Very Early Reaction Intermediates Detected by Microsecond Time Scale Kinetics of Cytochrome cd1-catalyzed Reduction of Nitrite*

Paracoccus pantotrophus cytochrome cd1 is a nitrite reductase found in the periplasm of many denitrifying bacteria. It catalyzes the reduction of nitrite to nitric oxide during the denitrification part of the biological nitrogen cycle. Previous studies of early millisecond intermediates in the nitrite reduction reaction have shown, by comparison with pH 7.0, that at the optimum pH, approximately pH 6, the earliest intermediates were lost in the dead time of the instrument. Access to early time points (∼100 μs) through use of an ultra-rapid mixing device has identified a spectroscopically novel intermediate, assigned as the Michaelis complex, formed from reaction of fully reduced enzyme with nitrite. Spectroscopic observation of the subsequent transformation of this species has provided data that demand reappraisal of the general belief that the two subunits of the enzyme function independently.

Paracoccus pantotrophus cytochrome cd 1 is a nitrite reductase found in the periplasm of many denitrifying bacteria. It catalyzes the reduction of nitrite to nitric oxide during the denitrification part of the biological nitrogen cycle. Previous studies of early millisecond intermediates in the nitrite reduction reaction have shown, by comparison with pH 7.0, that at the optimum pH, approximately pH 6, the earliest intermediates were lost in the dead time of the instrument. Access to early time points (ϳ100 s) through use of an ultra-rapid mixing device has identified a spectroscopically novel intermediate, assigned as the Michaelis complex, formed from reaction of fully reduced enzyme with nitrite. Spectroscopic observation of the subsequent transformation of this species has provided data that demand reappraisal of the general belief that the two subunits of the enzyme function independently.
Cytochrome cd 1 is a homodimeric enzyme found in the periplasm of denitrifying bacteria such as Paracoccus pantotrophus and Pseudomonas aeruginosa. It catalyzes the one electron reduction of nitrite to nitric oxide, which is the first committed step in the denitrification pathway of the biological nitrogen cycle (1,2). There are two classes of enzyme that catalyze this reaction. The copper nitrite reductases, comprising the first category, contain an electron-accepting type I copper center and a type II copper center at the active site (3). Cytochromes cd 1 belong to a second group of nitrite reductase (1,2). These are heme-containing enzymes with one c heme and one d 1 heme per monomer. The c heme is the site of electron donation from external electron donor proteins; for the enzyme from P. pantotrophus, these have been shown to be pseudoazurin and cytochrome c 550 (4). The d 1 heme forms the active site of the enzyme. The 1.55-Å crystal structure of oxidized cytochrome cd 1 from P. pantotrophus reveals that in its oxidized as-isolated state, the c heme is axially ligated by histidines 69 and 17, and the d 1 heme binds histidine 200 and tyrosine 25 (5). The c heme is located in a predominantly ␣-helical domain of the enzyme, whereas the d 1 heme resides in a ␤-propeller structure. The tyrosine ligand to the d 1 heme is part of the N-terminal c heme domain and is connected to the c heme distal ligand, His-17, by a short polypeptide loop of just 8 amino acids. Upon reduction of the enzyme, His-17 is replaced by Met-106, and Tyr-25 dissociates leaving the active site pentacoordinate and able to bind substrate (5).
Our previous study of nitrite reduction by P. pantotrophus cytochrome cd 1 described the use of stopped flow methodology to study the kinetics of nitrite reduction by P. pantotrophus cytochrome cd 1 (6). At the earliest time point (2-3 ms) using a conventional stopped flow apparatus, at pH 6.0, a significant proportion of the enzyme had already undergone one turnover. At this time point three separate species were assigned at the d 1 heme, Fe(II)-NO, Fe(II)-NO ϩ , and ferrous d 1 heme, with neither substrate nor NO bound (6). The latter, it was postulated, was an intermediate of the enzyme that had released product generated in the first turnover. The findings of this conventional stopped flow study indicated that a faster technique was required to observe intermediates of the first turnover.
