A remarkably stable phosphorylated form of Ca2+-ATPase prepared from Ca2+-loaded and fluorescein isothiocyanate-labeled sarcoplasmic reticulum vesicles.

After the nucleotide binding domain in sarcoplasmic reticulum Ca2+-ATPase has been derivatized with fluorescein isothiocyanate at Lys-515, ATPase phosphorylation in the presence of a calcium gradient, with Ca2+ on the lumenal side but without Ca2+ on the cytosolic side, results in the formation of a species that exhibits exceptionally low probe fluorescence (Pick, U. (1981) FEBS Lett. 123, 131-136). We show here that, as long as the free calcium concentration on the cytosolic side is kept in the nanomolar range, this low fluorescence species is remarkably stable, even when the calcium gradient is subsequently dissipated by ionophore. This species is a Ca2+-free phosphorylated species. The kinetics of Ca2+ binding to it indicates that its transport sites are exposed to the cytosolic side of the membrane and retain a high affinity for Ca2+. Thus, in the ATPase catalytic cycle, an intrinsically transient phosphorylated species with transport sites occupied but not yet occluded must also have been stabilized by fluorescein isothiocyanate (FITC), possibly mimicking ADP. The low fluorescence mainly results from a change in FITC absorption. The Ca2+-free low fluorescence FITC-ATPase species remains stable after addition of thapsigargin in the absence or presence of decavanadate, or after solubilization with dodecylmaltoside. The remarkable stability of this phosphoenzyme species and the changes in FITC spectroscopic properties are discussed in terms of a putative FITC-mediated link between the nucleotide binding domain and the phosphorylation domain in Ca2+-ATPase, and the possible formation of a transition state-like conformation with a compact cytosolic head. These findings might open a path toward structural characterization of a stable phosphorylated form of Ca2+-ATPase for the first time, and thus to further insights into the pump's mechanism.

The SERCA1a 1 Ca 2ϩ pump is a P-type membrane ATPase, whose catalytic cycle comprises several intrinsically transient auto-phosphorylated forms, the processing of which is tightly coupled to the binding or dissociation of calcium and hydrogen ions at distant transport sites. Twenty years ago, Pick and Karlish (1) showed that the use of fluorescein isothiocyanate (FITC) as a fluorescent covalent label of Ca 2ϩ -ATPase made it possible to monitor conformational changes of the protein. It was subsequently found that FITC specifically labels lysine 515 in the ATPase nucleotide binding domain (2). The fluorescence changes observed upon vanadate or Ca 2ϩ binding to FITC-ATPase were generally of relatively small amplitude, but they nevertheless have been widely exploited. Pick also described the formation under specific conditions of an FITC-ATPase species with an exceptionally low fluorescence (3). Surprisingly, however, these latter results were not, to our knowledge, much exploited or mentioned later.
We report here that this FITC-ATPase species of low fluorescence has very unusual functional and energetic characteristics; it is a remarkably stable phosphorylated species, with Ca 2ϩ binding sites empty, but oriented toward the cytosolic side and of high affinity. These results are discussed in relation to the putative mechanism for ion transport by Ca 2ϩ -ATPase and in relation to the environment of FITC in the ATPase nucleotide binding pocket. The stability of the low fluorescence FITC-ATPase species makes it a good candidate for helping the long-sought structural characterization of a phosphorylated form of Ca 2ϩ -ATPase, which would provide significant insight into the conformational flexibility of an ion transport ATPase during its catalytic cycle.

EXPERIMENTAL PROCEDURES
In most experiments, the medium contained 100 mM KCl, 5 mM Mg 2ϩ , and 50 mM MOPS-Tris at pH 7 and 20°C (buffer A). SR vesicles (prepared as in Ref. 4) were labeled with FITC (Sigma F 7250) as described previously (5). In most cases, this incubation was followed by pH neutralization, centrifugation, and resuspension at 20 mg/ml protein in buffer A to which 0.25 M sucrose had been added. Certain aliquots of resuspended labeled vesicles at 20 mg/ml protein were passively loaded by 1-2 h of equilibration with 5 mM Ca 2ϩ in buffer A without Mg 2ϩ , and frozen without sucrose. Control vesicles were treated similarly but without FITC.
