Recent structural insights into the function of copper nitrite reductases

Copper nitrite reductases (CuNiR) carry out the first committed step of the denitrification pathway of the global nitrogen cycle, the reduction of nitrite (NO2 ) to nitric oxide (NO). As such, they are of major agronomic and environmental importance. CuNiRs occur primarily in denitrifying soil bacteria which carry out the overall reduction of nitrate to dinitrogen. In this article, we review the insights gained into copper nitrite reductase (CuNiR) function from three dimensional structures. We particularly focus on developments over the last decade, including insights from serial femtosecond crystallography using X-ray free electron lasers (XFELs) and from the recently discovered 3-domain CuNiRs.


Introduction
Copper nitrite reductases (CuNiRs) are key enzymes in the nitrogen cycle, occurring in a wide range of denitrifying bacteria and fungi. 1 The denitrification pathway in these organisms couples ATP synthesis with the reduction of nitrate (NO 3

À
) or nitrite (NO 2 À ) to dinitrogen (N 2 ), via nitric oxide (NO) and nitrous oxide (N 2 O) intermediates. 2,3This pathway is critical in the return of fixed nitrogen to the atmosphere and in controlling the level of biologically available nitrogen in the soil.The extensive use of nitrogen containing fertilisers from the Haber-Bosch process can cause major environmental problems if the denitrification pathway becomes overloaded. 2 Moreover, the N 2 O intermediate in the denitrification pathway is a potent greenhouse gas. 4 The roles of different metalloenzymes in denitrification and their environmental and agricultural impacts have been recently reviewed. 5he first committed step (i.e. the first effectively irreversible reaction in the pathway that commits the organism to generate the final product) in denitrification is the reduction of nitrite to NO (eqn (1)) and this is carried out either by Fe-containing cytochrome cd1 nitrite reductases, penta-or octa-haem nitrite reductases or by CuNiRs, encoded by the nirk gene.

NO 2
À + e À + 2H + -NO + H 2 O (1) Many detailed biochemical, spectroscopic, kinetic and structural studies have investigated the catalytic mechanisms of the CuNiRs and we describe only the key points in this review, focusing on structural insights.We describe the three-dimensional structures of CuNiRs and how these have led to profound insights into enzyme function and catalytic mechanism.We particularly focus on developments in the last decade including the discovery of extended CuNiRs, the application of serial femtosecond crystallography (SFX) and approaches to characterise the enzyme mechanism within protein crystals.

Overall fold and oligomeric state of CuNiRs
The majority of CuNiRs are trimeric, with each B37 kDa monomer comprising two cupredoxin (b-sandwich) domains, Fig. 1A and B. 6 Each of these domains has a similar architecture to that of monomeric cupredoxin proteins, such as azurin, pseudoazurin and rusticyanin, that act as electron donor partner proteins to CuNiRs. 7The C-terminal region of each cupredoxinlike monomer forms an extended loop that wraps around an adjacent monomer contributing to the stability of the trimer. 6hese well-characterised two-domain CuNiRs are assigned as classes I and II, depending on the colour arising from their type 1 Cu (T1Cu) centre, with class I being blue and class II being green.Class III has an additional T1Cu domain with a hexameric oligomer and is greenish-blue due to the presence of two Cu centres, one each of class I and II. 8This class III CuNiR is an example of a recently discovered group of extended (threedomain) CuNiRs, 9 that exist with additional fused N-or C-terminal domains containing a T1Cu or cytochrome haem redox centre, respectively (Fig. 1). 10 Regardless of such variations, the core structure of T1Cu and type 2 Cu (T2Cu) sites in the trimeric core is largely conserved.We will first discuss the structure and function of the trimeric CuNiRs before highlighting variations in the more recently characterised extended CuNiRs.

