The Metal Sites on Sarcoplasmic Reticulum Membranes That Bind Lanthanide Ions with the Highest Affinity Are Not the ATPase Ca2+ Transport Sites*

We attempted to establish whether lanthanide ions, when added to sarcoplasmic reticulum (SR) membranes in the absence of nucleotide, compete with Ca" for binding to the transport sites of the Ca2+-ATPase in these membranes, or whether they bind to different sites. Equilibrium measurements of the effect of lanthanide ions on the intrinsic fluorescence of SR ATPase and on ''Ca2+ binding to it were performed either at neutral pH (pH 6.8), Le. when endogenous or contam- inating Ca2+ was sufficient to nearly saturate the ATPase transport sites, or at acid pH (pH 5.5), which greatly reduced the affinity of calcium for its sites on the ATPase. These measurements did reveal apparent competition between Ca" and the lanthanide ions La", Gd", Pr3+, and Tb3+, which all behaved similarly, but this competition displayed unexpected features: lanthanide ions displaced Ca" with a moderate affinity and in a noncooperative way, and the pH dependence of this displacement was


The Metal Sites on Sarcoplasmic Reticulum Membranes That Bind
Lanthanide Ions with the Highest Affinity Are Not the ATPase Ca2+ Transport Sites* (Received for publication, January 27, 1992) Fernando HenaoS, Stephane Orlowski, Zalika Merahs, and  We attempted to establish whether lanthanide ions, when added to sarcoplasmic reticulum (SR) membranes in the absence of nucleotide, compete with Ca" for binding to the transport sites of the Ca2+-ATPase in these membranes, or whether they bind to different sites. Equilibrium measurements of the effect of lanthanide ions on the intrinsic fluorescence of SR ATPase and on ''Ca2+ binding to it were performed either at neutral pH (pH 6.8), Le. when endogenous or contaminating Ca2+ was sufficient to nearly saturate the ATPase transport sites, or at acid pH (pH 5.5), which greatly reduced the affinity of calcium for its sites on the ATPase. These measurements did reveal apparent competition between Ca" and the lanthanide ions La", Gd", Pr3+, and Tb3+, which all behaved similarly, but this competition displayed unexpected features: lanthanide ions displaced Ca" with a moderate affinity and in a noncooperative way, and the pH dependence of this displacement was smaller than that of the Ca" binding to its own sites. Simultaneously, we directly measured the amount of Tb3+ bound to the ATPase relative to the amount of Ca2+ and found that Tb3+ ions only reduced significantly the amount of Ca" bound after a considerable number of Tb3+ ions had bound. Furthermore, when we tested the effect of Ca2+ on the amount of Tb3+ bound to the SR membranes, we found that the Tb3+ ions which bound at low Tb3+ concentrations were not displaced when Ca2+ was added at concentrations which saturated the Ca2+ transport sites. We conclude that the sites on SR ATPase to which lanthanide ions bind with the highest affinity are not the high affinity Ca2+ binding and transport sites. At higher concentrations, lanthanide ions did not appear to be able to replace Ca2+ ions and preserve the native structure of their binding pocket, as evaluated in rapid filtration measurements from the effect of moderate concentrations of lanthanide ions on the kinetics of Ca2+ dissociation. Thus, the presence of lanthanide ions slowed down the dissociation from its binding site of the first, superficially bound 4sCa2+ ion, instead of specifically preventing the dissociation of the deeply bound 46Ca2+ ion. These results highlight the need for caution when interpreting, in terms of calcium sites, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 5 Present address: Laboratoire de Biologie de 1'Ecole Centrale de Paris, 92295 Chitenay-Malabry Cedex, France.
ll To whom correspondence should be addressed. experimental data collected using lanthanide ions as spectroscopic probes on SR membrane ATPase.
Active transport of calcium from the cytoplasm of muscle cells into the lumen of sarcoplasmic reticulum (SR)' is mediated by membranous Ca2+-activated transport ATPase. The cloning and sequencing of this enzyme's cDNA allowed the definition of its amino acid sequence and, combined with low resolution structural data, the prediction of ita secondary structure and gross topography relative to the membrane plane (MacLennan et al., 1985). This ATPase was suggested to be composed of three main domains: a globular ATPbinding cytoplasmic domain, a pentahelical stalk, and a transmembrane domain. However, the high affinity Ca2+-binding sites whose saturation results in ATPase activation and ion transport have not been unambiguously localized on the predicted structure (Brandl et aZ., 1986;Clarke et aZ., 1989), and the available ATPase diffraction data obtained with 2D or 3D crystals do not yet provide sufficient resolution to allow such localization (e.g. Stokes and Green, 1990). In the meantime, one reasonable approach to this question is to try to replace Ca2+ with a metal ion with properties allowing distance measurements and triangulation. In this connection, lanthanide ions with magnetic or optical properties have already proved useful as calcium analogs for the study of various soluble calcium-binding proteins (Reuben, 1975;Martin and Richardson, 1979;Rhee et al., 1981;Evans, 1990;Petersheim, 1991). These ions, which have a relatively high atomic mass, are also potentially useful for anomalous x-ray scattering or other xray absorption spectroscopy techniques (Powers, 1982;Fairclough et al., 1986).
The effect of lanthanide ions on SR Ca2+-ATPase has been the subject of several publications during the last few years. Some time ago, these ions were shown to inhibit SR ATPase activity in the presence of ATP (e.g. Yamada and Tonomura, 1972). However, it now appears that this inhibition does not result from competition for occupancy of the calcium transport sites, but rather from the binding of Ln. ATP in place of Mg.ATP at the catalytic site (Fujimori and Jencks, 19901, or possibly also from the binding of lanthanide ions to sites distinct from both the transport sites and the catalytic site (Ogurusu et al., 1991). In the absence of ATP, lanthanide ions might nevertheless bind to the ATPase Ca2+ transport sites with high affinity. This was indeed suggested, and as a result, fluorescent or paramagnetic lanthanide ions were used to estimate the position of the two calcium-binding sites in the ATPase relative to the membrane surface or to other sites on the ATPase, including the catalytic site and individually labeled amino acids (Stephens and Grisham, 1979;Highsmith and Murphy, 1984;Scott, 1985;Herrmann et ul., 1986a;Teruel and G6mez-Fernhdez, 1986;Squier et al., 1987Squier et al., , 1990Jona et al., 1990; see also a discussion in Martonosi et al., 1990). Lanthanide ions have also been used to estimate the distance between the two Ca2+-binding sites (Scott, 1985;Herrmann et al., 1986a;Herrmann and Shamoo, 1988) as well as their mean hydration (Scott, 1984;Klemens et al., 1986;Gangola and Shamoo, 1987;Lockwich and Shamoo, 1990). Lastly, resonance x-ray diffraction studies have been performed to detect the bound lanthanide ions (Asturias and Blasie, 1991). The results of these last studies, as well as of some of the fluorescence transfer ones, suggested that the highest affinity lanthanide-binding sites were located in the stalk domain of the ATPase, where the Ca2+-binding sites had originally been predicted to reside (Brandl et al., 1986). However, in subsequent directed mutagenesis experiments, amino acids critical for calcium control of the catalytic events were discovered in the putative transmembrane region of the ATPase and not in the stalk region (Clarke et al., 1989;Green, 1989).
