Interaction of Cytochrome c, Ferrous Ion, and Phosphate ELECTRON TRANSFER WITHIN A STOICHIOMETRIC COMPLEX*

The rate and extent of electron transfer from ferrous ion to ferricytochrome c are enhanced by the presence of inorganic orthophosphate at concentrations compa-rable to those of reductant and oxidant. Evidence, ob- tained by the method of continuous variations, shows that the electron transfer occurs within a stoichiomet- ric complex composed of cytochrome c, ferrous ion, and phosphate in molar proportions of about 1:l:l. The incorporation of the anion into this complex appears to result in a modulation of the extent and rate of cyto- chrome c reduction. The rate of electron transfer obeys a first order rate law, characterized by an apparent first order rate constant of 1.4 min-‘. The complex has kinetic significance only; equilibrium dialysis, gel fil- tration, and sedimentation velocity experiments yielded no evidence for stable binding of phosphate and iron, or of aggregation, on a significant scale. The extent of reduction is limited (for reasons not yet known) to about one-half of the available cytochrome molecules. Reduction in excess of 50% can be achieved only when both, ferrous ion and phosphate, are present in excess of the cytochrome concentration. Kinetic data indicate that reduction to extents and below 50% occurs by different mechanisms.

Phosphates can interact with cytochrome c and affect its functional properties. For example, Pi markedly enhances the reducibility of cytochrome c by ferrous salts (1,2). The effect is appreciable even when the Pi concentration is a mere fraction of the cytochrome concentration (on a molar basis) and is far below buffer ion concentrations which can be several orders of magnitude higher (1).
The suggestion of specificity, implicit in this kind of sensitivity, invites considerations of possible functional significance. Indeed, there are many, frequently specific ion effects on cytochrome c on record and suggestions abound concerning their physiological relevance (3). For instance, an ion carrier function has been suggested for cytochrome c on the basis of differential ion binding by its ferri and ferro forms (4,5). Nucleotide-cytochrome c interactions have been viewed as reflections of possible regulatory devices (6,7). Ionic bonds are believed to provide the principal link of cytochrome c to neighboring members of membrane-bound electron transport chains (8)(9)(10)(11)(12)(13)(14). Anion effects on the oxidation-reduction potential (l&17), the structure of the protein and its heme conjugate (18)(19)(20), and the interaction of the protein with diverse * This work was supported by Research Grant PCM 72-01998 of the National Science Foundation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
oxidation-reduction reagents (21-28) have received sustained attention for years and the apparently specific binding of an anion to ferricytochrome c crystals has been noted (29, 30)all of these in vitro studies of anion-cytochrome c interactions having produced insights into the properties of the heme protein as an electron carrier.
Against such a background, it seemed desirable to extend our earlier study (l), with a view to the particular possibility that the observed effects of Pi, at low concentrations, on the reducibility of cytochrome c, may be reflections of the formation of a stoichiometric complex composed of oxidant (ferricytochrome c), reductant (Fe"), and anionic effector (Pi). Schematically, this possibility may be depicted by the following reactions: [Fe"-P,-Cy?] [Fez+-Pi-Cyt*+] --) Fe3+ + P, + Cyt*+ This scheme implies that there is direct interaction between Pi and cytochrome c (Cyt) (Reaction 1) and it postulates that electron transfer is an intracomplex process (Reaction 2). The scheme implies no commitment regarding the arrangement of ligands in the iron's coordination sphere except that Pi and protein are considered as possible ligand contributors.
Alternatives to Reaction 1 are not excluded. Indeed, as we shall see, some partitioning of the initial reactants (particularly Fe" and Pi) among competing reactions is likely. Such alternatives could include the formation of other complexes (involving the same constituents in different proportions, or others which are present in the reaction mixture) which may, but need not, be productive of cytochrome reduction via reaction alternatives to Reaction 2. For example, some Fe2+ may donate electrons to an acceptor other than the cytochrome (e.g. 02) or it may reduce cytochrome c without prior complex formation. Finally, the scheme, as shown, is intended to leave the question open for the moment concerning the reversibility of the reactions. In the following, we present evidence for 1) the formation of a stoichiometric complex composed of equimolar proportions of cytochrome c, Fe2', and Pi and 2) electron transfer being, in this case, an intramolecular (intracomplex) process. We applied this method in our search for evidence for the possible formation of a complex involving cytochrome c, Fe'+, and P,. We assumed that complex formation may be revealed not only in terms of some physical or chemical characteristic of the complex itself but also in terms of some consequence of complex formation, such as complex-dependent electron transfer. Accordingly, cytochrome solutions (0.05 and 0.10 mM, respectively, in two series of experiments) were supplemented with FeS04 and Pi, varying the concentrations of these two constituents so that their combined concentrations were constant at a level twice that of the cytochrome concentration (0.10 and 0.20 mM, respectively).