The copper centres of CuNiRs in the resting state
All CuNiRs characterised to date contain an electron-providing T1Cu centre and a catalytic T2Cu centre.At the T1Cu site, the Cu is coordinated by two His, one Cys and an axial Met forming a tetrahedral geometry, Fig. 2. 6 The structural changes within T1Cu sites between the Cu(II) and Cu(I) states are small (consistent with an entactic/rack state), and close to the level of precision at the typical resolutions of crystal structures.
Atomic resolution structures provide more accuracy for bond length parameters but usually at the expense of higher X-ray dose which can readily reduce the T1Cu to the Cu(I) state.EXAFS studies suggest an increase in the Cu-Cys bond length on reduction of the blue copper protein azurin of some 0.07 Å. 11 As the main chromophore of CuNiR, the T1Cu site is responsible for the colour of the protein in the cupric state.In general, T1Cu sites in copper proteins can vary widely in colour, including blue, green, red and purple. 12,13The relationship between T1Cu colour and geometry has been systematically studied.CuNiRs from different species appear blue, green or 'bluish-green' despite possessing an identical set of ligand residues.Insights into the relationship between T1Cu structure, colour and redox potential have been obtained by comparison of structures of blue and green CuNiRs and by mutagenesis of first and second coordination shell residues.Blue CuNiRs typically exhibit a long/weak Cu-Met bond B2.9 Å while in green CuNiRs this bond is much shorter.
Work from Solomon and co-workers revealed that in green CuNiRs, two spectral forms were present, with the ratio between these being strongly temperature dependent. 14At low temperatures, a green form with a strong Met-Cu interaction is dominant, while at higher temperatures, a blue form becomes more populated, with a weaker Cu-Met interaction.The T1Cu geometry, colour and redox potential are profoundly affected by mutation of ligand residues with consequent effects on enzyme activity.
The catalytic T2Cu is separated from the T1Cu by B12.6 Å and the centres are connected by a Cys-His bridge (for rapid  Image generated from the structure of resting state AcNiR at 0.9 Å resolution (PDB 2BW4). 17lectron transfer) comprised of the T1Cu ligand Cys and a T2Cu ligating His. 6The T2Cu has (His) 3 -H 2 O ligation in the resting state with a tetrahedral geometry, Fig. 2 and 3A.One of the three His ligands is provided from an adjacent monomer of the CuNiR trimer.Upon reduction (in the absence of substrate), the T2Cu loses the coordinated H 2 O resulting in Cu-(His) 3 ligation. 15The T2Cu centre possesses only a weak UV-visible absorption spectrum that is obscured by the much stronger absorbance of the T1Cu centre. 16The T2Cu ligand binding pocket contains several conserved residues with proposed functional roles related to substrate access, specificity, activity, electron transfer & proton transfer. 16The Cu-coordinated water molecule is H-bonded to an aspartic acid residue (Asp CAT ) and, via a water bridge, to a conserved His (His CAT ). 6Asp CAT has been proposed to play roles in substrate guidance/stabilisation and to act as a proton donor, 17 while His CAT acts as a second proton donor.Crystal structures show that Asp CAT can adopt two different conformations, named 'gatekeeper' -pointed away from T2Cu and 'proximal' -oriented towards T2Cu and H-bonding to the bound water molecule in the resting state. 18utation of Asp CAT to asparagine (Asn) in AxNiR eliminated catalytic activity providing strong evidence for its role in proton transfer. 19The properties of Asp CAT and His CAT may provide the rationale for the observed pH dependence of CuNiR activity.Solomon and coworkers have reported a pK a of 6.4 for Asp CAT (in AxNiR) when NO 2 À is bound. 20Finally, an isoleucine (Ile CAT ) residue that lies above the T2Cu binding site has been shown by structural studies 21 to profoundly influence substrate orientation and consequently the catalytic activity.

Enzymatic mechanism of nitrite reduction: structural insights
The consensus mechanism for two domain CuNiRs begins with the resting state enzyme, with a water bound to the T2Cu centre, Fig. 3A.On the basis of X-ray crystallography and EXAFS studies on native and site-selected AxNiR mutants, supported by kinetics measurements, a catalytic mechanism was initially proposed in which NO 2 À coordinates to the T2Cu(II), displacing the water molecule in the native enzyme, Fig. 3B.In the absence of nitrite at the T2Cu there is a limited potential for electron transfer (ET) from the T1Cu, but following nitrite binding ET from T1Cu(I) to the T2Cu site occurs, 15,22,23 which reduces the T2Cu.Two protons required for the catalytic reaction are thought to be provided by residues His254 (His CAT ) and Asp97 (Asp CAT ) and the proton-coupled electron-transfer (PCET) from T1Cu to T2Cu, involving the His CAT and Asp CAT residues, is thought to be the rate-determining step. 24,25In this PCET process, the two protons are provided to the bound NO 2 À along with an ET from the reduced T2Cu, to yield the NO product, Fig. 3C.This product then dissociates from the T2Cu, allowing water to rebind to reform the oxidised resting T2Cu state, Fig. 3A.Structural and kinetic studies on a number of other CuNiRs gave apparently conflicting evidence on whether the intramolecular ET occurred before or after NO 2 À binding [e.g.ref. 15, 26 and 27].These conflicts have been mostly reconciled by a kinetic scheme for catalytic turnover in which two alternative mechanistic pathways are possible, varying in the sequence of ET and binding of nitrite, depending on the pH and nitrite concentration. 28][31][32][33][34] Key features relevant to this mechanism that are informed by structures include the modes of substrate and product binding, the likely electron and proton transfer pathways and steps and the conserved water structure within the active site pocket.

