NosL is a dedicated copper chaperone for assembly of the CuZ center of nitrous oxide reductase† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c9sc01053j

The Cu(i)-binding protein NosL functions specifically as an assembly factor for the unique CuZ centre of nitrous oxide reductase (N2OR).


Introduction
Nitrous oxide (N 2 O) is a signicant greenhouse gas with a $300 fold greater global warming potential than CO 2 and an ability to deplete stratospheric ozone. 1 Agriculture produces 65% (6.8 Tg N-N 2 O per year) of the total N 2 O emitted each year. The main contributor within this sector is the soil microbial community, which produces 40% of these emissions. 2 The surge in atmospheric N 2 O from 270 ppb to 324 ppb over the last 100 years correlates strongly with the use of anthropogenic nitrogenbased fertilisers in farming to improve crop yield. 3 The doubling of available nitrogen in the environment has enriched a class of soil dwelling microorganisms called denitriers, which respire anaerobically by reducing nitrate to dinitrogen (N 2 ) gas via the free intermediates nitrite, nitric oxide and N 2 O using different metallo-enzymes. Environmental factors such as soil pH, moisture, carbon to nitrogen ratio, temperature and a lack of copper 4-6 have all been identied as factors leading to increased N 2 O emissions from these microbes.
Encoded by the nosZ gene, the cupro-enzyme nitrous oxide reductase (N 2 OR or NosZ) catalyses the 2-electron reduction of N 2 O to N 2 . Two distinct and approximately equally abundant clades of N 2 OR-containing bacteria and archaea have been identied. 7 Importantly, clade II members act as an N 2 O sink, while members of clade I, such as a-, band g-proteobacteria, are able to produce and remove N 2 O under optimum conditions. In order to assist with future strategies for the control of emissions from soil ecosystems, a key task is to explore how the enzyme is produced and matured in N 2 O emitting bacteria. 3 N 2 OR in the denitrifying a-proteobacterium Paracoccus denitricans (PdN 2 OR) is exported to the periplasm via the twinarginine translocation (TAT) pathway, 8,9 where its two copper centers, Cu A and Cu Z , are assembled. Several crystal structures are now available for N 2 OR, [10][11][12][13] revealing that the Cu A site is housed in a C-terminal cupredoxin domain, while the Cu Z site lies within the N-terminal seven bladed b-propeller domain.
The Cu A center contains two copper ions that are bridged by two conserved Cys residues and further coordinated by Met, His and Trp ligands to form a site that closely resembles the electron transfer Cu A site present in subunit II of cytochrome c oxidase. 14 In N 2 OR the Cu A site also acts as an electron shuttle, accepting electrons from small electron donors such as cytochrome c 550 (ref. 15) or pseudoazurin, in P. denitricans, 16 for the reduction of N 2 O at the Cu Z site. This center comprises four copper atoms coordinated by seven conserved His residues and bridged by one ([4Cu:S]) or two ([4Cu:2S]) suldes, depending on the presence or absence of O 2 , respectively, during purication. 10,12 A major challenge that is particularly important for addressing N 2 O emissions from soil is to understand how the cofactor sites of N 2 OR are assembled and, in doing so, identify assembly/chaperone systems that are involved. For clade I organisms, including P. denitricans, any such systems must be periplasmic, as mutations in the TAT leader sequence of N 2 OR results in the protein remaining in the cytoplasm, in a folded but copper-free state. 17 Studies of the biogenesis of copper cofactor sites of eukaryotic cytochrome c oxidase have identied the proteins Cox17 and ScoB, which are involved in the assembly of the Cu A site, 18 and prokaryotic homologues PCu A C and SenC have been implicated in the maturation of the Cu A site of the aa 3 -type cytochrome c oxidase from Rhodobacter sphaeroides. 19 In comparison, little is known about the assembly of the Cu Z site. The nos gene cluster of clade I N 2 O reducing bacteria vary between denitrifying phyla but contain, in addition to nosZ, ve other core genes (nosRZDFYL), while predominantly aand bproteobacteria members also contain a further two genes (nosC and nosX). nosDFY are predicted to encode an ABC-type transporter homologous to ATM1 from eukaryotes, which transports a sulfur-containing species out of the mitochondrion for Fe-S cluster assembly in the cytoplasm, 20 suggesting that these proteins are likely involved in supplying sulde for assembly of the Cu Z center. In P. denitricans, downstream of nosDFY are two further accessory genes that are part of the same operon, nosL and nosX. The only mutation analysis of nosL has been in P. stutzeri 9 and of nosX in P. denitricans, 21 both of which led to the conclusion they alone are not important for whole-cell N 2 O reduction. However, nosL is part of the denitrication core gene cluster, its expression is responsive to cellular copper status, 22 and the NosL protein has been shown to bind copper, 23 suggesting that it plays a role in maturation or activation of N 2 OR.
