Role of a disulfide bond in the gamma subunit in activation of the ATPase of chloroplast coupling factor 1.

The relationship between activation of the latent ATPase activity of isolated chloroplast coupling factor 1 (CF1) and reduction of a disulfide in the gamma subunit has been assessed. The sulfhydryl residues involved in the disulfide bond are distinct from residues normally accessible to maleimide modification during incubation of thylakoids in the dark or the light. Dithiothreitol-induced activation is time dependent, and correlates with reduction of the disulfide. Sulfhydryl residues exposed during activation can be reoxidized to disulfide by incubation with iodosobenzoate , with a concomitant loss of ATPase activity. Activation and deactivation are reversible, but deactivation is prevented by treatment of the reduced enzyme with N-ethylmaleimide. Heat activation does not reduce the disulfide bond unless dithiothreitol is present during activation. Prior heating of CF1, which partially activates the enzyme, renders the disulfide more susceptible to subsequent dithiol reduction. The activity obtained when heat and dithiothreitol are used together is approximately equal to the sum of the partial activations obtained with heat or dithiothreitol alone. Iodosobenzoate has no effect on heat-activated CF1. Enzyme activated by heating in the presence of dithiothreitol can be partially deactivated, consistent with reversal of the activity attributable to the dithiol effect. Fluorescence polarization of anilinonaphthylmaleimide bound to the reduced enzyme indicates that the sulfhydryl residues involved in the disulfide are in a less rigid environment than the other two sulfhydryl residues in the gamma subunit. Polarization of anilinonaphthylmaleimide bound to these sulfhydryls is reduced by heat treatment of CF1. The increased susceptibility of the disulfide to reduction upon heat treatment, and the activation of ATPase activity with or without disulfide bond cleavage are indicative of conformational changes within the gamma subunit that occur during the conversion of CF1 from a latent to an active ATPase. In addition the results are consistent with at least two distinct conformational forms of CF1 that can hydrolyze ATP.

Role of a Disulfide Bond in the y Subunit in Activation of the ATPase of Chloroplast Coupling Factor 1" (Received for publication, August 31, 1983) Carlo M. NalinS and Richard E. McCarty The relationship between activation of the latent ATPase activity of isolated chloroplast coupling factor 1 (CF,) and reduction of a disulfide in the y subunit has been assessed. The sulfhydryl residues involved in the disulfide bond are distinct from residues normally accessible to maleimide modification during incubation of thylakoids in the dark or the light. Dithiothreitolinduced activation is time dependent, and correlates with reduction of the disulfide. Sulfhydryl residues exposed during activation can be reoxidized to disulfide by incubation with iodosobenzoate, with a concomitant loss of ATPase activity. Activation and deactivation are reversible, but deactivation is prevented by treatment of the reduced enzyme with N-ethylmaleimide.
Heat activation does not reduce the disulfide bond unless dithiothreitol is present during activation. Prior heating of CF,, which partially activates the enzyme, renders the disulfide more susceptible to subsequent dithiol reduction. The activity obtained when heat and dithiothreitol are used together is approximately equal to the sum of the partial activations obtained with heat or dithiothreitol alone. Iodosobenzoate has no effect on heat-activated CF,. Enzyme activated by heating in the presence of dithiothreitol can be partially deactivated, consistent with reversal of the activity attributable to the dithiol effect.
Fluorescence polarization of anilinonaphthylmaleimide bound to the reduced enzyme indicates that the sulfhydryl residues involved in the disulfide are in a less rigid environment than the other two sulfhydryl residues in the y subunit. Polarization of anilinonaphthylmaleimide bound to these sulfhydryls is reduced by heat tretment of CF,.
The increased susceptibility of the disulfide to reduction upon heat treatment, and the activation of ATPase activity with or without disulfide bond cleavage are indicative of conformational changes within the y subunit that occur during the conversion of CF, from a latent to an active ATPase. In addition the results are consistent with at least two distinct conformational forms of CF, that can hydrolyze ATP.
CFl' catalyzes the synthesis of ATP using energy derived _____ * This work was supported in part by Grants PCM 79-1196 and PCM 82.14011 from 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.