One critical issue is to determine clearly whether or not lanthanide ions bind to the Ca2+ transport sites. Several regions in SR membranes are obvious candidates as binding sites for trivalent ions: besides the high affinity Ca2+ binding and transport sites, these regions include at least the magnesium-binding site in the catalytic domain of the ATPase, the phospholipid headgroups in the membrane interface region, and the stalk domain in the ATPase, which contains up to 20 negatively charged amino acid residues. A difficulty when trying to design experiments clearly demonstrating true competition between Ca2+ and Ln3+ for binding to ATPase is that at neutral pH contaminating or endogenous Ca2+ is usually sufficient to saturate the ATPase Ca2+ transport sites. As no chelating agent is available which would selectively bind Ca2+ but not Ln3+, it is not a simple matter to study the binding of Ln3+ to Ca2+-free ATPase. T o overcome this difficulty, previous investigators used Ca2+-chelating resins for preliminary membrane and buffer depletion of Ca2+ (Chevallier and Butow, 1971;Jona and Martonosi, 1986;Itoh and Kawakita, 1984;Imamura and Kawakita, 1991a).
In the present work, we specifically addressed the question of whether or not lanthanide ions bind to the Ca2+ transport sites. T o solve the problem created by the saturation of the Ca2+ transport sites by contaminating Ca2+ a t neutral pH, we conducted some of the experiments at acidic pH (pH 5.5), i.e. under conditions in which the ATPase affinity for calcium was so greatly reduced (Verjovski-Almeida et al., 1977;Watanabe et ul., 1981;Guillain et ul., 1982) that contaminating or endogenous Ca2+ was no longer an obstacle. On the one hand, we performed equilibrium and kinetic measurements of the effect of lanthanide ions on 45Ca binding to the ATPase and on the changes in SR intrinsic fluorescence induced by Ca2+ (all the lanthanide ions tested behaved similarly), and on the other, we measured directly, by a fluorimetric assay using dipicolinic acid, the amount of T b 3 + bound to the ATPase and the effect of Ca2+ on this bound T b 3 + .
Our results show that although the trivalent lanthanide ions did reduce the amount of Ca2+ bound to the ATPase, they only did so with a moderate affinity and after a considerable number had been bound, corresponding to 4-10 times the stoichiometry of high affinity Ca2+ binding to the transport sites. Moreover, the Tb3+ ions which bound to the ATPase first, at low T b 3 + concentrations, were not displaced when Ca" bound to its own sites. In addition, when we studied the effect of lanthanide ions on the kinetics of dissociation of the two Ca" ions bound to the ATPase, we found that moderate concentrations of these lanthanide ions slowed down the dissociation of the first Ca2+ ion to leave its binding pocket, which is clear evidence for binding at a site different from the Ca2+-binding sites. We thus conclude that the sites on the ATPase to which lanthanide ions bind with the highest affinity are not the Ca2+ transport sites. Even at higher concentrations, Ln3+ did not behave like Ca2+ because the presence of Ln3+ in the dissociation medium did not block dissociation of the second Ca2+ ion. These results highlight the need for caution when interpreting experimental data concerning calcium site localization on the ATPase using lanthanide ions.

EXPERIMENTAL PROCEDURES
Sarcoplasmic reticulum vesicles were prepared from rabbit skeletal muscle, and the protein concentration was determined spectrophotometrically at 280 nm in the presence of 1% sodium dodecyl sulfate, as previously described (Champeil et al., 1985). Lanthanide ions were obtained as chlorides from Aldrich; EGTA and NTA were from Sigma. The dissociation constants for Ca2+-EGTA and Mp'f-EGTA were assumed to be 0.94 PM and 57 mM, respectively, at pH 6.8 and 380 PM and 0.97 M at pH 5.5. The dissociation contants for Ca2+-NTA and Me-NTA were assumed to be 400 PM and 4 mM, respectively, at pH 6.8 and 8.5 mM and 85 mM at pH 5.5. The dissociation constants for complexes of NTA with La3+, Pr3+, and Tb3+ at pH 6.8 were, respectively, assumed to be 52, 13 and 2.6 nM (Martell and Smith, 1974;Tsien and Pozzan, 1979).
Steady-state fluorescence measurements were performed at 20 "C in a continuously stirred cuvette with a Perkin-Elmer Cetus MPF 44A (or a SLM 4000s) fluorimeter, using excitation and emission wavelengths of 290 and 330 nm, respectively. Changes in fluorescence levels were corrected for dilution. Binding of 45Ca2+ was measured in filtration experiments as described by Champeil and Guillain (1986), using [3H]glucose as a marker of the amount of fluid wetting the filter. Unless otherwise indicated, 0.3 mg/ml protein SR vesicles were preincubated in a medium containing 1 mM [3H]glucose and either 40 or 100 p M "Ca2+ (in the form of a CaCIZ solution), 0.3 mg of protein was layered onto a Millipore HA filter, and the 3H and '5Ca radioactivities on the filter were counted by liquid scintillation. As described previously, 45Ca2' dissociation rates were measured at 20 "C with a Biologic rapid filtration apparatus in which the filters with adsorbed vesicles were perfused with the final dissociation medium for various electronically controlled periods (Orlowski and Champeil, 1991).
Throughout this work we used two main suspension media. The acid "pH 5.5 medium" contained 150 mM MES-Tris, no potassium, and, unless otherwise indicated, 20 mM Mp'f (20 "C, pH 5.5 & 0.1). The pH 6.8 or neutral pH medium contained 50 mM MOPS-KOH, 80 mM KC1, and 10 mM Mp'f (20 'C, pH 6.8 f 0.1). As regards contaminating Ca2+ in these media, it is known that the walls of glass vessels adsorb large amounts of Ca2+ which under certain conditions may be released back into the medium after prolonged storage. As we found that such release was particularly critical for the experiments performed at pH 5.5, in which we wanted as little contaminating Ca2+ as possible, the stock solutions for some of these experiments were stored in plastic vessels.
The lanthanide chloride salts were dissolved in 10 mM HCl and stored at 4 "C in plastic containers at different concentrations, all in the same 10 mM HCl medium (dos Remedios, 1981). The measured absorption properties of the lanthanide stock solutions were as expected from previously published extinction coefficients: for Pr3+, t444 = 10.4 M".cm"; for Gd3+, e212 = 3.4 M".cm"; for Tb3+, = 0.31 M" .cm", and CZM = 320 M".cm" (Carnall, 1979). In the case of T b 3 + , the accuracy of the concentration of the lanthanide stock solution was also confirmed by the fact that the fluorescence of its complex with DPA (dipicolinic acid, see below) was maximal at the expected stoichiometry of one T b 3 + for three DPA (Barela and Sherry, 1976). In the case of La3+, the concentration of its stock solution was established by observation of ita stoichiometric association with NTA or EGTA (see murexide experiments below, Fig. 48).
