Y25S variant of Paracoccus pantotrophus cytochrome cd1 provides insight into anion binding by d1 heme and a rare example of a critical difference between solution and crystal structures.

Tyr25 is a ligand to the active site d1 heme in as isolated, oxidized cytochrome cd1 nitrite reductase from Paracoccus pantotrophus. This form of the enzyme requires reductive activation, a process that involves not only displacement of Tyr25 from the d1 heme but also switching of the ligands at the c heme from bis-histidinyl to His/Met. A Y25S variant retains this bis-histidinyl coordination in the crystal of the oxidized state that has sulfate bound to the d1 heme iron. This Y25S form of the enzyme does not require reductive activation, an observation previously interpreted as meaning that the presence of the phenolate oxygen of Tyr25 is the critical determinant of the requirement for activation. This interpretation now needs re-evaluation because, unexpectedly, the oxidized as prepared Y25S protein, unlike the wild type, has different heme iron ligands in solution at room temperature, as judged by magnetic circular dichroism and electron spin resonance spectroscopies, than in the crystal. In addition, the binding of nitrite and cyanide to oxidized Y25S cytochrome cd1 is markedly different from the wild type enzyme, thus providing insight into the affinity of the oxidized d1 heme ring for anions in the absence of the steric barrier presented by Tyr25.

There are two strikingly different forms of bacterial nitrite reductase used in denitrification to catalyze the one-electron reduction of nitrite (NO 2 Ϫ ) to nitric oxide (NO) and water, namely, cytochrome cd 1 (NirS) and the copper-containing enzyme NirK. Cytochrome cd 1 is a soluble tetraheme periplasmic homodimer. Each monomer contains a covalently bound c-type heme and a non-covalently bound d 1 heme (a cofactor unique to this class of enzyme). The c heme is the site of electron entry, whereas the d 1 heme binds and reduces nitrite. In addition, each monomer is divided into two distinct domains. The Nterminal domain is predominantly ␣-helical and binds the c heme, whereas the C-terminal domain forms an eight-bladed ␤-propeller in which the d 1 heme sits (1). The crystal structure of the oxidized "as isolated" Paracoccus pantotrophus cytochrome cd 1 nitrite reductase had unexpected ligands at each of the heme centers (1). The c heme centers on each monomer are bis-histidinyl, coordinated by His 69 (belonging to the CXXCH motif) and His 17 . The d 1 heme has His 200 as an axial ligand along with Tyr 25 ; the latter is from the c heme domain on the same extended N-terminal polypeptide loop as His 17 . The possibility that the axial ligands His 17 and Tyr 25 were an artifact of crystallization was eliminated by spectroscopic (magnetic circular dichroism (MCD) 1 and electronic paramagnetic resonance (EPR)) studies in solution; these unambiguously confirmed the bis-histidinyl coordination of the c heme and were entirely consistent with the presence of tyrosine at the d 1 heme (2). Upon reduction, the crystal structure demonstrated a switch in the heme ligands of both heme centers. The His 17 ligand to the oxidized c heme was displaced by Met 106 , and there was loss of the Tyr 25 ligand to the d 1 heme, leaving the latter pentacoordinate and available for NO 2 Ϫ binding (3). Subsequent to the initial spectroscopic and crystallographic characterization of the enzyme, it was discovered that significant catalytic activity of P. pantotrophus cytochrome cd 1 required activation by reduction (4). This could be understood in terms of the bis-histidinyl ligands to the oxidized c heme resulting in this center having an E°Ј of approximately ϩ60 mV. The latter value is not obviously compatible with acceptance of electrons from the physiological electron donors, cytochrome c 550 and pseudoazurin, both of which have E°Ј values of approximately ϩ250 mV. In addition, Tyr 25 would present a steric barrier against nitrite binding to the oxidized enzyme. An oxidized form of the enzyme with His/Met coordination at the c heme and absence of Tyr 25 binding at the d 1 heme can be generated (5,6). This conformation of cytochrome cd 1 is catalytically active, not least because of the His/Met coordinated c heme center with an estimated E°Ј value of ϳ250 mV (7), very similar to those of either of the two physiological electron donor proteins; also, the d 1 heme is available to bind substrate in this conformation (6). These observations cast doubt on the proposal that binding and dissociation of Tyr 25 and the bis-histidinyl coordination at the c heme played roles in the catalytic activity of the protein (1,3). On the other hand, the finding that Tyr 25 readily displaces cyanide from the d 1 heme upon oxidation of the enzyme in the crystal or in solution provided further evidence for the strong tendency for this tyrosine to bind to the d 1 heme iron (8).
