Electron Transfer Associated with Oxygen Activation in the B2 Protein of Ribonucleotide Reductase from Escherichia coli*

Each of the two /3 peptides which comprise the B2 protein of Escherichia coli ribonucleotide reductase (RRB2) possesses a nonheme dinuclear iron cluster and a tyrosine residue at position 122. The oxidized form of the protein contains all high spin ferric iron and 1 .O-1.4 tyrosyl radicals per RRBB protein. In order to define the stoichiometry of in uitro dioxygen reduction catalyzed by fully reduced RRBS we have quantified the reactants and products in the aerobic addition of Fe(I1) to metal-free RRBB,, utilizing an oxygraph to quantify oxygen consumption, electron paramagnetic resonance to measure tyrosine radical generation, and Mossbauer spectroscopy to determine the extent of iron oxidation. Our data indicate that 3.1 Fe(I1) and 0.8 Tyr"' are oxidized per mol of 0 2 reduced. Mossbauer experiments indicate that less than 8% of the iron is bound as mononuclear high spin Fe(II1). Further, the aerobic addition of substoichiometric amounts of "Fe to RRBB,, consistently produces dinuclear clusters, rather than mononuclear Fe(II1) species, providing the first direct spectroscopic evidence for the preferential formation of 280 nm ((280 = 120 mM" cm"). Bovine erythrocyte superoxide dismutase and bovine liver catalase were purchased from Sigma. Dioxygen consumption experiments were performed on a Gilson Model K-ICT C Oxygraph with a 1.5-ml water jacketed cell equipped with a Clark electrode. Experiments were conducted at ambient temperature and controlled with a circulating water bath. All solu- tions were prepared with 25 mM HEPES buffered at pH 7.6, unless otherwise noted. The oxygraph was calibrated by addition of a known amount of an anaerobic solution of protocatechuate to an oxygen-saturated solution containing protocatechuate 3,4-dioxygenase (-5 p ~ ) . For reconstitution of RRBZ.,, four equivalents of Fe(I1) from an anaerobic aqueous solution of Fe(NH4)z(S04)z (14 mM) were added to an air saturated solution of RRBZ, (50 p ~ ) . Immediately follow- ing the O2 consumption experiment, 500 p1 were transferred to an epr tube or Mossbauer cup and frozen in liquid nitrogen. Reconstituted samples had specific activities of 3000 units. Tyrosyl radical concentration was determined by epr at 77 K using a Bruker ER/300 epr spectrometer by quantification of the double-integrated epr signal at g = 2 uersw diphenylpicrylhydrazyl hydrate (129.9 PM). FeEDTA was used as a standard for the g = 4.3 signal of high spin mononuclear Fe(II1). Mossbauer experiments were performed as previously de- scribed (10). 67Fe-reconstituted (Y122F)RRB2 was obtained by adding 3 equivalents of 67Fe(II) to (Y122F)RRB2., in buffer and used without further purification. B2 protein of E. coli ribonucleotide reductase reconstituted with "Fe ((Y122F)RRB2). The spectrum was recorded at 4.2 K in a 0.06 T parallel applied magnetic field. The solid line is a simulated spectrum using the Mossbauer parameters of the wild-type protein. The additional absorption between 0 and +1 mm/s double-velocity results from a minor ( 4 % ) impurity, possibly an Fe(II1) aggregate in solution.

Each of the two / 3 peptides which comprise the B2 protein of Escherichia coli ribonucleotide reductase (RRB2) possesses a nonheme dinuclear iron cluster and a tyrosine residue at position 122. The oxidized form of the protein contains all high spin ferric iron and 1 .O-1.4 tyrosyl radicals per RRBB protein. In order to define the stoichiometry of in uitro dioxygen reduction catalyzed by fully reduced RRBS we have quantified the reactants and products in the aerobic addition of Fe(I1) to metal-free RRBB,, utilizing an oxygraph to quantify oxygen consumption, electron paramagnetic resonance to measure tyrosine radical generation, and Mossbauer spectroscopy to determine the extent of iron oxidation. Our data indicate that 3.1 Fe(I1) and 0. 8 Tyr"' are oxidized per mol of 0 2 reduced. Mossbauer experiments indicate that less than 8% of the iron is bound as mononuclear high spin Fe(II1). Further, the aerobic addition of substoichiometric amounts of "Fe to RRBB,, consistently produces dinuclear clusters, rather than mononuclear Fe(II1) species, providing the first direct spectroscopic evidence for the preferential formation of the dinuclear units at the active site. These stoichiometry studies were extended to include the phenylalanine mutant protein (Y 122F)RRB2 and show that 3.9 mol-equivalents of Fe(I1) are oxidized per mol of O2 consumed. Our stoichiometry data has led us to propose a model for dioxygen activation catalyzed by RRBS which invokes electron transfer between iron clusters.
