Catalytic and Regulatory ATP-binding Sites of the

Catalytic and Regulatory ATP-binding Sites of the Red Cell Ca2+Pump Studied by Irreversible Modification with Fluorescein Isothiocyanate* (Received for publication, June 11, 1982)

Shmuel Muallem and Steven J. D. Karlish From the Biochemistry Department, The WeizmannInstitute of Science, Rehovot 76100, Israel

bound with high affinityat cytoplasm-oriented sites and which site. is phosphorylated by ATP at a highaffinitybinding Calcium is probably bound to the protein in the form Ca. EIP which is a high energy phosphoenzyme capable of transferring the phosphate groupback to ADP. A spontaneous conformational transition leads to appearance of a low energy phosphoenzyme, E2P,which can be hydrolyzed by water and from which Ca’+ dissociates a low affinity site oriented to the exterior. A second conformational transition between forms EPand E, reorients theCa’+-binding sites toward the interior, completing the cycle. The most characteristic feature of the plasma membrane Ca2+ pumps is regulation by calmodulin (14). Combination with calmodulin greatly increases the turnover rate, and theCa’+-binding affinity to theEl form (15).In cahodulin-activated red cell membranes or in resealed red cell ghosts (but not in calmodulin-stripped membranes), we have observed that ATP activates (Ca’+ + Mg”)-ATPase activity with bothhigh (1-2 PM) and low (200-400 PM) apparent affinities (16, 17). The high affinity activation parallels that for phosphorylation.The low affinity activationis thought to involve stimulation by ATP of one or more of the subsequent stages (1, 2, 16-20). “Negative-cooperative’’ activation kinetics have also been observed with both (Na’ + K+)-ATPase (21, 22) and sarcoplasmic reticulum Ca”-ATPase (23). For both of these systems, particularly (Na’ + K+)ATPase, there is evidence that low affinity regulatory effects of ATP are accounted for by acceleration of conformational transitions between Ez and E, forms of the active transport proteins (24, 25). One objective of the present paper is to It is now clear that most mammalian cell membranes con- define the point of regulation by ATP for the red cell (Ca” tain a calmodulin-activated (Ca2+ + Mg’+)-ATPase, respon- + Mg’+)-ATPase. sible for active Caz+extrusion from the cell and maintenance For both (Na+ + K+)-ATPase (26, 27) and sarcoplasmic of the characteristicallylarge transmembrane gradients(3-8). reticulum Ca”-ATPase (28), evidence for a dimeric (or oliThe best studiedof these systems is undoubtedly the redcell gomeric) subunit structure and half-of-the-sites binding for membrane (Ca2+ Mg”)-ATPase (9). pump ligands, particularly ATP orcovalent phosphorylation, The reaction mechanism of the red cell (Ca’+ Mg‘+)- have led to formulationof half-of-the-sites or flip-flop models. ATPase is similar to that of other ATP-dependention pumps, Intrinsic or ligand-induced negative binding interactions for such as sarcoplasmic reticulum Ca-ATPase or (Na’ + K+)- ATP or phosphorylation on the opposing halves of the dimer ATPase, with participation of a covalent phosphoenzyme, and is suggested to force out-of-phase coupling of different stages in particular two major conformational formsof phosphoryl- of the reactionsequence. This could involve coupling of phosated and nonphosphorylated protein. Phosphorylation exper- phorylation with dephosphorylation, or any steps involving iments with red cell membranes and analogy with the other the action of high and low affinity ATP sites. Thus, energetiion-transporting ATPases have led to the following formulacally or kinetically favorable steps could drive less favorable ADP steps with consequent benefit forcontrol of the entireprocess. tion (1, 2, 10-13): ATP Cab + E l +&)Ca.EIPe For the red cell (Ca” + Mg‘+)-ATPase, structuraland P, ligand binding data are meager, but the negative-cooperative E2P + C&+E, El. El is theformto which Ca2+is ATP activationposes the same questionsas to thevalidity of the half-of-the-sites concept. A crucial requirement for such * The costs of publication of this article were defrayed in part by models is that both high and low affinity ATP-binding sites the payment of page charges. This article must therefore be hereby co-exist on the same pumpmolecule. marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The tool we have chosen to look at these problems in the

