Activation of the Cytochrome cd1 Nitrite Reductase from Paracoccus pantotrophus

Cytochromes cd1 are dimeric bacterial nitrite reductases, which contain two hemes per monomer. On reduction of both hemes, the distal ligand of heme d1 dissociates, creating a vacant coordination site accessible to substrate. Heme c, which transfers electrons from donor proteins into the active site, has histidine/methionine ligands except in the oxidized enzyme from Paracoccus pantotrophus where both ligands are histidine. During reduction of this enzyme, Tyr25 dissociates from the distal side of heme d1, and one heme c ligand is replaced by methionine. Activity is associated with histidine/methionine coordination at heme c, and it is believed that P. pantotrophus cytochrome cd1 is unreactive toward substrate without reductive activation. However, we report here that the oxidized enzyme will react with nitrite to yield a novel species in which heme d1 is EPR-silent. Magnetic circular dichroism studies indicate that heme d1 is low-spin FeIII but EPR-silent as a result of spin coupling to a radical species formed during the reaction with nitrite. This reaction drives the switch to histidine/methionine ligation at FeIII heme c. Thus the enzyme is activated by exposure to its physiological substrate without the necessity of passing through the reduced state. This reactivity toward nitrite is also observed for oxidized cytochrome cd1 from Pseudomonas stutzeri suggesting a more general involvement of the EPR-silent FeIII heme d1 species in nitrite reduction.

The enzyme monomer comprises two distinct domains, one of which binds heme c, the other heme d 1 (4). Heme c transfers electrons from donor proteins to the active site heme d 1 , an iron-dioxoisobacteriochlorin unique to this class of enzyme. In the crystal structure of the aerobically isolated ("oxidized") cd 1 from Paracoccus pantotrophus (Pp), each heme is in the Fe III state with two protein-derived axial ligands, His 17 and His 69 to heme c and His 200 and Tyr 25 to heme d 1 (see Scheme 1). MCD and EPR spectroscopy confirmed that these are the ligands in solution (5). Crystallography further showed that reduction of both hemes to the Fe II state triggers a conformational change, which results in the replacement of the heme c ligand His 17 with Met 106 and the dissociation of Tyr 25 from heme d 1 (6), thus creating a vacant coordination site at which substrate can bind (shown in top line of Scheme 1). Both dissociating heme ligands are provided by a 7-residue c-domain loop, which extends into the d 1 domain (6,7). The switch to His/Met ligation significantly raises the reduction potential of heme c, more closely matching the donor proteins cytochrome c 550 and pseudoazurin (8 -13). This led to the speculation that this facilitates electron transfer (8) but a low heme c potential is not strictly an obstacle because the overall electron transfer from donors to NO 2 Ϫ is thermodynamically favorable (13,14). The purpose of the ligand switch is therefore unclear. Other cytochromes cd 1 have His/Met liganded hemes c and lack a residue corresponding to His 17 . In these species, reduction appears sufficient to trigger dissociation of the distal heme d 1 ligand without a coordinated ligand switch at heme c (10,(15)(16)(17).
Reoxidation of reduced cytochromes cd 1 using NO 2 Ϫ yields a mixture of half-reduced species containing stable Fe II -nitrosyl heme d 1 (18 -21), although a recent report described rapid NO dissociation from the fully reduced enzyme (22). Reoxidation of fully reduced cd 1 using dioxygen gives rise to oxoferryl (Fe IV ϭO) heme d 1 and a protein radical (23). Hydroxylamine, however, is a convenient two-electron non-physiological substrate, which stoichiometrically reoxidizes both hemes in a single turnover (11,12,24,25).