Substrate binding to the T2Cu centre and specificity
The first structure of a nitrite-bound CuNiR complex adduct was determined in 1991. 6 Nitrite has typically been observed to bind either in an asymmetric bidentate binding geometry with its O1 and O2 atoms ligating to T2Cu(II) or in a 'side-on' geometry with similar distances from T2Cu to the three atoms of nitrite.One of the highest resolution structures of a CuNiRnitrite complex is that of AcNiR, to 1.10 Å resolution, 18 where nitrite was observed with bidentate binding to T2Cu and dual conformations of Asp98 CAT were present, with the proposal that the gatekeeper conformation of Asp98 (Asp CAT ) may be important for guiding NO 2 À to its binding site at the T2Cu.The proximal conformation, on the contrary, is significant for proton transfer, bond-cleavage and formation and release of the product, NO.The ] resolutions, respectively, with nitrite bound to reduced T2Cu, revealing few structural changes relative to the oxidised form. 41n the different structures of nitrite complexes, bidentate nitrite has been modelled in both 'top-hat' and 'side-on' positions.XFEL SFX 54 data have revealed that the 'top-hat' mode is the binding geometry initially present in the T2Cu(II) form of the enzyme and that the side-on may be the result of T2Cu reduction.This highlights that great care must be taken in crystallographic studies of nitrite-bound CuNiRs due to the propensity for X-ray generated photoelectrons to initiate catalysis within the crystal, leading to the conversion of substrate to NO at the T2Cu site.Structures determined from CuNiR nitrite complexes using high X-ray doses in many cases reveal a significant population of the T2Cu-NO product. 27Most recently this phenomenon has been exploited to drive in crystallo reactions to generate 'structural movies' of enzyme catalysis in green AcNiR 55,57 and thermostable Geobacillus thermodenitrificans (GtNiR). 582Cu site mutations in AfNiR significantly altered the binding mode of nitrite 21 and these were related to changes in enzyme activity.Mutation of this Ile CAT to Val or Leu led to asymmetric bidentate NO 2 À binding while mutation to Met, Thr, Gly or Ala resulted in monodentate coordination, allowing a water molecule to also bind to T2Cu in a position close to that usually occupied by the O2 atom of nitrite.Monodentate nitrite binding was associated with greatly reduced catalytic activity, pointing to a key role for Ile CAT in promoting a nitrite binding geometry conducive during catalysis.Crystal structures were determined of the oxidised and chemically-reduced Asp CAT to Asn mutant in Af NiR. 35In these structures, nitrite was bound in a side-on manner with an increased level of disorder, ascribed to a shift of Asn relative to Asp producing an unfavourable geometry for H-bonding with the bound protonated nitrite.However, this disorder could simply represent a mixture of top-hat and side-on nitrite molecules within the crystal.The His CAT to Asn mutant of Af NiR showed no displacement of the water molecule and the nitrite is bound to the T2Cu in a monodentate manner via an O atom.Nitrite is oriented away from Ile257 and towards an Ala residue and the pocket H-bonding network is drastically altered in the H255N mutant, indicating an essential role for nitrite binding and function. 35 somewhat different binding mode for nitrite was observed in a structure of the C135A mutant (which eliminates T1Cu binding) of the thermophile GtNiR, 52 a close homologue of GkNiR.The 1.90 Å structure showed nitrite bound in a monodentate manner via its O2 atom with this atom being positioned to form a H-bond to Asp CAT .His CAT adopted a rotated conformation likely due to the formation of a H-bond to a Gln residue rather than the Thr which interacts with His CAT in other CuNiRs.Interestingly, in this enzyme, the gatekeeper conformation of Asp CAT and the 'vertical' conformer of His CAT are sterically restricted by the unusual position of a Phe residue near to the catalytic pocket which provides additional rigidity to the active site in this enzyme from a thermophile, Fig. 4A.Intriguingly, when the structure of the nitrite complex of GtNiR was determined at 320 K, the observed substrate binding mode was side-on, similar to that observed in other CuNiRs, 53 a result ascribed to increased dynamic flexibility of the pocket at the higher temperature.

Substrate and copper access to the T2Cu site
The T2Cu site, as mentioned previously, makes up the active substrate-binding site and is coordinated by three His and one H 2 O/OH À ligand. 38In the resting state nitrite has access to the T2Cu centre via a channel some B12 Å deep from the protein surface and displaces the H 2 O ligand to initiate catalysis.Intriguingly, mutation of the surface-exposed Phe306 to a Cys This journal is © The Royal Society of Chemistry 2017 residue some 12 Å away from T2Cu in AxNiR resulted in a change to the second coordination sphere of the Cu with consequent effects on the rate-limiting step of the enzyme reaction. 59In this F306C mutant, the substrate access channel is more open and contains an additional water bound to the T2Cu-coordinating water molecule.The specific activity was remarkably increased four-fold while the apparent nitrite affinity and the rate of intramolecular ET were lower than in native AxNiR.In this mutant, intramolecular ET became the rate-limiting step.
Binding of small molecule ligands to the T2Cu centre Native CuNiR can bind small molecules including acetate, nitrate, formate, azide and N 2 O, of which complex structures have all been determined crystallographically.The binding mode of such ligands 60 is strongly influenced by the pocket residues Ile CAT and Asp CAT and these, together with the dimensions of the substrate access channel, act to prevent the binding of larger ligands.Monoatomic species such as chloride have also been shown to bind to T2Cu. 42,54 recent study of GtNiR at 1.20 Å resolution 61 revealed electron density consistent with the interpretation that an oxygen ligand was bound at the T2Cu site, in the absence of nitrite or NO, Fig. 4B.CuNiRs have the ability to catalyse the two-electron reduction of O 2 to hydrogen peroxide (H 2 O 2 ) but no structural data for this molecular oxygen ligand has previously been observed.In the GtNiR structure, the O 2 ligand was modelled in a side-on manner with one short (2.1 Å) and one longer (2.4 Å) Cu-O bond length.Oxygen forms H-bonds with Asp CAT and a water molecule and also makes van der Waals contacts with Val CAT .This latter hydrophobic interaction appears to be related to the side on binding mode as previously proposed for NO binding.Oxygen binding has not been observed in other CuNiRs and was proposed to be observable in GtNiR due to the presence of a Val residue in place of Ile CAT , which creates an unusually open substrate binding pocket and/or the slower O 2 -reduction rate in the CuNiR from this species.Interestingly, the structure of a CuNiR from Nitrosomonas europaea 50 revealed a much more restricted substrate channel than in other CuNiRs, proposed to be the basis of the unusual O 2 tolerance of this enzyme enabling it to function in an aerobic environment.In addition, a second water channel was observed and assigned as a water exit channel in this enzyme.
It is a notable feature of CuNiRs that they possess a significant superoxide dismutase activity, 15,62 for example AxNiR at a remarkable B56% activity compared to bovine Cu, Zn superoxide dismutase (CuZnSOD), one of the most efficient enzymes known.The structural similarity between the Cu(I) form of the T2Cu site and that of CuZnSOD has been noted from both EXAFS and crystallographic data. 15Interestingly, CuZnSOD does not display NiR activity.

