NMR studies of electron transfer mechanisms in a protein with interacting redox centres: Desulfovibrio gigas cytochrome c3

Em. J. Biochem. 141, 283 - 296 (1 984) 9 FEBS 1984

NMR studies of electron transfer mechanisms in a protein with interacting redox centres : Desulfovibrio gigas cytochrome c3

Helena SANTOS, Jose J. G. MOURA, Isabel MOURA, Jean LeGALL, and Antonio V. XAVIER Centro de Quimica Bstrutural, Universidade Nova de Lisboa; Gray Freshwater Biological Institute, University of Minnesota; and Department of Biochemistry, University of Georgia

(Received December 5 , 1983) - EJB 83 1307

The proton NMR spectra of the tetrahaem cytochrome c3 from Desulfovibrio gigas were examined while varying the pH and the redox potential. The analysis of the NMR reoxidation pattern was based on a model for the electron distribution between the four haems that takes into account haem-haem redox interactions. The intramolecular electron exchange is fast on the N M R time scale (larger than lo5 s- I ) . The NMR data concerning the pH dependence of the chemical shift of haem methyl resonances in different oxidation steps and resonance intensities are not compatible with a non-interacting model and can be explained assuming a redox interaction between the haems. A complete analysis at pH* = 7.2 and 9.6, shows that the haem-haem interacting potentials cover a range from - 50 mV to +60 mV. The midpoint redox potentials of some of the haems, as well as some of their interacting potentials, are pH-dependent. The physiological relevance of the modulation of the haem midpoint redox potentials by both the pH and the redox potential of the solution is discussed.

Cy tochromc c3 ( M , 13 000) is a multihaem protein found in anaerobic sulfate-reducing bacteria belonging to the genus Desulfuvibrin. Each molecule contains four haems with an unusually low redox potential. They are covalently attached to the polypeptide chain by thioether linkages provided by cysteinyl residues and two histidines are used as axial ligands.

Cytochrome cj was first isolated from Desulfovibrio vulgaris in 1954 by Postgate [I] and Ishimoto et al. [2]. Homologous proteins were isolated from Desdfovibrin de- sulfuricans [3, 41, Desulfovibriu gigas [5], Desulfovibrio salexi- gens [6] and Desuljbvibrio africanus [7]. The amino acid sequences of cytochrome c3 isolated from six different species are known [X - 101. An alignment of the amino acid sequences for several cytochromes c3 from different Desulfovihrio species was proposed by Ambler et al. [ X I based on amino acid analogies. More recently, X-ray diffraction structures of cytochrome c3 from D. vulgaris (Miyazaki) [I I] and D. desuljiuricans (Norway 4) [12, 131 have been reported at 0.25-nm resolution. A careful comparison of those structures suggested that the relative orientation and arrangement of the four haems is conservcd in spite of the low degree of sequence homology. With this information, a new proposal for the sequence alignment between five cytochromes c3 was presented, together with a general three-dimensional model for cytochrome c 3 [14].

Although cytochrome c3 was the first electron transfer carrier isolated from Desulfovibrio, its physiological role is not yet fully understood. In Desulfbvihrio spp. which grow on colloidal sulfur plus organic sulfate, it was shown that cytochrome c3 is the sulfur reductase [15]. More generally, it is believed that cytochrome c3 is a cofactor for hydrogenase [16]. It was shown to be required for the reduction of ferredoxin,

Part of this work was presented at the First h ternuthnal Conference on Bioinorganic Chemistry, Florence, 1983 [H. Santos, J . 1. G. Moura, I. Moura, J. LeGall, and A. V. Xavier (1983) Inorg. Chim. dctu 79, 167- 1681.

Abbreviation. EPR, electron paramagnctic resonance.

flavodoxin and rubredoxin by hydrogenase plus H, [17]. NMR studies also demonstrated that there is an interaction between cytochrome c j and ferrodoxin I1 from D. gigas [18], as well as with rubredoxin and flavodoxin, from the same species [19]. It has also been reported that cytochrome c3 stimulates the proton-deuterium exchange reaction catalysed by hydrogenase [20]; however, this is not a universal observation [21].

Several physico-chemical techniques, mainly Mossbauer spectroscopy [22], circular dichroism [23], electron paramagnetic resonance (EPR) [24- 271, NMR [28- 321, cyclic voltam- metry, differential pulse polarography [33 - 381 and pulse radiolysis [39] have been applied to elucidate the mechanism of electron transfer in cytochrome c3.

The midpoint redox potentials of the four haems are different, in general. EPR measurements coupled with poten- tiometric titrations were performed to determine the midpoint redox potentials of the individual haems of D. gigas cytochrome c3 [27] (-235mV, -235mV, -306mV, -315mV), D. vulgaris cytochrome c j [26] (- 284mV, - 310 mV, -319mV, -324mV) and D. cfesulfuricans (Norway 4) [40] (- 125 mV, - 125 mV, - 305 mV, - 325 mV). Extensive electrochemical studies have also been carried out: there is a rapid electron transfer between cytochrome c3 at both the mercury [34] and the 4,4’-bipyridyl-modified electrodes [38]. For D. vulgaris (Miyazaki) cytochrome c3 the electrochemical behavior has been evaluated in terms of a modcl with four reversible redox centres [35]; the experiments were fitted by digital simulations and the best fit was obtained with -220mV, -272mV, -292mV, and --310mV, for the four macroscopic midpoint redox potentials. Bianco et al. [36] reported values of -170mV, -310mV, -360mV and -400mV for the half-wave potentials of cytochrome c3 from D. desidfuricans (Norway 4). However. it should be emphasized that the individual potentials obtained from electrochemical studies are macroscopic rather than microscopic parameters and can not be compared directly with the values obtained from EPR measurements. Furthermore, thc EPR measurements

284

werc carried out a t 8 I( and. a s pointed out by Palmer and Olson, for ;I multiredox centre protein this cannot be compared easily with the room tcinperature values [41].

Preliminary NMR studies were reported for D. vulgaris and D. gigas cy(ochromes c ' , ~ [28, 291 but are rather limited. Recently, a more detailed study was undertaken [32] on the electron transfer mechanism of cytochronic c3 from D. vulgaris.

A four-redox-centre molecule originates 16 different redox states in the multiredox electron distribution equilibria [41]. In the general casc of intcracting centres, the electron affinity of one centre differs from that of the others and may also be affected by the ovidation state ofthe adjacent haems. Thus, the number of u n k n o w n parameters is very large and a full characterization of. each individual centre i n such a complex systcni is rather difficult or even impossible by solution equilibria methods. Spectroscopic methods (namely EPK and NMR) probing each site individually: can provide valuable contributions to the elucidation of electron transfer mechanisms i n iiiultiple-redox-centre enzymes. This paper presents an cxtcnsivc study o f D. gigtrs cytochroinc c3 by NMR. Redox titrations at different values of temperature and pH were followed by N M R . The rcsults are interpreted in terms of a complete four-rcdox-centre model for the elcctron distribution and further iindci-standing about the intramolecular and intermolecular electron exchange mechanisms was achieved.

MATERIALS A N D METHODS

lh.sul/i,vihrio , g i ~ q m cytochrorne i '3 was isolated as previously described 1421, and the purity index was (A;',d,- A ?;(,)/A";,,, == 3.0. The protein was dialysed against distilled water at 4 C7 and IyophiliA twice from 2H,0.

For the N M R measurements thc protein was dissolved in 'H,O to a concentration of about 2mM and the pII was adjusted with NaO1.I and/or DCI. Quoted pH* values are meter readings uncorrected for the isotope effect.

Solutions of the fully reduced proteins were obtained by addition of ;I small amount of solid disodium dithionite and were kept under argon, to delay autoxidation. The samples were allowed to reoxidize very slowly by introducing small amounts of air into the N M R tube with a Hamiltoii syringe through serum caps. As the decomposition products of dithio- nitc may alter the pH ol'thc solution, some redox titrations were performed by varying the hydrogen pressure in the presence of hydrogenasc puriticd I'roni /I. gigas [43]. The reoxidation patterns were found to be identical in both methods.

The proton N M K spectra were obtained using a Bruker C'XP-300 spectrometer equipped with an Aspect 2000 computer in which mathematical inanipulations were carried oul. Some of the spectra were run in a Bruker AM-500 at 297K, equipped with an Aspect 3000 computer. Saturation transfer experiments were performed as previously dcscribed [32] in intermediate oxidaiion stages. Typically four free induction decays were acquircd aftcr an ahout 0.25 s irradiation at the frequency of ;I hacni methyl resonance; the next four free induction decays, with the sanie period of irradiation i n an empty region of the spectrum, were then subtracted from the previous ones. The sequence was repeated up t o 1000 times in order to obtain ;I good signal-to-noise ratio in the difference spectrum. The temperature a t which the spectra were obtaincd was 273

Thc chemical shifts are quoted downficld from the methyl resonance of sodium 2,2-diinetIiyl-l-silapentane 5-sulplionate but dioxanc w a s used a a an internal standard.