Tyr 25 was subsequently mutated to a serine residue (9) to probe the consequences of the loss of this tyrosine ligand on protein conformation and activity. The visible absorption spectrum of the Y25S variant cytochrome cd 1 showed the loss of all peaks associated with signal from a high spin d 1 heme species. This was to be expected, following the loss of Tyr 25 , which puts the d 1 heme in a state of high/low spin thermal equilibrium (2). Upon removal of Tyr 25 , there was only a signal associated with a low spin d 1 heme species (2,9). A 1.4 Å structure of this variant protein showed only very localized changes in the immediate vicinity of residue 25 (9). The protein still had bishistidinyl coordination at the c heme, and the serine had effectively replaced Tyr 25 in the d 1 heme pocket, but a sulfate ion (presumably from the precipitant) was bound to the d 1 heme rather than the ϪOH of Ser 25 . Kinetic assays revealed this enzyme did not require pre-reduction for maximal substrate turnover, unlike wild type cytochrome cd 1 (4,9), possibly as a result of the increased accessibility of the d 1 heme permitting nitrite to bind and subsequently raise the reduction potential of that heme (9). The latter might allow electron transfer from a donor protein even with an energetically uphill initial step due to His/His coordination of the c heme (9,10). It was assumed that onset of catalytic turnover could also trigger the switching of the c-type center to His/Met coordination. An alternative possibility, that the Y25S enzyme had different heme ligands in solution (e.g. His/Met at the c heme center) compared with the crystal, was deemed very improbable because, as discussed above, the wild type enzyme retained the same heme ligands in the crystal and solution for the oxidized as isolated state. Furthermore, the Y25S protein retained essentially all those structural features in the crystal that can be recognized to stabilize the binding of His 17 to the c-type center (9). However, additional spectroscopic studies of the Y25S protein have, as described in the present work, unexpectedly undermined this conclusion.
Because of Tyr 25 binding to the d 1 heme in the oxidized wild type enzyme, studies of cyanide binding to the d 1 heme of P. pantotrophus cytochrome cd 1 have only been possible in the reduced form (8). Oxidation of the reduced cyanide bound complex results in the displacement of CN Ϫ by Tyr 25 . It was thought that P. pantotrophus Y25S cytochrome cd 1 might allow the study of cyanide and nitrite binding to ferric d 1 heme without interference from Tyr 25 , thus providing insight into the relative affinities of the ferric and ferrous forms of d 1 heme for these two anions. Nitrite will bind only to the activated form of oxidized wild type cytochrome cd 1 (6); the binding affinity of nitrite to ferrous d 1 heme could not be determined because the physiological catalytic reaction would reduce nitrite to nitric oxide and water.

MATERIALS AND METHODS
Protein Purification-The P. pantotrophus Y25S cytochrome cd 1 variant was expressed from P. pantotrophus strain EG6202 complemented with a Y25S nirS expression vector, pEG760 (9). Cells were grown for 16 -24 h on minimal media supplemented with 20 mM succinate and 20 mM nitrate under anaerobic conditions to ensure maximum d 1 heme production and incorporation into the protein (9). A standard P. pantotrophus cytochrome cd 1 protocol was employed to obtain the periplasmic fraction and purify Y25S cytochrome cd 1 from the cell pellets (9,11). Concentrations of oxidized Y25S cytochrome cd 1 were assayed spectroscopically using the absorbance at 410 nm (⑀ ϭ 255,000 M Ϫ1 cm Ϫ1 ). P. pantotrophus pseudoazurin was purified from Escherichia coli XL1-Blue (Stratagene) that contained the plasmid pJR2 (12). Concentrations of oxidized P. pantotrophus pseudoazurin were determined at 590 nm (⑀ ϭ 1360 M Ϫ1 cm Ϫ1 ).
Spectroscopy-EPR spectra were recorded on an X-band ER-200D spectrometer (Brü ker spectrospin) interfaced to a computer and fitted with a liquid helium flow cryostat (ESR-9; Oxford Instruments, Oxford, UK). MCD spectra were recorded on either a circular dichrograph (JASCO-J-500D) for the wavelength range 280 -1000 nm or a laboratory-built dichrograph for the range 800 -2500 nm. Samples were mounted within an Oxford Instruments SM4 split-coil super-conducting solenoid capable of generating magnetic fields of up to 5 T for low temperature measurements and in an Oxford Instruments SM1 6 tesla superconducting solenoid with an ambient temperature bore for room temperature MCD measurements. All solutions used in this study were dissolved in deuterated 50 mM potassium phosphate buffer at the pH* values specified in the text (pH* is the apparent pH of D 2 O solutions measured using a standard glass pH electrode). Samples for MCD included deuterated glycerol (C 3 H 5 O 3 D 3 ) in a 1:1 (v/v) ratio, as a glassing agent. Electronic absorption spectra were recorded on a PerkinElmer Life Sciences Lambda-2 spectrophotometer or a Hitachi instrument.