A rapid freezing technique known as microsecond freezehyperquenching (MHQ) 2 has been developed in the de Vries laboratory (7,8). This technique allows mixing and freezing of enzyme and substrate on a microsecond time scale, with an effective dead time of 75 s (8), and thus enables previously undetectable intermediates to be observed. We report a completely novel species observed at the d 1 heme just 130 s after mixing of the fully reduced enzyme with nitrite in the absence of excess reductant. The formation of other early intermediates was also observed within the first 800 s of the reaction, and novel insight into their involvement in the catalytic cycle is presented; these intermediates provide evidence that the long standing assumption that the two monomers are independent needs reappraisal.

EXPERIMENTAL PROCEDURES
Production of Cytochrome cd 1 -P. pantotrophus was grown under anaerobic conditions at 37°C. Cytochrome cd 1 was purified from the periplasm of the cells according to the method of Moir et al. (9) as modified by Koppenhöfer et al. (10). 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 of Ͼ1. 25. The concentration of the enzyme was determined at 406 nm for the oxidized enzyme and 418 nm for the reduced, with the respective extinction coefficients of 142.5 mM Ϫ1 (11) and 161.5 mM Ϫ1 (10). These extinction coefficients refer to the concentration of the enzyme monomer. Throughout this work the monomeric concentration will be reported.
Anaerobic Preparation of the Enzyme for Rapid Kinetic Experiments-The purified enzyme was transferred to an anaerobic glove box (Coy Laboratory Products Inc.) that was maintained at less than 2 ppm O 2 . Cytochrome cd 1 was reduced with a small excess of sodium dithionite and then passed down a desalting column packed with P6-DG resin (Bio-Rad) and equilibrated with 50 mM potassium phosphate of the desired pH. All buffers were sparged overnight in the anaerobic glove box to remove oxygen. The enzyme was loaded into a gas-tight syringe that had been presoaked in sodium dithionite and washed to remove traces of excess reductant. The absence of excess dithionite was confirmed by testing buffer expelled from the syringe with methyl viologen, which turns blue on contact with dithionite. 10 mM potassium nitrite was made up in 50 mM potassium phosphate buffer of the desired pH and loaded into a second gas-tight syringe that had undergone the same process as the enzyme-containing syringe to remove traces of oxygen. This was important because cytochrome cd 1 also functions as an oxidase (10).
MHQ-MHQ measurements were performed as described previously (7) and as modified by Ref. 24. Optical measurements were performed on an SLM-Aminco DW2000 scanning spectrophotometer, which was adapted for low temperature measurements of spectra of the frozen powders obtained from the MHQ experiments; the entire MHQ design and set up of the spectrophotometer are described in Ref. 7.

RESULTS
Fully reduced cytochrome cd 1 was mixed with potassium nitrite in the absence of excess reductant in the MHQ apparatus. Because cytochrome cd 1 also functions as an oxidase (10), great care was taken to ensure the equipment was sufficiently anaerobic to prevent re-oxidation of the enzyme, from which excess reductant had been removed, before mixing. To test that this requirement had been met, the enzyme was mixed with anaerobic phosphate buffer in the absence of excess reductant, and the optical spectrum produced was compared with the spectrum recorded when the enzyme was mixed with phosphate buffer in the presence of excess reductant. The position of the c heme Soret band (418 nm), which is characteristic of reduced c heme in this enzyme, and the d 1 heme spectral features at 460 and 653 nm both indicated that the enzyme remained fully reduced in the apparatus, even in the absence of excess reductant ( Fig. 1 and Fig. 3 (t ϭ 0 spectra)).
