Pseudoazurin Dramatically Enhances the Reaction Profile of Nitrite Reduction by Paracoccus pantotrophus Cytochrome cd1 and Facilitates Release of Product Nitric Oxide*

Cytochrome cd1 is a respiratory nitrite reductase found in the periplasm of denitrifying bacteria. When fully reduced Paracoccus pantotrophus cytochrome cd1 is mixed with nitrite in a stopped-flow apparatus in the absence of excess reductant, a kinetically stable complex of enzyme and product forms, assigned as a mixture of cFe(II) d1Fe(II)-NO+ and cFe(III) d1Fe(II)-NO (cd1-X). However, in order for the enzyme to achieve steady-state turnover, product (NO) release must occur. In this work, we have investigated the effect of a physiological electron donor to cytochrome cd1, the copper protein pseudoazurin, on the mechanism of nitrite reduction by the enzyme. Our data clearly show that initially oxidized pseudoazurin causes rapid further turnover by the enzyme to give a final product that we assign as all-ferric cytochrome cd1 with nitrite bound to the d1 heme (i.e. from which NO had dissociated). Pseudoazurin catalyzed this effect even when present at only one-tenth the stoichiometry of cytochrome cd1. In contrast, redox-inert zinc pseudoazurin did not affect cd1-X, indicating a crucial role for electron movement between monomers or individual enzyme dimers rather than simply a protein-protein interaction. Furthermore, formation of cd1-X was, remarkably, accelerated by the presence of pseudoazurin, such that it occurred at a rate consistent with cd1-X being an intermediate in the catalytic cycle. It is clear that cytochrome cd1 functions significantly differently in the presence of its two substrates, nitrite and electron donor protein, than in the presence of nitrite alone.

Cytochrome cd 1 is a nitrite reductase found in the periplasm of many denitrifying bacteria, including Paracoccus pantotrophus and Pseudomonas aeruginosa (1,2). The enzyme is a homodimer; each monomer contains one c-type heme center and one d 1 heme cofactor (3). The c heme is located in the N-terminal region of the enzyme, which is predominantly ␣-helical (with a fold similar to that of mitochondrial cytochrome c). As in the vast majority of c-type cytochromes (4,5), the heme (Fe-protoporphyrin IX) is covalently bound to the polypeptide via two thioether bonds that form between the heme vinyl groups and cysteine thiols that occur in a Cys-Xaa-Xaa-Cys-His motif. In cytochromes cd 1 , the c heme is the site of electron acceptance from the external donor proteins (6,7), which are, in P. pantotrophus, the cupredoxin pseudoazurin and cytochrome c 550 (8). The d 1 heme is a dioxoisobacteriochlorin (a four-electron reduced porphyrin relative to "normal" b heme and with other variant features) (9) and is unique to this class of enzyme. It is noncovalently bound to the polypeptide chain and is at the active site (10); the d 1 heme binding domain has an eight-bladed ␤-propeller structure. Nitrite (NO 2 Ϫ ) binds via its nitrogen atom to the iron of the d 1 heme, where it is reduced by one electron to form nitric oxide (10); the other reaction product is water, with the two required protons being provided by two highly conserved histidine residues in the d 1 heme pocket (11). In addition to reduction of nitrite to NO, cytochrome cd 1 is capable of catalyzing the two-electron reduction of hydroxylamine to ammonia and the four-electron reduction of oxygen to water. The enzyme was long thought to be an oxidase; however, it is now widely accepted that its physiological role is the reduction of nitrite to nitric oxide during denitrification (12)(13)(14)(15).
The crystal structure of oxidized P. pantotrophus cytochrome cd 1 "as isolated" revealed that the c heme is axially ligated by histidines 69 and 17 and that the d 1 heme iron is coordinated by histidine 200 and tyrosine 25 (3). The distal tyrosine ligand to the d 1 heme is provided by the c heme domain of the protein and is connected to the c heme distal ligand, His-17, by a short loop. Upon reduction of the enzyme, a remarkable switch in heme coordinations occurs. His-17 is replaced by Met-106 as a ligand to the c heme, resulting in His/Met coordination, and Tyr-25 dissociates from the d 1 heme, leaving the iron pentacoordinate and able to bind substrate (10). The oxidized "as isolated" state of the enzyme is catalytically inert, but the enzyme can be activated by reduction (12,16). When reduced enzyme is reoxidized, the oxidized form generated (which retains His/Met c heme coordination) is catalytically active in the presence of electron donor and substrate but, in their absence, reverts slowly (over a few min) to the inactive "as isolated" conformation (which has His/His-coordinated c heme) (16,17). There is no evidence for c heme ligand switching occurring during the catalytic cycle of the enzyme.
