Mitochondrial ATP Synthase INTERACTION OF A SYNTHETIC 50-AMINO ACID, @-SUBUNIT PEPTIDE WITH ATP*

A 50-amino acid peptide predicted by chemical mod- ification studies of F1 and by comparison with adenylate kinase to comprise part of an ATP-binding domain within the &subunit of mitochondrial ATP synthase has been synthesized and purified. In the numbering system used for bovine heart B, the peptide consists of amino acid residues from aspartate 141 at the N-ter-minal end to threonine 190 at the carboxyl end. In Tris-C1 buffer, pH 7.4, the peptide undergoes a dramatic reaction with ATP resulting in precipitate formation. Analysis of the precipitate shows it to contain both peptide and ATP. Similar to the ATPase activity of F1 and the binding of nucleotide to the enzyme, the capacity of ATP to induce precipitation of the peptide is decreased markedly by lowering pH. Interaction of the peptide with the fluorescent ATP analog, TNP-ATP (2’(3’)-0-(2,4-6-trinitrophenyl)- adenosine 5’-triphosphate), can be demonstrated in solution at low concentrations. A 7-fold enhancement in fluorescence is observed when 2.5 PM TNP-ATP interacts with 2.5 )IM peptide. Divalent cation is nei-ther required for ATP-induced precipitation of the peptide nor for demonstrating interaction between TNP-ATPand peptide, just as Mg2+ is not required for nucleotide binding to F1. These results indicate that the &subunit peptide studied here comprises at least part of a nucleotide- binding domain within

system used for bovine heart B, the peptide consists of amino acid residues from aspartate 141 at the N-terminal end to threonine 190 at the carboxyl end.
In Tris-C1 buffer, pH 7.4, the peptide undergoes a dramatic reaction with ATP resulting in precipitate formation. Analysis of the precipitate shows it to contain both peptide and ATP. Similar to the ATPase activity of F1 and the binding of nucleotide to the enzyme, the capacity of ATP to induce precipitation of the peptide is decreased markedly by lowering pH. Interaction of the peptide with the fluorescent ATP analog, TNP-ATP (2'(3')-0-(2,4-6-trinitrophenyl)adenosine 5'-triphosphate), can be demonstrated in solution at low concentrations. A 7-fold enhancement in fluorescence is observed when 2.5 PM TNP-ATP interacts with 2.5 )IM peptide. Divalent cation is neither required for ATP-induced precipitation of the peptide nor for demonstrating interaction between TNP-ATPand peptide, just as Mg2+ is not required for nucleotide binding to F1.
These results indicate that the &subunit peptide studied here comprises at least part of a nucleotidebinding domain within the mitochondrial ATP synthase complex.
The mitochondrial ATP synthase consists of two major components, one called Fo which spans the inner membrane and the other called F, which projects into the matrix space. Fo is thought to direct protons derived from the respirationdriven electrochemical proton gradient to F1, a water-soluble complex consisting of five different subunits in the stoichiometric ratio a3@,Gyt (for recent reviews, see Refs. [1][2][3][4][5]. F1, upon binding ADP and Pi synthesizes ATP on its surface with an equilibrium constant near 1, and is thought to utilize the proton gradient to effect release of bound ATP (6)(7)(8)(9)(10).
These events are generally believed to take place on the psubunit as this subunit binds ATP and hydrolyzes this nucleoside triphosphate at a low rate (11)(12)(13).
Recent work from several laboratories (14)(15)(16)(17)(18) has emphasized structural similarities between the @-subunit of F, and certain nucleotide-binding proteins, in particular adenylate * This work was supported by National Institutes of Health Grant CA 10951 (to P. L. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked ''aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. " kinase, the ras p21 transforming protein, and elongation factor Tu. All of these proteins show some sequence homology in a region near a glycine-rich flexible loop. X-ray crystallographic studies of elongation factor T u indicate that at least part of the nucleotide-binding domain may reside near the glycine-rich flexible loop (17,18). Similar conclusions h a w been reached from NMR studies of MgATP binding to a 50amino acid peptide of adenylate kinase (15,16).
In extending these analogies to the @-subunit of F,, Duncan et al. (19) proposed a three-dimensional model for a nucleotide-binding domain, presumably the catalytic site, within the @-subunit of ATP synthase. Support for this model came from both chemical modification and genetic studies. Significantly, in this model much of the ATP-binding domain is contained within the 50-amino acid stretch depicted in Fig. M. In analogy to the three nucleotide-binding proteins described above, the tripolyphosphate portion of ATP is placed near a glycine-rich flexible loop, in this case with the a-phosphate within binding distance of lysine 162. The adenosine portion of the molecule is embedded within the same 50-amino acid stretch between an a-helical stretch and a 0-pleated sheet.
