insight into catalysis of nitrous oxide reductase from high-resolution structures of resting and...

doi:10.1016/j.jmb.2006.06.064 J. Mol. Biol. (2006) 362, 55–65

Insight into Catalysis of Nitrous Oxide Reductasefrom High-resolution Structures of Restingand Inhibitor-bound Enzyme fromAchromobacter cycloclastes

Konstantinos Paraskevopoulos1,2, Svetlana V. Antonyuk2,R. Gary Sawers3, Robert R. Eady2 and S. Samar Hasnain1,2⁎
1School of BiomolecularSciences, Liverpool JohnMoores University,Liverpool L3 5AF, UK2Molecular Biophysics Group,CCLRC Daresbury Laboratory,Warrington, CheshireWA4 4AD, UK3Max-Planck-Institute forTerrestrial Microbiology,D-35043 Marburg, Germany
Abbreviations used: AcN2OR, Achnitrous oxide reductase; PnN2OR, Pnitrous oxide reductase; PdN2OR, Pnitrous oxide reductase; AxN2OR, Anitrous oxide reductase; N2OR, nitroDFT, density functional theory; EPRparamagnetic resonance.E-mail address of the correspondi

[email protected]

0022-2836/$ - see front matter © 2006 E

The difficult chemistry of nitrous oxide (N2O) reduction to gaseous nitrogen(N2) in biology is catalysed by the novel μ4-sulphide-bridged tetranuclearCuz cluster of the N2O reductases (N2OR). Two spectroscopically distinctforms of this cluster have been identified as CuZ and CuZ*. We haveobtained a 1.86 Å resolution crystal structure of the pink-purple species ofN2OR from Achromobacter cycloclastes (AcN2OR) isolated under aerobicconditions. This structure reveals a previously unobserved ligation withtwo oxygen atoms from H2O/OH– coordinated to Cu1 and Cu4 of thecatalytic centre. We ascribe this structure to be that of the CuZ form of thecluster, since the previously reported structures of two blue species ofN2ORs, also isolated aerobically, have characterised the redox inactive CuZ*form, revealing a single water molecule at Cu4. Exposure of the as-isolatedAcN2OR to sodium iodide led to reduction of the electron-donating CuA siteand the formation of a blue species. Structure determination of this adductat 1.7 Å resolution showed that iodide was bound at the CuZ site bridgingthe Cu1 and Cu4 ions. This structure represents the first observation of aninhibitor bound to the Cu1-Cu4 edge of the catalytic cluster, providing clearevidence for this being the catalytic edge in N2ORs. These structures,together with the published structural and spectroscopic data, give freshinsight into the mode of substrate binding, reduction and catalysis.

© 2006 Elsevier Ltd. All rights reserved.

Keywords: denitrification; catalysis; nitrous oxide binding; copper chem-istry; electron gating
*Corresponding author
Introduction

The biochemical conversion of “inert gaseoussubstrates” is of substantial importance to life onthe planet. A number of metalloenzymes haveevolved to activate the otherwise inert substrates

romobacter cycloclastesseudomonas nauticaaracoccus denitrificanslcaligenes xylosoxidansus oxide reductase;, electron

ng author:

lsevier Ltd. All rights reserve

N2 and N2O, enabling their biological utilisation.The active sites of these enzymes often involvecomplex metal–ligand clusters. How these clustersperform the difficult chemistry of such conversionshas attracted the attention of a wide range ofdisciplines. Nitrogenase is a paradigm example,where structural studies have enabled the rationa-lisation of a tremendous body of spectroscopic dataproviding insightful understanding of how nitrogenfixation is achieved at ambient temperature andpressure by a unique metallo-cluster. Likewise, thecrystal structures of N2OR from Pseudomonas nauticaand Paracoccus denitrificans at 2.4 Å and 1.6 Åresolutions, have revealed a novel μ4-sulphide-bridged tetranuclear CuZ cluster.1 However, under-standinghow this cluster, unique in biology, performsthe difficult chemistry of N2O reduction remains achallenge. Recent spectroscopic and computational

d.

mailto:[email protected]

56 Insights into Catalysis of Nitrous Oxide Reductase

efforts have provided a significant insight into thesequestions.2

The reduction of nitrous oxide is the last step inthe denitrification process of the geo-biologicalnitrogen cycle:

N2Oþ 2Hþ þ 2e� → N2 þH2O ðEoðpH7:0Þ¼ þ1:77 VÞ

N2ORs have been characterised from a wide varietyof bacteria including Achromobacter cycloclastes(AcN2OR), Alcaligenes xylosoxidans (AxN2OR), P.nautica (PnN2OR) and P. denitrificans (PdN2OR).3,4

The reduction of N2O is thermodynamically favour-able but it is kinetically inert, and is a poor ligand fortransition metals. N2ORs are a highly conservedprotein family of ∼130 kDa homodimers containingmultiple Cu atoms. Since the original spectroscopicwork in the 1980s,5,6 more recent, extensive spectro-scopic and mutational studies have shown these Cuatoms to be organized into two distinct types of Cucentres.3,4 Subsequent X-ray crystal structure ana-lysis of PnN2OR and PdN2OR confirmed thepresence of two multi-nuclear Cu centres permonomer.7–9 This includes a CuA site that, inanalogy to the role of this centre in cytochrome coxidase,10 is proposed to mediate electron transferto the second multi-nuclear catalytic CuZ centre. TheCuZ centre is a novel μ4-sulphide-bridged tetra-nuclear Cu cluster ligated by seven His ligands andis the site of N2O reduction. The mechanism bywhich CuZ binds, activates and reduces N2O hasbeen the subject of extensive theoretical efforts,2,11,12

