The low temperature magnetic circular dichroism spectra of iron-sulphur proteins I. Oxidised rubredoxin

Biochimica et Biophysica Acta, 493 (1977) 122-131

© Elsevier/North-Holland Biomedical Press BBA 37704 T H E L O W T E M P E R A T U R E M A G N E T I C C I R C U L A R D I C H R O I S M SPECTRA OF I R O N - S U L P H U R P R O T E I N S I. O X I D I S E D R U B R E D O X I N

J. C. RIVOAL", B. BRIAT", R. CAMMACK b, D. O. HALLb, K. K. RAO b, I. N. DOUGLAS c and A. J. THOMSON ¢ a Labaratoire d'Optique Physique*, EPCI, lO, rue Vauquelin, 75231 Paris Cedex 05 (France), b School of Biological Sciences, University of London King's College, 68 Half Moon Lane, London, SE24 9JF and c School of Chemical Sciences, University of East Anglia, Norwich, NRA 7TJ ( U . K )

(Received October 7th, 1976) (Revised manuscript received March 21st, 1977)

SUMMARY Variable temperature magnetic circular dichroism spectra have been measured on oxidised Clostridium pasteurianum rubredoxin. Evidence has been obtained for the presence of two one-electron charge-transfer transitions, sulphur to ferric ion, in the region 15 000 to 28 000 cm -1. The first moment of the lower energy band is consistent with it being the orbital transition tl non-bonding sulphur orbital, to the 2 e ferric d-orbital. The magnitude of the spin-orbit coupling constant in the lower excited state has been determined and shown to be small compared with the axial distortion. The splitting of the low energy band observed in the absorption spectrum can therefore be equated directly with the axial distortion of the lowest excited chargetransfer state. Finally, the potential utility of making saturation experiments at very low temperatures has been examined.

INTRODUCTION The structure of the three basic functional units of the iron-sulphur proteins are now well characterised and their properties compared with those of synthetic analogues [1]. They are the one-iron, pseudo-tetrahedral [Fe-(S-R)4]"- of rubredoxin, the two-iron [FeES2(S-R)2] n- seen, for example, in spinach ferredoxin, and finally the four-iron [Fe4S4(S-R)4] n- found either singly or in pairs. Here we report the magnetic circular dichroism (MCD) spectra of the one-iron centres in proteins. The spectra are of potential interest for two reasons. First, we seek to show that the M C D spectra are diagnostic of the particular unit. Since there are many complex proteins which contain a number of chromophoric centres, possibly including one or more iron-sulphur units, the ability to recognise the units in situ would be of considerable " ER 5 of the CNRS.

123 value. In principle, MCD spectroscopy is more discriminating than absorption spectroscopy with unpolarised light, since the signals have a sign as well as magnitude. It remains, however, to be demonstrated that this is a secure method of identifying iron-sulphur centres in heterogeneous proteins. Secondly, a sound interpretation of the electronic spectrum can furnish information about the iron valences in the cluster. Two limiting models may be considered. One, the so-called valence trapped model, describes the cluster as a mixture of ferrous and ferric ions whereas the other description, the valence delocalised model, assigns a valence to each iron which is an average of the total oxidation level of the cluster. The two-iron centre is undoubtedly a valence trapped cluster [2]. We first report the MCD spectra as a function of the temperature of oxidised Clostridium pasteurianum rubredoxin (a one-iron protein) and assign the sulphur to ferric charge-transfer spectrum. This assignment is then used in part II as the basis of a possible assignment of the spectrum of both oxidised and reduced Spirulina maxima two iron ferredoxin. The relationship to the spectra of spinach ferredoxin and adrenodoxin, both two-iron ferredoxins, is also examined. EXPERIMENTAL

