-micrograph astigmatism), 3D-refinement and postprocessing led to a 2.73 Å (map sharpening B-factor of −66 Å2) reconstruction. Subsets 4–6 were then processed similarly to above, and the best particles were joined with subsets 1–3 (totaling 326,211 particles). A 3D-refinement and postprocessing job led to a 2.49 Å (−67 Å2) reconstruction. CTF refinement (similar fashion as above) and polishing (trained with 10,000 particles and using all frames) were performed before 3D-refinement and postprocessing produced a 2.19 Å (−36 Å2) reconstruction.
At this point, a C2 symmetrized map (produced using relion_align_symmetry and relion_image_handler commands) was used as a reference in a 3D classification job (K = 3, mask diameter = 200 Å, Limit resolution E− step = −1). 292,309 particles were carried forward for a further round of 3D classification, with C2 symmetry imposed, leaving 269,127 particles. 3D-refinement and postprocessing lead to a 2.13 Å (−39 Å2) reconstruction. CTF refinement (similar fashion as above) and polishing (trained with 10,000 particles and using all movie frames) were performed before 3D-refinement and postprocessing produced a 2.06 Å (−32 Å2) reconstruction. Particles were extracted at full pixel size and subjected to two rounds of 3D refinement, CTF refinement (same as above, including four-fold astigmatism (tetrafoil)), and polishing, producing a 1.96 Å (−33 Å2) reconstruction (job170). Subsets 7-10 were then processed similarly as described above and the best particles were merged with those from sets 1-6. Two cycles of 3D classification, 3D refinement, CTF refinement, polishing, and 3D refinement led to a final resolution of 1.89 Å (−34 Å2) from 357,535 particles.
For the Henderson-Rosenthal plot analysis (Supplementary Fig. 6), particles from the final reconstruction were randomly split into subsets of 500, 1000, 2000, 5000, 10,000, 20,000, 50,000, 75,000, 100,000, and 200,000 particles, respectively. Each subset was then refined and sharpened using a soft mask encompassing the whole molecule. The squared value of the attained resolution was plotted against the natural logarithm of each particle subset. The B-factor was calculated by straight-line fitting.
Prior to running 3D Variability Analysis, particle stacks from the final 3D refinement were transferred from RELION to cryoSPARC using UCSF pyEM73. The final 3D volume and a custom-made mask to exclude the detergent micelle were imported via the cryoSPARC GUI. 3D Variability Analysis was then performed using the default parameters, aside from a filter resolution of 4 Å. Results were outputted in simple mode using 20 frames and volume down sampling to a 256 pixel box. Motion movies were created in UCSF Chimera v.17.1 using the volume series tool.
NmqNORBRIL monomer
Image processing was performed using cryoSPARC v4.1.174,75, with the workflow shown in Supplementary Fig. 7. All movies were subject to patch-based Motion Correction and patch-based CTF estimation. 104,877 particles were picked from 100 micrographs using the circular blob picker (minimum diameter 50 Å, maximum diameter 150 Å). These were used to generate 2D references for template-based picking against all micrographs, where a total of 4,640,985 particles were picked. 3,963,954 particles were extracted using a 360-pixel box rescaled to 90 pixels (pixel size of 3.008 Å/px). A subset of 150,000 particles were used for ab initio reconstruction (K = 3, maximum resolution 7 Å, initial resolution 9 Å). Heterogenous refinement was performed to completion and then ran once again up until one iteration had been completed to generate “decoy/bait” 3D volumes. The remaining particles were subject to two rounds of 2D classification (K = 200, maximum alignment resolution 10 Å, maximum resolution 8 Å, force max over poses and shifts = false, mask inner diameter 130 Å, batchsize per class 300, no. of EM iterations = 40 and no. of full iterations = 1) leaving 1,530,505 particles. These were then used for heterogeneous refinement (using the above created “good” and “decoy/bait” classes), and the top two populated classes (795,501 particles) were refined using non-uniform refinement. Particles were re-extracted in a 120 px box (pixel size 2.25 Å/px), before ab initio reconstruction (K = 3, maximum resolution 8 Å, initial resolution 10 Å). Heterogenous refinement was performed with the ab initio models, the previous non-uniform refinement map, and two “decoy” classes from a previous hetero refinement. Two classes (484,196 particles) were subsequently refined and re-extracted in a 180 px box (1.54 Å/px). Ab initio reconstruction and heterogenous refinement with 6 classes led to 2 classes (384,052 particles) which were separately refined (dynamic mask near and far distances of 8 and 20 Å, respectively). The two classes were identical, aside from the handedness which was corrected using the volume tools. Particles were re-extracted to 240 px (1.13 Å/px) and processed similarly as described. For the two best classes (total of 286,686), non-uniform refinement with the integrated fitting of beamtilt, trefoil, anisotropic magnification, and per-particle defocus led to a 2.30 Å reconstruction. After re-extracting in a 324 px box, hetero refinement and non-uniform refinement (similar to above, with fitting of spherical aberration and four-fold astigmatism (tetrafoil)) led to a 2.25 Å (map sharpening B-factor of −64 Å2) reconstruction from 274,346 particles. All resolutions were calculated according to the gold standard FSC = 0.143. 3D Variability Analysis was performed using the default parameters, aside from a filter resolution of 4 Å, using a mask excluding the detergent micelle. Results were outputted in simple mode using 20 frames and volume down sampling to a 256 pixel box. Motion movies were created in UCSF Chimera v.17.1 using the volume series tool.
