Tetrahedral Iron in the Active Center of Plant Ferredoxins And

Proc. Nat. Acad. Sci. USA Vol. 68, No. 12, pp. 3015-3020, December 1971

Tetrahedral Iron in the Active Center of Plant Ferredoxins and Beef Adrenodoxin* (iron-sulfur proteins/rubredoxin/near-infrared circular dichroism/ligand-field spectrum/EPR) WILLIAM A. EATONt, GRAHAM PALMERI, JAMES A. FEEt§, TOKUJI KIMURA'[, AND WALTER LOVENBERG11 t Laboratory of Physical Biology, National Institute of Arthritis and Metabolic Diseases, National In- stitutes of Health, Bethesda, Md. 20014; t Biophysics Research Division, Institute of Science and Tech- nology, University of Michigan, Ann Arbor, Mich. 48105; 'I Department of Chemistry, Wayne State University, Detroit, Mich. 48202; and 11 Experimental Therapeutics Branch, National Heart and Lung Institute, National Institutes of Health, Bethesda, Md. 20014 Communicated by Harry B. Gray, September 30, 1971

ABSTRACT The coordination structure of the iron- with a net spin of 2. Recent reviews by Tsibris and Woody sulfur complex in spinach ferredoxin and adrenodoxin is Palmer and have discussed the chem- investigated by optical spectroscopy. The circular-dichro- (6) and Brintzinger (7) ism and absorption spectra of these two-iron iron-sulfur ical, magnetic resonance, magnetic susceptibility, and M6ss- proteins reveal weak electronic transitions in the near- bauer studies, which have all been consistent with this model, infrared wavelength range, 0.8-2.5 ,Am (12,500-4000 cm-l). though no demonstration of the coordination geometry of On the basis of the low absorption intensities and large the iron-sulfur complex was possible. The basis for investigat- anisotropy factors, d d transitions of the iron can be identified in the reduced proteins at about 4000 cm-' and ing this and other aspects of the proposed structure by optical 6000 cm-. The low energy of these one-center ligand-field spectroscopy was recently developed by Eaton and Loven- transitions, together with the similarity to the ligand- berg, from their absorption and circular-dichroism studies in field spectrum of the one-iron protein rubredoxin, leads the near-infrared on the one-iron protein rubredoxin (8). to the conclusion that the reduced two-iron iron-sulfur and reduced are known from proteins also contain a high-spin ferrous ion in a distorted Oxidized rubredoxin x-ray tetrahedral site. diffraction (9), magnetic susceptibility (10), M6ssbauer (10), and optical studies (8) to contain a high-spin ferric and A currently attractive proposal for the structure of the active high-spin ferrous ion, repsectively, coordinated to four sulfurs center of the two-iron iron-sulfur proteins such as the plant of cysteinyl residues in an approximately tetrahedral complex ferredoxins, adrenodoxin, and putidaredoxin, was put forth (Fig. 1). The d - d electronic transitions of the tetrahedral several years ago, in part to explain the absence of electron ferrous ion, predicted from ligand-field theory to be char- paramagnetic resonance (EPR) of the oxidized protein and acteristically very optically active, were found by Eaton the unusually low average g value observed upon one-electron and Lovenberg at about 6250 cm-' (1.6 jm) (8). In the ab- reduction (4, 5). The proposed structure consists of a bi- sence of a very strong interaction between the irons, the above nuclear complex in which each iron atom is bonded to four model would predict that a similar one-center ligand-field sulfur atoms tetrahedrally disposed, as depicted schematically spectrum exists in the two-iron iron-sulfur proteins. In this in Fig 1. Two sulfur atoms-the "acid-labile sulfides"-are communication we demonstrate that this is indeed the case common bridging ligands to the metal ions, while the four for plant ferredoxins and adrenodoxin, thereby providing terminal mercaptide sulfur atoms are provided by the cys- strong evidence that the reduced proteins contain a high-spin teinyl residues from the protein. It was further proposed (5) ferrous ion in an approximately tetrahedral site. that the oxidized protein contains two high-spin ferric (d$, S = 5/2) ions antiferromagnetically coupled to give a net MATERIALS AND METHODS spin of zero in the ground state, while in the reduced protein a Rubredoxin from the bacterium Cloetridium pasteurianum, high-spin ferrous (d6, S = 2) ion is antiferromagnetically ferredoxin from spinach chloroplasts, and adrenodoxin from coupled to a high-spin ferric ion, resulting in a ground state beef adrenal glands were prepared by previously described

