ATP Regulation of Sarcoplasmic Reticulum Ca2+-ATPase

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 263,No. 25, Issue of September 5, pp. 122&3-12294,1988 Printed in U.S.A.

ATP Regulation of Sarcoplasmic Reticulum Ca2+-ATPase METAL-FREE ATP AND 8-BROMO-ATP BIND WITHHIGH AFFINITY TO THE CATALYTIC SITE OF PHOSPHORYLATEDATPase AND ACCELERATE DEPHOSPHORYLATION* (Received for publication, October 26,1987)

Philippe Champeil, Sylvie Riollet, Stephane Orlowski, and Florent Guillain From the Service de Biophysique, Departement deBiologie, CEN Saclay, 91 191 Gif-sur- YvetteCedex, France

Christopher J. Seebregts andDavid B. McIntosh From the Medical Research Council Bwmembrane Research Unit, Departmentof Chemical Pathology, University of Cape Town Medical School, Observatory 7925, Cape Town, South Africa

To localize and characterize the regulatory nucleotidesite of skeletal muscle sarcoplasmic reticulum Ca2+-ATPase,we have investigated the effects of ADP, ATP, and analogues of these nucleotides on the rateof dephosphorylation of both native ATPase and ATPase modified with fluorescein 5’-isothiocyanate (FITC), a reagent which hinders access of nucleotides to the ATPase catalytic site without affecting phosphorylation from Pi. Dephosphorylation of the phosphoenzyme formed from Pi was monitored by rapid filtration or stopped-flow fluorescence, mostly at 20 OC, pH 6.0, and in the absence of potassium. Fluorescence measurements were made possible through theuse of 8-bromoATP, which selectively quenched certain tryptophan residues of the ATPase, thereby allowing the intrinsic fluorescence changes associated with dephosphorylation to be measured in the presence of bound nucleotide. ATP, 8-bromo-ATP, and trinitrophenyladenosine diand triphosphate, but not ADP, enhanced the rate of dephosphorylation of native ATPase 2-3-fold when added in the absence of divalent cations. Millimolar concentrations of Mgz+eliminated the acceleratingeffects. Acceleration in theabsence of Mg2+ was observed at relatively low concentrations of ATP and 8-bromoATP (0.01-0.1 mM) and binding of metal-free ATP and ADP, but not Mg. ATP, to the phosphoenzyme in this concentration range was demonstrated directly. Modification of the ATPase with FITC blocked nucleotide bindingin thesubmillimolar concentration range and eliminated the nucleotide-induced acceleration of dephosphorylation. These results show that dephosphorylation, under these conditions, is regulated by catalytic ATP but not by Mg .ATP or ADP, and that the site is the locus of this “regulatory”ATP binding site.

are accelerated by ATP binding to a “regulatory” site on the enzyme, at concentrations higher than thatrequired for phosphorylation and turnover of the pump (1-7). Several intermediate steps of catalysis have been identified as being modified by ATP, as well as by various analogues of ATP. Thus, the conformational change associated with Ca2+binding to the high-affinity sites on the ATPase is accelerated by ATP binding (8-10). The rate of Ca2+ release to the lumen of isolated SR vesicles is likewise increased (11, 12). The subsequent dephosphorylation step hasbeen found to be inhibited (13), accelerated (12-14), or not affected (15)by ATP binding, depending on conditions. In the latter study, ADP, rather than ATP, accelerated phosphoenzyme processing (15). The phosphoenzyme intermediates must be major targets for the stimulating effects of high concentrations of ATP, since, under most conditions, the predominant steady-state species of ATPase is a phosphorylated form; significant modulation of overall hydrolysis can only be expected if the rate-limiting processing of the predominant species is stimulated. The locus of the regulatory nucleotide site is uncertain. It could be an allosteric site, physically distinct from the catalytic site. Alternatively, it could be the catalytic site of intermediates which develop following phosphorylation and ADP dissociation. There is evidence for both possibilities (13, 1620). The issue is complicated by current uncertainties regarding the quarternary structure of the membranous ATPase, and the possibility that regulation by ATP may depend, in part, on subunitinteractions of an oligomer (13, 20, 21). Recently, the idea that theregulatory nucleotide site resulted from transformation of the catalytic sitegained support from experiments with the high-affinity fluorescent ATP analogue TNP-AMP (22) and from intramolecular cross-linking of the catalytic site (23). In this paper, we investigate the requirements for nucleoCalcium transport and ATP hydrolysis catalyzed by the Ca*+-ATPaseof skeletal muscle sarcoplasmic reticulum (SR)’ tide-induced acceleration of dephosphorylation of phosphoenzyme formed from Pi, and attempt to distinguish between the * The costs of publication of this article were defrayed in part by models for the locus of the regulatory nucleotide binding site the payment of page charges. This article must therefore be hereby by investigating the effect of modifying the active site with marked “advertisement” in accordance with 18 U.S.C. Section 1734 fluorescein isothiocyanate (FITC), which blocks access of solely to indicate this fact. to the catalytic site without hindering phospho’ The abbreviations used are: SR, sarcoplasmic reticulum; Mes, 2- nucleotides rylation from Pi (24-27). Acceleration of dephosphorylation (N-morpho1ino)ethanesulfonicacid EGTA, [ethylenebis(oxyethylenenitri1o)ltetraacetic acid FITC, fluorescein 5’-isothiocyanate; 8- of native ATPase was demonstrated by two methods: either Br-ATP, 8-bromo-ATP; Azals7,calcimycin; TNP-AMP, -ADP, and - by measuring 32P-labeledphosphoenzyme levels in conjuncATP, 2’(3’)-0-(2,4,6,-trinitrocyclohexadienylidene)-adenosine5’- tion with a rapid filtration device (12) or by monitoring mono-, di-, and triphosphate; CazEIP andE,P, ADP-dependent and intrinsic fluorescence change by stopped-flow fluorimetry ADP-independent phosphoenzymes, the two main conformations of phosphorylated enzyme in the cycle; Mops, 3-(N-morpholino) (28). The latter measurements are not possible using ATP as propanesulfonic acid AMP-PCP, adenylyl P,y-methylene diphos- the effector nucleotide, as the decrease in intrinsic fluorescence associated with dephosphorylation is counterbalanced phate.