0.5 K, unless otherwise stated.

Model for the electron distrihutio/i in n letrahfieriz qtochronze

The general scheme for the equilibrium distribution of electrons in a four-centre molecule is depicted in Fig. 1, with all the pathways by which the 16 different states can interconvert during the oxidation-reduction process [41].

By successive losses of one electron, according to the following scheme :

~ l e ~ l c ~ Ic ~ I r

Step 0 + Step I g Step I1 2 Step 111 Step I V

live different oxidation steps can be obtained, starting from the fully reduced statc (step 0). Step I dcvelops by loss of one electron and includes oxidation states 1 1 , 12, I3 and 14, all of them with only one haern oxidized. Loss of one inore electron leads to step TI and six possible oxidation states can be generated, from 5 to 10. Step 111 includes states 1 - 4 in which three haems are oxidized and one is reduced. Finally, the fully oxidiLed state, 0, is the only state of step IV. A macroscopic standard potential can be defined for each of the four equilibria involving two adjacent steps [35]. These are the redox potentials obtained by electrochemical methods.

Microscopic standard potentials (ei, e, $, c q ' , where i, j , k , 1 = 1 - 4) can be defincd for the 32 equilibria between each pair of states shown in Fig. 1, where ci is the microscopic midpoint redox potential of haem i , when all thc other haems in the same molecule rcmain reduced : ci is the microscopic midpoint redox potcntial of haern iwhen haemjis oxidized and the other two haeins (not spccified) are reduced, and so on for

I t is worth pointing out here that a unique microscopic midpoint redox potential can be defined for each haem in the equilibria step 0 e step 1, and step 111 e step 1V. For instance, e, is the midpoint redox potential for haern i i n the equilibrium step 0 $ step I, and <kl is the midpoint redox potential of the same haem for the equilibria in step I11 + step IV: however, for the other cquilibria involving step 11 no unique microscopic midpoint rcdox potential can be defined for each haeni (see Fig. 1).

The four haems can be numbered from 1 to 4, according to their redox potentials ei ( 6 7 , 5 e, I c j I el) and P, designates the molar fraction of state / I (n = 0, . . . , 15) for a given solution redox potential, E. For example, Po is the molar fraction of the population ofthe fully oxidized state and P , is the population of the fully reduced statc. Now, 32 Nernst equations can be written for the 32 rcdox couples involved; all the populations, P,t, can be expressed inathematically as a function of two of them, e.g. P , and P I , , and of the differences between the haem microscopic midpoint redox potentials. The expressions for the 16 populations are dcscribed in Appendix A.

The interacting potential between two haems, e.g. haem i and haem j , is expressed by the change in the midpoint redox potential ofhaem icaused by the oxidation ofhaem j . Thus, the interacting rcdox potential between haems i and j , Z,;, is defined a s follows:

$ and p!k ' ,

The interacting redox potentials are assutned to be additive, i.e:

Since the Gibbs energy is a state function, i t is easily demonstrated that the effect on the midpoint redox potential of haem i caused by oxidation of haem j is similar to the effect of

28 5

Fig. 3 .E/ectron dixiistrihurioti scheine for o terrahaem cyfochrome. Oxidation states 0 and 15 refer respectively to the fully oxidized and fully reduced protein. ep' is the microscopic midpoint redox potential of haem i when haems j , k and I remain oxidized (0 = haem reduced: o = haem oxidized)

the oxidation of haein i on the midpoint redox potential of haem . j :

e.-e? = p . - $ 1 1 J

E;'k -ef = e'k-pk J I '

15

With the additional condition that C P, = 1, it is possible

to calculate all the populations P,, for each value ofthe solution redox potential, E, if the absolute values for the four microscopic midpoint redox potentials ei and the interacting potentials are provided. A full description of the general case requires ten parameters, i s . el, e,, e3, e,, and the six interacting potentials, fij, between the different haems.

If (here is no interaction between the different haems in the sense that the redox potential of one haem is not affected by the oxidation state of the other haems, then:

n = O

p . = (J = pP = $ 1 i, j , k , 1 = 1 - 4 1 1

and only four parameters (el, e2, e3 and e,) are needed to characterise the system, the expressions for P,, (see Appendix A) become much simpler.

Plots of the 16 populations P,, as a function of E are shown in Fig. 2: a non-interacting system is considered and the effect of the diffcrciices between the midpoint redox potentials,

= e,-e,, , d 3 , = e3-e4, on thepopulation distribution of the different states is illustrated.

= e2-el ,

RESULTS

N M R redox titrution qf' Desulfovibrio gigas cytochrome c j

The general features of the NMR spectra of D. gigas cytochrome c3 in the fully oxidized and fully reduced forms were discussed previously [IS, 281. Fig. 3 and 4 show the low-field region of the N M K spectra of the reoxidation pattern

1.0

0.8

c 0 .- c

e L

4 0.6

0.4

0.2

0.0 - 0.4 - 0.2

Redox potential [ V I

Fig, 2. Population distribution CUYIW , jbr ihi. 16 o.xitbtion states reprpruscwled in Fig. 1, assuiniizg u !ion-interacting nrodcl und jbr two hy/iorherical cases, A tirzd R. (A) Pour equivalent haems, el = e, = e3 = e4; (B) four different haein midpoint redox potentials: ez-ei = 30mV; e2-e3 = -50mV; e3-e4 = - - 50mV

of D. gigas cytochrome c j at 273 K and at pH* 6.7 and 9.6. In the fully oxidized protein (upper spectra) the low-field region (10- 35 ppm) shows several resonances shifted out from their position in the spectrum of the fully reduced protein by interaction with the paramagnetic ions. The resonances M:' -

Miv0 have integrals corresponding to three protons and have been assigned to the methyl groups of the haems [28, 291.

The total paramagnetic shift felt by each haem methyl group in the fully oxidized state is the combined result of the paramagnetic interactions with the ferric ion of its own haem (intrinsic shift, A,,,,) and the ferric ions of the neighbour haems (extrinsic shifts, 4,J. For instance, for a haem methyl group I ,

belonging to haem 1, the total paramagnetic shift (relative to the diamagnetic position) is:

The intrinsic shift, lint is dominated by the contact contribution, and is in genernl much larger than ,A,,,, which is only due to the pseudocontact contribution thus being dependent on the relative positions of the methyl group considered and the paramagnetic centre [44].

The spectrum of the fully reduced protein is characteristic of a diamagnetic cytochrome. By a careful addition of small amounts of air, intermediate oxidation stages could be obtained and the NMK spectra were acquired after the equilibrium was reached. In this way, the reoxidation of D. gigas cytochrome c3 can be followed by NMR. As the reoxidation proceeds, some resonances of the haern methyl groups start appearing in the low-field region. The most striking feature of the reoxidation patterns is the presence of resonances in intermediate stages of reoxidation which are not present in the

286

n

L I I 1 I I

35 2 5 15 ppm

Fig. 3. 300-MHz sprc~ i~o of D. gigas clytoclirome c3 it? .wveral oxidation- reducliioi7 srqg~.s, ihc iippcr .spec,tritnz being flint ofthe fuli,v oxidizedstate. The redox titration was carried out at 273 K and pH* = 6.7. Some relevant resonances are identified (see text). Haem methyl resonances arc numbered from 1 to I6 according to their chemical shifts in the spectrum o f thc fully oxictiked protein

I I I I I I 3 5 25 15 PPM

Fig. 4. 300-MHz spectra o f D . gigas cytoc,hrornc c j in several oridurion- redurlion stages, .starring ,from the ,firlly reduccti (lower spectr im/ aiitl going t o the fully oxidizedprotein (upper . q i r ~ m m ) . ‘The titration wiis carried out at 273K and pH* = 9.6. Some relevant resonances ;ire identified (see text)

spectrum of the fully oxidized protein. This can be more easily observed in Fig. 5, whcre a detail of the NMR reoxidation pattern of cytochromc c3 from D. gigas, at pH* 7.2 and 273 K, is shown. The internicdiatc resonances appear, grow and later on disappear, bur their chemical shift and linewidths do not change while the reoxidation proceeds. This means that the intermolecular electron exchange between oxidation states belonging to different oxidation steps (see Fig. 1) is slow on the NMR time scale. In fact. i C the intermolecular electron exchange was fast, the averaged peaks would shift throughout the reoxidation, I‘rom their positions in the fully reduced protein to their final positions in the fully oxidized slate. The last siluation is observed i n cytochrorne c j from Desdfovibrio iksidfiuic.cin,r.i,s (strain 0974) [27] and De.su(/ovihvio desulfirricans (Norway 4) [45]. At this point it is possible to conclude that the intermolecular clcctron exchange is slow on the NMR time scale for D. gigrrs cytochrome c j at 273K and at 2mM concentration as was the case for Desdjbvibrin vulgaris cytochronie c . ~ [32].

