Indi an Journal or Biochemistry & Biophysics Vol. 37, December 2000, pp. 459-469

Role of metal ion ·. in coupling between chemical catalysis and conformational -.L... -

changes in A TP synthase •

, - / . - - ,-

~1~~Hochman 1 Xiao-Min.. bong.\Y~e Lifshitz and Chanoch Carm.e li *#

Department or Biochemi stry, Tel Aviv Uni versity, Tel Aviv 69978, Israel.....l ,,--.

Received 19 March 2000; accepted 30 May 2000

Iso lated chl oroplast ATP synthase (CFoF 1) was used for determination of the structure- fun cti on relation by measuring th~effect of divalent metal ions on the properties or ATPase. Mg2+ ions were more efticient catalys ts than Ca2+ ions as indi ­cated by KCIII / K", or 55.2 and 5.4, respecti vely. Other activity parameters related to binding, such as the K", or MATP and K; or MADP, indicated a stronger binding in the presence of Mg2+ as seen from a Mg2+/Ca2+ rati o of 2.8 and 3.8, respectively. Strong binding of Ca2+ ions with a K" of 0.03 ± 00.6 ~M· I was detected only in the presence of ADP probably because of the positi ve interactive effect of CaADP as indicated in the inhibition properties. Mg2+ ions were more erticient catalysts also in other forms of the enzyme such as in the thylakoid membrane, in isolated CFoFI and in CF 1• The Mg2+/Ca2+ rat io of Ken'! K", was 5.3, 10.2 and 1.5 for the thyl akoid membrane enzyme, the isolated CFoFI and the so luble CF 1 respectively. This indi ­cated that Ca2+ ions became less efficient catalysts in the more intact and integrated enzyme whi le Mg2+ ions were as effi­cient in ail forms of the enzyme. Un like Mg2+, Ca2+ ions also did not support proton-coupled ATP synthesis and ATP driven proton pumping. It is suggested that the differences in the ligand structure of these two ions mi ght be the reason for the dif­ferentia l function . An average 0.3A shorter bond length of octahedral first coordi nati on in Ca2+ ions caused a weaker bind­ing of CaATP than that of MgATP. The effect of differenti al binding is discussed in re lati on to the binding of the transiti on state intermediate and to the rate of product release.) .- , \,.I s-1

I&#"' ~ (" J Introduction

The A TP synthases constitute a group of homolo­gous enzymes responsible for ATP synthesis in mito­

chondri a, chloroplasts and bacteria. These proton translocating revers ibl e ATPases consist of two majo r regions: hydrophob ic protei ns (Fo) situated in the membrane, serving as a proton channel, to which coupling factor I (F I ) is attached I . The so luble cata­lytic part of the H+-ATPase (FI) is composed of five subunits2

: 3cx, 3 ~ , y, 8 and £. The structure of the F I -

ATPase from bovine heart mitochondria, at a 2.8 A reso luti on, was determined by crystallography' . The structure supports a mechani sm in which the three catalytic s ites, at the ~ subunits, are alternately in dif­feren t states of the cata lytic cyc le4

• The mechani sm of the ATP synthase is probably bes t described by the "binding change" hypothes is4. ATP is spontaneously

#This work was partially supported by grant # 199943 1 from BSF *Author for correspo ndence; Email :ccarmcli@post.tau. ac.il Abbreviations: CFoFJ, ATP synthase of chl oroplast; CFJ, coupling factor 1 of ATP synthase; Tri ci ne, N- tris(hydroxymcth yl)-methyl glycine; lf EPES, N-2- Hydroxycthylpiperazine-N-2-ethanesulfoni c acid; DTT, 1,4-Di thi othrei tol: DCCD, di cyclocar­bodi imide; MATP, divalent metal ion-ATP complex : AMPPNP, adcnylylimidodiphosphnte.

formed from tightl y bound ADP and Pi , alternately, at the three catalytic sites on Fl . The re lease of the

product is induced by a conformationa l change caused by the e lectrochemical potenti al of protons across the membrane during synthes is and by nega­ti ve cooperati ve interaction on binding of ATP during

hydrolysis . Support for thi s hypothesis came from the finding that ATP hydro lys is with equilibrium con­stant of about one) and the concomitant catalysis of

H2180-ATP exchange occur when the nucleotide is

tightly bound to a sing le site on the enzyme. The se­quence of reactions is coordinated through coopera­ti ve interaction among the three active sites of the

enzyme. Interaction among the subunits was exempli ­fied by the ability to quenc h the activated state of the CF I 6 and to modulate ATPase activity in the iso lated CFI

7 by positive cooperati ve binding of ADP. H 180-A TP exchange was also modul ated by negative coop­erati ve binding of ATPx. The y subunit is the best

candidate for transmi ss ion of the e lectrochemical po­tenti al through conformati onal changes. There is a

grow ing body of direct and indirect ex perimenta l evi­dence demonstrating that Fo mi ght function as hydro­gen ion turbine transmitting the conformational c hanges through the rotation of the y subunit9

-11

.

460 INDIAN J. BIOCI-IEM. BIOPI-IYS., VOL. 37, DECEMBER 2000

Divalent metal ions are also essential for catalysis and the study of the role of metal s in catalysis is, therefore, crucial for the understanding of the mode of action of the enzyme. In recent years, metals have been used by us as probes for the active site of CF,' 2. 1.1. Correlation between the kinetic constants of ac tivation and inhibition of ATPase by dival ent metal ions and the binding constants enabled the identifi ca­tion of the metal binding sites that are involved in catalysis'4. EPR measurements indicated that CF, contains three loose, non interacting and three tight Mn2+ binding sites which interact cooperatively'5. Cooperative binding of Mn2+ was also demonstrated by the use of the NMR relaxation technique '6. It was suggested that this cooperativity is a result of interac­tion among the three active sites of the enzyme. Ki­netic analysis of the binding and release of divalent metal ions from the interacting sites indicated that the on and off rate constants '7 are similar to the rate con­stant of the acceleration of pre-steady state cata lys is in the presence of these metal s' 8. It was, therefore, suggested that the conformational changes which were invol ved in cooperative metal binding, al so controlled the modulation of the activity of the en­zyme.