Our data have been collected on a home-made instrument [3] allowing the measurement of both MCD and absorption on the same sample. Samples were prepared as clear glasses by quickly freezing a sucrose-saturated buffered solution of the material in a gas-flow cryostat (see, for example, ref. 3). In the case of oxidised rubredoxin, the magnetic field (0.83 tesla) was provided by an electromagnet whereas a superconducting magnet was used in the case of ferredoxin and adrenodoxin, since the MCD signals are considerably smaller. Our data are plotted in terms of AA or A A M = AA/t, BB versus the wavenumber tr. Both #BB and a are expressed in cm -~. A stands for the decadic absorbance and AA for the differential decadic absorbance. The proteins were prepared by well characterised methods fully described elsewhere [4-7]. Unfortunately, no universal terminology has yet been put forward to describe MCD spectra. It is therefore appropriate to describe that used in the present papers as well as in a previous contribution on the applications of MCD [8]. We define nth order moments of A/a and AAM/a relative to a wavenumber ~ as:

(A).

= ~ (A/a) Or -- ~)" da

(AAM). = J" (AAMI
AAM = al(--6A/6a) + (bo + co/kT)A One then has A Mo ~ bo + co/kT and A M1 ~ al.

124 Co occurs for paramagnetic species only and arises due to the difference of population among the Zeeman sublevels in the ground state; a~ is present for both diamagnetic and paramagnetic compounds and is due to the Zeeman splitting of the lines; b0 occurs when either the ground or the excited state mixes with another state under the action of B. A positive Co or bo value implies a positive bell-shaped MCD whereas a positive a, value corresponds to a derivative-like shape with a positive maximum at higher energy. When several bands overlap dMo and AM1 are still convenient descriptions, although some care should now be taken regarding their precise significance in terms of molecular parameters. Also, when Zeeman splittings are comparable to kT, then normalised amounts should be determined as (AA),,/(A~o as a function of i~aB/kT. THEORETICAL MODEL FOR OXIDISED RUBREDOXIN The X-ray structural determination shows that this protein contains an iron atom surrounded by the -CH2-S- groups of four cysteinyl residues in an approximately tetrahedral array [9]. The X-ray data first suggested a distorted tetrahedral unit with one short Fe-S bond. However, this result has been challenged with the aid of new structural data supplied by the XAFS* technique [10]. The exact nature of the distortion of the ground state is not yet clear. However, low temperature electron paramagnetic resonance (EPR) studies show that the ground state is 6A 1 derived from the high-spin d 5 configuration, with an almost completely rhombic environment, D = 1.76 cm -~ and E = 0.485 cm -~ [11]. As a consequence the ground state consists of three almost equally spaced pairs of Kramer's doublets, spanning an energy range of approx. 12 cm -1, with the ± 5/2 doublet lying lower. The analysis of the optical spectrum carried out by Eaton and Lowenberg with the aid of polarised single crystal spectra and room-temperature MCD spectra led to the following conclusions [12]. Two charge-transfer transitions, from sulphur to ferric ion, dominate the spectrum from 15 000 to 30 000 cm-~. The transitions are both allowed, being 6A 1 ~ 6T z of tetrahedral parentage. Both excited states undergo an axial distortion, the peaks at 26 200 and 20 200 cm -~ being assigned a s 6A1 --> 6E and those at 28 800 and 17 700 cm -1 arising from 6A 1 ~ 6A 2. The rhombic splitting of the excited state was estimated to be less than half the axial distortion. By the use of the polarised single crystal data it was possible to relate the axis of axial distortion to the ground state geometry. The unique distortion axis runs almost along the $4 axis of the Fe(S-R)4 tetrahedron suggesting an effective excited state geometry of D2a symmetry. An excited state, 6T2, of tetrahedral parentage can be split by static distortions of the environment and by spin-orbit coupling. To illustrate the effect of both perturbations acting together we have calculated the matrix of a static uniaxial distortion along the $4 axis, generating a D2d field, and of spin-orbit coupling. The resulting energies are plotted in Fig. 1 as a function of the ratio of two parameters, the axial distortion parameter, 11, and a spin-orbit coupling parameter, A, both defined in Fig. 1. In the limit of zero axial distortion the three spin-orbit states arising from 6T 2 a r e separated in energy by 5 All2 (J ~ 7/2), --2 All2 (J--~ 5/2) and --7 All2 XAFS; X-ray absorption fine structure.