Titrations with Anionic Ligands-Equilibrium binding parameters of cyanide were determined by titrating potassium cyanide into an air-tight cuvette containing a solution of oxidized P. pantotrophus Y25S cytochrome cd 1 essentially as described in Jafferji et al. (8). Using an excess of potassium cyanide, it was shown that only the d 1 heme significantly bound cyanide at the range of concentrations of cyanide used in the titration. On addition of cyanide to the cuvette, using a Hamilton syringe, the solution was allowed to equilibrate, and then the cyanide binding affinity was monitored spectroscopically using the absorption at 672 nm. Note that the visible absorption spectrum of the Y25S cytochrome cd 1 variant is different from that of the wild type, and therefore it was not appropriate to use the absorption at 632 nm as described in Jafferji et al. (8). The same method was used when titrating dithionite-reduced Y25S cytochrome cd 1 that had been passed down a PD10 desalting column (Amersham Biosciences). The equilibrium binding parameters of nitrite were measured in the same fashion using potassium nitrite. Fig. 1 shows the oxidized and reduced visible absorption spectra of Y25S P. pantotrophus cytochrome cd 1 . The Söret band of the Y25S cytochrome cd 1 spectrum is red shifted from 406 nm for the wild type to 410 nm, a value consistent with, but not reliably diagnostic for, His/Met coordination of the c heme as in other cytochromes cd 1 (2,13,14), including activated P. pantotrophus cytochrome cd 1 (6,15). As seen in Gordon et al. (9), the Y25S spectrum has a single broad peak above 600 nm at ϳ640 nm; this is consistent with the disappearance of the high/low spin thermal equilibrium at the d 1 heme caused, in the wild type enzyme, by Tyr 25 coordination (2), leaving only a low spin species. The spectra of reduced P. pantotrophus wild type and Y25S cytochrome cd 1 are identical, leading to the assumption they have the same structure in the vicinity of the absorbing cofactors. The absorbance maximum at ϳ 650 nm, in the reduced spectra of both proteins, allows direct comparison of d 1 heme content with the wild type protein. Y25S consistently contains less d 1 heme per dimer. Exact amounts varied from preparation to preparation but were never Ͻ80% that of the wild type.

Visible Absorption Spectroscopy-
Room Temperature MCD Spectroscopy-The ultraviolet and visible region room temperature MCD spectrum of oxidized Y25S cytochrome cd 1 is shown in Fig. 2. As with all cytochromes cd 1 , the signal between 300 and 580 nm is typical of low spin ferric protoheme (16); contributions in this region from ferric d 1 hemes are significantly weaker than those of the ferric c hemes (17). In this case, the spectrum is identical to that of Pseudomonas stutzeri cytochrome cd 1 at room temperature (2), a protein with His/Met coordination at the c heme. Fig. 3 shows the room temperature near-infrared MCD spectrum of oxidized "as isolated" wild type and Y25S cytochrome cd 1 at room temperature in the region between 700 and 2000 nm. The wild type protein has a predominant peak at 1530 nm, whereas the Y25S protein has this peak red shifted to 1775 nm. Both these peaks have higher energy shoulders. The peaks observed for each protein are consistent with the near-infrared charge transfer bands associated with low spin ferric protohemes (i.e. the c heme of cytochrome cd 1 ) and can be used as a diagnostic tool to assign heme ligands in b/c-type cytochromes (16). The Y25S cytochrome cd 1 band at 1775 nm lies in the region indicative of His/Met coordination, whereas the wild type cytochrome cd 1 band at 1530 nm lies in the region diagnostic of His/His coordination. These results indicate that, at room temperature, the c heme of Y25S cytochrome cd 1 has His/Met coordination in solution, differing from the crystal structure (9). The near-infrared MCD spectrum of P. pantotrophus Y25S cytochrome cd 1 is identical to that of the wild type P. stutzeri (2) and Pseudomonas aeruginosa (14) cytochromes cd 1 , both of which have His/Met axial ligands to their c hemes. Fig. 3 shows a significant broad signal at ϳ1000 nm in the spectrum of the wild type P. pantotrophus cytochrome cd 1 that is not present in the Y25S protein. This feature has been attributed to the high spin ferric d 1 heme that is present in thermal equilibrium with the low spin species at room temperature because of Tyr 25 coordination in wild type P. pantotrophus cytochrome cd 1 (2).