* This is one in a series of papers describing the electronic proper- ties of spinach ferredoxin and other iron sulfur proteins; related publications (refs. 1-3) contain the results of Mossbauer, electron- nuclear double-resonance spectroscopy, and variable-temperature magnetic-susceptibility measurements. A summary of all our physical data, their bearing on a discrete model for two-iron iron- sulfur proteins, and an analysis of the consequences of this model for nuclear magnetic resonance contact-shifted proton resonances will be published in Biochim. Biophys. Acta, vol. 253. § Present address: Department of Chemistry, Rensselaer Poly- FIG. 1. Schematic structure of the active centers of one- and technic Institute, Troy, N.Y. two-iron iron-sulfur proteins. 3015 Downloaded by guest on September 23, 2021 3016 Chemistry: Eaton et al. Proc. Nat. Acad. Sci. USA 68 (1971)

Wavelength (micrometers) 0.8 1.0 1.5 2.0 2.5 light in the two cell compartments. At wavelengths shorter I than about 1.8 pm, the measured circular-dichroism mag- 61-E, (M'1cmr|) RED nitudes at the extrema are probably within 10-20% of the +8 true values. However, only qualitative estimates -+6 Circular Dichroism of the circular -+4 dichroism could be made in the 2.1-2.5 um region because -+2 of high D20 and protein absorption from vibrational overtone transitions and because of considerable deviation from perfect -2 ox circular polarization. The low degree of polarization would E'(M'Icm|) result in a measured value that is lower than the true value Absorption (14). The absorption and circular-dichroism spectra of reduced e (M-1cm-,) rubredoxin and reduced adrenodoxin showed no change during II % the 2- to 5-hr period of data collection. Some samples of ferredoxin exhibited a slowly increasing absorbance over the I ,T 300 entire near-infrared spectral range (0.8-2.5 Mum); this absorbance did not, however, significantly alter the circular- f dichroism spectrum. We have not yet determined the nature of the optically inactive material responsible for this, but it Rubredoxin dI could possibly be a polymeric complex of iron and sulfide, -200 ', which might be expected to show intense near-infrared absorption. In any event, the spectra for ferredoxin reported 1l RED below have no contribution from this undefined material. RESULTS -100 '., II Figs. 2-4 show the near-infrared absorption and circular- dichroism spectra of the oxidized and reduced proteins: rubredoxin, ferredoxin, and adrenoxdoin. In the region from 4000 cm-' to about 9000 cm-', the oxidized proteins exhibit 14,000 12,000 10,000 8,000 only optically-inactive vibrational overtone bands of the Frequency (cm-I) protein and some residual HOD. The broad electronic absorption of the reduced proteins in this spectral region is FIG. 2. Circular-dichroism and absorption spectra of oxidized more clearly seen in the reduced-minus-oxidized difference (OX) and reduced (RED) Clostridium pasteurianum rubredoxin in spectra shown in the middle of the Figures, where the con- D20-0.1 M Tris buffer (pH 7) at room temperature. The middle tribution from the relatively sharp vibrational transitions curve (RED - OX) is the difference absorption spectrum, (reduced rubredoxin) minus (oxidized rubredoxin). Extinction has been subtracted out by use of the oxidized proteins as a coefficients are based on the value 8800 Al-' cm-' for the 4900- reference. band of oxidized rubredoxin (11). For all three proteins, several of the electronic transitions in the near-infrared have large anisotropy factors (greater than 0.01, see Table 1) compared to the spectra in the visible methods (11-13). After two cycles of freeze-drying and soaking and ultraviolet regions (8, 15), where the (point) anisotropy in D20, the optical ratios for rubredoxin, ferredoxin, and factor does not exceed 0.005. Furthermore, there are very important similarities among the spectra of the reduced adrenodoxin were A4go/A2so = 0.35-0.4, A42o/A276 = 0.43- forms of and 0.46 and A414/A274 = 0.65, respectively. These ratios for the rubredoxin, ferredoxin, adrenodoxin, par- oxidized proteins are somewhat less than the best possible ticularly in the region 4000-8000 cm-', where all three proteins values, but should not influence the spectral results in the show weak absorption bands with large anisotropy factors. near-infrared, where the apoproteins exhibit no electronic Quantitative aspects of the spectra are summarized in absorption. Reduction was performed with excess solid sodium Table 1, together with the spectral assignments discussed dithionite and, in the case of adrenodoxin, methyl viologen below. The center of each transition (v) is obtained from either the maximum or the (about 10jum) was added as a redox mediator. absorption circular-dichroism In cases are All spectral measurements were made with D20 as the extremum. where both reasonably well resolved, solvent, since it absorbs much less strongly in the near- the agreement is within 300 cm-'. Most of the oscillator infrared than H20. Spectra could be measured to 1.8 jum strengths (f) and, consequently, the anisotropy factors (g), (5600 cm-') with 25-mm pathlength cells; to penetrate the are only reported to one significant figure, reflecting the con- in of of 2.1-2.5 ,tm (4800-4000 cm-') D20 "window" 2- or 5-mm siderable ambiguity the resolution certain parts the into its cells were used. Absorption spectra were obtained with either a absorption spectra component bands. Similar data Cary 14 or Zeiss DMR 21 recording spectrophotometer. to that shown have also been obtained with the rubredoxin from elsdenii and the ferredoxin Circular dichroism was measured with an accessory to a Cary Peptostreptococcus from 14 spectrophotometer equipped with a pen period control parsley. and a slidewire with 0.05 absorbance units full-range. The DISCUSSION circular-dichroism accessory, previously described by Eaton Metal d - d transitions of optically-active complexes may and Lovenberg (8), uses Glan prisms as linear polarizers have the diagnostic property of exhibiting rather weak ab- and mica plates to create right and left circularly-polarized sorption bands with relatively intense circular dichrosim, Downloaded by guest on September 23, 2021 Proc. Nat. Acad. Sci. USA 68 (1971) Tetrahedral Iron in Ferredoxin 3017