Direct measurement of the environment of manga­nese by X-ray absorption fine structure analysis (EXAFS) at the three interacting sites indicated direct binding of ATP or ADP to the metal in CF, . The Mn2+ was bound to six oxygens with average bond di stance of 2. 14 ;". . The average distance between the Mn2+ and the nucleotide phosphorus was 3.6 ;"" 9.20. This finding was recently confirmed by EPR measurements which determined the distance be­tween the nucleotide phosphate oxygen and V02+ as an analog of the divalent metal ion2'.2' . It was sug­gested ''! that the strong electrophylic effect induced by the metal renders the phosphorus more susceptible to nucleophylic attack by water. It was also suggested that an exchangeable water molecul e detected in the CF, MnATP complex '6. '9 might serve as the nucleo­phyl. The metal of the catalytic site, as seen from the crystals of F" GTPases and myosin ATPase

24, has a

stri ct structural arrangement. The Mg2+ is bound to a serine, a threonine or to the two amino acid residues and to the oxygen atoms of the triphosphate chain of the nucleotide. The oxygen atoms of the y-phosphate are al so bound to arginine or to lysine nitrogens which stabilize the pentacoordinated phosphorus. A water molecule, in line with the y, ~-phosphate oxy-

gen bridge, undergoes genera l base cata lysis by the carboxyl of a glutamic acid or by a glutamine res idue. This strict arrangement enables effici ent hydrolysis of the anhydride bond of the nucleotide. In case of ATP synthase, it also facilitates formation of the anhydride bond during synthesis of ATP. This arrangement might be perturbed when Mg2+ is substituted by Ca2+ ions. Such perturbation by Ca2+ is suspected because thi s ion supports hydrolysis in isolated CF, but did not support synthes is25 and was a competitive inhibi­tor to Mg2+ supported A TP synthes is26 in the mem­brane bound CFoF, . Ca2+ was a poor activator co m­pared to Mg2+ in the membrane bound27 and in the phospholipid reconstituted CFoF, 2 .2'! . Yet Ca2+ ions support high rates of ATPase activity in iso lated CF I. These data also indicate that Ca2+ binds to the cata­lytic site of the enzyme. However C}+ ions do not support ATP synthesis, Pi-ATP exchange reacti on and ATP driven proton pumping in FoF, 26'31 . The dif­ferences in binding properti es and in ligand arrange­ment of these two ions mi ght be the reason for the differential functi on. Mg2+ and Mn 2+ ions maintain an octahedra l arrangement in the first coordinati on sphere with very similar bond length to the six li g­ands. Ca2+ ions in solution, in inorganic and organic complexes and in protein have between 6 to 9 coordi­nation number with longer bond lengthJ2-J4.

Materials and Methods Enzyme preparatioll

Chloroplasts were prepared from lettuce (v r. ro­maine) leaves6. CF, was iso lated from the chl oro­piastsJ5 and stored as an (NH4hS04 suspension. Stored CF,' 7 was passed through a centrifuged G-50 Sephadex column equilibrated with 40 111M Tricine­NaOH, pH 8 and I mM ATP. The enzy me was heat activated,6 by incubation in a medium containing 40 mM ATP, 5mM DTT, 40 mM Tricine-NaOH, pH 8 and I mM EDT A for 4 min at 60°C. Fo ll owing incu­bation , the enzyme was passed twi ce through centri­fuged G-50 sephadex columns equilibrated with 40 mM Tricine-NaOH, pH 8 and 0.1 mM EDT A. For binding the desalted latent enzyme was activated in the presence of 50 mM DTT by incubati on for 5 hrs at 25°C. The enzyme was passed twice through centri ­fuged G-50 sephadex columns equilibrated with 40 mM HEPES NaOH, pH 8 and 0.1 mM EDT A and a third time without EDT A. Protein was concentrated by ultrafi ltration up to I mM. Prote in concentration was determined by UV absorption,6, assuming Mr of

HOCHMAN e l a l. : ROLE OF METAL IONS IN COUPLING IN ATP SYNTHASE 461

400,000 or by a colorimetric method "7 . CFoF I was prepared from lettuce chlorop lasts ac­

cording to our detergent solubilization method28

which was later modified29. Essentially, thylakoids were solubilized in cholate and octylglucoside. The enzyme was fractionated by ammonium sulfate pre­cipitation and purified on a sucrose gradient in the presence of 0.2% Triton-X I 00. Protein concentration was determined "7 assuming a Mr of 535,000.

ATPase activity Ca2+-dependent activity was assayed in a medium

containing: 40 mM Tricine-NaOH, pH 8, 3 mM ATP or as indicated, 1.5 mM CaCh or as indicated and 12 I1g CFI or CFoFI in a total volume of 0.5 ml for 1- 10 min at 37°C. The reaction was started by the addition of the enzyme and was terminated by the addition of I ml of reagent for spectrophotometric determination of Pi"s. Rates of activity were determined from linear regression of the time course.

Mg2+-dependent activity was assayed in a medium containing: 40 mM Tricine-NaOH, pH 8, 10 mM ATP, I mM MgCI2. or as indicated ( 100 mM Na2S0.1 only with CFI) and 1211g CFI, or CFoFI in a total vol­ume of I ml for 5 min at 37°C.

Light-triggered A TPase activity The activity was assayed 6 in a medium containing:

40 mM Tricine-NaOH, pH 8, 50 mM KCI , I mM H2P04' , pH 8, 10 mM MgCI2, or CaCI2 as indicated, 50 11M phenazine methosu lfate, 5 mM OTT, I mM NH4C1 and chloroplasts (containing 50 I1g Chl/ml) . The ch loroplasts were il luminated for 5 min at 1.25 x 106 ergs / cm2 / sec in a total volume of I ml at 25°C and the indicated C2py]ATP concentration (contain­ing 106 cpm) was added on turning the light off. The reaction proceeded for I min in dark and was tenni­nated by the addition of trichloroacetic acid (final concentration 5%). Pi was separated from ATP for radioactivity counting").