With Pp cd 1 , the product of this reaction, freeze-trapped after ϳ10 ms, displays Fe III heme c EPR g values different from those of the oxidized enzyme (11). Because these g values are almost identical to those observed for the His/Met liganded heme c in other cytochromes cd 1 (see Table 1), it was suggested that the Pp enzyme has been trapped in a transient form (cd 1 * on Scheme 1) in which heme c retains the coordination normally associated with the reduced state (11,12). Unambiguous identification of a His/Met ligand pair by MCD spectroscopy confirmed this to be the case (12). The transient cd 1 * reverts to the oxidized conformation with His/His heme c over a period of ϳ20 min (11). Until this reversion has occurred, the enzyme is capable of NO 2 Ϫ reduction using the physiological electron donor pseudoazurin (11). It is reported that for full activity, oxidized Pp cd 1 requires preactivation by chemical reduction (11,13). Consequently, His/Met ligated heme c is presumed to be a prerequisite of activity, and it has been proposed that the oxidized enzyme is a resting form bypassed in catalysis as the enzyme cycles rapidly compared with the rate of reversion to the His/His form (13,26). The oxidized state would therefore not exist in vivo without an appropriate reactivation mechanism (8,12,26). The only candidate suggested to date is reduction by NapC, a tetraheme protein with reduction potentials ranging down to Ϫ235 mV (27,28).
The intermediate cd 1 * differs from oxidized Pp cd 1 in that it is reported to bind NO 2 Ϫ immediately following its formation by hydroxylamine reoxidation. The absorption spectrum of this putative NO 2 Ϫ -bound form is characteristic of low-spin Fe III heme d 1 (29). It was concluded that NO 2 Ϫ binds to the distal side of Fe III heme d 1 before the relatively slow rebinding of Tyr 25 can occur. However, several observations make it unlikely that the product (hereafter cd 1 *-X) is the simple NO 2 Ϫ -bound Fe III heme d 1 species depicted in the lower left of Scheme 1. The long timescale of formation is not consistent with a process, which is preventing the reversion to the oxidized state (29). The EPR spectrum contained no signals characteristic of low-spin Fe III heme d 1 . A single species was observed with similar g values to those of His/Met heme c in cd 1 * (29) ( Table 1). It was proposed that NO 2 Ϫ bound at heme d 1 was blocking the rebinding of Tyr 25 and preventing the conformational change to His/His ligation at Fe III heme c. This is not strictly proven because ligand orientation, not identity, is the primary determinant of the form of the EPR spectrum (30). Rotation of the histidine ligands from the perpendicular conformation observed in oxidized cd 1 could yield the same g values. The work described here was therefore undertaken with two primary objectives: to determine the identity of the ligands to heme c in cd 1 *-X and to provide an explanation for the EPR silence of the heme d 1 .

EXPERIMENTAL PROCEDURES
Purification of cd 1 -Samples of cd 1 from Pp and Pseudomonas stutzeri (Ps) strain ZoBell (ATCC 14405) were purified according to published methods (10,31). Initially, samples of the Pp cd 1 *-X were prepared according to reported methods (11,29): oxidized Pp cd 1 was reduced in an anaerobic hood (Faircrest Engineering; operating at Ͻ1 ppm O 2 in an N 2 atmosphere) using aliquots of buffered solutions of sodium dithionite; excess reductant was removed by chromatography on a PD-10 desalting column; the enzyme was reoxidized using aliquots of buffered hydroxylamine solution. Aliquots of buffered potassium nitrite solutions, freshly prepared, were added within 10 -15 s of this reoxidation. Assays of nitrite reductase activity were performed following published methods (13) using reduced equine cytochrome c as electron donor.
EPR spectra were measured with a Bruker ER300D spectrometer fitted with a dual mode cavity type ER4116DM interfaced to an ELEXSYS computer control system (Bruker Analytische Messtechnik GmBH) and equipped with a variable temperature cryostat and liquid helium transfer line (Oxford Instruments). EPR simulations were performed using the Bruker program WINEPR SimFonia (v1.25). MCD spectra were recorded on JASCO circular dichrographs models J810 and J730 for the UV-visible and near-infrared (NIR) regions, respectively, used in conjunction with an Oxford Instruments SM4 split-coil superconducting solenoid capable of generating magnetic fields of up to 5 tesla.
The intensities of spectra presented are referred to concentrations of cd 1 monomer, calculated using the Soret absorption intensity of the oxidized enzymes and extinction coefficients of ⑀ ϭ 141 mM Ϫ1 cm Ϫ1 for the Ps enzyme (32) and ⑀ ϭ 148 mM Ϫ1 cm Ϫ1 for the Pp enzyme (10). Samples for MCD spectroscopy were prepared in deuterium oxide solutions containing 50 mM BTP buffer at pH* ϭ 6.5 to which equivalent volumes of glycerol had been added as glassing agent (33). pH* is the apparent pH of the D 2 O-based solutions measured using a standard glass pH electrode.