Wavelength (micrometers) Wavelength (micrometers) 0.8 1.0 1.5 2.0 2.5 0.8 1.0 1.5 2.0 2.5 I -I I ,E -E, (M'§cm'1) Circular Dichroism +2_ Circular Dichroism OF EI-6, (Mccm- ) +22d t -2O E'(M-'c m'1) -4L OX RED - 150 -2 _ RED -4_ ox Absorption | 100 Absorption

E E (M'cm'|) 'E (M'lcmg) (M-1 cm-,) 100 400 - 400r \ - 50

RED-OX

I RED-OX I

300 300 F 1. II Ferredoxin II I Adrenodoxin II II 200 2001- I 11I I\ RED ox ", RED 100 1 100

I ,t -.I_ 12,000 10,000 8000 6,000 4,000 12,000 10,000 8,000 6,000 4,000 Frequency (cm') Frequency (cm |') FIG. 3. Circular-dichroism and absorption spectra of oxidized FIG. 4. Circular-dichroism and absorption spectra of oxidized (OX) and reduced (RED) spinach-chloroplast ferredxoin in (OX) and reduced (RED) beef-adrenal-cortex adrenodoxin in ~~~~~~~~~~~~room D20-0.1 M Tris buffer (pH 7) at room temperature. The middle D20-0.1 M Tris buffer (pH 7) at temperature. The middle - curve (RED - OX) is the difference absorption spectrum, (re- curve (RED OX) is the difference absorption spectrum, duced ferredoxin) minus (oxidized ferredoxin). Extinction coeffi- (reduced adrenodoxin) minus (oxidized adrenodoxin). Extinc- cients are based on the value 9400 M-1 cm per molecule for the tion coefficients are based on the value 9800 M-1 cm-' per 4200-A band of oxidized ferredoxin (15). molecule for the 4140-A band of oxidized adrenodoxin (16).