Catalytic and regulatorybinding sites for ATP onthe red cell Ca2’ pump have been investigated using fluorescein isothiocyanate (FITC).Both (Ca2+ + M&’)ATPase activity and ATP-dependent Ca2+flux are selectively and irreversibly inactivatedby FITC and the pump is protected from FITC by the presence of ATP. The time course of inactivation by FITC is characteristically biphasic. Analysis of the kinetics of inactivation by FITCand protection by ATPreveals the participation of both high and low affinity binding sites for ATP and FITC. The sites binding ATP or reacting with FITC do not, however, appear to co-exist on the same enzymemolecules. Thus, “flip-flop” mechanisms for (Ca2++ M&’)-ATPase, involving negative interactions between high and low affinity ATP sites, are considered unlikely. The two affinities for ATP are most simply explained by assuming that the Ca2+ pump protein exists in alternative conformational forms, El having a high affinity for ATP and Ez having a low affinity for ATP. Ca2+pumping and (CaZ++ Mg“)-ATPase involve interconversion between these forms. It is suggested that regulation of Ca2+pump activity by Mg-ATP reflects acceleration of the conformational transition between the El and E2 forms, as well as a previously described acceleration of phosphoenzyme hydrolysis (Muallem, S., and Karlish, S. J. D. (1981)Biochim. Biophys. Acta 647, 73-86; Garrahan, P. J., and Rega, A. F. (1978)Biochim. Biophys. Acta 513, 59-65).

+

+

+

169

Red Cu2+ Cell

170

Pump/FITC Conformational Changes

red cell membrane is FITC.' Recently, we have shown that FITC is a highly selective label for the ATP-bindingregion of both (Na' + K')-ATPase (29) and sarcoplasmic reticulum ea"-ATPase (30). Enzyme activity is inhibited and thefluorescence properties of the labeled proteins have allowed detailed study of the major conformational transitions. Similar modification of the ATP-binding site for red cell ATPase is to be expected. However, the paucity of (Ca" + Mg"+)-ATPase in the red cell membrane (perhaps 0.2% of the membrane protein) (31) effectively precludes selective fluorescence labeling of only this protein. We have therefore focused our attention on the kinetics of inactivation of(Ca" + MgL+)ATPase by the FITC. The FITC turns out to be a valuable tool for studying catalytic and regulatory ATP-bindingsites, because the experiments show clearly that these sites do not co-exist onthe (Ca" + Mg2')-ATPase protein,andthe method does not depend on estimates of ligand-binding stoichiometries. EXPERIMENTALPROCEDURES