Electron exchange between the haems inside the same molecule (intramolecular electron exchange) results only in

interconversion between the oxidation states belonging to the same oxidation step. The ratios of‘ the populations of the oxidation states in each oxidation step are independent of the solution redox potential (see Appendix A) and remain constant for each step. The observed indcpendcnce of the posi tions of the intermediate resonances on the reoxidation stage of the sample is thcn compatible with either a slow or a fast intramolecular electron exchange. T n fact, if this excl-range were slow, each intermediate resonance would correspond to an individual oxidation state and obviously its intensity would change throughout the reoxidation but its position would be constant. Each haem methyl group would give rise to a maximum of 16 resonances throughout the reoxidation. If the intramolecular exchange were fast, each resonance would then be the average of the resonances corresponding to the oxidation states between which the intramolecular electron exchange was fast: similarly, its position would still not change in this case since the ratios of the populations involved are always constant.

As will be discussed below, the pH dcpendence of the chemical shifts of the resonances in intermediate stages of reoxidation is not compatible with a slow intramolecular

287

IV

Mk I h I1

I I I I 35 30 25 2 0 PPM

Fig. 5. 300-MHz NMR sprc~ra of D. gigas cytochrome c j in different oxidution stuges: detail o f the low-jield region. The redox titration was carried out at 273K and pH* : 7.2

electron exchange. Furthermore, not more than five resonances are ever observed for each heam methyl group (see next section).

Saturation transfer experiments

Saturation transfer experiments offer an effective method for the assignment of the methyl resonances at intermediate stages of oxidation [32]. Performed at different oxidation stages, they allow the cross-assignment of the haem methyl resonances in different oxidation steps.

The resonances are assigned as Mf, where i varies from 1 to 16, according to the position of the haem methyl group resonance in the spectrum of the fully oxidized protein and k represents the oxidation step to which the resonance i belongs (0, T,I1, III, or IV). For instance, MY refers to the haem methyl group 1 in the oxidation step IV.

For clarity oftheexposition let us foilow the resonances M'; (Fig.5). M:, at 23.43ppm is one of the first resonances to appear in the low-field regon of the spectrum, when oxidation starts. Saturation transfer experiments, performed by irradiation of this resonance in stages similar to the one corresponding to the second spectrum from the bottom in Fig. 5,

Table 1 . Chemical shlfs for haem methyl resonances corresponding to dfferent oxidation steps, assigned by satiiration traasjier experiments (pH* = 7.2, T = 273K)

Methyl MP Mf M!' MI" MI" Haem E'OUP resonance (4

PPm

1 3.1 23.43 24.44 33.59 34.07 1 2 - - - 6.51 32.74 4 3 3.1 8.41 26.59 31.24 31.60 2 4 3.2 8.01 22.20 25.60 25.91 2 5 - - 11.0" 21.58 23.11 3

5.41 20.90 4 6 - - -

a Rather weak saturation transfer.

show that there is transfer of saturation to a resonance in the position expected for haem methyl groups of the fully reduced protein, MY (step 0). Proceeding with the reoxidation, M i increases in intensity and other resonances start appearing. Transfer of saturation from M', is then observed to a resonance positioned at 24.44ppm and later on to another resonance at 33.59ppm. This last resonance is later cross-assigned to M:" (at 34.04ppm). Thus the resonance at 33.59 ppm was assigned to MY' and the resonance at 24.44ppm to MI:.

Several other haem methyl group resonances have been assigned in the same way and their positions are shown in Table 1. A maximum of five different positions have been assigned to any haem methyl group. From the above discussion it is concluded that they correspond to the five different oxidation steps involved, implying that the intramolecular electron exchange rate is fast between all the four haems. This will be confirmed later. A similar pattern is observed in NMR spectra obtained at 500 MHz (not shown) giving a lower limit of lo5 s - ' for the intramolecular electron exchange rate.

At present it is impossible to assign all the four methyl groups belonging to each of the four haems, but a careful analysis of the reoxidation patterns and a comparison of the saturation transfer effects during the reoxidation allows the assignment of a sufficient number of haem methyl groups to understand the NMR redox titration. The resonances Mt belong to haem 1. Indeed, the resonance corresponding to step I, Mi, is one of the first peaks to appear in the low-field region, at 23.43 ppm (see Table 1) in the NMR reoxidation titration. This strong paramagnetic shift must necessarily involve a large contribution of an oxidation state with intrinsic shift, since extrinsic shifts, when present, must be small : thus, haem methyl group 1 belongs to the haem with the lowest midpoint redox potential, haem 1 .

For haem methyl group 3, a different situation is observed: the first paramagnetically shifted resonance of this methyl group, ML, appears only moderately shifted, at 8.47 ppm, and it is only the resonance corresponding to step 11, M:, that iirst appears strongly shifted to low field. Thus, the haem methyl 3 must belong to the haem with the second lowest midpoint redox potential, haem 2. Similar reasoning leads to the conclusion that methyl group resonance 5 belongs to haem 3. In the latest stages of the reoxidation, four peaks, M',V, Mk", Mi; and Mi; appear and grow; from these it is possible to transfer saturation

IV

to resonances situated bctwecn 4ppin and 8 ppm. Sincc for these haem methyl goups , the lirst peaks which are affected by ;I large paranxignctic shift bclong to step IV, they must belong to the haem with the least negative midpoint redox potential, i.e. haem 4. Thesc assignments arc later confirmed by thc pH dependence curves.

A fortuitous obscrvatiun provides a conlirniation for the assignment 01'thc methyl group resonances 2,6, 10 and 12. The N M R spectrum o f ; i solution of D. , y i p s cytochrome c j kept at p H * z 9 for severd weeks showed that the intensity of resonances 2. 6. 10 and 12 was very small and four new peaks appear i n the luw-field region of the spectrum (not shown). We conclude t h a t hacm 3 w a s selectively affected, possibly by a change i n a11 axial ligand.

p H r1t~prirtlciic.c o/ 7hc c h r r i i i c d shifis

Since the isoelectric point of D . gigfzpns cytochrome c3 is 5.2 [42] at pH* < 5.5 the protcin starts to precipitate; NMR redox titrations were performed a t pH* values between 5.6 and 9.6.

Fig. 6 shows tho chemical shifts of scveral haem methyl reson;inces plotled ;LS a function of the pll" for the different oxidation stcps. Although the pH dependence of the haem nietliyl group resonances in the fully oxidizcd protein is small

(see the curves for MiV) some resonances corresponding to intermediate oxidation steps show very large shifts. For instance, the chemical shifts of M i and MY changc about 9 ppm and 8 ppni, respectively, with the pH; however, some in- termediatc resonances (e.g. MY, MY' and MY') hardly shift with the pH. Somc resonances (e.g. MIl, MI;. M:') shift to low field with increasing pH while others (e.g. MY', Mi:' and MY:,) show the reverse dependence.

The fact that the pH dependence of some of the resonances in inlcrinediate steps are much larger than those of the corresponding resonances for the fully oxidized state can only be explained if each intermcdiatc resonance results from an averagc of resonances separated by a large chemical shift and whose population ratiochanges with the pH. It was pointed out before that this can only be obtained ifcxch observed resonance is the result of a fast intramolecular electron exchange between thc states belonging to the same step. Within each step for the states i n which a particular niethyl group is attached to an oxidizcd haem a large intrinsic shift to low field is induced to its resonance. However, for the states in which the same haem is reduced, the resonance will bc at higher field, close to the diamagnctic position. Thus, thc chcmic:il shift of the average resonance will depend on the population ratio between these two groups of states. As an examplc, the chemical shift of the resonance MIL rclative to its diainagiictic position is given by

where 2'1 = P l I + P12 + PI, + P , , and I:,, is the intrinsic paramagnetic shift felt by haem methyl group 1 and I :~ ,~ is the cxtrinsic one.

Substituting the relation for the populations P,, described in Appendix A in Eqn (2), this expression becomes:

where C = R?'/F(Fis the Faraday constant). This last equation shows that 6M', depends only on the differences between the hacm midpoint redox potentials and the par:iniagnctic shifts. Since the paramagnetic shifts are little at'fected by the solution pH (see curve for MY i n Fig.6), the large pH dependence observed for 6Mf. implies that the midpoint redox potential of at least one of the haems changes with thz pH.