12288

12289

ATP Regulation of SR Ca2+-ATPase by an increase on ATP binding. In order to circumvent this problem, we synthesized 8-Br-ATP, which we demonstrated to be almost as good a substrate as ATP in terms of CaZ+ transport rates,K,,, for phosphoenzyme formation, and binding affinity, but does not show the increase in intrinsic fluorescence on binding, due most probably to the fluorescencequenching propertiesof the bromine atom (29). ATP and 8-Br-ATP stimulated the hydrolysis rate of Piderived phosphoenzymein the absence butnot in the presence of M P . This regulatory effect of ATP was not observed after modification of the ATPase catalyticsite with FITC,suggesting transformation of the latter type of site into a regulatory site in the native ATPase upon phosphorylation.In addition, we found that, in the absence of divalent cations, the stimulating effect ofadding ATP to phosphorylated unmodified ATPase was characterized by a relatively high affinity (between 0.01 and 0.1 mM). Using the fast filtration device, this enabled us to demonstrate direct binding of metal-free [3’P] ATP or [I4C]ADP(but not Mg.ATP) to phosphorylated ATPase.

L Y

0

0

.1 1 [ATP] or[Mg.ATP]

10

,mM

(lop . a 1 0 1

FIG. 1. Modification of SR ATPase by FITC. In A, SR vesicles were modified with FITC as described under “Experimental Procedures,” except for different concentrations of FITC (plotted along the Materials-FITC, acetyl phosphate, and ATP (Nos. A5394 and abscissa). For very low FITC concentrations, the incubation period A0270) were purchased from Sigma, bromine from Kanto Chemical was slightly longer than the 20 mn specified under “Experimental Co., Tokyo. [3zP]Pi,[“CIADP, and [-p3*P]ATPwere purchased from Procedures,” since parallel measurements of the labeling kinetics Amersham Corp. showed that the reaction was not completed within 20 min. Open General Procedures-SR vesicle preparation (281, measurement of circles and triangles: Residual activity of ATPase (open circles) and ATPase activity and Ca2+ transport activity (30, 31), steady-state phosphoenzyme formation (28), nucleotide binding and Ca” release calcium transport (open triangles) after labeling. The activities were (12) have been described. Stopped-flow fluorescence measurements measured a t 20 “C after dilution of the reacted enzyme to 0.1 mg of 7 medium which contained 5 mM oxalate, 1 mM were either performed on a Dionex system or a Biologic stopped-flow protein/ml in a pH fluorimeter, with similar results. In the Biologic unit, the optical ATP, 100 mM KCI, 5 mM M$+, and either 50 mM Mops buffer for Ca2+ uptake measurements or5 mM buffer for ATPase activity pathway is very short because the observation chamber is perpendicmeasurements. To measure the Ca2+uptake rate, 0.08 mM murexide ular to the excitation beam; this is an advantage for strongly absorbing was added to the above medium (optical monitoring with an Aminco or turbid solutions. Time-resolved filtration measurements were carDW2 spectrophotometer, using the wavelength pair 550 nm/487 nmf. ried out using a Biologic fast filtration unit (12), either by flushing a dephosphorylation solution through a filter loaded with 32P-phos- To measure the ATPase activity, the amount of protons released phoenzyme, or by flushing a radioactive nucleotide-containing solu- during ATP hydrolysis was monitored with a pH meter (pH only varied between 7.0 and 6.8). The reaction was initiated by adding 0.1 tion through afilter loaded with an unlabeled enzyme suspension. Modification with FZTC-SR vesicles (2 mg/ml) were incubated mM Ca2+.Open squares: Ability of FITC-reacted SR vesicles to be for 20 min in a medium containing 0.3 M sucrose, 1 mM M%+, 0.01 phosphorylated by Pi in a medium containing 15% dimethyl sulfoxide, mM Caz+,and 10 or 100 mM Tricine-Tris, pH 8.0 and 20°C, in the 0.1 mM [3zP]Pi,20 mM M$+, 2 mM EGTA, and 130 mM Mes-Tris presence (unless otherwise stated) of 0.016 mM FITC, i.e. 8 nmol of (pH 6, 20°C). For this assay, the phosphorylation reaction was FITC/mg of protein. The FITC stock solution was made daily, and quenched with acid, and precipitated 3ZP-labeledphosphoenzyme was consisted of 2 mM FITC in dimethyl formamide. The reaction was measured after filtration through a glass fiber Gelman A/E filter. generally stopped or slowed down bylowering the temperature to 0 “C Closed squares: Stoichiometry of labeling. After incubation with different concentrations of FITC, vesicles were diluted to 0.1 mg/ml in and/or changing the pHfrom 8 to 7 or 6. a pH 7 solution and then filtered through a Millex GS filter. The FITC modification inactivated Ca2+transport andATPase activity monophasically, with the extrapolation to zero activity occurring a t concentration of unreacted FITC was deduced from the measured approximately 6 nmol of FITC/mg of protein (Fig. L4, open circles fluorescence of the filtrate (excitation and emission wavelengths were and triangles). In separateexperiments, the stoichiometry of labeling 495 and 525 nm). The amount of FITC which reacted with the was directly measured by fluorescence after filtration (closed squares, ATPase was obtained by subtraction and expressed as nanomoles per Fig. L4).The amount of reacted FITC a t increasing FITC concentra- milligram of protein (closed squares, right-hand scale). In B, ATPtions leveled off at about 6nmol of FITC/mg of protein. These results induced fluorescence changes of bound FITC areshown. After vesicle are consistent with 1:l stoichiometric modification of active ATPase labeling under standard conditions, unreacted FITC was removed by polypeptides, as has been found by others (18, 27, 32, 33). Ca2+ centrifugation; the modified SR was resuspended at a final concentransport with acetyl phosphate (data not shown) and Pi phospho- tration of 0.01 mg/ml in a medium containing 150 mM Mes-Tris (pH rylation (open squares, Fig. 1A) were virtually unaffected by the 6, 20°C) and either 1 mM EDTA (circles), 20 mM M 2 + and 2 mM modification, in disagreement with a recent report (34),but in agree- EGTA (triangles), or 20 mM M%+ and 0.1 mM Ca2+(squares). The fluorescein fluorescencewas monitored in a PerkinElmer fluorimeter ment with the original findings of Pick and Bassilian (26). It will be shown (Fig. 2) that FITC modification eliminates ATP (excitation and emission wavelengths were 500 and 530 nm, respecand 8-Br-ATP binding in the submillimolar concentration range. tively). ATP from a concentrated solution was added, either as M F However, nucleotide binding appears to occur a t higher concentra- free ATP (Na2ATP, buffered with Tris), or as Mg.ATP (equimolar tions. This can be demonstrated by monitoring the fluorescence mixture ofMgC12 and the previous ATP solution) (triangles and properties of the bound fluorescein: as much as 30% decrease of the squares). The indicated changes have been corrected for dilution and bound FITC fluorescence can be observed upon addition of very high for the small change observed upon addition of a corresponding concentrations of ATP to the modified SR vesicles (Fig. 1B). The amount of NaCl. large amplitude of the fluorescence drop and its sensitivity to ionic conditions (Fig. 1B) suggest that thelocus of this low-affinity binding Synthesis and Characterization of 8-Br-ATP-The synthesis and site is close to the fluorescein moiety, indeed in the catalytic site, as purification of 8-Br-ATP was carried out asdescribed (35). [y3’P]8will be discussed again later. This implies that access of ATP to the Br-ATP was synthesized enzymatically by a procedure developed for “high-affinity” nucleotide site is not strictly forbidden by FITC ATP (36). modification, but only perturbed to a considerable extent, so that its Fig. 2 shows that 8-Br-ATP is a comparable substrate to ATPfor affinity is lowered by 2 or 3 orders of magnitude (see also Fig. 6C). the ATPase. A shows dual wavelength experiments comparing the EXPERIMENTAL PROCEDURES