Model building, refinement, and validation
NmqNORBRIL dimer
The starting model (PDB 6l3h) was initially rigidly docked into the final cryoEM map using the UCSF Chimera “Fit in Map” function. The structure was refined using Phenix v.1.15 real_space_refinement module, with the inclusion of restraint files for the heme molecules. Models were manually rebuilt within Coot and subject to further rounds of real space refinement, before validation in MolProbity. Statistics of image processing and refinement are summarized in Supplementary Table 1.
NmqNORBRIL monomer
The starting model (Chain A from the NmqNORBRIL dimer) was initially rigidly docked into the final cryoEM map using UCSF Chimera “Fit in Map” function and the structure was refined using Phenix v.1.20 real_space_refinement module76. To aid initial model building and main chain placement in the disordered regions (residues 245–265 and 567 to 590), the unfiltered cryoEM density map from non-uniform refinement was used. Finer amino acid side chain placement and rotamer fitting were performed using the sharpened map where necessary. Refinements were done using the unfiltered map, with the inclusion of restraint files for the heme molecules. Models were manually rebuilt within Coot77 and subject to further rounds of real-space refinement, before validation in MolProbity78. Statistics of image processing and refinement are summarized in Supplementary Table 1. CAVER79 analysis (via a PyMOL plugin) was performed on both dimer and monomer models using a probe radius of 0.9 Å and a shell radius of 3 Å. Ca2+, Fe3+ and water molecules were removed before analysis. Analysis was conducted from the heme b3 iron as a starting point.
CryoEM density map depictions and protein model figures were prepared using UCSF ChimeraX v1.4, v1.680 and PyMOL (Schrödinger). UCSF Chimera v1.17.3 was used to create the 3DVA movies81. Consurf-DB server82 and ESpript 3.083 were used to generate sequence conservation and alignment data in Supplementary Fig. 11. Final figure collation was performed within Inkscape 1.1.2 (b8e25be8, 2022-02-05, www.inkscape.org).
Statistics and reproducibility
Activity assays were performed with several technical replicates and biological replicates (separate purifications). Data is presented with standard deviations in Table 1. CryoEM data were taken from one grid, after screening 2–3 grids of the same condition.
Data availability
Accession codes for the NmqNOR dimer structure are as follows; PDB: 8zgp, EMDB: EMD-60086. Accession codes for the NmqNOR monomer structure are as follows; PDB: 8zgo, EMDB: EMD-60085. Uncropped images of the gels in Fig. 7 and Supplementary Fig. 1 can be found in Supplementary Fig. 18. All other data are available from the corresponding author on reasonable request.
References
Marreiros, B. C. et al. Exploring membrane respiratory chains. Biochim. Biophys. Acta Bioenerg. 1857, 1039–1067 (2016).
Zumft, W. G. Cell biology and molecular basis of denitrification. Microbiol. Mol. Biol. Rev. 61, 533–616 (1997).
Kakishima, K., Shiratsuchi, A., Taoka, A., Nakanishi, Y. & Fukumori, Y. Participation of nitric oxide reductase in survival of Pseudomonas aeruginosa in LPS-activated macrophages. Biochem. Biophys. Res. Commun. 355, 587–591 (2007).
Yoon, S. S. et al. Pseudomonas aeruginosa anaerobic respiration in biofilms. Dev. Cell 3, 593–603 (2002).