Ititetisity meusurenzeiits

A qualitative analysis of Fig. 5 shows that the resonance MY' reaches a larger intensity than the maximum reached by either M1! or Mi . As some of the important resonances overlap, quantitative dcterminations of several resonance intensities throughout the reoxidation were performed with a computer- Gtting programme. This programme enabled the decomposition of the spectrum into Lorentzian curves and the determination of the intensities of the individual resonances. The intensity of each methyl group resonance divided by the intensity of the resonance corresponding to the same inethyl

289

Table 2. Pavunzagnetir chemirul shifis rfresonanrcr M i , M;', hl;", M;', MY1 and Mf', the corre.cpoiirr'ingpoulation rniias und the values measured for the intensity maxima The uncertainty for the Rf values is 2 0.001 (see also text). The intensity maxima S, were measured on resonance MIl (i 10 y;), S,, on resonances M I ; and MY ( 2 10 7;) and S,,, on resonances MI,',, MY' and MY' ( k 5 x) pH* Methyl i ,lint Paramagnetic chemical Population ratios (R:) Intensity

shift for for maxima (S,)

step I step 11 step 111 step I step I1 step 111 S, s,, SI,,

PP" p l l - 0 1.2 1 31.0 20.33 21.34 30.49 R', = 0.656 R7=0.689 I - ,985

- ,927

9.6 1 31.1 25.73 25.08 31.10 Rr1 =0.827 RY =0.806 - ,998

- .968

R111 - 0 - - - R111 - 0

Rl11 - 0

3 28.5 - 23.49 28.14 - R Y ~ 0 . 8 2 4 - ,987 0.4 0.3 0.7 5 20.0 - 18.48

3 28.3 - 24.04 27.90 - R: = 0.848 R:"=0.984 0.5 0.3 0.7 5 20.0 19.36 - - - - Rl11 - 0

group in the fully oxidized state was plotted as a function of an arbitrary unit that represents the evolution of the oxidation (not shown). The intensities of the resonances M',, MY, M;, ,, ,, 4 , , z , ,andM',Vwerefollowed. Foreach

spectrum, the intensity values of MY', M:', M:' and MY are identical (within the uncertainty of these intensity measurements) and the same is true for M:", M y , Mi", and MY. Table 2 shows the measured fractional intensity maxima, S,, for step k , where :

~ 1 1 1 MIII MIII MIV MIV MIV

S, = max (P11+P,2+Pl .3+P14) ,

S,,, = max (P4 + P , + P , + P, 1. The maximum of the intensity of the curves obtained for M:" is Sill = 0.7 at pH* = 9.6 and is approximately constant with pH. The pH values at which the resonance intensity measurements can be carried out precisely is limited by the degree of the resonance overlapping and also by the attainment of stable intermediate reoxidation stages. This latter requirement was difficult to fulfil at pH* < 6.5 for the earliest stages of reoxidation.

S,, = max ( P I + P, + P, + P, + P, + PSI,

DATA ANALYSIS

General description

It was shown above that a full description of the electron distribution for a four-redox-centre molecule is rather com- plicated, involving many unknown parameters : the four microscopic midpoint redox potentials, ei ( i = l - 4), and the six interacting potentials, lij (i, j = 1-4, i < I ] .

As shown in the preceding sections, the intramolecular electron exchange rate between the four haems is fast on the NMR time scale. The maxima for the resonance fractional intensities, S,, measured on the NMR spectra, are then the sum of the maximum populations of the oxidation states belonging to step k ; thus they are related to the differences between the haem midpoint redox potentials and the interacting potentials, through the expressions for the population states presented in Appendix A. Furthermore, since the intermolecular electron exchange rate is slow, for each step k , the induced paramagnetic shift for a haem methyl group resonance, 6MT, is the weighted average of the chemical shifts of'the resonances due to the same haem methyl group, i, in the different oxidation states belong-

ing to that step. Thus the NMR data concerning the values of the chemical shifts for the different intermediate resonances (at different pH values) also contain information about the haem midpoint redox potentials and interacting potentials through relations like Eqn (3). However, the expressions for the chemical shifts involve four further parameters. The new unknowns are the intrinsic paramagnetic shifts of each haem methyl group and the lhree extrinsic paramagnetic shifts caused by the other three iron atoms in the same molecule.

At this point, some assumptions must be introduced and subsequently tested in order to reduce the number of unknowns. The extrinsic hyperfine shift felt by a given haem methyl group being due to the dipolar contribution that arises from a through-space interaction, falls off with the inverse cube of the distance to the iron of a neighbour haem [44]. This hyperfine interaction is then short-ranged and the induced extrinsic paramagnetic shifts felt by the haem methyl group are expected to be small [44], particularly when compared to the largest intrinsic ones. A first approximation assuming them to be zero seems reasonable. The intrinsic shift for each haem methyl group can now easily be calculated from the difference between its chemical shift in the fully oxidized state and its diamagnetic position ( z 3 ppm). For example, the intrinsic paramagnetic shift for haem methyl 1 is 31.0ppm at pH* = 7.2 (Table 2).

With this assumption (that the extrinsic shifts are negligible), the expressions for the intermediate resonance positions are much simpler, although nine parameters (three differences between the midpoint redox potentials and six interacting potentials) are still involved for each pH value. The expression for the chemical shift of M: (a methyl group of haem 1) relative to the methyl 1 diamagnetic position, 6M',, becomes:

Analogously for dMY :

Indeed, M', is the average between one resonance corresponding to the haem methyl group 1 in the oxidation state 14 (whose paramagnetic shift is I f n , ) and three other resonances due to the states 13, 12 and 11 collapsing in the diamagnetic position. A similar reasoning applies to MY, and so on: the ratios Rf

290

corresponding to a methyl group resonance M: belonging 10 haem .j, is simply given by the relation between the state populations for which haemj is oxidized and the sum of all the state populations present in step k . As a further example, the ratio corresponding to Mi is given by P , 3 / ( P 1 4 + P , 3 + P , ,

+ P, ,) (see Fig. 1 ) . From the above relations it is apparent that at each step the resonances of the four methyl groups belonging to the same haem must show a similar pH dependence (again if the extrinsic shifts are negligible) and this provides further confirmation for the assignment of the methyl groups to the different haems. The observation of Fig. 6 indicates that resonances M,, M, and M, belong to methyl groups attached to the same haem, that resonances M2, M, and MI, all belong to one other haem, and that M, belongs to yet another one, supporting the assignments of Table 1.

Furthermore. as haem methyl groups 2 ,6 and 10 have been shown to belong to the same haem (see above), the similar pH dependences for MY, MF and MY; further supports the assumed hypothesis about the magnitude of the extrinsic shifts.

The experimental values for the ratios R: can be directly calculated from thc chemical shift of each resonance hlf, for each pH value:

The analysis of the electron distribution in cytochrome c, was based on six independent experimental ratios calculated for resonances Mi (ratio R;), M': (ratio R:'), M! (ratio R;) MY' (ratio R:'), MY (ratio R:') and MY' (ratio R;') and three resonance fractional intcnsily maxima, S,, STI, S,,, (see Table 2). The goal is to obtain quantitative values for the differences between the haem midpoint redox potentials and for the haem- haem interacting potentials. It is important to stress at this point that methyl 1 was assigncd to the haem with the lowest midpoint redox pokntial, hacm 1 methyl 3 was assigned to haem 2, and methyl 5 t o haem 3.

Four identical arid non-iritercictirig liaenw

A simplistic model that considers cytochrome c3 as a molecule with four identical and non-interacting centres [35] can promptly be ruled out. In that case, el = e, = e, = e4 and lij = 0, the maximum for the relative intensity predicted for a haem methyl resonance in an intermediate oxidation state would bc 0.42 (see Fig. 2). As shown in Table 2, the maximum intensity measured for step I11 is 0.7, and the difference is well above the experimental error. This high value is an immediate indication that the population of oxidation state 4 must be much larger than those of states 2,2 and 3, implying that haem 4 must have a midpoint redox potential in step 111 much higher than that of the othcr haems. Taken together, these obser- v;itions show that a model which considers four identical and non-interacting haems is not suitable.

The results obtained by redox titrations coupled to EPR measurements [27] had already revealed that the haem midpoint redox potentials were not identical although a non- interacting model was then assumed.