but an experimental structural basis for this remainsto be elucidated and continues to be a challenge.Spectroscopically, two distinct forms of the tetra-

nuclear Cu centre, CuZ and CuZ*, have beenidentified.13 CuZ is predominant in anaerobic pre-parations and undergoes an electrochemicallymediated single-electron oxidative redox process.In contrast, CuZ* predominates in aerobic prepara-tions (with 66% in the CuZ* form and 34% in the CuZform) and, despite being at least as catalyticallyactive, does not undergo an oxidative redox changeunder similar experimental conditions.8 Spectro-scopic data suggest that the basic Cu4S structure iscommon to both forms but the structural relation-ship between them remains unclear.Crystallographic structures of PnN2OR at 2.4 Å

and PdN2OR at 1.6 Å resolutions were those of theblue aerobic preparations of the enzyme and in thesestructures, the tetranuclear catalytic centre wasassigned to the CuZ* form of the cluster.7 Despitethe extensive spectroscopic studies on severalN2ORs, kinetic and mechanistic information islimited in scope, and the mechanism of action ofthis enzyme remains poorly understood. Magneticcircular dichroism spectroscopy, Cu K-edge X-rayabsorption spectroscopyand electron paramagneticresonance (EPR) spectroscopy, in combination withdensity functional theory (DFT) calculations haveshown the resting state of the catalytic centres inboth CuZ and CuZ* to be a 1CuII/3CuI state with an

Stotal=1/2 ground state.13–16 The specific activity ofN2OR in steady-state assays has been shown tocorrelate with the slow reduction of the 3CuI/1CuII

state of the catalytic centre, characteristic of thedithionite-reduced enzyme to the super-reduced4CuI state.13 The fact that the S=1/2 form of CuZcould be reduced further was demonstrated first byGhosh et al.11 Once reduced, exposure of the isolatedenzyme to N2O results in EPR and optical changesconsistent with the oxidation of both the CuA andcatalytic centres, the latter returning to the 3CuI/1CuII state.17 Dinitrogen was identified as a productof the reduction of N2O, but only in a stoichiometryof 0.13 N2 per monomer. The reason(s) for the lowyield is unclear, but since no N2 was formed whenthe as-purified, reductant-free enzymewas used, the4CuI state has been assigned as the catalyticallyrelevant state of the catalytic cluster. However inview of the extremely slow rate of formation of thisoxidation state (0.07s−1), it remains puzzling how itcould be formed andmaintained in normal turnover.This, once-only reductant activation that appears tooccur in turnover argues in favour of ligandexchange being a prerequisite for subsequent rapidredox cycling of the activated enzyme.We report here the structures of AcN2OR in its

aerobically isolated pink-purple form and with theinhibitor iodide bound, at 1.86 Å and 1.7 Å resolu-tion, respectively. The structures of the catalytic sitein both these species represent new forms of thecatalytic cluster. The structure of this cluster in theaerobically isolated enzyme from this organismshows both Cu1 and Cu4 ligated to oxygen-donating ligands (H2O/OH–), which we assign tothe CuZ form of the catalytic cluster, distinct fromthe previously characterised CuZ* form. The inhi-bitor complex reveals iodide binding bridging thesetwo copper atoms. The molecular envelope of thetwo oxygen ligands at Cu1 and Cu4 can accom-modate a bent N2O with O and terminal N in thesame positions as the two observed oxygen atoms inthe native AcN2OR structure. Upon N2 release, theremaining O in an intermediate could occupy abridging position similar to that of the iodideobserved in the inhibitor-bound complex.

Results and Discussion

Spectroscopic data

The aerobically purified AcN2OR had a pink-purple appearance and gave rise to an opticalspectrum exhibiting main peaks at 480 nm, 540 nmand 800 nm ascribed to the oxidised CuA centre(Figure 1, trace a). The minor peak at 650 nm arisesfrom catalytic copper cluster with no signature ofthe oxidised cluster at 560 nm as defined by Ras-mussen et al.18 The optical spectrum of the singlecrystal grown from this material retained these spe-ctral features, again with no suggestion of the CuZcentre being oxidised (Figure 1, trace b). The EPRspectrum of this material at 15K showed the

Table 1. Crystallographic data collection, processing andrefinement statistics of native and inhibitor iodidecomplex of AcN2OR

Native-AcN2OR

Inhibitorcomplex

Resolution, Å (last shell) 50.0–1.86(1.94–1.86)

50.0–1.70(1.76–1.70)

Unit cell parametersa (Å) 70.3 71.1b (Å) 118.2 120.9c (Å) 131.1 137.3

Completeness (%) 95.9 (75.9) 98.4 (87.5)Redundancy 3.6 (2.5) 4.2 (2.6)Unique reflections 88,191 128,667Rmerge (%) 12.6 (43.3) 14.4 (48.0)I/σ(I) (last shell) 7.9 (2.1) 7.8 (1.9)Wilson B-factor (Å2) 18.4 19.6Overall B-factor (Å2) 20.1 20.5R-factor (Rfree) (%) 21.0 (27.1) 16.8 (20.6)r.m.s. deviations from ideal

Bond lengths (Å) 0.014 0.016Bond angles (deg.) 1.6 1.6

ML positional ESU (Å) 0.111 0.073

ML, maximum likelihood; ESU, estimated standard uncertainty.

Figure 1. Optical spectra of AcN2OR: trace a, enzymeas isolated; trace b, crystal of enzyme as isolated; trace c,enzyme after incubation with NaI; and trace d, enzymeafter incubation with NaI treated with excess K3Fe(CN)6.The similarity of traces a and b is evident, showing that theoxidation status of CuA and CuZ in the crystal is the sameas in the AcN2OR as isolated.