Stevanin, T. M., Moir, J. W. B. & Read, R. C. Nitric oxide detoxification systems enhance survival of Neisseria meningitidis in human macrophages and in nasopharyngeal mucosa. Infect. Immun. 73, 3322–3329 (2005).
Pizza, M. & Rappuoli, R. Neisseria meningitidis: pathogenesis and immunity. Curr. Opin. Microbiol. 23, 68–72 (2015).
Favazzo, L. J., Gill, A. L., Farnsworth, C. W., Mooney, R. A. & Gill, S. R. The response of nor and nos contributes to Staphylococcus aureus virulence and metabolism. J. Bacteriol. 201, e00107–19 (2019).
Nishida, Y. et al. Identifying antibiotics based on structural differences in the conserved allostery from mitochondrial heme-copper oxidases. Nat. Commun. https://doi.org/10.1038/s41467-022-34771-y (2022).
Gopalasingam, C. C. et al. Dimeric structures of quinol-dependent nitric oxide reductases (qNORs) revealed by cryo–electron microscopy. Sci. Adv. 5, eaax1803 (2019).
Jamali, M. A. M. et al. The active form of quinol-dependent nitric oxide reductase from Neisseria meningitidis is a dimer. IUCrJ 7, 1–12 (2020).
Matsumoto, Y. et al. Crystal structure of quinol-dependent nitric oxide reductase from Geobacillus stearothermophilus. Nat. Struct. Mol. Biol. 19, 238–245 (2012).
Gonska, N. et al. Characterization of the quinol-dependent nitric oxide reductase from the pathogen Neisseria meningitidis, an electrogenic enzyme. Sci. Rep. 8, 3637 (2018).
Hino, T. et al. Structural basis of biological N2O generation by bacterial nitric oxide reductase. Science https://doi.org/10.1126/science.1195591 (2010).
Chorev, D. S. et al. Protein assemblies ejected directly from native membranes yield complexes for mass spectrometry. Science 362, 829–834 (2018).
Vercellino, I. & Sazanov, L. Structure and assembly of mammalian mitochondrial supercomplex CIII2CIV. Nature 598, 364–367(2021).
Maldonado, M., Guo, F., Letts, J. A. & Carter, A. P. Atomic structures of respiratory complex III2, complex IV, and supercomplex III2-IV from vascular plants. Elife 10, 1–34 (2021).
Letts, J. A., Fiedorczuk, K., Degliesposti, G., Skehel, M. & Sazanov, L. A. Structures of respiratory supercomplex I+III2 reveal functional and conformational crosstalk. Mol. Cell 75, 1131–1146.e6 (2019).
Hartley, A. M. et al. Structure of yeast cytochrome c oxidase in a supercomplex with cytochrome bc 1. Nat. Struct. Mol. Biol. 26, 78–83 (2019).
Mühleip, A. et al. Structural basis of mitochondrial membrane bending by the I–II–III2–IV2 supercomplex. Nature 615, 934–938 (2023).
Davies, K. M., Blum, T. B. & Kühlbrandt, W. Conserved in situ arrangement of complex I and III 2 in mitochondrial respiratory chain supercomplexes of mammals, yeast, and plants. Proc. Natl. Acad. Sci. USA 115, 3024–3029 (2018).
Zheng, W., Chai, P., Zhu, J. & Zhang, K. High-resolution in situ structures of mammalian respiratory supercomplexes. Nature https://doi.org/10.1038/s41586-024-07488-9 (2024).
Guo, R., Zong, S., Wu, M., Gu, J. & Yang, M. Architecture of human mitochondrial respiratory megacomplex I2III2IV2. Cell 170, 1247–1257.e12 (2017).
Rathore, S. et al. Cryo-EM structure of the yeast respiratory supercomplex. Nat. Struct. Mol. Biol. 26, 50–57 (2019).
Moe, A., Dimogkioka, A.-R., Rapaport, D., Öjemyr, L. N. & Brzezinski, P. Structure and function of the S. pombe III–IV–cyt c supercomplex. Proc. Natl. Acad. Sci. USA 120, e2307697120 (2023).
Tsukihara, T. et al. The whole structure of the 13-subunit oxidized cytochrome c oxidase at 2.8 A. Science 272, 1136–1144 (1996).
Shinzawa-Itoh, K. et al. Monomeric structure of an active form of bovine cytochrome c oxidase. Proc. Natl. Acad. Sci. USA 116, 19945–19951 (2019).