Ribonucleotide reductase (RR)' catalyzes the reduction of all four ribonucleotides to their corresponding deoxyribonucleotides, an essential step in the synthesis of DNA in all living cells (1)(2)(3). The Escherichia coli enzyme contains two proteins, designated B1 (a2) and B2 (82). The RRBl protein contains the substrate and effector binding sites as well as * This work was supported by National Science Foundation Grant DMB-8804458 (to L. Q.), National Institutes of Health Grant GM-22701 (to E. M.), and the Swedish Cancer Society (to B.-M. S.). J. B. L. is grateful for a National Institutes of Health Predoctoral Traineeship (GM-07323). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
)I To whom correspondence should be addressed Dept. of Chem- several redox-active thiols (4)(5)(6). The active form of the RRB2 protein contains a stable tyrosine radical, essential for enzymatic activity, and a dinuclear nonheme iron-oxo cluster, involved in the generation and stabilization of the radical (7). Site-directed mutagenesis has been employed (8) to construct a radical-free, inactive mutant in which the native tyrosine at position 122 was substituted with a phenylalanine residue, (Y122F)RRB2, clearly identifying Tyr."' as the position of the tyrosyl radical in the wild-type enzyme.
The radical free diferric form of wild-type RRB2 (RRB2,,J has peen crystallized and the x-ray structure determined to 2.2-A resolution (9). The reported three-dimensional structure supports a stoichiometry of one dinuclear iron-oxo cluster per polypeptide, as previously proposed by Lynch et al. (10); this raises the possibility of having one tyrosyl radical associated with each dinuclear cluster. However, the highest ty-rosy1 radical content per RRB2 protein reported (10) is 1.4 as determined by epr spectroscopy.
The fluctuation in the tyrosyl radical concentration in the cell in response to its growth cycle suggests the availability of cellular mechanisms to reduce the radical and then to regenerate it (11). In uitro, various forms of the RRBB protein have been prepared, as sbown in Fig. 1. Reichard et al. (12) have shown that aerobic incubation of RRBB,,, with a crude extract from E. coli will regenerate the tyrosyl radical; however, simple exposure to oxygen will not. Two of the three components from this extract have been isolated and identified as superoxide dismutase (12) and an NAD(P)H:flavin oxidoreductase (13). This observation has led to the proposed mechanism that the oxidoreductase may be capable of reducing RRB2 to a form which could then react with dioxygen followed by oxidation of Tyr'". The tyrosyl radical together with the (r-0xo)differic cluster may also be generated by exposure of the diferrous (RRB2,,d) form of the protein to dioxygen (14) or the aerobic Fe(I1) reconstitution of metal-free (RRBB,,,) protein (7, 15). However, the dinuclear site is not reconstituted by the addition of Fe(II1) (14). The apparent inefficiency of tyrosyl radical production has led us to further investigate the radical generating chemistry catalyzed by this nonheme dinuclear iron cluster. Specifically, this study focuses on the stoichiometry of dioxygen activation catalyzed by RRB2.
The reduction of dioxygen to water requires four reducing equivalents. RRB2,,d appears to provide three of these reducing equivalents by the oxidation of one dinuclear ferrous cluster and Tyr'". 2Fe" + Tyr"' + X + 0, + Fe"'-O-Fel" (Reaction 1) The source of the fourth reducing equivalent ( X in Reaction 1) is unknown. However, in a recent study of mouse ribonucleotide reductase (16), it was speculated that the fourth equivalent is provided by adventitiously bound Fe(I1). We have sought to further define the stoichiometry of in vitro dioxygen activation associated with the RRB2 protein. We report here our findings for the stoichiometry of Reaction 1 for wild-type RRB2 and the (Y122F)RRB2 mutant which suggest that the fourth reducing equivalent is not supplied by adventitiously bound iron in the E. coli protein. A model is proposed to account for our stoichiometric data which invokes electron transfer between dinuclear clusters.