12290

ATP Regulation of SR Ca2+-ATPase 0

Time, s I

"-"

"

"

I

"""___

9 -I

r

" 1

A

Nucleotide1

, uM

FIG. 2. 8-Br-ATP as a substrate of the Cas+-ATPase. A , Ca2+ transport anddissociation during hydrolysis of ATP (upper trace) or 8-Br-ATP (lower trace) by SR vesicles. Vesicles (0.4 mg/ml) were suspended in a medium containing 150 mM Mes-Tris (pH 6.0,20 "C), 20 mM M e , 0.09 mM Ca2+,and 0.08 mM murexide. Medium Ca2+ concentration was monitored by dual wavelength spectroscopy (wavelength pair 550 nm/487 nm). After calibration with 2 X 0.005 mM Ca2+,addition of 0.005 mM ATP (upper trace) or 8-Br-ATP (lower trace) initiated Ca2+uptake, which could be reversed with ionophore (0.4% (w/w) A2S187). After further addition of 4% (w/w) A23187, which made the vesicles completely leaky, either 0.005 mM (single arrow) or 0.05 mM (double arrow) of ATP or 8-Br-ATP revealed that, in both cases, the accumulation of Ca2+-depletedintermediates (E2 and E2P) during steady-state ATPase turnover led to the transient dissociation of Ca2+ from the ATPase. B , equilibrium binding of ATP (closed symbols) and 8-Br-ATP(open symbols) to SR vesicles. HA Millipore filters were first loaded with 0.3 mg of SR protein and then perfused with a solution containing various concentrations of [3ZP]ATP(closed symbols) or [3ZP]8-Br-ATP(open symbols). The medium contained 150 mM Mes-Tris (pH 6.0,20"C), 5 mM EDTA, and [3H]sucroseto allow estimation of the volume trapped in the filter. The signal/noise ratio is of course low in measurements made at high nucleotide concentrations. Experiments were performed with native vesicles (circles) or FITC-modified vesicles (triangles). ability of small amounts of nucleotide to promote either Ca2+accumulation into the vesicle lumen (first addition of substrate),or transient Ca2+dissociation from the high-affinity Ca2+ sites upon further additions of substrate if the preparation is made leaky with a CaZ+ ionophore. The traces show thattransport efficiency, Ca" dissociation, and hydrolysis rates aresimilar for ATP and8-Br-ATP. B shows that both nucleotides bind to the extent of 5-6 nmol/mg of protein in the presence of EDTA with equal affinity. Prior modification of the ATPase with FITC prevented high-affinity nucleotide binding in both cases. Under different conditions, the K,,, for phosphoenzyme formation from [y-32P]8-Br-ATPwas 0.4 p M (100 mM Mops-Tris, pH7.0,O "C, 1mM MgC12,O.l mM CaC12, 4%(w/w) A23187).

I

A

4.