57Insights into Catalysis of Nitrous Oxide Reductase

characteristic spectrum with g⊥ 2.066 and gll 2.184.19

These spectroscopic data are consistent with thecatalytic cluster being in the 3CuI/1CuII state.It has been shown that the inhibitor iodide has

a high affinity for N2OR, and its binding resultsin the loss of the spectral features of CuA and theformation of a blue adduct.19 Incubation of nativeAcN2OR with NaI over an extended periodresulted in complete loss of features associatedwith CuA resulting from the reduction of thiscentre (Figure 1, trace c) and the emergence of astrong 650 nm peak, consistent with the CuZcluster remaining in the 3CuI/1CuII state. Sub-sequent oxidation with K3Fe(CN)6 resulted inpartial restoration of the CuA features of theoxidised centre while retaining the 650 nm peakassociated with the 3CuI/1CuII state of the CuZcluster (Figure 1, trace d). The retention of the650 nm peak after treatment with K3Fe(CN)6suggests that the 3CuI/1CuII state in the iodideadduct of the CuZ cluster is redox-inert.

Quality of the models

The structures of the aerobically isolated pinkAcN2OR and its iodide inhibitor complex have beendetermined at 1.86 Å and 1.7 Å resolution, respec-tively. The final R-factor and Rfree are, respectively,21.0% and 27.1% for the native AcN2OR, and 16.8%and 20.6% for the iodide complex (Table 1). Bothcrystals are in the P212121 space group with two

monomers in the asymmetric unit. The average B-factors for the native and iodide-bound complexwere 20.1 Å2 and 20.5 Å2, respectively, withsignificant disorder of the N terminus in bothstructures. The quality of the model was assessedby examining the detailed stereochemistry usingPROCHECK. For the pink native AcN2OR, out ofthe 1036 non-glycine and non-proline residues, 86.5% of them occupied positions in the most favouredregions of a Ramachandran plot, while for theinhibitor-bound AcN2OR, 902 amino acid residuescorresponding to 87.1% of the non-glycine and non-proline residues fell within the most favouredregions, with 12.2% (126 residues) inside theadditionally allowed regions. Each structure has anumber of hetero-atoms such as calcium, sodiumand chloride. In addition, the inhibitor-complexedAcN2OR model contains iodide ions.

Overall crystallographic structures

Like the two previously determined structures ofnitrous oxide reductases,AcN2OR is a homodimer inwhich the two monomers are in close contact. Eachmonomer exhibits the main structural motifs foundpreviously, and is composed of two domains; a C-terminal domain that contains theCuA cluster and anN-terminal domain that contains the catalytic CuZsite (Figure 2). Overall, bothmodels retain the typicaltwo-domain structure observed previously. This isreflected by the small r.m.s. deviations calculatedfrom the superposition of their corresponding Cα

atoms. The two domains of eachmonomer pack anti-parallel to each other, so that the cupredoxin-likedomain of one monomer contacts the β-propellerdomain of the other monomer directly and vice versa.The result is a tightly packed globular proteinmolecule with an extensive monomer–monomerinterface. Analysis of this dimer interface using the

Figure 2. Overall structure of amonomer of AcN2OR showing twodomains; a C-terminal domain thatcontains the CuA cluster (left) andan N-terminal domain that containsthe catalytic CuZ site (right). Cop-per atoms are shown as yellowspheres.

Figure 3. The CuA centre of AcN2OR.


Protein-Protein Interaction (PPI) server20 revealedthat it covers ∼26.5% of the total solvent-accessiblesurface area of the monomer, corresponding to∼6250 Å2. The two monomers are held togetherstrongly with a combination of polar and non-polarinteractions, with around 63% of the atoms in theinterface being non-polar. In addition to the severalvan derWaals interactions between the hydrophobicresidues, over 50 hydrogen bonds ensure the stronglinkage of the two monomers.For both structures, the data were collected at the

SRS and examination of the crystals before and afterthe data collection did not show any visibledifference in colour. Our experience with datacollection on the SRS for a number of other copperproteins, where we have collected Cu X-ray edgedata, persuades us that the copper clusters in ourstructures have not been photo-reduced, consistentwith the lack of colour change. We note that inprevious studies of N2OR undertaken at ESRF,7 amuch more intense source, and EMBL Hamburg, nophoto-reduction was reported.

The CuA site and the cupredoxin domain

The CuA centre of AcN2OR is located in the loopregion connecting two of the β-strands of the β-sandwich of the cupredoxin fold near the C terminusand shows close similarity to the CuA centres of bothPnN2OR and PdN2OR (Figure 3; and see Table 2A).The structural resemblance of the CuA site of N2ORto the CuA site in cytochrome c oxidase was notedpreviously. As with the other two N2OR structures,the CuA site consists of two copper atoms ligated tothe protein by a number of residues; namely, theCys582 and Cys 578 Sγ atoms, Met589 Sδ, the His543andHis586 Nε2 atoms and the Trp580 carbonyl atom(Figure 3). Both cysteine residues participate in theligation of both Cu atoms, while His543, Met589,His586 and Trp580 ligate only Cu1 and Cu2,respectively.