Materials-FITC (isomer I) was obtained from Sigma, vanadatefree ATP was from Boehringer, and 45CaC1was from New England Nuclear. "'P, was purchased from the Israel Atomic Energy Center, Beer Sheba, and [y-"'PIATP was prepared as described before (32). Preparation of Membranes a n d Calmodulin-Human erythrocytes membranes stripped of their endogenous calmodulin and purified calmodulin were prepared according to our previous reported methods ( I , 33), respectively. One-step inside-outvesicles were prepared essentially as described before (34) with the following modifications. Human red blood cells were washed three times with a solution containing 130 mM NaCl, 10 mM KC1, and 10 mM Hepes-Tris (pH7.4 a t 0 "C) and were lysed in 40 volumes of an ice-cold medium containing 5 mM Hepes-Tris, 1 mM EGTA-Tris (pH7.4 a t 0 "C). The lysate was incubated for 30 min at 37 "C, afterwhich the vesicles were collectedby 10-min centrifugation at 45,000 X g (0-4 "C), resuspended into 2 ml with the lysing solution, and passed one time througha 27-gauge needle. The ion-tight vesicles were washed three times by 5-min centrifugation a t 45,000 X g with 50 volumes of a cold solution of 5 mM Hepes-Tris (pH 7.4 a t 37 "C), 10 mM KC1 and resuspended with the same solution to 5070 of the original packed cell volume. Assay-The FITC was dissolved in dry dimethyl formamide to the necessary concentration. Incubation of the red cell membranes and vesicles (equivalent to 30 p1 of original cell volume) was at 20 "C in a medium containing 10 m~ Hepes, 100 mM NaCl, 200 mM sucrose, 50 mM Tris, pH 9.0, in the presence of either 0.1 m~ EGTA or 0.025 mM CaC12. The FITC was added to thissuspension a t the required concentration, and after different times, the incubation with FITC was terminated by the additionof ATP toa final concentration of 3 mM and by the transfer of the tubes to an ice-cold water bath. Then, the (Ca" + Mg"')-ATPase activity of the membraneswas measured directly or after theremoval of the excess FITC by two washes with 50 volumes of a cold solution composed of 10 mM Hepes-Tris (pH 7.4 at 37 " c ) , 0.1 mM MgCI,. The (Ca" + Mg")-ATPase activity and the phosphoas well as themaximal level rylation and the dephosphorylation rates of the phosphoenzyme were estimated as previously described (1). For theCa influx measurements, 20 p1 of FITC-treated or untreated inside-out vesicles were suspended into 100 plof a medium containing 10 mM NaCl, 10 mM KCI, 3 mM MgCl?, 0.025 mM CaCL (containing 5 x cpm of 45Ca),25 mM Tris-HC1 (pH 7.4 a t 37 "C) with or without 1 mM ATP. The incubationswere initiated by the addition of vesicles into prewarmed reaction mixtures, and after 3 min, the vesicle SUSpensions were applied to Dowex 50W-X8 columns for the separation between intravesicular and extravesicular *"Ca.The description of the Dowex method for ion transport assays with vesicles was given in detail in Ref. 35. _

_

~

I The abbreviations used are:FITC, fluorescein isothiocyanate; AMP-PNP, 5'-adenylyl-P, y-imidodiphosphate; EGTA, ethylene glYCOI bis (a-aminoethyl ether)-N,N,N',N"tetraacetic acid; Hepes, 4-(2hydroxyethy1)-I-piperazineethanesulfonicacid.

RESULTS

+

Fig. 1 shows a time course of inactivation of (Ca2+ Mg")ATPase activity, following preincubation with FITC in an EGTA-containing medium. The inhibition is described by two exponentials with rate constantsk , = 1.36 min" and k2 = 0.14 min", respectively. ATP, 3 mM, in the preincubation medium gave essentially complete protection. The protectiveeffect is not exclusive for ATP since the analogue AMP-PNP is also effective (see below). The irreversible modification withFITC was done on membranes stripped of their native calmodulin (16), but the FITC inactivated equally well (Ca2++ MgL+)ATPase of these membranes or those after subsequent rebinding of pure calmodulin (16). Thus, the ATP sites which become modified are normally functional in both calmodulinfree or rebound membranes. The fall in initial rate of inactivation might be caused, in principle, by a gradual reduction in the effective FITC concentration, due either to spontaneoushydrolysis (which may be appreciable at the alkaline pH of the preincubation medium) or to a slow solubilization of FITC in the lipid domain of the membranes. We havemeasuredthedistribution of FITC between the medium and the membranes attwo incubation times andin the presence or absenceof ATP (Table I). Under the conditions of the experiment, almost 50% of the FITC is associated with the membranes but the distribution ratio was the same at 0.5 and 10 min and it was not affected by the presence of ATP. Other experiments showed that an increase in the membrane concentration produced a proportionate decrease in concentration of free FITC and rate of inactivation of (Ca" + Mg'+)-ATPase. These findings (as well as the protective effect of ATP) make it clear that inactivation of (Ca2+ Mg")-ATPase is caused by free and not by membrane-bound FITC, and the concentrationof the free FITC is constant during the preincubationperiod. The rate of inactivation of (Ca" + M$')-ATPase was greatlyincreased by raising the pH of thepreincubation mixture from7 to 9, as expected foran isothiocyanatereaction.

