Iron and Free Radical in Ribonucleotide Reductase

Ribonucleotide reductase from Escherichia coli consists of two nonidentical subunits (Protein Bl and Protein B2), lcoth required for activity. Protein B2 contains 2 iron atoms and a lesser, variable amount of a paramagnetic species, characterized by g = 2.0047 EPR and 410 nm electronic absorptions. Except for the 410 nm band, Protein B2 much resembles oxyand methemerythrins in electronic spectra. The paramagnetic species is probably organic free radical rather than metal, is dependent upon the presence of iron in Protein B2, is directly correlated with enzymatic activity, and is destroyed by NH20H and hydroxyurea (EHRENBERG, A., and REICHARD, P. (1972) J. Biol. Chem. 247, 3485-3488). We here describe complete removal of iron from Protein B2, reconstitution to active metalloenzyme, and Mijssbauer spectroscopy of [“7Fe]Protein B2. Addition of excess Fe”+ to metal-free protein yielded reconstituted/reactivated protein with 2: 1 iron:subunit stoichiometry and a hemerythrinlike electronic spectrum, but with 120 to 39Ocj, of the original specific enzymatic activity and correspondingly increased 410 nm and EPR absorptions. NHPOHor hydroxyurea-inactivated protein was also converted to highly active enzyme subunit by the same iron substitution procedures. Miissbauer and electronic spectra suggested that Protein B2 contains 2 nonidentical high spin Fe(II1) ions in an antiferromagnetically coupled binuclear complex that resembles both methydroxohemerythrin and oxyhemerythrin. Below 195°K the iron in Protein B2 was completely diamagnetic, in confirmation of the previous assignment of the 410 nmand EPR-absorbing species as stable organic free radical of unknown structure.

We propose that the function of iron in Protein B2 is the initial generation of radical from a protein-bound group, and that stability of the radical depends upon some continuing interaction with the iron center. The radical is apparently less stable than is metalloprotein structure: this would account for the variability of radical content in native protein and for the > 100% yield of radical obtainable by substitution of the iron.
We further propose that the free radical participates in the reduction of ribonucleotides by the enzymatically active Protein Bl-Protein B2 complex.
Ribonucleoside diphosphate reductase from Escherichia co2i uses the reduced form of thioredoxin,' a small protein, as the electron donor for the irreversible reduction of the four commori ribonucleoside diphosphates (1,2). The ribonucleotide reductase system consists of the enzyme plus thioredoxin and thioredosin reductase, as shown in Fig. 1. The enzymatically active form of ribonucleotide reductase is a 1 :I complex of two dissimilar subunits, called Proteins Bl and B2 (3)(4)(5).
Either prot,ein possesses no known enzymatic activity in the absence of the other subunit (3).
Per molecule of 78,000 daltons, Protein 132 contains two identical or nearly identical polypeptide chains that. do not dissociate except under denaturing conditions, and 2 iron atoms (5, 6). Except for a sharp 410 nm band of widely variable absorbance (in different near-homogeneous preparations; see Refs. 6 and 7), the electronic spectrum of protein 82 at >320 nm much resembles spectra of oxy-and met-forms of hemerythrius,2 intracellular Oz-carrying proteins from certain species in four primitive animal phyla (8,9).
In addition to iron, Protein B2 coutains a variable amount of a paramagnetic species that gives a characteristic EPR ab- sorption centered at g = 2.0047; double integration of the EPR signal aud assumptjion of spin one-half yielded ~0.6 unpaired electron per subunit molecule (7). The EPR and 410 nm absorptions and the specific enzymatic activity of Protein B2 were linearly correlated properties, and treatment of the protein with hydroxyurea or hydroxylamines concurrently destroyed all of these properties (6,7,10). From characteristics of the EPR signal, the 410 nm and EPR species was tentatively assigned as stable organic free radical rather than as metal (7).
We here report Miissbaucr studies that, along with complementary magnet,ic studies to be reported elsewhere,3 confirm the assignment of the paramagnet,ic species in Protein B2 as organic free radical.
For convenience we shall henceforth refer to the EPR-and 410 run-absorbing species as "radical."
Neutral NHzOH solution for inactivation of Protein B2 was prepared immediately before use from solid NHZOI-I-HCl and aqueous NaOH.
Other chemicals were of the highest quality obtainable.
Presoaked SM 13200 collodion bags and SM 16304 glass holders (vacuum dialysis apparatus from Sartorius-Membranfilter GmbH) Ivere used for the iron removal procedure and other dialyses.