The CuZ site and β-propeller catalytic domain

The CuZ centre of N2OR is located nearly in themiddle of the central channel of the β-propeller(Figure 3) and is composed of four Cu ions arrangedin a distorted tetrahedral arrangement, and abridging inorganic sulphide ion residing aboveand in the middle of the Cu ions within bondingdistance of all of them (Figure 4(a) and (b)). As withthe PdN2OR structure, we find that a “channel”extending from the CuZ centre to the top surface ofthe dimer is present through which the substrate canenter and reach the active site. Examination of theactual position of the copper centre with regards tothe surface of the molecule shows that the distance

Table 2. Stereochemical details

Bond distances (Å) AcN2OR (1.86 Å) AcN2OR-iodide complex (1.7 Å) PdN2OR (1.6 Å)

A. Stereochemical details of the CuA- site in the two structures reported hereCu1-His543 Nδ1 2.2 2.3 2.0Cu1-Met589 Sδ 2.5 2.5 2.5Cu1-Cys578 Sγ 2.4 2.3 2.3Cu1-Cys582 Sγ 2.6 2.5 2.4Cu2-His586 Nδ1 2.2 2.1 2.0Cu2-Trp580 O 2.5 2.5 2.6Cu2-Cys578 Sγ 2.6 2.6 2.3Cu2-Cys582 Sγ 2.4 2.3 2.5Cu1-Cu2 2.5 2.5 2.5Bond angles (deg.)Met589 Sδ-Cu1-His543 Nδ1 102 97 101His543 Nδ1-Cu1-Cys582 Sγ 101 100 98Met589 Sδ-Cu1-Cys578 Sγ 106 123 113Met589 Sδ-Cu1-Cys582 Sγ 103 100 102Cys578 Sγ-Cu1-Cys582 Sγ 120 120 116His543 Nδ1-Cu1-Cys578 Sγ 123 115 124Cys582 Sγ-Cu2-His586 Nδ1 122 129 98Cys582 Sγ-Cu2-Trp580 O 80 101 97His586 Nδ1-Cu2-Trp580 O 97 97 96Cys578 Sγ-Cu2-Cys582 Sγ 122 116 113Cys578 Sγ-Cu2-Trp580 O 80 80 82His586 Nδ1-Cu2-Cys578 Sγ 120 115 116

B. Stereochemical details of the CuZ- site in the two structures reported hereCu1-His285 Nε2 2.2 2.1 2.0Cu1-His340 Nε2 2.2 2.1 2.1Cu2-His142Nε2 2.2 2.1 2.1Cu2-His93 Nε2 2.2 2.2 2.1Cu3-His94 Nε2 2.5 2.4 2.1Cu3-His391 Nε2 2.2 2.1 2.1Cu4-His452 Nδ1 2.1 2.2 2.0Cu1-S 2.5 2.4 2.3Cu2-S 2.1 2.2 2.2Cu3-S 1.9 2.0 2.2Cu4-S 2.1 2.1 2.2Cu1-CuZ3 3.4 3.4 3.4Cu1-CuZ4 3.6 3.0 3.4Cu2-CuZ3 2.7 2.7 2.6Cu2-CuZ4 2.4 2.6 2.6Cu3-CuZ4 2.9 2.8 3.0Cu1-W1 2.2 2.8Cu4-W2 (W1 for PdN2oR) 2.5 2.5Cu1-IOD 2.5Cu4-IOD 2.8Bond angles (deg.)W1(IOD)-Cu1-His340 Nε2 101 109 110W1(IOD)-Cu1-His285 Nε2 119 128 141W1(IOD)-Cu1-S 112 105 88His340 Nε2-Cu1-S 125 126 137His285 Nε2-Cu1-S 95 95 93His340 Nε2-Cu1-His285 Nε2 106 96 95His142 Nε2-Cu2-S 96 86 107His93 Nδ1-Cu2-S 162 164 155His93 Nδ1-Cu2-His142 Nε2 99 109 102His94 Nε2-Cu3-His391 Nε2 110 109 105His391 Nε2-Cu3-S 118 118 112His94 Nε2-Cu3-S 132 130 142His452 Nδ1-Cu4-S 148 147 156His452 Nδ1-Cu4-W2(IOD) 103 110 108W2(IOD)-Cu4-S 108 103 96

A. Stereochemical details of the CuA- site in the two structures reported here. For comparison, the structure of the CuA site in PdN2OR isgiven.B. Stereochemical details of the CuZ- site in the two structures reported here. For comparison, the structure of the CuZ site in PdN2OR isgiven. The second water molecule, W2, is seen only in the 1.86 Å AcN2OR structure.


from the bottom surface (along the channel axis) isaround 25 Å (from Cu3 to Lys189), whereas theminimum distance from the top surface is around11 Å (Cu1 to Gln388).

As with both previous N2OR crystallographicstructures, the copper atoms are ligated to theprotein by seven histidine residues, His93, His94,His142, His285, His340, His391 and His452 (Figure

Figure 4. CuZ cluster of the native and inhibitor-bound AcN2OR structures. Electron density maps are shown,contoured at 1σ: (a) pink native AcN2OR; copper atoms are shown in cyan and the bridging sulphur atom is shown inyellow. The Cu1 and Cu4 atoms of the CuZ cluster are ligated by OH– and water, respectively. A N2O molecule has beenincluded in such a way that the O atom fromN2O ligates Cu4 and the terminal N binds to Cu1 and (b) the inhibitor-boundcomplex: The Cu1 and Cu4 atoms of the CuZ cluster are bridged by an iodide atom.