Binding afcalcillm ions to CF, and to CFoF, Binding measurements of Ca2+ to CF, and to CFoF,

were done with the aid of ion spec ific chelator dyes. Arsenazo III having a K" of 5 11M for Ca2+ at pH 8 was used for determination of ti ght ion binding si tes40

. The titration mixture contained 20 11M CF, or CFoF, . 40 11 M ADP, 50 mM HEPES-NaOH, pH 8 and 40 11M Arsenazo III. It was titrated with CaCl2 and the

absorption changes of Arsenazo III at 648 nm were used for determination of the concentration of the free C 2+ a .

Data analysis The kinetic parameters of ATPase activity were

determined using the Dixon ', Hi1l2, and Michaelis­Menten " equations . Binding data were calculated and plotted according to the Scatchard4 equation. The data fit and plotting were done by the "Ka leidaGraph 2. 1", a graphic software program by Synergy Software.

I I [I] (Eq. I) -=--+ , ..

v V max V max K ;

v [S]" = (Eq. 2)

Villa. Kill + [S]" , ..

v [S] (Eq.3)

Villa. K ill + [S] , ..

where v and Vn~lx are specific activities, [S] and [I] are substrate and inhibitor concentrations, Kill is the Mi­chaelis constant, K; is the inhibitor dissociation con­stant and n is the Hill coefficient.

(Eq.4)

where r is moles of ligand bound per mo le of enzyme, (C) is the concentration of the free ligand and nl and n2 are the number of sites having association constant KlI , and K1I2 , respectively.

Results Inhibition by CaADP alld by ji'ee Ca2

+ ions A TP synthesis is not supported by ci+ ions25 yet

this ion binds to the enzyme since it is a competitive inhibitor to Mg2+ in ATP synthes is26 in the membrane bound CFoF,. C}+ ions are also poor acti vators in

. M ?+. h b b d?7 d' comparison to g- 111 t e mem rane oun - an In phospholipid reconstituted CFoF, 28.2') . Yet Ca2+ ions support high rates of ATPase activity in iso lated CFI. These data indicate that Ca2+ binds to the catalytic site of the enzyme. However, the mode of action and binding might differ in various forms of the enzyme. Since direct binding of Ca2+ ions to the membrane­bound enzyme might be obscured by binding to other proteins, it was necessary to use isolated CFoF, for binding measurements. For correlation of binding to

462 INDIAN 1. BIOCHEM . BIOPHYS ., VOL. 37, DECEMBER 2000

function, it was important to determine the kinetic parameters. Only some parameters were determined for the activity of reconstituted CFoFI in the presence of Mg2+ ions41 but no data are available in the pres­ence of Ca2+ ions.

ATPase activity in the presence of CaATP was in­hibited by CaADP in an allosteric manner. This was indicated by the sigmoidal response curve that was obtained by plotting the rate of activity as function of the concentration of CaATP (Fig. I ). A calcu lated Hill coefficient of 2.3 may indicate that positive in­teraction between more than two binding sites in the enzyme enhanced the inhibition . A similar effect was earlier observed for the inhibition by CaADP (Ki 165 ~) of ATP hydrolysis in CF1 42. Dixon analysis indi­cated a Ki of 35 ±2.8 ~M for the inhibition by CaADP in CFoFI. The activity was also inhibited by free Ca2+ ions (Fig. 2) . The possibility that the free cation in­hibits indirectly by forming a complex with ADP was evaluated by determination of the inhibition con­stants. Dixon analys is of the data indicated an essen­tially non competitive type inhibition with Ki of 11 .25 ±1 .2 mM. The results should be compared to a com­petitive type inhibition with a similar Ki in CF114 . However, a Kill of 0.7 ± 0.08 mM and Vmax of 3.8 ± 0.3 ~moles/mg/min for CaA TP in CFoFI were lower than in CF11 4. An indication of the effici ency of CaATP hy­drolysis that is related to the binding constant of the transition state intennediate is given by K({I/KlIlof 5.4.

2.5

2.0 r::

"§ -.. 0 0 co .§ 1.5 i:l: '" Q)

(3 1.0 6 :t

0.5

2 3 4 Ca-ATP [mM]

Fig. I-Allosteric inhibition by CaADP of ATPase acti vity in isolated CFIIF 1• [ATPase activity was measured as dcscribed under "Materi als and mcthods" in the presence of various concentrat ions of CaATP, 2 mM Ca2+ free, and CaADP as foll ows: (e), 0 mM, (0), 0. 16 mM and C.A.), 0.5 mM. Positi ve interaction is indicated by the Hill coeffi cient of 0, 1.8 and 2.3 in the presence of free Ca2+ concentrati ons of 0, 0.16 and 0.5 mM, respectively. The coefli cients were obtained from a tit of the data to the Hill eq ua­tion].

Inhibition by MgADP and by .Ji"ee Mi+ ions Unlike reconstituted CFoF128.29 the iso lated enzyme

used in these experiments was much more sensitive to inhibition by free Mg2+ ions (Fig. 3). Dixon analysis of the data gave a Ki of 5 ± 0.07 ~M for free Mg2+ ions in the presence of MgATP. The Ki for the free cations and the Kill of 0.25 ± 0.03 mM for MgA TP was much smaller than that for Ca2+ ions. The stronger binding of Mg2+ and Mg-nucleotides was clear from the values of 2.8, 3.7 and 2250 of

• 3 •

o o

Fig. 2-lnhibition by free Ca2+ ions of ATPase activ ity in iso lated CFIIF 1• [ATPase acti vity was measured as described under "Mate­rials and methods" in the presence of various co ncentrations of CaATP and either (e), 2 mM or (0), 12 mM, free Ca2+ ions. In the inset the data were plotted as l/v vs. I/CaATP, ind icating a non­co mpetiti ve type of inhibition with K; 11.25± 1.2.4 mM and K", 0.7±0.07 mM] .

• • 0.6 • c 0 'Ej 0

---OIl 6

---0.4

i:l: 0 '" Q)

(3

§. 0.2 {).

0~----~2-----L4----~6-----78 -----710

Mg-ATP [mM]

Fig. 3-lnhibition by free Mg2+ ions of ATPase ac ti vity in iso­lated CFIIF 1• [ATPasc activi ty was measurcd as described under "Material s and Methods" in the presence of various concen tra­ti ons of MgATP, and either (e), 0.8 or (0), 0.16 or (t. ), 0.35 mM, free Mg2+ ions. Analysis of the kinetic parameters ind icates a mi xed type inhibit ion by the free Mg2+ ions].