Binding of T b 3 + to SR membranes was measured by the difference between the total amount of T b 3 + in the membrane suspension and the amount of T b 3 + not bound to the membranes, using a protocol derived from a previously published one (Fairclough et al., 1986). To separate the membranes from the medium, however, we did not spin but simply filtered the suspension through a Millex GS filter. The Tb(DPA)3 fluorescence assay (Barela and Sherry, 1976) was used to quantify both the T b 3 + concentration in the filtrate and its total concentration. Aliquots of the various samples were sequentially added to a cuvette containing 200 pM DPA (the dilution was taken into account for subsequent calculations). Filtration of a control of Tb3+ binding to the filter itself.
membrane-free T b 3 + sample allowed correction for the small amount For example, for the experiment illustrated in Fig. 58, the initial T b 3 ' concentration was 10 p~, and 2 0 4 aliquots of the various samples were added to a 2-ml DPA-containing cuvette. In this particular trace, comparison of CF with CT showed that after passing 0.8 ml of a control membrane-free 10 pM T b 3 ' solution through the Millex, 8.6 p~ was recovered in the CF filtrate (see legend to Fig. 5A for definition of these samples, CT, CF, EF, and &); similar measurements, performed in duplicates for various Tb3+ concentrations, enabled us to establish a plot allowing Tb3+ binding to the Millex system to be estimated for all T b 3 + concentrations. Similarly, referring again to the trace in Fig. 58, comparison of EF with CT showed that after filtration through the Millex of a 10 p M T b 3 + solution containing SR membranes (here, 0.1 mg/ml protein was present), the T b 3 + concentration in the filtrate (EF) was 5 pM. From the above curve for T b 3 + binding to the Millex system itself, we then deduced that the free concentration of Tb3+ in equilibrium with the SR membranes before filtration was slightly higher, i.e. 6 p~. Therefore, we concluded that at a total Tb3+ concentration of 10 pM, the concentration of T b 3 + bound to the membranes was (10 -6) = 4 p~ (+0.4), corresponding to 40 + 4 of nmol Tb3+ bound/mg of SR protein. To compute this amount of bound Tb3+, the fluorescence observed upon addition of the ET aliquot was not explicitly taken into account, as it was always consistent with the one observed upon addition of the CT sample, although with a slightly reduced amplitude probably due to sample turbidity. Note, in addition, that the trace corresponding to addition of ET in Fig. 5 8 clearly shows that before reacting with DPA, a small fraction of the bound T b 3 + dissociated at a relatively slow rate from the SR membranes.
The Tb(DPA)3 assay was also used to ascertain whether a significant fraction of the added lanthanide ions bound to the vessel's walls, as such binding reduces the amount of Ln3+ interacting with the ATPase. By aliquoting a given solution of T b 3 + into different types of vial and then assaying the vial's contents for T b 3 + , we did find that in some cases, a small fraction of the added T b 3 + was lost, although for magnetic bar-containing spectrophotometer cuvettes and at our volume/surface ratio, this fraction was smaller than 25% for all the T b 3 + concentrations tested from 2 to 50 p~ Tb3+ (surprisingly, the Teflon-coated magnetic bar appeared to be the main target of this low affinity adsorption and not the glass or quartz walls of the spectrophotometer cuvettes). In contrast, there was no detectable Tb3+ binding to our disposable plastic tubes. Binding of T b 3 + to Millipore GS filters (see Fig. 5 and above) was also assayed in a similar way after passing 0.8 ml of a T b 3 + solution through the filter; the fraction of T b 3 + bound at 100 p~ T b 3 + was negligible and did not exceed 20% at 1 p~ Tb3'.
Finally, to ascertain what proportion of the lanthanide ions was really free in solution and available for reaction with the ATPase, rather than being complexed with other components of the solution (the sulfonate-containing buffers might have been candidates for such complexation), we ran two additional controls. First, in the case of La3+ and Tb3+, we evaluated the concentration of free lanthanide ions through their interaction with murexide, which has a poor affinity for these ions (Ohnishi, 1978). We found that the presence of 150 mM MOPS or MES did not perturb the interaction of La3+ or Tb3+ with murexide to a greater extent than that expected from the ionic strength increase. Second, in the case of Gd3+, we examined this metal's fluorescence and vibronic spectra, which reflect its interactions with the liganding water molecules (MacGregor, 1989). Again, 50 mM MOPS or MES did not modify these spectra excluding the possibility of significant complex formation between Gd3+ and these buffers'; as regards 50 mM MOPS, this result fits with the earlier finding that MOPS has no effect on the luminescence lifetime of EuS+ (Gangola and Shamoo, 1987).

Intrinsic Fluorescence Changes Induced by
Binding of Ca2+ or L a 3 + Ions to SR ATPase at Acid pH-At neutral pH, contaminating and endogenous Ca2+ (5-30 PM, depending on water quality, conditions of solution storage, protein concentration, etc.) is usually sufficient for virtual saturation of the high affinity SR ATPase transport sites. Lowering the medium pH greatly reduces the affinity of these sites for calcium (Watanabe et al., 1981;Guillain et al., 1982), so that at acid pH and in the absence of extra Ca2+, most sites can be expected to remain unoccupied. This expectation is confirmed by trace A of Fig. 1, which shows that at pH 5.5 k 0.1, addition of excess EGTA to SR vesicles only slightly reduced their intrinsic fluorescence level, known to be a reliable index of site occupancy by Ca2+, whereas addition of excess Ca2+ did ' This was not the case for PIPES which did quench Gd3+ fluorescence and vibronic side bands. Using the Tb(DPA)a assay, we also found that interaction of T b 3 + with high concentrations of PIPES resulted in the formation of a slowly dissociating complex with unusual binding properties to the cuvette's walls (not shown). This contrasts with the previous observation that the Gd3+ ESR spectrum did not change in the presence of PIPES (Stephens and Grisham, 1979). However, in agreement with this previous observation, the present measurements of Gd3+ fluorescence confirmed the existence of some interaction between Gd3+ and high concentrations of TES. raise this level. The poor affinity for Ca2+ of the sites under these conditions was further demonstrated by the fact that in the absence of EGTA direct addition of Ca2+ to SR vesicles, up to a total concentration of 100 PM, significantly raised their fluorescence but not to its maximal level (trace B, single and double arrows). Titration of the SR fluorescence level with various concentrations of added Ca2+ allowed us to draw fluorescence versus Ca2+ concentration3 curves, as shown in punel A of Fig. 2 (circles). The half-saturating Ca2+ concentration, [Ca2+ILh, was about 70-80 PM in the absence of M 8 + and 100-120 PM in the presence of 20 mM M e at pH 5.5. This is consistent with our previous measurements, in which the apparent affinity for free Ca2+ at pH 6 in the presence of 20 mM M$+ was 15-20 PM (Orlowski and Champeil, 1991). Therefore, at pH 5.5 and in the absence of added Ca2+, most On this curve, the abscissa takes into account both the amount of Ca2+ added to the cuvette and the estimated small amount of contaminating or endogenous Ca". In the absence of La3+, the agreement between the resulting plot and the one deduced from experiments in which the free Ca2+ concentration was precisely buffered by adding large amounts of Ca2+ to millimolar concentrations of EGTA was excellent (not shown) and in fact allowed us to refine our initial rough estimation of the amount of contaminating and endogenous Ca2+. transport sites in the ATPase remained Ca2+-free. Under these conditions, addition of 0.1 mM La3+ resulted in a small rise in SR fluorescence, as illustrated by truce C in Fig. 1.