3). In the previous structures of aerobically preparedblue forms of PnN2OR and PdN2OR,1 an oxygenatom of a water molecule or a hydroxyl group wasalso bonded to Cu4. These structures of the bluespecies were assigned to the CuZ

* form of thecluster.4

In contrast, in our 1.86 Å resolution native pinkAcN2OR structure, clear electron density isobserved for two oxygen atoms. Oxygen 1 isligated to Cu1 and oxygen 2 is ligated to Cu4 at2.2 Å and 2.5 Å, respectively. Even though both ofthese oxygen atoms may originate from watermolecules, the longer distance of 2.5 Å is sugges-tive of it being from a water molecule while theshorter distance is compatible with it being from ahydroxyl group. It is noted that the correspondingcoordinate error for this structure as defined by theE.S.U. values (calculated by the refinement programREFMAC)21,22 is 0.11 Å; therefore, the distancedifference of 0.3 Å is thought to be significant. Thisstructure was predicted to be a stable structure on thebasis of the computational methods, whereas thestructure with two water molecules ligated to theseCu atoms was not.2 The observed distance betweenthe two oxygen atoms is 2.3 Å, a distance able to

accommodate an O-N-N molecule where Cu4 wouldligate to the O atom while Cu1 would ligate to theterminal N. This gives a bent configuration of boundN2O, as suggested by the DFT calculations,2 with anangle of 150° (see Figure 4(a)). In view of thedifferences between our structure and the previouslyreported structures of blue species,1 we suggest thatin our case the structure represents the CuZ form ofthe cluster. As noted above, aerobic preparations ofN2ORs contain some 30% of the CuZ form of thecatalytic cluster. In our case, it would appear that thisform has crystallised preferentially.Examination of the catalytic centre in the inhibitor-

bound complex crystal structure revealed an iodideion bound to Cu1 and Cu4 in a bridging mode(Figure 4(b)).

Structural insight into the redox behaviour ofCuZ species, substrate binding and catalysis

Our structure of the CuZ, together with theprevious structures of the CuZ*, allows us the firststructural insight into the differences in redoxbehaviour of the two CuZ species. In the CuZ*structure, in the catalytic edge of the cluster, Cu1


does not have an exogenous ligand, while Cu4 isbound to a water molecule. In our CuZ structure,Cu1 and Cu4 both have oxygen-donating ligands,presumably from a hydroxyl group and a watermolecule, respectively (see Table 2B). The presenceof an exogenous ligand on Cu1 enables access to theoxidative route from the reduced (3Cul1Cull) to theoxidized (2Cul2Cull) state of the cluster, a findingwith potential relevance to the provision of the twoelectrons required for N2O reduction. Recent DFTcalculations have shown that the binding of N2O tothe 1Cull3Cul state of the catalytic cluster isunfavourable compared to the “super-reduced”4Cul state due to the effective competition bywater for the site.2 Binding of N2O in the μ1,4bridging mode to the super-reduced cluster allowsthe high activation energy barrier for N2O reductionto be overcome by a fast two-electron transferprocess. We suggest that this two-electron processis facilitated by the anchoring of the substrate to

Figure 5. Superposition of the catalytic copper clusterfrom native (shown in magenta) and inhibitor-bound(shown in blue) AcN2OR. The calculated r.m.s. deviationwas 0.28 Å. The iodide atom (from inhibitor-boundcomplex) and the water and hydroxide molecules (fromthe native AcN2OR) were excluded from the r.m.s.calculations.

Figure 6. Superposition of the catalytic copper clusterfrom native AcN2OR (shown in magenta) and PdN2OR(shown in green).

both Cu1 and Cu4, allowing the catalytic centre tocycle between the super-reduced 4Cu1 and theoxidized (2Cul2Cull) state during catalysis. Theposition of Cu1 and Cu4 atoms in our CuZ centreof the native AcN2OR structure has adjusted suchthat an N2O molecule can be accommodated readilywithin the envelope of the experimentally observedwater molecule and hydroxyl group. The bridgingmode exhibited in the iodide–inhibitor complexstructure clearly demonstrated for the first timehere, would then represent the O-bound intermedi-ate (O derived from N2O) following N-O bondcleavage and the loss of N2 and thus represent the[CuZ.O] species in the catalytic cycle.2 Modelling anO atom positions it at ∼2.2 Å from Cu4 and ∼2.6 Åfrom Cu1.A detailed comparison of the CuZ and CuZ*

structures reveals insight into the possible commu-nication between the CuA and the catalytic centres.The calculated overall r.m.s. deviation for thecatalytic cluster of our native AcN2OR structurewith the catalytic clusters of all the other structures,where the catalytic centre represents the CuZ* form,indicates close correspondence in the positions of the


copper atoms and their corresponding proteinligands. However, a number of significant differ-ences are observed for the native AcN2OR. Inparticular, the imidazole ring of His452 (AcN2ORnumbering) has been rotated by∼30° towards His93with a slight increase in its distance from Cu4 (2.04instead of 2.25 Å). It should be noted that His452 isthe only protein ligand that is bonded to Cu4. In thisstructure, the position of the bridging sulphide ionconnecting the four copper atoms of the catalyticcluster shows a concerted shift together with Cu4and His452 (Figures 5 to 7). Differences are observedfor all four Cu-S bond lengths. These differences areseen also when a comparison is made between thenative AcN2OR and the iodide-bound structure. Thecommon feature between the previous structures1

and the iodide complex of AcN2OR is the fact that inall of these cases the CuA centre is reduced, while inthe native pink AcN2OR structure CuA is oxidised.Thus, these differences in structure for CuZ in thenative AcN2OR structure compared to all otherstructures reflects the ability to sense the oxidationstate of the CuA centre in the neighbouring mono-mer. In the PnN2OR structure,9 an electron transferroute was proposed from His504 (His502 in

Figure 7. Superposition of the catalytic copper clusterfrom the inhibitor-bound complex of AcN2OR (shown inblue) and PdN2OR (shown in green).