+

1 - A '

5

15 Tlme-min

FIG. 1. Time course of inhibition of the (Ca'+ + MgZ+)-ATPase activity by FITC. Calmodulin-stripped membranes were preincubated in the presence of 0.1 mM EGTA, 20 p~ FITC at pH 9.0 as described under "Experimental Procedures." After various times a t 20 "C, the labeling with FITC was terminated by a quick addition of ATP to a final concentration of 3 mM. When the protecting effect of ATP was tested, the ATPwas added toa final concentration of 3 mM before the addition of FITC. At the endof preincubations, the (Ca2+ + Mg'+)-ATPase activity of the membranes was measured either in the presence (0,0)or absence (0,W) of calmodulin as previously described (16).

Red Cell Ca" Pump/FITC Conformational Changes

171

Thus, preincubation with FITC was done routinely at pH 9 FITC concentration mustbe due, at least in part, to a differential occupancy of the sites. Since it was not possible to and then the membranes were transferred to the standard reaction mixture at pH 7.4 for estimation of enzyme activity. obtain k S at a saturating FITC concentration (see Table II), one cannotexclude the additionalpossibility that theprotein In view of the complex inactivation kinetics, itseemed side chains at the two sites show also differential reactivity important to ascertain whether FITC inactivates ATP-dependent Ca2' transport in the same manner as the (Ca'+ + toward FITC when it is bound. To characterize further the properties of the reactivesites, Mg2+)-ATPaseactivity. ATP-dependent Ca2+ uptake been has measured in inside-out vesicles (Fig. 2) and the pattern of we have looked at the ATP concentrations required to protect inactivation of the active Ca" transport is quite similar to against the FITC. Membraneswere incubated in the presence of EGTA, 20 p~ FITC, and different concentrations of ATP that of (Ca" + Mg2+)-ATPase seenin Fig. 1. Inactivation of (Ca2++ Mg")-ATPase has been studied a t for 1 or 10 min, and the remaining (Ca'+ + Mg"+)-ATPase different FITC concentrations; the dependence of the two rate activity was then measured (Fig. 3). A striking difference in requirement for ATP was found between membranes incuconstants k , and kl on FITC concentration is recordedin Table 11. The faster rate constant reaches a maximum (23.45 bated for 1 or 10 min. At the shortertime, a simple saturation . the rnin") a t FITC concentrations equal to or higher than 20 p ~ , curve was observed with an apparent K A r p = 4.6 p ~ For while the slower rate constant continues torise linearly up to longer preincubation period, ATPprotectedagainstFITC the 30 p ~It . is reasonable with two distinctlyseparate affinitiescalculatedfrom the highest FITC concentration studied, ~ ~ L Mand K Z A ~2p180 p ~ The . fraction of to assume that the FITC binds to the enzyme reversibly and curve as K ~ A--T8.9 quickly compared to the subsequent slow covalent modifica- ATP hydrolyzed during the10-min incubation never exceeded tion. The dependence of the inactivation rate on FITC con- 1%,thus excluding the possibility that thebiphasic protection curves result artifactually from a reduction of the ATP concentration should therefore reflect the bindingaffinityfor FITC and thisis clearly differentat the two sites, respectively. centration. It was of interest to look at effects of ATP on The observed difference in rate of inactivation a t a particular protection in the presence of Ca2+ions, but thiswas precluded by the fact that ATP is hydrolyzed in thepresence of the Ca2+ ions. However, protection against FITC by the nonhydrolyzed TABLE I Partitioning of FITC between membranes and medium Calmodulin-stripped membranes were incubated in thepresence of 0.1 mM EGTA, thedifferent FITC concentrations, and with or without 3 mM ATP. After 0.5 or 10 min, the mediumwas separated from the membranes by centrifugation a t 103,000 X g for 5 min,using the Beckman air-driven ultracentrifuge. The supernatantswere couected carefully and the absorption at 490 nm was measured and compared with that of a 2 PM standard FITC solution. Incubation time min