TABLE 1. Spectroscopic data for near-infrared electronic transitions Molecule i (cm-,) 103f* R (DBM)t gt Electronic origin Oxidized rubredoxin 13,400 2.5 -0.008 0.004 S -k Fe charge transfer Reduced rubredoxin <4, 000 - - >0.01§\ Fed-d 6,250 0.96 O.090T 0.050/ Oxidized ferredoxin 10,70011 1 -0.018 0.015 ? Reduced ferredoxin 3,800 1 - \ 11,0005,800 30.65 +0.017-0.051 0.006 ?Fe Fechargetransfer Oxidized adrenodoxin 10,50011 0.9 -0.022 0.038>0.02 ? Reduced adrenodoxin 4,500 0.5 >0.01§\ Fe d d 6,000 0.69 -0.031 0.022/ 11,000 3 +0.015 0.004 ?Fe Fechargetransfer

* The oscillator strength is calculated fromf = 4.32 X 10-9fedi, and is unitless. t The rotational strength is calculated from R = 0.0514/if(el- *r)di, in Debye-Bohr magnetons. IPE -r)d t The anisotropy factor is calculated from g = fedp . § This value is estimated from the point-anisotropy factor, (ei - er)/e, and is a minimum value since the polarization of the light in the 2.1-2.5 /Am spectral region deviates considerably from circular polarization (14). ¶ This value is the sum of the absolute magnitudes of the positive and negative components. 11 Wilson found a distinct maximum at 10,700 cm-' and a shoulder at 12,000 cm-' in the absorption spectra of oxidized ferredoxin and oxidized adrenodoxin at 770K (17). Downloaded by guest on September 23, 2021 3018 Chemistry - Eaton et al. Proc. Nat. Acad. Sci. USA 68 (1971)