Here, we report genetic and biochemical studies of P. deni-tricans nosL/NosL. The data demonstrate that NosL is a Cu(I)binding protein that is required for efficient assembly of the N 2 OR Cu Z center, and thus represents the rst characterised assembly factor for this unique metal center in biology.

NosL is required for N 2 OR activity under copper-limited conditions
Wild type PD1222 (WT), DnosZ (lacking the gene encoding N 2 OR) and DnosL strains (Table S1 †) were cultured under both Cu-sufficient and Cu-limited anaerobic conditions and their growth characteristics investigated. For the WT strain, growth yield was affected under Cu-limited conditions (Fig. 1A). For the DnosZ strain, growth rate and yield were affected under both conditions, consistent with the importance of N 2 OR for optimal growth under anaerobic nitrate-sufficient conditions (Fig. 1B). Growth of the DnosL strain was similar to WT under Cusufficient conditions, but was attenuated under Cu-limited conditions, exhibiting a maximum OD 600 nm similar to the DnosZ strain (Fig. 1C).
The impact of nosL deletion on N 2 OR activity in vivo was assessed by measuring N 2 O levels in the headspace of cultures. Consistent with previous studies, no N 2 O was accumulated by the WT strain under Cu-sufficient conditions and only a transient low level of N 2 O (<1 mM at $16 h) was observed under Cu-limited conditions (Fig. 1D), where the latter is likely due to lower transcription of nosZ under these conditions. 22 In contrast, the DnosZ mutant exhibited a marked Nos-negative phenotype (Nos À ) regardless of copper levels, with N 2 O emitted from cultures to >4 mM aer 24 h (Fig. 1E). Like the WT, essentially no N 2 O was emitted from DnosL cultures under Cu-sufficient conditions. Importantly, however, aer 24 h of growth and once the cells had reached stationary phase, up to 4 mM N 2 O was detected in the headspace of DnosL cultures under Cu-limited conditions, similar to the DnosZ strain (Fig. 1F).
The Nos À phenotype of the DnosL strain under Cu-limited conditions was almost fully complemented by expression of a plasmid-borne functional nosL gene copy from a taurine inducible promoter (Fig. S1 †). In particular, the extent of N 2 O release from the complemented DnosL mutant closely resembled the WT strain in the transient accumulation of N 2 O at $16-20 h.

Absence of NosL results in a copper-and catalytically-decient N 2 OR
To investigate the functional properties of NosL in relation to N 2 OR biogenesis, N 2 OR (NosZ) with a C-terminal Strep-II tag was overproduced in P. denitricans strains grown under Cusufficient conditions and puried. N 2 OR from DnosL cells contained on average 4 Cu atoms per monomer, compared to $6 for the enzyme from DnosZ cells (Table 1). N 2 OR from a DnosZL double mutant contained $3.5 copper atoms per monomer, demonstrating that the presence of the chromosomal nosZ gene in the DnosL cells had little effect on the copper content of the tagged N 2 OR (NosZ) ( Table 1).
As purication of N 2 OR was carried out aerobically, the enzyme from DnosZ cells contained the Cu * Z (pink, form II) form of the active site, which is catalytically inactive. 24 To activate the isolated N 2 OR to the fully reduced form, the enzyme was incubated with excess reduced methyl viologen at room temperature 25,26 for 150 min, at which point N 2 O reductase activity had reached a maximum. N 2 OR from DnosL and DnosZL cells exhibited signicantly lower maximum activities than N 2 OR from DnosZ, see Table 1.