Hamaguchi, T. et al. A new cryo-EM system for single particle analysis. J. Struct. Biol. 207, 40–48 (2019).
Fislage, M., Shkumatov, A. V., Stroobants, A. & Efremov, R. G. Assessing the JEOL CRYO ARM 300 for high-throughput automated single-particle cryo-EM in a multiuser environment. IUCrJ 7, 707–718 (2020).
Flynn, A. J., Antonyuk, S. V., Eady, R. R., Muench, S. P. & Hasnain, S. S. A 2.2 Å cryoEM structure of a quinol-dependent NO Reductase shows close similarity to respiratory oxidases. Nat. Commun. 14, 3416 (2023).
Blomberg, M. R. A. & Ädelroth, P. The mechanism for oxygen reduction in cytochrome c dependent nitric oxide reductase (cNOR) as obtained from a combination of theoretical and experimental results. Biochim. Biophys. Acta Bioenerg. 1858, 884–894 (2017).
Moenne-Loccoz, P. & de Vries, S. Structural characterization of the catalytic high-spin heme b of nitric oxide reductase: a resonance Raman study†. Methods 7863, 5147–5152 (1998).
Kim, E. et al. Superoxo, μ-peroxo, and μ-oxo complexes from heme/O 2 and heme-Cu/O 2 reactivity: Copper ligand influences in cytochrome c oxidase models. Proc. Natl. Acad. Sci. USA 100, 3623–3628 (2003).
Hino, T., Nagano, S., Sugimoto, H., Tosha, T. & Shiro, Y. Molecular structure and function of bacterial nitric oxide reductase. Biochim. Biophys. Acta Bioenerg. 1817, 680–687 (2012).
Punjani, A. & Fleet, D. J. 3D variability analysis: resolving continuous flexibility and discrete heterogeneity from single particle cryo-EM. J. Struct. Biol. 213, 107702 (2021).
Dong, M. et al. Structure and dynamics of the active Gs-coupled human secretin receptor. Nat. Commun. 11, 4137 (2020).
Liang, Y.-L. et al. Structure and dynamics of adrenomedullin receptors AM 1 and AM 2 reveal key mechanisms in the control of receptor phenotype by receptor activity-modifying proteins. ACS Pharmacol. Transl. Sci. 3, 263–284 (2020).
Papasergi-Scott, M. M. et al. Time-resolved cryo-EM of G-protein activation by a GPCR. Nature 629, 1182–1191 (2024).
Tôei, K., Motomizu, S. & Korenaga, T. Nitrosophenol and nitrosonaphthol derivatives as reagents for the spectrophotometric determination of iron and determination of micro-amounts in waters with 2-nitroso-5-dimethylaminophenol. Analyst 100, 629 (1975).
Michel, L. V. et al. Heme attachment motif mobility tunes cytochrome c redox potential †. Biochemistry 46, 11753–11760 (2007).
Bhagi-Damodaran, A. et al. Heme redox potentials hold the key to reactivity differences between nitric oxide reductase and heme-copper oxidase. Proc. Natl. Acad. Sci. USA 115, 6195–6200 (2018).
Cordas, C. M., Duarte, A. G., Moura, J. J. G. & Moura, I. Electrochemical behaviour of bacterial nitric oxide reductase—Evidence of low redox potential non-heme FeB gives new perspectives on the catalytic mechanism. Biochim. Biophys. Acta Bioenerg. 1827, 233–238 (2013).
Terasaka, E. et al. Dynamics of nitric oxide controlled by protein complex in bacterial system. Proc. Natl. Acad. Sci. USA https://doi.org/10.1073/pnas.1621301114, 201621301 (2017).
Malström, B. G. Cytochrome c oxidase Structure and catalytic activity. Biochim. Biophys. Acta Rev. Bioenerg. 549, 281–303 (1979).
Busenlehner, L. S., Salomonsson, L., Brzezinski, P. & Armstrong, R. N. Mapping protein dynamics in catalytic intermediates of the redox-driven proton pump cytochrome c oxidase. Proc. Natl. Acad. Sci. USA 103, 15398–15403 (2006).
Flock, U., Watmough, N. J. & Ädelroth, P. Electron/proton coupling in bacterial nitric oxide reductase during reduction of oxygen. Biochemistry 44, 10711–10719 (2005).
Salomonsson, L. et al. Proton transfer in the quinol-dependent nitric oxide reductase from Geobacillus stearothermophilus during reduction of oxygen. Biochim. Biophys. Acta Bioenerg. 1817, 1914–1920 (2012).