In order to test directly the above model, we have synthesized the 50-amino acid, F1 @-subunit peptide described above and examined its interaction directly with ATP, TNP-ATP,' and a variety of other ligands.

Materiak;
All the Boc-protected amino acids, Boc-Thr(OBz1)-OCH,-PAMresin, and the reagents and solvents for the peptide assembly were purchased from Applied Biosystems, Inc. Trifluoroacetic acid, pcresol, p-thiocresol, and dimethyl sulfide were supplied by Aldrich. Hydrogen fluoride was from Matheson Gas Products, Inc., and acetonitrile and methylene chloride (both HPLC grade) were obtained from American Burdick & Jackson, Inc. Amino acid standards and phenylisothiocyanate were obtained from Pierce Chemical Co. ATP and ADP were obtained from Pharmacia LKB Biotechnology Inc. ITP, GTP, AMP-PNP, MgC12, Tris-C1 and Tris acetate were purchased from Sigma, whereas Pi was obtained from J. T. Baker Chemical Co. TNP-ATP was purchased from Molecular Probes and its purity confirmed by chromatography on polyethyleneimine-cellulose plates (Cel-300 PEI, Brinkmann Instruments) in a solvent system containing 2 M formic acid and 0.5 M LiCl. SDS, acrylamide, and bisacrylamide were purchased from Bio-Rad. Radioactive nucleotides were obtained from ICN Radiochemicals and Budget-Solve from Research Products International.  (19,29). The region indicated in bhck has been proposed to comprise at least part of an ATP-binding domain (see text for discussion). This region comprises residues in the &subunit from aspartic acid 141 to threonine 190 and includes lysine 162 and glutamate 188, known to be essential for F,-ATPase activity (30)(31)(32). It is this 50-amino acid residue peptide (PP-50) that was synthesized and used in these studies. B, amino acid sequence of PP-50. Note that in addition to lysine 162 and histidine 177 (closed boxes), which are conserved in F,-@ and adenylate kinase, there are four additional positively charged amino acid residues (dotted boxes).

Methods
Synthesis of the 50-Amino Acid Residue Peptide (PP-50)-The peptide was synthesized on an Applied Biosystems Model 430A peptide synthesizer by the solid phase method developed by Merrifield (20). Peptide assembly was carried out starting with Boc-Thr(OBz1)-OCH,-PAM-resin and the Boc-protected amino acids using a symmetric anhydride activation procedure. Side chain protecting groups were as follows: benzyl ether for the hydroxyl group of threonine and serine; benzyl ester for the carboxyl group of aspartic and glutamic acids; p-tosyl for the guanidine of arginine and imidazole of histidine; 2-chlorobenzyloxycarbonyl for the c-amine of lysine; 2-bromobenzyloxycarbonyl for the phenolic hydroxyl group of tyrosine. Asparagine, glutamine, and arginine were coupled as their l-hydroxybenzotriazole esters.
The synthetic protected peptide-resin was deprotected and cleaved by liquid hydrogen fluoride containing the following scavengers in the ratio, HF:p-creso1:dimethyl sulfide:p-thiocresol, 101:1:0.2 by volume. After stirring at 0 "C for 1 h, the HF was removed in uacuo, the peptide was precipitated with ether and filtered. It was extracted with 15% aqueous acetic acid and freeze dried. The crude peptide was subjected to purification by HPLC on a semipreparative CIS column (10 pm, 30 X 0.78 cm). Buffer A was water, 0.03% trifluoroacetic acid and buffer B was acetonitrile, 0.03% trifluoroacetic acid. The column was equilibrated with buffer A at a flow rate of 2 ml/min and about 200 mg of crude peptide was loaded onto the column. The column was eluted with a gradient of 0-60% B over 60 min to fractionate the peptide. The fractions were analyzed and those containing only the pure peptide were pooled and freeze-dried. The yield of pure peptide was 12.5 mg.
Amino Acid Analysis-This was performed using the PICO-TAG amino acid analysis system of Waters-Millipore. The peptide (about 120 pg) was hydrolyzed with 6 N HC1 containing 1% phenol by volume at 110 "C for 24 h. The hydrolysate was dried and the amino acids derivatized with phenylisothiocyanate to yield the corresponding phenylthiocarbamyl derivatives following standard procedures (21). These derivatives were analyzed on the PICO-TAG amino acid analysis system which had been previously calibrated with a standard mixture of amino acids.