0-

Time, s FIG. 3. Effect of nucleotides on the rate of dephosphorylation of native ATPase and FITC-ATPase, measured by rapid filtration. SR vesicles (0.3 mg of protein/ml) were incubated in a phosphorylation medium containing 15% (v/v) dimethyl sulfoxide, 20 mM M$+, 2 mM EGTA, 0.2 mM ["PIPi, and 130 mM Mes-Tris (pH 6.0, 20°C). 1 mlof this suspension was layered onto an HA Millipore filter in the Biologic fast filtration device, and thenflushed for various periods (the flow rate was 2 ml/s for the shortest periods) with a dephosphorylation solution containing 5 mM EDTA (except for panel D), 150 mM Mes-Tris (pH 6.0), and various nucleotides. The whole experiment was performed a t 20°C in a temperatureregulated room. A , rate of dephosphorylation of native ATPase, in the absence (circles) or presence (triangks)of 1 mM ATP. B , experiments performed under control conditions (circles) or in the presence of 1 mM ADP (squares) 0.005 mM TNP-ATP (V),or 1mM 8-Br-ATP (A). For short perfusion periods, duplicates were measured as in A ; only the average is shown. C, rate of dephosphorylation of FITCATPase, in the absence (circles) or presence (triangles) of 1mM ATP. D, same experiment as in A , except that the dephosphorylation medium contained 20 mM M$+ instead of EDTA.

no radioactive Pi was then flushed through the filter for various periods with a fast filtration device, and the residual amount of 32Pbound to the membranes on the filter was measured. When ATP (1 mM), 8-Br-ATP (1 mM), or TNPATP ( 5 p ~ was ) included in the EDTA-containing dephosRESULTS phorylation medium, dephosphorylation was accelerated apThe effect of ATP, 8-Br-ATP, TNP-ATP, and ADP on the proximately 2-fold, whereas inclusion of ADP (1 m M )had no rate of dephosphorylation of phosphoenzyme formed from Pi effect. TNP-ADP ( 5 PM) was as effective as TNP-ATP (reis shown in Fig. 3, A and B . In these experiments, Millipore filters were first loaded with native SR ATPase previously sultsnot shown). AMP-PCP (1 mM) had an intermediate phosphorylated by incubation with [32P]Piin a Ca2+-deprived effect. When the dephosphorylation medium contained Mg+ medium, in the presence of M e and dimethyl sulfoxide, at (20 mM) instead of EDTA, the dephosphorylation rates in the see also Ref. pH 6.0 and 20 "C; to measure the rate of ATPase dephospho- presence or absence of ATP were the same (D; rylation, a dephosphorylation medium containing EDTA but E ) , suggesting that metal-free ATP, and not the Mg.ATP

12291

ATP Regulation of SR Ca2+-ATPase

0

complex, stimulates dephosphorylation.' Fig. 3C compares the rate of dephosphorylation of FITCATPase in the absence and presence of 1mM ATP, andshows that thepresence of the fluorescein moiety at the active site substantially blocked the nucleotide effect, although not completely at this concentration of ATP (see also Fig. 6, B and

C).

2

Time, 8

'1

I

480

FIG. 4. Intrinsic fluorescence changes upon phosphorylation and nucleotide binding. Emission wavelength was 330 nm. Excitation wavelength was298 nm, to limit the drop in detected fluorescence due to absorption of excitation light by 8-Br-ATP; this drop was larger with 8-Br-ATP thanwith ATP, because the absorption spectrum of 8-Br-ATP is shifted toward long wavelengths by a few nanometers in relation to the ATPspectrum. In this figure, the fluorescence levels indicated were corrected for changes in detected light intensity due to sample dilution. Measurements were with SR vesicles (0.1 mg of protein/ml) in 150 mM Mes-Tris, pH 6.0 (20"C), 0.03 mM CaC12 and additions, where indicated, as follows: 4 mM EGTA, 11.25 mM Pi, 11.25 mMMgC12, 2 mM (single arrow) or 12.5 mM (double arrow) EDTA, 0.2 mM (single arrow) or 2 mM (double arrow) ATP, 0.2mM 8-Br-ATP. In C and D the medium contained 2 mM EDTA and in D it also contained 8 mM ATP.

1

t

lD4t

\\

FITC-SR

-

IO2 ID0 01 +Br~ 2ATP"

4-

4

8

10

Time, s

FIG. 5. Effect of 8-Br-ATP on the rate of dephosphorylation of native and FITC-ATPase measured by stopped flow fluorescence. A, diagram showing the relative fluorescence levels of Ca2+-deprived ATPase (E2), Pi-derived phosphorylated ATPase (EZP), and the noncovalent complex between unphosphorylated ATPase and either ATP (E2.ATP) or 8-Br-ATP (E2.8-Br-ATP). This diagram was based on measurements made in Fig. 4. B and C, typical stopped-flow fluorescence recordings of dephosphorylation experiments, after averaging 5-10 individual shots. Excitation wavelength was295 nm; emitted light was measured after its passage through an interference filter (Balzer 330 nm). A Biologic stoppedflow fluorimeter was used in this experiment (see "Experimental Procedures"). The enzyme syringe contained either 1 mg of protein/ ml ( B ) or 0.6 mg of protein/ml (C), 11.25 mM Pi, 11.25 mM MgCI,, 4 mM EGTA, and 150 mM Mes-Tris, pH 6.0. The "substrate" syringe contained 20 mM EDTA, 8.75 mM Pi, and 150 mM Mes-Tris, so that the final pH after mixing was 6.0; 8-Br-ATP was included in this syringe. Temperature was 20°C. B , native SR vesicles, in the absence (upper trace) or presence (lower trace) of 8-Br-ATP (0.3 mM final concentration). C, same experiment with FITC-ATPase. Note change in time scale.