Thorndycroft, F. H., Butland, G., Richardson, D. J. & Watmough, N. J. A new assay for nitric oxide reductase reveals two conserved glutamate residues form the entrance to a proton-conducting channel in the bacterial enzyme. Biochem. J. 401, 111–119 (2007).
Blomberg, M. R. A. & Ädelroth, P. Mechanisms for enzymatic reduction of nitric oxide to nitrous oxide - A comparison between nitric oxide reductase and cytochrome c oxidase. Biochim. Biophys. Acta Bioenerg. 1859, 1223–1234 (2018).
Barth, K. R., Isabella, V. M. & Clark, V. L. Biochemical and genomic analysis of the denitrification pathway within the genus Neisseria. Microbiology 155, 4093–4103 (2009).
Falsetta, M. L. et al. Transcriptional profiling identifies the metabolic phenotype of gonococcal biofilms. Infect. Immun. 77, 3522–3532 (2009).
Cardinale, J. A. & Clark, V. L. Determinants of nitric oxide steady-state levels during anaerobic respiration by Neisseria gonorrhoeae. Mol. Microbiol. 58, 177–188 (2005).
Lewis, A. M. et al. Examination of the Staphylococcus aureus nitric oxide reductase (saNOR) reveals its contribution to modulating intracellular NO levels and cellular respiration. Mol. Microbiol. 96, 651–669 (2015).
Unemo, M. et al. WHO global antimicrobial resistance surveillance for Neisseria gonorrhoeae 2017–18: a retrospective observational study. The Lancet Microbe 2, e627–e636 (2021).
Bald, D., Villellas, C., Lu, P. & Koul, A. Targeting energy metabolism in Mycobacterium tuberculosis, a new paradigm in antimycobacterial drug discovery. MBio 8, e00272–17 (2017).
Cook, G. M., Hards, K., Vilchèze, C., Hartman, T. & Berney, M. Energetics of respiration and oxidative phosphorylation in Mycobacteria. Microbiol. Spectr. 2, 10.1128/microbiolspec.MGM2-0015-2013 (2014).
Abe, K. et al. Deep learning driven de novo drug design based on gastric proton pump structures. Commun. Biol. 6, 956 (2023).
Lin, H. et al. Correction: DiffBP: generative diffusion of 3D molecules for target protein binding. Chem. Sci. 16, 6523 (2025).
Hamaguchi, T. et al. Structure of the far-red light utilizing photosystem I of Acaryochloris marina. Nat. Commun. 12, 2333 (2021).
Li, M. et al. Physiological and evolutionary implications of tetrameric photosystem I in cyanobacteria. Nat. Plants 5, 1309–1319 (2019).
Çoruh, O. et al. Cryo-EM structure of a functional monomeric Photosystem I from Thermosynechococcus elongatus reveals red chlorophyll cluster. Commun. Biol. 4, 304 (2021).
Huokko, T. et al. Probing the biogenesis pathway and dynamics of thylakoid membranes. Nat. Commun. 12, 3475 (2021).
Tsukihara, T. et al. Structures of metal sites of oxidized bovine heart cytochrome c oxidase at 2.8 A. Science 269, 1069–1074 (1995).
Kolbe, F. et al. Cryo-EM structures of intermediates suggest an alternative catalytic reaction cycle for cytochrome c oxidase. Nat. Commun. 12, 6903 (2021).
Hiser, C., Buhrow, L., Liu, J., Kuhn, L. & Ferguson-Miller, S. A conserved amphipathic ligand binding region influences K-path-dependent activity of cytochrome c oxidase. Biochemistry 52, 1385–1396 (2013).
Guo, Y. et al. Monomer and dimer structures of cytochrome bo 3 ubiquinol oxidase from Escherichia coli. Protein Sci. 32, 0–2 (2023).
Hayashi, T. et al. Higd1a is a positive regulator of cytochrome c oxidase. Proc. Natl. Acad. Sci. USA 112, 1553–1558 (2015).
Timón-Gómez, A., Bartley-Dier, E. L., Fontanesi, F. & Barrientos, A. HIGD-driven regulation of cytochrome c oxidase biogenesis and function. Cells 9, 2620 (2020).
Mastronarde, D. N. Automated electron microscope tomography using robust prediction of specimen movements. J. Struct. Biol. 152, 36–51 (2005).