125

o6

%E:O;~-o.6

05

- 0.5

02 sE ~

- 02

v~

o

-o ~

-0"1

-0"1

-0-2

.02

-Oz

- -o.4

-0-~

-05

-0.6

-0.6

--0 7

0

I

I

I

0.5

IO

05 v A

A v

"1-07 0

Fig. l. Energy level diagram showing the effect of applying both a spin-orbit perturbation and axial (D2d) distortion to a 6T2 state. In the case of oxidised rubredoxin the allowed sulphur to ferric chargetransfer excited state are 6T2 under the point group Ta. The energy scales are normalised and the parameters are defined as V, axial distortion,

2 and A, spin-orbit coupling parameter

-

-

-2 V <6T211ZIs,u[16T2> 35

(J = 3/2). On the other hand, at the limit of zero spin-orbit coupling two states 6E and 6B 2 arise. A similar diagram has been constructed for a static distortion opposite in sign to that shown. By studying the temperature variation of the MCD spectrum, the contribution of spin-orbit coupling to the distortion of the 6T 2 state can be measured. In the limit of zero spin-orbit coupling the MCD spectrum corresponding to the transitions 6A x --~ 6E and 6B 2 will be temperature-independent and consist of an al term and a b0 term respectively. At the other extreme of zero axial distortion and finite first order spin-orbit coupling, the spectrum will consist essentially of three temperature dependent Co terms, two positive and one negative, corresponding to transition from the ground state spin-orbit components of 6At, t o the three components of 6T2, namely, U~/2, E;/2 + U~/2 and ET/2 + E~i2 + U7)2. Moreover, a moment analysis of the complete band allows an estimate to be made of the magnitude and sign of the excited state spin-orbit coupling constant [13]. The sign of this parameter may in turn provide an indication of the orbital assignment of the charge-transfer transition.

126 Therefore a study of the temperature dependence of the MCD spectrum of oxidised rubredoxin enables two extra pieces of information to be obtained. First, the sign and magnitude of the excited state spin-orbit coupling constant can be determined provided that the transitions are sufficiently well resolved to enable a moments analysis to be carried out. Secondly, an assignment of the configurational origin of the transition is possible in principle. RESULTS AND DISCUSSION

The MCD spectra of oxidised C. pasteurianum rubredoxin at room temperature and at 17.5, 10.2 and 6.1 K, are shown in Fig. 2a. The room temperature absorption spectrum is also given at 300 K (7 K), no significant change occurring upon cooling. Our MCD spectrum at room temperature is in good agreement with that previously reported [12, 14]. These workers also measured the spectrum of reduced rubredoxin, demonstrating that no intense absorption bands appear below approx. 27 000 cm-1. The region of the spectrum of the oxidised species between 15 000 and 23 000 cm -~ contains Co terms and is remarkably similar in shape and sign with the results

a

.,.'"~,= O . 8 3 T ""

-a~ -10

",.

\

3 0 0 K ---.=-

/ ,"'~, 2. 7 K " ' " < 8

OXIDISED RUBREDOXIN

30 ..j/~2,5

! / ~ a C M'IN},,

•f \

..

..

..= ~-

,:..--.)~

,;..:-..~

-- 6.1K

< '~ -'40

\ '"" ~ / ~ /

\ "I ',,,. /

.... 1 0 . 5 K - ' - 17.2 K

o204:1~

• -40

(1/K T ) / C M 0.1 !

0.2 I

Fig. 2. (a) The absorption ( × - × .) and M C D spectra o f oxidised C. pasteurianum rubredoxin in Tris buffer. Concentration is 10 -5 M, path length 1 ram, B = 0.83 testa. Temperature at which M C D spectra were measured, - - , 300 K , . . . . , 17.2 K ; . . . . . , 10.5 K ; and . . . . ,6.1 K. (b) Plot o f the first m o m e n t , AMt, measured from 15 000 to 23 5 0 0 c m -1, against 1/kT.