Similar results were obtained using intrinsic fluorescence changes to monitor dephosphorylation. These experiments were made possible through the use of 8-Br-ATP, instead of ATP, to enhance dephosphorylation. Fluorescence monitoring of dephosphorylation in the presence of nucleotide is dephosphorylation experiments performed in the presence of difficult because the fluorescence levels of the Pi-derived nucleotide. phosphoenzyme (E2P) (28, 37) andthe dephosphorylated Experiments monitoring the rate of dephosphorylation in enzyme in the presence of ATP (E'. ATP) (38,39) are similar. the presence of excess EDTA by stopped-flow fluorimetry are This is shown in Fig. 4; both Pi-induced phosphorylation ( A ) shown in Fig. 5, B and C. 8-Br-ATP enhanced the rate of and ATP binding ( B ) cause approximately a 3% increase in dephosphorylation of nativeATPase approximately 2-fold intrinsic fluorescence. In contrast, binding of 8-Br-ATP did (Fig. 5 B ) . 8-Br-ATP was equally effective in stimulating not increase the fluorescence of the ATPase. Addition of this dephosphorylation of native ATPase in the presence of 100 nucleotide (0.2 mM), in fact, caused a 9.7% drop in detected mM KC1 at 6.0 or in the absence of KC1 at pH 7.0 (data not light intensity (C) whichwas shown to be almost entirely shown). Prior modification of the ATPase with FITC again attributable to absorbance by the nucleotide since an approx- eliminated this stimulating effect (Fig. 5 C ) . In the example imately equivalent change (9.1%) was observed in the pres- shown, the concentration of 8-Br-ATP is 0.3 mM and the ence of excess ATP (D). The slightly lower fluorescence level result cannot be compared directly with that shown in Fig. 3, of E2.8-Br-ATP relative to E2 may indicate that thebinding where 1mM ATP is used and aslight stimulation was observed of 8-Br-ATP marginally lowers the fluorescence level of the (see, however, Fig. 6). To obtain reproducible results, careful control of the temATPase (schematic diagram in Fig. 5 A ) . These effects of 8perature was a crucial requirement because the dephosphoBr-ATP areprobably due to a quenching effect of the bromine rylation rate is highly temperature-sensitive (Ea= 30 kcal/ atom on a neighboring tryptophan residue(s), possibly Trpmol, Ref. 37); this means that the relative margin of error in 552 in the proposed nucleotide binding domain (40). 8-Brthe rate will be 20% if the temperature is regulated to within ATP could therefore be used as a substitute for ATP in 1"C. In Fig. 6, the observed rate constantsof dephosphorylation * In our present experiments, the dephosphorylation rate in the absence of nucleotides was slightly stimulated when dephosphoryla- of native and FITC-ATPaseobtained both from stopped-flow tion was performed in the presence of M e compared with in the fluorescence ( A ) and rapid filtration ( B and C) experiments presence of EDTA and absence of added M$+ (cf. Fig. 3, D and A ) . are plotted as a function of nucleotide concentration. The 2-

A T P Regulation of SR Ca2+-ATPase

12292

0 3

r'

.5

2

1

!r

I

v)

I

1

1

1

3r

v)

P 0

Y

3 6

+ H h

0

I

1 .

?

P [ATP]

,

pM (log scale)

FIG. 6. Nucleotide concentration dependence of the observed rate constant of dephosphorylation of nativeand FITC-ATPase. A, 8-Br-ATP dependence of the observed dephosphorylation rate constant, deduced from the observed drop in fluorescence. Circles, native ATPase; triangles, FITC-ATPase. Conditions as for Fig. 5. Note that the significant absorbance by 8-Br-ATP prevents investigating the effects of very high concentrations of this nucleotide. B and C: ATP dependence of the measured rate constant of dephosphorylation, deduced from rapid filtration experiments, as in Fig. 3. To improve precision, the first points in each dephosphorylation curve were determined in triplicate and averaged. Circles, native ATPase; triangles, FITC-ATPase.In B, NaOH-neutralized ATP was used.The continuous line shows the trueeffect of ATP per se, in relation to C. In C the effects of NaOH-neutralized ATP (open symbols) are compared with Tris-neutralized ATP (closed symbols) using native ATPase (circles) and FITC-ATPase (triangles). The temperature for this experiment was probably slightly lower than the one for the experiment described in B, resulting in slightly lower rate constants.

fold increase in observed rate constant of the native ATPase occurred, using either method, in the10-100 pM concentration range. No stimulation was observed with the FITC-ATPase in this range. A pronounced stimulation of dephosphorylation of both native and FITC-ATPase was observed at high concentrations of Na .ATP (Sigma No. A5394, neutralized with NaOH) using the rapidfiltration method (Fig. 6 B ) . This acceleration was not observed with the native ATPase if Tris . ATP (Sigma No. A0720) was used instead of the sodium salt (Fig. 6C), and is evidently related to the known accelerating effect of Na+ andK+ ions on dephosphorylation (41,42). Note that 8-Br-ATP (A) was prepared as the Li+ salt. In the case of the FITC-ATPase, a %fold stimulation of dephosphorylation was observed at very high concentrations of Tris. ATP, suggesting that the effect of modification by FITC was not the complete abolition of nucleotide binding, but only a shift in binding affinity. The relatively high affinity with which the phosphoenzyme binds ATP allowed the direct measurement of phosphoenzyme-bound nucleotide by the rapid filtration technique (Fig.

O I

1

01

.5

1

2

'53

Time ,s FIG. 7. Time course of ATP and ADP binding to phosphorylated and unphosphorylated native ATPase. The procedure used was similar to thatdescribed in the legend to Fig. 3, except that the ATPase on the filter had been preincubated either inthe absence of Pi (open symbols, unphosphorylated ATPase, designated E2) or with 2 mM nonradioactive Pi instead of 0.2 mM ["P]Pi (closedsymbols, fully phosphorylated ATPase, E'P), and that theperfusion medium contained 0.02 mM radioactive nucleotide ([y3'P]ATP for A and B, ["CIADP for C ) , a [3H]sucrose tracer, 150 mM Mes-Tris (pH 6.0, 20 "C), and either 5 mM EDTA ( A and C ) or 20 mM MgClz and 2 mM For all the above curves, the final binding level obtained EGTA (B). at thechosen nucleotide concentration (0.02mM) was consistent with the equilibrium binding level measured separately, using a procedure identical to theone described for Fig. 2B.