Fully reduced cytochrome cd 1 was mixed with potassium nitrite in the MHQ apparatus at pH 7.0, and the reaction was quenched at various time points between 130 s and 11 ms after mixing to build up a profile of the early reaction interme-FIGURE 1. Optical spectra measured at the times indicated after mixing of fully reduced P. pantotrophus cytochrome cd 1 (in the absence of excess reductant) and 10 mM potassium nitrite (which was in at least 10-fold excess over enzyme monomer) at pH 7.0. The dashed lines indicate 653 nm, the position of the d 1 heme peak in the fully reduced spectrum. The dotted lines indicate 620 nm. The region between 500 and 700 nm is multiplied by 2.5. Mixing was carried out at 25°C. Spectra were measured at 77 K. diates ( Fig. 1). A spectrum taken at the fastest possible time point (ϳ75 s) was not different from the 130-s spectrum, and therefore we have concentrated our analysis on time points Ն130 s, which can be reported with the greatest accuracy (Ϯ5%). 130 s after mixing of enzyme and substrate, a peak at 620 nm was observed. This peak is shifted by 33 nm from the corresponding d 1 heme absorbance at 653 nm in the starting spectrum of the fully reduced enzyme (Fig. 1). The d 1 heme peak at 460 nm became a shoulder. Complexes of NO and CO bound to the ferrous enzyme from P. aeruginosa lose absorbance at 460 nm altogether (12), and the spectrum of the complex of CN Ϫ bound to ferrous P. pantotrophus cytochrome cd 1 is also significantly perturbed in this region (13). We therefore interpret the loss of absorbance at 460 nm as ligand binding at the active site. At this time point, the c heme was still fully reduced, is judged by the position of the Soret band. The k cat of cytochrome cd 1 is 72 s Ϫ1 per enzyme monomer at pH 7.0 (14), which means that the catalytic cycle takes place over a period of about 14 ms. 130 s was therefore extremely early in the catalytic cycle of cytochrome cd 1 -catalyzed nitrite reduction. An absorbance maximum at around 620 nm has not been observed during conventional studies of the kinetics or ligand binding for this enzyme, and thus it is concluded that the species observed 130 s after mixing of enzyme and substrate was very probably the Michaelis complex of fully reduced cytochrome cd 1 with nitrite bound at the d 1 heme. This assignment is supported by further spectroscopic evidence that will be discussed later. At pH 7.0, between 130 and 780 s, the 620 nm peak began to red shift but was still predominant in the spectrum until 780 s. By 1.31 ms, two separate peaks of roughly equal intensity were visible at 630 and 660 nm. This spectrum is comparable with that seen in the earliest time point, ϳ3 ms, observed in conventional stopped flow studies (6,15). However, presumably because the current spectra were recorded at low temperature, the two separate peaks observed at this time point are much more clearly resolved than in the room temperature spectrum. The 630-nm peak can be assigned with confidence as arising from d 1 Fe(II)-NO ϩ because its formation has previously been observed to occur with the same rate constant as the species shown to be Fe(II)-NO ϩ by FTIR; note that Fe(II)-NO ϩ , while isoelectronic with Fe(III)-NO, is a distinct species, and the previously reported FTIR band observed at 1913 cm Ϫ1 is assigned to the former (15). It has not proved possible to assign unequivocally the 660 nm species to date. However, the best current interpretation is that it arises from a high spin ferrous d 1 heme that is pentacoordinate or more probably has a 6th weak field ligand such as water (6).
The c heme remained essentially reduced for the first 410 s of the reaction and then underwent oxidation between 410 s and 1.31 ms. It then continued to oxidize further but at a much slower rate until the final time point at 11 ms. This second slower phase of oxidation was interpreted as a balance of continued oxidation in conjunction with the back reduction of the c heme shown by George et al. (15). After 11 ms, the c heme had reached ϳ55% oxidation (Fig. 1), which is again in good agreement with the conventional stopped flow studies of the nitrite reduction reaction at this pH (6,15).
Conventionally, the percentage oxidation of P. pantotrophus cytochrome cd 1 is judged by the relative intensity of the c heme ␣-band. However, this requires that spectra are first normalized according to protein concentration. Using the MHQ technique, it was not possible to determine the concentration of the enzyme in the final sample from which the optical spectra were produced because an unknown and slightly variable amount of cold reacted powder was mixed with cold isopentane. The percentage oxidation of the c heme of the enzyme can therefore only be judged by the position of the Soret band. The Soret band of P. pantotrophus cytochrome cd 1 shifts from 410 to 418 nm between its fully oxidized (His/Met coordinated) and fully reduced forms. Assuming a linear relationship between oxidation state and Soret position, the extent of c heme oxidation was estimated from the percentage Soret shift; however, the values obtained are inevitably approximate.