127 obtained by two of us for the first charge transfer band of tetrachloro- and tetrabromocomplexes of iron (III)[13]. These observations confirm Eaton and Lovenberg's suggestion [12] that this region corresponds to a single transition, 6 A 1 ~ 6T2, of tetrahedral parentage. Four further Co terms are resolved at low temperature between 23 500 and 19 000 c m - L It seems quite unlikely from our predictions that they arise from a single 6A~ ~ 6T 2 transition. Our interpretation of the spectrum thus differs from that proposed by Eaton and Lovenberg who concluded, from the less well-resolved room temperature spectrum, that the above spectral region contains only one transition of tetrahedral parentage. The lowest energy band system is sufficiently well separated from its neighbours to allow a moment analysis to be carried out from 15 000 to 23 000 c m - L A plot of A Mt against 1/kT is shown in Fig. 2b. The plot is linear within experimental error. This result seems surprising at first sight since the total splitting between the ground state doublets is about 17 K and it would seem reasonable to observe nonlinearity. The expressions for the zeroth and first moments for the ligand to metal charge-transfer bands of a tetrahedral, high spin, ferric complex have been given by Rivoal and Briat [13]. In the limit of first-order spin-orbit coupling, they found: (1)

AMo = 0

A M I = 2l

35 d 18 k T

(2)

where l is the orbital moment in the excited state, given by (6T 2 I]LzI6T2 1) and d is the total spread of the spin-orbit components; A is taken as positive when the level with maximum multiplicity lies highest. Note that AM1 is independent of static distortion of the excited state provided that second order mixing of the state with neighbouring states can be ignored. The above model is certainly not perfectly appropriate for rubredoxin where relatively large zero-field splitting is known to exist in the ground state. We have therefore considered a more realistic situation with an isolated 6 T 2 excited state (as in the previous model) but with a zero-field splitting 2D cm -1 between 4-3/2 and 4-1/2 and similarly 4D c m - ' between 4- 5/2 and 4- 3/2 in the 6 A 1 state (axial situation). Assuming now that the mixing between 6 A 1 and 2S+~T~ via spin-orbit coupling is negligible, the coupling coefficients listed in ref. 13 can be used to obtain: A

( A A ) I / ( A ) o = -- I ~ Z (B, T, D)

(3)

for the temperature dependent part of the first MCD moment. The Z function can be written: Z = {sh x + 3e -y sh 3x 4- 5e -3y sh 5x} {ch x 4- e -r ch 3x ÷ e -3y ch 5x} -1

(4)

with x = glzsB/kT and y = 2D/kT; 2D is taken as positive when the ± 5/2 doublets lies highest.

128

Z

1

3

0

2

21D/CM'~ 2

1

Fig. 3. A plot of function Z (see text) against x = gl~aB/kTforseveral values of 2D; g was taken as 2 and B = 0.83 tesla; 2D is chosen as positive when the 5/2 doublet lies highest. Z is plotted in Fig. 3 for 0 < x < 1 and for values o f 2D ranging from --10 to + 10 cm-1. Fig. 4 shows an enlargement o f the expectations in the x region covered in our experiments. These data clearly demonstrate the interest o f saturation experiments, i.e., measurement o f (AA)I/(A)o for increasing values o f x. Our experimental data in Fig. 2b cannot by any means be understood i f 2 D is positive since, for example, Z varies little or not at all when T is changed from 10.5 K to 6.1 K. On the other hand, negative values o f 2D (12Dr ~< 4) lead to an almost perfectly linear plot o f Z versus x for the three temperatures considered in agreement with experiment. The choice o f 2 D = - - 4 cm -1 leads to an estimate o f A ~ 85 cm -1. The orbital m o m e n t l is quite certainly negative but its precise value is uncertain. We note that inclusion of some orbital character into the 6A 1 ground state is expected to change only slightly the A parameter [13]. The above result is independent o f the axial distortion, whether tetragonal or trigonal. One may now wonder what happens in the case o f rhombic symmetry. We indeed performed the calculation o f Z (B, T. E, D) in the case of 2 = E/D = 1/3 at the

Z 17'.2

16.5 T / K ~

-2

1

I

i

I

Fig. 4. An enlargement of the Z = f(x, D) plot in the x region covered in our experiment.