RESULTS
Preparation and Absorption Spectra of cd 1 *-X-Samples of the Pp cd 1 *-X species were prepared anaerobically at pH ϭ 7.0 according to the methods described by Allen et al. (29). After 2-h incubation, the same type of absorption and EPR spectroscopic changes were observed. However, identical results were obtained during control experiments in which oxidized cd 1 was incubated with the same concentration (5 mM) of NO 2 Ϫ . This was the case whether the incubation was performed anaerobically or aerobically, but at pH 7.0, the reaction still required several hours to go to completion. Further control experiments SCHEME 1. The two-electron reoxidation of reduced Pp cd 1 by hydroxylamine followed by slow reversion to the oxidized state (within box) and the proposed trapping of cd 1 * by NO 2 ؊ .
revealed that the rate of this reaction of resting enzyme with NO 2 Ϫ was strongly pH-dependent, increasing significantly at lower pH. Although there are some variations in the rate between preparations, at pH 6.5, the product is fully formed in 5-10 min but at pH 9.0 negligible reaction was observed over 24 h. The substrate itself (pK a 3.35) is fully deprotonated over the pH range investigated, and the pH dependence may be a characteristic of NO 2 Ϫ binding and/or the subsequent reaction. No significant changes in the absorption spectrum of the oxidized enzyme were observed over the pH range 5.5-9.0. Assays of NO 2 Ϫ reductase activity using published methods (13) confirmed that cd 1 *-X prepared by this method is catalytically active, as is reported for the species formed by NO 2 Ϫ addition to enzyme reoxidized with hydroxylamine (29). All samples used for the spectroscopic characterization of cd 1 *-X described in this work were prepared in D 2 O-based buffers at pH* 6.5 by aerobic additions of buffered solutions of potassium nitrite. The absorption spectra of oxidized Pp cd 1 and of cd 1 *-X are shown in Fig. 1a. In oxidized Pp cd 1 at room temperature, Fe III heme d 1 exists in a thermal equilibrium between a low-spin ground state (S ϭ 1 ⁄ 2) and an excited high-spin state (S ϭ 5 ⁄ 2), which give rise to absorption bands at 644 and 702 nm, respectively (5). The cd 1 *-X spectrum lacks the high-spin Fe III heme d 1 band at 702 nm, but the low-spin band has increased in intensity and is blue-shifted to 631 nm. Low-spin Fe III heme c bands between 500 and 600 nm are present in both forms. It therefore appears that all Fe III heme d 1 is in the low-spin state following reaction with NO 2 Ϫ . The absorption spectrum of Pp cd 1 *-X is very similar to those of the oxidized cytochromes cd 1 from Pa and Ps, enzymes in which the heme d 1 is fully low-spin at all temperatures (5,19). This reactivity of oxidized Pp cd 1 is unexpected given that it is generally accepted that this state does not interact with NO 2 Ϫ (34). The heme c ligand switch of the Pp enzyme is unique among known cytochromes cd 1 , and this raises the question as to whether the ability to react with NO 2 Ϫ in this manner is also unique. We found that the cd 1 from Ps also reacts with NO 2 Ϫ at pH 6.5. When the Ps enzyme is treated with NO 2 Ϫ under identical conditions, the heme d 1 absorption band undergoes a small shift from 643 to 640 nm (Fig. 1b). However, in this enzyme, formation of a cd 1 *-X species, which contained low-spin Fe III heme d 1 would involve no change of spin-state at either heme. Less significant absorption changes would therefore result. That, these small changes do indicate the formation a cd 1 *-X, species is confirmed by EPR and MCD spectroscopic studies described below.