$ Eppley Foundation Postdoctoral Fellow. The abbreviations used are: CF,, chloroplast coupling factor 1; from the proton gradient across the thylakoid membrane (1)(2)(3). When isolated from the membrane, the enzyme loses its capacity for net ATP synthesis, but retains a latent ATP hydrolysis activity. This activity is expressed after treating the enzyme at room temperature with detergents (4), reducing agents (5), or proteases (6), or after heating CF, in the absence or presence of reducing agents (6,7). CF1 consists of five distinct subunits ( w " ) in a stoichiometry that remains controversial. The molecular weight has recently been redetermined by various techniques (8); the value obtained (400,000) and the molecular weight of the subunits suggest a subunit stoichiometry of 01~p~y6c.
The role of sulfhydryl residues in energy transduction remains a focus of investigation. At least one sulfhydryl residue in the y subunit is essential for ATP synthesis (9, 10). Modification of this residue in the membrane-bound enzyme under energized conditions (i.e. illuminated thylakoids) inhibits ATP synthesis and also inhibits ATP hydrolysis by both the membrane-bound and soluble CF1. In addition to the essential residue, a nonessential sulfhydryl group is exposed under both energized and nonenergized conditions in the membrane-bound enzyme (10, 11). These residues have been shown to be distinct, and are not different conformers of the same reactive group (12). Controversy over the stoichiometry of sulfhydryl residues is, in part, related to changes in the oxidation-reduction states of these residues that occur during activation of the latent ATPase activity. Dithiothreitol-induced activation causes reduction of a disulfide bond in the y subunit of soluble CF1 (13). Recently, a model has been proposed for heat activation in which a disulfide exchange between the y and 01 subunits occurs, rather than a net reduction (13). Since ATPase activity can be expressed by heating CF1 in the absence of added reducing agents (6, 7), an activation mechanism that does not require net reduction bears consideration.
Until recently, the energy-dependent activation of thylakoid ATPase by thiols was generally considered to be a curiosity of little physiological significance. However, the finding that illumination of intact chloroplasts (14) or of leaves (15) causes a similar induction of the ATPase activity of thylakoids derived from them, has rekindled interest in the thiol-activated or reduced state of CF,. The active form of CF1 in vivo is likely to be the reduced form.
In this paper, we compare the activation of the latent ATPase activity of soluble CF1 by dithiothreitol and by heat. Although heating in the presence of dithiothreitol promotes the reduction of a disulfide bond in the y subunit, no change in the oxidation state of this subunit was detected when heating was carried out in the absence of the reducing agent.

7275
While we agree that heat activation and reductive activation occur by different mechanisms (18), in our hands, only in the presence o f an added reducing agent is the disulfide bond in the y subunit reduced.

MATERIALS A N D METHODS
Chloroplast thylakoids were prepared from market spinach (16).
('I?, was isolated essentially as described in (17). The sucrose gradient centrifugation step was omitted. The enzvme was determined to he more than 9 V i pure hy SIX-polyacrylamide gel electrophoresis. with ribulose bisphosphate carboxylase being the major contaminating protein.
Modification of soluble CFI with ANM was performed in 50 mM Tris-CI (pH 8.0). 1 mM EDTA, and 1 mM ATP buffer. CF, was incuhated at 0.5-2.5 mg of protein/ml with 30-50 p M ANM for 3 min. Either 0.5 mM dithiothreitol or 1 mM N-acetylcysteine was added to stop modilication and the samples were passed through a 1 X 10 cm Sephadex (i-50 column, equilihrated with the Tris-EDTA-ATP so-Itflion. When small volumes were used, Sephadex centrifuge columns (18) were used to remove the unhound reagents. Where premodificat ion of sullhydrvl groups with N-et hvlmaleimide was required, CFt in the Tris-EI)TA-ATf' solution was inquhated with 1 mM maleimide lor 15-30 min prior to the treatments. Reduction of disullide bonds was accomplished by incuhating the latent enzyme with -50 mM dithiothreitol for :3-5 h, or 10 mM dithiothreitol for 15-30 min lor denatured CF,. Where SDS was added to denature protein, a final concentration of 1 ri was present.