The final species seen after 11 ms at pH 7.0 still contains clear contributions from the 630 and 660 nm species that are also observed in the conventional time stopped flow experiments at this time point. It has been shown previously from later time points (6,15) that this species decays over a period of 100 ms to produce a spectrum containing a single peak in the d 1 heme region with maximum absorbance at 630 nm. Stopped flow FTIR shows that the peak corresponding to d 1 Fe(II)-NO ϩ forms at the same rate as the optical species at 630 nm, and therefore, as previously explained, the 630 nm absorbing species is assigned as containing d 1 Fe(II)-NO ϩ (15).
The mixing of fully reduced cytochrome cd 1 (in the absence of excess reductant) with potassium nitrite was repeated at pH 6.0 because our recent study of the pH dependence of the nitrite reduction reaction on a slower time scale showed that at pH 6.0 the final reaction product was different and formed approximately three times faster compared with pH 7.0. The data in the current ultra-rapid reaction study show that at pH 6.0 a similar pattern of early intermediates to those at pH 7.0 was observed (Fig. 3). However, their lifetimes differed significantly from those at pH 7.0. As with the reaction at pH 7.0, a 620-nm peak was observed in the optical spectrum of the fastest time point (130 s), which as before is assigned as the Michaelis complex of c Fe(II)⅐d 1 Fe(II)-NO 2 Ϫ . At pH 6.0, oxidation of the c heme was similar to pH 7.0 with the majority of this oxidation occurring between 780 s and 1.31 ms (Fig. 2). The most notable difference between pH 6.0 and 7.0 is that at pH 6.0 the species with a peak at 660 nm decayed on a much faster time scale (compare Fig. 1 and Fig. 3). At both pH 6.0 and pH 7.0, a small amount of absorbance at 660 nm was evident in the spectrum at 780 s. By 1.31 ms in the pH 6.0 reaction, which is the time point at which the greatest level of c heme oxidation was observed, clear resolution of a peak at 630 nm, which was assigned as d 1 Fe(II)-NO ϩ , and an increase in intensity of the 660 nm peak occurred. Unlike at pH 7.0, where both the 660 and 630 nm absorbance peaks remained for longer than 11 ms after mixing of enzyme and substrate, at pH 6.0, the 660 nm peak decayed much faster and had substantially disappeared by 2.2 ms, leaving a species generating a single 630 nm peak by 11 ms.

DISCUSSION
This work exemplifies the novel insight into enzyme mechanisms, in this case cytochrome cd 1 , which is available through the use of the recently developed MHQ apparatus. The novel species observed just 130 s after mixing of enzyme and substrate, with an optical d 1 heme signature at 620 nm, is argued to be the Michaelis complex of fully reduced enzyme with nitrite bound at the d 1 heme, cFe(II)⅐d 1 Fe(II)-NO 2 Ϫ . A peak at 620 nm has not previously been observed during the catalytic cycle of cytochrome cd 1 , and indeed there are very few references to d 1 heme complexes with absorbance maxima below 630 nm. The closest reported d 1 heme absorbance maximum to 620 nm is that of the d 1 heme pyridine hemochrome, d 1 Fe(II)-bis-pyr, the absorbance maximum of which has been reported at 617 nm (16) and 620 nm (17). Also of note is that ferrous, cyanidebound cytochrome cd 1 has an absorbance maximum at 628 nm (13,18). Both the pyridine hemochrome and the ferrous CN Ϫbound species are examples of ferrous d 1 heme with strong field ligands. Nitrite is a strong field -acceptor ligand, and the fact that these two complexes of ferrous d 1 heme, with strong field axial ligands, result in observed absorbance maxima below 630 nm supports the proposal that the 620-nm peak seen in this study arises from nitrite bound to ferrous d 1 heme, the Michaelis complex. The time scale on which the 620-nm peak is present (between 130 and 780 s after mixing of enzyme and substrate) is also consistent with it being the Michaelis complex, given the length of the catalytic cycle, which is in the order of 14 ms at pH 7.0 and 8 ms at pH 6.0, as determined by Richter et al. (14). This also makes it unlikely that the first