Phosphate Binding by Cytochrome c SPECIFIC BINDING SITE INVOLVED IN THE FORMATION AND REACTIVITY OF A COMPLEX OF FERRICYTOCHROME c, FERROUS ION, AND PHOSPHATE*

Electron transfer from ferrous ion to ferricytochrome c is affected by inorganic orthophosphate under conditions when the reductant, the oxidant, and the anionic effector combine to form a stoichiometric complex (Taborsky, G. (1979) J Biol. Chem 254, 5246-5251). The hypothesis that a specific phosphate binding site may be involved was tested by exploiting a known side reaction which accompanies the ferrous ion-ferricytochrome c electron exchange. This reaction results in the modification of threonine side chains, of which horse heart cytochrome c has 10. Most of them are in the vicinity of one or more of the even more numerous cationic side chains of the protein. Since the putative phosphate binding site can be expected to be cationic, it was supposed that adjacent to any potential phosphate binding site there would likely be a threonine residue. (This supposition is generally borne out by an inspection of the cytochrome’s three-dimensional structure.) Were the reaction of iron, phosphate, and cytochrome to involve such a cationic site, any adjacent threonine might be expected to undergo the modification reaction preferentially. In fact, one residue, threonine 89, was found to suffer modification about 6 to 20 times more extensively than any one of the other nine threonines. The surface region of which this residue is a part includes the cationic side chains of lysines 86, 87,88, and a&nine 91. Singly or in combination, these side chains are proposed to constitute the specific site involved in the formation of the phosphate-modulated oxidation-reduction complex. The site itself is undoubtedly of biological significance in that it is known to be part of the domain which interacts with other respiratory chain components. As a phosphate binding site it may also serve in some mitochondrial regulating event.

(This supposition is generally borne out by an inspection of the cytochrome's three-dimensional structure.) Were the reaction of iron, phosphate, and cytochrome to involve such a cationic site, any adjacent threonine might be expected to undergo the modification reaction preferentially.
In fact, one residue, threonine 89, was found to suffer modification about 6 to 20 times more extensively than any one of the other nine threonines.
The surface region of which this residue is a part includes the cationic side chains of lysines 86, 87,88, and a&nine 91. Singly or in combination, these side chains are proposed to constitute the specific site involved in the formation of the phosphate-modulated oxidation-reduction complex. The site itself is undoubtedly of biological significance in that it is known to be part of the domain which interacts with other respiratory chain components.
As a phosphate binding site it may also serve in some mitochondrial regulating event.
Proteins become oxidatively modified when exposed to ferrous ion in the presence of air (1). The reaction was shown to occur with a variety of proteins and has the striking feature that the identity of the reaction product is a function of the ionic composition of the reaction medium. In Tris buffer, the major targets of modification are lysine side chains; in phosphate buffer, the principal effect is on side chains of threonine (and serine) residues.
Compared with other proteins, cytochrome c is a special case. In addition to the effect of its aerobic interaction with ferrous ion on some of its amino acid side chains, it can also accept electrons from the added iron when its heme group is initially in the ferric form. Furthermore, both of these concurrent reactions are facilitated or enhanced when phosphate is present. The facilitation of heme reduction appears to be the consequence of the formation of a stoichiometric complex involving cytochrome c, ferrous ion, and phosphate (2). The enhanced susceptibility of its threonine(s) to oxidative modification, compared with other proteins (l), implied the possibility that this reaction too reflects the specific interaction of the heme protein with ferrous ion and phosphate. A specific interaction being potentially of functional significance, we sought to establish whether any of the numerous threonine residues of cytochrome c show exceptional reactivity. We found this to be the case, and we present the relevant evidence in this paper. The results are interpreted in terms of a specific phosphate binding site on the surface of the cytochrome c molecule.

EXPERIMENTAL PROCEDURES'
Ferricytochrome c was reduced with excess ferrous sulfate in sodium phosphate buffer, pH 7.5, in air. Under the conditions of these experiments, the conversion of ferricytochrome c to the ferro-form is nearly complete (about 90%), without an opportunity for the reduced heme iron to undergo appreciable autoxidation (1,3). We had also ascertained earlier that these conditions ensure the essential completion of the oxidative side reaction affecting threonine residues of the protein (1  (I The given numbers refer to positions within the primary structure of horse heart cytochrome c (50). 'The given designation permits identification of the threoninecontaining peptide with reference to additional analytical data contained in the miniprint supplement (see Footnote 1 in the main text). ' No peptide containing Thr-49 was isolated. Given that the peptide isolation procedure screened out any peptide fragments with very low radioactivity, we assume that Thr-49 did not contain radioactivity in excess of the lowest specific activity actually measured (cfi Thr-47).
data permitted a firm identification of the peptide fragment in terms of its location in the cytochrome primary structure, and where it could be viewed as certain that the threonine residue of the peptide received no appreciable contribution of radioactivity from some other, contaminating, fragment.