The course of cytochrome reduction was followed as described above. The procedure can be expected to reveal the formation of a "productive" complex even if it is unstable and dissociates upon completion of the electron transfer. The procedure will not reveal the formation of products of side reactions (alternatives to Reaction 1) which are "nonproductive" so far as cytochrome reduction is concerned, unless they make distinctive contributions to absorbance at the two monitored wavelengths. Analysis of complete spectra of reaction mixtures at various stages of the reaction gave no evidence of such alternative products (or intermediates). pendent of the cytochrome concentration (0.05 mM, lower curve; 0.10 InM, upper curue). As noted before (see "Experimental Procedures"), the finding of such an optimum is consistent with the notion that the reducibility of cytochrome c depends on the formation of a cytochrome . Pie Fe'+ complex.

AND
The complex constituents appear to be bound in the molar proportions of about 1:0.8:1.2. There is reason to propose, however, that the actual stoichiometry is simpler than that. There is significant cytochrome reduction even in the absence of Pi. This must. occur by way of some alternative to Reactions 1 and 2 (see introduction).
A contribution to the overall reduction by a phosphate-free reaction must, therefore, be expected, to a maximal extent when Pi is absent and approaching insignificance as we progress from left to right in Fig. 1. A phosphate-free reaction should cause the experimental curves to be skewed. Were the curves in Fig. 1 to depict only the phosphate-dependent reaction, they would be more symmetrical and their maxima more to the right than they are in fact. It seems reasonable, therefore, to postulate that we are dealing with a ternary complex of which the constituents are equimolar.
Reaction Rates in Continuous Variation Experiments-The experiments on which Fig. 1 is based were subjected to kinetic analysis. All reactions in which the Pi:cytochrome ratio was 0.5 or larger obeyed a fist order rate law, through four or more half-lives. The value of the apparent fist order rate constant showed no significant dependence on P, (and Fe'+) concentration over the indicated concentration range and was found to be 0.62 + 0.16 min-' at 0.05 IIIM, and 0.66 f 0.11 mini' at 0.10 IIIM cytochrome c. In other words, the rate constant appeared to be insensitive to the cytochrome c concentration as well, within the indicated standard deviations. Reactions in the complete absence of Pi also obeyed first order kinetics. The apparent rate constant was 0.27 + 0.04 min-' and was independent of Fe'+ and cytochrome c concentrations.
The kinetic course of the reactions within the transition range of molar ratios from 0 to 0.5 (Pi/CytOChrOIIle) was less clear cut. Some kinetic complexity was indicated by occasionally observed, small but significant, initial bursts of cytochrome c reduction, followed by a slower process by which the reduction reached its final level in an apparently fist order fashion.
It is of special interest to find that most of these reactions proceed by first order kinetics and are characterized by a rate constant which appears to be independent of the concentrations of the constituents of the postulated ternary complex. This finding reinforces our view that electron transfer from ferrous ion to ferricytochrome c occurs after a non-rate-limiting association of reductant and oxidant, accounting thereby for the apparently unimolecular reaction mechanism. In the presence of Pi, the complex incorporates this anion. A consequence is the modulation of electron transfer; it becomes more extensive and more rapid. In the absence of Pi, the first order rate law implies a reaction which is similarly unimolecular.