EPR Spectra of cd 1 *-X- Fig. 2 shows the 10 K X-band perpendicular mode EPR spectra of the oxidized cytochromes cd 1 from Pp and Ps, and of the species formed following treatment of each enzyme with NO 2 Ϫ . No signals were detected in the parallel mode for any of these samples. Heme c in oxidized Pp cd 1 (Fig.  2a) gives rise to a "large g max " spectrum with a g z feature at ϳ3.06. This is one of two limiting types of EPR observed for low-spin Fe III c-type hemes (5,30). The other two g values are often difficult to detect but spectral simulations, presented below, show that broad g y features are present in the g ϭ 2.2 -2.3

FIGURE 1. Electronic absorption spectra of cd 1 from Pp (a) and Ps (b) in the oxidized state (---) and after the addition of potassium nitrite (--).
Samples were in 50 mM BTP, D 2 O, pH* 6.5. Where present, NO 2 Ϫ ion concentration was 5 mM. region. Heme c of oxidized Ps cd 1 gives rise to an EPR spectrum of the second limiting type. This "rhombic" spectrum ( Fig. 2c) comprises three detectable features at g ϭ 2.99, 2.29, 1.61. Note that, although the hemes c in these two enzymes have different axial ligation, it is not this which leads to the different EPR spectra. The occurrence of a large g max or rhombic spectrum is dictated primarily by ligand orientation and not ligand identity. Thus both ligand sets could give rise to either type of EPR spectrum. In the case of the His/His-coordinated heme c of oxidized Pp cd 1 , the observation of a large g max spectrum is consistent with the perpendicular ligand arrangement observed in the crystal structure (4,30).
At 10 K, only the low-spin state of Fe III heme d 1 in oxidized Pp cd 1 is significantly populated, and minor traces of the highspin state are observed at g values of 6.86 and 4.99 (5). The low-spin Fe III state of heme d 1 gives rise to EPR spectra, which are unlike either of the limiting cases described for heme c, having narrow features and low g value anisotropy (5). These are observed in the oxidized cd 1  , the EPR spectra (Fig. 2, b and d) are devoid of the characteristic low-spin Fe III heme d 1 features, although this state of d 1 is apparent in the absorption spectra (Fig. 1). The Ps cd 1 *-X EPR spectrum (Fig. 2d) retains the signals which were assigned to heme c in the oxidized state, and it is reasonable to assume that that these g values do indicate His/Met ligation because the cd 1 from this species does not exhibit ligand switching. For NO 2 Ϫ -treated Pp cd 1 (Fig. 2b), the large g max spectrum of heme c has been replaced by a rhombic spectrum, with g values of 2.93, 2.32, 1.42 as reported by Allen et al. (29), who proposed that these features originate from Fe III heme c with His/Met ligation. The validity of this assignment will be addressed below using MCD spectroscopy. It has not previously been shown that this EPR spectrum accounts for a single heme, although, given its unusual electronic ground state, low-spin Fe III heme d 1 is unlikely to give rise to these signals (5,30,35). The EPR signals assigned to heme c in the cd 1 *-X forms (Fig. 2, b and d) were quantitated by double integration of the spectra and comparison to a Cu II -EDTA spin standard using established methods (36). Each represents a single low-spin Fe III heme per monomer confirming that heme d 1 is EPR-silent.
In the EPR spectra of the oxidized enzymes (Fig. 2), several features appear to have incompletely resolved structure suggesting minor heterogeneities at each heme. These features are preparation-dependent and of unknown origin. Because cd 1 is a dimeric protein and we have identified a form in which both hemes are apparently in the low-spin Fe III state but heme d 1 is EPR-silent, this raises the possibility that oxidized samples actually include substoichiometric levels of cd 1 *-X so that in some dimers one monomer contains an EPR-silent heme d 1 . This would constitute a heterogeneity, which could be responsible for the EPR splitting. However, if this was so then the contribution of heme d 1 to the overall EPR envelope would be less than that of heme c. To the best of our knowledge, it has never been determined that the EPR signals arising from heme c and heme d 1 represent equivalent concentrations of heme. This has therefore been tested by simulating the EPR spectrum of each oxidized enzyme. The spectra for simulation were recorded under non-saturating conditions. This required recording at a temperature of 20 K, and because the spin-lattice relaxation of heme d 1 is slower than that of heme c, a microwave power of 0.64 milliwatt. The microwave power-dependence of the two hemes at 10 and 20 K is shown in Fig. 3. The spectrum of each oxidized enzyme was simulated using contributions from two low-spin Fe III heme c species (c a and c b ) and two low-spin Fe III heme d 1 species (d a and d b ). The position and g value for the low-field feature of each simulated species is also indicated on the simulations in Fig. 4. All the g values used in these simulations are included in Table 1, where the weighting for each species is shown in brackets as percentage figures cal-  culated after setting the total heme c to 1.00 in each case. For the Ps enzyme, the heme d 1 contributions to the simulations represent only ϳ2/3 of those of heme c, which would be consistent with the presence of some EPR-silent heme d 1 . However for the Pp enzyme, which shows a more pronounced EPR heterogeneity, the ratio of total heme d 1 to total heme c is 0.99. Thus in the Pp enzyme at least, both hemes are fully EPR detectable, and cd 1 *-X is not the cause of the observed EPR heterogeneities. Furthermore, it is not apparent from these data that the heterogeneities at the two hemes are correlated because in none of the spectra does the percentage distribution between c a and c b match that between d a and d b .