129 three experimental temperatures. D and E have their usual meaning and the ground state wave functions were taken from ref. 15. We chose 12 cm -1 for the total spread of the ground state Kramer's doublets, i.e., the value estimated from EPR. Neglecting again the mixing of 6A1 with orbital triplets, we also found a linear dependence of Z vs. 1/kT for 6.1 K < T < 17.2 K, the slope (and thus 4) being slightly altered from that found in the axial case. We did not pursue our investigation further in view of our experimental data being limited to only three temperatures. It appears however that new experiments over a wider range of temperatures are likely to help in deciding upon whether the symmetry is axial or rhombic. The configurational origin of the first transition in oxidised rubredoxin can only be established by comparison of our estimates for d and l with the results of a calculation. The order of the highest occupied orbitals mainly of ligand character has been determined in a number of analogous metal complexes such as tetrahedral metal halides [13], oxyanions and thioanions [16]. In all of these examples the lowest energy ligand to metal charge transfer transition involves the non-bonding tl ligand level and the 2e d-level. Disregarding the fl-carbon atoms of the four cysteinyl residues surrounding the ferric ion the centre can be treated as the pseudotetrahedral [Fe(SH)4]- unit. Strictly the tl, sulphur Pn, ligand orbitals, although non-bonding with respect to the metal ion, will involve same fl-carbon character. Indeed, the effective D20 symmetry determined for the first excited state from polarised single crystal spectroscopy may arise from the geometry of the cysteinyl residues. However, inclusion of the fl-carbon character in the sulphur non-bonding t~ orbitals would be cumbersome and involve the assumption of a number of parameters. Therefore we choose to analyse the spectrum as a pseudo-tetrahedral unit allowing us to take over the analysis carried out for the halides and thioanions. The expressions for A deduced by Rivoal and Briat are general. For the transition h -+ 2e, l is given by 3/10 ~PL, where (PL is the spin-orbit coupling constant of the ligand. Thus, in this instance, no molecular orbital coefficients of the metal-ligand combinations are involved. Similarly, the value of l for this orbital transition is shown to be --0.25 B.M. and independent of any molecular orbital coefficient involving the mixing of metal and ligand orbitals. This occurs because the ligand orbitals are nonbonding with respect to the metal d-orbitals. Presumably some admixture of fl-carbon character will enter and may be significant in determining the precise numerical value. The value of (pL is 380 cm -~ for the 3s2 3p 4 configuration of the free atom of sulphur. The value deduced from the experimentally determined value of d is (p~ 300 cm -~. Thus the sign is consistent with the assignment of the lowest energy transition to h--~ 2e and the magnitude seems quite reasonable. The difference could perhaps be accounted for by the low symmetry of the sulphur atom in the cysteinyl residue. The sign found experimentally for l again agrees for this assignment. Evaluation of A for other possible transition could in principle be carried out with the expressions of Rivoal and Briat, although in these cases knowledge is required of the molecular orbital coefficients and the metal ion spin-orbit coupling constants. The former data are not available. However, we can point out that the second transition, between 23 500 and 28 000 cm -~, has a positive value of AM~ which is inconsistent with its assignment to the t~ -~ 2e transition.