RESULTS'
The principal finding in this study is presented in This nonspecific susceptibility is likely to be a reflection of the disposition of threonine side chains on the surface of the cytochrome molecule, as a rule (4). It is of interest that Thr-49, the only residue with which we found no significant radioactivity to be associated, is an "internal" residue, hydrogenbonded within the heme crevice and not accessible to interactions at the protein surface.
As reported earlier (l), threonine modification depends on the presence of phosphate.
In Tris buffer, threonine (and serine in other proteins) is completely inert to oxidative modification under otherwise identical conditions. The effect cannot be ascribed to other ions present. Sulfate, introduced as the anion of the added iron salt, cannot be the "effector" of the reaction, since it is without effect in Tris-buffered reaction systems to which it is added in identical amounts. (Sulfate, even at greatly increased concentrations, is ineffective also as a promoter of the electron transfer reaction between ferrous ion and heme iron which is greatly enhanced by phosphate (3).) Sodium ion, introduced as the cationic buffer component of the phosphate system, cannot be held responsible for the "phosphate effect," because the addition of sodium salts other than sodium phosphate is without influence on the product obtained from reaction mixtures which lack phosphate. Conversely, chloride ion, the anionic component of Tris-buffered systems, cannot be considered to act as an "inhibitor" of a general, buffer-independent reaction, because the presence of chloride-containing salts has been noted not to override the phosphate.
Finally, it should be noted that threonine is the only amino acid component of cytochrome c which undergoes appreciable radioactive labeling in this reaction. This was shown earlier (1) in terms of the ion exchange chromatographic profile of the acid hydrolyzate of the phosphate-buffered reaction product of cytochrome c. Radioactive products other than threonine which were eluted within the resolving range of "standard" amino acid analyzer columns contained no more tritium, if any, than threonine residues shown in Table I other than Thr-89. In any case, the noteworthy result of this study appears to be the finding that one particular threonine among the 10 residues of the horse heart cytochrome c contains an exceptional amount of radioactive label.