Absence of NosL results in N 2 OR decient in the Cu Z center
The UV-visible absorbance spectrum of as isolated air-oxidized (pink, form II) strep-tagged N 2 OR puried from DnosZ P.
denitricans displayed bands at 480, 535 and 645 nm ( Fig. 2A and S2 †), in agreement with literature for N 2 OR enzymes from a range of bacteria. 11,26-29 These features arise from S 2À to Cu(II) charge-transfer transitions and transitions due to interactions between Cu(I) and Cu(II) ions. 27 The bands at 480 and 550 nm correspond to the Cu A centre, while that at 645 nm is characteristic of the Cu Z centre (in its Cu * Z form). 29 The spectra of N 2 OR enzymes puried from DnosL and DnosZL strains were very similar in the 450-550 nm region, but exhibited signicantly reduced intensity beyond 550 nm  a Total copper contents per monomer were determined using a BCS copper assay. b N 2 O reductase activity was determined using a reduced methyl viologen assay (mmol N 2 O min À1 mg À1 enzyme). Proteins were pre-incubated with a 500-fold excess reduced methyl viologen for 150 min prior to activity assay. All reactions were carried out in triplicate and SD is shown. N.D., not detectable. ( Fig. 2A). The apparent absorbance maximum was shied to $635 nm, consistent with the enhanced relative inuence of the underlying absorbance due to Cu A (maximum at 535 nm). Reduction with sodium dithionite resulted in the spectra shown in Fig. 2B. Bands at 480 and 540 nm were lost, consistent with the reduction of the Cu A center to its colorless diamagnetic Cu(I)/Cu(I) state. The remaining band is characteristic of the Cu Z center following addition of dithionite. 27 N 2 OR from both DnosL and DnosZL strains exhibited a much less intense Cu Z absorbance than that from DnosZ, indicating diverse occupancies of the center. Ferricyanide-oxidized minus dithionite-reduced difference spectra (Fig. 2C) closely overlay, particularly in the 450-550 nm region, demonstrating that the N 2 OR Cu A centers of the different enzymes are essentially identical, and close to fully populated, as estimated by the measured extinction coef-cients. 27 Together, these data demonstrate that N 2 OR isolated from a DnosL background is specically decient in its Cu Z center.

Absence of NosL under copper-limited growth conditions results in complete absence of Cu A and Cu Z centers in N 2 OR
The copper determinations, activity assays and spectroscopic characterizations of N 2 OR described above were performed with samples isolated from cultures grown in copper-sufficient media, in which only a minor growth phenotype for the DnosL strain was observed ( Fig. 1). Thus, N 2 OR characteristics under low copper, where N 2 O is generated from cultures, were investigated. N 2 OR from DnosZ cells contained $5 Cu per monomer (Table 1) with an absorbance spectrum indicative of complete, or near complete, Cu A center population, but a less than stoichiometric population of Cu Z (Fig. 3). In contrast, N 2 OR from DnosL and DnosZL cells contained no detectable copper (Table  1), and gave UV-visible absorbance spectra with no absorbance in the visible region (Fig. 3), indicating complete failure to assemble either of the copper centers of N 2 OR in the absence of NosL.

NosL binds Cu(I) with attomolar affinity
NosL contains a Type-II signal peptidase recognition sequence that, when cleaved, produces a protein with an N-terminal Cys residue that is predicted to bind lipid and anchor NosL into the outer membrane. 23,30 The NMR solution structure of NosL, lacking its membrane anchor sequence, from the b-proteobacterium Achromobacter cycloclastes revealed two independent homologous domains with an unusual bbab topology. 31 The same authors showed that the protein binds Cu(I) specically and XAFS data were consistent with a Cu(I) coordination consisting of S and N/O ligands. 23 To determine the biochemical/ biophysical properties of P. denitricans NosL, the protein lacking its periplasmic export signal sequence and its predicted N-terminal Cys residue 23 was puried resulting in a metal-free form of the protein, which gave a mass of 18 890 Da (predicted mass 18 891 Da) by LC-ESI-MS (Fig. S3 †). The nal, gel ltration step of purication resulted in a broad elution band that suggested a mixture of monomer/dimer association states for NosL, a result conrmed by native PAGE (Fig. S4 †), which showed two species of NosL. Similar observations were made for apo-NosL from A. cycloclastes. 23 Titration of apo-NosL with Cu(I) resulted in the series of spectra shown in Fig. 4A, in which broad absorbance in the near-UV region of the spectrum was observed to gradually increase and saturate at a level of 1 Cu(I) per NosL (Fig. 4B). The absorbance is characteristic of charge transfer transitions involving Cu(I) coordinated to a cysteine thiolate. 32 Cu(I)binding was also investigated by CD spectroscopy, which conrmed the tight association of one Cu(I) per protein but also suggested that further Cu(I) can associate with NosL, albeit weakly (Fig. S5 †). Gel ltration and native PAGE analysis of Cu(I)-NosL (Fig. S4 †) also demonstrated that Cu(I) binding does not signicantly affect the association state equilibrium.