Proteill B2 was prepared and assayed for enzymatic activity according to Brown et al. (3). Originally Protein Bl was prepared as described earlier (3); in later experiments we used a a C. L. Atkin, A. Ehrenbery, and T. IT. Moss, manuscript in preparation. modified procedure (5). Purified &. coli thioredoxin (1, 2) was a gift from A. Holmgren (Karolinska Institute) and thioredoxin reductase was purified as before (11).
Electronic Spectra and Protein Determinations of Protein B2-Ultraviolet-visible spectra were made with microcuvettes in a Cary 14 spectrophotometer fitted with a Cary 1443400 microcell attachment.
Spectra from 240 to 700 am were usually obtained at room temperature on 0.02 to 0.2 mg of Protein B2 in 0.15 ml of buffer by use of both 0 to 0.1 and 0 to 1.0, 1.0 to 2.0 absorbance unit full scale slidewires.
Iron Assays-The calorimetric ferroin method of Massey (12) was scaled down to final assay volumes of 125 ~1. o-Phenanthroline was replaced by an equivalent concentration of sodium bathophenanthrolinesulfonate for b ureater sensitivity and trichloroacetic acid was a special low iron product.
Blair and Diehl gave es35 = 22,140 &f-l cm-l for the absorption maximum of the Fe(I1) complex of this chelator (13). However, the order of mixing precluded cuvette corrections before addition of iron, and we therefore measured A535 and A650 as well as ran blank assays. Standard curves with iron wire dissolved in aqueous HCl were taken with each run and gave ecPS -eesO = 21,700 M-l cm-l.
Protein B2 with exactly 2 iron at,oms per subunit contains 0.143% iron or 25.6 ng atoms iron per mg, according to Thelander's molecular weight data (5).
Saturated Lithium S-Hydroxyquinoline-M'ulfonate Solution for Removal of Iron from Protein Bg-A suspension of 100 g of 8-hydroxyquinoline-5-sulfonic acid dihydrate in 700 ml of HZ0 was dissolved and neutralized by addition of approximately SO ml of saturated aqueous LiOH and warming to 60". The solution was filtered, cooled, and repeatedly extracted first with CHCl, and then with Et20 to remove odiferous impurities and dark residue at the phase interfaces.
The aqueous phase was vacuum-evaporated to 300 to 400 ml aud residual ether was removed by boiling.
The resulting solution was supersaturated at room temperature but required generous seeding and long standing (2 1 week) to precipitate yellow crystals at equilibrium with clear dark orange saturated solution of the salt at pH 7.0. The saturated solution at 25" was 0.4i .\I in lithium %hydroxyquinoline-5-sulfonate (by dry weight). The stock mixture of crystals and solution was stored protected from light. The method developed for removal of iron from Protein 1~2 (see "Results") employed a IO-fold dilution of the stock saturated solution of chelator.
["'Fe]Protein B2 was prepared and precipitated as described under "Results"; the precipitate was evenly suspended in part of the supernatant solution and the resulting thick suspension (0.70 ml total volume) and a blank sample (0.70 ml of the supernatant solution) were frozen with powdered solid CO? in polyethylene Xjssbauer cells (14) that gave 3.2.mm sample thicknesses in the y-beam direction.
These samples, prepared in Stockholm, were packed in solid CO2 and transported immediately to HarLvell for Miissbauer studies.
The blank, the [j'Fe]Protein B2 sample, and smaller fractions subsequently derived from the latter (see "Results") and frozen with liquid Nz in Miissbauer cells were individually masked with lead foil and maintainrd at 5 195°K during spectroscopy or storage.
Counters, other equipment, and sample handling technique have been described in detail (14,16). The y-ray sources were "7Co diffused in palladium foils. Velocity calibrations were rnade with "'Fe-enriched iron foil absorbers at room temperature, and velocities herein are specified relat,ive to the center of symmetry of their spectra (17). The background rate was determined at the start of each run by insertion of an aluminum filter in the y-ray beam (14) ; the absorption scales on Fig. 5 are background corrected.
Most of the runs were made with the samples in low transverse magnetic fields from permanent magnets (cf. Refs. 14 and 18); these low fields in fact had negligible effect since iron in the present case turned out to be diamagnetic. High transverse magnetic field (30 kG) was obtained with a superconducting solenoid.
Low field spectra were run for -48 hours at each temperature, and the high field spectrum required -72 hours running time.