N-(I-Anilinonaphthvl-4)-maleimide was purchased from Wako I'ure Chemical Industries, I,td., .Japan, and used without further purification. On the hasis of its reactivity with cysteine or N-acetylcysteine, the ANM was greater than % r i pure. Stock solutions in dimethyl sulfoxide remained stahle for weeks at 0 "C. such as ANM provides a simple, visual means of locating the sulfhydryl residues and disulfides in the enzyme. Fig. 1 shows an example of' this approach. Soluble, nondenatured CFI contains ANM-reactive groups (presumably sulfhydryls) that are predominantly in the y and c subunits (Refs. 10 and 11; Fig. 1, I,anc2 1 ). Trypsin treatment of this enzyme (23) releases a peptide from the 9 subunit having M, -6000 and containing all the fluorescent label (not shown). A similar pattern of digest ion of y was observed using CF, isolated from thylakoids modified with ANM in the dark (12). The similarity in the size o f t he ANM-labeled peptides released by trypsin digestion suggests that the reactive sulfhydryl in the y subunit of the soluble enzyme is the same one that is accessible on the membrane-bound enzyme in the deenergized state. The fluorescent label associated with the c subunit is presumably bound to the single sulfhydryl residue (11, 17, 24) in that subunit. When CF, is denatured in SDS (Fig. 1, Lane 2 ) , additional sites in (Y, /$, and 9 subunits become accessible to labeling. Only the d subunit does not show any covalent incorporation of ANM, consistent with another published report that this subunit does not contain sulfhydryl groups ( 2 5 ) . When denatured CF, is first exposed to a large excess of N-ethylmaleimide, none of the sulfhydryl residues will sub- sequently react with ANM (Lane 3 ) . However, if after incubating with N-ethylmaleimide, the enzyme is treated with dithiothreitol prior to ANM addition, ANM is incorporated specifically into the y subunit (Lane 4 ) . These results are indicative of the presence of a disulfide bond only in the y subunit. No evidence for intersubunit disulfide bonds was obtained, in agreement with previous reports (13. 25). T h e increased electrophoretic mobility of the y subunit observed after modification is likely caused by bound ANM, and not by cleavage of the polypeptide chain.

/,omtion of Sulfhydpl Residues and the Disulfide Bond in
The sulfhydryl residues on the y subunit exposed by dithiothreitol reduction are distinct from the dark-or light-accessible sulfhydryl residues of the membrane-bound enzyme (9,10). Thylakoids were incubated with N-ethylmaleimide in the light or the dark and CF, was isolated. The purified CFI preparations were denatured in 1% SDS and incuhated in the presence (+) and absence (-) of 10 mM dithiothreitol. Following removal of the dithiothreitol, the preparations were incubated with ANM. Although some ANM was incorporated into the y subunit of CFI from thylakoids incubated with Nethylmaleimide in the dark, increased ANM incorporation was seen after dithiothreitol reduction (Fig. 2, 1,anc.s 1 and  2). Much less ANM was incorporated into the 1' subunit of CF, from thylakoids incubated with N-et hylmaleimide in the light (Lane 3 ) , but substantial ANM fluorescence was seen after reduction (Lane 4 ) . Since the prior alkylation of the sites accessible to N-ethylmaleimide in the light did not affect ANM incorporation after reduction, it is unlikely that these groups are part of the disulfide bond in the y subunit and that the disulfide bond is reduced upon illumination of thylakoids. Consistent with this result, dithiothreitol reduction is still required to elicit trypsin sensitivity of the 7 subunit ( 2 3 ) of enzyme released from the membrane after modifying in the dark or the light with N-ethylmaleimide (data not shown). 2. ANM labeling of CF, from N-ethylmaleimide-labeled thylakoids. Thylakoids were incubated with 2 mM N-ethylmaleimide lor 2 min in the dark or in the light as described in (12).