130 CONCLUSION Low temperature experiments with oxidised rubredoxin have provided useful evidence showing the presence of two one-electron charge-transfer transitions between 15 000 and 28 000 cm -1. The first moment of the lower energy band is consistent with it being the orbital transition tt --~ 2e. The value of A ~ 85 cm -1 determined experimentally can be used to assess the relative importance of the axial distortion parameter V, required to fit the energy separation between the peak at 17 800 cm -~ and that at 20 500 cm-1. Examination of Fig. 1 shows that the spectrum can only be fitted with a value of A/V much less than 0.1. Since d is so small, V must be responsible for producing the large separation of the two peaks in the absorption spectrum. Thus V can be equated directly with the splitting of 2500 cm -~. Therefore A/V is <~0.01. This confirms Eaton and Lovenberg's view that the axial distortion is responsible for the band splitting. Spin-orbit coupling is significant only in producing the temperature dependence of the M C D spectrum. It is important to note that this analysis provides only a conclusion about the geometry of the lowest excited charge-transfer state. This is not directly related to the ground state geometry as probed by EPR, X-ray, and XAFS [10] techniques. Further since our analysis has been in terms of the moment of the M C D spectrum which is, to a first approximation, independent of static distortions, whether axial or rhombic, of the excited state we can conclude nothing about the direction of the axis of distortion relative to the molecular framework. However, the form of the spectrum does indicate that the axial distortion far outweights any rhombic component. Another interesting conclusion of this study is our demonstration of the potential utility of saturation experiments down to pumped helium temperature in the case of orbitally non degenerate ground states suffering from low symmetry distortions of either uniaxial or rhombic symmetry. In this respect, M C D experiments provide an important link between E P R measurements and the optical spectrum. ACKNOWLEDGEMENTS Thanks are due to the Science Research Council for support for preparative work (King's College) for travel (Norwich) and for a fellowship (to I.N.D.). One of us (A.J.T.) was the recipient of a Ciba-Geigy Senior Fellowship. REFERENCES 1 Hall, D. O., Rao, K. K. and Cammack, R. (1975) Sci. Prog. Oxf. 62, 285 2 Herskovitz, T., Averill, B. A., Holm, R. H., Ikers, J. A., Phillips, W. D. and Weiher, J. F. (1972) Proc. Natl. Acad. Sci. U.S. 69, 2437 3 Badoz, J., Billardon, M., Boccara, A. C. and Briat, B. (1969) Syrup. Farrday, Soc. 3, 27 4 Rao, K. K., Evans, M. C. W., Cammack, R., Hall, D. O., Thompson, C. L., Jackson, P. J. and Johnson, C. E. (1972) Biochem. J. 129, 1063 5 Rao, K. K., Cammack, R., Hall, D. O. and Johnson, C. E. (1971) Biochem. J. 122, 257 6 Cammaek, R., Rao, K. K., Hall, D. O. and Johnson, C. E. (1971) Biochem. J. 125, 849 7 Hall, D. O., Cammack, R. and Rao, K. K. (1972) Biochem. Biophys. Res. Commun. 47, 798 8 Briar, B. (1975) in Electronic States of Inorganic Compounds: New Experimental Techniques, (Day, P., Ed.), D. Reidhel, Dordrecht, Boston

131 9 Watenpough, K. D., Sieker, L. C., Herriott, J. R. and Jensen, L. H. (1971) Cold Spring Harbor Symp. Quant. Biol. 36, 359 10 Sayers, D. E., Stern, E. A. and Herriott, J. R. (1976) J. Chem. Phys. 64, 427 11 Peisach, J., Blumberg, W. E., Lode, E. T. and Coon, M. J. (1971) J. Biol. Chem. 116, 5587 12 Eaton, W. A. and Lovenberg, W. (1973) in Iron-Sulphur proteins (Lovenberg, W., ed.), Vol. II. Academic Press 13 Rivoal, J. C. and Briat, B. (1974) Mol. Phys. 27, 1081 14 Ulmer, D. D., Holmquist, B. and Vallee, B. L. (1973) Biochem. Biophys. Res. Commun. 51, 1054 15 Hollis Wickman, H., Klein, M. P. and Shirley, D. A. (1965) J. Chem. Phys. 42, 2113 16 Petit, R. H., Briat, B., Mflller, A. and Diemann, E. (1974) Mol. Phys. 27, 1373