MATERIALS AND METHODS
RRBZ was isolated from E. coli strain N6405/pSPS2 (17), a heatinducible overproducer, as previously described (18) and its concentration was determined by absorbance at 280 nm (ezm = 141 mM" cm") (10). The procedure used to isolate the mutant (Y122F)RRB2 from E. coli strain K38/pMK5/pGP1-2 (8, 19) was identical to that used for the wild-type. The RRBZ protein was purified to homogeneity as indicated by the presence of a single band on a sodium dodecyl sulfate-polyacrylamide gel electrophoresis. RRBZ isolated by this method displays a specific activity of 4700 determined spectrophotometrically by monitoring the disappearance of NADPH (0.16 mM) in the presence of RRBl (60 pg/ml), thioredoxin (50 pg/ml), thioredoxin reductase (0.7 pg/ml), CDP (1 mM), and ATP (1.6 mM) as previously described (10). Iron removal was accomplished by dialysis against the lithium salt of 8-hydroxyquinoline-5-sulfonic acid in 1.0 M imidazole as previously described (7). The concentration of RRBB., was determined by absorbance at 280 nm ((280 = 120 mM" cm").
Bovine erythrocyte superoxide dismutase and bovine liver catalase were purchased from Sigma.
Dioxygen consumption experiments were performed on a Gilson Model K-ICT C Oxygraph with a 1.5-ml water jacketed cell equipped with a Clark electrode. Experiments were conducted at ambient temperature and controlled with a circulating water bath. All solutions were prepared with 25 mM HEPES buffered at pH 7.6, unless otherwise noted. The oxygraph was calibrated by addition of a known amount of an anaerobic solution of protocatechuate to an oxygensaturated solution containing protocatechuate 3,4-dioxygenase (-5 p~) .
Immediately following the O2 consumption experiment, 500 p1 were transferred to an epr tube or Mossbauer cup and frozen in liquid nitrogen. Reconstituted samples had specific activities of 3000 units. Tyrosyl radical concentration was determined by epr at 77 K using a Bruker ER/300 epr spectrometer by quantification of the double-integrated epr signal at g = 2 uersw diphenylpicrylhydrazyl hydrate (129.9 PM). FeEDTA was used as a standard for the g = 4.3 signal of high spin mononuclear Fe(II1). Mossbauer experiments were performed as previously described (10). 67Fe-reconstituted (Y122F)RRB2 was obtained by adding 3 equivalents of 67Fe(II) to (Y122F)RRB2., in buffer and used without further purification.

RESULTS
In order to define the stoichiometry of dioxygen reduction catalyzed by RRB2,d (Reaction l), we have quantified the reactants and products in the aerobic addition of Fe(I1) to RRBB.,,, utilizing an oxygraph to quantify oxygen consumption, epr to measure tyrosine radical generation, and Mossbauer spectroscopy to determine the extent of iron oxidation. Listed in Table I are results from the addition of four equivalents of Fe(I1) to an air-saturated solution of RRB2,,. These results indicate that 3.1 mol-equivalents of Fe(I1) and 0.8 molequivalent of Tyr"' are oxidized per mol of O2 reduced for wild-type RRB2. In vitro reconstitution experiments were run by adding an anaerobic ferrous ammonium sulfate solution to an oxygraph cell filled with an aerobic RRBZ,, solution. Aliquots were removed from the oxygraph cell and frozen immediately in liquid nitrogen for epr analysis. The ratios in Table I are not affected by the addition of catalase or superoxide dismutase to the RRBS,, solution, consistent with results reported by Salowe (18), indicating that diffusive peroxide or superoxide species are not produced during the reduction of dioxygen by RRB2d.
Because Fe(II1) has been shown not to be incorporated into the dinuclear binding sites in RRB2 but can bind to the protein adventitiously (14), it is essential to this study to monitor and quantify the fate of the Fe(I1) added to RRBB., upon exposure to 0 2 . Thus, in order to assess the integrity of the dinuclear clusters upon reconstitution and permit the quantification of adventitiously bound Fe(III), oxygraph experiments were run with 67Fe(II) to obtain the O2 stoichiometry and the samples were then subjected to Mossbauer spectroscopic quantification of these iron species. Mossbauer spectra of these samples contain two quadrupole doublets with h E B = 1.62 and 2.44 mm/s and 6 = 0.55 and 0.45 mm/s, respectively; these parameters are identical to those observed for the differic site in RRB2,, (7). The amount of mononuclear high spin ferric iron bound can be determined by quantification of the outer features of its magnetic hyperfine spectrum obtained in 60 kilogauss externally applied field. With this procedure, the mononuclear high spin ferric component accounts for no more than 8% of the total iron content. Aerobic addition of substoichiometric amounts of 57Fe(II) to RRB2., consistently produces dinuclear clusters, rather than any mononuclear Fe(II1) species.