HOCHMA N el al. : ROLE OF METAL IONS IN COUPLI NG IN ATP SYNTHASE 463

Table I-Kinetic parameters for ATPase ac ti vity in isolated CFIlF I . [The parameters were calculated from dat a as described under "Mate­rials and Methods" and presented in the figures . M represents a divalent-metal ion]

Metal Kill MATP Kj MADP (mM)

Mo2+ '"

0.25±0.03 0.0093±0.0004

Ca2+ 0.70±0.OS 0.0350±0.OO28

Mg2+/Ca2+ rati o of Kill for MATP, K; for MADP and K; for free metal respectively. Yet the activity in the presence of MgATP with a Villa x of 13.8 ± 1.5 ~mo l es/ . mg/min was much higher than with CaATP. Similar to CaADP, MgA DP inhibition of the activity in the presence of MgA TP was enhanced because of a strong positive interacti on (data not shown). The re­sult of Di xon analys is of the inhi bition by MgADP gave a K; of 9.33±0.47 ~M (Tab le I). The str ikingly large Km/ Kill of 55.2 indicated a much more effi cient catalys is in the presence of Mg2+ compared to that of Ca2+ ions.

Ligh t triggered ATPase in thylakoids

The nove l findin g of the properties of ATPase ac­ti vi ty in the presence of CaATP in isolated and recon­stituted CFoF, mot ivated us to rein vesti gate the hy­drolyt ic activ ity in thylakoids in the presence of Ca2+ ions. Light dependent ATPase ac ti vity was inhibited by DCCD that binds to the c subunit of CFo. Some proton pumping was also measured in the presence of CaATp4:l . These findings indicated th at C}+ ions funct ion as an acti vator of the membrane bound CFoF,. Yet it is possible that no light triggered Ca­ATP hydrolysis was detected in these experiments because the substrate was added onl y two minutes after ac ti vati on following partial decay of the ac ti va­tion. Indeed , in the present experiments, CaATP was added i mmed iate l y after the act i vated I ight was turned off, a ll owin g low but measurab le activity to be detected (Fig. 4). The activit y was inhibited by free Ca l + ions (fi g. 4, inset) also indicating an interacti on of the free ions with the membrane bound enzyme. The inhib ition by Ca2+ ions in the presence of CaATP was much stronger than that of free Mg2+ ions in the presence of MgATP (Table 2). The low rate of hy­dro l ys i ~ of CaA TP cou ld be a result of further decay of the ac tivat ion state, the maintenance of which de­pend s on L1~H+ . Since very little proton pumping is supported by CaATP hydro lysis , the ac ti vity should have dropped during the measurements. T he surpri s-

Kj M free Vma,,( K(,(/ /K", ( ~lmoles/mg/min )

0.005±0.0007 13.S± I. 5 55 .2

I 1.25± 1.2 3.S±0.3 5.4

6 .-------------------------------.

S ..c: 4 U 01)

E --~ en <U

(3 E ::i

2

00 0.1

2 0

00 4 S 12 16 20 Ca(f) [mM]

0.2 0.3 0.4 Ca-ATP [mMJ

Fig. 4-The effect of CaCI, on li ght tri gge red ATPase acti vity in thyl akoids. [Thylakoids were li ght tri ggered as described under "Materials and Methods" except for the omi ssion of di valent metal ions at th is stage. The acti vit y was measured in the dark in the presence of various concentrati ons of CaATP and 0-0.5 mM total CaCI, . In the inset, the effect of free Ca2

+ ions on the rale o f ac ti vity was measured in the dark in the presence of I mM CaATP and various concentrati ons of free Ca2+ ions].

Tahle 2- Kinetic parameters for li ght-triggered ATPase ac ti vity in thylakoid bound CF 1• IThe parameters were calculated from

data as describeu under "Material s anu Methous" and presented in the li gures. M represent s a di valent-metal ionl .

Metal Kill MATP Kj Mfree \fllla ,", Kw/ KIII mM (fl moles/mg(c hl )/hr )

Mo 2+

'" 0.19±0.03 20± 1 gO± 1.5 -1-2 1

Ca2+ O.O7±O.0 1 1.5±O.2 6±0.9 gO

ing low Kill of 0.07 mM for CaATP cou ld al so have been a result of the low activity . A simil ar decrease in Kill was observed in the presence of Mg2+ iOll s at low concen trati ons of the act ivator sulfite44 or at hi gh t,~H+ 4). Although the Kill of CaATP was lower than that of MgATP, the K,,,/ Kill was 5.3 fo ld lower in the presence of ci+ ions. A much more effic ient cataly­sis resulted in V""" ratio of 13 .3 in the presence of MgATP and CaATP, respect ive ly. It mi ght be ind ica­tive fo r a weaker binding of the transition state inter­mediate in the presence of C}+ ions.

464 IND IA J. BIOC HEM. BIOPH YS. , VOL. 37, DECEMBER 2000

Binding of Ca2+ ions 10 CFoF, and CF,

The K; for inhibition by free Ca2+ ions was low, indicating a weak binding affinity of the free ca ti on in all the forms of the enzyme. Yet the K; for in hib ition by CaADP and the Kill for CaA TP, although larger than those observed in the presence of the Mg2+ ions, were still reasonabl y low. Thus, there were good in­dica ti ons that at least in the presence of nucleotides, Ca2+ ions bind strongly to the enzyme. It was, there­fore, necessary to directly measure the binding of C}+ ions to the enzyme. Binding measurements of Ca2+ to CFI and to CFoFI were done with the aid of the ion specifi c chelator dye Arsenazo III having a K" of 5 flM for Ca2+. The titration with CaCI 2 all owed determ inati on of the free metal after equilibrium was reached. Binding data were ca lcul ated and plotted according to the Scatchard equati on. In the presence of two moles of ADP per mole of CFoFI, one strong and several weak binding sites with K" of 0.03±0.006 flM-' and >50 flM-' respecti vely were detected (Fig. 5). The loose sites could not be determined accurately because their di ssoc iati on constant was apparent ly out of the range for measurement with th is dye. More than 90% of the ADP was bound to the enzyme under the ex perimental condit ions according to our calcul a­ti on when the K; for inhibition of CaA DP was used as K" for bind ing. The binding to the enzyme was much stronger than to free ADP which has onl y a K" of 1.53 mM-' . Therefore, very littl e interfe rence fro m binding to free ADP coul d be ex pected. Without ADP the binding of Ca2+ ions was very weak and could not be accurately measured (data not shown). Thi s fi nding is in harmony with the observati on of rather hi gh K; fo r inhibition of ATPase activity by free Ca2+ ions.