Identification of intramolecular electron and proton transfer pathways
The CuNiR enzyme reaction requires that one electron and two protons are supplied to the catalytic T2Cu centre and it has been demonstrated that this occurs in a coupled (PCET) manner. 25ectroscopic and mutational studies showed that electrons are transferred from physiological redox partner proteins such as azurin, pseudoazurin or cytochrome c 551 to reduce the T1Cu centre.Intramolecular ET occurs between the Cu sites to deliver the electron to the T2Cu centre, in the process reoxidising the T1Cu.The pathway for intramolecular ET was identified as a Cys-His bridge between ligating residues of the two Cu centres, providing a facile through-bond pathway.
T1Cu site mutations have been characterised revealing effects upon enzyme colour, redox potential and ET.A M150E mutation in AfNiR resulted in a protein binding Zn to the T1Cu site 39 with a 1000-fold loss of activity, while a C130A mutation in AxNiR led to a vacant T1Cu site 23 and abolition of activity.
Analysis of high resolution crystal structures led to proposals for proton channels connecting the protein surface (and bulk solvent) with the T2Cu site.A 'primary' proton channel in AxNiR linked the T2Cu site residue Asp CAT with the surface via a wellordered network of water molecules. 42A 'secondary' channel was later identified in a structure at high pH where residues Asn90, Asn107 and Ala131 together with water molecules also provide a H-bonding network to the surface. 43Supporting evidence was provided from activity and structural work on mutated enzymes.For example, a H254F mutation abolished the 'primary' channel yet retained near-native levels of activity.In contrast the N90S variant targeting the secondary channel lowered activity by 70%.These data led to the proposal that this secondary channel was in fact the most important route for protons. 22,25Similarly, the ET pathway has been investigated by structural and related studies on mutants.As well as the C130A AxNiR mutation described above, a H129V mutation in AxNiR resulted in an unprecedented bis-His T2Cu binding site 63 while disrupting the Cys-His bridge leading to a complete loss of enzyme activity.

The structure of the T2Cu-NO species
All crystal structures of CuNiR-NO complexes have revealed a side-on binding mode to T2Cu, Fig. 3C. 18,45,55Structures have been determined of NO complexes by several approaches: (i) NO-soaking into AfNiR crystals, 45 (ii) endogenously present NO, in crystals grown from protein isolated from Achromobacter cycloclastes cells grown under denitrifying conditions 18 and (iii) produced from nitrite in situ by X-ray driven catalysis using 'multiple-structure one crystal' (MSOX) serial crystallography. 55he structures determined by each method consistently show side-on binding with near-equidistant bond lengths to Cu for the N and O atoms and reveal that bound NO interacts with active site residues.Several issues regarding the NO complex remain challenging to address from the electron density alone.The electron density for the N and O atoms is very similar and structures have been determined with different assignments of the density peaks modelled. 18,45,55Consideration of bonding further suggests that the NO molecule interacts with Asp CAT which adopts its 'proximal' position in the product complex.Similarly, crystal structures are not yet able to give a definitive answer to the identification of the side-on NO complex as Cu(I)-NO or Cu(II)-NO À .
An outstanding mystery is the suggestion that solution spectroscopic data supports an end-on mode of NO binding [Cu-N-O] rather than side-on binding.In particular, ENDOR data 64 of native and the I289V and I257A mutants of RsNiR suggests a Cu-N-O angle of 1601 in native RsNiR with magnetic interaction of O with the C d1 atom of Ile257.Simulations of the complex show 3-10 kcal mol À1 energetic preference for the end-on form. 65A subsequent EPR study of model compound analogues found a clear dependence of g z upon the Cu-N-O angle. 66Taken together with the previous EPR data of RsNiR, 64 this work suggested that a strongly bent (B1361) Cu-N-O binding geometry is feasible in CuNiRs.One proposal from computational simulation was that Ile257 dictates the side-on binding mode observed in structures with the proposal that a small structural change when the protein is in solution would allow the end-on binding mode to be sterically feasible. 67