AcN2OR) to His638 (ligand of Cu2 at the CuA site)via a hydrogen-bonded water molecule. This watermolecule-mediated pathway is observed also in ourcrystal structures of AcN2OR. Site-directed muta-genesis of the gene encoding N2OR from Pseudomo-nas stutzeri has provided further strong support forcommunication between the two centres wheremutation of the ligands of the CuA centre affectsthe spectroscopic properties of the catalytic coppercluster.23 Despite this linkage, it remains puzzlingthat in the isolated enzyme, the reduced CuA site isnot capable of transferring electrons to the catalyticcentre for it to achieve the functionally importantsuper-reduced 4Cu+1 state. How this gating ofelectron transfer is mediated is not clear and remainsan important question to resolve.

Conclusions

The structure of the pink native AcN2OR hasrevealed the presence of a water molecule and ahydroxyl ion bound to the Cu4-Cu1 edge of theCuZ cluster of the resting enzyme. This structurehas provided the definition of the CuZ form of thecatalytic centre, distinct from the CuZ* formdescribed previously. The molecular envelope ofthese two O-donating ligands to Cu1 and Cu4 issuch that the O atom from N2O could bind to Cu4with the terminal N binding to Cu1, giving rise toa bent N2O geometry as predicted by several ofthe DFT calculations. The 1.7 Å resolution structurerepresents the first structure of an inhibitor-boundcomplex of a nitrous oxide reductase. It showsiodide bound in a bridging mode. We suggest thatfollowing bond cleavage N-O and N2 release, theremaining O in an intermediate occupies a brid-ging position similar to that of iodide observed inthe inhibitor-bound complex. A similar proposalhas been made recently by Gorelsky et al.2 Thestructural comparison of the native AcN2OR, thefirst in which the CuA centre is oxidised, withprevious structures and the iodide-bound AcN2ORstructure has revealed structural changes at theCuZ centre in response to the redox state of theCuA centre.

Experimental Procedures

Growth of the organism and protein purification

Achromobacter cycloclastes strain 1013 was obtained fromthe IAM culture collection, University of Tokyo, Japan. Forthe studies reported here, cultures were grown underdenitrifying conditions in a nutrient broth, yeast extract,glycerol medium containing 0.5 g/l of KNO3,

24 modifiedby the inclusion of 5 mM CuSO4.

25 Cultures were grownanaerobically at 30 °C for 50 h in a 200 l fermenter stirredat 50 rpm. In order to minimize the accumulation of toxiclevels of nitrite, three separate additions of 0.5 g/l ofKNO3 and 5 ml of glycerol were added at ∼12 h intervals.


Cells were harvested using a continuous-flow centrifugeand frozen in liquid nitrogen before storage at –80 °C.Crude extracts were prepared by re-suspension of the

frozen cells in 50 mM phosphate buffer, pH 7 (2 ml/gcells) followed by disruption using two passagesthrough a Manton-Gaulin homogeniser at 600 kg/cm2.A small amount of DNase was added before centrifuga-tion at 15,000g for 90 min to remove cell debris andunbroken cells. To prevent auto-reduction of proteins inthe extract, hydrogen peroxide (10 μl of 30%, v/v) wasadded to the resulting crude extract (110 ml). The crudeextract was then loaded onto a column of DEAEcellulose (5 cm×15 cm) equilibrated with 20 mMphosphate buffer (pH 7). The column was washedwith 500 ml of 20 mM phosphate buffer to removeunbound proteins and then with 500 ml of 100 mMphosphate buffer. It was then developed with a 2l lineargradient of NaCl from 100 mM to 250 mM in phosphatebuffer. Separation of coloured proteins on the columnwas followed visually. Pseudoazurin did not bind underthese conditions and was present in the flow-throughduring loading of the crude extract. The gradientresulted in a predominantly greenish band containingnitrite reductase activity that was partially resolved froma purplish band containing nitrous oxide reductase(N2OR) activity.This fraction from the DEAE column was loaded onto

a 2.5 cm×5.5 cm column of hydroxyapatite equilibratedwith 100 mM Tris–HCl (pH 7.4) and, after washing with50 ml of loading buffer, the column was developedstepwise with buffer containing increasing concentrationsof phosphate (50 ml of 5 mM; 30 ml of 10 mM; 40 ml of20 mM and 40 ml of 50 mM). A purple-pink band ofN2OR was eluted at 20–50 mM phosphate. Followingbuffer-exchange into 50 mM Tris–HCl (pH 8), thisfraction was loaded onto a Mono Q10 anion-exchangecolumn, which had been equilibrated with the samebuffer. N2OR was eluted using a NaCl gradient from 0 to1 M. The final purification step involved gel-filtrationchromatography using a Sephadex (S)-200 column(Amersham Biosciences) equilibrated with 50 mM Tris–HCl (pH 8.0), 200 mM NaCl. The column was developedat a flow-rate of 1.0 ml/min, and the eluate wasmonitored by measuring the absorbance at 530 nm and635 nm. The material obtained from the gel-filtration stepwas homogeneous, as judged by SDS PAGE, and had aspecific activity of 1.1 μmol min–1 mg–1 of protein whenassayed in the standard dithionite methyl viologenassay.26