To our knowledge, thi s is the first report of direct binding measurement of Ca2+ ions to CFoFI. It was, therefore, of interest to compare it to the binding of Ca2+ ions to CFI . As in the case of isolated CFoFI, the binding of ci+ ions to CFI in the absence of ADP was very weak and could not be accurately deter­mined (data not shown). However, in the presence of two moles of ADP per mole of CFI, one strong and five weak binding sites with K" of 0.1 ±O.O I flM-' and IO I±2 1 flM ' respectively were detected (Fig. 6). The three-fo ld weaker binding was ex pected since the in­hibiti on by CaA DP was also fi ve-fold weaker in CFI than in CFoFI as seen from the K; values of 165 flM42

and 35 flM (Table I) respectively. There was a good correlation between the bi ndi ng of Ca2+ ions to the

enzyme in the presence of ADP and the inh ib ition of acti vity by CaADP. Based on thi s correlati on, it is suggested that the cations bind to the ac ti ve site of the enzy me. The dissoc iation constant for bi nding was expected to be lower than the K; fo r inhibi tion by CaADP because the inhibi tion was measured in the presence of A TP that weakened the binding of sub­strate and products to the ac ti ve si tes . There was a major difference between the binding of Mn2+ and Ca2+ ions to CFI. The pos iti ve interac ti on among three

5 r 10 15

Fig. 5-Bindi ng of Ca2+ ions to iso lated CFoF I . [Bindi ng to

CFoCFI in the presence of 2 moles of ADPI CFoCFI was meas­ured by ti trat ion wi th CaCI2 as described under ··Materials and Methods" . The concentrati on of the free Ca2

+ ions was determined fro m the absorpt ion change of Arzenazo II I at 648 nm. Scatchard analys is was used to present the data, where moles of bound Ca2

+

per mo le of CFoFI (r) are plotted against r/Ca(free). The calcula­ti ons indicated one strong and many loose bind ing sites with K" 0.03±0.006 o/" IJU I, > SO pU I, respectively!.

10.---.---,----,---,---,,---.

8

6 --. Q) Q) "- 4 ---~ 0 ">:::

2

0 0 1 2 3 4 5 6

r Fi g. 6-Binding o /" Ca2

+ ions to iso lated CFI. [Binding to CFI in the presence o /" 2 mo les o /" ADP/CFI was measured by tit ration with CaCI2 as described under "Materi als and Methods" and in Fig. 5. The calcul ations indicated one strong and li ve loose bind­ing sites wit h K" of O. IO±O.O I 11M-I and IOI±2 1 IJU I. respec­ti vely].

HOCHM AN et 01. : ROLE OF METAL 10 S IN COUPLI NG IN ATP SYNTH ASE 465

si tes in the binding of Mn2+ was not observed ' 7 in the case of Ca2+ ions. It is poss ible that the limitation in the sensitivity of the dye used for the detection of free Ca2+ ions prevented the measurement of a wider range of sites and , therefore, obscured the interacting sites. We are presentl y trying other dye systems in order to improve the binding measurements.

Simlllation of C((2+ iOIl billding to the active site of FI A visualization of the poss ible change in structure

on substitution of Mg2+ by Ca2+ ions was done by simulation. The structure of the phosphate-binding loop (P-Ioop) of ~TP of bovine mitochondria FI which contains MgAMPPNP " overlapped the struc­ture of myosin containing MgA DP-vanadate46

. Thi s appeared to be reasonab le since it was demonstratecf24

that there is a considerabl e conservat ion in the di stri­bution and nature of the li gands that coordinate the triphosphate moiety in many purine nucleotide-

binding proteins. The Mg2+ of FI that is bound to onl y three resolved li gands, was repl aced by the octahedral Mg2+ from myosin or by octahedral Ca2+ deri ved from the seven coordinated cations bound to GDP-AIF4 in transducin47

. Superimposition of the two cations and their ligands revealed that binding of Ca2+ ions caused an increase of 0.3 A in the average bond length (Fig. 7). The alterati on is due to an average length change of 0.16 A and 0.44 A in the li gands to the phosphate oxygens and the threonine and to the water molecu les respectively. It also caused a di spl acement of the ca­tions by 0.49A. The interacti ons of the three water molecules, bound in the first coordination sphere of the metal , with amino acid side chains were al so changed with the cation. It should be noted that , in the presence of Ca2+, two such water molecul es were 0.46 and 0.88 A closer to the carboxy l oxygen of Glu 188 and Asp 256 respecti vely (Fig. 8 A, B). The two water molecules were approximately 4 A from the y

) 2 . ~ .

~~ ~ ' .-2.44 --"' . ' , ") \ -.~ " -< Ca

/ .~. \

('-......, H20

\~

N~d \i Thr

I I .t~ 25:::-.

Fig. 7-Simulation of Ca2+ ion bind in g to the acti ve site of Fl' [The Mg2+ in MgAMPPNP of PTP of bovine mitoc hondri a FI (ref. 3) was repl aced by the octahedral Mg2+ from myosin MgADP- vanada te~ (, or by octahedral Ca2+ derived fro m the seven coordinated cati on bound to GDP-AIF~ in transd ucin47 The Mg2+ ion and the three bound water molecu les were labeled in bl Jck and Ca2+ ion and the bound waleI' molecu les wcrc labcled in gray. PA, PH and PG are the a , p, and y phosphorus, respectively, of MgAM PPNP of PTP of Fl' The MgA MPPNP and the threonine retained the coordinates as in the crystal struclUre of Fl ' The coordinates used were from the I BM F, I YOM and ITAD liles of the Protei n Data Bank for F" myosin and transducin , respect ively. The Insight II 98.0 so ft ware program from Molecular Simulati on Inc. was used for simulationl .