Quantification of PP-50-PP-50 was quantified by the method of Lowry et al. (22) using bovine serum albumin as standard. This procedure gave values for PP-50 standards within 5% of those given by quantitative amino acid analysis. For stoichiometry calculations a molecular weight for PP-50 of 5815 was used. This was based on the known amino acid composition of PP-50 and corrected for water split out during peptide bond formation and trifluoroacetic acid salt formation with seven amino functional groups.
Interaction of PP-50 with TNP-ATP-This was monitored fluorometrically in a 4-ml quartz cuvette (Starna Suprasil) containing in a total volume of 2 ml, 10 mM Tris-C1, pH 7.4, and concentrations of TNP-ATP and PP-50 indicated in the legend to Fig. 5. A Gilford Fluoro-4 fluorometer set at an excitation wavelength of 410 nm and emission wavelengths indicated in the legend to Fig. 5 was used for these studies.
Gel Electrophoresis of PP-50 in SDS-SDS-PAGE was carried out by a modification of the Weber and Osborn procedure (23) in 10% polyacrylamide gels exactly as described previously (24).
Methods Relevant to F,-ATPase-F, was purified to homogeneity from rat liver mitochondria by the procedure of Catterall and Pedersen (25). Binding of a-labeled [32P]AMP-PNP to this enzyme was assessed by the column centrifugation procedure of Garrett and Penefsky (26) exactly as described previously (27). Initial ATPase rates were monitored by quantifying the release of inorganic phosphate by the procedure of Gomori (28).
Measurement of Radioactivity-Radioactivity was measured in a Beckman LS lOOC series spectrometer. PP-50 precipitated in the presence of a-labeled [32P]ATP was dissolved in 150 p1 of NaOH; 50 pl was added to 10 ml of Budget-Solve and assessed for radioactivity and 100 p1 was used to determine protein as described above. When binding of a-labeled [32P]AMP-PNP to F1 was carried out, 100 pl of the eluate from the centrifugation column was added to 10 ml of Budget-Solve and assessed for radioactivity.

RESULTS AND DISCUSSION
Purity and Characterization of the F , ("bunit Peptide (PP-50) Used in These Studies-Procedures employed in the synthesis and purification of PP-50 are described in detail under "Methods." Two batches of PP-50 have been synthesized de novo and exhibit essentially identical purity and behavior patterns. As shown in Fig. 2, three methods were used to examine the purity of PP-50 prior to investigating its capacity to interact with ATP. Fig. 2.4 shows that, upon HPLC chromatography on a pBondapak Clx column, PP-50 elutes as a single peak, nearly gaussian in shape. The inset in Fig. 24 shows that PP-50 migrates as a single band upon SDS-PAGE in a 10% polyacrylamide gel system. As expected from its predicted molecular weight of 5815, PP-50 migrates noticeably faster than cytochrome c ( M , = 11,700). Significantly, there is no suggestion from either the HPLC chromatogram or from SDS-PAGE of larger molecular weight species that might have resulted during chemical synthesis. As shown in Fig. 223 the amino acid composition of PP-50 compares favorably (within experimental error) with the predicted amino acid composition of the peptide. Finally, in data not presented here we have confirmed the sequence from aspartic acid 141 at the N-terminal end of PP-50 to alanine 176.
Interaction of PP-50 with ATP-Initial experiments showed that when ATP is added to PP-50 in 50 mM Tris-C1 buffer, pH 7.4, a rather dramatic precipitation of PP-50 occurs which can be observed visually. Precipitation is visually ob-@-Subunit Peptide of Mitochondrial A T P Synthase

A, HPLC chromatogram of PP-50.