Schorb, M., Haberbosch, I., Hagen, W. J. H., Schwab, Y. & Mastronarde, D. N. Software tools for automated transmission electron microscopy. Nat. Methods 16, 471–477 (2019).
Yonekura, K., Maki-Yonekura, S., Naitow, H., Hamaguchi, T. & Takaba, K. Machine learning-based real-time object locator/evaluator for cryo-EM data collection. Commun. Biol. 4, 1–8 (2021).
Zivanov, J., Nakane, T. & Scheres, S. H. W. Estimation of high-order aberrations and anisotropic magnification from cryo-EM data sets in RELION-3.1. IUCrJ 7, 253–267 (2020).
Wagner, T. et al. SPHIRE-crYOLO is a fast and accurate fully automated particle picker for cryo-EM. Commun. Biol. 2, 1–13 (2019).
Asarnow, D., Palovcak, E. & Cheng, Y. asarnow/pyem: UCSF pyem v0.5. Preprint at https://doi.org/10.5281/zenodo.3576630 (2019).
Punjani, A., Zhang, H. & Fleet, D. J. Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction. Nat. Methods 17, 1214–1221 (2020).
Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. CryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296 (2017).
Afonine, P. V. et al. Real-space refinement in PHENIX for cryo-EM and crystallography. Acta Crystallogr. Sect. D Struct. Biol. 74, 531–544 (2018).
Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. Sect. D Biol. Crystallogr. 60, 2126–2132 (2004).
Williams, C. J. et al. MolProbity: more and better reference data for improved all-atom structure validation. Protein Sci. 27, 293–315 (2018).
Petrek, M. et al. CAVER: a new tool to explore routes from protein clefts, pockets and cavities. BMC Bioinform. 7, 316 (2006).
Pettersen, E. F. et al. UCSF ChimeraX: structure visualization for researchers, educators, and developers. Protein Sci. 30, 70–82 (2021).
Pettersen, E. F. et al. UCSF Chimera–a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612 (2004).
Ben Chorin, A. et al. ConSurf-DB: an accessible repository for the evolutionary conservation patterns of the majority of PDB proteins. Protein Sci. 29, 258–267 (2020).
Gouet, P., Robert, X. & Courcelle, E. ESPript/ENDscript: extracting and rendering sequence and 3D information from atomic structures of proteins. Nucleic Acids Res. 31, 3320–3323 (2003).
Acknowledgements
We thank Ms. Sachiko Hashimoto for culturing of all NmqNOR samples and isolating membrane fractions for purification. We also thank JEOL Ltd. for the provision of SerialEM macros for CRYO ARMTM 300 operation. This work was supported by the Japan Society for Promotion of Science KAKENHI grants JP19H05761 (YS), JP20K22633 (CCG) and JP21H02064 (T.T.). CryoEM data collection on the Glacios and CRYO ARMTM 300 at RIKEN SPring-8 was facilitated by the Platform Project for Supporting Drug Discovery and Life Science Research (Basis for Supporting Innovative Drug Discovery and Life Science Research (BINDS)) from the Japan Agency for Medical Development (AMED) under grant number JP21am0101070 and JP24ama121001 (C.C.G., C.G., H.S., and M.Y.).
Author information
Authors and Affiliations
Contributions
H.E., M.S., C.C.G., and K.F. purified and characterized NmqNORBRIL samples, whilst C.C.G., M.S., and T.T. purified and characterized site-directed variants of NmqNOR. C.C.G. and H.E. froze grids for cryoEM analysis. C.G., M.Y., and H.S. provided overall cryoEM technical support. C.C.G. performed grid screening for dimer and monomer samples and data collection for the NmqNOR dimer data set. H.S. performed data collection for the NmqNOR monomer data set. C.C.G. and H.E. performed image processing. C.C.G. and K.M. built, refined and validated the atomic models. C.C.G., Y.S., K.M., and T.T. conceived the study. C.C.G. prepared the figures. C.C.G., Y.S., and T.T. prepared the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Communications Biology thanks Haichun Gao, Stephen Muench and the other anonymous reviewer(s) for their contribution to the peer review of this work. Primary handling editors: Xiaoling Xu and Tobias Goris. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Gopalasingam, C.C., Egami, H., Shigematsu, H. et al. Structural basis of Neisseria meningitidis quinol dependent nitric oxide reductase activation by dimerization. Commun Biol 9, 433 (2026). https://doi.org/10.1038/s42003-026-09754-0
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s42003-026-09754-0