There remain many mechanistic uncertainties for cytochrome cd 1 , prominent among which is the mode of release of nitric oxide from the d 1 heme. Full, catalytically competent dis-sociation of NO has so far not been observed from the P. pantotrophus enzyme other than in steady-state turnover experiments. George et al. (18) demonstrated that fully reduced cytochrome cd 1 in the absence of excess reductant forms a stable complex with an optical absorbance maximum at 632 nm when mixed in a stopped-flow apparatus with 5 mM potassium nitrite (key spectroscopic data in this paper are summarized in Table 1). The complex was assigned as a mixture of cFe(II) d 1 Fe(II)-NO ϩ and cFe(III) d 1 Fe(II)-NO and was shown to persist over many h in the dark, although in intense light it decayed with a rate constant of about 0.2 s Ϫ1 . Throughout this work, this complex will be referred to, for simplicity, as cd 1 -X. 3 Steady-state turnover of fully reduced P. pantotrophus cytochrome cd 1 can be achieved in the presence of the reduced electron donors cytochrome c 550 , pseudoazurin, and the nonphysiological horse heart cytochrome c; thus, product release must occur, and, by implication, the presence of electron donor must contribute to effecting product release (12). This may be through a protein-protein interaction, or it may be as a result of the electron donation itself. To test these possibilities, we have mixed fully reduced cytochrome cd 1 plus potassium nitrite, in the absence of excess reductant, with fully oxidized pseudoazurin (in the absence of excess oxidant) and also with pseudoazurin from which the normal copper had been extracted and replaced with zinc. The results of these experiments are presented here, and implications for the physiological mechanism of nitric oxide dissociation from cytochrome cd 1 are discussed.

EXPERIMENTAL PROCEDURES
Preparation of Cytochrome cd 1 -P. pantotrophus was grown in anaerobic conditions at 37°C. Cytochrome cd 1 was purified from the periplasm of the cells according to the method of Moir et al. (19), as modified by Koppenhöfer et al. (20). The purity of the enzyme was determined by the R z value (A 406 /A 280 ), and all cytochrome cd 1 used in this work had an R z Ͼ 1.25. The concentration of the enzyme was determined at 406 nm for the oxidized enzyme and 418 nm for the reduced, using the respective extinction coefficients of 142.5 mM Ϫ1 cm Ϫ1 (21) and 161.5 mM Ϫ1 cm Ϫ1 (20). These extinction coefficients refer to the concentration of the enzyme monomer; throughout this work, the enzyme concentration will be given as monomer concentration. Fully reduced cytochrome cd 1 , which has oxidase activity (20), was produced by reduction with sodium dithionite in an anaerobic glove box (Ͻ2 ppm of O 2 ; Faircrest Ltd.); the excess reductant was removed by passing the enzyme down a desalting column packed with P6-DG resin (Bio-Rad) and equilibrated with 50 mM potassium phosphate buffer.
Preparation of Pseudoazurin-Pseudoazurin was purified according to the method of Moir et al. (19) from the total soluble cell extract of Escherichia coli XL-1 Blue transformed with the plasmid pJR2 (22). Cells were grown aerobically for 16 h on 2ϫ TY medium containing 2 mM CuSO 4 and 100 g ml Ϫ1 ampicillin. To ensure that the pure pseudoazurin was fully oxidized for use in kinetics experiments, a small excess of potas-sium ferricyanide was added. The excess oxidant was separated from the protein using a desalting column packed with P6-DG resin (Bio-Rad).
Preparation of Zinc Pseudoazurin-Apopseudoazurin was prepared essentially according to the method previously described (23). The apopseudoazurin was washed by passing down a desalting column packed with P6-DG resin and equilibrated in water. Buffering apopseudoazurin in 50 mM potassium phosphate solution was found to cause protein precipitation. The apopseudoazurin was incubated with 20 mM ZnCl 2 for 1 h at room temperature and then washed with water as above. The concentration of zinc pseudoazurin, which is colorless, was estimated from its predicted extinction coefficient at 280 nm calculated using the program ProtParam (available on the World Wide Web).
Characterization of Zinc Pseudoazurin-Zinc pseudoazurin, like reduced copper pseudoazurin, is colorless; therefore, it was not possible to verify the presence of zinc holopseudoazurin by any characteristic absorbance in the visible region of the spectrum. However, a comparison of the UV region of the spectrum with that of copper holopseudoazurin indicated the same absorbance pattern (data not shown). MALDI-TOF mass spectrometry indicated that the zinc pseudoazurin was of the expected molecular weight and had not been degraded during the preparation process, although the metal dissociates from the polypeptide in this experiment. The addition of potassium ferricyanide, in an attempt to oxidize any residual copper bound to the protein, did not cause any increase in absorbance at 590 nm, indicating that the process of making the zinc pseudoazurin had not merely yielded reduced copper pseudoazurin. The addition of copper to the zinc pseudoazurin, followed by oxidation with potassium ferricyanide, did not result in any increase in absorbance at 590 nm, indicating that apopseudoazurin was not present; apopseudoazurin takes up copper to reform holopseudoazurin, which absorbs maximally at 590 nm in the oxidized state (19).