The Mossbauer data suggest that less than 8% of the Fe(I1) is oxidized and ultimately observed as mononuclear high spin Fe(III), a result which is corroborated by quantitation of the epr signal at g = 4.3. Consequently, the oxygen consumption data must be reduced by at most 2% assuming that 0 2 is reduced to water in generating the mononuclear high spin Fe(II1) component. This seems likely since the oxygen consumption results do not change when the reconstitution experiments are run in the presence of catalase, indicating that peroxide is not generated as a result of partial oxygen reduction. Dioxygen consumption is complete 30 s after the addition of Fe(I1) to RRBB,, (Fig. 2 A ) . Addition of Fe(I1) to the oxygraph cell in the absence of protein requires 4 min to run to completion indicating a much slower oxidation rate (Fig.  2B). The oxygen consumption which results from the addition of Fe(I1) to RRB2,,t follows a trace similar to that shown in Fig. 2B. Ascorbate is commonly used in the reconstitution of RRB2., with Fe(I1) (7,14), presumably to keep the iron reduced and to achieve the highest possible degree of iron reconstitution. In order that the number of reducing equivalents may be accurately quantified, we have not used ascorbate in these experiments.
The results for similar stoichiometry measurements are shown in Table I for the addition of Fe(I1) to (Y122F)RRB2,,,; this enzyme has a phenylalanine residue substituted at position 122 in place of the native tyrosine residue. These experiments indicate that 3.9 mol-equivalents of Fe(I1) are oxidized per mol of O2 consumed. The Mossbauer spectrum of the oxidized form of (Y122F)RRB2, shown in Fig. 3, is indistinguishable from that of the wild-type protein. Table I  alone and Fe(I1) is incorporated into RRB2 predominantly, if not almost exclusively, as dinuclear clusters even when substoichiometric amounts are added. The Mossbauer parame-ters for the diferrous cluster are unique for RRB2. Since addition of Fe(II1) to RRBP,,, does not generate this unique dinuclear cluster, the diferrous unit must form preferentially at the active site and O2 reduction must occur only at this site. Otherwise, we should observe a greater proportion of mononuclear Fe(II1) units, particularly in the substoichiometric additions. Therefore, it seems likely that the formation of the diferrous center facilitates 0 2 binding and reduction. The carboxylate-rich nature of the active site (9) would be expected to lower the reduction potential of the diiron center and enhance its oxidation.

Listed in
Based on the values listed in Table I, we report the following stoichiometry for the reduction of dioxygen by the diferrous clusters in RRB2. This stoichiometry is reported in terms of dioxygen consumption in order to track the number of reducing equivalents, which totals 3.9 per mol of 0 2 consumed. Therefore, our results appear to be consistent with that reported for the mouse enzyme by Thelander et al. (16) in which three Fe(I1) and one tyrosine combine to contribute the four electrons required to reduce O2 to water.
The results for analogous stoichiometry measurements for the addition of Fe(I1) to (Y122F)RRB2,,,, shown in Table 1, indicate that 3.9 mol-equivalents of Fe(I1) are oxidized per mol of O2 consumed. This result suggests that the iron clusters are capable of providing all four of the reducing equivalents required to reduce 0 2 . Thus, it appears that one dinuclear cluster is capable of transferring its reducing equivalents to another cluster. This suggests that the activated iron-oxygen species, which results from the binding of O2 and subsequent oxidation of one iron cluster, has a potential high enough that electron transfer from the remaining diferrous cluster takes precedence over the binding and activation of O2 at the latter diferrous cluster.