Equivalent titrations with Cu(II) followed by absorbance spectroscopy resulted in spectra very similar to those for Cu(I) (Fig. S6 †), suggesting that Cu(II) may undergo auto-reduction upon binding to NosL. This possibility was investigated using EPR ( Fig. S7 and Table S2 †). Addition of Cu(II) to NosL resulted in a characteristic S ¼ 1/2 Cu(II) signal that, by comparison with a Cu(II) standard, corresponded to only $8% of the Cu(II) initially added. The same sample in the presence of EDTA (a Cu(II) chelator), which prevents Cu(II) ions in solution dimerizing to form EPR-inactive species, resulted in a Cu(II) concentration of $11%. Similar experiments but with the addition of Cu(I) in place of Cu(II) resulted in 4% and 8% in the absence and presence of EDTA, respectively (Table S2 †). As isolated NosL contained <1% Cu(II). Together, these data indicate that Cu(II) undergoes auto-reduction to EPR silent Cu(I) upon binding to NosL.
To further characterise Cu(I)-binding to NosL, ESI-MS under non-denaturing conditions, where non-covalent interactions are preserved, was also employed. Fig. 5 shows the deconvoluted spectrum of apo-NosL with the major peak at 18 890 Da (as observed by LC-ESI-MS, Fig. S3 †), along with a number of lower intensity peaks to the higher mass side, due to noncovalent sodium and ammonium adducts. Attempts to remove these adducts from the non-denaturing MS, via buffer exchange, changes in pH and ionic strength, were unsuccessful. NosL was loaded with a single Cu(I) ion and the peak envelope of the resulting deconvoluted mass spectrum was at +63 Da relative to that of the apo-NosL spectrum (Fig. 5), consistent with the binding of a single Cu(I) ion.
Competition binding experiments using the high-affinity chelator BCS were used to determine the dissociation constant for Cu(I)-binding to NosL. Cu(I)-NosL was titrated with BCS, and the partition of Cu(I) between NosL and BCS determined from measured A 483 nm , due to [Cu(BCS) 2 ] 3À , together with the well-characterised formation constant for Cu(BCS 2 ) 3À , b 2 ¼ 10 19.8 (Fig. S8 †) 33 . From these, an average K d value of $4 Â 10 À18 M was determined 34 (Table S3 †), demonstrating very tight binding of Cu(I) to NosL.

Discussion
Despite nosL being a core component of the nos gene clusters in a range of microorganisms, mutational studies 9 have so far failed to reveal a function in N 2 O reduction. A nosL mutant of Pseudomonas stutzeri (containing an insertion towards the 3 0 end of the gene) exhibited a slightly lower growth rate but produced active, holo-N 2 OR. 9 The presence of a CXXC motif in the encoded protein prompted the suggestion that NosL might be a protein disulde isomerase, but analysis of the sequence of NosL from Sinorhizobium meliloti 35 showed that one of the Cys residues of the P. stutzeri protein is not conserved.
Here, a nosL deletion mutant of P. denitricans was generated and compared to the WT and nosZ deletion strains, as growth and N 2 O benchmarks, under both Cu-limited and Cu-sufficient conditions. The nosL mutant exhibited a Cu-limited growth phenotype relative to WT, associated with a deciency in the activity of N 2 OR, leading to accumulation of N 2 O in the culture headspace. The phenotype was complemented in trans by nosL and also under copper replete conditions, strongly suggesting that NosL functions in an aspect of copper metabolism. The reason why the previously reported P. stutzeri nosL mutant did not present a phenotype is unclear; one possibility is that there was sufficient copper in the growth medium to mask the phenotype, but growth conditions were not reported. 9  Strep-tagged N 2 OR enzymes puried from DnosZ, DnosL and DnosZL mutants exhibited clear differences. Under Cu-sufficient conditions, N 2 OR from DnosL cells contained only $4 Cu per N 2 OR monomer and was substantially less active (though it was apparently sufficiently active in vivo to mask any obvious in vivo phenotype). Importantly, spectroscopic characterisation revealed that the N 2 OR protein from DnosL was specically decient in its Cu Z center, demonstrating that NosL functions in the assembly of this unique biological metal center required for N 2 O destruction.