Surface properties and intermolecular electron transfer
The electrostatic surfaces of CuNiRs are variable, with distinct differences between the blue and green classes of 2 domain enzymes.This, together with comparable analysis of the electrostatic surfaces of the partner proteins 68 provides a rationale for the observed specificity of CuNiRs from different species for their cognate electron donors. 68Azurins reacted rapidly with blue CuNiRs but slowly with green CuNiRs.In contrast, pseudoazurin reacted effectively with either colour of CuNiR.Notably Af NiR (green) has a strong surface negative charge complementing the positive charge of the pseudoazurin complementary surface. 68oreover, a common structural feature of CuNiRs is a hydrophobic patch on the enzyme surface near to the location of the T1Cu centre, which has been proposed to form part of the charged protein-protein interface in the ET complex. 69,70dentifying the precise route(s) of intermolecular ET between the redox centre of the electron donor protein (T1Cu or haem Fe) and the CuNiR T1Cu centre is more challenging, in part due to the difficulty of obtaining crystal structures of transient ET complexes.Mutational analysis of putative pathway residues have provided insights.For example, a W138H mutation in AxNiR produced a large drop in activity with the physiological redox partner azurin but retained full activity with the artificial electron donor methyl viologen. 36This supported a role for Trp138 in intermolecular ET, consistent with a docking model of the azurin-AxNiR complex which positioned Tyr197 close to the azurin T1Cu ligand His117.Tyr197 is close to Trp138 which is adjacent in sequence to a T1Cu His of AxNiR. 36

Structural characterisation of complexes between CuNiRs and redox partners
Structural characterisation of the ET complexes between CuNiRs and their partner proteins have remained elusive due to the transient nature of the complexes.Computational docking and mutagenesis have attempted to identify interface regions between CuNiRs and their specific azurin, pseudoazurin or c-type cytochrome partners e.g. 67Paramagnetic nuclear magnetic resonance (NMR) data have been used to characterise the interacting region in the complex between Af NiR and its partner pseudoazurin [Paz]. 71In this case, distance restraints derived from NMR data of gadolinium added to engineered surface cysteines were used to restrain computational docking simulations.The resulting structural model positions the T1Cu atoms of Paz and Af NiR some 15.5 (AE0.5)Å apart.
One crystal structure available for such an ET complex is that between AxNiR and a functional partner protein, Cyt c551 , at 1.7 Å resolution 49 (PDB 2ZON).This CuNiR has been shown to accept electrons either from azurin, psuedoazurin or a cytochrome electron donor.In the binary AxNiR-Cyt c551 complex, the edge of the haem moiety lies some 10.5 Å from T1Cu in an arrangement likely to promote efficient intermolecular ET, Fig. 5A.The protein-protein interface involves the previously proposed hydrophobic patch surrounded by a semicircle of interface water molecules. 49Analysis of the structure suggested the main ET route to be from the haem CBC methyl carbon through space to Pro88 of the CuNiR.Through-bond ET then proceeds to the adjacent T1Cu ligand His89.
Recently, a heterogeneous complex between AxNiR and a pseudoazurin from H. denitrificans as the electron donor was solved at 3.0 Å resolution (PDB 5B1J).The intermolecular contact at the hydrophobic patches adjacent to the T1Cu sites positions the Cu centres of the two proteins to within B16 Å.The solvent-exposed His81 Cu-ligand of the pseudoazurin lies B2.6 Å from the Ala86 oxygen atom of AxNiR, an arrangement that would allow through-space ET between the proteins, 72 Fig. 5B.

3-Domain fused CuNiRs
A recent development has been the discovery of CuNiRs containing additional domains.The first example was the structure of the hexameric CuNiR from Hyphomicrobium denitrificans (HdNiR), 73 determined at 2.2 Å resolution, 74 Fig. 1.HdNiR has high sequence homology to AniA, while the structure revealed a novel additional N-terminal cupredoxin domain, containing an additional T1Cu(T1Cu N ) centre.The extended enzyme forms a 340 kDa hexameric prism-shaped oligomer.Interestingly, the additional T1Cu N cupredoxin domain has low sequence homology to the other cupredoxin domains of HdNiR but high homology with those of Af NiR.The separation of the T1Cu N centres is 14 Å in a head to head arrangement, suggesting that ET proceeds by initial reduction of T1Cu N by Cyt c550 followed by ET between T1Cu N and T1Cu C , Fig. 6A.The additional cupredoxin domain in HdNiR may not be involved in the catalytic mechanism 74 due to the large distance separating it from the T2Cu centre where catalysis occurs.Subsequent bioinformatics analysis led to the identification of other N-and C-terminal 3-domain CuNiRs. 9The C-terminallyextended enzymes possess an additional haem-cytochrome domain.A crystal structure of a C-terminally extended CuNiR from Pseudoalteromonas haloplanktis (PhNiR) was deposited in the Protein Data Bank in 2008 (PDB 2ZOO) but remains unpublished.The first detailed and published structural analysis of a C-terminally extended CuNiR, came from the atomic resolution (1.01 Å) structure of the 499 residue CuNiR from Ralstonia picketti (RpNiR) by Hasnain and co-workers. 75The 1.01 Å resolution structure of RpNiR 75 revealed a domain binding motif that positioned the T1Cu and Fe atoms only 10.6 Å apart, notably This journal is © The Royal Society of Chemistry 2017 closer than the separation of centres in the AxNiR-Cyt c551 complex, Fig. 5 and 6.A series of water molecules were present between the cytochrome domain and the surface above the T1Cu site, suggesting a role in electronic coupling across the interface.This hypothesis was tested by structural and functional analysis of sitedirected mutants which confirmed the roles of Met92 and Pro93 in interface ET.As well as the differences in domain structure, notable differences are apparent between RpNiR and the 2-domain NiRs in the vicinity of the catalytic T2Cu centre.In contrast to 2-domain NiRs where the T2Cu-bound water is H-bonded to Asp CAT and His CAT , in RpNiR this water instead forms an H-bond to Asp CAT and a second water molecule.In PhNiR, the water is only H-bonded to Asp CAT .
Notably, in the substrate binding channel of RpNiR, a Tyr residue acts to 'plug' the substrate binding channel, excluding water molecules. 75The atomic resolution structure of RpNiR provided clear evidence for the observed preference for nitrite binding to reduced T2Cu in this enzyme, rather than the preferred oxidised state in 2-domain CuNiRs.