0.5

10.0

lFITCIT

ATP

[FITCIr,e, measured

(3 mM)

[FITClmemh,m,. calculated

PM

10.0 10.0 40.0 40.0 10.0 10.0 40.0 40.0

TABLE I1 The dependence of the rate constants of inhibition on the FITC concentration Time courses of inhibition of (Ca" + Mg")-ATPase activity by the different FITC concentrationsindicated were measured asin Fig. 1. From the obtained curves, k , and k p were calculated. Rate constants of inhibition min"

[FITCIr,,,

k,

kp

PM

-

5.30 5.32 21.80 21.32 5.30 5.32 21.00 21.18

+ + + +

4.70 4.68 18.20 18.68 4.70 4.68 19.00 18.82

2.5 0.029 5.0 0.058 7.5 10.0 0.153 20.0 0.460 0.714 30.0

I

0.17 0.38 I .04 1.66 3.31 3.45

0.081

I

I

I

250

500

750

I,/

I

1000 2703

I

ATP Concentmtlon - p M

IO

5

IO

Time - rnm

FIG. 2. Time course of inhibition of Ca2+influxby FITC. Preincubation of inside-out vesicles with 20 FM FITC was performed under the conditionsof Fig. 1, after which the Ca" influx capacity of the vesicles was estimated in the presence (0,O) or absence(0, ) . of calmodulin as described under "Experimental Procedures."

FIG. 3. Protection by ATP against inactivation by F€TC of (Ca2++ Mg2')-ATPase activity. The membranes were added to preincubation media similar to those of Fig. 1 but containing the different A T P concentrations indicated. The FITC was added to a final concentration of 20 PM and after either 1 (0)or 10 min (0), inactivation by FITC was stopped by addition of ATP to a final of the preincubations, concentration of 3 mM in each tube. At the end the (Ca" + Mg")-ATPase activity was measured. The curves have ~ for ATP werecalculated as been drawn by eye, and K O ,values described in Ref. 17.

Red Ca2+ Cell

172

Pump/FITC Conformational Changes

7'-

2

l

l

I

i

I

240

560 7;O AMP PNP Concentrotlon

Id00& 'k d

- pM

FIG. 4. Protection by AMP-PNP against inhibition of (Ca2+ Mg2+)-ATPase activity by FITC in the presence of Ca". Preincubations of membranes for 1 (0) or 10 min (0)with20 p~ FITC were done as described in Fig. 3 except that 0.025 mM Cai+ and the appropriate concentrations of AMP-PNP replacedthe EGTA and ATP. TheAMP-PNPconcentration in each tube at the end of preincubations was 3 mM. 10 ,dof membrane suspensions were taken from each tube for the measurement of the (Ca" + Mg'+)-ATPase activity in a reaction medium that contained finally 1.5mM ATP, 0.15 mM AMP-PNP.

+

[ I " . " 15 T i m e - mln

30

FIG. 5. The rate of inactivation of (Ca2+ + Mg'*)-GTPase activity by FITC in the presence of a low concentration of ATP. Treatment of membranes with 20 p~ FITC and experimental procedures were as in Fig. 1, except that 50 p~ ATP was included in the preincubation medium of one set of tubes. Control ( O ) ,+50 PM ATP (0). +2.7 mM ATP (A).