466 IND IAN J. BIOCHEM. BIOPHYS ., VOL. 37, DECEMBER 2000

phosphorus. Their close proximity to the carbox yl in the presence of Ca2+ ions could make them better nu­cleophyles during hydro lys is of ATP.

Discussion Both Mg2+ and Ca2+ ions support hydrolys is that is

(A)

ANP

(B)

AN P

PA

the reverse reaction to ATP synthesis. However, any activity that requires a forward reacti on such as ATP synthes is and partial reactions, which support isotopic exchange or proton pumping, were catalyzed onl y by Mg2+ ions. Thus onl y Mg2+ ions support ATP synthe-

. 2S ATP d . . ?7 "048 P' ATp?7 49 SIS ' , rIven proton pumpll1g- .. " 1- - .

2 '~8"'" Asp256

Glu188

HOH

Fig. 8-S imul ati o n o r Mg2+ and ci+ ion binding sites at the PTP subunit or F J [A visuali za tion o r the puss ibl e inte racti uns or the ami no

acid s ide chain s with the water molecu les bound to the metal in MAMPPNP of the PTP subunit or F J are presented . The amino acid side chains that interact within 5 A rrum eit her (A), Mgh or (B), Ca2+ ion are shown. The th ree wate r mo lecules added to the metal in F J were not numbered while those of the origi nal structure were nlJlnbered. Other details were as described under Fig. 71·

HOCHMAN el al. : ROLE OF METAL IONS IN COUPLING IN ATP SYNTH ASE 467

and Pi-HI ROH exchanges5o. It seems that an under­

standing of the reasons for the differential effect of these two cati ons might give a clue to the mode of coupling between the chemical catalys is and struc­tural changes at the acti ve site. A closer scrutiny of the catalytic properties revealed that Mg2+ ions were more effi cient catalysts than ci+, even in ATP hy­drolys is. The VMglCa for hydrolys is was larger than one in ::111 forms of the enzyme indicating that Mg2+ was also superi or in catalyzing the reverse reacti on. Smaller differences were found in the Kill of MgATP and CaATP in the hydrolytic acti vities. A smaller KJ for substrate binding in the presence of Ca2+ could be a reason for the higher Kill of the CaATP supported acti vity. Indeed there was al so higher K; for inhibition by CaADP as compared to MgADP, that can also be attributed to a weaker binding. A large difference in the inhibition of ATPase acti vity by the two free ca­ti ons in some fo rms of the enzyme was found . Such di fferences between the two cations might be related to the mode of binding and, therefore, to structural changes. In the thylrrkoids, light- tri ggered ATPase acti vity in the presence of MgATP was poorl y inhib­ited by free Mg2+ ions, while free Ca2+ ions were much better inhibitors of the acti vity in the presence of CaATP. A rati o of 12.5 and 0.006 in the K; for in­hibiti on of ATPase acti vities in the thylakoids and the isolated CFI by free Ca2+ and Mg2+ ions respecti vely was determined. The strong inhibition by free Mg2+ ions that caused an apparentl y hi gher ac tivity led to an erroneous conclusion that the VlIlax for CaATP was hi gher than that of MgATP in the iso lated CFI. Free Mg2+ and Mn2+ are the strongest inhibitors among the various metal ions in CFI. However, oxyanions such as carboxy lic ac ids, sulfite and also the detergent oc­tylglucoside greatl y increased the K; and also caused a 10-fold decrease in the Kill for MgATP. However, these agents increased the Kill for CaATpIJ.51

. In the presence of oxyanions the properties of the iso lated CF1 resembled th ose of the membrane bound enzyme. These are the onl y known conditi ons under which

18 some H OH-Pi exchange was shown to proceed during single site catalys is in the presence of both MgATP and CaATP in CFI (ref. 52).

The inhi bition by free di valent-metal cati ons could result from interference of the cati on with the binding of M-ATP complexes at the acti ve sites. The incom­ing complex will have to be exchanged with a bound metal before catalys is can take place. Since the rate of ligand exchange of Ca2+ is 300 times faster than in

Mg2+ (ref. 53), thi s could ex plain why free Mg2+ ions are stronger inhibitors than ci+ ions in isolated CFI. However, if this were the case, free Mg2+ ions should have been better inhibitors than free Ca2+ for ATPase activity also in thylakoids. However, thi s was not the case as free Ca2+ ions were better inhibitors of ATP­ase activity in thylakoids.

The binding of free di valent metal-cati ons coul d cause the acti ve sites to assume an inhibited confo r­mati on. The strong positi ve cooperati ve binding of Mn2+ ions at the three active was proposed to induce inhibitory conformationl7

. Such an inhibition was expressed in a 20-fold increase in the rate of pre­steady state acceleration of ATPase acti vity in the presence of metal ions l8

. We have proposed that the ti ghtly bound ADP at the acti ve site is the first bind­ing site fo r the metal. Indeed the acti vati on was also correlated to the rate of release of ti ghtl y bound ADP2.

It is poss ible, therefore, that the inhibition was in­directly caused by binding of MADP formed from the free cations and ADP. Such a complex can be formed either with enzyme bound ADP or with ADP that was produced during the hydrolys is of ATP. We have proposed that the enzyme can be in two states of ac ti­vation modul ated by negati ve cooperativity of MADP binding to the ac ti ve sites. Since however, the K; fo r MADP was almost 10-fold hi gher than the K" of the free metal ion, it seems that both free metal and the MADP contribute to the inhibitory effect. Binding of MADP caused an inacti ve conformation of the en­zyme. Pos iti ve cooperati vity, induced by binding of MATP, releases the product during hydro lys is. Re­lease of bound products is also induced by the ~~H+ formed by light-induced electron transport during pre­ac tivation of the membrane bound enzyme or during ATP synthes is. The binding of CaADP was weaker than that of MgADP as indicated by the ratio of K; of 3.8 (ref. 54) and 7.6 in the iso lated and the membrane bOllnd CFI, respecti ve ly.