The circular dichroism spectrum of the zinc pseudoazurin was compared with that of copper pseudoazurin (data not shown). The spectrum of copper pseudoazurin was as previously published (23), and that of the zinc pseudoazurin was very similar, although the spectrum of the latter possibly indicated the presence of a minor amount of unfolded protein, which is likely to have been produced during the process of extracting the copper. The zinc pseudoazurin was also denatured with 4 M urea; the unfolded form gave a completely different CD spectrum showing that, although there may have been a small percentage of misfolded protein present in the zinc pseudoazurin stock, the majority was not unfolded.
Stopped-flow UV-visible Spectroscopy-Stopped-flow experiments were performed on a Hi-Tech SF-61 DX2 double mixing stopped-flow spectrophotometer interfaced with a CU-61 control unit (Hi-Tech Scientific). The path length of the optical cell was 10 mm, and the dead-time of the instrument was ϳ2 ms. The stopped-flow unit was housed entirely within an anaerobic glove box maintained at Ͻ2 ppm O 2 . Unless otherwise stated, all solutions of enzyme and substrate were prepared in 50 mM potassium phosphate buffer, pH 7.0, and experiments were carried out at 25°C. Both single and multimixing experiments were carried out, and details of these are given under "Results" and in each of the figure legends. For multimixing experiments, enzyme and substrates were prepared at 4 times the final concentration required, and in single mixing experiments, they were prepared at double the final required concentration.
EPR Spectroscopy-EPR spectra were recorded using conditions similar to those previously described (18) on a Bruker ELEXSYS 500 spectrometer with an ER049X SuperX microwave bridge and shq cavity, fitted with an Oxford Instruments ER-900 liquid helium-cooled cryostat. Integrations of EPR signals were performed according to the method of Aasa and Vänngård (24).

RESULTS
Pseudoazurin Promotes Decay of cd 1 -X-A stopped-flow multimixing experiment was carried out in which fully reduced P. pantotrophus cytochrome cd 1 was mixed with 4 mM potassium nitrite to form cd 1 -X, which is an approximately equimolar mixture of cFe(III) d 1 Fe(II)-NO and cFe(II) d 1 Fe(II)-NO ϩ (18). This mixture was allowed to age for ϳ82 ms, the time scale on which George et al. (18) reported full formation of cd 1 -X, and was then mixed with oxidized pseudoazurin. When a stoichiometric ratio of initially oxidized pseudoazurin to d 1 heme was used, the disappearance of cd 1 -X was lost in the dead time of the instrument (ϳ2 ms), and, therefore, the experiment was repeated using pseudoazurin at one-tenth the stoichiometry of enzyme monomer. Diode array spectra were collected on a logarithmic time base for 2 s after mixing with pseudoazurin ( Fig. 1). The first time point after mixing with pseudoazurin, at which a peak at 632 nm is clearly visible and a significant proportion of the c heme is still reduced, as judged by the characteristic split ␣-band absorbance around 550 nm, is shown in boldface type. These spectral features indicate the presence of cd 1 -X (18) ( Table 1). At later time points after mixing with pseudoazurin at one-tenth the stoichiometry of enzyme monomer, the peak at 632 nm attenuated and was red-shifted to 641 nm, with oxidation occurring at the c heme. The final species was almost completely oxidized at the c heme and had a d 1 heme absorbance maximum at 641 nm ( Fig. 1; Table 1). In other experiments (e.g. Fig. 2), the final product was fully oxidized at the c heme, and the presence of the residual absorbance around 550 nm in Fig. 1 is probably explained by the presence of a small amount of reduced semiapocytochrome cd 1 (i.e. lacking the d 1 heme) (13). The most likely assignment of the final product of the reaction between pseudoazurin and cd 1 -X is all-ferric cytochrome cd 1 with nitrite bound (i.e. cFe(III) d 1 Fe(III)-NO 2 Ϫ ), as was suggested by George et al. for the product of photodissociation of cd 1 -X (18). If so, the enzyme has been oxidized relative to cd 1 -X, and NO has been released. The final products of photocatalyzed and pseudoazurin-dependent NO dissociation from cytochrome cd 1 have essentially identical absorption spectra (Fig. 2); this was the same whether the pseudoazurin concentration was stoichiometric with d 1 heme or at one-tenth the stoichiometry. Another possible assignment for the species obtained after reaction of initially oxidized pseudoazurin with cd 1 -X would be the all-ferric enzyme with NO bound at the d 1 heme. However, this is unlikely, since the spectrum does not vary with pH in the manner that we have observed for the all-ferric NO-bound form of FIGURE 1. Pseudoazurin catalyzes further turnover by, and product dissociation from, the stable complex formed when P. pantotrophus cytochrome cd 1 reacts with nitrite. Shown are the optical spectra obtained after mixing fully reduced P. pantotrophus cytochrome cd 1 (in the absence of excess reductant) with 2 mM potassium nitrite, followed by mixing with initially oxidized pseudoazurin at one-tenth the stoichiometry of enzyme monomer. The concentration of cytochrome cd 1 was ϳ16 M before mixing. The aging time between the first and second mixing was 82 ms. After the second mixing, 200 spectra were recorded over a period of 2 s with a logarithmic time base; the minimum data interval was 2 ms. For clarity, not all spectra are shown. The first time point after mixing with pseudoazurin (0.003 s) is shown in boldface type, and the final spectrum (1.9 s) has the lowest absorbance at 550 and 632 nm. All solutions were prepared in 50 mM potassium phosphate buffer, pH 7.0. The reaction was carried out anaerobically at 25°C.