analogue AMP-PNP could be investigated in the presence of Ca'+ ions (see Fig. 4). AMP-PNPprotects,and a higher concentration of AMP-PNP was required to protecta t 10 min compared to 1 min of preincubation with FITC, although the curve at 10 min is not so well separated intotwo concentration ranges as observed forATP. AMP-PNP protects against FITC with lower apparent affinities than does ATP and other experiments showed that the requirement for AMP-PNP did not appear to be affected by the presence or absence of Ca" ions. The experiments of Figs. 1-4 and Table I1 point to the existence of both high and low affinity ATP-binding sites, the former having also the higheraffinity for FITC and being irreversibly modified at the faster rates.If one were to incubate membranes with FITC in the presence of ATP at a concentration low enough to protect only the high affinity FIG.6. ATP dependence of (Ca" + Mg")-ATPase in FITCsite, one might observe a simple exponential inactivation curve with a slow rate correspondingto kZ,measured in the absence treated or untreated membranes.The FITC-treated membranes were prepared by preincubation of membranes with 20 PM FITC for of ATP. Such an experiment, using 20 ,UM FITC and 50 ,UM 1 min, followed by two washes with 50 volumes of a cold solution ATP, is shown in Fig. 5. Of course, if the high affinity site containing 10 mM Hepes (pH 7.4 at 37 "Cj, 0.1 mM MgCI,. The were to be protected completely by the ATP, inactivationby activation by ATP of the (Ca" + Mg")-ATPase activity, in the and unFITC would be incomplete. Fifty ,UM ATP occupies a site of presence of 7.5 pg of calmodulin, of the FITC-treated (0) apparent affinity 4.6 ,UM to only about 90% saturation. Thus, treated (0)membranes was determined as in Ref. 16. the rate of reaction with FITC at this site should be reduced modified memby only some 10-fold (ie.to about the rate h,) and hence, as are essentiallyidentical, bothcontroland observed, all of the sites and (Ca')+ Mg")-ATPase activity branes displaying the characteristic high and low ATP actishould disappear with approximately the same slow rate con- vationregions (see Ref. 16). Another property of partially Mg")-ATPase that has been tested is stant. The experimentof Fig. 5 suggests that thehigh and low inactivated (Cas+ beaffinity sites for ATP are independently protected by ATP Ca"-dependent phosphorylationandtherelationship tween the remainingphosphoenzyme level and (Ca" + Mg2+)and are independently inactivated by FITC. The functional relationshipbetween the ATP-binding sites ATPase activity (Table 111). The 'lP incorporation is measobserved from the inactivationkinetics has been investigated ured a t 2 ,UM ATP and must thereforeoccur at a high affinity by looking at kinetic properties of enzyme activities remaining site for ATP. When measured in the presence of La"' ions, the maximal phosphoenzyme levels are estimated (1,36), this after partial inactivation with FITC. In a short incubation with FITC, only the high affinity ATP sites shouldbe modi- being a measure of the number of available sites. The results in Table I11 show that thereis an excellent correlation between fied, while the low affinity sites should remain intact. Fig. 6 shows ATP activation curves of calmodulin-stimu- the maxima1 phosphoenzyme level and (Ca" + Mg")-ATPase lated (Ca" + Mg")-ATPase activity of control membranes activity remaining after different periods of incubation with and membranes treatedwith 20 ,UM FITC for1 min. The (Ca'+ FITC. Furthermore, as seen in Fig. 7A, the rate constant for Mg")-ATPaseactivityremaining after this exposure to EP formation is the samefor control and partially inactivated FITC is 4040% of control activity, but otherwise the curves membranes kt,,,,= 0.09 s", although thelevel of E P is reduced

+

+

+

Red Cell Caz+Pump/FITCConformational Changes

173

rescein in the rapid phaseof inactivation do not subsequently

TABLEI11

not sustain Phosphoenzyme level and (Ca2++ Mg“)-ATPase activity of FITC- bind ATP,cannot be phosphorylated,anddo treated and untreatedmembranes (Ca” + Mg”)-ATPase activity. Calmodulin-stripped membranes were incubated for30 s or 10 min Fig. 8 shows that the rateof inactivation of (Ca2++ Mi”)with 20 p~ FITC as described in Fig. 1. The FITC-labeled membranes ATPase is not affected by the presence or absence of Ca2+ were washed twice with 50 volumes of a solution composed of 10 mM ions or Cas+plus calmodulin during the preincubationperiod. Hepes-Tris (pH 7.4), 0.1 mM MgC12 andresuspended in the same From previous experience ( l ) ,we might have expected Ca2+ solution to a protein concentration of 2 mg/ml. The (Ca” + Mg”)ATPaseactivityand the maximal level of phosphoenzymein the or Ca2++ calmodulin to stabilize an E , form and so the lack presence of 0.1 mM LaCI.