Under Cu-limited conditions, tagged N 2 OR puried from the DnosZ mutant contained nearly 5 Cu/monomer, while those from DnosL or DnosZL mutant strains lacked copper entirely, indicating complete failure to assemble either N 2 OR copper center in the absence of NosL. Thus, although the Cu A can be reconstituted using copper alone, 36 when copper is limited, NosL may also supply Cu for incorporation into Cu A .
The N-terminal sequence of NosL contains a periplasmic export signal and a lipobox that is predicted to be processed by Lgt, Lsp and Lnt enzymes, resulting in a mature protein that is membrane-anchored via a triacylated N-terminal Cys residue (indicated by a yellow arrowhead in Fig. S9 †). 37 This is likely a substrate of the Lol system, 37 such that it is located in the outer membrane, with its soluble domains facing the periplasm. Studies of a soluble form of NosL, lacking its periplasmic targeting sequence and the N-terminal Cys residue that is acylated, revealed that it binds Cu(I) with attomolar affinity, a characteristic of many copper chaperones.
Our data are consistent with Cys coordination, in agreement with previous studies of A. cycloclastes NosL using thiol-specic reagents and EXAFS, with data from the latter consistent with a three or four coordinate Cu(I) center. The best t was obtained for three coordinate Cu(I) with (O/N)S 2 ligands. 23 Alignment of NosL sequences from the two clades of N 2 O reducing bacteria identies two conserved Cys residues: one through which the protein is believed to be anchored to the membrane (see, Fig. S9, † yellow arrowhead); and, one within a CXM motif that is likely to participate in Cu(I)-binding. 23 The Met residue of this motif is also strictly conserved and was proposed to provide the second sulfur ligand identied by EXAFS. 23 There is one other absolutely conserved Met residue but, as acknowledged by McGuirl et al., there is no absolutely conserved His residue (Fig. S9 †).
Separation of NosL proteins according to the clade to which the organism belongs may shed some further light on this, as it reveals residues that are conserved within, but not between, clades. Clade I NosL proteins, which include P. denitricans and A. cycloclastes, contain a well-conserved His residue close to the CXM motif, resulting in a CXMX 3 H motif. Clade II NosL proteins do not contain this His but instead contain a conserved Cys residue at the N-terminal side of the CXM motif (resulting in a CXXCXM motif). P. stutzeri NosL is the exception to this, as it comes from a Clade I organism but is more similar to Clade II NosL proteins.
In summary, we identify here the rst Cu chaperone with a specic role in Cu Z center assembly. Our data indicate that NosL, which binds Cu(I) with attomolar affinity, is a key part of a high-affinity Cu trafficking pathway that enables assembly of the Cu Z center of N 2 OR. The pathway functions under Cusufficient conditions, but becomes essential under Cu-limited conditions. During copper limitation, the NosL pathway may also play a key role in supplying Cu for the Cu A center. How NosL delivers Cu to N 2 OR is unknown; it is likely that NosL functions directly in the transfer of Cu(I) to N 2 OR, and is thus a copper chaperone, but further studies are now needed to investigate this. The identication of NosL as a key component of a Cu-trafficking pathway for the assembly of the active holoreductase N 2 OR, the sole pivotal enzyme for N 2 O destruction, is a signicant advance towards the long-term aim of mitigating microbial emissions of N 2 O into the atmosphere.