Tkrr four k G ~ m iii c vtoc lrrvrne cj (Ire interactii2g centres

I n order to know whether the centres are interacting, a simple computer programme was written which allows the yearch for the domains of 1 2 3 and 1 34 that simultaneously satisfy the SIX experimciital ratios shown in Table 2. For each value of the ratio R f - l 2 was expreqsed as a function of A Z 3 and

0 0.02 0.04 0.06 0.08 0.10

A23

Fig. I. Plots of !I,, versus l , , , f o r R', = 0.83 (- I urzd R{/' = 0.80 ( . . ~ 1 , crssuming a nun-imerucring model ( I , = 0) . In each domain the upper curve is plotted for = 0 and the lower curve for

= -200mV. Two completely distinct domains are defined, showing that 110 set of values for A , , , I , , and I,, can satisfy simultaneously R\ and RY at pH* = 9.6

A 3 4 (see Appendix B); /I2, can then be plotted as a function of 4,3 (or 4,,)foragivenvalue of 434(or .I,,). Forinstanceifwe consider 0 to ~ 200 mV as a possible range for 3 -,+, the plots of ,1,,asafunctionof I,,(for I,, = Oandfor.I,, = -200mV) will limit the domain of values for I , , and d z 3 that satisfy the ratio R:. If a non-interacting situation holds for cytochrome c3, then the domains defined for the different experimental ralios Rf should show an intersecting region that contains the three values, .A,, , and / J3 , , which satisfy all the experimental ratios R?. However, even if we only consider the ratios R: (0.83) and RY (0.80) at pH* = 9.6, there is no domain of values ofA,,, .I2, and / J X 4 which satisfy these ratios siinultaneously(if a non interacting model is considered) (see Fig. 7). The same conclusion holds for other pH values.

The necessity to consider a model with interacting potentials could have been anticipated in another way. If the interacting potentials were zero, the ratio R: = P , 4 / ( P , 4 + P,, + P I , + PI,) would vary between 0.25 when all the state populations in each step are identical ( e , = e2 = e3 = e,) and 1.0 when el 4 e2 , e3, e4. On the other hand, the ratio RY =

vary between 0.5 and 1.0 for the same limiting cases. Furthermore, the values of RY would always be larger than those of R:. Tn other words, a non-interacting model would predict a much smaller pH dependence for MI: than for Mi, without any possibility of crossing between the two curves. Thus, such a model is incapable of explaining the experimental results shown in Fig. 6. Indeed, the chemical shift of resonance MY changes by about 9ppm in the pH range studied, while resonance MY changes about 8 ppm; furthermore, above pH* % 8.0 the experimental ratio R : becomes larger than RY.

The redox centres in cytochromc c3 must then interact (i.e. the microscopic midpoint redox potential of a haem depends on the redox state of its neighbours in the same molecule). We will attempt to quantify these interactions later.

(PI" + P, + PJ(P1" + P, + P , + P , + P , + P5) would

Quuntitalive evidence for the p H d e p d r i i L P

of the haenz inic@oint redox potential,

The strong pH dependence observed for some resonnnce~ in intermediate oxidation steps shows that the midpoint redox

291

- > E - 5 0 - t

'a" - 7 0

- 9 0

1 -

-

-

-

- PH

Fig. 8. p H depetidence of the dfJerence between the midpoitit redox potential5 of haems 3 and 4, defitiedfor the equilibrium step III Z step IV. 1z4 was calculated from the experimental ratios RY', RY1 and RY'

potential of at least one of the haems changes with the pH (see Results). It is possible to quantify this effect on the basis of the present NMR data. As the ratios between the populations belonging to step I (P,,, PI,, P,,, Pl l ) are independent of the interacting potentials, Ii j (see Appendix A) some definite conclusions concerning the ranges of variation of A the differences between the midpoint redox potentials e l and e,, for different pH values, can be extracted immediately. Fig. 7 shows that the range of values of ,A,, which satisfies the experimental ratio R', at pH* = 9.6 is contained in a narrow region between 37 mV and 64 mV. At pH* = 6 (curves not shown) the allowed range for / j 2 , is from 10mV to 35mV.

The pH dependence of the haem midpoint redox potentials, defined for the equilibrium step I11 2 step IV, can be calculated more precisely. Since three experimental ratios are available for step I11 (RY', RY' and R!') it is possible to calculate the three differences between the haem midpoint redox potentials for the equilibrium step I11 + step IV, i.e. ei34-e:34 = ,4,*,, e i34-

expressions it was necessary to express the stale populations as a function of the haem midpoint redox potentials 4k'. The procedure is similar to that described in Appendix A, but now the state populations P, are expressed as a function of Zij, 4;,, ~~Az3 and ,43, instead of I . . I , , , , I z 3 and 434. Under these it' conditions, RY', R;' and R,' are independent of Zij in the same way as R', was independent of Iij, in the first treatment (Appendix A), and depends only on A s , , 4,*3 and AQ4. Solving a simple system of three equations and three unknowns, it is possible to calculate the differences between the haem midpoint redox potentials, A ; , , A T 3 and .IT4, for each pH value.

The pH dependence of 1Q4 = e:24-ek2s is shown in Fig. 8. The uncertainty inherent in these values is & 5 mV. A ; , changes from - 50 mV to - 80 mV when the pH* changes from 6 to 9.6. The uncertainty in .1 z, and ,I t3 can be quite large for some pH values (see below).

4 2 4 = I d * 2 3 and e:24-ei23 = A;4. In order to simplify the

T h e order of'magnitudejbr [he haem-haem interacting potentials

In order to quantify the haem-haem interacting potentials a more elaborate method of calculation is necessary. For each pH value, the experimental data available are the six population ratios (RY, RY, R!, RY', RY and RY') and the three resonance intensity maxima corresponding to steps I, I1 and I11 (see Table 2). Again the three ratios, RY', RY' and RF, enable us to calculate and 4T4 for each pH value; ii convenient mathematical manipulation of the expressions for the remain-

Table 3. Solutionsjound by computer search that bestfit the experimental data in Table2 The uncertainty values for the interacting potentials I,, and for the differences (ei - ej) were calculated taking in to account the extent of the family of solutions

Parameter Value at

pH* = 7.2 pH* = 9.6

solution A solution B

mV

> 49 - 16 - 80

14+3 - 2 9 k 4

41k3 -31+3

o + 1 5 1 + l

- 5 & 1 -75+1

36+4

> 49 - I6 - 80

7* 4 - 2 2 5 5

37* 3 5 3 5 5

-43+ 4 I + 3

43* 2 -33+10 -43+10

0.41 0.50 0.49 0.42 0.42 0.42 0.66 0.69 0.69

ing three experimental ratios, R:, RY and R: allows us to express three interacting potentials, e.g. I,,, I , , and II4 as a function of the other three : 123, I,, and Is4. Then, a computer programme was written that searches for the sets of values Zij which generate state populations that fit the measured maxima, S,, for the resonance intensities in the three intermediate steps. The complete analysis was done for two pH* values, pH* = 7.2 and pH* = 9.6 and the solutions found are shown in Table 3. Based on the values for the interacting potentials and the differences between the midpoint redox potentials obtained for pH* = 7.2 (Table3), the sum for the molar fractions of the oxidation states belonging to each oxidation step, as a function of the solution redox potential, was calculated (Fig. 9).

Only one solution family was found for pH* = 7.2, but for pH* = 9.6 two families ofsolutions satisfied the data within the experimental error. In order to rule out one of the solutions at this last pH, further information is required. The calculated ratio Pl3,'(PI4 + P,, + P,, + P , is 0.09 for solution A and 0.1 3 for solution B. At pH* 9, in the early stages of reoxidation a resonance at 6.1 ppm was cross-assigned to MY by saturation transfer; it was then assigned to Mi giving an experimental ratio R i = 0.10. With this support we favour solution A.

Further evidence for the existence of haem-haem interacting potentials is provided by a simple calculation based on the data shown in Table3. As we have explained before. the differences between the haem midpoint redox potentials ( Iz l , It, and AZ4) defined for the equilibrium step I11 s! step IV are directly calculated from experimental ratios RY', R:' and Rt'. Now, the ratios R:, R:' and R': and the intensity maxima S,, S,, and S,,, can be calculated (Table 4), assuming a non-interacting model (Zij = 0). The calculated values (see Table4) show a considerable discrepancy with the experimental ones. Both positive and negative interacting potentials must be introduced in order to obtain a good accordance with the experimental data (Table 3) .

However, the uncertainty in is much smaller and its valucs can be trusted within 5 mV (scc Fig. 8).

There is further information which can be used to test thc above analysis and results. For oxidation step I l l we have assigned a t least one resonance belonging to each hacm: MI;' (haem l) , M!' (haem 4), MY' (hacm 2) and MY1 (haem 3). A simple c a l c u l a h n shows that the theorctical value for the sum of the population ratios corresponding L O each haem in step 111 is 3.0. From the experimental data shown in Tiible 1. obtained assuming all 4,,, = 0, we obtain 3.01 for this sum at pH* = 7.2. This supports the assumption that the extrinsic chemical shifts for the haem methyl groups with the largest paramagnetic shifts arc very small and ncgligible when compared to the intrinsic chemical shifts. Thus the values shown i n Table 3 are reliable at least for pH* = 7.2. At p H * = 9.6 the position of MY1 is very closc to the water pcak, an exact assignmeni. of its position was not possible, and wc can not use this last test.