TABLE 1
Absorption maxima for P. pantotrophus cytochrome cd 1

and its complexes referred to in this work
Unless stated, data are for d 1 heme absorption bands, were obtained at pH 7.0, and are for wild type enzyme. Rows i and ii refer to oxidized "as isolated" cytochrome cd 1 and fully reduced cytochrome cd 1 , respectively, and are given for reference. Rows iii-vi refer to spectra observed during the formation of the complex called cd 1 -X (iii and iv) and after the decay of cd 1 -X (v and vi). To obtain the spectra of these complexes, fully reduced cytochrome cd 1 was mixed with nitrite in a stopped-flow apparatus, and spectra were recorded over a period of 2 s with a minimum data interval of 2 ms. cd 1 -X was either allowed to decay in the intense light of the diode array spectrophotometer (v) or was mixed with initially oxidized pseudoazurin (vi). Rows vii-ix refer to standard (model) cytochrome cd 1 complexes used in the assignment of spectra described in this work; vii and viii describe spectra of the Y25S variant of P. pantotrophus cytochrome cd 1 (27,34). cytochrome cd 1 ( Table 1). The final product of the reaction between pseudoazurin and cd 1 -X at pH 7.0 had a visible absorption peak at 641 nm, and when the same experiment was repeated at pH 6.0, the peak was at 639 nm. Model spectra for both nitrite and NO bound to fully oxidized cytochrome cd 1 were obtained using the Y25S variant of P. pantotrophus cytochrome cd 1 , which, unlike the "as isolated" oxidized wild type enzyme, is fully competent for exogenous ligand binding (25)(26)(27). The d 1 heme peak in the spectrum of nitrite bound to all-ferric Y25S enzyme varied from 639 to 642 nm as the pH shifted from 6.0 to 7.0 (Table 1). However, in the spectrum of NO bound to all-ferric Y25S cytochrome cd 1 , the d 1 heme peak varied between 632 nm at pH 6.0 and 640 nm at pH 7.0. The possibility of the final species produced by the reaction between pseudoazurin and cd 1 -X being ferrous d 1 heme with NO bound was ruled out using EPR spectroscopy. A signal corresponding to d 1 Fe(II)-NO (28) was observed in the spectrum of a sample of cd 1 -X reacted with pseudoazurin, but quantitation showed that it represented only ϳ1% of the total d 1 heme that was present. Upon close inspection of the data shown in Fig. 1, it is apparent that there are two distinct kinetic phases of the reaction between cd 1 -X and initially oxidized pseudoazurin at one-tenth the stoichiometry of the cd 1 monomer. In the region above 600 nm, where only the d 1 heme absorbs (21), the phases can be clearly separated from one another. Up to 100 ms after mixing cd 1 -X with pseudoazurin, the first phase was dominant, and an isosbestic point was apparent at 642 nm. From 100 ms onward, the second phase dominated the absorbance change with an isosbestic point at 635 nm.

Protein or protein complex
The first 100 ms of the absorbance change at 635 nm ( Fig. 3A) fitted to a single exponential with k obs ϭ 24 s Ϫ1 . The rate constant of the second phase of the reaction was determined by plotting the absorbance change at 642 nm, the wavelength at which there is no change during the first phase of the reaction, against time (Fig. 3B). After an initial lag of ϳ100 ms, the change at 642 nm fitted to a single exponential with k obs ϭ 5 s Ϫ1 . Oxidation of the c heme was measured by subtraction of the change in absorbance at 539 nm from the change in absorbance at 549 nm (Fig. 3C). This enables determination of the absorbance change at the c heme by substantially eliminating the d 1 heme contribution and is necessary because there is no wavelength at which the c heme absorbs without some contri-  bution from the d 1 heme (see also Ref. 18). The two separate phases of the 549 nm minus 539 nm trace (up to 100 ms and from 100 ms onward) were fitted to separate single exponentials with k obs1 ϭ 10 s Ϫ1 and k obs2 ϭ 4 s Ϫ1 . The first phase of oxidation of the c heme appears to occur at about half the rate of the first phase occurring at the d 1 heme in which decay of the 632 nm peak is observed. The second phase of oxidation of the c heme occurs at a rate similar to that of the second phase occurring at the d 1 heme.