As we shall see in the following section, the suggestion that P, is, in this case, replaced by another anionic constituent of the reaction medium has an experimental basis. Effects of Other Constituents of Fe'+/Ferricytochrome c Solutions-In addition to reductant, oxidant, and phosphate, the reaction mixtures included the components of the Tris-HCl buffer system. Chloride ion is known to interact with cytochrome c (e.g. Refs. 4 and 28). Tris does not interact with the heme protein (4) but it chelates iron, as suggested, for example, by Tris inhibition of the reduction of cytochrome c by the xanthine oxidase system in which an iron atom appears to be essentially involved at the enzyme's cytochrome-reducing site (36). It was, therefore, indicated that we establish what influence, if any, Cl-and Tris may have on the interaction of cytochrome c, Fe'+, and Pi. Tris-HCl buffer itself provides an appreciable concentration of Cl-: about 40 mM in a 50 mM buffer at pH 7.5. It is clear, however, that Cll exerts no major effect, through competition with phosphate, because the effect of Pi is marked even when its concentration is as low as 0.02 mM. This is about 2000 times lower than the concentration of Cl-in the buffer. Nevertheless, Cll has some effect. When a typical, 1:l:l mixture of cytochrome c, Fe'+, and Pi, was supplemented by an additional 50 InM Cl-(as NaCl), the extent of cytochrome reduction was depressed to about 80% of the level noted in the absence of the added Cl-. Stimulation by Pi, at 1/1800th of the Cl-concentration, remained appreciable even if somewhat depressed.
Chloride has no appreciable inhibitory or stimulating effect by itself. With 1:l mixtures of cytochrome and iron, in the absence of Pi, the extent of cytochrome c reduction was the same (9 to 10%) whether the solution was, or was not, augmented by 50 mM NaCl.
The possibility that the Tris component of the buffer had an influence on the reaction was tested by replacement of Tris-HCl with sodium cacodylate. Cacodylate is a nonbinding ion (4). In this buffer system, in the absence of P,, cytochrome c was reduced to a somewhat greater extent than in Tris buffer. Since Pi was not present, we attribute this enhancement to the absence of Tris, a known iron chelator, rather than to the absence of Cl-, the effect of which we have already attributed to its ability to provide anionic competition, when present in massive excess over P,. In the presence of Pi, the buffer replacement was of barely significant consequence. The relatively small amounts of sulfate ion, introduced as the counterion of Fe'+, can be safely ignored. Up to 10 times greater amounts of sulfate had been shown to be without effect (l), consistent with the report that sulfate affects the electrophoretic mobility of cytochrome c just ahead of cacodylate, the last member of a series of ions in order of their decreasing effectiveness (37).
We conclude that both buffer constituents may exert a small influence without affecting the principal thrust of our interpretation of Fig. 1. The weak competition which Clappears to be capable of, uis o uis Pi, makes the first order reaction rates in the absence of Pi explicable. Chloride may take the place of Pi as the anionic component of the complex. In a Cl--containing complex, modulation of electron transfer may be less effective, as suggested by the small observed rate of reduction (0.27 min-') in the absence of Pi. Tris may provide a relatively weak competition for iron, possibly by way of one of the alternatives to Reaction 1 considered in the introduction. This competition appears to suffice to make a portion of the Fe2+ unavailable for cytochrome reduction, but it has no effect on the enhancement of cytochrome c reduction by P,.

Consideration
OfAlternatives to the Complex Hypothesis-Although the postulation of a ternary complex offers the most direct and simple way to account for Fig. 1, and for the rates of the underlying experiments, an alternative interpretation not invoking a complex should be considered. To find that the reduction becomes increasingly effective as the relative concentration of Fe'+ is raised accords with normal expectations; there is more reduction with more reductant. A problem arises as we note that the reaction extent, after it passed through an optimum, falls even while the Fez+ concentration continues to rise. We might suppose that a nonreductive, Fe'+-consuming reaction is promoted at high Fe'+ concentrations. Such a reaction would cut down on the availability of Fe'+ for cytochrome reduction and produce results like those in Fig. 1. For this to be a realistic alternative, however, it would require that the nonreductive Fe2+-consuming process be switched from a secondary role (at low Fe'+ concentrations) to a dominant role (at high Fe'+ concentrations).