When titrated as a function of the La3+ concentration, from micromolar concentrations up to 5 mM, this La3+-induced fluorescence rise provided no evidence for saturation, a result consistent with the binding of La3+ to many different sites (data not shown). Nevertheless, in this concentration range, the rise in fluorescence remained much smaller than the rise induced by maximal doses of Ca2+, and when 20 mM M e was present in the medium it was even weaker than in the absence of M e . Trace C in Fig. 1 also shows that when the ATPase reactivity to Ca2+ was tested in the presence of La3+ by sequential additions of Ca2+ (single and double arrows in trace C), the affinity of ATPase for Ca2+ was reduced (compare to truce B ) ; panels A and B in Fig. 2 illustrate this La3+dependent reduction in the ATPase apparent affinity for Ca2+.
Conversely, trace D in Fig. 1 shows that the successive addition of increasing amounts of La3+ to SR in the presence of a nearly saturating Ca2+ concentration gradually reduced SR fluorescence. Apparent competition between La3+ and Ca2+ for binding to ATPase sites was suggested by the facts that high Ca2+ concentrations counteracted the La3+-induced reduction in SR fluorescence and high La3+ concentrations apparently totally reversed the Ca2+-induced fluorescence change, as shown in panels C and D of Fig. 2. Note that despite the acid pH and the presence of La3+, no irreversible denaturation took place on the time scale of these experiments, since (i) this La3+-induced reduction in fluorescence was reversed upon addition of the lanthanide chelator NTA and (ii) subsequent addition of EGTA and Ca2+ still allowed monitoring of the previously described Ca2+-dependent fluorescence changes (see trace D in Fig. 1).
It should be stressed that although lanthanide ions are known to induce aggregation of membranes under certain conditions (Jona and Martonosi, 1986), no turbidity artifact was present in the above experiments. In fact, to our surprise, we found that whereas millimolar concentrations of La3+ or Tb3+ induced large changes in the turbidity of SR suspensions containing KCl, no such change was observed in the absence of KC1 (data not shown). Nevertheless, to confirm the intrinsic fluorescence results, we also performed direct measurements of the amount of Ca2+ bound to the ATPase high affinity sites, using 45Ca2+ and filtration methods.
Effect of Various Lanthanides on Equilibrium Binding of 45Ca2+ to the SR ATPase Transport Sites- Fig. 3 shows equilibrium 45Ca2+ binding measurements performed at pH 5.5 in the presence of 150 mM MES-Tris and 20 mM M F . Either 40 PM 45Ca2+ (triangles) or 100 PM (circles) was added to the assay, which, due to the poor affinity under these conditions, resulted in incomplete site saturation (see Fig. 2 4 , circles; two 45Ca2+ ions bound/ATPase chain correspond to 10-14 nmol/ mg protein, see Orlowski and Champeil, 1991). As illustrated by the closed symbols in panel A of Fig. 3, increasing concentrations of La3+ reduced to zero the amount of bound 45Ca2+. The same was true for all the lanthanide ions tested, i.e. La3+, Pr3+, Gd3+, and Tb3+, all of which had similar affinities (results for Pr3+ are shown in panel B, and those for Tb3+ are shown below in Fig. 6A). The concentrations required to chase half the initially bound 45Ca2+ were 30 and 65 FM La3+ in the presence of 40 and 100 PM added 45Ca2+, respectively, and were consistent with the [La3+Izh values deduced from the fluorescence experiments above (compare asterisks to circles in Fig. 2 0 ) .
To check that lanthanide ions were added in large excess relative to the number of their binding sites, the experiment described above in the presence of La3+ (panel A ) was also performed using an SR concentration three times lower than the one we used in the previous experiments. Results were similar (open versus closed symbols inpanel A), thus excluding the possibility that most of the La3+ ions would bind to membrane components under these conditions (see also below, Figs. 5 and 6). As previously observed on fluorescence titrations (Fig. 2C), there was no evidence for positive cooperative interaction between the La3+ ions that displaced 45Ca2+ from the transport sites. Different ionic media were also tested. The efficiency of La3+ was similar in our standard medium or in the presence of 100 mM KC1 and 5 mM M$+; in the absence of M$+, La3+ was slightly more efficient in chasing "Ca", although more 45Ca2+ was initially bound to the ATPase because of the slightly higher affinity of ATPase for Ca2+ in the absence of M e (data not shown).
We considered the possibility that the relatively low efficiency with which lanthanide ions displaced 45Ca2+ reflected the rather acid pH (5.5) of our experiments. We then moved back toward neutrality and for a large range of pH values measured the amount of 45Ca2+ bound to the ATPase after adding 40 p~ 45Ca2+ in the presence or absence of 100 p~ of the lanthanide ions tested. In the presence of Mg2+, this is shown in Fig. 3 as the inset to panel A for La3+ and as the inset to panel B for Pr3+, and similar results were obtained with Gd3+ and T b 3 + (not shown), In the absence of lanthanide (open histograms), the amount of bound 45Ca2+ increased with the pH up to the value corresponding to two Ca" ions/ ATPase, as expected from the increased affinity of 45Ca2+ for the ATPase. However, at neutral or alkaline pH, 100 PM of the La3+ or Pr3+ ions was proportionally less efficient than at acid pH in displacing 45Ca2+. At least for La3+, the same was true of all the media tested (e.g. circles in the inset to panel A).4 For comparison, we also tested the pH dependence of the efficiency of Sr2+ in displacing bound "Ca2+ since Sr2+, which binds with a low affinity but can be transported into the SR lumen, is known to be a true competitor of Ca2+ (Berman and King, 1990). In contrast with the results obtained with the lanthanides, Sr2+ proved to be equally efficient at all pH in displacing bound 45Ca2+ (not shown). The above results imply that the apparent ATPase affinity for the lanthanide ions which displace Ca" is enhanced by raising the pH to an extent smaller than the ATPase affinity for Ca2+.
What Is the True Efficiency of Free La3+ in Displacing Ca2+ Binding to SR ATPase ut Neutral pH? Effect of NTA-To exclude the possibility that the relatively low efficiency of the lanthanide ions in displacing bound 45Ca2+ resulted from a dramatic lowering of their free concentration due to their binding to nonspecific sites, especially under neutral or alkaline conditions, we assessed the effect of adding La3+ on 45Ca2+ binding both in the absence and presence of NTA, a strong chelator for lanthanides but a relatively poor one for Ca2+. These experiments were performed under a set of conditions commonly used for the study of SR ATPase, i.e. at pH 6.8 in the presence of KC1 and M e . As illustrated in panel A of Fig. 4, at 0.1 mM NTA (triangles) or 0.4 mM NTA (squares), addition of La3+ only reduced the amount of bound 45Ca2+ when the concentration of La3+ exceeded that of NTA (compare to the control curve in the absence of NTA, circles in panel A ) . As is clear from these results and illustrated in the inset to panel A, the efficiency with which La3+ displaced 45Ca2+ only depended on the difference [La3+Itotal - [NTAItot,l, and the true potency of free La3+ in our experiments was certainly not in the range of La3+ concentrations which can be precisely buffered with NTA (whose apparent dissociation constant at pH 6.8 in the presence of 10 mM M$+ is close to 180 nM). This contrasts with previously reported results (Squier et al., 1987(Squier et al., , 1990. We also considered the possibility that La3+ might bind to the sulfonic acid moiety of the buffer used, which would have reduced the free concentration of La3+ available for binding to the ATPase. When measuring the concentration of free La3+ with murexide, we found that this was not the case (see "Experimental Procedures"). Incidentally, a control experiment in this series, particularly relevant for our study, made it clear that there was no mistake in the evaluation of the relative concentrations of La3+ and NTA in our solutions because in the presence of NTA (or EGTA) murexide only detected the presence of La3+ when the concentration of the latter exceeded that of the chelator (Fig. 4, panel B, triangles  and squares, respectively).