This gave about 6 X lo6 and 4 x lo6 counts per channel for spectra at' low and high applied magnetic field, respectively.
Least squares computer fits to low field da.ta were made using a program written by B. Window.4 The high field spectra were calculated using a program which computed the Mossbauer spectrum averaged over all possible orientations of the molecule relative to applied field and y-beam directions.j

Stability of Iron and Radical in Protein B%-Protein
132 was insensitive to a number of chemical treatments that effect distinct oxidation-reduction or ligation changes in other iron proteins. Thus, neither the enzymatic activity nor the ultraviolet spectrum (see later) of Protein B2 in 0.5 M Tris-HCl, pH 7.6, was affected by the following: deaeratioa-oxygenation cycling; treatment at 25" for 1 hour with 1 atm CO in the dark, or with the reductants mercaptocthanol, dithiothreitol, NaBH4, or Na2S20b; treatment for 2 days at 0 to 4" with 1 nr KF, 0.1 M KCN, 0.01 M NaNa, or with a variety of 0.001 M siderochrome polyhydroxamate chelators (19) and ferroin chelators (20) with or without ascorbate or dithiothreitol reductant.
Higher concentrations of NaNa caused inactivat,ion of an unknown sort,, but observed gradual losses of Protein B2 absorption at >320 nm excluded the presence of anything resembling metazidohemerythrin (cf. Ref. 9). Passage of Protein B2 through columns of sodium Cheles 100 resin at pH 6.5 to 9.0 was without effect. Even the very powerful chelating resin Nopchelate A failed to remove iron from Protein B2 at pI-I 6.5 to 9.0 but caused inactivation at the lower pH, apparently because the resin hydrolyzed to yield NH20H.
Iron in Protein B2 was insensitive to H202 and other oxidants, but some of these destroyed radical and activity. pH values much outside the range 6.5 to 9.0 caused losses of iron and activity (cf. Ref. 6).
The state of the iron in Protein B2 was not detectably affected by treatment with Protein Bl, thioredoxia-(SH)I or -SZ, or any other components or combination of protein components, substrates, and allostcric effecters of the complete ribonucleotide reductase system.
The radical and associated enzymatic activit'y decayed spontaneously with a half-life of roughly 2 weeks for concentrated Protein B2 (210 mg per ml) stored at 4" in 50 mbf Tris-HCl, pH 7.6; decay was much faster at higher temperatures and dilutions.
Aged Protein B2 resembled parti&\-NHZOH-in activated protein in ultraviolet spectra and in capacity to be reactivated (see later).

Removal of Iron from Protein Bs-Two
previous procedures resulted only in partial iron depletion and caused loss of denatured, insoluble protein (6). We sought conditions to cause rapid dialysis of 5"Fe label from [jgFe]Protein B2 (6) to yield metal-free protein that could be reconstituted.
In a typical experiment 50 to 200 ~1 of Protein B2 solution (10 mg per ml) were pipetted into the tip of a collodion vacuum dialysis bag, and the bag was then dipped to the level of the internal liquid into 10 ml stirred chelator solution.
With the [""Fe]Protein B2, radioactivity outside the bag approached a constant level after 3 hours dialysis.
Dialyzed protein could be vacuum-concentrated at least 2-fold directly within the bag before filtration through a small column of Sephades G-25 equilibrated with 50 mM TrisHCl, pH 7.6. This method gave, in 80 to 90% yield, protein with less than 2% of the original iron content.
From here on this metal-free Protein 132 will also be called Apoprotein 132.
Imidazole and 8hydrosyquinoline-5-sulfonate were both essential to the procedure.
Rather than serving as an iron ligand, imidazole more likely induced conformational changes that exposed iron to chelator (see Ref. 21). NH2OH was not essential but gave marginally faster removal of iron.
Properties of Apoprotein B%-Metal-free protein obtained by the above procedure had no enzyme activity, showed great loss of absorption above 320 am as compared to native protein (see later), and gave no EPR signal; similar properties were reported for partially iron-depleted protein (7). Apoprotein prepared from NH2011-or hydroxyurea-inactivated Protein 1~2 (cj. Refs. 6, 7, and 10) appeared identical with apoprotein prepared from native Protein B2 (with omission of NHSOI-I in the foregoing procedure) and could be stoichiometrically reconstituted to highly active metalloenzyme.