compared with, say, weak charge-transfer transitions (18-20). with an anisotropy factor of 0.004, presumably corresponds This occurs because electronic promotions between d orbitals to a sulfur -- iron charge-transfer transition. The d5 high- are inherently electric-dipole forbidden, but may give rise spin ion is unique among the d' ions, in that all the d -) d to substantial magnetic-dipole transition moments (21-23). transitions are forbidden by the strict spin-selection rule, All the transitions listed in Table 1 may be considered as AS = 0 (21-23). Thus, the result that the ligand-field bands essentially electric-dipole forbidden on the basis of their low of reduced rubredoxin "disappear" upon one-electron oxida- oscillator strengths. For octahedral or tetrahedral complexes, tion is consistent with the demonstration by magnetic-sus- ligand-field theory predicts that the lowest-energy spin-allowed ceptibility measurements (10) that oxidized and reduced d d transitions of the d' ions are inherently magnetic-dipole rubredoxin contain high-spin ferric ion and a high-spin ferrous allowed, whereas most of the higher-energy d -- d transitions ion, respectively. are magnetic-dipole forbidden (18-23). The anisotropy factor The transitions near 6000 cm-' in both reduced adreno- (a (E - Er)/e) for magnetic-dipole allowed transitions doxin and reduced ferredoxin have anisotropy factors that generally exceeds a value of 0.01, and provides a powerful are large enough to establish that they are inherently mag- experimental criterion for spectral assignments (18-20, 24, netic-dipole allowed d -- d transitions. With adrenodoxin, 25). Thus, from a combination of absorption and circular- as with rubredoxin, the observed lowest-energy transition dichroism measurements, together with predictions of ligand- exhibits an anisotropy factor greater than 0.01, again in- field theory, it is possible to make rather specific assignments dicating substantial magnetic dipole character, and, therefore, of d d transitions in complex spectra. For a given metal, d -- d transitions.** The low energy of these one-center ligand- the energies at which certain d -o d transitions appear can field bands, together with the similarity to the ligand-field be highly characteristic of the coordination geometry, valence spectrum of ferrous rubredoxin, lead to the conclusion that state, and even type of ligand bonded to the metal (26). the reduced two-iron iron-sulfur proteins contain a high-spin In this way, Eaton and Lovenberg were able to assign the ferrous ion in an approximately tetrahedral site, which is prob- 6250 cm-' band in reduced rubredoxin as arising from com- ably provided by four sulfurs. ponents of the magnetic-dipole allowed 6E -E 'T transition Additional evidence for this conclusion comes from a more of an approximately tetrahedrally-bonded high-spin ferrous detailed consideration of the d-orbital energy levels, based ion (8). Their interpretation was based on the low intensity, on the recent M6ssbauer results (1) and the magnetic dipole low energy, and, most convincingly, on the large anisotropy selection rules for electronic promotions between individual factor (Table 1). The 5E -p- 5T transition arises from the pro- d-orbitals. The M6ssbauer studies have shown that reduced motion of an electron from a filled e orbital (mainly d52, ferredoxin contains a high-spin ferrous ion, in which d32 is d,2-y2) into a half-filled t orbital (mainly d3., dye, day). In per- the lowest lying d-orbital, and the next lowest d-orbital is fect tetrahedral symmetry, and in the absence of spin-orbit about 500 cm-' above de2. The most common geometry for a coupling, the static crystal-field model (27) predicts the ap- high-spin ferrous ion with d.2 lowest is tetrahedral; however, pearance of a single electronic transition at an energy of it is conceivable that these results would be obtained with a -10 Dq, where 10 Dq is the energy difference between the square-planar complex. The only inherently magnetic-dipole doubly-degenerate e orbitals and the triply-degenerate t allowed electronic promotions from d52 are into d.. and d,,. orbitals. Thus, we may conclude that the d-orbital character of the The measurements of Eaton and Lovenberg (8) were con- 4000 cm-', 6000 cm-' transitions is mostly d,2 dr,, dy..tt fined to frequencies greater than 5500 cm-', where D20 ab- These optical and M6ssbauer results yield the following level sorption is not a serious experimental problem. As can be diagram for the four "observed" d-orbitals in ferredoxin and seen in Fig. 2. reduced rubredoxin has at least one additional adrenodoxin. electronic transition below 4000 cm-'. This transition has dYZ 6000 cm-' an anisotropy factor greater than 0.01, indicating that it is mgnetic-dipole allowed. The separation between the ligand- dXZ 4000 cm- field band at 6250 cm-' and the first intense charge-transfer d? 500 cm-' band at 30,000 cm-' is so large that even in the absence of d.2-- 0 supporting circular dichroism data, this transition centered below 4000 cm-' must almost certainly correspond to one of This level diagram is inconsistent with octahedral coordina- the split components of the 5E -* 5T transition. Conse- tion and indicates that tetrahedral coordination (d? = quently, a better estimate for the average ligand-field strength d.2,2) is much more likely than a square-planar arrangement of the mercaptide sulfurs in reduced rubredoxin is Dq of ligands. Again, like rubredoxin, the large separation of -500 cm-'. Moreover, since the observed components of the do and d,3 orbitals indicates a considerable distortion the upper 5T state are split by at least 2500 cm-', which is from perfect tetrahedral symmetry. Part of this distortion about half of the tetrahedral splitting, there must be a con- can be attributed to the presence of two kinds of sulfur-sulfide siderable distortion from perfect tetrahedral symmetry. This and mercaptide-bonding to the iron (Fig. 1). large distortion is seen in the polarized single-crystal absorption spectrum of oxidized rubredoxin (Eaton and Loven- ** We were not able to obtain comparable circular-dichroism data and is also reflected in the rhombic on ferredoxin in the 4000-4800 cm-' region, because the in- berg, unpublished), creased absorbance with time referred to EPR at = 4.3. et al. have in the Experimental spectrum g Recently, Watenpaugh Section resulted in optical densities sufficiently large that the that one of the iron-sulfur bonds in oxidized rubre- reported signal to noise level was unacceptable. doxin is about 0.4-A shorter than the other three (28). tt The lack of circular-dichroism data, together with the differ- The near-infrared spectrum of oxidized rubredoxin shows ence in bandshape for the 5800-cm-l and 3800-cm-1 bands, no magnetic-dipole transitions; the band at 13,400 cm-', makes the 3800-cm-1 assignment in ferredoxin less certain. Downloaded by guest on September 23, 2021 Proc. Nat. Acad. Sci. USA 68 (1971) Tetrahedral Iron in Ferredoxin 3019 TABLE 2. "Effective" spin-orbit coupling constants

9z- gs AE...2 (cm-') rx (cm-') z -gy AEyz-.2 (cm-') By (cm'-) Ferredoxin* 0.159 3800 300 0.088 5800 260 Adrenodoxint 0.091 4500 200 0.086 6000 260

* 9g = 1.886, gy = 1.957, gz = 2.045 (37, 38). t g, = 1.930, gy = 1.935, gz = 2.0214 (15, 39).