NIR-MCD Spectra: the Heme c Ligands in cd 1 *-X-Heme c gives rise to almost identical rhombic EPR spectra in cd 1 * and in cd 1 *-X (Table 1 and Fig. 2), but as explained above, this does not necessarily indicate that heme c also has His/Met axial ligation in cd 1 *-X because these g values could also result from a change in the relative orientation of two histidine ligands from perpendicular to parallel. As was reported for cd 1 * (12), the identity of the heme c ligands in cd 1 *-X can be unambiguously determined using MCD spectroscopy in the NIR region (700 -2000 nm). At these wavelengths, low-spin Fe III hemes give rise to a characteristic porphyrin-to-Fe III charge-transfer (CT) transition, the exact energy of which is diagnostic of the axial ligands to the heme (37,38). However, for the unusual (d zx , d yz ) 2 (d xy ) 1 ground state of low-spin Fe III heme d 1 , the band is symmetry forbidden and has not been detected (5,35). Consequently, the transitions observed in this region are solely due to low-spin Fe III heme c. NIR-MCD spectra, recorded at 4.2 K, are shown in Fig. 5 for the four forms of cd 1 reported in Figs. 1 and 2. The hemes c in oxidized Pp cd 1 and oxidized Ps cd 1 give rise to CT bands with peaks at approximately 1550 and 1775 nm, wavelengths diagnostic of His/His and His/Met coordination, respectively, as is previously reported (5). For Ps cd 1 *-X, the band remains at 1775 nm confirming retention of His/Met ligands. However, on formation of cd 1 *-X in the Pp enzyme, the CT band is red-shifted from 1550 to 1795 nm proving that a ligand switch to His/Met at heme c has occurred.
The Electronic State of Heme d 1 in cd 1 *-X-This work has established that the low-spin Fe III heme c in cd 1 *-X has His/Met ligation and so, in the case of the Pp enzyme, a ligand switch has been triggered in the oxidized enzyme by the reaction with NO 2 Ϫ . The nature of heme d 1 in cd 1 *-X is complex in that it appears to be low-spin Fe III on the basis of absorption spectra but is EPR-silent. The electronic and magnetic properties of this center have therefore been investigated using variable magnetic field variable temperature MCD spectroscopic methods. The UV-visible region MCD spectrum of a heme contains an extremely detailed pattern of electronic bands, and it is well established for b-and c-type hemes that this pattern is diagnostic of the spin and oxidation state of the iron, including the EPR-silent Fe II states (38). Although heme d 1 is unique to this class of enzyme, and the available data is limited, it is clear that heme d 1 MCD spectra are also sensitive to changes in spin and oxidation state (5, 32, 39 -42). The UV region MCD of cd 1 is dominated by an intense heme c Soret band but at visible wavelengths, transitions of both hemes are readily observed. Fig. 6a shows MCD spectra of the cd 1 *-X form of the Pp enzyme recorded in this wavelength region using a magnetic field of 5  (5,30). b Numbers in parentheses represent the variation in the second decimal place in g values reported for each feature. This includes spread due to a 2-fold heterogeneity often observed for cd 1 EPR features. c -, g-values not determined from the spectrum. d Percentages in parentheses are the weightings used for each species in the simulations of Fig. 4. tesla and at three temperatures, 1.7, 4.2, and 10 K. All bands are strongly temperature-dependent indicating that they arise from paramagnetic centers. The MCD band pattern of low-spin Fe III c-type hemes is well known (38). The major transitions attributable to low-spin Fe III heme c are therefore readily assigned and are indicated in the upper part of Fig. 6a by blocks, which are offset vertically to illustrate bands of positive and negative intensity. Thus, in the region 520 -590 nm, the negative feature at 557 nm and the two positive bands to either side are solely due to low-spin Fe III heme c. Blocks in the lower part of Fig. 6a illustrate the positions and signs of transitions, which have previously been observed for heme d 1 in the low-spin Fe III state (5,32,39,40,42). Transitions from both hemes overlap at 450 -520 nm but features in the 590 -700 nm region are pure heme d 1 transitions. 3 The overall pattern of bands in the MCD spectrum here is virtually identical to that previously reported for oxidized cd 1 from Ps, in which both hemes are already entirely low-spin Fe III (5). These spectra therefore further support the conclusion that in cd 1 *-X, heme d 1 is in the low-spin Fe III state and is responsible for paramagnetic MCD intensity despite the fact that it does not give rise to a detectable EPR spectrum at X-band.