For such a switch, we need to assume that the nonreductive reaction proceeds with a higher order of dependence on Fe"+ concentration than the reductive process. Although we can speculate reasonably about the occurrence of higher order reactions of iron (e.g. the formation of some polynuclear iron complex), we can rule out its applicability in this case on experimental grounds. We showed earlier (1) that cytochrome reduction proceeds to monotonously increasing extents as the concentration of Fe'+ is raised, even to levels many times higher than the maximal levels reported in this paper. We conclude, therefore, that the postulated ternary complex remains the simplest hypothesis by which the relatively efficient reduction of cytochrome c, at particular P, and Fe'+ concentrations, can be accounted for. conditions. The incompleteness of the reduction and the greater reaction extent at the higher cytochrome concentration suggest this. They invite consideration of the question whether, and under what conditions, the Pi effect might be "saturating," that is produce a true maximum of extent and possibility even of rate of reduction.
We might suppose that Reaction 1 is an equilibrium process which fails to ensure complete association of the complex constituents because Kamociation is unfavorable. Alternatively, we might suppose that Reaction 1, as a practically irreversible process, may occur at a rate which is slow enough to permit alternative reactions to occur on a significant scale (cf introduction). In either case, the implied limitations on the reductive success of Reactions 1 and 2 could be overcome by concurrent concentration increases of all of the reactants of which the complex is constituted. This prediction presupposes that Reaction 1 is the only process which involves all three of the complex constituents and no other reactant. Alternatives to Reaction 1 presumably depend, in part, on other components of the reaction mixture (e.g. Tris, Cl-, 02) of which the concentrations would be left unchanged. (Given the inferences drawn from Fig. 1, the originally considered possibility of complexes which differed only in the molar ratios of identical constituents can be ignored.) Fig. 2A shows results of experiments performed with the above rationale; the absolute concentrations of ferricytochrome c, Pi, and Fe'+ were varied but equimolar in every case. It is clear that the extent of the reaction can be made to show the effect of saturation although only at the level of about half-reduction of the available cytochrome c. While puzzling, this apparent "half-of-the-sites" reactivity was not unexpected. shows, confiiing earlier findings (l), that when a 1:l mixture of oxidant and reductant is supplemented by Pi in increasing proportions, the reaction extent increases sharply at first, approaching half-reduction and then drops gradually as the Pi concentration rises to great molar excesses. Fig. 2B also shows the results of a parallel experiment (upper curve) in which the Fe'+:cytochrome ratio was fixed at 2:l (i.e. a 100% molar excess of Fe2+). This experiment makes it clear that the barrier to complete cytochrome reduction can be breached when both Pi and Fe2+ are provided in excess. However, it is clear, too, that reduction beyond the 50% level is achieved with greater difficulty in the sense that the reaction extent becomes a less emphatic function of Pi concentration beyond the 50% reduction level. It is noteworthy that the facility with which low Pi concen- trations can promote up to about 50% cytochrome reduction is associated with equimolar mixtures of cytochrome c and ferrous ion (CL Fig. 2, A and B). Reduction beyond the 50% level, promoted with lesser ease by higher concentrations of P, and only when Fe '+ is present in excess (iron:cytochrome, >l), must proceed by a different reaction mechanism.
It is instructive to consider the kinetics of the reactions depicted in Fig. 2. Reactions which involve equimolar reactants at different absolute concentrations (cf. Fig. 2-4) obey first order kinetics at concentrations up to 0.2 mM. Beyond this concentration, the kinetics become somewhat complex, coincident with the increasingly depressed reaction extent as the reactant concentrations rise above the 0.2 mM level. (Both kinetic complexity and depressed reaction extent are likely to be manifestations of some aggregation phenomenon since the specificity of the protein-Pj-iron interaction may lose its dominant significance.) Reactions which involve iron and cytochrome in 1:l and 2:l molar ratios and varying relative amounts of Pi (cf Fig. 2B) also obey first order kinetics so long as the reaction extent stays below 50%. Beyond 50% reduction, the kinetics becomes emphatically complex, as one might expect in view of the speculation (cfi preceding paragraph) that at excessive Pi and Fe'+ concentrations some new reaction mechanism becomes operative, superimposed on the intracomplex mechanism. First order rate constants are summarized in Table I. The "saturation effect" which was noted in terms of the reaction extent can also be recognized in terms of reaction rates. Before kinetic complications are allowed to obscure the picture (at high concentrations of all reactants, and at high concentrations of Pi when Fe" is also present in excess), the first order rate constant approaches a maximal value of about 1.3 to 1.5 min-'. This appears to be the true rate of the reaction within the ternary complex. From the rates calculated for the reactions in the continuous variation study which we had reason to believe occurred under less than optimal conditions, it is clear that the apparent first order rate constants calculated for those reactions must have represented a sum of first order reactions including electron transfer within the cytochrome. Pi. iron complex and also the less facile reaction within a less effectively modulated complex containing Cl-as the anionic component.