The procedures used for ordinary fluorescence or stopped-flow fluorescence measurements, for 45 Ca 2ϩ binding measurements (either at equilibrium or during rapid filtration with Bio-Logic equipment), and for [ 32 P]EP measurements (either without acid quenching or after acid quenching), have already been described (4 -8). FITC fluorescence * 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. 1 The abbreviations used are: SERCA1a, sarcoplasmic or endoplasmic reticulum ATPase, type 1a; SR, sarcoplasmic reticulum; C 12 E 8 , octaethylene glycol monododecyl ether; DM, ␤-D-dodecyl maltoside; MOPS, 4-morpholinepropanesulfonic acid; A23187, calcimycin; P i , inorganic phosphate; FITC, fluorescein 5Ј-isothiocyanate; AcP, acetylphosphate; VO 4 , vanadate; TG, thapsigargin; E1P, E2P, names given to the postulated different conformations of phosphorylated ATPase.
(Spex fluorolog or PerkinElmer Life Sciences 650-40 instrument) was generally recorded with excitation and emission wavelengths of 495 and 520 nm, respectively (2-and 5-nm bandwidths), and plotted as percentage of the value in the presence of Ca 2ϩ without any correction for dilution effects (dilution was kept below 1.2% for every addition, except for that of 10 mM AcP, which resulted in 4% dilution). The stopped-flow experiments were performed with a Biologic SFM 3 stopped-flow instrument equipped with a short pathlength optical cell FC15), and data points were collected every 2 or 5 ms. The excitation wavelength was 460 nm, and the emission filter was a broad MTO 531 filter (Massy, France). For the experiment at the final free concentration of about 30 M, the low fluorescence species was first formed by addition of 6 mM EDTA instead of 2 mM EGTA, and then mixed with 2 mM Ca 2ϩ plus 4 mM Mg 2ϩ . For phosphorylation experiments performed in the presence of detergent, the phosphorylation reaction was quenched with 15 mM P i and 0.5 M perchloric acid (i.e. 4% v/v) instead of our usual 0.12 M perchloric acid, to precipitate the detergent-solubilized ATPase more easily.
Ca 2ϩ uptake into SR vesicles was measured in different ways, among which through changes in absorbance of the calcium-sensitive dye antipyrylazo III (Fluka no. 10795; e.g. Ref. 9). Absorbance and turbidity were measured either at a single wavelength or at multiple wavelengths with a diode array HP 8453 spectrophotometer, in a continuously stirred temperature-controlled cuvette. AcP hydrolysis was deduced from proton release, using a pH meter (PHM 62, Radiometer) whose analogic output was amplified (thanks to G. Lecointe in Saclay) to allow acquisition and digitalization.
Acetylphosphate (AcP) was from Sigma (catalog no. A-0262) and was freshly prepared at 250 mM. Orthovanadate solutions were prepared freshly as 100 mM alkaline (colorless) solutions at pH 12, by simple dissolution in water (Sigma S-6508). Decavanadate (yellow) solutions were prepared by titrating and diluting the former solutions first to pH 2 and then to pH 7 (and 50 mM), or preferably to pH 6 to minimize the presence of other forms of vanadate (10). The nonionic detergents C 12 E 8 and DM were obtained from Nikko and Calbiochem (or Anatrace), respectively. Free Ca 2ϩ concentrations were computed as described previously (8,11).

RESULTS
The Puzzling Properties of an Unusually Stable Low Fluorescence FITC-ATPase Species Formed from AcP in the Presence of a Ca 2ϩ Gradient-Pick (3) reported previously that, after AcP-dependent Ca 2ϩ uptake by FITC-labeled SR vesicles and an additional EGTA-induced rise in the Ca 2ϩ gradient across the membrane of these vesicles, an FITC-ATPase species with an unusually low fluorescence was formed. Trace A in Fig. 1 shows that this low fluorescence species can be very stable, much more in fact than initially reported by Pick. The poorer stability of the low fluorescence species in Pick's original report appears to be due to the fact that, in his experiment, EGTA addition did not result in a free Ca 2ϩ concentration as low as in the experiment illustrated in our trace A; trace B in Fig. 1 demonstrates this effect of a higher free Ca 2ϩ . The low fluorescence species was back-converted to a species of higher but intermediate fluorescence as soon as subsequent addition of Ca 2ϩ raised the external Ca 2ϩ , and the addition of ionophore immediately brought the fluorescence level up to about 100%. When the ionophore was added before AcP and EGTA, it completely prevented the appearance of any low fluorescence species (trace C in Fig. 1). Traces A--C therefore imply that large drops in the fluorescence level depend on the presence of a high calcium concentration on the lumenal side of the vesicles, as concluded previously by Pick. In agreement with this view, we found that the half-time for the AcP-induced initial slow drop in fluorescence (in the absence of any Ca 2ϩ -precipitating anion) was similar to that required for the accumulation of 45 Ca 2ϩ in FITC-labeled SR vesicles (20 -30 s, data not shown).