7). For this purpose, an ATP concentration of20 PM was chosen which is close to theK d for binding to theATPase in EDTA (see Fig. 2B (unphosphorylated ATPase)). The ATPase was loaded onto the filter and either phosphorylated with non-radioactive Pi or simply incubated in a Ca2+-deprived Mg+-containing medium, and the filter was then perfused with a medium supplemented with radioactive ATP in the absence (A) or presence ( B ) of M P . In Fig. 7B, it is shown that, in the presence of M P in the perfusion medium, ATP bound rapidly to the unphosphorylated enzyme (open circles) but did not bind to the enzyme that was initially phosphorylated until it was dephosphorylated (closed circles; compare with Fig. 30). If EDTA was substituted for M P (Fig. 7 A ) , then rapid binding of ATP occurred to the phosphorylated enzyme (compare with Fig. 3A); in fact, the amount bound to thephosphorylated ATPase after 0.2 s was slightly higher than to the unphosphorylated enzyme. Comparison of the closed circles in A with those in B establishes unambiguously that metal-free ATP binds to thephosphorylated ATPase and,assuming rapid equilibrium, that thebind-

ATP Regulation of SR Ca2+-ATPase

12293

0.3 mM ATP to native ATPase argues in favor of an identical binding locus in both cases, whose occupancy results in similar effects. The present results (Fig. 7) directly confirm that nucleotides bind to the phosphorylated catalytic site of the E2P phosphoenzyme. Inhibition of active site cross-linking on formation of this phosphoenzyme has led to the suggestion that the active site cleft of this intermediate is closed (23). Evidently, if there is indeed closure it is not sufficient to prevent nucleotide binding, despite evidence for apronounced DISCUSSION increase in hydophobicity of the site (47). ADP and ATP bind In this study, we have established that metal-free ATP with apparently equal affinity (Fig. 7, A and C) and yet only binds to Pi-derived phosphoenzyme with relatively high affin- the latter nucleotide destabilizes the phosphoryl group of the ity and accelerates dephosphorylation. The Mg .ATP complex phosphoenzyme. This is presumably due to the steric condoes not bind to thisphosphoenzyme with comparable affin- straints in accomodating the extra phosphoryl group. Identiity. The affinity of the phosphorylated ATPase for ATP in fication of the regulatory site as the phosphorylated active the EDTA-containing medium is in fact slightly higher than site is also in line with previous suggestions that ATP enters that of the metal-deprived unphosphorylated form, as shown the catalytic cycle before it is used as a phosphorylation by the higher level of bound ATP on the former at a total substrate (13, 18). The finding that metal-free ATP, and not Mg- ATP, binds nucleotide concentration of 20 p M (Fig. 7). One purpose of this work was to allow discrimination be- to thePi-derived phosphorylated active site fits with previous tween possible sites for the binding of “regulatory” ATP. First evidence that the E2P intermediate contains a tightly bound of all, our results cannot be explained by ATP accelerating “occluded magnesium which is involved in catalyzing the dephosphorylation through binding to a nonphosphorylated hydrolysis of the phosphoenzyme (14,48). Evidently, a second magnesium ion cannot be accommodated in the site, and it polypeptide of a dimer or oligomer,whose othersubunits would be phosphorylated. The initial amount of phosphoen- provides further evidence that ATP does bind at thecatalytic zyme in our experiments (4-6 nmol/mg of protein, in Figs. 3 site. Similar Mg2+ dependence has been reported for ATP and 7) is close to the totalamount of active ATPase polypep- binding to the vanadate-ATPase complex (49). The nonextides (5-6 nmol/mg of protein; Fig. 2B and Refs. 28, 33, 43). changeability of the magnesium may be the result of active Therefore, ATP must bind to phosphorylated polypeptide site closure. Characteristics of Pi e HOH oxygen exchange catalyzed chains. The lack of ATP and 8-Br-ATP effects on FITC-modified by the ATPase in the absence of Ca2+have shown that the ATPase, as well as the relatively high binding affinity of the rate constantof phosphoenzyme hydrolysis was little affected native phosphoenzyme for these nucleotides, strongly suggests by millimolar concentrations of ATP in the presence of a high that the locus of the regulatory nucleotide binding site is at concentration of M e (20 mM) (13), findings in accord with the active site. The FITC modification has been well charac- the present results. However, the partitioning of bound Pi terized, and shown to predominantly involve Lys-515, which between phosphoenzyme formation and release from the enis located in a conserved region of the primary structure (32, zyme was found to be significantly increased under these conditions (13). Evidently, Mg. ATP may bind to thecatalytic 40), and to result in the inhibition of ATP binding in the submillimolar concentration range to thenonphosphorylated site of E2.Pi, if not of E2P, and this could be due to an catalytic site(25,27;present results, Fig. 2). These character- increased exchangeability of M e at the catalytic site of the . intermediate. istics have led to the generally accepted view that the modi- noncovalent E OPi There is a paradox in ourresults: we demonstrated that the fication is at, or close to, the active site, in the region of the nucleotide binding locus (18, 25, 27, 32, 40). The same inhi- phosphorylated catalytic site in its E2P conformation binds bition of nucleotide binding to the regulatory site, as evi- metal-free ATP in the absence of divalent cations with a denced from our present results, suggests that the regulatory relatively high affinity ( K = 10-100 PM, see Figs. 6 and 7), ATP site is at theactive site, and supports otherevidence for and simultaneously we consider that this site is identical to the so-called regulatory site, which is usually demonstrated this being its locale (13, 22, 23,44,45). A single nucleotide binding site per ATPase polypeptide to operate with a relatively low affinity in magnesium-rich need not contradictthe finding by nuclear magnetic resonance solutions, i.e. we suggest that modulation of ATPase activity that ATP, at high concentrations, interacts with the FITC- does arise from ATP binding to the locus that we have ATPase (46), or our own observation that, at similar high characterized here. Two aspects need to be considered for the concentrations, it binds to the unphosphorylated (Fig. 1B, solution to thisparadox. First, it mustbe recognized that the circles) or phosphorylated (Fig. 6C, closed triangles) ATPase main target for “regulatory” nucleotides, the phosphorylated despite FITC modification. Such “low-affinity” binding need enzyme, may exist under various conformations during turnnot reflect the existence of a second, physically distinct, over. The one that we have studied here, the so-called E2P nucleotide binding site; rather, itmay beexplained by assum- conformation, is predominant during ATPase turnoverat pH ing that, contrary to theusual belief, FITC labeling does not 6 in the presence of a high magnesium concentration and in totally inhibit nucleotide binding to its ATPase site, close to the absence of potassium (12, 31); since Mg. ATP does not Lys-515, but lowers its affinity by 2 orders of magnitude Or bind to this form (Fig. 7B), stimulation of ATPase activity more. Two pieces of evidence support this possibility. First, will not be observed under these conditions (see Fig. ID in nucleotide binding to the FITC-ATPase at high concentra- Ref. 31). On the other hand, most conditions used to demontions lowers the fluorescence of the fluorescein moiety (Fig. strate low-affinity stimulation of hydrolysis by ATP include l B ) , suggesting a close proximity of nucleotide and fluoro- potassium and magnesium as well as a buffer whose pH is phore. Second, the fact that addition of 25 mM ATP to FITC- close to neutrality;theseare conditions under which the ATPase results in the same rate enhancement as addition of Ca2ElP toE2Ptransition is the rate-limiting stepat low ATP