Crystallisation and data collection

An extensive crystallisation screening using the pur-ified pink form of N2OR delivered crystals under twoconditions: condition 1, 20% (w/v) PEG 3350, 0.2 Mpotassium thiocyanate, 0.1 M bis-Tris–propane (pH 6.5);condition 2, 20% PEG 3350, 0.2 M sodium iodide, 0.1 Mbis-Tris–propane (pH 6.5) with an Innovadyne crystal-lisation robot using a protein concentration of 12 mg/ml.Manual optimisation of these conditions using thehanging-drop method with a protein concentration of6 mg/ml yielded crystals for data collection within twodays when 2 μl of protein solution in 10 mM Tris–HCl(pH 8.0) was mixed with 1 ml of reservoir solution fromthe two different conditions and suspended over a 500 μlreservoir at 21 °C. Crystals grown from crystallisationsolution 1 were pink/purple, whereas crystals grown inthe presence of sodium iodide under condition 2 were

light blue. Crystals were soaked in a cryoprotectantsolution containing paratone oil before flash-cooling to100K in the cryostream for X-ray data collection. For theinhibitor-bound complex crystal, all the data werecollected from a single crystal using an ADSC Quan-tum-4 CCD detector on station 9.6 at the SRS, DaresburyLaboratory using an X-ray wavelength of 0.87 Å. A totalof 200 images were collected with an oscillation range of0.5° and an X-ray exposure time of 30 s. In the case of theaerobically isolated pink N2OR crystals, all the data werecollected from a single crystal on station 10.127 at the SRSusing a MAR 225CCD detector and an X-ray wavelengthof 1.28 Å. The data were indexed, scaled and mergedusing HKL2000.28

Structure solution and refinement

The structure was solved by the molecular replace-ment method implemented in MOLREP,29 using the1.6 Å crystal structure of PdN2OR7 as the search model(over 85% amino acid sequence similarity). For the iodide-bound complex crystal form ofAcN2OR, a partially refinedmodel of a monomer from the native pink AcN2ORstructure was used as the search model. Refinement ofthe structures was carried out using the maximum-likelihood method as implemented in REFMAC5.22 Man-ual rebuilding of the model was done using both themolecular graphics programs Coot30 and O.31 Datacollection statistics are summarised in Table 1.

Spectroscopy

Flash-frozen protein crystals of native pink AcN2ORwere cryoprotected in paratone oil, and frozen at 100Kusing a liquid nitrogen cryostream and were examinedusing a diode array microspectrophotometer (4DX Sys-tems, Uppsala). Optical and EPR spectra for the solutionprotein were also obtained. In particular, the reaction ofN2OR with NaI under anaerobic conditions was mon-itored spectrophotometrically using a Perkin-ElmerLambda 35 spectrophotometer. NaI (100 μl of 1 M) wasadded into a stoppered cuvette containing 400 μl of N2ORsolution (24 mg/ml in 25 mM Bis-Tris buffer, pH 6.5).Before mixing, both the protein and NaI solutions wereflushed extensively with gas nitrogen and equilibrated in aglove box for 12 h. Following the prolonged incubation ofN2OR with NaI for a total of seven days (at 4 °C) and itssubsequent reduction, a small excess of K3Fe(CN)6 wasadded to the reaction mixture to investigate the possiblere-oxidation of N2OR.

Protein Data Bank accession codes

Coordinates have been deposited in the Protein DataBank under accession: 2iwf (1.86 Å as-isolated pinkAcN2OR) and 2iwk (1.7 Å blue AcN2OR).

Acknowledgements

The authors thank CCLRC and BBSRC forprovision of facilities at the Daresbury Laboratory


and the John Innes Centre, respectively. We arepleased to acknowledge members of the MolecularBiophysics group (http://biophysics.dl.ac.uk) fortheir help and interest. We are particularly indebtedto Dr Michael Hough for his help. The work wasundertaken on the SRS high-throughput MADbeamline 10.1, which has been funded by BBSRCgrants 719/B15474 and 719/REI20571 and anNWDA project award N0002170.

References

1. Brown, K., Djinovic-Carugo, K., Haltia, T., Cabrito, I.,Saraste, M., Moura, J. J. G. et al. (2000). Revisiting thecatalytic CuZ cluster of nitrous oxide (N2O) reductase.J. Biol. Chem. 275, 41133–41136.

2. Gorelsky, S. I., Ghosh, S. & Solomon, E. I. (2006).Mechanism of N2O reduction by the mu4-S tetra-nuclear CuZ cluster of nitrous oxide reductase. J. Am.Chem. Soc. 128, 278–290.

3. Zumft, W. G. (1997). Cell biology and molecular basisof denitrification.Microbiol. Mol. Biol. Rev. 61, 533–616.

4. Eady, R. R. & Hasnain, S. S. (2003). Denitrification.Compr. Coord. Chem. 8, 759–786.

5. Coyle, C. L., Zumft, W. G., Kroneck, P. M., Korner, H.& Jakob, W. (1985). Nitrous oxide reductase fromdenitrifying Pseudomonas perfectomarina. Purificationand properties of a novel multicopper enzyme. Eur. J.Biochem. 153, 459–467.

6. Kroneck, P. M., Antholine, W. A., Riester, J. & Zumft,W. G. (1988). The cupric site in nitrous oxidereductase contains a mixed-valence [Cu(II),Cu(I)]binuclear center: a multifrequency electron paramag-netic resonance investigation. FEBS Letters, 242,70–74.

7. Haltia, T., Brown, K., Tegoni, M., Cambillau, C.,Saraste, M. & Djinovic-Carugo, M. K. (2003). CrystalStructure of nitrous oxide reductase from Paracoc-cus denitrificans at 1.6 Å resolution. Bioch. J. 369,77–88.

8. Prudencio, M., Pereira, A. S., Tavares, P., Besson, S.,Cabrito, I., Brown, K. et al. (2000). Purification,characterization, and preliminary crystallographicstudy of copper-containing nitrous oxide reductasefrom Pseudomonas nautica 617. Biochemistry, 39,3899–3907.

9. Brown, K., Tegoni, M., Prudencio, M., Pereira, A. S.,Besson, S., Moura, J. J. et al. (2000). A novel type ofcatalytic copper cluster in nitrous oxide reductase.Nature Struct. Biol. 7, 191–195.