I t is worth stressing hcre that this does not mean that all the extrinsic chemical shifts are negligible. I n fact preliminary evidence indicates that this is not the case for other haem methyl groups with smaller intrinsic shifts, which were not used in the present calculations. Bcsidcs, sinall extrinsic shifts (z 0.2 ppm) have been detected in cytochromes containing two haems [45 a].

R', 0.460 0.656 0.840 0.827 R:' 0.861 0.689 0.958 0.806 R:l 0.860 0.834 0.67 0.84X s, 0.47 0.41) & 0.03 0.56 0.50 i 0.03 .XI 0.49 0.35 & 0.03 0.49 0.40 f 0.03 SIII 0.62 0.70? 0.03 0.6Y 0.70+0.01

Although thc search for the interacting potentials was made in a broad range (from - 100 mV t o +200mV), the solutions found show interacting potentials roughly between - 50 mV and +60mV. At both pH the valucs for the interacting potentials can be considered similar, except for the interactions between haems 1 and 4 ( f l , ) and between haems 3 and 4 (f3,) which show significant changes with the pH.

Since all the values shown in Table 3 were calculated assuming negligible cxtrinsic paramagnetic shifts for haem methyl groups 1. 3 and 5. the analysis of the uncertainty introduced by the e\,entual prcscncc of intrinsic shifts different from zero, is pcrtinent. The uncertainty of thc values calculated for , I T l , ..IT3 and depends only on the error affecting the experimental ratios obscrved for step 111 : R';', R!' and RY'. The maximum for the resonance inlcnsities in step 111 is rather high, S,,, = 0.7 (Table 2) showing that P, $ P,, P , , P,. Taking this fact into account i t is ciisy to demonstrate that RY', RY1 and Rf:' can only be signiiicantly affected by , I e x l 4 , i.e. the extrinsic shift caused by haeni 4 o n methyl groups 1. 3 and 5, respectively.

Although the expected extrinsic shifts are small [44], the fact that R'," and R!:' are close to 1 (see Table2) leads to a considerable uncert;tinty i n the values calculated for l:, and

At pH' >8.0 the csperimenlal ratio RY' is 2 1.0 (since 6MLI:" = OM':) and only :I lower limit for ' I T I is obtained.

DISCUSSION AND CONCLUSIONS

A considerable effort has been riiade to characteri7e the electron transfer processes that occur in inultiredox centre proteins [41], but the large number of parameters generally involved in describing the electron distribution within the various redox centres requires a selective experimental tech- nique able to providc specific information about the individual ccntres. Optical spectroscopy is in general an ambiguous mctbod. EPR has been shown to be one of the most useful techniques. since different pararnagnctic centres tend t o exhibit well resolved spectra. However, EPR has the disadvantage that the measurements usually have to bc carried out at cryogenic temperatures [41]. Furthermore, in cases like cytochrome c3 , where four similar centres are prescnt, EPR may not be able to resolve the four centres [27].

NMR is a very convenient techniquc for the study of the electron exchange mechanism in cytochronie cj . In fact, each redox centre can be probed through the behaviour of the different resonances of methyl groups belonging to different haems and the experiments can be performed at temperatures close to the physiological values.

From our prescnt data on the electron transfer mechanisms of Desulfovibrio gigas cytochrome c., five main conclusions can be extracted.

a ) The haem microscopic midpoint redox potentials arc different.

b) Thc intcrmolecular electron cxchange rate is faster than

c) There is a redox interaction between the different haems,

d) Some haem midpoint redox potentials are pH dependent. e) At least two of the interacting potentials also c h a n g with

the pH. The data analysis reported for cytochronie c3 by either EPR

[24- 271 or electrochcmical methods [33 - 351 is based on the assumption that 11 non-interacting model holds. This assumption allows great simplification of the analysis. [However, the

1 0 5 ~ - 1 .

the interacting potentials are not negligible.

293

change in the linewidth of the EPR resonances observed for Desulfhvihrio vulgaris (Hildenborough) cytochrome c3 [26] is an indication of such interactions. Early Mossbauer studies [22] also suggcstcd a haem-haem interaction in D. vulgaris (Miyazaki) cytochrome cj and it was recently pointed out by van Leeuwen et al. [39] that the published EPR data [26] on D. vulguris (Hildenborough) cytochrome c j indicate that the differences between the haem midpoint redox potentials are dependent on the degree of reduction. From the data shown in Table 3 it is apparent that both negative and positive haem- haem interactingpotentiak are present in D. gigascytochrome cj . Since the interacting potentials were defined as &j e i - e , positive interacting potentials suggest that the mechanism for the haem-haeni interaction is more complex than a simple electrostatic one and a redox-linked protein conformational change must be involved. It has been reported that some proteins show different conformations in different redox states, e.g. horse heart cytochrome c [46] and Pseudomonas cytochrome oxidase [47].

Evidence for the presence of interactions between the different centres of a multiredox centre electron carrier has been reported earlier, namely for cytochrome c oxidase [48 - 521, cytochrome c peroxidase [53], cytochrome h dimer [54], succinate dehydrogenase [55] and xanthine oxidase [41]. Several kinds of interaction have been postulated : redox interaction, in which the spectral characteristics [56] and the oxidation-reduction potential of one centre is affected by the oxidation state of its neighbours [51] and magnetic interactions between the different paramagnetic centres [57]. However, quantitative data on the redox interactions are rare due to the ambiguity generally involved in this type of analysis [41, 491. Values between - 55 mV and + 90 mV have been reported for the interacting potentials between haems c and d in Pseudomorius cytochrome oxidase [58] and an interaction of about 120 mV was found between the two haemsin cytochrome oxidase 1591.

A comparison of our prcsent data with the values reported for the haem midpoint redox potentials obtained by EPR measurements for the same cytochrome [27] is rather difficult since the experimental conditions, namely temperature, are different. Furthermore, the previous analysis was based on a non-interacting model and in the presence of interactions a unique midpoint redox potential of one haem cannot be defined independently of the stage of the redox titration. However, the approximate Nernst behaviour shown by the EPR-redox titration curvcs [26, 271 suggests that the haem-haem interacting potentials are not very large, in agreement with our results.

Electrochemical measurements are now being performed in D. gigas cytochrome cj (H. A. 0. Hill, personal communi- cation) and it will be interesting to compare them with the NMR data. Wc want to stress here that the values obtained by those methods are macroscopic midpoint potentials and can only be compared with the coresponding parameters which can be calculated from the data in Table 3 (see Table 5). Also the macroscopic midpoint potentials and the microscopic midpoint potentials may show distinct pII dependences, the dependenccs of the macroscopic ones being less accentuated.

In order to extract information about the absolute values for the midpoint redox potentials, a more sophisticated experi- ment has to be carried out so that a measure of the absolute solution redox potential is performed in the NMR tube during the reoxidation. Once this is achieved, the absolute macroscopic midpoint redox potentials can be read directly from the intersections of the curves that represent the resonance intensity in the different oxidation steps versus the solution redox

Table 5. Differencer between [he macroscopic midiioint redox potentiah calculated f rom data in Table 3

1E Value at

pH* = 7.2 pH* = 9.6

rnV

E2 -El 22 37

E3 - Eci ~ 66 - 73 E,- E , - 25 - 23

potential. This would be a simple and direct way to obtain by NMR the information about the macroscopic redox potentials provided by electrochemical methods.

Our present data on D. gigas cytochrome c j show a pH dependence of the haem midpoint redox potentials; however, the definitive identification of the haems whose midpoint redox potentials are affected by the solution pH is not possible until the solution absolute redox potential is measured.

The pH dependence of the midpoint redox potential (redox- Bohr effect) is not a new effect and, on its basis, models for redox proton pumps have been postulated for complex I11 of the respiratory chain [60] and for cytochrome oxidase [61]. Several monohaem cytochromes, namely cytochrome c2 [62] cytochrome cSs1 [63], cytochrome c’ [64], the cytochrome h dimer in complex I11 [54], the multihaem D. vulgaris (Hildenborough) cytochrome c, [32] and many iron-sulfur proteins exhibit pHdependent midpoint reduction potentials [65]. The haem propionic acidic groups have been postulated as the ionizing groups responsible for the redox-Bohr effect in monohaem cytochromes since their pK,, values depend on the oxidation state of the haem iron [63]. The N1-H group of histidine has also been suggested as a possible candidate [66].

The physiological role of cytochrome c3 is still rather poorly understood and a pertinent discussion about the biological significancc or the existence of interacting potentials between the haems and the redox-Bohr effect is necessarily limited. In spite of these limitations, the redox properties of cytochrome c3 described in this article can be anslysed in the light of the physiological activity of Desulfovibrio spp.