Under exactly the same ionic conditions, we also explored, in the presence and absence of NTA, the effect of adding La3+ on SR intrinsic fluorescence in the presence of Ca2+ (data not shown). Here again, the presence of 0.4 mM NTA shifted by This was also the case in the absence of MgZ', but in that case analysis was less straightforward because, at neutral or at alkaline pH, 45Ca2f binding at the Ca2+ concentration used was not restricted to the high affinity transport sites. Effect of La3+ on 4aCaz+ binding at neutral pH in the absence or presence of NTA. Panel A, "Caz+ binding to the ATPase was measured at pH 6.8 and 20 "C in the presence of various amounts of La3+, in a medium containing 80 mM KC1, 10 mM M e , 50 WM total Ca2+ and 50 mM MOPS-KOH, in the absence of NTA (circles) or the presence of either 0.1 mM NTA (triangles) or 0.4 mM NTA (squares). W a Z + (40 WM had been added) and [3H]glucose were also present together with 0.3 mg/ml SR protein. The amour indicates the abscissa corresponding to 0.4 mM La3' . Zmet, the same data plotted as a function of the excess of La3' over NTA. Similar results were obtained when SR intrinsic fluorescence was measured in the presence of various amounts of La3+, without NTA or with 0.4 mM NTA (not shown). Panel B, control experiment, performed with murexide but without SR membranes, showing the change in murexide absorbance (OD4,5-OD,,0) as a function of the La3+ added in the absence of NTA (circles) or the presence of 0.4 mM NTA (squares) or EGTA (triangles). The medium contained 5 mM MOPS-Tris (pH 6.8) and 40 WM murexide. Arrow indicates the abscissa corresponding to 0.4 mM La3+.

Lanthanide
an equivalent amount the total concentration of La3+ required to reduce SR fluorescence to a given extent. In this medium containing KCl, it was difficult to explore, with intrinsic fluorescence, the effect of high La3+ concentrations because of drift and turbidity problems (see above). Nevertheless, the La3+-induced reduction in fluorescence remained smaller than the drop induced in the absence of La3+ by Ca2+ chelation, as previously observed at acid pH (see trace D in Fig. 1). A distinct feature of the La3+-induced drop was its relatively slow rate, compared to the very fast rate (on this time scale of minutes) of the transition induced by Ca2+ removal. This slowness can also be distinguished on trace D of Fig. 1, obtained at acid pH, and has already been observed under neutral conditions (cf. Fig. 3 in Girardet et al., 1989). These fluorescence measurements were repeated with Pr3+ in the presence or absence of NTA under the same conditions, and similar results were obtained (not shown).
Effect of Solubilization and Partial Delipidatwn-A series of fluorescence experiments was performed in the same neutral medium containing KC1 but in the presence of solubilizing and delipidating concentrations of the nonionic detergent CIZE~. At 5 mg/ml of this detergent, removal of Ca2+ from the ATPase high affinity sites resulted in irreversible ATPase inactivation on a time scale of minutes, concomitantly with a large gradual reduction in intrinsic fluorescence intensity (Andersen et al., 1982); the presence of Mg2+ afforded partial protection from this inactivation. We found a similar gradual reduction in fluorescence intensity when 0.3 mM La3+ was added to solubilized ATPase, and this too was tentatively attributed to irreversible ATPase inactivation. It was interesting to observe that the initial rate of this detergent-induced "inactivation" in the presence of La3+ was even faster than the rate noted in the presence of EGTA (data not shown), suggesting that La3+ did not simply act by removing Ca2+ from the ATPase high affinity sites but by binding to one or several other sites. Since the ATPase was largely delipidated at the concentration of detergent used (de Foresta et al., 1989), the site or sites concerned are probably located on the ATPase itself and not on the lipids.

Comparison between Tb3+ Binding to the SR Membranes and the Tb3+-induced Displacement of Bound 45Ca2+: Effect of Ca2+ on Bound
Tb3+-In order to make a final check of whether or not massive binding of lanthanide ions to the membrane components was responsible for the low apparent potency of lanthanides in displacing 'Ta2+ at neutral pH, and to correlate the lanthanide-induced displacement of 45Ca2+ with the binding of the lanthanide ion itself, we then also measured, under identical conditions, the amount of bound lanthanide ions. This was previously attempted by using radioactive 153Gd3+ (Krasnow, 1977;Squier et al., 1990) or fluorescent Tb3+ (Fairclough et al., 1986;Sprowl and Thomas, 1989). We selected Tb3+ as a prototype lanthanide ion; Tb3+ binding to the membrane was assayed by first passing the suspension through a Millipore filter and then measuring the resulting reduction of the Tb3+ concentration in the filtrate, using the fluorescence Tb(DPA)s assay to detect T b 3 + (Barela and Sherry, 1976). Panel A in Fig. 5 shows a diagram of our protocol, and panel B shows typical recordings. At moderate Tb3+ concentrations in the pH 6.8 medium containing KC1, the amount of Tb3+ bound to the SR membranes could be reliably measured. At higher concentrations, there was more uncertainty because the measured amount of bound Tb3+ resulted from a difference measurement (Tbgu,=,d = TbZal -Tb;;:) and because only a relatively small proportion of Tb3+ ions was bound to the membrane at these high concentrations. Nevertheless, panel C makes it perfectly clear that for Tb3+ concentrations which affected the binding of 45Ca2+ the binding of Tb3+ to the membrane only reduced the free Tb3+ concentrations moderately (as illustrated by the horizontal bar adjacent to the triangles in panel C of Fig. 5). A first conclusion, therefore, is that the apparent efficiency of lanthanide ions in displacing 45Ca2+ deduced from experiments in which only [Ln3+Itotal is plotted, as in Fig. 4 for La3+, is not far from the true value.
A second, even more important conclusion, is that when the amount of bound Tb3+ (triangles in Fig. 5, panel C, left  scale) was compared in quantitative terms to the amount of 45Ca2+ bound to the ATPase (crosses, right scale), the number of bound Tb3+ ions vastly exceeded the number of bound Ca2+ ions (compare triangles and squares, now plotted on the same (left) scale). In addition, under these conditions, overall Tb3+ biding to the membranes occurred with an affinity higher than the one with which Tb3+ displaced 45Ca2+. Note that both the Tb3+ and 45Ca2+ binding measurements were performed under exactly the same conditions, after adding 40 PM 45Ca2+, so that comparison of these affinities is perfectly valid.