By the criterion of reconstitution to active enzyme (XC below), apoprotein was as stable as native Protein B2. Aliquots of apoprotein taken over a 4-day period from a solution (5 mg per ml) stored at 4" gave reconstituted fractions wibh identical ultraviolet absorbance ratios and specific enzymatic activities. i2poprotein could be precipitated with 3.2 M (N&),80+ frozen and stored in liquid N2, thawed, and recovered in desalted solution with the same good efficiency as could native protein (3).
Apoprotein resembled native protein in several other respects. Both forms showed the same mobility during (nonstacking) disc gel electrophoresis at ~1-1 8.7 (3) and in sucrose density gradient centrifugation.'j One difference b t e ween apoprotein and native protein, however, appeared upon reaction of sulfhydryl groups with excess 5,5'-dithiobis@nitrobenzoate) under nondenaturing conditions. Fig. 2 shows that only two such groups were oxidized per molecule of native protein, while additional -SH groups were slowly titrated in apoprotein.
The same experiment with the addition of 8 tir guanidine HCl (not shown) gave 8.4 and 7.2 rapidly titrating sulfhydryl equivalents per mole of apoprotein and native protein, respectively, approaching the value of 10 half-cystincs found by amino acid analysis (5) Anaerobic reaction at 25" was performed and followed suectrouhotornctric:allv as described elsewhere for Protein Bl (5) . _ except that, in the experiments shown, we omitted guanidine HCl.
The uroteiu fractions used are fwther described in Table  I, Batch 1. L ascorbute and 10 Mel Fc(~II~)L(SOl)a.6E-I,O was freshly prepared in dcaerated 50 IllhZ Tris-HCl, pH 7.6, from weighed-out salts and left standing several minutes to permit reduction of any residual Fe(II I). The solution turned deep purple due to formation of nnFc(II)-oxidized ascorbate comples (22). When j7Fe or j"Fe in I-I<:1 solution was the source of 10 InM iron, a similar procedure was used but buffer concentration was increased to yield a final 111-I near 7.6, and ascorbate was increased to 150 mM. X volume of Fe(II)-ascorbate mixture, containing 50 to 100% es(aess iron, was added to Apoprotein U2 solution and after 5 miu at 25" the mixture was filtered immediately at 4" through a column of Scphades G-25 equilibrated with 50 mM Tris-HCl, pH 7.6. 111 a typical cxpcriment, 4 11 of Fe(U)-ascorbate misture were added to 100 ~1 of protein solution (10 mg per ml) and filtered through a gel column of l-ml bed volume.
Scaling the experiment up by a factor of 10 or more worked equally well. Protein yields after the gel filtration approached 90%. Specific enzymatic activity of fully reconstituted preparations was always 120 to 390% of that of native starting material, but repetition of the iron removal and reconstitution sequence gave decreasing activity.
We now routinely improve activity of purified Protein 132 by the described iron substitution procedures and have obtained specific activities up to 30,000 units per mg, compared with the best activities of 20,000 units per mg found earlier for native Protein B2 (cf. Rcfs. 3 and 7). Table I shows typical analytical results for native, inactivated, metal-free, and recoilstituted Protein 132 (Eatch 2 refers to the Miissbauer esperiment below, and a comparison of ultraviolet spectra is given later).
Ascorbate was unnecessary for reconstitution/reactivation of apoprotein with various ionic Fe(I1) salts but was necessary to prevent adventit,ious binding of iron (cf. Ref. 6). The purple color of the Fe(II)-ascorbate mixture described above visibly decreased as iron was taken up by apoprotein, but was still faintly visible as gel filtratiou was begun. Ascorbate used therefore nppcared sufficient to maiutain excess iron in the more soluble ferrous state and to prevent advcnt,itious binding; this was coilfirmed by iron assays such as shown in Table I  studies of reconstituted protein that showed negligible g = 4.3 iron and magnctic susceptibility corresponding to that espected from radical content. of the protein according to EPR.3 Treatment of apoprotein with limiting amounts of "9Fe gave partially reconstituted fractions with specific enzymatic activities directly proportional to iron conteut (Fig. 3). Chemical and radioactivity assays for iron agreed. Specific absorbances at 365 and 410 nm, as well as the rest of the characteristic spectrum of active Prot,ein B2 at >320 nm, also increased proportionately with iron content (not shown).
In light of our assignment of much of the electronic spectrum-notably A365to a hinuclear iron center (see "Discussion"), it appeared that only binuclear and no mononuclear iron centers were formed in partially reconstituted protein.