Materials and methods
Construction of mutant nosL and nosZL decient strains of P. denitricans Unmarked deletions of nosL in P. denitricans wild type or DnosZ backgrounds (Table S1 †) were produced by the method of allelic replacement. 22 Briey, regions anking nosL (Table S4 †) were cloned into pK18mobsacB using EcoRI and PstI sites to generate a suicide plasmid (pSPBN1, Table S1 †) that was subsequently conjugated into P. denitricans PD1222 or DnosZ PD2303 using the helper E. coli pRK2013 strain. Single cross-over recombination events were screened using spectinomycin (25 mg ml À1 ) and kanamycin (50 mg ml À1 ). Primary Spec R /Km R transconjugants were grown to stationary phase in Luria Bertani broth (LB) with no antibiotic. Double cross over events were selected for using a high-salt modied LB agar supplemented with 6% (w/v) sucrose. Sucrose resistant colonies with Spec R were screened using colony PCR and gene deletion conrmed by sequencing. Deletion strains were named PD2501 (DnosL) and PD2505 (DnosZL). For complementation of DnosL cells the coding sequence of Pden_4215 was synthesised (Genscript) with anking 5 0 NdeI and 3 0 EcoRI restriction sites and sub-cloned into a taurine-inducible modied pLMB509 plasmid with gentamicin resistance (20 mg ml À1 ) to generate pSPBN2. The complementation plasmid was conjugated into the mutant strain using the helper E. coli pRK2013 strain and successful conjugants were Spec R /Gm R . Expression of nosL from the plasmid was induced by adding 1 mM taurine to the medium at the start of the growth experiment.

Growth and phenotype analysis of cultures
Anaerobic minimal media batch cultures (400 ml) were grown in sealed Duran asks (500 ml total volume), tted with a septum to allow for gas-tight sample extraction. Minimal media consisted of: 30 mM succinate, 20 mM nitrate, 11 mM dihydrogen orthophosphate, 29 mM di-sodium orthophosphate, 0.4 mM magnesium sulfate, 1 mM ammonium chloride, pH 7.5. The minimal media was supplemented with a 2 ml l À1 Vishniac and Santer trace element solution 38 where copper sulfate was present (Cu-sufficient, 12.8 mM) or excluded from the original recipe (Cu-limited, <0.5 mM), as previously described. 22 Media were inoculated using a 1% inoculum from a starter culture to give a starting OD 600 nm of 0.02 and incubated at 30 C. Samples of the liquid culture were taken in 1 ml aliquots and OD 600 nm measured. 3 ml gas samples were removed from the headspace of the cultures and stored in preevacuated 3 ml Exetainer® vials. A 50 ml gas sample was injected into a Clarus 500 gas chromatograph (PerkinElmer) with an Elite-PLOT Q (30 m Â 0.53 mm internal diameter) and an electron capture detector. Carrier gas was N 2 , make-up gas was 95% (v/v) argon, 5% (v/v) methane. Standards containing N 2 O at 0.4, 5, 100, 1000, 5000, and 10 000 ppm (Scientic and Technical Gases) were measured and total N 2 O was determined as previously described. 22 Purication and characterisation of affinity-tagged N 2 OR from P. denitricans strains N 2 OR (NosZ) was expressed in trans in P. denitricans using pLMB511, a derivative plasmid of the taurine-inducible expression vector pLMB509 for a-proteobacteria (Table S1 †). 39 The EcoRI site at position 1107 bps in pLMB509 was removed by PCR-based site-directed mutagenesis to generate pMSL001 (see Table S4 † for primers). Subsequently, pMSL001 was modied by cloning of an NdeI-EcoRI fragment (Table S1 † for sequence) to yield pLMB511, which has a unique NdeI-BamHI-XmaI-EcoRI multiple cloning site that also contains the Strep-II tag sequence. As high-GC content precluded PCR gene amplication, the coding sequence of Pden_4219 (nosZ) was synthesised (GenScript) and cloned into pLMB511 as a NdeI-XmaI fragment, yielding pMSL002 from which N 2 OR (NosZ) with a C-terminal Strep-tag II was overproduced. The pMSL002 plasmid with Gen R (20 ml ml À1 ) was conjugated into PdDnosZ (PD2303), PdDnosL (PD2501) and PdDnosZL (PD2505) using the E. coli pRK2013 helper strain. Conjugants were screened for both Gen R /Spec R and rst cultured in LB and subsequently in 4 L minimal media supplemented with 2 ml l À1 trace element solution, at 30 C. Overproduction of strep-tagged N 2 OR was initiated by the addition of 10 mM taurine when the culture reached OD 600 nm $ 0.6 and cultures were incubated at 30 C for 24 h. Cells were harvested by centrifugation at 5000 Â g and resuspended in binding buffer (20 mM HEPES, 150 mM NaCl, pH 7.2) with a protease inhibitor (cOmplete™ from Roche, 1 tablet per 50 ml resuspended cells) and lysed using a French pressure cell at 1000 psi. The cell lysate was centrifuged at 205 000 Â g for 1 h at 4 C and the supernatant applied to a Hi-Trap HP Strep II affinity column (5 ml, GE Healthcare). N 2 OR-Strep-tag II was eluted using elution buffer (20 mM HEPES, 150 mM NaCl and 2.5 mM desthiobiotin, pH 7.2) and exchanged back into binding buffer using a 30 kDa MWCO Centricon lter unit. Purity of the sample was conrmed using SDS-PAGE analysis and LC-MS. Protein concentrations were determined using the Bradford assay (BioRad) 40 and bovine serum albumin as a protein standard.