Membrane associated CuNiRs
The crystal structure of the soluble domain of the outer membrane AniA protein from Neisseria gonorrhoeae was determined in 2002. 44e AniA protein has around half the specific activity of Af NiR, is essential for growth in low-oxygen conditions with nitrite present and protects against human-sera killing of the pathogen.The soluble domain has a similar fold to the typical CuNiRs but has two shortened loops, with one loop possibly related to interactions with a membrane anchored azurin electron donor and the other likely to interact with the membrane itself.The Cu centres in AniA are similar to those of 2-domain CuNiRs, but with a visible colour intermediate between that of green and blue sites.While bioinformatics identified related proteins to AniA across a range of species, further structures have not yet been forthcoming.

Variations on the classical CuNiR structure and further classification of CuNiRs
In the vicinity of the metal centres, CuNiRs have been divided into classes based on the length of the two loops, the linker loop between the cupredoxin domains forming a monomer and the 'tower loop' adjacent to the T1Cu site. 44The majority of structures are of class I enzymes, while class II includes AniA. 44ome recently discovered CuNiRs fall between these two classes with intermediate length loops, e.g.GkNiR.Several CuNiRs from thermophilic bacteria have been structurally characterised.The structure of Geobacillus kaustophilus HTA426 CuNiR (GkNiR) 51 revealed a novel 28 residue additional a-helical region close to the N-terminus (residues 41-68).Other notable features of GkNiR include the presence of a Val residue in place of Ile CAT and differences in the hydrogen bonding network around the T2Cu centre.
Much of the structural data presented above has come from single, static crystal structures.Recent developments in X-ray sources, rapid detectors and data handling have allowed the determination of both 'damage free' single structures from large numbers of microcrystals and also multiple (potentially hundreds of) sequential datasets from just one protein crystal, making it possible to follow reactions in crystallo.Both approaches show promise for determining structures of relevant catalytic intermediates.

Insights from multiple-structures one crystal (MSOX) serial crystallography
Multiple Structures from One Crystal (MSOX) is a data collection technique which uses solvated electrons generated during X-ray diffraction experiments to drive reactions requiring electron transfer. 76By taking sequential diffraction datasets over the same volume of one crystal it is possible to observe a reaction, occurring through a series of high resolution crystal structures with progressively increasing X-ray dose, Fig. 7.For NiRs, this approach was used to obtain 3 (low, medium and high-dose) datasets from tobacco assimilatory NiR, a Fe-S cluster and sirohaem containing protein that converts nitrite to ammonium. 77A series of this sort for a CuNiR, comprising 45 consecutive datasets collected in 19 s each for AcNiR, between 1.07 Å and 1.62 Å resolutions, clearly showed conversion of bound NO 2 À , to side-on NO 55 within the crystal.At the end of the 45 structure series, a Cu(II) species with bound water is observed after loss of the NO ligand.Crystal structures are not yet able to give a definitive answer to the identification of the observed side-on NO complex as Cu(I)-NO or Cu(II)-NO À .
Little to no change to the overall fold of the enzyme or in the vicinity of the T1Cu site were seen throughout the MSOX series.The low dose atomic resolution starting structure exhibited a previously unseen dual conformation of NO 2 À , in ''top-hat'' and ''side-on'' bidentate orientations, of which the top-hat conformation matches the vertical substrate coordination from 'radiation-damagefree' XFEL structures, 54,58 corroborating the idea that this vertical binding mode represents the initial nitrite binding position.Overall, the 45-dataset MSOX series demonstrated several structural changes at the T2Cu site, Fig. 7. Starting with the dual conformation nitrite and both gatekeeper and proximal Asp CAT conformations, by dataset four the series showed nitrite adopting a single side-on conformation and increasing proximal Asp CAT occupancy, lending credence to the operation of the sensor loop in response to Fig. 6 Proposed electron transfer routes between the redox centres of 3-domain CuNiRs.In (A) À H À dNiR the external electron donor partner protein cyt.c 550 is proposed to bind to a hydrophobic patch spanning the N-terminal (near Ala89, Ile90) and cupredoxin domains (near Leu216), where it is positioned for electron transfer to both the T1Cu N and T1Cu C sites (indicated by arrows). 74In (B) RpNiR, the proposed internal electron transfer route is from the haem (CBC atom) to Met92 and to the T1Cu site. 75Interface and T2Cu water molecules have been omitted from the figures for clarity.the copper oxidation states during catalysis. 23By dataset eleven a second previously unseen conformation was observed, with either NO 2 À , NO or a H 2 O molecule at the T2Cu, indicative of catalytic turnover in the crystal.The changes in gatekeeper and proximal positions of Asp CAT provide some structural evidence for the role of Asp CAT in substrate delivery and product release.