DISCUSSION
The preferential modification of one threonine residue of cytochrome c, under the conditions described here, suggests a mechanism which involves the interaction of iron and phosphate with a particular surface site on the cytochrome. This view is reinforced by our previously described finding (1) that the reduction of the heme iron by ferrous ion is facilitated by phosphate because cytochrome c, phosphate, and iron constitute a complex within which electron transfer occurs between the added iron and the heme. Phosphate being the apparently specific promoter of both reactions, it is reasonable to suppose that its effect is a reflection of a specific phosphate binding ability of the cytochrome.
Presumably, the interaction involves a particular, positively charged, anion-binding locus on the protein surface.
Cationic clusters, formed by lysine and other positively charged amino acid side chains on the cytochrome c surface, have often been ascribed significance regarding the native structure of the protein, its interactions, and its physiological function. An anion binding site is associated with the crystalline protein (4,5). In solution, P, and nucleotides appear to bind at two loci (6). NMR data implicate His-26 in this binding an extent appreciably below the extent of modification of residues in the cytochrome (1). The specific susceptibility of Thr-89 appears to account for this difference. by guest on July 9, 2020 http://www.jbc.org/ Downloaded from (7). A site near this histidine, possibly involving the neighboring Lys-27, and an additional site centered on Lys-87, emerged as likely Pi binding sites from experiments concerned with electrostatic interactions of cytochrome derivatives with selectively modified lysine side chains (8).
Nucleotide binding is of particular interest to us because we found earlier that ATP and ADP, but not AMP, affect cytochrome reduction by Fe'+ in a manner similar to Pi (3). Cytochrome c affects the phosphorus resonances of ATP and ADP, but not AMP (9). A similar, differential response to nucleotides is revealed in terms of their enhancement of the 695 nm absorption of the form of ferricytochrome c which is dominant at neutral pH (10). A reducing nucleotide, NADH, also prefers that form (11).
Cationic sites are implicated in the promotion of electron exchange with anionic reactants. Iron hexacyanide, in stable complex with cytochrome c, shows enhanced electron exchange (12) and binds at sites which depend on free lysine side chains (13). Speculations concerned with the involvement of Lys-79 have come into question recently (14, 15), but alternatives have been proposed in terms of two clusters formed by . One of these may be identical with one of the Pi sites. This would be consistent with the reported competition between Pi and hexacyanide (16). Pi also competes with dithionite (17). This site may be identical with one of the hexacyanide sites; kinetic differences between the two anionic reductants have been attributed to differences in the lifetimes of their respective complexes with the cytochrome (18).
Such complex formation is, of course, central to our interpretation of the Fe"'-cytochrome interaction.
The formation of an analogous oxidation-reduction complex would be consistent with the kinetics of the reduction of cytochrome c by catecholate (19). A similar complex may be involved in the reaction of cytochrome c with cortisol (this reaction being completely suppressed by Pi) (20).
Anionic free radical reductants have also been considered to bind at positively charged loci (21), but a unique interpretation of their reactions is elusive because they may proceed by a variety of mechanisms, at diverse sites and with various cytochrome conformers (22-24).
Anions affect cytochrome reduction when the reductant is a positive ion, the reaction under study here being an example Anion catalysis of cytochrome reduction by Cr'+ is accounted for in terms of anionic bridges linking the reductant either directly to the heme iron or to another cytochrome site (such as the heme edge), depending on the relative rates of heme iron ligand replacement and the reduction (25). P, binding to a chromium.cytochrome complex has been shown, but not the site of this interaction (26). (In the absence of an interacting anion, Cr has been shown to be bound to Tyr-67 and Asn-52 (27).) Anionic metal complexes will not bind to cytochrome c indiscriminately.
We surmised earlier that certain coordination complexes of Fe2+ (e.g. EDTA) may be unsuited for complex formation with the cytochrome (3). It is now known that blockage of all lysine side chains of cytochrome c is without consequence on its reduction by Fe(EDTA)"-(28), when compared with the reduction rate measured with tht unmodified protein (29).
Functional involvement of cationic sites is well established for interactions of cytochrome c with components of the electron transfer chain. Findings such as the Pi-dependent dissociation of cytochrome c from Keilin-Hartree preparations (30), the effects of Pi on the rate of cytochrome oxidation by cytochrome oxidase (31), and the effects of polycations (30, 32), pH, buffers, and ionic strength (33) on the reductase and oxidase reactions, left little doubt early on that electrostatic interactions are essential. More recently, it has become clear that these interactions involve specific loci of positive charge. The domain of the cytochrome c surface which interacts with cytochrome oxidase is now defined with appreciable precision. It involves lysine residues 8 (34, 35), 13 (34-38), 25 (39), 27 (34, 35), 79 (34), and 87 (35). In contrast, residues 22 (35, 37, 39), 39 (35), 55 (38), 60 (35), 99 (35, 38), and 100 (34) appear to fall outside the interaction domain. (Lysine residues 5, 7,53, 73,86, and 88 have not been covered by these efforts.) It is noteworthy that four of the five invariant lysine residues and a conservatively variable one (40) are implicated in the interaction.
This domain includes, at its periphery, the P, binding site at Lys-87 suggested by an earlier study (8). That this may be a physiologically significant site is suggested by the observation that certain anions, Pi and nucleotides among them, affect the kinetics of the cytochrome c-cytochrome oxidase reaction at concentrations within the physiological range (41). The site may be of consequence in terms of regulation.
The outlines of the interacting domain with respect to cytochrome reductase are also emerging. Extensive blocking of lysine side chains prevents cytochrome c from complexing with cytochrome c, (42). In this interaction, the specific involvement of lysines 8,13,27,72, and 79 has been indicated (43), whereas lysines 22, 25, 55, 99, and 100 appear not to be involved (39,43). It is striking to what extent the two domains of the oxidase and reductase, respectively, appear to overlap. This is of obvious significance for the elucidation of the biological mechanism of electron transfer involving cytochrome c (for relevant discussions, see Refs. 35,[43][44][45].
In most cases of observed anion binding to cytochrome c, the precise location of the binding site is either a matter of speculation or is inferred from findings of interference with binding by the chemical modification of potential sites. The cytochrome c:Fe'+:Pi system appeared to provide an opportunity for the identification of a phosphate binding site in a cytochrome molecule of which the functional integrity remains unimpaired.
In fact, the cytochrome c e phosphate complex has an enhanced electron transfer capability. We consider that the modified residue, Thr-89, marks the site of phosphate binding. This is based on the supposition that any side chain which would be preferentially modified when the protein interacts with Fe'+ and Pi must be in the vicinity of the site of interaction.
The results presented in this paper are well accommodated within the overall context of our knowledge regarding cationic binding sites on the cytochrome surface. Cationic charges proximal to Thr-89 are provided by lysines 86, 87, 88, and arginine 91. We propose that the putative phosphate binding site involved in the reactions described here and earlier (2) is constituted of these residues in some combination.
It is noteworthy that, of these cationic side chains, two (Lys-87 and Arg-91) are invariant in the evolutionary sense (40). It is noteworthy too that a recent study of arginine models has shown them to be particularly suitable for the specific binding of phosphate (49). This proposal is in specific agreement with one of the two likely phosphate binding sites which emerged from experiments concerned with the electrostatic interactions of cytochrome c derivatives with selectively modified side chains (8).
If, as we believe, this phosphate binding site is involved in the binding of Fe'+ and the subsequent electron transfer to the heme iron (2), then the location of the binding site would seem to require a mechanism of electron transfer which is indirect. The binding site is too distant to accommodate direct reductant-oxidant contact between Fe'+ on the one hand and either the heme iron or the porphyrin ring on the other hand 14.

29.
It might also be argued that the preferential modification of Thr-89 reflects only a fortuitous juxtaposition of that par-30 ticular residue to a phosphate binding site which may lack functional significance; "significant" binding may be else-31.
where, remote from the Thr-89 locus and remote also from 32.
other threonines, which would be, therefore, nonreactive in 33.
this svstem. However. the 10 threonine residues of cvtochrome c find themselves in nearly every case in the vicinity of one or 34.
vealed by the x-ray crystallographically determined structure of the protein (40,48). In addition, the finding that phosphate 36. also promotes the electron transfer between Fez' and the heme group within a cytochrome c.Fe2+.Pi complex (2) rein- 37. forces the view that the site marked by modified Thr-89 is a functionally significant phosphate binding site.