EPR Spectroscopy-The X-band EPR spectrum at 10 K of P. pantotrophus Y25S cytochrome cd 1 is shown in Fig. 4; there are multiple features. For P. pantotrophus cytochrome cd 1 , the three features at g ϭ 2.93, 2.34, and ϳ1.4 can be attributed to a low spin ferric c heme with His/Met axial ligands (5,15). The semi-apo form of cytochrome cd 1 (with the d 1 heme removed) has near identical g values; this type of cytochrome cd 1 has been described as having His/Met ligands to the c heme (15). Furthermore, an oxidized holo form of wild type cytochrome cd 1 , generated by oxidation of the reduced enzyme by hydroxylamine, displaying a g y value of 2.33, was shown to have His/Met ligation at the c heme (5,18). Additionally, these features are also very similar to those seen in the X-band EPR spectra of P. stutzeri and P. aeruginosa cytochromes cd 1 , both of which have His/Met coordination at their c hemes (2). In previous work, it has been shown that the c-type cytochrome center of wild type P. pantotrophus cytochrome cd 1 has a relatively unusual EPR spectrum of the "large g max type," in which g z is Ͼ3, and the other two g values are not easily detected (2,15). This feature is thought to be due to bis-histidinyl axial ligation with the orientation of the imidazoles closer to perpendicular than parallel (19), as is observed in the crystal structure of wild type P. pantotrophus cytochrome cd 1 (1). There is no corresponding feature in the spectrum of P. pantotrophus Y25S cytochrome cd 1 . These EPR observations are absolutely consistent with Y25S cytochrome cd 1 having His/Met axial ligands at low temperature, but without low temperature MCD it cannot be said so categorically. It remains possible, but unlikely, that the c heme of Y25S cytochrome cd 1 may switch from His/Met at room temperature to His/His at low temperature, due to a freezing artifact or thermal equilibrium. Note that the x-ray crystal structure of Y25S cytochrome cd 1 was determined at liquid N 2 temperature (9).
The remaining EPR features at g ϭ 2.57, 2.44, 2.25, 1.87, and 1.61 (Fig. 4) are consistent with previously observed characteristics of low spin ferric d 1 heme and can be assigned to a rhombic species with g ϭ 2.57, 2.25, and 1.87 together with an axial species with g ϭ 2.44, 2.44, and 1.61 (2,15). These g values are thought to represent His/OH Ϫ coordination at the d 1 heme, which is highly likely in oxidized Y25S cytochrome cd 1 because the d 1 heme is exposed. It is clear that there are two different species of low spin ferric d 1 heme possibly sharing a g value of 2.44. These two species could be caused by two different orientations of OH Ϫ .
Features at g ϭ 6.95 and 4.99 are characteristic of rhombically distorted high spin ferric d 1 heme present in trace amounts. These g values have been seen in varying amounts in all wild-type P. pantotrophus cytochrome cd 1 preparations (2,5).
Low Temperature MCD Spectroscopy-It was clear from the experiments described above that at room temperature the ligands to the oxidized c heme were His/Met in solution for the Y25S protein. In addition, the evidence from the EPR measurements also strongly suggested that the oxidized c heme ligands were His/Met. However, although the room temperature near-infrared MCD showed that the coordination was His/Met, the EPR cannot be taken as certain evidence for retention of this coordination at the much lower temperature used with the latter technique. Consequently, low temperature near-infrared MCD analysis was undertaken, using two pH values such that any influence of proton concentration on the c-type cytochrome center could be assessed at the same time. The visible low temperature MCD spectra of samples at pH 6.5 (a pH close to physiological pH and that (pH 5.8) which produced maximal activity (4)) and pH 8.5 are shown in Fig. 5. In both spectra it is clear there is no signal above 700 nm. Cheesman et al. (2) showed that any signal of Ͼ700 nm is due to high spin ferric d 1 heme that is in thermal equilibrium with low spin ferric d 1 heme in wild type P. pantotrophus cytochrome cd 1 at room temperature. This observation confirms that the previously mentioned EPR features of the possibly two species of d 1 heme are both from low spin hemes. The low temperature near-infrared MCD spectra of Y25S cytochrome cd 1 at pH 6.5 and pH 8.5 are shown in Fig. 6. It is clear that the sample at pH 8.5 is close to 100% His/Met, based on the presence of a band at 1775 nm and the absence of a band at 1530 nm. However, the sample at pH 6.5 has a heterogeneous spectrum, with bands characteristic of both His/Met (75%) and His/His (25%). The absence of a "large g max " species in the EPR spectrum (pH 7.0) signifies (unless there is dramatic difference in the amount of this form at pH 7.0 compared with pH 6.5) that this species of bis-histidinyl-coordinated c heme does not have perpendicular imidizole rings, thus implying that they are in a parallel conformation. The crystal structure of P. pantrotrophus Y25S cytochrome cd 1 shows the histidine residues are perpendicular to each other. This highlights another difference between the solution and crystal forms of the protein.