There is no precedent for an isolated low-spin Fe III heme not giving rise to a detectable EPR spectrum but there are cases of Fe III hemes being rendered EPR-silent by a spin-spin interaction with a second paramagnetic center. For example, the heme at the active site of bacterial nitric oxide reductase gives rise to MCD bands characteristic of the high-spin Fe III state but it is EPR-silent as a result of spin-coupling with a nearby non-heme Fe III iron (43). Such an explanation that Fe III heme d 1 in cd 1 *-X is spin-coupled to a second paramagnet would reconcile the EPR and MCD data, and this possibility was investigated by further MCD measurements.
The pattern of bands in the MCD spectrum reports fundamental properties local to the heme, such as spin and oxidation state, and this band pattern will not be significantly altered if the spin of the heme is spin-coupled to a second paramagnetic species. But even a relatively weak interaction of this nature could preclude the observation of EPR signals. Such an interaction could, however, be detected via the magnetic field and temperature dependence of the MCD intensity. These properties have, therefore, been studied for cd 1 *-X using the "ratio data" method of MCD spectroscopy (RD-MCD). This method was originally applied to cytochrome bo 3 to locate and analyze heme o 3 MCD bands from the active site Fe III heme o 3 -Cu B II pair against an intense background of low-spin Fe III heme b bands (44). RD-MCD allows the immediate identification of transitions in the MCD spectrum, which arise from paramagnetic species for which the spin S 1 ⁄ 2. Details of the method are given in Ref. 44, and only a brief summary will be presented here. For the case of an S ϭ 1 ⁄ 2 chromophore, the MCD intensity varies as tanh(Ϫ␤H⌫/2 kT), where ␤ is the Bohr magneton, k is Boltzmann constant, H the magnetic field, and T the absolute temperature. ⌫ is a function of the g values, which determines the magnitude of the Zeeman splitting between the ground state components, m S ϭ Ϯ 1 ⁄ 2, at any orientation of the molecule to the external magnetic field expressed as the spherical polar angles and .
⌫ ϭ [g zz 2 cos 2 ϩ (g xx 2 cos 2 ϩ g yy 2 sin 2 )sin 2 ] 1/2 (Eq. 3) Except at extreme high magnetic field or low temperature, outside of those used in this work, tanh(Ϫ␤H⌫/2kT) Ϸ ␤H⌫/2kT, and so the MCD intensities of S ϭ 1 ⁄ 2 systems will be proportional to H/T. Therefore if the MCD spectrum is recorded at different temperatures, but the magnetic field is adjusted so as to maintain the H/T ratio, then the contribution to the spectrum of S ϭ 1 ⁄ 2 species remains constant. Variations in intensity will be observed only for bands arising from paramagnetic centers, which are not simple S ϭ 1 ⁄ 2 species. Fig. 7 shows RD-MCD spectra of cd 1 *-X, from Pp and Ps, recorded at four different combinations of H/T. It is immediately apparent, for both enzymes, that the spectra are closely matched in intensity across the 520 -590 nm region. This is entirely consistent with the previous assignment of these bands in Fig. 6 to the low-spin Fe III heme c. In the regions to either side, where heme d 1 bands occur, there are obvious variations in the MCD intensities. This divergence of the spectra is most pronounced in the 590 -700 nm region where the bands are exclusively from heme d 1 . These RD-MCD data therefore demonstrate that heme d 1 in cd 1 *-X is part of a paramagnetic but EPR-silent system. The S ϭ 1 ⁄ 2 lowspin Fe III heme d 1 must be interacting with another half-integer (Kramers) paramagnet. Heme c is clearly not involved; it remains EPR detectable and in the RD-MCD spectra gives rise to bands, which overlie as expected for an isolated S ϭ 1 ⁄ 2 spe- cies. Therefore, the species interacting with heme d 1 can only be a radical resulting from the reaction between oxidized cd 1 and NO 2 Ϫ .