The detailed anal ys is of the Kill for the M-ATP and of the K; fo r the MADP and free metal ions clearl y indicated that Ca2+ binding was weaker than Mg2+ binding to all forms of the enzyme. Thi s was expected as the K,J for the CaADP and CaATP complexes were approximately two-fold larger than the K" for MgADP and MgATP. The weaker binding was probabl y due largely to the fact that the average bond di stance is larger by approx imately 0.3 A in the Ca­nucleotide than in the Mg-nucleotide complexes. A

468 INDI AN 1. BIOCHEM. BIOPHYS., VOL. 37, DECEMBER 2000

stronger binding of substrate could indicate a st ronger binding of the transit ion state intermed iate with Mg2+ ions than in the presence of Ca2+ ions whi ch would result in a more efficient catalysis . A Mg2+/Ca2+ ratio of K",IKm of 5.3 , 10.2 and 1.5 for the thylakoid mem­brane enzyme, the isolated CFoFI and the so luble CF 1

respectively were determined. These values indicated that Ca2+ ions became less effi cient cata lysts in the more intact and integrated enzyme while Mg2+ ions were equally efficient in all forms of the enzy me. The higher KmlKm in the presence of Mg2+ ions in a ll forms of the enzyme could be correlated to a stronger binding of the transition state intermedi ate in the presence of these ions. Indeed we have shown that divalent cation-ADP-vanadate was transition state inhibitor that bind strongly to the enzyme. This was ev ident from the fact that MADP decreased the K; for vanadate inhibition 30-fold. An average K" of 0.4 nM fo r three interacting binding sites was measured for vanadate in the presence of MADP. The vanadium assumes a pentacovalent tri gonal bipyramidal ge­ometry on binding to the enzyme. Such is the pre­dicted transition state intermed iate when the hydroly­sis involves an on line attack of water on the y­phosphate of A TP.

Although the krill is determined by the K" of the transi ti on sta te intermediate, it was shown that the rate limiting step of the overall catalys is was deter­mined by the rate of release of the enzyme-bound product to the so luti on. The release rate of products was great ly enhanced by positive cooperati ve inter­act ion on the binding of MATP during hydrolysis and on the conformati onal changes induced by the 6pH+ during ATP synthesis . These conformational changes were sequentiall y transmitted to the three ~ subunits by rotation of the y subunit. Since rotation of the y subunit was essenti al for release of the products from the acti ve sites of the enzyme, binding of CaATP should cause rotation and therefore, pumping of pro­tons during ATP hydrolysis. Yet very little or no proton pumping is supported by CaATP hydrol y­sis27Jo. Comparati ve analysis of the relation between substrate binding and the rate of CaATP hydrolys is might exp lain why there was little hydrol ys is in the membrane bound CFoFI as compared to the isolated CF1• It is proposed that the rate of hydrolysis of Ca­ATP is in verse ly related to the degree of coupling between CF 1 and the turbine. I. In isolated CF1 the y subunit could rotate freely and support the release of the product during CaATP

hydro lys is. 2. A loose coupling between the y and c subunits of the CFo turbine in the iso lated CFoFI preparati on could sustain rotation of the y in the CF1 without ro­tation of the c subunits that drive proton pumping and, therefore support a moderate rate of CaATP hy­drolysis. 3. In the membrane-bou nd enzy me, CaATP was hy­drol yzed slowly because the rotation of the y subunit was slow. The process was slow because there was a tight coupling between the rotation of the y subunit and the c subunits of the turbine in the membrane. 4 . In the membrane-bound enzyme, CaATP bi ndi ng was not as strong as MgA TP and did not produce the required fo rce to drive the turbine. Although the overall K eq of ATP hyd rolys is was simil ar in the pres­ence of the two cat ions, the conformatio nal energy required for rotati on of the turbine could not be pro­duced by binding of CaATP and some of the energy of hyd rolys is was lost as heat. 5. Inefficient synthes is was due to the fact that the conformati onal changes produced by the proton driven rotation of the y subunit fa iled to produce a suffic iently ti ght binding of the CaADP. Because the bond distance to the first shell li gands in Ca2+ was longer than in Mg2+, the binding of the former was weaker. The di va lent ca ti ons were bound either di­rectly to oxygen of the phosphate in ADP and in the side chain threonine or indirec tl y through water molecul es to other side chain am in o ac ids. The strong binding at the active site of the ~ subun it is requi red in order to change K(,(, of the reaction to 2.5 41, as was found in the presence of ti ghtl y bound MgATP and MgADP. Such a change was required in the interme­diate catalytic step for production of tightly bound ATP from ti ghtl y bound ADP and Pi. 6. It mi ght be assumed that the equili brium constant for the reactions of the bound substrate and product was larger in the presence of Ca2+ than in the pres­ence of Mg2+ ions. For example, a IO-fold larger K, in

'+ M ' + '11 I the presence of Ca- compared to g- WI C lange the K eq to 25. Such a change could result from a faster backward reaction that is unfavorab le for the synthe­sis of ATP when there is a less ti gh t binding of the substrate to the active site. 7. Alternatively, the transition state conformation cou ld be similar in the forward and backward direc­tions. The transiti on state conformati on of any of the MATP complexes was expected to resemble others in the same enzyme. If the transition state in MgA TP

HOC HMAN et al.: ROLE OF METAL 10 S IN COUPLI 'G IN ATP SYNTHASE 469

was similar to that of CaATP, the reactions in the latter should also proceed in both direc ti ons. Since thi s was not the case differences in the transition state conformations of MgATP and CaATP might ex ist. Such differences could explain why the enzyme catalyzed hydro lys is but not synthes is from CaADP and Pi . Unlike Mg2+ ions, Ca2+ can have more than six ligands. For example, in the transiti on state analogue CaGDP-AIF-I of transducin-l7, the Ca2+ is bound to seven li gands, th ree of whi ch are from the analogue, compared to on ly two in MgADP-vanadate of my­os in46

. Addition of a seventh li gand in the transition state of CaATP mi ght favor hydrolysis rather than synthesis. For example, if the seventh li gand was a water molecule, the dehydration of a pentacovalent phosphorus intermediate might be less favorabl e than the hydrolysis. As seen in the simulation, the longer bond length of the water molecule bound to ci+ places them closer to carboxy lic acid side chains that can serve as general base catalysts. These water molecules became potentially better nucleophyles and favor hydro lys is of ATP in the presence of Ca2+ ions.