The difference between the low temperature near-infrared MCD spectra at pH values of 6.5 and 8.5 prompted further examination of the visible absorption spectra under the same conditions. In each case the Söret band maximum was at 410 nm (as in Fig. 1), rather than at 406 nm (as for the wild type protein). Although this wavelength cannot be used in a general sense to assess coordination at heme, the fact that there is no difference between pH 6.5 and pH 8.5 indicates that for Y25S cytochrome cd 1 , there is an insignificant increase in His/His coordination on going from pH 8.5 to pH 6.5 in solution at room temperature. Thus, on the basis of the data described in the present study, it seems that the His/His coordination is favored by low temperature and lower pH values.
Reduction with Physiological Electron Donor-Given that the ligands to the c heme of oxidized Y25S cytochrome cd 1 in solution had changed from those in wild type cytochrome cd 1 , it was important to investigate whether the in vivo electron donors to cytochrome cd 1 (4, 20) could, unlike the wild-type (4), reduce Y25S cytochrome cd 1 before substrate binding. To determine whether reduced pseudoazurin could pass electrons to oxidized Y25S cytochrome cd 1 , an excess of reduced pseudoazurin (from which the excess reductant, ascorbate, had been removed by gel filtration) was mixed with oxidized Y25S cytochrome cd 1 under anaerobic conditions and in the absence of nitrite. Pseudoazurin was used because its absorbance is small at the relevant wavelengths in comparison with those of the alternative electron donor, cytochrome c 550 . Fig. 7 shows the resultant spectrum of the mixture 30 s after mixing. The appearance and magnitude of the split ferrous c heme ␣-band at ϳ550 nm indicates that the Y25S cytochrome cd 1 has been completely reduced by the pseudoazurin. The peak at 460 nm, characteristic of reduced high spin d 1 heme, also illustrates the change in the oxidation state of cytochrome cd 1 . The peak assigned to reduced d 1 heme at 650 nm is difficult to identify  because of the broad pseudoazurin absorbance at ϳ590 nm.
Cyanide and Nitrite Binding Titration-Oxidized "as isolated" wild type P. pantotrophus cytochrome cd 1 is unable to bind CN Ϫ at either the c or d 1 hemes, but the reduced enzyme binds the ligand at the d 1 heme (K d ϭ 0.7 ϫ 10 Ϫ6 M) (8). It was thought, in the absence of Tyr 25 , that oxidized P. pantotrophus Y25S cytochrome cd 1 would be able to bind CN Ϫ . The enzyme was exposed to 10 mM KCN to assess whether it could bind the CN Ϫ anion. In contrast to the wild type protein, oxidized Y25S cytochrome cd 1 was able to bind CN Ϫ . The d 1 heme ␣-band at 640 nm in the oxidized "as isolated" P. pantotrophus Y25S cytochrome cd 1 shifted to 631 nm upon addition of cyanide. The position of this d 1 heme band is the same as that for oxidized wild type P. aeruginosa cytochrome cd 1 with cyanide bound (14). The visible absorption spectrum of reduced Y25S cytochrome cd 1 with cyanide bound was identical to that of reduced wild type protein with cyanide bound (8). Upon oxidation with K 3 [Fe(CN) 6 ], the reduced d 1 heme peak of Y25S cytochrome cd 1 remained at 631 nm rather than returning to 640 nm (Y25S cytochrome cd 1 oxidized peak), which would have been expected if CN Ϫ had been displaced from the d 1 heme by Ser 25 ; cyanide displacement by Tyr 25 is seen in wild type cytochrome cd 1 (8).
Potassium cyanide was titrated into a solution of oxidized "as isolated" P. pantotrophus Y25S cytochrome cd 1 at 25°C to determine equilibrium binding parameters. The dissociation constant (K d ) was found to be 4.4 ϫ 10 Ϫ5 M at pH 7.0 using a Hill plot; this value is an order of magnitude larger than that of the reduced wild type cytochrome cd 1 (ϳ1 ϫ 10 Ϫ6 M) (8). It is noteworthy that only the d 1 heme was affected by the addition of cyanide over this titration range. In addition, a titration of cyanide against reduced Y25S cytochrome cd 1 was also performed. This showed that the reduced Y25S protein-cyanide complex had a similar dissociation constant to the reduced wild type protein (in each case, ϳ1 ϫ 10 Ϫ6 M at pH 7.0).