XFEL structures of intact redox states of CuNiRs by serial femtosecond crystallography
Reduction of metalloproteins by X-ray generated solvated electrons has been well-characterised, presents a major challenge for structural biology of such proteins, 78,79 and has been demonstrated to occur in CuNiRs. 271][82] Strategies to mitigate such reduction include merging of composite, low dose partial datasets 83 or helical data collection approaches. 84Use of rapid X-ray detectors and cryogenic temperatures -typically 100 K -minimises but does not eliminate the effects of photoreduction or radiation damage to crystals.However, intact fully-oxidised structures may require the use of femtosecond X-ray pulses that can only be provided with sufficient brilliance by an X-ray free electron laser (XFEL). 85The first CuNiR XFEL structures at room temperature were recently obtained using the Spring-8 Compact Free Electron Laser (SACLA). 54,58omparison of the 1.43 Å resolution serial femtosecond crystallography (SFX) structure of resting state GtNiR 58 with conventional synchrotron radiation (SRX) structures revealed a 101 rotation of the imidazole ring of His CAT , in the SRX data in relation to the SFX structure.This led to the suggestion that His CAT may function as a proton-relay switch, whereby T2Cu reduction causes His CAT rotation thus altering the H-bonding network of His CAT with Thr268 and Glu267 to destabilise the positive charge of His CAT and facilitate proton transfer to the bridging water.
Subsequently, SFX structures were determined of Alcaligenes faecalis NiR (AfNiR) in its resting and nitrite bound (PDB 5D4I) states at 2.03 Å and 1.60 Å resolutions, respectively. 54The SFX structure of intact nitrite bound AfNiR revealed a single, vertical 'top-hat' binding mode of NO 2 À compared to the side on conformation typically observed in SRX structures, 18

Outlook and remaining questions
Several exciting new discoveries of CuNiR enzymes together with methodological developments have taken place in recent years.The advent of rapid mixing technologies at XFEL sources may allow time-resolved structural analysis of the CuNiR reaction with damage-free structures for reaction intermediates.The everincreasing number of sequenced genomes promises the discovery of new variations of CuNiRs with differing domain structures.However, several outstanding questions remain that structural analysis will play a key role in addressing.For example, the protonation states of the important 'catalytic residues' have yet to be unequivocally determined for all stages of the reaction mechanism.Neutron diffraction -another way to obtain room temperature radiation-damage free structures -may help provide this information in the future and its potential for studying CuNiRs, in combination with high resolution X-ray structures, has already been shown. 86Establishing the precise chemical nature of the sideon NO intermediate also remains challenging as do the routes of substrate access and product escape.While significant insights have been gained from the current structures of protein-protein electron transfer complexes, improved structural resolution is required to clarify details of the emerging picture.The dynamic behaviour of CuNiRs is clearly of vital importance and can be addressed by combining structural data, measured at different temperatures, with molecular dynamics and QM/MM simulations.
RPG-2014-355 to MH and RWS.We acknowledge Professors S. S. Hasnain and R. R. Eady and Dr S. Antonyuk for fruitful discussions of CuNiRs over many years.

Fig. 1
Fig. 1 The domain and oligomeric structures of CuNiRs.In class I and II proteins, two cupredoxin domains are present per monomer, here coloured green and purple.Class III CuNiRs have an additional C-or N-terminal domain (yellow) containing T1Cu or haem.The two-domain proteins form a trimeric oligomer, shown in panels A and B. In the threedomain proteins from PhNiR and RpNiR the additional C-terminal cytochrome domain, panels C and E, forms trimeric oligomers, panels D and F, respectively.In the three-domain HdNiR an additional T1Cu N-terminal domain is present, panel G, and the biological unit appears to be a hexameric oligomer, panel H.