The "disappearance" of the low-energy one-center ligand- field transitions. In the case of the oxidized proteins, however, field bands upon one-electron oxidation implies that the the anisotropy factor indicates substantial magnetic-dipole reducible iron atom is high-spin ferric in the oxidized proteins, character, which suggests some other origin for the transition. as was previously noted for rubredoxin. This result is con- Thus, the considerable magnetic-dipole transition moment sistent with the coupled-tetrahedral-iron model outlined (greater than 0.2 Bohr magnetons) is suggestive of a one- earlier, in which both iron atoms are high-spin ferric. center spin-allowed d d transition. If the oxidized proteins Both oxidized and reduced proteins exhibit weak transi- contain only ferric ions, this would require that the spin tions at about 11,000 cm-', the region of the onset of relatively state of at least one of the irons is low, which is not in keeping intense, broad absorption. Interpretation as to the origin of with recent Mossbauer (1) and magnetic-susceptibility results these transitions presents considerable difficulties because (3). In the case of the reduced proteins, a more plausible ex- of the many possible transition types, and, consequently, planation for the transition at 11,000 cm-' is that it cor- the following discussion should be considered largely specula- responds to the lowest-energy intervalence charge-transfer tive. On the other hand, an understanding of this part of the transition, i.e., the transfer of an electron from the lowest d electronic spectrum is potentially very important, because orbital of the ferrous ion to the lowest d orbital of the ferric is may directly reflect the extent of interaction between the ion (34, 35). In contrast to the spectra between 13,000 cn-' metal ions. It is well known in high-spin d5 manganese and 30,000 cm-', where the oxidized proteins absorb more systems, that exchange coupling between the metal ions may strongly, the greater absorption of the reduced proteins at result in considerable enhancement of the absorption intensity about 11,000 cm-' supports this interpretation. If this ex- for the spin-forbidden sextet-quartet transitions of the single planation is correct, then the low intensity of the intervalence ion, with only a small effect on the absorption frequencies band indicates that the optical electron, the "eleventh'' d (29-30). More recently, Schugar et al. have observed marked electron, is highly localized on the ferrous metal site in the intensification of these transitions in oxobridged iron dimers ground electronic state. More convincing evidence for the (31). For a dc5 high-spin ion in a tetrahedral ligand-field, the nonequivalence of the two metal sites is provided by the lowest energy sextet-quartet transition is 6A -> 4T (4G), existence of the single ferrous-ion type ligand-field spectrum, and is estimated to appear at about 11,000 cm'- in ferredoxin and also by the observation of two quadrupole pairs in the and adrenodoxin.-T Furthermore, a tetrahedrally-bonded high-temperature M6ssbauer spectrum of reduced ferredoxin high-spin ferrous ion could also exhibit spin forbidden d d (1). transitions (6E 3T) in the neighborhood of 11,000 cm-', Since EPR spectroscopy has played such a central role in and exchange coupling could again result in marked intensifi- characterizing the iron-sulfur proteins, it would seem im- cation. The transitions near 11,000 cm-' in the oxidized and portant to test the consistency of the interpretation of the reduced proteins are conceivably not too intense to correspond EPR results and the limited amount of optical data. The to components of these multidegenerate one-center ligand- EPR g-values can be calculated from (5): 9. = go + 7/3Ag9 The energy of the 4T (4G) state can be obtained from the tt gy = solution of the weak-field secular determinant (32): g.-2r/AE2 = 4p 4F BIG 9g g,- 2r/AE._., 35,100w - x 0 -4A/5 Dq where Ag, is the difference between the (assumed) isotropic 0 52,100 0 - X - 2/5 Dq g-value of the ferric ion and the free electron value, AEx-z22 -4A/5 Dq -2 ,/5 Dq 32,000w - x and AEyzz2 give the energies of the optical transitions, which where the diagonal energies are the free ferric-ion term energies, are considered to be mainly dci2 -- dcoand dc,2 -- dc, electronic relative to the IS ground term (31), multiplied by a reduction promotions on the ferrous ion, and r is the one-electron spin- factor fl to account for the decrease in the free-ion values on orbit coupling constant for the ferrous ion; it has the free- forming a complex. Because of the large nephelauxetic effect of ion value of 400 cm-' (36). Anticipating the existence of sub- sulfur ligands, the differences between the free-ion term energies stantial covalency in this system, Gibson et al. (5) chose may be considerably reduced (33). Estimating#8 = 0.5 and Dq = to use a value of 300 cm'- for r and, using a set of approximate -600 cm' , we calculate the 4T (4G) state to be 11,100 cm'I above g-values for spinach ferredoxin, they predicted the optical the 6A ground state. However, in addition to ignoring lower- transitions at 3750 cm-' and 6000 cm-'. The agreement with symmetry ligand fields, this result is very sensitive to the choice of parameters p and Dq. For example, variation of the parameters the optical data seems remarkable, if we make the critical , and Dq over the physically reasonable ranges 0.4-0.6 and assumption that the bands at 3800 cm-' and 5800 cm-' -500 cm-' to -700 cm-', respectively, results in a variation of correspond to the transitions dc2I,- d,c, and d42 --o d, respec- the calculated energy of 4T (4G) from 6700 cm-' to 14,400 cm-'. tively, as discussed above. By using more accurate g-values Downloaded by guest on September 23, 2021 3020 Chemistry: Eaton et al. Proc. Nat. Acad. Sci. USA 68 (1971)