The "as isolated" oxidized form of wild type P. pantotrophus cytochrome cd 1 will not bind nitrite due to the steric barrier presented by Tyr 25 . However, if the alternative active form (i.e. with His/Met coordination at the c heme) of the oxidized enzyme is prepared, then substrate addition of nitrite results in a stable oxidized enzyme with nitrite bound; the effective K d is 2 mM (6). It was therefore of interest to test whether the oxidized Y25S protein would also bind nitrite. Using the same technique for cyanide binding, nitrite was titrated into a solution of oxidized Y25S cytochrome cd 1 . From this experiment, it can be calculated that the K d for nitrite dissociation from the enzyme is ϳ7 ϫ 10 Ϫ5 M.
In previous work, the activated ferric form of wild type enzyme was, when complexed with nitrite, unexpectedly EPR silent with respect to the d 1 heme (6). The absence of an EPR spectrum was also seen in the present work for the nitrite complex with oxidized Y25S cytochrome cd 1 . The explanation for this EPR silence is not obvious and requires further study, although the present results reinforce the idea that the observation reflects an intrinsic feature of the ferric d 1 hemenitrite complex. DISCUSSION MCD and EPR spectroscopies categorically identify the axial ligands at the c heme of P. pantotrophus Y25S cytochrome cd 1 as His/Met in solution at room temperature and the majority population at two different pH values at very low temperature. It is reasonable to assume the methionine residue is Met 106 , an axial ligand to the reduced wild type cytochrome cd 1 c heme (3). This conserved residue is also an axial ligand to the oxidized c heme of P. aeruginosa cytochrome cd 1 (21). This result is in clear disagreement with the 1.4 Å crystal structure of oxidized Y25S cytochrome cd 1 that showed bis-histidinyl axial ligation of the c heme (9). The reasons for, and implications of, these results are discussed below.
There have been many discussions since the advent of protein crystallography as to whether solution and crystal structures are the same. In general they are. This study, therefore, reports a relatively rare case of a protein that is crucially different in structure between the solution and crystalline states. As with most proteins, the crystallization conditions (here including 2.3 M ammonium sulfate and high protein concentration) are rather different from those used for functional studies in solution. Thus we cannot distinguish between high salt/protein concentration or formation of crystals per se as being the determining factor. However, high protein concentration may not be a major factor because our spectroscopic studies were carried out at enzyme concentrations (120 M) comparable to those used for crystallization (22).
There is no doubt that the solution form of oxidized Y25S cytochrome cd 1 does not have bis-histidinyl coordination of the c heme; rather, His/Met coordination at the c heme is seen. The loss of essentially just one interaction between the phenolate oxygen of Tyr 25 and the d 1 heme iron appears to alter the protein folding energy sufficiently such that in the solution conditions used for functional studies, the position of the N-terminal arm and, consequently, the appropriate position of His 17 as a heme ligand for the c-type cytochrome center, are lost. There are, for crystalline wild type cytochrome cd 1 , 19 (subunit A) or 20 (subunit B) direct hydrogen bonds and salt bridges, plus the Tyr 25 to d 1 heme interaction, between the N-terminal domain (up to residue 135 which comprises the N-terminal arm and the cytochrome c domain) and the d 1 (C-terminal) domain (1,23). Of these interactions, 10 and 11, respectively, are between the N-terminal arm (residues 9 -48) and the d 1 domain. Thus, loss of 1 interaction, Try 25 to d 1 , of 19 or 20 for each subunit is sufficient to switch the Y25S enzyme from the bis-histidinyl coordination of the c heme to the His/Met state on going from the crystalline to the solution conditions used in the present work. The coordination of the hemes in this enzyme is evidently very delicately balanced. The bis-histidinyl coordination in the Y25S cytochrome cd 1 crystal could be a consequence of the crystallization conditions and/or the phase change to the crystalline state. However, both low temperature and pH also promote this conformation. This view is supported by the low temperature MCD at pH 6.5 and pH 8.5. At pH 6.5, there is ϳ25% of the protein in the bis-histidinyl coordination; a figure that appears to decrease upon an increase in pH. Therefore, if this effect was predominantly temperature-dependent, the crystal structure would be expected to be His/ Met in the majority form with Ͻ25% having bis-histidinyl coordination because the structure was determined at pH 7.0 (9).