It is known that Desulfovibrio spp. can be grown on hydrogen, in the presence of sulfate. It is also known that when grown on pyruvate they reduce sulfate to sulfide and when sulfate is absent, they reduce protons to molecular hydrogen [67]. In the absence of sulfate their growth on lactate is dependent on a co-culture with methanogenic bacteria [68]. In this case, a stoichiometric amount of methane is produced by Metfianosurcinu hurkeri from the acetate, CO, and H, produced from lactate by the sulfate-reducing bacteria. The need for the presence of methanogenic bacteria is apparently a consequence of the high potential of the electrons produced by the oxidation of lactate to pyruvate (Ed = -190mV) in comparison with that for the production of 112 ( E i = - 440 mV). The methanogenic bacteria serve as a scavenger for the H, produced which drives forward the reactions, leading to further formation or H, [68].

Our present knowledge of Desulfovibrio metabolism is consistent with the following hypotheses concerning the physiological role of cytochrome cj .

a) Cytochrome c j may be linked to the formation of molecular hydrogen from both lactate and pyruvate. In this

294

case a charge separation device is not necessary, since all the energy produced is obtained via the formation of acetylphos- phate from the oxidation of lactate and pyruvate [69].

b) In view of the recent hypothesis of hydrogen cycling in Desulfovihrio spp.. both the periplasmic hydrogenase and cytochrome c3 are part of a charge separation device, which is necessary for the net production of ATP during the reduction of sulfate [69]. This hypothesis is supported by the fact that other sulfate-reducing bacteria, such as Desulfotonzuculurn, which do not have cytochrome c3 are unable to perform oxidative phosphorylati on. Also the periplasmic hydrogenase is only active in the reduction of cytoplasmic redox carriers, such as ferredoxin and flavodoxin, when cytochrome c3 is present.

The present additioual information, that the haem midpoint redox potential i s sensitive to the p H of the solution and that there are interactions between the haems, supports the idea that cytochrome c3 plays a complex role in a reaction which has to be carried out at different redox potentials and possibly also involves proton transfer. Although a mechanism for the interaction between the different haems in cytochrome c3 cannot be anticipated from the present study, it is reasonable to speculate that a conformational change induced by the redox stage of the protein can influence the different haem midpoint redox potentials and also the pK, of dissociable groups.

At least two reasons can be mentioned to justify such an enzyme having four c-type haems instead of another mono- haein cytochrome : (a) the capacity of donating or accepting more than one electron at the same time; (b) the possibility of performing a fine tuning of the midpoint redox potential in order to optimize the interaction with other electron transfer partners in different physiological situations which would require a broad range of redox potentials (see Fig. 9). This can be achieved by the modulation of the midpoint redox potentials by different interaction potentials and different responses to the solution pH, enabling the transfer ofelectrons in a very wide range of redox potentials. This capacity has implications for the different possible pathways of Drsulfuvibrio spp. metabolism enabling the involvement of cytochrome c3 in both the production and the consumption of 13,.

One of the potential effects derived from the presence of interacting potentials is the possibility of changing the way in which electrons are accepted and donated. Indeed, although a full discussion is postponed until there is information about the absolute midpoint rcdox potentials, it is apparent (Table 3) that at pH* = 7.2, i n the first step, haems 3 and 4 have similar midpoint redox potentials, and in the last oxidation step their midpoint redox potentials are remarkably different. Also, the midpoint redox potential of haem 4 is affected by the presence of interactions in such a way that the electrons are more easily accepted by a fully oxidized molecule, as well .as being more easily released when the molecule is fully reduced.

The pH effect on the midpoint redox potentials of the haems might also play an important role. Indeed, the protons produced by the dehydrogenases could increase temporarily the midpoint rcdox potentials of the haems, favouring the subsequent acceptance of electrons. Alternatively, the release of protons by cytochrome c j could facilitate its subsequent oxidation. By this mechanism, the proton transfer would assist the electron transfer between cytochrome r3 and hydrogenase, which itself contains. redox centres with pH-dependent midpoint potentials [70].

Cytochrome (‘ .3 is then an extraordinarily versatile molecule and more physiological studies are necessary in order to elucidate complclcly the purpose of this specialized structure.

APPENDlX A

From the 32 Nernst equations that can be written for the 32 electron equilibria between each pair of microstates repre- sented in Fig. 1, the 16 state populations may be expressed as a function of the differences between the midpoint redox potentials e, and the interacting potentials, Iz,:

121 + 2 ’ 2 3 + 4 % + I 1 2 + I , , + z14+ z,, + 124+ I,, c

and F is the Faraday constant. The populations P, can be readily expressed as a function of

the solution redox potential E, since from the Nernst equation for thc equilibrium between states 13 and 15 :

295

APPENDIX B

The six ratios of populations, R',, RY, Rf: , RT', RY' and R Y , defined respectively for the resonances M:, MY, M:, MY, MY' and Mf:' are expressed as follows:

p14

P 1 4 f PI 3 f p 1 2 + '1 1 R: =-

PI 0 + p9 + ps RY = PI,+ P, + P, +P,+ P,+ P,

R: = Pl, + p7 + P6

P,, + P, + P, + P, + P, + P ,

P4+ P,+ P2-t P, RT1 = P4+P3+P2

P4+P3+ P , P, + P3 + P2 + P ,

P4 + P2 + P1 Rt' = P4 + P3 + pz + '

Introducing now the relations for the populations P, given in Appendix A, these ratios can be expressed as a function of and Iij. By a simple rearrangement one can obtain:

) +exp( c ' 2 3 + /' 3 4 + I24

and the expressions for RY', R:' and R!' can be obtained by a similar procedure.

The authors thank Dr Fernando Queiroz Teixeira from Znstituto Gulbenkiuii de Cihcia for the execution of a program to calculate the resonance intensities and Dr Luis Vilas Boas for help in the develop- ment of the other computer programs. We are also grateful to Bruker Company for instrumental facilities with the AM-300 spectrom-

eter, to Mr F. C. Matos for spectrometer maintainance, to Mrs Isabel Carvalho for technical assistance in the protein purification and to Mrs Margarida Martinez for typinz the manuscript. Financial support from Znstituto Nacional de Invesrigapio Cientifica, Juntn Nacional de Znvesriga@o Cientifiea e Tecnolbgica, the National Institute of Healh (GM 28579 to AVX) and the Calouste Gulbenkian Foundation are gratefully acknowledged.

1 .

2.

3. 4.

5. 6.

7.

8. 9.

10. 11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

REFERENCES ' Postgate, J . R. (1954) Biochem. J . 58, xi.

Ishimoto, M. IS., Koyama, J. & Nagai, Y. (1954) Bull. Chem. Soc.

Postgate, J. R. (1956) J . Gen. Microbiol. 14, 545-572. Bruschi, M., Hatchikian, C. E., Golovleva, L. A. & LeGall, J.

LeGall, J. (1963) J . Bucteriol. 86, 1120. Drucker, H., Trousil, E. B. & Leon Campbell, L. (1970)

Singleton, R., Leon Campbell, L. & Hawkridge, F. M. (1979) J.

Ambler, R. P. (1973) System. Zool. 22, 554-565. Shinkai, W., Hase, T., Yagi, T. & Matsubara, H. (1980) J .

Bruschi, M. (1981) Biochinz. Biophys. Acta 671, 219-226. Higuchi, Y., Bando, S., Kusunoki, M., Matsuura, Y., Yasuoka,

N., Kakudo,M.,Yamanaka,T.,Yagi,T. &Inokuchi,H.(1981) J. Biochem. ( Tokyo) 89, 3 659 - 1662.

Haser, R., Pierrot, M., Frey, M., Payan, F., Astier, J. P., Bruschi, M. & LeGall, J. (1979) Nature (Lond.) 282, 806-810.

Pierrot, M., Haser, R ., Frey, M., Pay an, F. & Astier, J.-P. ( 1 982) J . Biol. Chenz. 257, 14341 - 14348.

Higuchi, Y., Kusunoki, M., Yasuoka, N., Kakudo, M. & Yagi, T. (1981) 1. Binchenz. (Tokyo) 90, 1715-1723.

Fauque, G., Herve, D. & LeGall, J . (1919) Arch. Microbiol. 121,

Peck, H. D., Jr & LeGall, J . (1982) Phil. Trans. R. SOC. Lond. B298,

Bell, G. R., Lee, J.-P., Peck, H. D., Jr & LeGall, J. (1978) Biochimie

Moura, J . J. G., Xavier, A. V., Cookson, D. J., Moore, G. R. &

Moura, I., Moura, J. J. G., Santos, H. & Xavier, A. V. (1980)

Yagi, T., Tsuda, M. & Inokuchi, H. (1973) J. Biochem. (Tokyo)

Berlier, Y. M., Fauque, G., Lespinat, P. A. & LeGall, J. (1982)

Ono, K., Keisaku, K. & Kimura, K. (1975) J . Chenz. Phys. 63,

Drucker, H., Leon Campbell, L. & Woody, R. W. (1970)

LeGall, J., Bruschi-Heriaud, M. & DerVartanian, D. V. (1971)

DerVartanian, D. V. & LeGall, J . (1974) Biochim. Biophys. Actu

DerVartanian, D. V., Xavier, A. V. & LeGall, J. (1978) Biochimie

Xavier, A. V., Moura, J . J . G., LeGall, J. & DerVartanian, D. V.