Measurements of the amount of bound Tb3+ were repeated under acid conditions at pH 5.5, and the results were again compared to the Tb3+-induced displacement of bound 45Ca2+. The same picture emerged as at neutral pH, i.e. Tb3+ only reduced the amount of 45Ca2+ bound to the ATPase after a considerable number of Tb3+ ions had bound to the mem- containing 45Ca2+, pH 6.8 (40 p~ added 45Ca2+) was mixed with a given amount of Tb3+. 1 ml of the resulting Tb3+ solution was added to water, to serve as control ( C ) , and 4 ml was mixed with SR membranes to give an enzyme suspension ( E ) with a final SR protein concentration of 0.1 mg/ml. After 1 min of SR incubation with Tb3+, 3 ml was filtered onto an HA filter and the filter was counted to estimate bound 45Ca2+ ( E ) , and another 0.8 ml was filtered through a Millex system so that the filtrate (Ep) could be assayed for Tb3+ and compared to the total amount of Tb3+ (ET). 0.8 ml of the control sample run in the absence of membranes was also filtered through the Millex system to measure how much less Tb3+ was in the filtrate ( C p ) than the total initial amount of Tb3+ ( C T ) . In some cases, SR membranes were preequilibrated with 45Ca2+ before being incubated for 1 min with T b 3 + , but this made no difference to the amount of bound T b 3 + . Similarly, although we kept the incubation period constant for the various Tb3+ concentrations, we checked that incubation periods between 15 s and 15 min did not modify the amount of bound Tb3+. Panel E shows the actual assay, when CT, CF, EF, and ET fractions were assayed for Tb3+ by adding aliquots to a 2-ml fluorimeter cuvette containing 200 p~ DPA (see "Experimental Procedures" for analysis). Panel C shows the results: triangles, Tb3+-binding data (left scale); crosses, 45Caz+-binding data (right scale). Squares refer to the same 4sCa2+ data as the crosses but are plotted here on the same scale (left) as the Tb3+-binding data. Free T b 3 + , not total Tb3+, is plotted on the abscissa. Fig. 6). Moreover, under these acid conditions, it was possible to measure the amount of Tb3+ bound with or without simultaneous saturation of the transport sites by Ca2+, by simply adding Ca2+ at concentrations which, according to control fluorescence experiments similar to those shown in Fig. 1 for La3+, did saturate the transport sites despite the presence of T b 3 + in the medium. It is highly significant that in those experiments addition of Ca2+ did not measurably reduce the amount of bound Tb3+ (panel B ) . This agrees with similar results previously reported (Sprowl and Thomas, 1989).

branes (panel A in
From a theoretical point of view, in an equilibrium situation, one would expect that if the presence of T b 3 + reduces the affinity of ATPase for Ca2+, the presence of Ca2+ would also reduce its affinity for Tb3+, at least for the Tb3+ ions responsible for displacement of the bound Ca2+, leading to a Ca2+-induced reduction of the amount of bound Tb3+ ions observable under certain conditions. However, due to the relatively modest affinity with which Tb3+ (like other lanthanide ions) displaced Ca", this Ca2+-induced reduction of the amount of bound Tb3+ ions is probably only observable at relatively high Tb3+ concentrations, i.e. when the expected drop in bound Tb3+, which is likely to concern only one T b 3 + ion/ATPase molecule (see "Discussion"), is smaller than the error bar in our difference measurements (see Fig. 5 or 6). This probably explains why the reduction expected at high Tb3+ concentrations could not be resolved in our experiments. Conversely, the clear absence of any significant Ca2+-induced reduction of bound Tb3+ at low Tb3+ concentrations (Fig. 6 B ) shows that those Tb3+ ions which bind to the ATPase first, at low Tb3+ concentrations, are not displaced when Ca2+ binds to its own sites. Time-resolved Measurements of 45Ca2+ Dissociation in the Presence of Ln3+-The above results for T b 3 + binding and 45Ca2+ displacement exclude the possibility of competition between Ca2+ ions and the Tb3+ ions which bind at the sites of highest affinity. They show that the apparent competition between Tb3+ and Ca2+ evidenced by the equilibrium measurements reported in Figs. 1-4 is due to Tb3+ binding to another class of sites. Even these sites, however, are not necessarily the Ca2+ sites. As is well known, equilibrium measurements do not generally permit discrimination between the various molecular mechanisms which lead to apparent competition, only some of which are really based on true competition for a common site at the molecular level. Under favorable conditions, kinetic measurements do allow such discrimination. As we already pointed out, addition of La3+ to Ca2+-saturated ATPase only reduced its fluorescence level slowly (Fig. ID), so that kinetic effects could be suspected. We therefore attempted to characterize, by rapid filtration techniques, the effect of lanthanide ions on the ATPase high affinity Ca2+-binding sites, using an experimental protocol capable of detecting any true competition between Ca2+ and the lanthanide ions for binding at these sites. Accordingly, 45Ca2+-equilibrated SR membranes were layered onto a filter in a rapid filtration apparatus and perfused for electronically controlled periods with an appropriate dissociation buffer. In such experiments, the presence of a calcium analog in the perfusion medium is expected to slow down dramatically the rate of dissociation of the second, deeply buried 45Ca2+ ion, whose dissociation is prevented by the binding of the analog to the superficial subsite after the first 45Ca2+ ion has left it, just as the binding of 40Ca2+ would (Dupont, 1984;Petithory and Jencks, 1988;Orlowski and Champeil, 1991).
The first functional indication that lanthanides might bind to the ATPase at a location other than the Ca2+ transport sites came from the previous observation that the presence of 10 mM La3+ instead of 40Ca2+ in the perfusion medium considerably reduced the rate of dissociation of the first 45Ca2+ ion . We were able to reproduce this result here (not shown), but it might have been obtained because of the very high La3+ concentration used. We therefore repeated the experiment using lower La3+ concentrations (IO0 p~ or 1 mM), and we also included 40Ca2+ in the perfusion medium so that the conditions during perfusion corresponded to those previously found, at neutral pH, to allow partial displacement of 45ca2+ (100 p M La3+) or total displacement (1 mM La3+), in accordance with the equilibrium measurements made under exactly the same ionic conditions (Fig. 4A). Fig. 7 shows that 100 p~ La3+ (open triangles) already reduced the initial dissociation rate of the first, superficially bound 45Ca2+ by more than half (compare to circles) and that the effect of 1 mM La3+ (closed triangles) was still greater. Moderate concentrations of La3+ therefore did perturb the dissociation of the superficially bound 45Ca2+ ion. Similar results were obtained at pH 6, both with La3+ and Tb3+ (data not shown).
Even clearer results were obtained at pH 5.5 (Fig. 8). Here again, "Ca2+-equilibrated SR membranes were perfused with various media. Unlike dissociation in a Ca2+-free medium (squares in panel A), the presence of 100 p~ 40Ca2+ in the perfusion medium (open circles in panel A ) rendered the "Ca2+ dissociation kinetics clearly biphasic, although dissociation of the second 45Ca2+ ion was not completely prevented at this concentration, in agreement with the high cooperativity but the poor affinity of the ATPase for Ca2+ at this acid pH (which is responsible for partial site saturation at time zero, see also Fig. 4 in Orlowski and Champeil, 1991). In contrast, the presence of 100 p M La3+ instead of 40Ca2+ (open triangles in panel A ) did not slow down the dissociation rate of the second "Caz+ ion in this way. If La3+ had been bound at the superficial Ca2+ site, it would have exerted the same slowing down effect as 40Ca2+, at least at this pH since the apparent affinity with which La3+ chases Ca2+ at pH 5.5 is in fact higher than the apparent affinity of the ATPase for Ca2+ itself (about 10-30 verszm 50-100 pM, as deduced from extrapolation to zero Ca2+ and La3+ in Fig. 2, panels D and B, respectively).