It is clear from the data presented so far that the addition of pseudoazurin causes the otherwise kinetically stable complex of cytochrome cd 1 and NO, referred to as cd 1 -X in this work, to react in a manner that is dependent on pseudoazurin concentration. We assign the product of this reaction as fully oxidized cytochrome cd 1 with nitrite bound to the d 1 heme (i.e. from which NO had dissociated). This product is apparently the same as the photodissociated product observed after a period of ϳ1 min when cd 1 -X was mixed with buffer and allowed to decay in the intense light of the diode array beam (Fig. 2) (18). Photolysis is a well known method of dissociating ligands such as CO and NO from heme (29), and the fact that the products of photocatalyzed dissociation and that dependent on pseudoazurin are essentially identical (Fig. 2) provides strong evidence to suggest that pseudoazurin causes NO release from the enzyme. It is also noteworthy that the pH dependence of the absorption maxima of the final products of photolysis-dependent and pseudoazurin-dependent reaction of cd 1 -X is comparable ( Table 1). The photodissociation reaction is slow and biphasic, indicated by a double exponential fit to the decay of the cd 1 -X 632 nm peak with rate constants of 0.25 s Ϫ1 and 0.06 s Ϫ1 . This is more than 10-fold slower than the slowest phase occurring during pseudoazurin-mediated dissociation of NO from cytochrome cd 1 , confirming that photolysis can be ruled out as an explanation of the effect seen with pseudoazurin.
Pseudoazurin Prevents Accumulation of cd 1 -X-The current work has shown that pseudoazurin causes product release from cytochrome cd 1 . Of at least equal importance is whether it prevents formation of the stable complex, cd 1 -X, between the enzyme and its product NO. At pH 7.0, cd 1 -X forms slowly in relation to the rate of steady-state turnover of nitrite by cytochrome cd 1 (18). The latter occurs at 72 s Ϫ1 per monomer (12), indicating that the length of the catalytic cycle is ϳ14 ms at pH 7.0, whereas cd 1 -X forms with k obs ϭ 40 s Ϫ1 at this pH. To determine whether this stable complex of enzyme and product is formed in the presence of pseudoazurin, fully reduced cytochrome cd 1 was mixed in the stopped-flow apparatus with a combination of 4 mM potassium nitrite and either one-tenth stoichiometric (Fig. 4) or stoichiometric (Fig. 5) initially oxidized pseudoazurin.
As can be seen from the spectra in Figs. 4 and 5, the concentration of pseudoazurin has a significant effect on the reaction. In the presence of pseudoazurin at one-tenth the stoichiometry of enzyme monomer, cytochrome cd 1 first reacts with nitrite to form a species with an absorbance maximum at 632 nm, the absorbance maximum of cd 1 -X, which then decays in a biphasic process to give the final product described earlier with a maxi-mum at ϳ641 nm and fully oxidized c heme (Table 1). Formation and subsequent decay of the 632 nm peak are shown separately in Fig. 4, A and B. In the presence of a 1:1 ratio of  pseudoazurin to enzyme monomer, separate formation and decay of the 632 nm peak is not observed, and there appears to be a smooth transition toward the final product (Fig. 5); the difference between this experiment and the reaction in the absence of pseudoazurin (Fig. 5, inset) is striking. Particularly noteworthy is that in the absence of pseudoazurin, the first observed intermediate of reaction between cytochrome cd 1 and nitrite in the stopped-flow apparatus has significant absorbance at 660 nm (Table 1); the absorbance at 660 nm then diminishes as cd 1 -X forms (Figs. 4 and 5). When fully reduced cytochrome cd 1 was mixed in the stopped-flow apparatus with pseudoazurin at one-tenth the stoichiometry of enzyme monomer and nitrite, the first time point showed significant absorbance at 660 nm, which decreased concomitantly with an increase in absorbance at 632 nm, as is seen in the absence of pseudoazurin. When the same experiment was conducted in the presence of stoichiometric pseudoazurin, the initial observed time point of the reaction showed much less absorbance at 660 nm than with substoichiometric pseudoazurin (Fig. 5). It appears that, remarkably, pseudoazurin, in a concentration-dependent manner, accelerates formation of intermediates in the reaction between cytochrome cd 1 and nitrite and promotes the release of product from the enzyme.
The reaction of fully reduced cytochrome cd 1 with potassium nitrite and pseudoazurin at one-tenth the stoichiometry of enzyme monomer was analyzed kinetically. The first phase was fitted at 635 nm, the wavelength at which no absorbance change occurs during the subsequent phases, and showed that formation of cd 1 -X occurred with a rate constant of 77 s Ϫ1 (Fig. 6A). This is faster than the rate of formation of cd 1 -X in the absence of pseudoazurin, which occurs with a rate constant of 40 s Ϫ1 at pH 7.0 (18) (this work). The decay of cd 1 -X was fitted to a double exponential at 650 nm, the wavelength at which no absorbance change occurs during the first phase of the reaction (Fig. 6B). The observed rate constants were 17 s Ϫ1 and 3 s Ϫ1 , comparable with those observed for the reaction between this concentration of pseudoazurin and preformed cd 1 -X (above), with respective amplitudes of 0.02 and 0.01. Changes in oxidation state at the c heme were measured by plotting the absorbance at 549 nm minus 539 nm to eliminate the contribution from the d 1 heme. The c heme appeared to rereduce rapidly (see Ref. 18) and then subsequently showed biphasic oxidation (Fig.  6C), the rate constants of which were 14 and 3 s Ϫ1 , with the same relative amplitudes as attributed to the two phases occurring at the d 1 heme. This strong agreement between both the rate constants and amplitudes of the phases occurring at the c and d 1 hemes indicates that the same biphasic process is being observed at both heme centers. The phases of the reaction of fully reduced cytochrome cd 1 with stoichiometric initially oxidized pseudoazurin and 4 mM potassium nitrite could not easily be kinetically resolved. However, it is clear (Fig. 5) that cd 1 -X does not accumulate; at this concentration of pseudoazurin, either it does not form, or its initial rate of decay exceeds its rate of formation. The latter is consistent with the observation above that the reaction between preformed cd 1 -X and stoichiometric oxidized pseudoazurin was lost in the dead time of the stopped-flow apparatus.