Stability of Cytochrome -P,. Iron Complex-The postulated complex appears to have only kinetic significance.
Dialysis equilibrium experiments conducted with 0.25 mM ferricytochrome solutions and with equimolar concentrations of Pi (in the usual buffer) showed that phosphorus concentrations on the two sides of the membrane were identical within analytical error. We estimate that an association reaction with K of the order of lo3 Mm1 would have been detected. This is not inconsistent with reported binding constants (5,38). Iron was not used in these experiments; precipitation of, presumably, iron hydroxides was inevitable during the long periods required. However, typical reaction mixtures including iron were gelfiltered. In experiments in which the Pi:cytochrome ratio ranged between 2 and 200, less than 0.1% of the phosphate was eluted in fractions containing the protein. The resolution of protein and anion was similarly complete when the column was charged with a cytochrome/Pi/iron mixture (in equimolar amounts). Sedimentation analysis was carried out primarily for the purpose of detecting any aggregation of the protein as a consequence of its interaction with Pi, iron, or both. The sedimentation coefficients (in six experiments) were identical within 0.05 S with the sedimentation coefficient of the protein under the same conditions. Also, the areas under schlieren "peaks" were identical, within the limits of planimetric integration, throughout an experiment. This indicated that no significant amount of protein was removed in the form of rapidly sedimenting aggregates. Remaining Questions-The evidence presented and discussed so far appears to establish that the phosphate-dependent enhancement of electron transfer from ferrous ion to ferricytochrome c is a consequence of the formation of an unstable but kinetically significant complex constituted of cytochrome c, ferrous ion, and inorganic phosphate in the molar proportions of 1:l:l. The anion appears to modulate the reaction by increasing its efficiency and facility.
Several unanswered questions remain. It is not known in what way, if any, mediation may be involved in the electron transfer. Under the conditions of these experiments, the initial transfer of electrons from Fe'+ to 02 must be considered, in view of the stimulating effect of 02 (2) and the well documented reducibility of cytochrome c by the superoxide anion (36,(39)(40)(41)(42). However, the reported lack of inhibition of cytochrome reduction by Fe" in the presence of 02-scavengers makes the involvement of superoxide unlikely (2). In addition, the possibility of direct O2 binding to the postulated 1:l:l complex would also argue against superoxide involvement in view of the proposal that superoxide generation is associated with aerobic transition metal systems when direct coordination of O2 to the metal is not feasible (43). We suspect that the autooxidation of ferrocytochrome c, in systems in which iron and phosphate are present in excess (l), may involve superoxide which has been suggested to be generated as a product of cytochrome autooxidation, causing its apparent deceleration (44). A second unanswered question relates to the apparent saturation of the Pi effect when the cytochrome is only halfreduced. An experimental accounting for this phenomenon is yet to be accomplished.
One clue may be provided by our earlier observation that cytochrome c reduction exceeding 50%, by excess Fe" (in Pi buffer), is followed by relatively rapid autooxidation, in contrast to cytochrome c reduced to extents below 50%. We suggested above that the mechanisms of cytochrome reduction by Fe'+ to extents below and above 50% differ. Exploration of the Pi effect under anaerobic conditions may be useful in an effort to elucidate the mechanism of reduction beyond the half-reduced stage. Finally, it remains to be shown where on the cytochrome surface the interaction with Fe" and Pi occurs. In view of the apparent specificity of the interaction, there must be a particular binding site associated with it. Locating the site is desirable because 1) it would confirm the specificity of the interaction, 2) it would permit useful comparisons with other known or suspected anion binding sites, and 3) it could assist in the definition of the electron transfer mechanism. We plan to report shortly on the results of experiments which bear on the identity of this anion binding site.'