However, this was not the case (compare triangles to circles). Furthermore, although adding 500 p M 40Ca2+ to the initial 100 p~ 40Ca2+ in the perfusion medium rendered the biphasic behavior even clearer, as expected (closed circles in panel B ) , adding 500 p~ La3+ instead of 40Ca2+ to the 100 p~ 40Ca2+ in the perfusion medium (closed triangles in panel B ) again only slowed down the dissociation of the first 45Ca2+ ion without specifically preventing the second from leaving its own subsite. Note that in equilibrium experiments under the latter conditions, 80% of the bound 45Ca2+ was displaced by La3+ ( Fig. 3A ) .

DISCUSSION
The experiments reported here were performed under both neutral and acid conditions. The latter conditions were designed to enable us to obtain easily an ATPase with Ca2+depleted binding sites and thus avoid preliminary treatment of the membranes with chelating resins. At pH 5.5 when no Ca2+ was added, the great pH dependence of the ATPase affinity for Ca2+ made it possible to add lanthanide ions to samples of ATPase with unoccupied transport sites. In preliminary fluorescence measurements, we found that lanthanide ions added to Ca2+-free ATPase slightly raised its intrinsic fluorescence (Fig. 1, trace C ) , an effect consistent with pre-vious observations by Jona and Martonosi (1986) and Imamura and Kawakita (1991a) but different from the absence of effect reported by Ogurusu et al. (1991), who removed endogenous Ca2+ from SR membranes but apparently did not eliminate contaminating Ca2+ from the buffers, On the time scale of minutes used here, our acid treatment did not reduce ATPase stability (see Fig. l), and the procedure was a convenient substitute for the more laborious technique of Ca2+ depletion with chelating resins (Itoh and Kawakita, 1984;Jona and Martonosi, 1986;Sprowl and Thomas, 1989;Imamura and Kawakita, 1991a). Under these conditions, we could then study the apparent competition between Ln3+ and Ca2+ ions through fluorescence (Fig. 2) and 45Ca2+ binding experiments (Fig. 3). The latter experiments showed that both Ca2+ ions were displaced from their binding sites by high concentrations of Ln3+ ions. However, the apparent competition between Ln3+ and Ca2+ ions had three puzzling features (Fig.  2): (i) La3+ ions reduced fluorescence without showing any sign of positive cooperativity between them (panel C in Fig.   2), whereas Ca2+-induced changes were clearly cooperative, as is well known (see circles in panel A); (ii) in panels B and D of Fig. 2, the plots of [Ca2+]% versw [La3+] and [La3+]% uersus [Ca"] were not straight lines, but their concavities were of opposite signs; this, combined with the observation (i) above, would be compatible with one La3+ ion being responsible for displacing the two Ca2+ ions from the ATPase-binding sites; (iii) positive cooperativity between Ca2+ ions was lost or reduced' in the presence of La3+ (compare squares to circles in panel A of Fig. 2); this observation is not easily reconciled with true competition for binding at a common site. In addition, when experiments were performed under various pH conditions, the pH dependence of the apparent competition between Ca2+ and lanthanide ions (see insets in Fig. 3) was different from that of the competition between Ca2+ and Sr2+, a true analog of Ca2+. Note also that La3+ and Ca2+ have different effects on ATPase intrinsic fluorescence (Fig. 1, C and D), whereas Sr2+ and Ca2+ have similar effects (Holguin, 1986 and data not shown), which is not in favor of La3+ being a good analog of Ca2+ for binding to the ATPase transport sites.
A distinct feature of our measurements both at acid and neutral pH was that Ln3+ displaced 45Ca2+ with a relatively modest affinity (see Figs. 3, 4A, 5C, and 6A). It has been previously suggested that the apparently poor efficiency of added lanthanide ions in displacing bound Ca2+ was misleading because it reflected the binding of these ions to membranes, vessels walls, or buffer components (Squier et al., 1990). In the presence of NTA, a chelator relatively specific for lanthanide ions, the free La3+ concentrations allowing displacement of half the bound Ca2+ at neutral pH were calculated to fall within the 15-100 nanomolar range as opposed to the 10-100 p~ range in terms of total concentrations (Squier et al., 1987(Squier et al., ,1990. We devoted much effort to checking this point but our results did not support this previous suggestion. As stated under "Experimental Procedures," we detected no binding of T b 3 + on the walls of plastic vessels in our experiments and no complex formation between La3+, Tb3+, or Gd3+ ions on the one hand and either MOPS or MES on the other. Although we did observe binding to glassware, quartz, Teflon, or cellulose esters, it was certainly not responsible for a large change in the free T b 3 + concentration in the corresponding experiments (in fact, binding was previously It is noteworthy that a change in the cooperativity of Ca2+ binding to SR membrane ATPase in the presence of Tb3' was also reported by Sprowl and Thomas (1989), although in the opposite direction. We do not know the reason for this discrepancy.
reported to take place on a time scale of days rather than, as here, of minutes; see Ellis and Morrison, 1975). In our experiments performed in the presence of NTA, the effects of lanthanide ions were only observed when their concentration exceeded that of NTA (this is an unexplained experimental discrepancy with the results reported by Squier et al. (1987,1990) and Girardet et al. (1989)), so that a submicromolar efficiency of lanthanide ions could not be confirmed. Lastly, Tb3+ binding to either the protein or the lipid constituents of SR membranes (see Herrmann et al., 1986b) did occur (Figs. 5 and 6) up to high levels, but these levels, which were similar to those previously reported for T b 3 + and Gd3+ (Krasnow, 1977; Sprowl and Thomas, 1989;Squier et al., 1990), were certainly not responsible for a drop by several orders of magnitude in the free Ln3+ concentration during the actual experiments.
We found that bound T b 3 + only reduced the amount of Ca2+ bound to the ATPase after a considerable number of Tb3' ions had been bound, corresponding to 4-10 times the stoichiometry of high affinity Ca2+ binding to the transport sites ( Figs. 5C and 6A). Moreover, we found that the T b 3 + ions which bound to the ATPase at low T b 3 + concentrations were not displaced when Ca2+ bound to its own sites (Fig. 6, panel  B ) . In a previous work, Sprowl and Thomas already suggested that measurement of the amount of lanthanide bound in the absence or presence of Ca2+ was one of the critical experiments to be done to describe the apparent competition between Ca2+ and Ln3+ for ATPase-binding sites. These authors, who first treated SR membranes and buffers with a Ca2+-chelating resin to reduce their Ca2+ contents, also made parallel measurements of bound 45Ca2+ and T b 3 + , but after equilibrium dialysis under neutral pH conditions (Sprowl and Thomas, 1989). In our laboratory, we found that an acid pH was suitable for depleting the ATPase transport sites of Ca2+, and we preferred filtration because it was a faster way of separating bound and free T b 3 + , thus avoiding possible problems due to lengthy dialysis in the absence of Ca2+. Both Sprowl and Thomas and ourselves found that addition of Ca2+ did not displace the T b 3 + ions bound at low T b 3 + concentrations. The different approaches by the two groups may have pitfalls, but different ones, so that the identical results they produced strengthen our conclusion that among the many sites on SR membranes to which T b 3 + ions bind, those to which these ions bind with the highest affinity are not the Ca2+ high affinity binding and transport sites. Since La3+, T b 3 + , Pr3+, and Gd3+ ions all behaved similarly in fluorescence and 45Ca2+ binding measurements (see "Results"), it seems reasonable to assume that this conclusion can be extended to all lanthanide ions. Indeed, a similar result has been very recently suggested for Gd3+ by Imamura and Kawakita (1991a) and Ogurusu et al. (1991), although in the absence of direct measurements of the amount of bound lanthanide. Note that it was previously shown that M&+, but not Ca2+, had the capability of displacing T b 3 + from a particular subset of their binding sites (Highsmith and Head, 1983).