DARK LIGHT
CF, was isolated and denatured in 1% SDS, and 10 mM dithiothreitol was added as noted. After 30 min, dithiothreitol was removed, and 50 p~ ANM was added for 3 min followed by 1 mM N-acetylcysteine. Fifteen pg of each sample were subjected to electrophoresis on a 10% acrylamide gel. Note that only the cu(Jy subunits are shown. Activation Effects on the Sulfhydryl Residues of the y Subunit-In its soluble form, CF, is a latent ATPase. Expression of ATP hydrolysis requires treatment with dithiothreitol, heat, or a combination of these two methods. Dithiothreitol activation is time-dependent and concentration-dependent (13). The rate and extent of activation are insensitive to medium ATP present during activation (not shown); however, these observations do not rule out possible involvement of enzyme-bound nucleotides in the activation process. Both the Ca2+ and M$+-dependent ATPase activities are stimulated by dithiothreitol activation. Hydrolysis rates are greater with Ca" than with M$+ a t all ATP concentrations tested.
Dithiothreitol treatment of CF, reduces the disulfide bond in the y subunit (13, 23). The sulfhydryl residues exposed by dithiothreitol treatment can be modified by maleimides without requiring that the protein first be denatured. Because of this accessibility, the extent of disulfide reduction can be monitored by measuring incorporation of ANM into the exposed sites. Fluorescence due to labeling of the accessible sulfhydryls on y and c subunit was reduced by pretreating latent CF, with N-ethylmaleimide. A correlation was observed between the extent of reduction and the rate of ATP hydrolysis (Fig. 3). SDS-polyacrylamide gel electrophoresis shows that virtually all the fluorescent label is incorporated into the y subunit (Fig. 3, inset).
Following removal of dithiothreitol, the ATPase activity of the reduced enzyme decays with time as the sulfhydryl residues are reoxidized. Iodosobenzoate, an oxidizing agent that converts vicinal sulfhydryl residues to a disulfide linkage, facilitates this deactivation. It was previously shown that treatment of activated CF, with 5 mM iodosobenzoate for 1 h has only a partial effect on deactivation (13). However, a t lower concentrations, up to 90% of the activity can be reversed in 10 min (Fig. 4). Iodosobenzoate-induced deactivation was prevented by reacting the reduced enzyme with N-ethylmaleimide (Fig. 4) or iodoacetamide (13). Under these conditions, N-ethylmaleimide does not react with the essential sulfhydryl. The partial effect a t high iodosobenzoate may be due to sulfhydryl oxidation to a sulfone or sulfinic acid, rather than to a disulfide. Treatment of latent enzyme with iodosoben- were passed through two successive centrifuge columns to remove dithiothreitol. ANM (30 p~) was added to each sample, and the fluorescence (W) was measured as described under "Materials and Methods." At the same time, aliquots were assayed for ATPase activity (0). Inset: aliquots from each sample used for the fluorescence measurement were precipitated with 2% trichloroacetic acid after removing unreacted ANM with 1 mM Nacetylcysteine. Twenty pg of each sample were subjected to electrophoresis on a 12% acrylamide gel. Iodosobenzoate effect on dithiothreitol-activated CFI. CFI a t 1.8 mg of protein/ml was activated with 50 mM dithiothreitol for 3 h. After removing dithiothreitol on a Sephadex 12-50 desalting column, half of the enzyme was treated with 2 mM Nethylmaleimide. The enzymes were then incubated with iodosobenzoate at the concentrations given for 10 min. ATPase was assayed as described. As a control, latent CFI was also incubated with iodosobenzoate. 0, -N-ethylmaleimide; U, +N-ethylmaleimide; 0, latent CFI. zoate also causes some activation, perhaps by direct oxidation of the disulfide. The rate of deactivation by low concentrations of iodosobenzoate is very fast relative to the rate of dithiothreitol activation even a t high dithiothreitol levels. This suggests that conformational changes make the vicinal sulfhydryls more accessible to the medium. These conformational changes are fully reversed by iodosobenzoate oxidation, since the rate of reactivation by dithiothreitol after iodosobenzoate treatment is the same as for the untreated enzyme. ATP had no effect on the efficacy of iodosobenzoate, nor is there a significant difference in the deactivation after short or long incubation with dithiothreitol (data not shown).