Fig. 2
Fig.2The T1Cu and T2Cu centres of AcNiR and the protein links between them.The T1Cu atom has (His) 2 CysMet ligation and is linked to the catalytic T2Cu centre by a Cys-His bridge.A second link is made between the T1Cu ligand His95 and T2Cu site residue Asp98, proposed to be a sensor loop, enabling intramolecular T1-T2 electron transfer.The T2Cu coordination is completed by three more His residues, one of which originates from an adjacent monomer of the trimer and a water molecule.Image generated from the structure of resting state AcNiR at 0.9 Å resolution (PDB 2BW4).17

Fig. 3
Fig. 3 Type 2 Cu site structures (A) resting state AcNiR with a water molecule bound to the T2Cu (PDB 2BW4); (B) nitrite bound to the T2Cu site in a side-on bidentate O manner (PDB 5AKR); (C) the T2Cu site of AcNiR bound to NO at 1.22 Å resolution (PDB 5I6N).The NO ligand is bound in a side-on manner with near-equidistant Cu-N and Cu-O bonds.Residue Asp98 forms a H-bond to the N atom of NO and is oriented toward the proton delivery channel.Panels A and B are from the structures described in ref. 18 while panel C is from ref. 55.
resting state has water bound to oxidised T2Cu and forms a H-bond to the negatively charged Asp CAT residue.Nitrite is observed to bind to the T2Cu ion via a O-coordinate bidentate mode, displacing the bound water molecule and forming H-bonds between the O d2 atom of Asp CAT and the O2 atom of nitrite.Proton transfer occurs from the Asp CAT to the nitrite to form the intermediate T2Cu + NOOH on reduction of the T2Cu site.His255 (His CAT ) facilitates bond cleavage of the N-ON bond to form the product, NO, by supplying the second proton in the widely-accepted mechanism.The NO product dissociates and leaves the resting water-bound oxidised T2Cu ion.Structures have been determined of ascorbate-reduced and reduced nitritesoaked A f NiR at 2.0 [PDB 1AQ8] and 1.85 Å [PDB 1AS8

Fig. 4
Fig.4The T2Cu site environment of GtNiR.(A) The nitrite-complex (PDB code 3WKP),52 note the monodentate binding of nitrite via a single O atom, the substitution of Ile CAT with a Val residue and the steric restriction of the gatekeeper Asp CAT (Asp98) conformation (seen in other CuNiRs) by Phe110, resulting in only the proximal conformation being observed; (B) oxygen binding to GtNiR (PDB code 3WNJ).61Note that partially occupied water molecules are omitted for clarity.Cu atoms are shown as cyan spheres.

Fig. 5
Fig. 5 Structures of electron transfer complexes between CuNiRs and electron transfer proteins.Cu atoms are shown as cyan spheres, while the haem group of cyt.c 551 is represented using red sticks (A) structure of the proposed electron transfer complex between AxNiR (gray) and cyt.c 551 (orange); (B) the interface region of the complex shown in (a), note the close contact between Pro88 of AxNiR and the haem edge of cyt.c 551 .(C) Structure of the proposed electron transfer complex between AxNiR (gray) and pseudoazurin from H. denitrificans (PAz, orange); (D) interface region of the complex shown in (C) here note the close contact between Ala86 of AxNiR and the T1Cu ligand His81 in Paz.

Fig. 7
Fig. 7 Selected frames from the 45-frame MSOX structural movie of AcNiR catalysis at 100 K. 55 (A) The starting nitrite-bound state with partial occupancies in the top-hat and side-on modes (dataset 1, PDB 5i6k); (B) fully side-on nitrite (dataset 4, PDB 5i6l); (C) side-on nitric oxide product complex (dataset 17, PDB 5i6n); (D) water rebinding to the T2Cu following dissociation of NO (dataset 40, PDB 5i6k).Electron density is shown, contoured at 0.53-0.36e À Å À3 , for the ligands and the Asp CAT residue, present in its 'gatekeeper' and 'proximal' orientations.The T2Cu atom is depicted by the cyan sphere and water molecules by red spheres.
suggesting the conformational change to side on NO 2 À results from photoreduction of the metal centres.AfNiR, as in GtNiR, shows a conformational change in His CAT in photoreduced SRX structures compared to oxidised SFX structures.However, unlike GtNiR where His CAT rotates 101 pivoting on N d1 and C b with a 0.3 Å shift in Cb, in AfNiR, His CAT rotates 201 about C g with no change in Cb, switching the H-bonding partner of His CAT N e2 from the Glu279 carbonyl to the hydroxyl of Thr280 which is both a longer and weaker H-bond, Fig. 8. Based on this conformational change between SFX and SRX structure 54 an updated reaction mechanism for CuNiRs was suggested in which nitrite binds to the T2Cu and is protonated by Asp CAT .This is followed by intramolecular ET to give side-on nitrite and rotation of His CAT , where weaker H-bonds facilitate proton transfer from His CAT to the bridging water by destabilising the positive charge of His CAT .With this the necessary second proton can be delivered to form NO in either end-on or side-on binding modes.However, the recently published MSOX series 55 showing X-ray induced ligand turnover from NO 2 À to sideon NO to water shows no change in the H-bond distances between His CAT and Glu279 (2.7 Å) between low and high X-ray doses and a gradual increase in the bond length between His CAT and Thr280 with increasing dose (2.8 Å at 0.69 MGy and 3.1 Å at 27.60 MGy).