If activity were indeed associated only with a binuclear iron structure in Protein B2, random rather than pairmise binding of iron ions to apoprotcin would have given activity proportional to the square of iron content (dotted line in Fig. 3 isolated, Protein R2 has a highly characteristic spectrum between 320 and 540 nm, with a sharp peak at 410 nm and broader peaks or shoulders at -320, 365, -380, aud -480 urn (6, 7); a very weak, broad band near 600 nm-similar to that in oxy-and methemerythritls (9)-may be present but was near our limit of detection (E z 200 ;\2-l cm-l).
Spectra of native Proteiii U2 in phosphate 01 Tris buffers at pH 6.5 to 9.0 did not differ significantly.
Inactivatiou of Protein U2 by hydrosylamine or hydrosyurea-which does not result in the loss of iroll-caused loss of the 410 nm band aud smaller losses in absorption near 380 nm and elsewhere (6, 7, 10). Fig. 48 gives a spectrum of native Proteiu 132. N11*0fI-illactivatcd protein is shown in Fig. 4B and a difference spectrum betwecu native and inactivated protein is given in Fig. 4C. Fig. 4, A and B (23)) ; and deoxy-and metal-free hemerythrins (---). Spectra A, experimental spectra (dots) and calculated spectra (solid lines) of a frozen suspension of citrate-precipitated enzymatically active [s7Fe]Protein B2 at given temperatures and applied transverse magnetic fields: a, solvent blank (supernatant from precipitated protein) at 4.2"K and 0.55 kG; b, c, and d, protein at 195, 77, and 4.2"K and 0.39, 0.39, and 0.55 kG, respectively; e, protein at 4.2"K and 30 kG [the solid line spectrum shown was calculated from low applied magnetic field spectrum d above (data in Table II) for only direct effects of high applied field on two equal populations of diamagnetic iron with negative "electric field gradients" (see "Results")]; f, calculated contribution to theoretical spectrum e from the iron population responsible for the outer pair of quadrupole split absorption lines in spectra at low applied magnetic field; g, similarly calculated contribution to e from the inner low field pair. Spectra B, experimental spectra of [67Fe]Protein B2 samples derived from that shown above; a, solution of protein after inactivation with NH20H (see "Results"), at 77°K and 0.39 kG; b, solution of protein after partial inactivation with hydroxyurea (see "Results"), at 4.2"K and 0.55 kG; and c, remaining suspension of citrate-precipitated enzymatically active [67Fe]-Protein B2, at 4.2"K and 0.55 kG. 0.63 pmole total QFe at a mean concentration of 750 to 900 pM. Contamination from solvent components was <0.2 PM 57Fe, and a Mijssbauer spectrum of supernatant solution from precipitated protein showed no absorption (Fig. 5Aa).
Miissbauer spectra of ["Fe]Protein B2 are shown in Fig. 5A, b to e.l The low field spectra strongly resembled that of 7 See reviews by Lang (14) and Debrunner (24) for general discussions of MCissbauer spectroscopy of iron proteins.
oxyhemerythrin (25)(26)(27) ; the presence of four lines of approximately equal intensity indicated two quadrupole-split pairs of absorptions, hence two inequivalent iron sites. The line widths were typical of iron in proteins.
The absence of magnetic features in the low temperature low field spectrum (Fig. 5Ad) suggested that no significant half-integral spin resided on the iron atoms. Assignment of the four lines to a narrow doublet at negative velocity and another doublet. at high positive velocity could be excluded.
The former would require an unusual high oxidation state of iron; the latter would be characteristic only of high spin Fe(I1) with regard to isomer shift while the quadrupole splitting would contradict such an assignment.
The successful simulation of the high field spectrum, described below, confirmed the identification of inner and outer quadrupole doublets. In least squares fits to the low field spectra, both lines of a doublet were constrained to have equal intensity and line width, but relative intensity of the doublets was allowed to vary: the results are presented in Table II. The quadrupole splittings were temperature-independent while the isomer shifts varied in a way characteristic of second order Doppler effect. Deviation of the intensity ratio from unity and its variation with temperature indicated different recoil-free fractions, hence different vibration amplitudes for the two classes of iron nuclei.
The high field spectrum of Fig. 5Ae provided a test for integral unpaired electron spin on the iron as well as a measure of the sign and symmetry of the electric field gradient interaction. Visual inspection suggested that only the direct effect of applied magnetic field on the iron nuclei was present, i.e. that the iron sites were diamagnetic.