ing exhibits a relatively high affinity, whereas Mg. ATP does not bind. This of course accounts for the absence of stimulation of the dephosphorylation rate by ATP when Mg2+ is present in the perfusion medium (Fig. 3 0 ) . Fig. 7C shows asimilarexperimentin which[14C]ADP rather than ATP was included in the EDTA-containing perfusion medium. Rapid binding of ADP to thephosphorylated ATPase was again observed, although ADP did not stimulate the dephosphorylation rate (Fig. 3B).

12294

ATP Regulation of SR Ca2'-ATPase

concentrations (Ref. 31). Therefore, the observed stimulation of hydrolysis by ATP under these conditions reflects binding of nucleotide to Ca2E1Pand not toE2P.We and othershave shown that, in the presence of magnesium, ATP accelerates the processing of CazEIPwith relatively low apparent affinity (11, 12). Since it is rather unlikely that SR ATPasehas physically different loci for regulation by ATP, one for binding to Ca2ElP and a second for binding to E2P,we conclude that the locus characterized in this paper through the binding characteristics of the E2Pform is identical to theone usually described as the regulatory site. The second consideration is whether the Ca2E1P to E2P conversion is accelerated by metal-free ATP, as with E2P hydrolysis, or by Mg. ATP. In the former case, the apparent low affinity observed for the regulatory ATP under most turnover conditions, which include magnesium in the medium (see, for example, Ref. 22), would reflect the small proportion of metal-free nucleotide compared with Mg.ATP. In the latter case, binding of Mg. ATP to Ca2E1Pbut not toE2P would reflect conformational differences between the two phosphoenzyme forms. Further experimentation will be needed to distinguish between these two possibilities. In this respect, it is worth noting that a low affinity is not necessarily expected as anintrinsic feature of the regulatory site. For example, AMP-PCP appears to bind to the unphosphorylated catalytic site and the regulatory site (which we consider to be the phosphorylated catalytic site) with equal affinity, i.e. in the 10-100 pbi range (18).This peculiarity may be related to thefact that thepresence of magnesium results in apronounced decrease in affinity of the unphosphorylated catalytic site for the analogue, contrary to what is observed with ATP (Ref. 23, Table I). In conclusion, we suggest that the biphasic ATP dependence of ATPase activity and ATP modulation of partial reactions of the cycle may beexplained by a single nucleotide binding site per ATPase, which changes its properties and function with the evolution of catalytic intermediates during a single cycle. In this model, intermediates whichdevelop following phosphorylation and ADP departure from the active site rebind either ATP or Mg. ATP at this locus with differing affinities, depending on the identity of the intermediate. Occupancy of this site in different ranges of ATP concentration and the attendent pertubations of catalytic eventscould result in acomplex dependence of enzyme turnover on ATP concentration in the absence of additional nucleotide binding sites and without subunit interactions of a putativeoligomer. Such a "single site-multiple function" mechanism of catalysis and regulation is also found for other enzyme systems in which a reaction product dissociates from the enzyme early in the catalytic cycle. For example, the deacylation of serine proteases is accelerated by substrate rebinding at the catalytic site (50, 51). Acknowledgments-Wewould like to thank Dr. M. Greenfor suggesting to search for ATP-induced changes in the spectrum of

bound FITC (Fig. lB),and Drs. U. Pick, J. Yon, and J. J. Lacapere for various discussions and P. Belle and M. Dreyfus for typing and editing an initial draft of the manuscript. REFERENCES 1. Weber, A., Herz, R., and Reiss, I. (1966)Biochem. Z.345,329-369 2. Inesi, G., Goodman, J. J., and Watanabe, S. (1967)J. Biol. Chem. 242, 4637-4643 3. Yamamoto, T., and Tonomura, Y. (1967)J.Biochem. (Tokyo)62,558-575 4. Neet, K. E., and Green, N. M. (1977)Arch. Biochem. Biophys. 178, 588597