The data for tyrosyl radical generation in Table I, 1.0 radical per RRB2 protein, is clearly less than the optimum yield of one radical per iron cluster. The value of 0.8 radicals per O2 is also less than the one radical per O2 predicted by a stoichiometry of 3.0 Fe(I1) and 1.0 Tyr"' per 0, consumed. This low tyrosyl radical content, consistent with previous reports for reconstitution experiments (7, lo), may be due to the inefficiency of radical formation, the instability of an intermediate formed during radical generation, or the availability of an alternative mechanism for the reduction of O2 which does not result in the oxidation of Tyr"'. The mechanism by which O2 reduction is achieved by (Y122F)RRB2 may contribute to the apparent inefficiency of tyrosyl radical production in the wildtype enzyme. Analysis of the wild-type data in Table I 0.7 Fe(1I) and 0.2 O2 closely approximates the 4 Fe(I1) per O2 stoichiometry of (Y122F)RRB2. This suggests that 20% of the O2 consumption may result from this alternative mechanism.
Thelander et al. (16) have reported studies on the mouse wild-type enzyme in which they conclude that 3 mol-equivalents of Fe(I1) are required to produce one tyrosyl radical. The third Fe(I1) ion, when oxidized, is designated as adventitiously bound Fe(III), which could be incorporated into the dinuclear site upon reduction by ascorbate. It should be noted that Fe(II1) is not incorporated into the E. coli RRB2,, dinuclear sites (14). The Mossbauer results reported here, the crystal structure (9) and a corrected iron content of 4 mol of Fe per RRB2 protein (10) previously reported, suggest that it is unlikely that adventitiously bound iron plays a role in this dioxygen consumption stoichiometry for the E. coli enzyme.
We thus propose a new model for the oxidation of RRB2,,d shown in Fig. 4 to account for this stoichiometry data. In this model, reducing equivalents can be supplied by another diferrous cluster via either intramolecular or intermolecular long range electron transfer. Impetus for an intercluster electron transfer mechanism comes from the observation that two diferrous clusters are oxidized per molecule of O2 for the Y122F mutant protein, which is best rationalized by such a mechanism. In the activation scheme proposed for the wildtype protein, the consumption of one equivalent of dioxygen results in the two-electron oxidation of one diferrous cluster, oxidation of one tyrosine radical and oxidation of one iron from the neighboring diferrous cluster, the latter yielding a mixed valence Fe(II).Fe(III) cluster (Reaction 2). The oxidation of this cluster to the mixed valence state effectively prevents it from oxidizing its own TyrlZ2 to the radical state.
The mixed valence state has been spectroscopically identified in other dinuclear nonheme iron proteins such as hemerythrin (20), purple acid phosphatase (21), and the hydroxylase component of methane monooxygenase (22). The chemical reduction methods used to generate the mixed valence state in these metalloproteins have not facilitated the isolation of this form of RRB2. However, Hendrich et al. (23) have recently characterized the mixed valence form of RRB2 generated by low temperature x-irradiation of an RRB2,,* sample. The epr signal associated with this state of the iron cluster disappears when the sample is thawed, indicating that it is a very reactive, unstable intermediate species.
The proposed scheme suggests that the mixed-valence RRB2 intermediate may disproportionate (Reaction 3) similar to that shown to occur for the mixed valence forms of hemerythrin (20,24). Reaction 4 of this proposed scheme is similar to that proposed for (Y122F)RRB2 in which 4 Fe(I1) are consumed per O2 resulting in no radical formation. A balanced stoichiometry for the proposed reaction scheme in Fig. 4 is 3.2 Fe(II), 0.8 Tyrl", 0.8 RRB2 consumed per mol of dioxygen reduced and a total of 1.25 mol of O2 consumed per mol of RRB2, assuming equal rates for all processes. These numbers are in good agreement with our experimental data for wildtype RRB2 reported in Table I.
The overall expected yield for the scheme in Fig. 4 is 1.0 tyrosyl radical per RRBB protein and thus appears to account for the lower than expected tyrosyl radical content consistently observed for the isolated RRBB protein. Reconstitution of RRBZ,,, in the presence of the crude extract from E. coli has generated the highest reported yield of 1.4 tyrosine radicals per RRB2 protein (10). This result implies that the presence of the three protein components of the cell extract described by Reichard et al. (12) must alter the reaction conditions to afford more efficient radical production. We are currently exploring how the stoichiometry of the oxygen consumption associated with oxidation of the diferrous clusters in RRB2 is affected by other protein components and by structural perturbations engenedered by appropriate mutations.