DISCUSSION
This work has shown that Pp cd 1 in the oxidized state will react directly with the physiological substrate NO 2 Ϫ to form cd 1 *-X, a complex in which heme d 1 is paramagnetic but EPRsilent. This reaction drives a ligand switch at Fe III heme c from His/His to His/Met, the conformation associated with enzyme activity (11)(12)(13)34). Hence, the enzyme has been activated without passing through the reduced state, something previously considered essential (6,11,34). Dissociation or displacement of Tyr 25 to allow NO 2 Ϫ binding does not therefore have to be driven by reduction of the two hemes, and the lack of appropriate P. pantotrophus cellular components with reduction potentials low enough to reduce cd 1 may not be an issue (27).
We have also shown that an equivalent cd 1 *-X species is formed in the reaction of NO 2 Ϫ with cd 1 from P. stutzeri. The reactivity is not, therefore, unique to enzyme with the ability to switch heme c ligands but may be a property of cytochromes cd 1 in general. Over twenty years ago, Muhoberac and Wharton reported that the addition of NO 2 Ϫ to Pseudomonas aeruginosa cd 1 results in the loss, from the EPR spectrum, of features associated with heme d 1 (45). The samples, at pH 7.0, were incubated for 24 h before spectra were recorded. There can be little doubt that this is the same reaction as has been described here in detail.
MCD showed that the EPR-silent Fe III heme d 1 is paramagnetic but that S 1 ⁄ 2. Because heme c is magnetically isolated, and the enzyme contains no other cofactors, heme d 1 must be interacting with a second paramagnet, presumably a radical species formed in the reaction with NO 2 Ϫ . The spin interaction is of sufficient strength to preclude observation of EPR signals at X-band, and so the magnitude of J, the spin-coupling constant, must therefore be several times that of the microwave photon energy (ϳ0.3 cm Ϫ1 ). However, the two spins are not coupled sufficiently strongly to yield a diamagnetic system; the MCD intensifies significantly between 4.2 and 1.7 K (Fig. 6) placing an upper limit on J of ϳ5 cm Ϫ1 (our preliminary analysis 4 of the RD-MCD suggests that J ϳ 3 cm Ϫ1 ). This order of magnitude would require that the radical is located in close proximity to heme d 1 . For example, in compound I of cytochrome c peroxidase, the exchange coupling between a tryptophan radical and an oxoferryl (Fe IV ϭO) heme separated by ϳ5 Å is Ͻ0.1 cm Ϫ1 (46,47). One possible explanation is that a single turnover of NO 2 Ϫ has taken place (Equation 1) using an electron abstracted from a nearby residue and so forming a radical. Displacement of the NO by a second NO 2 Ϫ would yield a species [R ϩ⅐ d 1 III -NO 2 Ϫ ] in which low-spin Fe III heme d 1 is coupled to the protein radical. Tyr 263 is the only fully conserved residue in close proximity to heme d 1 and has already been proposed as the location of a radical species formed during the reaction of reduced cd 1 with oxygen (23).