The Effect of Zinc-substituted Pseudoazurin-The data presented thus far have shown that initially oxidized pseudoazurin is capable of effecting dissociation of the product, nitric oxide, from cd 1 -X; the enzyme also becomes fully oxidized. Further- more, at sufficient concentration, initially oxidized pseudoazurin prevents accumulation of cd 1 -X during the reaction of fully reduced cytochrome cd 1 with potassium nitrite. These effects may be due to electron transfer processes dependent on pseudoazurin or simply because of protein-protein interaction between pseudoazurin and cytochrome cd 1 . To test whether or not electron transfer is required, the above experiments were repeated using redox-inactive zinc-substituted pseudoazurin (23) in place of normal pseudoazurin, which contains copper. The production of zinc pseudoazurin and its characterization is documented under "Experimental Procedures." Initially, multimixing experiments were carried out in which fully reduced cytochrome cd 1 in the absence of excess reductant was mixed first with 5 mM potassium nitrite in the stopped-flow apparatus, aged for ϳ70 ms, and then mixed rapidly with zinc pseudoazurin at one-tenth the stoichiometry of enzyme monomer. Over the course of 60 s, cd 1 -X was seen to decay; however, the rate of decay was comparable with the rate of photocatalyzed dissociation, indicating that the zinc pseudoazurin had little or no effect on dissociation of cd 1 -X. It was also necessary to determine whether zinc pseudoazurin affects formation of cd 1 -X. Fully reduced cytochrome cd 1 , in the absence of excess reductant, was mixed with a combination of one-tenth stoichiometric zinc pseudoazurin to enzyme monomer and potassium nitrite. Following mixing, formation of cd 1 -X occurred with the same rate constants as in the absence of pseudoazurin, and cd 1 -X did not decay above the rate of pho-tocatalyzed dissociation. Thus, zinc pseudoazurin had no detectable effect on the reaction.

DISCUSSION
The principal observations in this work are summarized in Scheme 1. As previously reported (18), when fully reduced cytochrome cd 1 , in the absence of excess reductant, is mixed with excess nitrite in a stopped-flow apparatus, an initial rapid phase is observed in which the c heme becomes largely oxidized and the d 1 heme forms a complex with two absorbance peaks, one at ϳ660 nm and one at ϳ630 nm (key spectroscopic data are summarized in Table 1). There then follows a second phase in which the 660 nm peak decays with concomitant increase in absorbance at 632 nm, and the c heme is partially rereduced, to yield a species that is ϳ50% reduced at the c heme and has a d 1 heme absorbance maximum at 632 nm. This second phase occurs with a rate constant of 40 s Ϫ1 at pH 7.0. The resulting product was assigned by George et al. (18) as a mixture of cFe(II) d 1 Fe(II)-NO ϩ and cFe(III) d 1 Fe(II)-NO and is referred to in this work as cd 1 -X (Scheme 1). cd 1 -X is stable over hours in the dark, but in the presence of intense light, as in the flow cell of the diode array spectrophotometer, the complex decays over a period of about 1 min. The current work has shown that the photocatalyzed dissociation of cd 1 -X is biphasic. The phases were of equal amplitude, which indicates that dissociation of NO may occur with greater ease from one monomer of the cd 1 dimer than the other. This is consistent with other evidence for inequivalence between the individual monomers of cytochrome cd 1 (10,26).