Obviously, the above conclusion has dramatic implications for the significance of the various previous attempts (quoted in the Introduction) to exploit the favorable physical properties of some of the lanthanide ions to learn something about the ATPase Ca2+ transport sites. In those studies, in order to collect information about a reasonably homogeneous class of binding sites, most authors added limited amounts of lanthanide ions to fairly high concentrations of ATPase, SO that the binding stoichiometry was often below 2 Ln3+ ions/ATPase chain. Under these conditions, only those sites with the highest affinity for Ln3+ were labeled, and unfortunately they do not appear to be the Ca2+ transport sites (Figs. 5 and 6). Irrespective of the controversy already discussed about effective free lanthanide concentrations, a similar situation in fact prevailed in the experiments of Squier et al. (1990), although in that case the total concentration of added Ln3+ greatly exceeded the concentration of ATPase active sites (Squier et al., 1990): in these experiments, 14 nmol of Gd3+ ions/mg protein had to bind to the ATPase before less than 2 nmol of 45Ca2+ ions/mg protein could be displaced (pGd = 8 in Fig. 6 of Squier et al., 1990). Therefore, for instance, when Pr3+ was added to ANS-maleimide-labeled SR, it is likely or at least possible, that the energy transfer from covalently bound ANSmaleimide to Pr3+ only monitored the binding of Pr3+ at this lanthanide's high affinity binding site(s), distinct from the high affinity Ca2+ transport sites. In fact, this energy transfer responded to slightly lower concentrations of Pr3+ than the displacement of bound "Ca" (Fig. 8 in Squier et al., 1990).
At pH 6.8, SR membranes displayed a slightly higher overall affinity for T b 3 + than at pH 5.5 (cf. Figs. 5C and 6A). The above considerations about the binding sites with the highest affinity for Ln3+ are based on the assumption that classes of Ln3+ sites with different affinities exist on SR membranes, which is the case at neutral pH when T b 3 + displaced Ca2+ with an apparent affinity poorer than the one for overall T b 3 + binding (Fig. 5) and where Gd3+ enhanced the intrinsic fluorescence of SR ATPase at concentrations lower than those displacing Ca2+ (Imamura and Kawakita, 1991a). At acid pH, however, this is not necessarily the case, based on the virtually indistinguishable affinities for overall binding of T b 3 + and for 45Ca2+ displacement (Fig. 6A) and the monotonous reduction by lanthanide ions of the fluorescence of ATPase in the presence of Ca2+ (Figs. 1D and 2C). Nevertheless, as clearly shown by the data in Fig. 6B, even at pH 5.5 the lanthanide ions which displace Ca2+ are only a very small fraction of all the lanthanide ions bound at a given concentration ( Fig. 6A; we suspect these ions to be in a 1 to 2 stoichiometry with respect to the bound Ca2+ ions, see above). At this pH too, the contribution of these particular ions is therefore not easy t o distinguish when measuring the overall spectroscopic properties of all bound ions.
In addition to these measurements of Tb3+ binding per se, our rapid filtration measurements of the rate of 45Ca2+ dissociation (Figs. 7 and 8) showed that at moderate concentrations lanthanide ions bind to sites where they affect the dissociation of '5Ca2+ from the transport sites (and also its binding)? In other words, such rapid filtration measurements suggest that moderate lanthanide ion concentrations exert allosteric rather than direct competitive effects on Ca2+ binding, a conclusion we were able to reach thanks to the use of kinetic methods and not only equilibrium methods. In addition, after dissociation of the superficially bound 45Ca2+ ion, Ln3+ did not specifically prevent the deeply bound 45Ca2+ ion from leaving its own subsite (Fig. 8).7 After completion of our work, experiments along the same rationale were also reported by Ogurusu et al. (1991) with similar results. Our conclusion is that although these results do not strictly exclude that at high concentrations Ln3+ ions might eventually bind to the trans-S. Orlowski port sites after Ca2+ has left, they certainly imply that if this is the case, the binding pocket, under these conditions, no longer has the narrow profile which in the native conformation of Ln3+-free ATPase is responsible for the ordered sequential dissociation of Ca2+ toward the cytoplasmic side of the membrane (Petithory andJencks, 1988, Orlowski andChampeil, 1991).
The claim that the high affinity Ca2+-binding site does not bind Ln3+ would be unusual for a conventional EF-hand soluble protein (e.g. see Evans, 1990;Petersheim, 1991). However, SR Ca2+-ATPase does not belong to this class of Ca2+binding proteins, and it was recently demonstrated, for instance, for annexin V, which does not belong to this class either, that lanthanum binds to two sites which are not the sites to which Ca2+ binds (Huber et al., 1991). Long ago, Silber suggested that the binding properties of lanthanide and calcium ions were fundamentally different when the effective environment of the potential binding sites had a dielectric constant lower than that of water (Silber, 1974; see also Evans, 1990); this is a situation likely to occur in membrane ion transport proteins. As regards the localization of the high affinity Ln3+-binding site on the ATPase, our data obviously do not provide any positive information. Nevertheless, from the results published by other groups (e.g. Asturias and Blasie, 1991), the highly negatively charged ATPase "stalk" (Brandl et al., 1986) appears to be a likely high affinity binding site for lanthanide ions. A further argument confirming that this Ln3+-binding site on the stalk is probably not the Ca2+-binding site is that a recently expressed chimeric ion pump which consisted of the N-terminal two-thirds of the a-subunit of Na+,K+-ATPase and the C-terminal one-third of the Ca2+-ATPase, and was therefore devoid of the Ca2+-ATPase stalk sector, remained sensitive to Ca2+ (Luckie et al., 1991), suggesting, as previously suspected (Clarke et al., 1989;le Maire et al., 1990), that the C-terminal portion of the polypeptide chain comprised most of the residues critical for ATPase control by Ca2+.
To sum up, the results of the experiments reported in this paper are fairly bad news as regards the chances of being able to obtain definite information about the ATPase high affinity Ca2+ binding and transport sites from the use of low concentrations of lanthanide ions. At higher lanthanide concentrations, since mutual exclusion of Ca2+ and Ln3+ binding is at least partly due to allosteric effects, even very carefully done experiments will be difficult to interpret to derive significant information about these sites (e.g. see Imamura and Kawakita, 1991b).