McDonald, C. C., Phillips, W. D. & LeGall, J. (1974) Biochemistry

Dobson, C. M., Hoyle, N. J., Geraldes, C. F., Bruschi, M., LeGall, J., Wright, P. 6. &Williams, R. J . P. (1974) Nature (Lond.) 249,

Xavier,A. V. & Moura, J. J. G. (1978) Biochimie (Paris) 60,327- 338.

Moura, I., Moura. J. J. G., Santos, H. & Xavier, A. V. (1980) Cienc. Bid. jPortuga/,i 5, 189 - 191.

Moura, J . J . Ci., Santos, H., Moura, I., LeGall, J., Moore, G. R., Williams, R. J. P. & Xavier, A. V. (1982) Eur. J . Biochem. 127,

Japan 27, 564- 565.

(1977) J. Bucteriol. 129, 30-38.

Biochemistry 9, 3395 - 3400.

Bacteriol. 140, 893-901.

Biochem. (Tokyo) 87, 1747-1156.

261 - 264.

443 - 466.

(Paris) 60, 315-320.

Williams, R. J. P. (1977) FEBS Lett. 81, 275-280.

Cieizc. R i d . (Portugal) 5 , 295-197.

73, 1069-1081.

PEBS Lett. 140, 185-188.

1640 - 1642.

Biochemistry 9, 1519- 1521.

Biochini. Biophys. Acta 234, 499- 512.

346, 79 - 99.

(Paris) 60, 31 5 - 320.

(1979) Biochiniie (Paris) 61, 689- 695.

13, 1952- 1959.

425 - 429.

151 - 155.

33. Niki, K., Yagi, T., Tnokuchi, H. 61 Kimura, K. (1977) J .

34. Niki. K.,Yagi,T.. Tnokuchi, H. & Kimura. K.(1979)J. Am. Cheni.

35. Sokol. W . F.. Evans, D. H., Niki, K . & Yagi, T. (1980) J . Elec/roana/. (%em / O X , 107 -- 1 15.

36. Bianco, P. &i Haladjian, . I . (1981) Elecrrochim. Actn 26, 1001- 1004.

37. Bitiiico, P. & HaIadjim. J . ( 1979) Riochim. Bioplzys. A m 54.5,86 - 93.

38. Eddo\vcs, M. J . , ElranowskiI, H . QL Hill, H. A. 0. (1979) Biochem. Soc. Trutrs. 7 , 735 - 737.

39. van Leeuwen, J . W., van Dijk. C.. Grande, H. J . & L‘eeger, C. (1982) Eiii.. J. Biochmr. 137. 631 - 637.

40. Fauque, G . (1979) Th&e r k D</<:/i)rctl de SpCciaEite, Marseille. 41. Palmer. G . &Olson, J . S. (1980) in Molybdeiium andhfolybdenurn-

Contuiiiing Eii:ynies (M. Coughlan, ed.) Chap. 5, Pergamon Press, New York.

42. LeGall, J.. Ma/r:i, G . & Dragoni, N . (1965) Biochinr. Biophys. Acra Y9> 385 ~ 387.

43. LeGall. J. , Ljungdahl, P., Moura, I . , Peck, H. D., Jr, Xavier, A. V., Moura, J . J . (3.. Teixeira, M., Huynh, B. H. & DerVartanian, D. V. (1982) Biocheni. Biophys. Re.\. Comnziin. 106, 610- 616.

44. Wuthrich. K. (1076) N M R in Biological Research: Peptides and Proteins, Chap. VL, North Holland Publishing Co., Amsterdam.

45. Moura, I . , Moura, J. J . G., Santos: H. & Xavier, A. V. (1979) Puoreedin,q.v of the 3 r d MeelDzg of /he Porruguese Chemical Socirrj., Porto.

45a. Moura, I.. Liu, M. C., LeGall, J., Peck, H. D. Jr, Payne, W. J., Xavicr. A. V. & Mour;i. 1. J . G . (1984) Eur. J . &OC/7C/71. 141. 397- 303.

46. Taka1io.T. & Dickerson, R. E. (1982)J. M o l . Bid . 153, 95-115. 47. Berger, II. & Whxton. D. C . (19x0) Biochirn. Biophys. Acta 622,

48. Vickery, 1.. E., Palmer, G. & Wharton, D. C. (1978) Biochem.

49. Lanne, D. & Viinngard, T. (1978) Bioc’hinr. Biophp. Acta 501,

Electrochem. Soc. 124, 1979- 1991.

SOC 101, 3335 - 3340.

355.

Bio/Tky,s. Rev. <‘(JrlIU7ZUi. 8O, 458 -463.

449-457.

50. Nicholls, P. &Peterson, L. Chr. (1974) Biochint. Biophys. ileta357.

52. Babcock, G. T., Vickery. L. E. & Palmer, G. (1978) J . Biol. Cherii.

52. Boclens, R. & Wever, R. (1980) FEBS Le/ l . 116, 223-226. 53. Ellfolk, N., Ronnberg, M., Aasa, R., Andreasson, L.-E. &

Vannghd, T. (1983) Biodzinr. Biophys. Acta 743, 23 -30. 54. von Jagow, G. & Engel, N. D. (1980) FEBS Lefl. 111, 1-5. 55. Salerno, J. C., Lim, J., King. T. E., Blum, H. & Ohnishi, T. (1979)

J . Biol. Cheni. 254, 4828-4835. 56. Wilson, D. F., Lindsay, J . G. & Brocklehursl, E. S. (1072) Biochinz.

Biophys. A r i a 256, 277 - 286. 57. Abragam, A. & Biedney, B. (1Y70) E/rctro!t f ~ ~ ~ ~ ~ / l l ~ ~ , ~ ~ r ~ ~ f ; [ ,

Resanuricr of Trunsition lons, p. 560, Clarendon Press, Oxford. 58. Blatt, Y. (1981) D. Phil. Thesis, Rehovot. 59. Wikstrom, M. K. F.; Harmon, H. J., Ingledew, W. J . &Chance, B.

60. Papa, S. (1976) Biochim. Biophys. Acra 455, 39--84. 61. Wikstrom, M. K. & Krab, K. (1979) Biochirri. Bioip/zys. Acra 549,

62. Pettigrew, G . W.? Meyer, T. E., Barcsch, R. G. & Kamen, M. D. (1975) Bioclzini. Biophys. Acta 430. 197- 208.

63. Moore, G. R., Pettigrew, (i. W., Pitt. R. C. & Williams, R. J . P. (1980) Biochitn. Biophphfs. Acta 590, 261 - 271.

64. Barakat, R. & Strckas, T. (1982) Biochim Biophys. A C I L I 679, 393 - 399.

65. Maggliozzo. R. S., McTntosh, B. A . Rr Sweeney, W. V. (1982) J . Biol. Chetn. 257, 3506 - 3509.

66. Ciadsby, P. M. A. & Thomson, A. J . (1982) FEBS Lett. 150, 59 -- 63.

67. LeGall, J. , Moura, J. J. G., Peck, H. D.. .Ir & Xavier, A. V. (1981) in Zron-Su(fur Proteins (T. G. Spiro, ed.) vol. 4, pp. 177 - 247. John Wiley and Sons, New Yurk.

68. Mclnerney, M. J. & Bryant, M. P. (1981) .4ppl. Oir . Microbial. 41.

69. Odoin, J. M. & Peck, H. D., J r (1981) FEBS Lett. 12. 37-50. 70. Cammack, K., Padl, D., Aguirre, R. & Hatchikian, E. C. (1982)

462 - 467.

2.53, 2400-2421.

(1976) FEBS I ~ t t . 65, 259-277.

177 - 222.

346- 354.

FEBS h i t . 142, 289 - 292.

H. Santos, J . J . G. Moura. 1. Moura, ;md A. V. Xavier, Centro de Quimica t..strutural, Complexo Interdisciplinar, Lnstituto Superior TCcnico, Aveiiida Kovisco Pais, P-1000 Lisboa. Portugal

J . LeGall, Ikpartment of Biochemistry. University or Georgia, Boyd Graduate Studies Research Center, Athcns, Georgia. 17S.A 30602

NMR studies of electron transfer mechanisms in a protein with interacting redox centres: Desulfovibrio gigas cytochrome c3

Documents