molecule of product NO is formed within the first 130 s. The c heme apparently remains fully reduced at the first time point of 130 s; thus no intramolecular electron transfer has occurred by this stage. Therefore, the only possibilities, other than the Michaelis complex, for the species absorbing at 620 nm, are ferric d 1 heme with either NO or nitrite bound; however, the above discussion of kinetic parameters argues against these possibilities. As reported in previous work (19), model spectra were obtained for the oxidized Y25S cytochrome cd 1 with nitrite and NO bound; it was not possible to use the wild type enzyme for these experiments because tyrosine 25 blocks the active site preventing exogenous ligand coordination to the oxidized enzyme. The Y25S enzyme is competent to exogenous ligand binding in its oxidized as-isolated state, and therefore circumvents this problem. No peak between 550 and 630 nm was seen for the Y25S enzyme with either nitrite or NO bound, further indicating that the 620 nm peak does not arise from either of these ligands bound to ferric d 1 heme. Recent work has invoked the possibility of electron donation from an amino acid side chain during nitrite reduction by cytochrome cd 1 (20), and therefore formation of cFe(II)⅐d 1 Fe(II)-NO was also addressed. However, when the 130-s sample was subjected to EPR, no signal arising from Fe(II)-NO was observed. The only logical assignment of the 620 nm peak, therefore, remains the Michaelis complex. At pH 7.0, following observation of the proposed Michaelis complex at 130 s, the 620 nm absorbance maximum is seen to red shift and flatten. This is concurrent with another species being formed at the d 1 heme, the absorbance of which is not sufficiently separated from 620 nm to distinguish formation of a separate peak. There is considerable evidence for non-equivalence between monomers of P. pantotrophus cytochrome cd 1 (13,21,22), but the general assumption has been that the monomers are kinetically independent. However, it is possible that the residual absorbance at 620 nm, in conjunction with increased absorbance at longer wavelength, reflects nitrite reduction occurring at monomer 1, whereas the Michaelis complex remains at monomer 2. By 1.31 ms, significant absorbance has become apparent at 660 nm, and the absorbance between 620 and 630 nm remains broad and flat although it is further shifted toward 630 nm. By 2.2 ms, the previously apparent absorbance at 620 nm appears to have significantly diminished, with the emergence of a much sharper peak at 630 nm and further increase in the absorbance at 660 nm. In previous work we have tentatively assigned the species with 660 nm absorbance as a high spin ferrous d 1 heme, which is either pentacoordinate or which has a 6th weak field ligand such as a water (6). The reason for the uncertainty of this assignment was because it is hard to justify how the d 1 heme, in the presence of a large excess of nitrite, could have anything bound other than this anion. However, in light of these new insights into much earlier time points in the reaction, several steps in the nitrite reduction reaction may, in fact, be gated by events occurring at the other cytochrome cd 1 monomer. The evidence for this is as follows: as already discussed, the earliest time point is assigned as the Michaelis complex of fully reduced enzyme with nitrite bound at the d 1 heme. Later time points show that this absorbance shifts to longer wavelengths, but no clear decrease in absorbance at 620 nm is observed until formation of the 660 nm species, at which time, clear absorbance at 630 nm is observed with a marked decrease in absorbance at 620 nm. A peak at 630 nm, later on in the reaction between cytochrome cd 1 and nitrite, has been shown to correspond to ferric d 1 heme with NO bound, formally d 1 Fe(II)-NO ϩ . An interpretation of the observed absorbance changes at the d 1 heme is that reduction of nitrite to NO at monomer 2 cannot occur until product has  Figs. 1 and 3) during the reaction of P. pantotrophus cytochrome cd 1 with potassium nitrite, which was in ϳ10-fold excess to