10. Iwata, S., Ostermeier, C., Ludwig, B. & Michel, H.(1995). Structure at 2.8 Å resolution of cytochrome coxidase from Paracoccus denitrificans. Nature, 376,660–669.

11. Ghosh, S., Gorelsky, S. I., Chen, P., Cabrito, I.,Moura, J. J. G., Moura, I. & Solomon, E. I. (2003).Activation of N2O reduction by the fully reducedμ4-sulphide bridged tetranuclear CuZ cluster innitrous oxide reductase. J. Am. Chem. Soc. 125,15708–15709.

12. Chen, P. C., Gorelsky, S. I., Ghosh, S. & Solomon, E. I.(2004). N2O reduction by the μ4-sulphide-bridgedtetranuclear CuZ cluster active site. Angew. Chem. Int.Ed. 43, 4132–4140.

13. Chen, P., Cabrito, I., Moura, J. J., Moura, I. & Solomon,E. I. (2002). Spectroscopic and electronic structurestudies of the μ4-sulphide bridged tetranuclear

CuZ cluster in N2O reductase: Molecular insight intothe catalytic mechanism. J. Am. Chem. Soc. 124,10497–10507.

14. Oganesyan, V. S., Rasmussen, T., Fairhurst, S. &Thomson, A. J. (2004). Characterisation of [Cu4S],the catalytic site in nitrous oxide reductase, by EPRspectrocopy. Dalton Trans., 996–1002.

15. Neese, F. & Solomon, E. I. (1999). MCD C-term signs,saturations behavior, and determination of bandpolarisations in randomly orientated systems withspin S >/=(1)/(2) and S=(5)/(2). Inorg. Chem. 38,1847–1865.

16. Chen, P., DeBeer, G., Cabrito, I., Antholine, W. E.,Moura, J. J. G., Moura, I. et al. (2002). Electronicstructure description of the μ4-sulphide bridgedtetranuclear CuZ center in N2O reductase. J. Am.Chem. Soc. 124, 744–745.

17. Chan, J. M., Bollinger, J. A., Grewell, C. L. & Dooley,D. M. (2004). Reductively activated nitrous oxidereductase reacts directly with substrate. J. Am. Chem.Soc. 126, 3030–3031.

18. Rasmussen, T., Berks, B. C., Butt, J. N. & Thomson, A.J. (2002). Multiple forms of the catalytic centre, CuZ, inthe enzyme nitrous oxide reductase from Paracoccuspantotrophus. Biochem. J. 364, 807–815.

19. Hulse, C. L. & Averill, B. A. (1990). Isolation of a highspecific activity pink, monomeric nitrous oxidereductase from Acromobacter cycloclastes. Biochem.Biophys. Res. Commun. 166, 729–735.

20. Jones, S. & Thornton, J. M. (1996). Principles ofprotein-protein interactions. Proc. Natl Acad. Sci. USA,93, 13–20.

21. Murshudov, G. N. & Dodson, E. J. (1997). Simplifiederror estimation a la Cruickshank in macromolecu-lar crystallography. In CCP4 Newsletter on ProteinCrystallography (Winn, M., ed.), Daresbury Labora-tory, vol. January, no. 33, pp. 31–39.

22. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997).Refinement of macromolecular structures by themaximum likelihood method. Acta Crystallog. sect. D,53, 240–255.

23. Charnock, J. M., Dreusch, A., Korner, H., Neese, F.,Nelson, J., Kannt, A. et al. (2000). Structural investiga-tions of the Cu-A centre of nitrous oxide reductasefrom Pseudomonas stutzeri by site-directed mutagen-esis and X-ray absorption spectroscopy. Eur. J.Biochem. 267, 1368–1381.

24. Liu, M. Y., Liu, M. C., Payne, W. J. & Legall, J. (1986).Properties and electron transfer specificity of copperproteins from the denitrifier Achromobacter cycloclastes.J. Bacteriol. 166, 604–608.

25. Libby, E. & Averill, B. A. (1992). Evidence that thetype-2 copper centers are the site of nitrite reductionbyAchromobacter cycloclastes nitrite reductase. Biochem.Biophys. Res. Comms. 187, 1529–1535.

26. Ferretti, S., Grossmann, J. G., Hasnain, S. S., Eady, R. R.& Smith, B. E. (1999). Biochemical characterization andsolution structure of nitrous oxide reductase fromAlcaligenes xylosoxidans (NCIMB 11015).Eur. J. Biochem.259, 651–659.

27. Cianci, M., Antonyuk, S., Bliss, N., Buffey, S. G.,Cheung, K. C., Clarke, J. A. et al. (2005). A high-throughput structural biology/proteomics beamlineat the SRS on a new multipole wiggler. J. Synchr. Rad.12, 442–452.

28. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode.Methods Enzymol. 276, 307–326.

29. Vagin, A. A. & Teplyakov, A. (1997). MOLREP: an

http://www.biophysics.dl.ac.uk


automated program for molecular replacement. J. Appl.Crystallog. 30, 1022–1025.

30. Emsley, P. & Cowtan, K. (2004). Coot: model-buildingtools for molecular graphics. Acta Crystallog. sect. D,60, 2126–2132.

31. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M.(1991). Improved methods for building proteinmodels in electron density maps and the location oferrors in these models. Acta Crystallog. sect. A, 47,110–119.

Edited by R. Huber

(Received 19 May 2006; received in revised form 26 June 2006; accepted 27 June 2006)Available online 12 July 2006

insight into catalysis of nitrous oxide reductase from high-resolution structures of resting and...

Documents