UV-visible absorbance spectra of N 2 OR-Strep-tag II were recorded on a Jasco V-550 spectrophotometer. Circular dichroism spectra were recorded using Jasco J-810 Spectropolarimeter. Samples were made anaerobic by sparging with nitrogen gas for 5 min and oxidised or reduced with 5 mg ml À1 stocks of potassium ferricyanide and sodium dithionite, respectively, in 20 mM HEPES, 150 mM NaCl, pH 7.2, by titrating concentration equivalents. Total copper content of the protein was determined using a colorimetric bathocuproinedisulfonic acid (BCS) assay. A 100 ml protein sample was heated to 95 C for 1 h with an equal volume of 20% (v/v) HNO 3 . The reaction was cooled and neutralised using 0.6 ml saturated ammonium sulfate solution. Copper was reduced using 100 ml hydroxylamine (100 mM) and 100 mL BCS (10 mM) added. The absorbance at 483 nm was recorded aer 30 min. A standard curve was produced using a standard copper sulfate solutions (Sigma).
Activities of N 2 OR-Strep-tag II enzymes was determined using an adapted methyl viologen assay 25,41 in which samples were incubated with a 500-fold excess of reduced methyl viologen. Reaction was initiated by adding N 2 O saturated buffer and the oxidation of blue (reduced) methyl viologen to its oxidised colourless form was followed at 600 nm as a function of time and data converted to specic activity using 3 600 nm ¼ 13 600 M À1 cm À1 for the reduced methyl viologen cation radical. 41

Purication and characterisation of NosL
A codon-optimised gene encoding an N-terminally truncated version of PdNosL (Pden_4215) was synthesised (Genscript) and sub-cloned into pET-21a(+) using 5 0 NdeI and 3 0 EcoRI restriction sites to generate pSPBN3. The truncation, which resulted in the replacement of the rst 16 residues with a Met, was designed to simplify the expression and maturation of the protein in E. coli as it yielded a soluble protein located in the cytoplasmic fraction. The pSPBN3 plasmid was used to transform E. coli BL21 (DE3) to ampicillin (100 mg ml À1 ) resistance. Typically, 2 L asks containing 500 ml LB supplemented with 100 mg ml À1 ampicillin were inoculated with 1% (v/v) of an overnight culture and grown for 2 h, 180 rpm, 37 C until OD 600 nm reached $0.6. Expression of the NosL-encoding gene was induced by addition of 500 mM IPTG and cultures were subsequently incubated at 37 C, 180 rpm for 5 h. Cells were harvested by centrifugation at 4000 Â g for 15 min at 4 C, resuspended with buffer A (50 mM MES, pH 6.5) and lysed by three rounds of sonication, each for 8 min 20 s (0.2 s intervals, 50% power), on ice. The cell lysate was centrifuged at 40 000 Â g for 45 min at 4 C and the supernatant applied to a DEAE column (HiPrep DEAE FF 16/10; GE Healthcare) equilibrated in buffer A. NosL was eluted using a 0-50% gradient of buffer B (50 mM MES, 1 M NaCl, pH 6.5). Fractions containing NosL were buffer exchanged using a 10 kDa MWCO Centricon into buffer A and applied to a Q-sepharose column (HiPrep Q FF 16/ 10; GE Healthcare) and eluted using a 20-50% gradient of buffer B. NosL-containing fractions (as determined by SDS-PAGE) were pooled and subsequently applied to an S-100 gel ltration column (120 ml, GE Healthcare) equilibrated in 100 mM MOPS, 100 mM NaCl, pH 7.5 (buffer C) and eluted in the same buffer. Fractions containing pure NosL were combined and dialysed overnight against buffer C containing 1 mM EDTA at 4 C, and subsequently back into buffer C.