enzyme monomer. Enzyme and substrate were mixed in the MHQ apparatus at 25°C, and the reaction was quenched at various time points. The percentage oxidation of the c heme was measured as the % shift between 418 nm (fully reduced His/Met coordinated) and 410 nm (fully oxidized His/Met coordinated). dissociated from monomer 1. Hence, residual absorbance at 620 nm, arising from the Michaelis complex at monomer 2, is observed until product dissociation occurs at monomer 1. If the assignment of the 660 nm species as ferrous pentacoordinate d 1 heme is accurate, then the product dissociated state of monomer 1 is evident from the 660 nm absorbance. As soon as the 660 nm absorbance becomes distinguishable in the spectral time course, the absorbance at 620 nm decreases markedly, and a 630 nm peak is apparent. This is consistent with the Michaelis complex at monomer 2 being converted to ferric d 1 heme with NO bound, as monomer 1 releases NO. As has been shown from later points in the time course of nitrite reduction by this enzyme (6,15), the subsequent disappearance of the 660 nm peak occurs concomitantly with an increase in intensity at 630 nm. Again, this is consistent with the requirement for product formation at monomer 2 before rebinding of substrate is possible at monomer 1. It has been suggested in our previous work that at early time points in the reaction, when the 660 nm peak is observed at its maximum intensity, essentially no FTIR signal for d 1 Fe(II)-NO ϩ is present (6). However, as is explained by George et al. (15), the FTIR signal is indicative of Fe(II)-NO ϩ and not Fe(III)-NO. Fe(II)-NO ϩ is predicted to adopt a linear geometry, whereas Fe(III)-NO is bent. It is therefore possible that in the absence of a driving force to effect product release, formation of Fe(II)-NO ϩ at monomer 1 may force monomer 2 to adopt a different geometry, if non-equivalence of the monomers is a strict requirement. This proposal requires a reconsideration of the cytochrome cd 1 mechanism. Some evidence for non-equivalence between the monomers of cytochrome cd 1 exists. Of particular significance to the current work is that Williams et al. (22) observed nitrite at monomer A and NO at monomer B when crystals of cytochrome cd 1 , which had been pre-reduced with sodium dithionite, were soaked in nitrite. In each crystal structure, monomer B appeared to have progressed further through the catalytic cycle than monomer A. Cooperativity between monomers has also been observed in cytochrome cd 1 from Pseudomonas stutzeri in which internal electron transfer is shown to be allosterically gated (23). No structure is available for the P. stuzeri enzyme; however, the N-terminal region of sequence shows significant difference to the P. pantotrophus enzyme. Our suggestion that catalytic events occurring at each monomer may be gated by the requirement for non-equivalence is supported by the previously described non-equivalence between monomers of P. pantotrophus cytochrome cd 1 . Further support for this view comes from the evidence for cooperativity between monomers within cytochromes cd 1 from other sources.
Analysis of the oxidation state of the c heme during the early phases of nitrite reduction also provides new mechanistic insight. At pH 7.0, the c heme remains essentially fully reduced until 410 s (Figs. 1 and 2). Between 410 and 780 s, the c heme substantially oxidizes, and a small amount of absorbance becomes visible at 660 nm. By 1.31 ms, the c heme has reached about 50% oxidation, and the two separate peaks are visible at FIGURE 3. Optical spectra measured at the times indicated after mixing of fully reduced P. pantotrophus cytochrome cd 1 (in the absence of excess reductant) and 10 mM potassium nitrite (which was in at least 10-fold excess over enzyme monomer) at pH 6.0. The dashed lines indicate 653 nm, the position of the d 1 heme peak in the fully reduced spectrum. The dotted lines indicate 620 nm. The region between 500 and 700 nm is multiplied by 2.5. Mixing was carried out at 25°C. Spectra were measured at 77 K. OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41