Heat treatment of CF, in the absence of added dithiothreitol also partially activates the latent ATPase (7). However, in our hands, this does not cause reduction of the disulfide bond, or formation of an alternate disulfide ( i e . disulfide exchange). This was determined by denaturing CF, and blocking free sulfhydryl residues with N-ethylmaleimide. Subsequent dithiothreitol treatment and exposure to ANM would label residues that had been part of a disulfide linkage. Both the latent enzyme (Fig. 5, Lane 1 ) and the heat-activated enzyme (Lane 2 ) contain a disulfide bond in the y subunit alone. No disulfide bond is detectable in the (Y subunit in either enzyme (the small, fluorescent band at the leading edge of the / 3 subunit is due to a ribulose bisphosphate carboxylase contamination). In contrast to heat-activated enzyme, dithiothreitolactivated CFI does not contain any disulfide bonds (Lane 3 ) . Iodosobenzoate treatment of dithiothreitol-reduced enzyme reforms a disulfide bond in y (Lane 4 ) .
Consistent with its effect as a sulfhydryl-oxidizing reagent, iodosobenzoate has no effect on the Ca*'-ATPase activity of the heat-activated enzyme, and only a partial effect on CFI heated in the presence of dithiothreitol (Table I). However, heating in the presence of iodosobenzoate further activates CF,. Although heat activation alone does not reduce the disulfide bond, it causes a permanent conformational change in the y subunit. The rate of further activation of the ATPase and, presumably, dithiothreitol reduction of the disulfide bond after heat activation is faster than is the rate of dithiothreitol activation of the latent enzyme (Fig. 6). This suggests that heat-induced conformational changes make the disulfide more accessible to the medium.
Conformational changes induced by activation have also been examined by fluorescence polarization of ANM bound to the sulfhydryl residues. Polarization and anisotropy of bound ANM (Table 11) indicate that the sulfhydryl residues exposed by dithiothreitol treatment are in a less rigid envi- Activation effects on the disulfide bond of CFI. Latent and activated CF, a t 1 mg of protein/ml was denatured in 1% SDS, and free sulfhydryl residues were modified with N-ethylmaleimide. Dithiothreitol (50 mM) was added to reduce the enzyme and remove unreacted N-ethylmaleimide. Following column centrifugation, the newly exposed sulfhydryls were labeled with 50 p~ ANM. Fifteen pg of each sample were subjected to electrophoresis on a 12% acrylamide gel. Lane 1, latent CFI; Lane 2, heat-activated CF,; Lane 3, dithiothreitol-activated CF,; Lane 4, CF, treated with iodosobenzoate after dithiothreitol activation, but prior to SDS addition.

TABLE I
Iodosobenzoate effect on ATP hydrolysis activity of CFI Soluble CFl a t 0.8-1.0 mg of protein/ml was activated by dithiothreitol, heat, or a combined method as described under "Materials and Methods." After removing the dithiothreitol and ATP, 0.5 mM iodosobenzoate (IBZ) was added to portions of each preparation, and the samples were incubated for 10 min. For the last entry in the table, 0.5 mM iodosobenzoate was present during heat treatment of the enzyme as well. Aliquots were removed and assayed f x ATP hydrolysis activity in buffer containing 5 mM CaCI, and 5 mM ATP.
ATPase activity

TABLE I1
Fluorescence polnrization of ANM-lnbeled CF, Fluorescence of ANM-labeled CFI was measured as described under "Materials and Methods." Dark-accessible and light-accessible refer to CFI modified while bound to thylakoids, followed by isolation of CF1 (12).  ronment than those accessible in thylakoids modified in the light or the dark (12,24). Heat activation, which has no effect on the dark-or light-accessible sites, diminishes the polarization of ANM bound to the dithiothreitol-exposed sites, indicating that they may be nearer the surface of the protein.

DISCUSSION
In this paper, we present a comparison of two different methods of activating CF, and the relationship between activation and the oxidation-reduction states of sulfhydryl residues in the enzyme. There are two striking observations made in these experiments. The first involves the existence of a disulfide bond in the y subunit that is distinct from the two sulfhydryl residues that are accessible in the membranebound enzyme in the light or dark. The second observation is related to the apparent differences between methods of activating the latent ATPase and their effects on the disulfide bond.