The data presented here clearly suggest that initially oxidized copper pseudoazurin brings about both the dissociation of NO from cd 1 -X and complete oxidation of the heme centers (Scheme 1). The rate of these processes is dependent on the concentration of pseudoazurin; in the presence of stoichiometric pseudoazurin, it was lost in the dead time of the stoppedflow apparatus and so was fast relative to k cat for the enzyme. The decomposition of preformed cd 1 -X in the presence of onetenth stoichiometric pseudoazurin to enzyme monomer was biphasic (Figs. 1 and 3). The two processes occurring at the d 1 heme can be clearly separated by isosbestic points, enabling the rates of these phases to be determined (rate constants of 24 and 5 s Ϫ1 ). Oxidation of the c heme was also a biphasic process with rate constants of 10 and 4 s Ϫ1 . Thus, it appears that the second phase observed at the c heme is the same as the second phase observed at the d 1 heme; however, the rates of the first phases do not agree. It is possible that this is the effect of electron  (18), referred to throughout this work as cd 1 -X. Formation of cd 1 -X is accelerated by the presence of initially oxidized pseudoazurin (Paz). When cd 1 -X is reacted with initially oxidized pseudoazurin, the product (III) is assigned as fully oxidized cytochrome cd 1 with nitrite bound (cFe(III) d 1 Fe(III)-NO 2 Ϫ ); the rate of reaction depends on [pseudoazurin], taking Ͻ6 ms when pseudoazurin is present at a stoichiometry of 1:1 with d 1 heme. Pseudoazurin with zinc substituted for the usual copper ion does not catalyze this effect above the background rate of photolysis. In the absence of pseudoazurin, over 1 min in the intense light of the diode array spectrophotometer (IV), or hours in the dark (V) (18), cd 1 -X decays to produce products again assigned as cFe(III) d 1 Fe(III)-NO 2 Ϫ (i.e. we presently believe that species III, IV, and V are identical but produced by different means). If fully reduced cytochrome cd 1 (I) is mixed simultaneously with nitrite and initially oxidized pseudoazurin stoichiometric with d 1 heme, the same product forms as when cd 1 -X reacts with pseudoazurin (species III); the data are consistent with models in which either cd 1 -X does not accumulate because its decay is faster than its formation or cd 1 -X formation is bypassed altogether; this uncertainty is shown by the dotted lines in the scheme. exchange between the c heme of cytochrome cd 1 and pseudoazurin. Such electron transfer would be consistent with the relative reduction potentials (19,30).
The observation that pseudoazurin with zinc substituted in place of the usual copper ion does not increase the rate of dissociation of cd 1 -X above the background rate of photolysis implies that electron transfer is essential for pseudoazurin-dependent NO release from cytochrome cd 1 . All analytical techniques used suggested that the zinc pseudoazurin was essentially fully folded. Even if a small amount of unfolded protein had been present, the fact that copper pseudoazurin increases the rate of NO dissociation ϳ100-fold above the rate of photolysis when present in just a one-tenth stoichiometric ratio to enzyme monomer suggests that if a protein-protein interaction alone were the driving force behind NO dissociation from cytochrome cd 1 , zinc pseudoazurin should have enabled NO dissociation even if a small percentage was not correctly folded.
Initially oxidized copper pseudoazurin is able to promote apparently full decomposition of cd 1 -X when present in a 10-fold substoichiometric ratio to enzyme monomer. Therefore, the effect of pseudoazurin on the enzyme is catalytic. The observation that zinc pseudoazurin has little or no effect on dissociation of NO from cd 1 -X indicates a crucial role for the presence of the redox-active metal, copper, in pseudoazurin to bring about product dissociation. These results imply that it is not the addition of electrons to the system per se that is required for dissociation of NO from cd 1 -X but more precisely electron movement. The introduction of pseudoazurin to preformed cd 1 -X must enable shuttling of electrons between monomers or individual enzyme dimers, which results in further turnover, oxidation of the cd 1 , and apparently release of the reaction product, NO, from the d 1 heme. From the available data, it also remains possible, if intuitively unlikely, that copper pseudoazurin (but not zinc pseudoazurin) plays a direct role in releasing NO from the d 1 heme of cytochrome cd 1 (e.g. by a ligand exchange between the heme and copper centers).
As shown in Fig. 2, the spectrum of the product of photocatalyzed decomposition of cd 1 -X is essentially identical to the spectrum following pseudoazurin-dependent decomposition, indicating that the products are likely to be the same. This was the case whether pseudoazurin was stoichiometric with d 1 heme or present at one-tenth of the stoichiometry. The final product of pseudoazurin-dependent decomposition of cd 1 -X is most likely to be the all-ferric enzyme with nitrite bound (Table  1 and Scheme 1). This assignment is based first on good agreement of the pH dependence of the optical absorbance maxima of the dissociated product of cd 1 -X with that of the all-ferric nitrite-bound Y25S cytochrome cd 1 . Further evidence is provided by Allen et al. (25), who reported the spectrum of reduced wild type P. pantotrophus cytochrome cd 1 , reoxidized with hydroxylamine and with nitrite added; the d 1 heme absorbance maximum of the enzyme was at 643 nm, which is also very similar to that of the final product resulting from pseudoazurincatalyzed dissociation of cd 1 -X. d 1 Fe(II)-NO ϩ has its absorption maximum at 632 nm (18); thus, the other plausible assignment of the cd 1 -X dissociation product is all-ferric enzyme with NO bound at the d 1 heme as d 1 Fe(III)-NO. However, as previously discussed, this explanation is not favored, because the pH dependence of absorbance maxima of the cd 1 -X-dissociated spectrum was very different from that of all-ferric Y25S cytochrome cd 1 with NO bound. Furthermore, there is considerable evidence that formation of d 1 Fe(II)-NO ϩ , rather than the isoelectronic state d 1 Fe(III)-NO, is favored in cytochromes cd 1 (18,31). Fe(III)-NO, Fe(II)-NO ϩ , and Fe(III)-NO 2 Ϫ are all EPRsilent complexes of d 1 heme, and therefore, none of these states can be unequivocally demonstrated (or eliminated) using this technique. However, d 1 Fe(II)-NO was ruled out as the final product of pseudoazurin-catalyzed cd 1 -X dissociation, because integration of the d 1 Fe(II)-NO EPR signal from a sample of cd 1 -X treated with pseudoazurin demonstrated that it constituted an insignificant portion of the total enzyme (ϳ1%). A significant presence of d 1 Fe(II)-NO in the final product is also inconsistent with our optical spectra; the absorbance maximum of d 1 Fe(II)-NO is broad but at ϳ631 nm, ϳ10 nm from the absorbance maximum of the final product of pseudoazurinmediated decay of cd 1 -X (Fig. 2). Notably, no d 1 heme signals other than the trace of Fe(II)-NO were observed in the EPR spectrum of cd 1 -X reacted with oxidized pseudoazurin, fully consistent with the final product being the (EPR-silent (25,27)) all-ferric state of the enzyme with nitrite bound.