Chromatography was carried out on a 0.78 X 30-cm pBondapak CI8 column. PP-50 (10 pg) was injected in a volume of 10 pl. Elution was effected with a gradient from 30 to 60% buffer B which consisted of acetonitrile + 0.03% trifluoroacetic acid. PP-50 in the eluate was detected at 220 nm using an ultraviolet detector. Inset, SDS-PAGE of PP-50 relative to that of cytochrome c (M, = 11,700). In each case, 2.5 pg of protein were applied to the gel (see "Methods" for details). B, amino acid composition of PP-50. The predicted amino acid composition of PP-50 is compared with that obtained experimentally by carrying out amino acid analysis by the PICO-TAG procedure (see "Methods").   Mg" is not required to effect binding of nucleotides to F1-ATPase (26). Table I provides a quantitative picture of the specificity of the ATP-induced precipitation of PP-50. Various ligands were added to PP-50 contained in 10 mM Tris-C1 buffer, pH 7.4, which promotes greater solubility of PP-50 than the more concentrated 50 mM buffer. As shown in the table, 1 mM ATP induces 67% of the original PP-50 in solution to precipitate. GTP and ITP induce similar precipitation responses. Significantly, the nonhydrolyzable ATP analog AMP-PNP also induces PP-50 to precipitate indicating that binding of ATP rather than ATP hydrolysis per se is the primary factor responsible for the induction of PP-50 precipitation. Again, AMP-PNP 17.0 63 the specificity of the precipitation phenomenon should be noted. Thus, AMP, Pi, and Mg2' have little capacity to induce precipitation of PP-50. Under these conditions, the ATP/PP-50 stoichiometric ratio, determined radioactively as described under "Methods," was found in five different experiments to be 0.91 f 0.08. Fig. 3 demonstrates that ATP-induced precipitation of PP-50 is sharply dependent on pH. Lowering the pH to 4 markedly suppresses the capacity of ATP to induce precipitation of PP-50 (Fig. 3A). Significantly, the capacity of F,-ATPase to bind AMP-PNP (Fig. 3B) and to hydrolyze ATP (Fig. 3C) is also markedly diminished as pH is decreased. It will be noted also in Fig. 3, A and B, that added M e is not required to facilitate either ATP-induced precipitation of PP-50 or binding of nucleotide to F,. Fig. 4 shows that the reaction described above can be readily demonstrated in solution provided ATP is replaced with the fluorescent ATP analog TNP-ATP. Significantly, at 2.5 p~ PP-50 and 2.5 p~ TNP-ATP, a 7-fold enhancement in fluorescence of the latter species is observed (Fig. 4A). This finding is of particular interest as Grubmeyer and Penefsky (34) reported previously that TNP-ATP is a strong competitive inhibitor of F1, undergoing a 7-fold enhancement in fluorescence when added in equimolar amounts to F,. Fig. 4B demonstrates the dependence of the fluorescence enhancement response on TNP-ATP concentration. A half-maximal response is observed a t less than 0.5 p~ TNP-ATP demonstrating a very high apparent affinity of PP-50 for this nucleotide analog. It is shown also in Fig. 4B that excess ATP added prior to TNP-ATP to PP-50 markedly reduces the fluorescence enhancement response, consistent with interaction of both nucleotides with the same binding domain on Taken together, results presented here are consistent with the view that PP-50 contains at least one part of a nucleotidebinding domain. The finding that precipitation of PP-50 is induced only by ligands with a pyrophosphate or tripolyphosphate moiety indicates that, contained within this peptide, is a region which interacts strongly with the phosphate ester region of nucleoside triphosphates. The two most likely amino PP-50. was carried out at room temperature for 1 h in 0.1 ml of the same buffers (see "Methods"). In C, ATPase activity of purified F, (17 pg) was monitored in the same buffers in a 1-ml system containing 10 mM ATP and 10 mM MgC12. After 1 min the reaction was terminated with 0.1 ml of 2.5 M perchloric acid. Pi was then quantified by the method of Gomori (28). acid candidates involved in this interaction are lysine 162 and histidine 177 as these residues are conserved in F1-@ and adenylate kinase, the enzyme on which the Duncan et al. (19) (29) model is based. Nevertheless, it should be noted that 4 additional positively charged residues are present within PP-50 (Fig. 1B). It should be noted also that regions of the F1-@ subunit homologous with adenylate kinase may not comprise a complete catalytic domain. This is emphasized by the recent results of Cross et al. (35) who have shown that 2-azido-ATP labels tyrosine 345, suggesting that this residue may also lie near or within the catalytic domain. Although tyrosine 345 lies outside the adenylate kinase homology region, Cross et al. (35) suggest that it may be juxtaposed to this region in the three-dimensional structure of F,.
The finding that interaction of the nucleoside triphosphates tested promote precipitation of PP-50 suggests that a conformational change is induced by binding substrate. Nonpolar residues normally contained within the interior of the peptide may then become exposed t o the aqueous medium. Along these lines it is of interest to note that ATP has been shown recently to induce a similar precipitation response when added to a 19-amino acid "active site" peptide of actin (36). It is tempting to suggest, therefore, that certain ATP-binding proteins may contain one or more regions, peripheral to nucleotide-binding domains, which are critical for maintaining such domains in solution throughout the catalytic/functional cycle.
Future experiments will focus on the effect of site-directed changes on the interaction of PP-50 with ATP and TNP-ATP as well as the capacity of other synthetic @-subunit peptides to facilitate ATP hydrolysis when combined with PP-50.