(37) and the same value of I, we calculate optical transitions 8. Eaton, W. A., and W. Lovenberg, J. Amer. Chem. Soc., 92, at 3750 cm-' and 6750 cm-'; this is, nonetheless, impressive 7195 (1970). 9. Herriot, J. R., L. C. Sieker, L. H. Jensen, and W. Loven- agreement. On the other hand, we may use the above equa- berg, J. Mol. Biol., 50, 391 (1970). tions, in conjunction with the EPR and optical spectra, to 10. Phillips, W. D., M. Poe, J. F. Wieher, C. C. McDonald, and calculate I, thereby providing a measure of the total effect W. Lovenberg, Nature, 227, 574 (1970). of covalency on the orbital part of the g-value. Table 2 sum- 11. Lovenberg, W., and W. M. Williams, Biochemistry, 8, 141 (1969). marizes the results. The large reduction in r from the free-ion 12. Petering, D., and G. Palmer, Arch. Biochem. Biophys., 141, value confirms the covalent nature of these complexes (1), 456 (1971). and suggests that adrenodoxin is somewhat more covalent 13. Kimura, T., in Structure and Bonding, ed. C. K. Jorgensen than ferredoxin. Furthermore, the apparent differences in ¢, et al. (Springer-Verlag, Heidelberg, 1968), Vol. 5, p. 1. 14. Abu-Shumays, A., and J. J. Duffield, Anal. Chem., 38, 29A calculated from and gy, may reflect different contributions gs (1966). of the ligand orbitals to the metal-ligand molecular orbitals 15. Palmer, G., H. Brintzinger, and R. W. Estabrook, Bio- (40). However, there is an apparent inconsistency in the chemistry, 6, 1658 (1967). above discussion, for adrenodoxin exhibits an axial EPR spec- 16. Kimura, T., and J. J. Huang, Arch. Biochem. Biophys., 137, trum, implying that the excited states responsible for the 357 (1970). 17. Wilson, D. F., Arch. Biochem. Biophys., 122, 254 (1967). inequivalence of gs and gy are essentially degenerate in adreno- 18. Moffitt, W., J. Chem. Phys., 25, 1189 (1956). doxin, whereas according to the optical assignments in Table 19. Mason, S. F., Proc. Chem. Soc. London, 137 (1962). 2, these excited states are split by 1500 cm-'. A possible solu- 20. Mason, S. F., Quart. Rev. Chem. Soc., 17, 20 (1963). tion to this dilemma is that the near degeneracy of g9 and gy 21. Griffith, J. S., The Theory of Transition Metal Ions (Uni- versity Press, Cambridge, 1964). in adrenodoxin is accidental, i.e., the EPR spectrum is not 22. Ballhausen, C. J., Introduction to Ligand Field Theory reflecting the rhombic symmetry of the complex. This acci- (McGraw-Hill Book Co., Inc., New York, 1962). dental degeneracy could come about if the covalent bonding 23. Figgis, B. N., Introduction to Ligand Fields (Interscience were sufficiently asymmetric so that D=/AE.,-,2 is approxi- Publishers, New York, 1966). mately equal to 2. 