An extinction coefficient at 280 nm of 11 923 AE 5.2 M À1 cm À1 , determined using a guanidine hydrochloride assay, 42 was used to quantify the NosL protein concentration. Copper analysis revealed that the protein was puried in a copper-free form.
Copper titrations were carried out using a 1 M stock solution of Cu(I)Cl (in 1 M NaCl and 0.1 M HCl) or CuCl 2 dissolved in water. The protein was titrated with copper to give increases in the ratio of Cu : NosL of 0.1 per addition and spectra were recorded between 240 and 600 nm aer each addition. Competition assays between Cu(I)-NosL and BCS were carried out to measure the dissociation constant, K d , for Cu(I)-binding to NosL, using the extinction coefficient 3 483nm ¼ 13 300 M À1 cm À1 to determine the concentration of [CuBCS 2 ] 3À , as previously described, 33 with 10 min incubation aer each addition. Absorbance spectra were recorded using a Jasco V550 spectrophotometer.
LC-MS was conducted using a Bruker microQTof-QIII electrospray ionisation time of ight (TOF) mass spectrometer calibrated in the m/z range 300-2000 using ESI-L Low Concentration Tuning Mix (Agilent Technologies). Samples were prepared by ten-fold dilution of 50 mM protein solution with 2% (v/v) acetonitrile and 0.1% (v/v) formic acid to 0.5 ml. Samples were loaded into the LC-MS via an autosampler using an UltiMate 3000 HPLC system (Dionex). A 20 ml injection volume of the protein was applied to a ProSwi reversed phase RP-1S column (4.6 Â 50 mm; Dionex) at 25 C. A gradient elution was performed at a ow rate of 200 ml min À1 using solvents A (0.1% formic acid) and B (acetonitrile, 0.1% formic acid). Once loaded the following chromatographic method was used: isocratic wash (2% B, 0-2 min), linear gradient from 2-100% B (2-12 min), followed by an isocratic wash (100% B, 12-14 min) and column re-equilibration (2% B, 14-15 min). Mass spectra were acquired throughout using the following parameters: dry gas ow 8 l min À1 , nebuliser gas pressure 0.8 bar, dry gas 240 C, capillary voltage 4500 V, offset 500 V, collision RF 650 Vpp. Mass spectra from manually chosen elution volumes were averaged and deconvoluted using a maximum entropy deconvolution algorithm in Compass DataAnalysis version 4.1 (Bruker Daltonik).
Samples for non-denaturing ESI-MS were prepared in volatile buffer, ammonium acetate, 50 mM, pH 7.8. Lines were washed with anaerobic buffer prior to sample loading to ensure all O 2 was removed and protein samples were loaded into a Hamilton syringe and directly infused into the ESI source at a rate of 300 ml h À1 . Data was acquired in 5 min increments with ion scans between 500 and 3000 m/z. NosL mass spectra (m/z 1000-3000) were recorded with acquisition controlled by Bruker qTOF Control soware, with parameters as follows: dry gas ow 4 l min À1 , nebulise gas pressure 0.8 Bar, dry gas 180 C, capillary voltage 4000 V, offset 500 V, quadrupole voltage 5 V, collision RF 1000 Vpp, collision cell voltage 20 V. Spectra were deconvoluted as above. Exact masses are reported from peak centroids representing the isotope average neutral mass. Predicted masses are given as the isotope average of the neutral protein or protein complex, in which Cu(I)-binding is expected to be charge compensated. 43 Continuous wave X-band electronic paramagnetic resonance (EPR) measurements were recorded using a Bruker EMX EPR Spectrometer equipped with an ESR-900 liquid helium ow cryostat (Oxford Instruments). Spectra were recorded at 10 K with the following instrumental settings: microwave frequency, 9.4652 GHz; microwave power, 3.18 mW; modulation frequency, 100 kHz; modulation amplitude, 5 G; time constant, 82 ms; scan rate, 22.6 G s À1 ; single scan per spectrum. A 98 mM Cu(II)-EDTA standard was used to estimate Cu(II) concentrations for protein samples by spin integration of signal area for spectra versus that of the standard, where all spectra were recorded under non-saturating power conditions.

Conflicts of interest
The authors declare no competing nancial interests.