Assuming such diamagnetism, we made a least squares fit with the program described under "Experimental Procedure." Quadrupole splittings, line widths, and isomer shifts were taken from the low field spectrum at 4.2"K. The electric field gradient was taken as negative (i.e. V,, < 0) at each sit'e, and the asymmetry parameters (q values) were allowed to vary. The best fit (curve in Fig. 5Ae) corresponded to 17 of -0.5 and +0.6 for outer and inner doublets, respectively. The mean square difference between this theoretical spectrum and the data points was only 1.3 times the value expected from counting statistics, so we regarded the fit as satisfactory.
The r] values, however, were not well determined: setting both to zero increased the mean square error by only 1%. Calculated individual contributions of the two types of iron are shown in Two features of the spectra further support the above evidence that iron in Protein B2 is diamagnetic at low temperatures. First, we estimate that a minority fraction of at most 20% of the iron in the sample could in principle have yielded a paramagnetic spectrum that was lost in the noise level of low field spectra, provided that it behaved as do known dilute ferric complexes. We would, however, have detected any paramagnetic iron at the concentration of the 9adical" (=0.55 unpaired electron per protein molecule) as determined by EPR.3 Second, the following theoretical estimate of absolute absorption strength shows that much or all of the iron in the sample contributed to the observed absorptions.
The fractional absorption of thin absorbers is given by the product of surface density of 5?Fe, peak absorption cross-section a,, the product f.f' of source and absorber recoil-free fractions, and two factors depending on the line widths and structures of emission and absorption spectra. Surface density was 1.4 to 1.7 x 1On nuclei per cm2 (from "Fe concentration estimate above and 3%mm sample thickness). We take c, = 2 x 1018 cm2 and use the rough estimate f.f' = 0.5 which is tvpical of hemoalobin at low temperature (16). Values for the [bTFe]Protein B2 subunit of E. coli ribonucleotide reductase from least squares computer fits to low applied magnetic field spectra b to d of Fig. 5A. "Inner" and "outer" refer to quadrupole-split doublets. Zero-field data for Golfilzgia gozilrlii oxyhemerythrin are taken from Garbett et al. (27). Procedure"). b J>efinite signs given only as determined for signs of the principal electric field gradients (V,, values) by sim\llat.ions of Rliissha\ler spectra in high applied magnetic fields.
If all lines lvere of natural width (0.19 mm 0) the next factor would be 0.5; we reduce this to 0.4 since the lines were slightly wider than natural.
Because the absorption is divided into four lines the last factor is 34. The predicted maximal absorption is thus 1.4 to 1.7a/,, where the spread does not include the uncertainty in f.f'. The maximal observed absorption (Fig.  5Ad) was about 1.4'%, well within the expected range.
Diamagnetism in a mononuclear iron complex would normally signify low spin Fe(H).
However, when a pair is present we have the possibility that 2 metal atoms of equal spin will couple antiferromagnetically to yield zero net spin. If such coupling occurs in Protein B2, it is unlikely that the iron is either low spin Fe (U) or high spin Fe(I1) because these are characterized by low lying excited orbital st'ates and resulting temperature-dependent quadrupole splittings. It might be argued that the present case does correspond to one of the above, and the temperature dependence is extremely small because the highly distorted environment has given rise to an unusually large orbital splitting.
Such a splitting would, however, suppress the mixing of orbital levels by spin-orbit coupling, and we would expect to see a quadrupole splitting characteristic of a pure crystal field orbital, approsimately 3.5 mm per s. Thus, the temperature independence of the quadrupole splitting combined with the fact that it is so small tends to argue against a low spin Fe(II1) or high spin Fe(H) assignment.
The high spin ferrous assignment is also unlikely in view of the isomer shift which is exhibited by B2. Most high spin ferrous isomer shifts are greater than 0.8 mm per s. The smallest of which we are aware is the ferrous component of spinach ferredoxin, which has an isomer shift of 0.54 mm per s at 4.2"K (28,29) probably associated with the fact that the ferrous atom is closely coupled to a ferric second neighbor.
The corresponding values for the two B2 components are 0.53 and 0.45 mm per s. The observation of net zero spin requires that both B2 irons have the same spin, so we camlot allow just one of them to be high spin ferrous.
Isomer shifts alone do not allow us to choose between the remaining alternatives-low spin Fe(I1) or spin-coupled binuclear high spin Fe(III)-while the large quadrupole splittings are unusual but not unheard of for either st'ate. The splittings and negative V,, values imply strong asymmetric bonding that produces a planar defect (or axial excess) of electron charge relative to the zero order symmetrical distribution which characterizes both low spin Fe(I1) and high spin Fe(II1).