24. Eaton, W. A., and E. Charney, J. Chem. Phys., 51, 4502 r,/AE,, (1969). Finally, we should point out that we have not concluded 25. Eaton, W. A., and E. Charney, in "Probes and Membrane anything about the coordination geometry at the ferric ion Function," Probes of Structure and Function of Macro- in these proteins, and that the representation of the geometry molecules and Membranes, ed. B. Chance, C. P. Lee, and as tetrahedral at both iron atoms in Fig. 1 is artistic license. J. K. Blasie (Academic Press, New York, 1971), Vol. 1, p. 155. It will require the application of x-ray crystallographic tech- 26. J0rgensen, C. K., Absorption Spectra and Chemical Bonding niques to establish the complete coordination structure of in Complexes (Pergamon Press, Oxford, 1962). these intriguing binuclear centers. 27. Slack, G. A., F. S. Ham, and R. M. Chrenko, Phys. Rev., 152, 376 (1966). We thank Drs. L. L. Lohr, H. Kon, H. B. Gray, and C. 28. Watenpaugh, K., L. C. Sieker, J. R. Herriott, and L. H. Marzzacco for helpful discussions. We are also grateful to Miss T. Jensen, Cold Spring Harbor Symp. Quant. Biol., 36, in press, Yamamoto for the loan of P. elsdenii rubredoxin. This work was (1971). supported by Career Development Award GM K3-21-213 to 29. Lohr, L. L., Jr., and D. S. McClure, J. Chem. Phys., 49, G. P., an NIH postdoctoral traineeship to J. A. F., and NIH 3516 (1968). research grants GM 12176 and 12713. 30. Ferguson, J., in Progress in Inorganic Chemistry, ed. S. J. 1. Dunham, W. R., A. Bearden, I. Salmeen, G. Palmer, R. H. Lippard (Interscience, New York, 1970), Vol. 12, p. 159, Sands, W. H. Orme-Johnson, and H. Beinert, Biochim. and references therein. Biophys. Acta, 253, in press. 31. Schugar, H. J., G. R. Rossman, C. G. Barraclough, and 2. Fritz, J. A., R. Anderson, J. A. Fee, G. Palmer, R. H. H. B. Gray, J. Amer. Chem. Soc., in press. Sands, J. C. M. Tsibris, I. C. Gunsalus, W. H. Orme- 32. Orgel, L. E., J. Chem. Phys., 23, 1004 (1955). Johnson, and H. Beinert, Biochim. Biophys. Acta, 253, in 33. Jorgensen, C. K., Inorg. Chim. Acta, Rev., 2, 65 (1968). press. 34. Hush, N. S., in Progress in Inorganic Chemistry, ed. F. A. 3. Palmer, G., W. R. Dunham, J. A. Fee, R. H. Sands, T. Cotton (Interscience, New York, 1967), Vol. 8, p. 391. lizuka, and T. Yonetani, Biochim. Biophys. Acta, 245, 201 35. Robin, M. B., and P. Day, in Advances in Inorganic Chem- (1971). istry and Radiochemistry, ed. H. J. Emeleus and A. G. Sharpe 4. Brintzinger, H., G. Palmer, and R. H. Sands, Proc. Nat. (Academic Press, New York, 1967), Vol. 10, p. 247. Acad. Sci. USA, 55, 397 (1966). 36. Dunn, T. M., Trans. Faraday Soc., 57, 1441 (1961). 5. Gibson, J. F., D. 0. Hall, J. H. M. Thornley, and F. R. 37. Fee, J. A., and G. Palmer, Biochim. Biophys. Acta, 245, 175 Whatley, Proc. Nat. Acad. Sci. USA, 56, 987 (1966). (1971). 6. Tsibris, J. C. M., and R. W. Woody, Coord. Chem. Rev., 5, 38. Palmer, G., and R. H. Sands, J. Biol. Chem., 241, 253 (1966). 417 (1970). 39. Watari, H., and T. Kimura, Biochem. Biophys. Res. Com- 7. Palmer, G., and H. Brintzinger, in A Treatise on Electron mun., 24, 106 (1966). Transfer, ed. M. Klingenberg and T. E. King (Marcel 40. Misetich, A. A., and T. Buch, J. Chem. Phys., 41, 2524 Dekker, New York, 1971), in press. (1964). Downloaded by guest on September 23, 2021