The structures of reduced wild type cytochrome cd 1 crystals soaked in nitrite have been solved (3). Nitrite or nitric oxide was bound to the d 1 heme, and His/His coordination of the c heme was observed. In contrast, rapid reaction solution studies, reacting reduced wild type cytochrome cd 1 with nitrite, showed no evidence of His/His coordination of the c heme but only His/Met coordination (24). It is possible, in light of the results presented here, that the same switching of ligands occurred in the nitrite-soaked crystals as proposed here for the Y25S cytochrome cd 1 crystals.
Pseudoazurin is unable to reduce oxidized "as isolated" wild type cytochrome cd 1 (4,5), whereas the results in the present study show that pseudoazurin can reduce oxidized Y25S cytochrome cd 1 . This result is in good agreement with the reduction potentials of the hemes in each enzyme. In the wild type "as isolated" enzyme, electrons must enter the enzyme via a His/His-coordinated c heme (1, 2) with a midpoint reduction potential of ϩ60 mV (7). Hence, pseudoazurin cannot act as a stochiometric reductant for the c-type center because it has a higher reduction potential of ϩ230 mV (11) that makes electron transfer thermodynamically unfavorable. The results presented now show that solution state oxidized Y25S cytochrome cd 1 has His/Met coordination, a change that is thought to raise the redox potential of a c-type heme by ϳ200 mV (25). A reduction potential increased by about this amount would favor electron transfer from pseudoazurin to Y25S cytochrome cd 1 . The possibility that the d 1 heme of Y25S cytochrome cd 1 may have a much higher reduction potential than pseudoazurin, making electron flow from pseudoazurin to the d 1 heme sufficiently thermodynamically favorable to overcome an uphill step from pseudoazurin to the cytochrome c heme of the enzyme (9), is not supported by the observation that the c heme became extensively reduced in the presence of reduced pseudoazurin.
Wild type cytochrome cd 1 requires preactivation to initiate maximal catalytic turnover (4). This is because reduction of the protein induces a conformational change to give the c heme His/Met coordination and make the d 1 heme pentacoordinate and ready to bind substrate (3). It is the binding of substrate to the d 1 heme that is thought to provide the thermodynamic driving force for the reaction (4). The model proposed in Gordon et al. (9), which is based on His/His coordination of the c heme in the Y25S cytochrome cd 1 crystal structure, suggests that, in the absence of Tyr 25 , nitrite is able to bind to the d 1 heme of the oxidized enzyme. The binding of nitrite would then raise the reduction potential of the d 1 heme to allow electron transfer from pseudoazurin by electron tunnelling (10). Thus reductive preactivation would not have been required for the Y25S variant (9). The results presented now demonstrate that the c heme is His/Met in solution (see Ref. 5 for diagrammatic representation), thus explaining why preactivation was not required and raising the possibility of reduction by physiological electron donor before binding of substrate, rather than substrate binding to the oxidized d 1 heme. It is shown that pseudoazurin can reduce oxidized Y25S cytochrome cd 1 ; therefore, substrate most likely binds to a reduced d 1 heme, which is the assumption in the wild type enzyme (3,26). In the case of Y25S cytochrome cd 1 , it is unlikely that nitrite must bind to oxidized d 1 heme to give the chain of redox centers its overall driving force. This observation is also in good agreement with the cyanide binding titration, which shows that cyanide (a model anion) has a higher affinity for ferrous, relative to ferric, d 1 heme by at least 1 order of magnitude. Nevertheless, it remains to be seen in future work whether cytochrome cd 1 has an obligatory mechanistic pathway in which nitrite binds to one oxidation state of the d 1 heme.
The c heme Söret band of oxidized "as isolated" wild type cytochrome cd 1 lies at 406 nm, whereas the Y25S band lies at 410 nm. This would not usually be cause to doubt a 1.4 Å crystal structure, although the Y25S cytochrome cd 1 crystal structure surprisingly showed no difference in c heme ligation or ligand orientation from the wild type. The visible absorbance c heme Söret bands of the oxidized wild type cytochromes cd 1 from P. aeruginosa and P. stutzeri both lie at 411 nm (2,27). It can now be proposed that an oxidized c heme Söret band of a cytochrome cd 1 at 410 -411 nm is a strong indication of His/ Met coordination at the c heme for cytochrome cd 1 . When oxidized P. aeruginosa cytochrome cd 1 was mixed with imidazole, the c heme Söret band shifted from 411 to 406 nm (28), reflecting the displacement of methionine by imidazoles; this extends the proposal to include a 406 nm c heme Söret band as an indication of His/His coordination. Erythrobacter sp. Och114 contains two different forms of cytochrome cd 1 , with distinct oxidized c heme Söret bands at 406 and 410 nm (29). It was suggested at the time of publication that these were two individual proteins; it is possible that they are just two different forms of the same protein, one with His/His coordination and the other with His/Met coordination. This hypothesis is reflected in the d 1 heme region of the spectra. A split d 1 heme ␣-band is observed in the 406 nm c heme Söret band form, a feature indicative of tyrosine coordination to the d 1 heme and consequentially His/His coordination at the c heme.