However, much to our surprise, we observed that when the ionophore was added to the cuvette after the formation of the low fluorescence species, the previous drop in FITC-ATPase fluorescence was not reversed, at least when the free Ca 2ϩ concentration on the cytosolic side of the ATPase (i.e. outside the vesicles) was very low (trace D in Fig. 1). Therefore, the existence of a Ca 2ϩ gradient is absolutely necessary for the formation of the low fluorescence species, but once this species is formed, it remains stable even if the gradient collapses. Trace E in Fig. 1 illustrates another puzzling feature of this low fluorescence species; although its formation is known (3) to require Mg 2ϩ , a cofactor of Ca 2ϩ -ATPase phosphorylation, once this species was formed, it no longer required Mg 2ϩ for stability. In fact, it was even more stable in the absence of Mg 2ϩ than in its presence, presumably partly due to EDTA-induced Ca 2ϩ chelation and clamping of the free Ca 2ϩ concentration at an extremely low value. The low fluorescence species also remained stable after complete detergent-induced solubilization of the vesicles (trace F), eliminating the possibility that the resistance to ionophore in panel D could be due to lack of permeabilization, and revealing an unusual stability for this detergent-solubilized ATPase in the presence of EGTA. The low fluorescence species remained stable, too, when thapsigargin was added to it (data not shown, 2 but see Fig. 2 below). Addi-2 Further details were submitted with the manuscript, but were FIG. 1. The low fluorescence FITC-ATPase species formed from AcP in the presence of a Ca 2؉ gradient has puzzling properties. FITC-labeled SR vesicles were suspended in 2 ml of buffer A, at 20 g/ml protein. First, as an internal control for each experiment, 40 M EGTA followed by 50 M Ca 2ϩ were added (upside-down triangles), resulting in the small well known Ca 2ϩ -dependent FITC fluorescence changes. In the experiment illustrated by trace A, 10 mM AcP was then added, followed by 2 mM EGTA (resulting in a low free Ca 2ϩ concentration of about 10 nM), 2 mM Ca 2ϩ , and finally 1 g/ml A23187. In the experiment illustrated by trace B, 150 M Ca 2ϩ was added before AcP (EGTA addition now resulted in a slightly higher free Ca 2ϩ concentration, about 40 nM). In the experiments illustrated by traces C and D, similar additions were made, but ionophore was added either before (C) or after (D) AcP and EGTA. In the experiment illustrated by trace E, 6 mM EDTA, followed by 6 mM Mg 2ϩ , were added after EGTA; in this case, 1 g/ml ionomycin was added at the end, instead of A23187 (iono). In the experiment illustrated by trace F, various aliquots of DM were sequentially added, resulting in final concentrations of 0.1, 0.2, 0.4, and 1 mg/ml; this experiment was performed in a medium containing 250 mM NaCl instead of 100 mM KCl, but qualitatively similar results were obtained in buffer A with C 12 E 8 . The first addition of DM was sufficient to reduce sample turbidity to a minimum. The starting time for each trace is arbitrary. tion of up to 10 mM ATP or ADP, either in the presence or absence of Mg 2ϩ , had no effect either on the low fluorescence species, including after its formation from partially labeled vesicles (data not shown).