References I Pcncfsky H S & Cross R L ( 199 1) Adv Enzymol Relat Areas

Mol Bioi 64, 173-2 14 2 McCart y R E & Carmeli C ( 1982) in PholO.ITnthesis: Energv

conversion by plants and bacteria (Govindjee, ed) pp 647-695 , Academic Press, Y

3 Abrahams J P, Leslie A G, Lutter R & Walker J E ( 1994) Na­tllre, London 370,62 1-628

4 Boyer P D ( 1993) Biochim Biophys Acta 1140, 215-250 5 Grubmcyer C, Cross R L & Penersky H S ( 1982) .I Bioi G elll

257,12092-12100 6 Carmeli, C & Lirshitz, Y ( 1972) Biochilll Biophvs Acta 267,

86-95 7 Nel son N, Nelson H & Racker E ( 1972) .I BiolChelll 247,

6506-65 10 8 Kandpal R P & Boycr P D ( 1987) Biochim Biophys Acta 890,

97-105 9 Duncan T M, Bulygin Y Y, Zhou Y, Hutcheon M L & Cross R

L ( 1995) Proc Natl A cad Sci USA 92, 10964-10968 10 Sabbert D, Engelbrecht S & Junge W ( 1996) Natll re. Lolldon

38 1, 623-625 II Noji H, Yasuda R, Yoshida, M & Kinosita K, Jr ( 1997) Na­

tllre, Londoll 386, 299-302 12 Frasch W D & Selman B R ( 1982) Biochellli.l'lrv 2 1, 3636-

3643 13 Hochman Y & Carmeli C ( 198 1) Biochelllisll ), 20,6293-6297 14 Hochman Y & Carmeli C ( 198 1) Biochelllisll), 20, 6287-6292 15 Hiller R & Carmeli C (1988) Prog Ciin Bioi Res 273 .307-3 14 16 Haddy A, Frasch W D & Sharp R R ( 1985) Biochemistry 24,

7926-7930 17 Hiller R & Carmeli C ( 1985) .1 Bioi Chelll 260, 1614-1 617 18 Canneli C , Lirshitz Y & Gutman M ( 198 1) Biochemistry 20,

3940-3944

19 CarmC\i C, Hu ang J Y, Mill s D M, Jagendorr AT & Lcwis A ( 1986) .I Bioi Chenl 26 1, 16969- 16975

20 Canneli C, Lewi s A & Jagcndorr A T ( 1990) in Progress ill photosynthesis research (Baltsche lTsky M, ed) pp 450-456, Kluwer Academic Publi shers, StJCkholm

2 1 Houseman A L, Lu Bruttu R & Frasch W D ( 1995) Biochelllis­tl )' 34, 3277-3285

22 Houseman A L, LoBrutto R & Frasch W D ( 1994) Biochelllis­II )' 33, 10000- 10006

23 Houseman A L, Morgan L, LoBruttu R & Frasch W D ( 1994) Biochelllistry 33, 4910-49 17

24 Smith C A & Raymcnt I ( 1996) Biophys .I 70, 1590- 1602

25 Jagendorr AT & Avron M ( 1959) Arch Biochelll Biophys 80, 246-254

26 Zhang S & JagendorrA T ( 1995) .1 Biul Cho ll 270, 6607-6614

27 Pick U & Weiss M ( 1988) Ellr.l Biochelll 173,623-628

28 Carmeli C & Racker E ( 1973) .1 Bioi G elll 248 , 828 1-8287

29 Pick U & Racker E ( 1979) .1 Bioi Gelll 254, 2793-2799

30 Casadio R MBA ( 1996) .1 Biuinorg G elll 1,284-291 3 1 Parageorgiou S, Mclandri A B, & So laini G ( 1998) .I Bioenerg

Biolllelllbr 30, 533-541

32 Burger N & Fuess H ( 1977) Acta Crrst B33, 1968-1 970 33 Holt C, Hasnain S S & Hukins D W L ( 1982) Biochilll Biophrs

Acta 719, 299-303 34 Taylor D A, Sack J S , Maune J F, Bccki ngham K & Quiocho F

A ( 199 1) .1 Bioi G elll 266, 2 1375-21380 35 Carmeli C & Lifshitz Y ( 1989) .1 Bacteriol 171 ,652 1-6525

36 Farran F & Rackcr E ( 1970) Biochelllistrv 9,3829-3836 37 Hartree E F ( 1972) Anal Biochelll 48. 422-427

38 Anton G E & Jagendorr A T ( 1983) Biochilll Biol,hys Acta 723,358-365

39 Sugino Y & Miyoshi Y ( 1964) .1 BioI Gelll 239, 2360-2364

40 Scarpa A (1979) in Methods in EnZYlllology (Fleischcr S & Packer L, eds) pp 30 1-338, Academic Press, New York

41 Fromme P & Graber P ( 1990) Biochilll Biophys Acta 1016, 29-42

42 Hochman Y, Carmeli S & Carmeli C ( 1993) .I Bioi G elll 268, 12373-12379

43 Mullet J E, Pick U & Arntzcn C J ( 198 1) Biochilll Biophys Acta 642,149-157

44 Larson E M & Jagendorr A T (1989) Biochilll Biophrs Acta 973, 67-77

45 Bickel-Sandkotter S & Stratmann H ( 198 1) FEBS Lell 125 , 188-192

46 Smith C A & Rayment I ( 1996) Biochemistry 35, 5404-54 17

47 Sondek J, Lambri ght D G, Noel J P, Hamm H E & Sigler P B ( 1994) Nature, Lonlioll 372, 276-279

48 Hartzog P E & Cain B D ( 1994) .I Bioi Gelll 269, 323 13-323 17

49 Carrneli C & Avron M ( 1967) Ellr.l Biochem 2, 3 18-326

50 Kohlbrenner W E & Buyer P D (1983) .I BioI G elll 258 , 10881-10886

5 1 Pick U & Bassilian S ( 1982) Biochelllistry 21,6144-6152

52 Wu D & Boyer P D ( 1986) Biochelllistl), 25 ,3390-3 396

53 Eigen M & Hammes G G (2000) J Alii Chelll Soc 82, 595 1-5952

54 Hochman Y & Carmeli C ( 1995) in PholOsynlhesis: /1'0111

Light 10 Biosphere (Mathi s P, ed) pp 67-70, Kluwer Aca­demic Publisher, The Nethcrl and s