Previous determinations of the stoichiometry of sulfhydryl residues in the y subunit were based on the incorporation of radioactive labeled sulfhydryl reagents or by quantitation of cysteic acid after oxidative treatment (7,11,13,25). Stoichiometries of 2-4 mol of cysteine/mol of y have been reported, based on the molecular weight of CF1 being 325,000 (27). This molecular weight has recently been shown to be an underestimate, with 400,000 being a more accurate value (8). The results in this paper and the accompanying reports (28, 29) provide strong evidence that the y subunit of soluble CFl contains four sulfhydryl residues, two of which are linked in a disulfide bond. Previous discrepancies in stoichiometry are likely due to incomplete modification, and to underestimation of the molecular weight of the enzyme (28). In addition, steric constraints, nearest neighbor effects, and/or detergent effects may have caused abnormal reactivity of these residues.
Dithiothreitol-induced activation of the latent. ATPase has been correlated with reduction of the disulfide bond in the y subunit. Fig. 7 is a simple model that depicts a possible relationship between the ATPase activity and oxidation-reduction of sulfhydryl residues in y. The location of the sulfhydryl groups in the model is based on reactivity with dithiothreitol and iodosobenzoate, and on fluorescent properties of ANM bound to these sites. Sulfhydryl residues shown close to the center of the circle are considered to be less exposed to the medium, and hence less reactive. Dithiothreitol-induced activation of CF, occurs by a reversible mechanism. It is sensitive to oxidizing agents that convert reduced vicinal sulfhydryls to a disulfide. Dithiothreitol-induced activation is distinct from heat activation, which does not reduce the disulfide and is insensitive to oxidizing agents. In addition, heat activation is irreversible, causing a permanent change in the exposure of the disulfide bond to the medium. When used together or sequentially, dithiothreitol plus heat transform CF1 into a still more active form, which is only partially sensitive to oxidizing agents. We cannot explain the discrepancy between our results and those of Arana and Vallejos (13), who found that heat activation caused an apparent disulfide-dithiol exchange between the cy and y subunits. No exchange was detected after heating, as determined from the thiol content of the subunits using [3H]N-ethylmaleimide, and using this reagent only the y subunit was found to contain a disulfide bond.' Recently, the amino acid sequence of the (Y subunit of CF, from tobacco chloroplasts was deduced from the gene sequence (30). Only a single cysteine residue per copy of the cy subunit was observed, in agreement with one published report for the Escherichia coli cy subunit (31), but in contrast to previous measurements for CF, (17,32) and E. coli (33). The presence of only one cysteinyl residue per cy subunit and the lack of evidence for intersubunit disulfides in CF1 would make a disulfide exchange mechanism appear unlikely.
The differences between the two methods of activation compared in this paper are important to consider in studying the catalytic mechanism and regulation of CF1. Although heating in the presence of dithiothreitol yields the most active preparation of the soluble ATPase, this treatment causes permanent changes in the structure of the enzyme. Therefore, attempts to study regulatory properties of CF1 after heat treatment may produce results that are unrelated to properties of the membrane-bound enzyme. Indeed, it has previously been demonstrated that heat-activated enzyme will not reconstitute photophosphorylation in thylakoid membranes stripped of endogenous CF1 (34) whereas the dithiothreitolactivated enzyme is more effective than oxidized CFI. Since the reduced enzyme is the physiologically significant form (14,15,29), dithiothreitol treatment clearly should be the method of choice for activating CF1 in experiments designed to study mechanistic features.
Finally, the importance of conformational changes in the y subunit during activation bears comment. The direct correlation between reduction of the disulfide bond and ATPase activity suggests that reduction causes conformational changes in y which are transmitted to the p subunit and the catalytic site. Considering the possible role of the y subunit as the proton gate in the mechanism of ATP synthesis, an understanding of these conformational changes in the soluble enzyme may be crucial to our understanding of the membranebound enzyme and to the coupling mechanism. The nature of the signal between the y subunit and the catalytic sites on subunits remains to be identified.