However, although the visible region absorption and MCD spectra of Pp cd 1 *-X are characteristic of low-spin Fe III heme d 1 , a band at this specific wavelength (ϳ631 nm, Fig. 1a) has previously been assigned to a Fe III heme d 1 nitrosyl species (18,48) formed in the reaction of NO 2 Ϫ with Fe II heme d 1 . The addition of NO to oxidized cd 1 from both Pa and Ps also results in absorption spectra similar to that of Ps cd 1 *-X (Fig. 1b) (19,49). This would, at first, appear to contradict our data because ferric nitrosyl hemes are best described as diamagnetic Fe II -NO ϩ species (50) and do not give rise to spectra characteristic of Fe III . The unpaired * electron of the NO is largely transferred to the Fe III d xz,yz orbitals. But we have previously proposed that the fully occupied d xz,yz orbitals in the unusual (d xz,yz ) 4 (d xy ) 1 ground state of low-spin Fe III heme d 1 could destabilize this mode of NO binding (5). Heme d 1 in cd 1 *-X is clearly paramagnetic and may represent a genuine Fe III -NO species, in which the NO ligand itself is the radical species interacting with the spin of Fe III heme d 1 . This interpretation would require that a transient protein radical is formed during the reaction but dissipates rapidly. A relatively weak coupling between a radical ligand and the spin of the metal ion to which it is bound is unusual but not without precedent. In horseradish peroxidase compound I, the coupling between the oxoferryl (Fe IV ϭO) iron and a radical on the porphyrin ligand is ϳ4 cm Ϫ1 (51,52). In the case of low-spin Fe III heme d 1 , the unpaired electron of metal is located not in d xz,yz orbitals, which readily interact with NO, but in the d xy orbital (5), which lies in the plane of the isobacteriochlorin and is potentially orthogonal to the NO * orbitals in a linear Fe-N-O conformation. However, these putative Fe III -NO species are stable, and so this property of heme d 1 alone is not sufficient to dissociate product. It was previously reported that addition of the chemical reductant ascorbate to as-prepared Pa cd 1 resulted in the appearance of the EPR sig- nature of Fe II -NO (53). This was attributed to the presence of NO 2 Ϫ in the enzyme, which was being reduced to NO. However, it cannot be excluded that the species responsible was equivalent to cd 1 *-X and contained Fe III -NO. This would also yield ferrous nitrosyl EPR signals upon reduction. EPR spectra of the as-prepared enzyme are not shown (53), and where they are presented elsewhere (17,32,39,45,54), no integrations are reported. Consequently it is not possible to say whether or not these samples contained EPR-silent heme d 1 . If they did, then it may be that Pa and Ps cytochromes cd 1 are prone to retaining product NO. The contrasting behavior of Pp cd 1 may be linked to an ability of Tyr 25 to displace distal heme d 1 ligands.
NO evolution is detected during cd 1 activity assays, which are performed in the presence of excess reductant (13,55). Yet, in several studies of the reaction of NO 2 Ϫ with the fully reduced enzymes from both Pp and Pa, the reaction stopped after a single turnover (18 -21). Evidently the role of reductant in product release is complex and may well be related to redoxlinked conformational changes at heme c.
The reaction with NO 2 Ϫ constitutes a mechanism by which the oxidized Pp enzyme can be activated by substrate but at a rate which is several orders of magnitude slower than turnover. It is unlikely therefore that this species is directly involved in the catalytic cycle in this exact form. However, the involvement of a transient protein radical in NO 2 Ϫ reduction would represent an additional redox center and raises the possibility that heme d 1 does not access the Fe II state during turnover of NO 2 Ϫ . This would perhaps explain the observations that carbon monoxide, a ligand to Fe II hemes, inhibits the reduction of O 2 by Pa cd 1 but not the reduction of NO 2 Ϫ . In contrast cyanide ion, a strong ligand to Fe III heme, inhibits both activities (56). The idea that the enzyme turns over too rapidly for reversion to the His/His heme c to occur (13, 26) may still be valid. However, it should be considered that this slow reversion is prevented by substrate binding rather than re-reduction of heme d 1 .
In summary, we have shown that cd 1 in the oxidized state reacts directly with its substrate NO 2 Ϫ driving the heme c His/ His to His/Met ligand switch required for activity. The reaction also occurs with cd 1 s that do not undergo ligand switching and do not require activation. The resulting novel species contains low-spin Fe III heme d 1 that is rendered EPR-silent by spin-coupling to a radical that we propose is bound nitric oxide formed in the reaction.