In magnitude of quadrupole splittings Protein 132 resembles osyhemerythrin (Table  II), methemerythrins, and oxo-bridged six-coordinate binuclear high spin Fe(II1) complexes wherein antiferromagnetic coupling was convincingly demonstrated by the appearance of paramagnetism at high temperatures (25-27, 3O).x Xlbhough the BZZ signs have not been determined for oxyhemerythrin, they were found to be negative in metisothiocyanatohemerythrin and the model compound (Fe(salen)Cl)z (25,27). It must be recognized that the concept of valence state is an idealization and seldom applies to a real chemical compound without at least some qualification.
In the present situation the spin coupling has even confused the usually distinct division between integral and halfintegral spin. By describing the B2 system in terms of a coupled high spin ferric binuclear complex we imply that a low lying excitation exists corresponding to the breaking of this coupling. This has not as yet been observed in B2 although, as mentioned above, the closely similar hemerythrins have exhibited it,.

Spectra of Inactivated
Protein &Z-The protein suspension used for spectra depicted in Fig. 5A was divided into three portions.
One portion (0.4 ml) was immediately refrozen in liquid NP and served as a control for freezing artifacts and radiation damage; its spectrum ( Fig. 5Bc) was unchanged as compared to the low field spectrum taken earlier at 4.2"K ( Fig.  5Ad).
This sample was then maintained at 5 195°K and later used for magnetic studies.3 Protein precipibates from the other two portions (0.15 ml each) were separately dissolved in 50 mM Tris-citrate, pH 7.6, and inactivated for 30 min at 4" with 10 mM NHZOH and 20 mM hydroxyurea, respectively. These manipulations involved about a-fold dilution of the original volume.
Miissbauer spectra of frozen solutions of the two treated samples are given in Fig. 5B, a and b. Even though these spectra are quite weak it is clear that their general features are the same as those of the untreated control.
EPR measurements and recordings of the 410 nm absorptions of the two illactivated samples demonstrated that the NHzOH treatment had nearly completely destroyed the radical, while the hydroxyurea-treated sample was only about 50% inactivated. These experiments further demonstrate that the radical had no effect on Miissbauer spectra of "7Fe in Protein B2.
by guest on March 23, 2020 http://www.jbc.org/ Downloaded from &phosphates) and allostcric effecters ( =nucleoside triphosphates) (4). The site of interaction between enzyme and the electron donor ( = thioredoxin) is unknown. hn obvious question is whether the second subunit, Protein B2, is an electron carrier and interacts wit,11 thioredosin.
Protein B2 contains 2 atoms of iron (6) and a lesser, variable amount of a paramagnetic species that from EPR studies was presumed to be au organic free radical (7). The main purpose of the present work was to define further the structure of iron and its relationship to the paramagnetic species. The result,s of Mossbauer spectra of ["7Fe]Protein B2 together with magnetic susceptibility studies3 clearly establish that the iron is diamagnetic and distinct from the paramagnetic entity, thus confirming the previous assignment (7) of the paramagnet'ic species as stable organic free radical.
These and previous results (6,7,10) also show that the enzymatic activity of Protein B2 depends upon the presence of radical and that the radical in turn depends upon the presence of iron. Removal of iron from Proteili B2 yielded diamagnetic, enzymatically inactive "izpoprotein 132." Treatment of apoprotein with excess Fez+ gave stoichiometrically reconstituted metalloprotein in which activity and radical had been regenerated wit'11 more than 100% over-all yield.
The mechanism of inactivation of Protein B2 by NHzOH and hydrosyurca clearly involves destruction of radical: hydroxyurea was recently found to be a powerful radical scavenger in another system (31). Iron is not removed nor apparently affected in any way by treatment of Protein B2 with these agents. Similarly, aging preparations of native protein showed spontaneous gradual losses of act,ivity and radical with no losses of or apparent changes in iron. Reactivation of protein was achieved only by removal of iron followed by recon stitution.
Other components of the enzyme system, osidationreduction agents, or treatrnent of inactivated protein with Fe(II)ascorbate showed no effect.
At present, the radical is characterized only by its EPR and magnetic susceptibility properties (7)3 and by its electronic spectrum (Fig. 4C), which last we assign from the port.ions of the spectrum of active Protein B2 that are sensitive to NHzOH or hydrosyurea.