5. Vezivski-Almeida, S., and Inesi, G. (1979)J. Biol. Chem. 264,18-21 6. Dupont, Y. (1977)Eur. J. Biochem. 72,185-190 7. Taylor, J. S., and Hattan, D.(1979)J. Bwl. Chem. 254,4402-4407 8. Scofano, H. M., Vieyra, A., and de Meis, L. (1979)J. Biol Chem. 264, 10227-10231 9. Guillain, F., Champeil, P., Lacapgre, J.-J., and Gingold. M. P. (1981)J. Biol. Chem. 266,6140-6147 10. Stahl, N., and Jencks, W. P. (1984)Biochemistry 23,5389-5392 11. Wakabayashi, S., Ogurusu, T., and Shigekawa, M. (1986)J. Biol. Chem. 261,9762-9769 12. Champeil, P., and Guillain, F. (1986)Biochemistry 26,7623-7633 13. McIntosh, D. B., and Boyer, P. D.(1983)Biochemistry 22,2867-2875 14. Shigekawa, M., Wakabayashi, S., and Nakamura, H. (1983)J. Bwl. Chem. 258,em-8707 15. Hobbs, A. S., Albers, R. W., Froehlich, J. P., and Heller, P. F. (1985)J. Biol Chem. 260.2035-2037 16. Coll, R. J., and Murphy, A. J. (1985)FEBS Lett. 187,131-134 17. Reynolds, J. A,, Johnson, E. A., and Tanford, C. (1985)Proe. NatL Acad. sei. u. S. A. 82,3658-3661 18. Cable, M. B., Feher, J. J., and Briggs, F. N. (1985)Biochemistry 24,5612""l"

a

19. Carvalho-Alves,P. C., Oliviera, C. R. G., and Vejovski-Almeida, S. (1985) J. Biol. Chem. 260,4282-4287 20. Dupont, Y.,Pougeois, R., RonJat, M., and Verjovski-Almeida, S. (1985)J. Bwl. Chem. 260,7241-7249 21. Msller, J. V.,Lind, K. E., and Andersen, J. P. (1980)J. Biol. Chem. 256, 1912-1920 22. Bishop, J. E., AI-Shawi, M. K., and Inesi, G. (1987)J. Biol. Chem. 262, 4fihX-4fifi3 .- - - ..-.

23. Ross, D., and McIntosh, D. B. (1987)J. Biol Chem. 262,12977-12983 24. Pick, U., and Karlish, S.J. D. (1980)Biochim. Biophys. Acta 626,255-261 25. Pick, U. (1981)Eur. J. Biochem. 121,187-195 26. Pick, U., and Bassilian, S. (1981)FEBS Lett. 123,127-130 27. Andersen, J. P.,M~ller,J. V., and J~rgensen, P.L. (1982)J . Biol. Chum. 257,8300-8307 28. Champeil, P., Guillain, F., Venien, F., and Gingold, M. (1985)Biochemistry 24,69-81 29. Berlman, I. B. (1973)J.Phys. Chem. 77,562-567 30. Riollet, S., and Champeil, P. (1987)Anal. Biochem. 162,160-162 31. Champeil, P., le Maire, M., Andersen, J. P., Guillain, F., Gingold, M.,Lund, S., and Mdler, J. V. (1986)J. Biol. 261.16372-16384 32. Mitchinson, C., Wilderspin, A. F., Tnnnlman, B. J., and Green, N. M. (1982)FEBS Lett. 146,87-92 33. Gafni, A., and Boyer, P. D. (1984)Biochemistry 23,4362-4367 34. Dux, L., Papp, S., and Martonosi, A. (1985)J. BioL Chem. 260, 1345413458 35. Ikehara, M., and Uesugi, S. (1969)Chum. Phnrm. Bull. 17,348-354 36. Glynn, I. M., and Chappel, J. B. (1964)Biochem. J. 90,147-149 37. LacapGre, J.-J., Gingold, M. P., Champeil, P., and Guillain, F. (1981)J. Biol Chem. 266,2302-2306 38. Dupont, Y., Bennett, B., and Lacapkre, J. J. (1982)Ann. N. Y. Acad. Sci. 402,569-572 39. Fernandez-Belda, F., Kurzmack, M., and Inesi, G. (1984)J. Biol. Chem. 269,9687-9698 40. Brand, C., Green, N. M., Konak, B., and MacLennan, D. H. (1986)Cell 44.592-607 41. Shigekawa, M., and Pearl, L. J. (1976)J. Biol. Chem. 251,6947-6952 42. Chaloub, R. M., and de Meis, L.(1980).J. Biol. Chep. 265,,6168-6172 43. Barrabin, H.,Scofano, H.M., and Inesi, G. (1984)Bmhemcstry 23,15421548 44. Nakamoto, R. K. and Inesi, G. (1984)J. BWL Chem. 269,2961-2970 45. Pick, U., and Bashian, S. (1983)Eur. J. Biochem. 131,393-399 46. Clore, G. M., Gronenborn, A. M., Mitchinson, C., and Green, N. M. (1982) Eur. J. Biochem. 128,113-117 47. Davidson, G. A., and Berman. M. C. (1987)J.Biol. Chem. 262,7041-7046 48. Garrahan, P. J., Rega, A. F., and Alonso, G. L. (1976)Biochim. Biophys. Acta 448,121-132 49. Andersen, J. P., and Msller, J. V. (1985)Biochim. Biophys. Acta 816. 915 50. Bechet, J. J., and Yon, J. (1964)Biochim. Biophys. Acta 89,117-126 51. Chevallier, J., and Yon, J. (1966)Biochim. Biophys. Acta 122,116-126

Chem.

~~