Y25S cytochrome cd 1 gives us the opportunity to study, for the first time, the binding properties of ferric d 1 heme in situ without the effects of Tyr 25 as a ligand. It is reasonable to assume Ser 25 is not obstructing entry to the d 1 heme iron because no structures even of wild type cytochrome cd 1 with His/Met coordination to the c heme have described Tyr 25 bound to the d 1 heme (3,8,30). In addition, the observed K d , of the order 10 Ϫ5 M, for cyanide dissociation from oxidized Y25S cytochrome cd 1 would imply little or no hindrance from Ser 25 . The equilibrium dissociation constant for cyanide binding to ferric d 1 heme of Y25S cytochrome cd 1 (K d ϭ 4.4 ϫ 10 Ϫ5 M) is more than an order of magnitude greater than that for cyanide dissociation from ferrous d 1 heme in wild type cytochrome cd 1 (ϳ1 ϫ 10 Ϫ6 M) (8). The result with Y25S cytochrome cd 1 is in excellent agreement (perhaps coincidentally) with the K d for cyanide dissociation from ferric d 1 heme when the latter was inserted into the heme binding pocket of myoglobin (4.2 ϫ 10 Ϫ5 M) (31). The decreased K d of cyanide binding to ferrous d 1 heme, compared with ferric d 1 heme, is in contrast to the usual cyanide binding properties of hemes. Cyanide generally binds very strongly to the ferric form of heme iron; examples of dissociation constants include 1 ϫ 10 Ϫ9 M for human hemoglobin and 2.5 ϫ 10 Ϫ4 M for mitochondrial horse heart cytochrome c (32). These compare with values of Ն1 M for the ferrous forms of the same proteins (32,33). Such vast contrasts are not seen for the d 1 heme of cytochrome cd 1 . These observations show that the d 1 heme, compared with protoheme, is tailored to bind anions in a ferrous rather than a ferric state. The covalent modifications distinguishing d 1 heme from "normal" heme are electronegative alterations, increasing the positive potential of the d 1 heme iron (34). The immediate environment of the d 1 heme active site also contains two highly conserved histidine residues (His 345 and His 388 in P. pantotrophus), which, if protonated, contribute to the positive electrostatic potential of the active site (8). These two factors can be argued to contribute to the increased ability of d 1 heme, especially in the ferrous state, of cytochrome cd 1 to bind anions.
Allen et al. (6) estimated that the dissociation constant of the complex of nitrite with the activated, oxidized form of wild type cytochrome cd 1 from P. pantotrophus was ϳ30-fold higher than that reported here for the analogous complex of the oxidized Y25S enzyme. Presumably, the tendency of Tyr 25 to religate to the d 1 heme and thus displace the nitrite from the wild type enzyme is reflected in a lesser affinity (i.e. higher dissociation constant) for nitrite than the Y25S variant protein. The results obtained with the latter enzyme are more likely to reflect the intrinsic affinity of the oxidized d 1 heme ring, along with the active site ligands such as His 345 and His 388 , for nitrite. The affinity of the oxidized enzyme for nitrite is therefore quite high, which implies that a reaction mechanism in which nitrite binds to oxidized d 1 heme warrants consideration (the K m (nitrite) with physiological electron donor is 7-19 M) (4). It is discussed elsewhere in this work that the affinity of the reduced d 1 heme for cyanide is at least an order of magnitude higher than that of the oxidized d 1 heme. The affinities of the oxidized d 1 heme for nitrite and cyanide are very similar (cf. 65 and 44 M). It could then, by extrapolation, be estimated that the affinity of the reduced d 1 heme for nitrite would be at least an order of magnitude higher than that of the oxidized heme. If this is indeed true for nitrite binding, we can conclude that a mechanism involving the binding of nitrite to the reduced d 1 heme also warrants careful consideration.
In conclusion, the most important outcome of general significance is that the crystal structure of the Y25S mutant of P. pantotrophus cytochrome cd 1 led us to an incorrect interpretation of the properties of the protein. The continuing need for solution spectroscopic techniques to confirm or question aspects of protein crystal structures is clearly very important, especially when the properties of a protein are not readily interpretable in terms of a crystal structure.