'The electronic spectrum of Protein B2 after destruction of the radical with NH20I-T (Fig. 4B) showed absorption bands at 320 and 365 nm in close resemblance to the spectrum of methydroxohemerythrin (Fig. 40). We assign absorption of inactivated Protein B2 at >320 nm directly to the iron center. The studies of Williams' and Klotz' groups on hemerythrins and model binuclear high spin Fe(II1) complexes (9, cf. Ref. 25) have provided strong evidence for binuclear iron centers in oxy-and methemerythrins, and they have assigned different absorption bands to specific bridging ligands.
The similarity of electronic spectra of Protein 112 and hemerythrins suggests the presence of a binuclear iron center also in our case, but we have no independent evidence for the presence or nature of any bridging ligands for iron in Protein B2. A~280 = 9000 M-l cm-l absorbance difference between metal-free and native Protein B2 (Fig. 4) is consistent with loss of intense absorption bands in this region from a binuclear iron center (see Ref. 9).
Our Mbssbauer results were again very similar to those for oxidized forms of hemerythrin  and were consistent only with two alternative struct'ures: (a) two low spin Fe(I1) complexes in two sites per Protein B2 molecule; or (b) one antiferromagnetically coupled binuclear high spin Fe(II1) complex per Protein B2 molecule.
Because of the similarities of Protein B2 and oxidized forms of hemcrythrin in both electronic and Mdssbauer spectra, we strongly favor a binuclear high spin Fe(II1) structure for iron in Protein B2. There is one important known difference between hernerythrins and Protein B2: in hemerythrin one subunit consists of a single polypeptide chain with 2 bound iron atoms (8) whereas in Protein B2 one subunit contains two apparently identical polypel)tide chains (5) and 2 iron atoms. It appears likely to us that, in the latter cast each iron is bound to a separate polypeptide chain but we have no evidence on this point.
Our Xjssbauer spectra showed the presence of two equal populations of iron. WC therefore assume that, as in the proposed oxyhemerythriii struct,urc (27), each binuclear iron complex in Protein B2 is intrillsically asymmetric and contains 1 iron ion of each class. Radical content clearly had no effect on the distribution into two iron populations.
What is the function of iron in Protein B2? The rnetal does not appear to participate as an electron carrier in t,he enzyme reaction.
The electronic spectrum of Protein B2 was not affected by addition of thioredoxin or any other components-alone or in combination-of the ribonucleotide reductase system. Furthermore, the part of the electronic spectrum due to iron was not influenced by a variety of oxidizing or reducing agents.
Our results also argue against a major structural role for iron in Protein B2. Thus, the apoenzyme and the native protein had identical electrophoretic and ultracentrifugal prollerties and both formed 1: 1 complexes with Protein Ul in the presence of Mg2+ (4, 5)." The apoprotein had, however, more exposed -SH groups than did Protein 1~2.
Instead, we suggest the hypothesis that the function of iron is to generate an organic radical from a protein-bound group as the metal binds to apoprotein and to stabilize that radical in the metalloprotein by some continued interaction. Radical formation might involve iron-catalyzed l-electron aerobic osidation of the radical precursor, or a l-electron reduction of the precursor by Fe(I1) to leave Fe(HI) in the protein2 r2ccording to this hypothesis the organic radical is the functional coiitribution of Protein 1122 to the enzymatically active Protein Bl-Protein 132 complex and the reduction of ribonucleosidc diphosphates by the B. coli enzyme would thus involve the participation of radical intermediates. The identification of the chemical nature of radical iii Protein 1~2 is now an important task.
In this connection we would like to recall that Blakley et al. recently postulated the involvernent of transient free radical intermediates in the reduction of ribonucleoside triphosphates by the reductase from Lactobacillus leichmannii (32,33). 111 this case cob(II)alamin coenzyme and not iron is part of the enzyme system and radical formation is thought to involve a homolytic scission of the cobalt-carbon bond of the coenzyme. Both the E. coli and the L. Zeichmannii enzymes catalyze the stereospecific replacement of the OH group of a secondary alcohol, i.e. reduction with retention of configuration (1,(34)(35)(36)(37). In organic chemistry there are to our knowledge no general methods for the direct reduction of alcohols. However, catalytic hydrogenolysis reduces substituted benzyl alcohols with retention of configuration and might be a model reaction (37)(38)(39). The active reductant in catalytic hydrogenolyses is likely to be chemisorbed hydrogen atom radicals (40).