A very informative result presented here is that obtained from mixing fully reduced cytochrome cd 1 with oxidized pseudoazurin and nitrite in the stopped-flow apparatus. In the presence of stoichiometric pseudoazurin, formation of cd 1 -X occurred at an accelerated rate relative to the absence of pseudoazurin. Subsequent decay of cd 1 -X occurred quickly enough to prevent observation of the discrete increase in absorbance at 632 nm that was seen with substoichiometric pseudoazurin and in the absence of pseudoazurin (Figs. 4 and 5) (although the data are also consistent with cd 1 -X being bypassed altogether at this concentration of pseudoazurin) (Scheme 1). The rate of cd 1 -X formation observed in the presence of even substoichiometric pseudoazurin was ϳ80 s Ϫ1 , putting it on a time scale that is relevant to the rate of steadystate turnover (the k cat of P. pantotrophus cytochrome cd 1 is 72 s Ϫ1 at pH 7.0 (12)). It is thus our current view that species identified following the addition of nitrite to fully reduced cytochrome cd 1 in the absence of electron donor proteins are relevant to the catalytic cycle, but pseudoazurin increases at least some of the rate constants for the formation and decay of these species.
The state of cytochrome cd 1 that normally releases NO is still the subject of much debate. It has been axiomatic that NO is released from oxidized d 1 heme, because "normal" hemes (e.g. b heme in hemoglobin) have a much greater affinity for NO when reduced. However, recent work has shown, unexpectedly, that NO can be released from the ferrous d 1 heme of P. aeruginosa cytochrome cd 1 , in the presence of excess ascorbate, at a rate greater than k cat for the enzyme from that source (32,33). In principle, given two hemes on each monomer of the enzyme, there are five possible overall oxidation states of the cd 1 dimer; these are the all-ferrous form (four-electron reduced), three-, two-, and one-electron reduced dimer, and the fully oxidized (all-ferric) state. In the partially reduced forms, multiple permutations of oxidation state are possible for the c and d 1 hemes (e.g. in the two-electron reduced dimer, the electrons could be distributed between the four hemes in six distinct ways, since the monomers are not absolutely equivalent (10)). It is clear that at least some of the partially reduced states of cytochrome cd 1 do not spontaneously release NO, irrespective of the oxidation state of the d 1 heme (18). This conclusion is reinforced by the current work in which the ability of initially oxidized pseudoazurin to shuttle electrons between cd 1 monomers (or dimers) is apparently an essential part of its role in catalyzing NO release from the enzyme. Our assignment of the final product of reaction between pseudoazurin and cd 1 -X as all-ferric enzyme with nitrite bound implies that fully oxidized cytochrome cd 1 is an effective NO-releasing state. It is conceivable that, for example, cytochrome cd 1 monomers work optimally when cycling between one-electron reduced and fully oxidized states; in the present experiments, the addition of oxidized pseudoazurin might have provided a sink for surplus electrons that was absent in earlier studies, allowing some or all of the cytochrome cd 1 molecules (depending on the pseudoazurin concentration) to maintain their favored oxidation state(s). It will be interesting to learn whether P. pantotrophus cytochrome cd 1 can, like its counterpart from P. aeruginosa (32,33), also release NO from the fully reduced enzyme at a catalytically competent rate.
The present work provides significant new insight into the mechanism of nitrite reduction by P. pantotrophus cytochrome cd 1 and shows that pseudoazurin enables what was previously thought to be a dead end complex of the enzyme (cd 1 -X) to rapidly undergo further turnover and apparently release the final product NO. Remarkably, the presence of pseudoazurin also accelerates formation of intermediates in reduction of nitrite by the enzyme. These effects are facilitated by a combination of electron transfer via the copper of pseudoazurin and protein-protein interaction between pseudoazurin and cytochrome cd 1 . It is very clear that cytochrome cd 1 functions significantly differently in the presence of its two substrates, nitrite and electron donor protein, than in the presence of nitrite alone.