The Similar Properties of the Low Fluorescence FITC-ATPase Species Formed from Inorganic Phosphate in the Presence of a Ca 2ϩ Gradient-Pick reported that a low fluorescence species could also be formed after adding both EGTA and P i (irrespective of the order) to FITC-labeled vesicles previously loaded passively with Ca 2ϩ , i.e. after phosphorylation from P i in the presence of a calcium gradient (3). We fully confirmed this observation ( Fig. 2A). The maximal amplitude of the observed signal was smaller than that obtained when experiments were performed with AcP and FITC-labeled vesicles. This smaller amplitude was presumably due to the fact that some of the vesicles had lost their impermeability during passive Ca 2ϩ loading (12) and/or freezing. The P i -dependence of the formation of the low fluorescence species revealed a relatively high apparent affinity for P i (half-amplitude was obtained for 0.2-0.3 mM P i , see traces A-D in Fig. 2), characteristic of gradientdependent phosphoenzyme formation (13). Note that orthovanadate (Fig. 2, trace E) was not able to replace P i and induced the same changes with Ca 2ϩ -loaded vesicles as those previously observed with nonloaded vesicles (14).
We found that the low fluorescence species formed from P i and Ca 2ϩ -loaded vesicles had puzzling properties, too. Although the formation from P i of this low fluorescence species was strictly dependent on lumenal Ca 2ϩ (as deduced from the fact that preliminary addition of ionophore completely prevented its appearance), once formed, this low fluorescence species was again resistant to the subsequent addition of ionophore (Fig. 2, traces F and G), or even to the addition of detergent at solubilizing concentrations, as already described for the species formed from AcP in Fig. 1 (data not shown). Similarly, although the formation from P i of this low fluorescence species was inhibited by thapsigargin, an inhibitor of Ca 2ϩ -dependent changes as well as of phosphorylation from P i (15), once formed, this species was resistant to the addition of thapsigargin (Fig. 2, traces H and I). Addition of orthovanadate to a preformed low fluorescence species brought the fluorescence back to a high level, at a concentration-dependent rate (data not shown), but decavanadate could bind to this species without destabilizing it; about the same relative quenching was observed when decavanadate was added to control FITC-ATPase (16) or to a low fluorescence TG-stabilized FITC-ATPase species formed from AcP (data not shown). 2 Traces J-N in Fig. 2 show how formation of the low fluorescence species was influenced by the final free Ca 2ϩ concentration in the medium. The low fluorescence species, once formed, was much more stable when the free Ca 2ϩ concentration dropped to a very low level than when it was only moderately low, as shown above when the low fluorescence species was formed after AcP-mediated calcium uptake (cf. traces A and B in Fig. 1). The pCa dependence of the amplitude of the P iinduced drop revealed a high affinity (pCa 1 ⁄2 was about 7, i.e. Ca 1 ⁄2 was submicromolar), which in fact was slightly higher than the overall affinity for Ca 2ϩ binding to SR ATPase under the same conditions (micromolar Ca 1 ⁄2; see Ref. 7).
Ca 2ϩ Is Not Bound to the Low Fluorescence FITC-ATPase, yet This ATPase Species Remains Phosphorylated and Phosphorylation Is More Stable than without FITC-Since the existence of a Ca 2ϩ gradient is required for the initial formation of a low fluorescence FITC-ATPase species, Pick initially concluded that this species would contain 2 or at least 1 bound Ca 2ϩ ion(s) (3). If this were the case, as this species, once formed, is resistant to the subsequent addition of ionophore or detergent, the putative bound Ca 2ϩ ion(s) should be occluded. We tested this possibility by loading FITC-labeled SR vesicles with 45 Ca 2ϩ , creating a low fluorescence species by adding EGTA, P i , and then ionophore, and measuring the amount of 45 Ca 2ϩ bound to the ATPase: under these conditions, however, this residual amount dropped to values smaller than 0.2 nmol/mg of protein, i.e. to values much smaller than the ATPase contents in SR vesicles, which typically is 5-7 nmol/mg ( Fig. 3A; see also Ref. 6). The results were similar when the Ca 2ϩ -loaded vesicles were first diluted in EGTA and P i was added afterward, or if partially labeled vesicles were used instead of fully labeled vesicles (data not shown). Therefore, Ca 2ϩ is not occluded in the low fluorescence FITC-ATPase species; in this ATPase species, the transport sites are not occupied by Ca 2ϩ at all.
Nevertheless, we found that the low fluorescence FITC-ATPase species was phosphorylated to a high level (Fig. 3C), in fact higher than that measured for unmodified ATPase under similar conditions (Fig. 3D). This phosphorylation level (close to 4 nmol/mg, i.e. a significant proportion of the above-mentioned ATPase contents) remained stable, whereas the level of the phosphoenzyme formed from unmodified vesicles slowly declined with time over minutes, presumably due to the dissi-withdrawn for the sake of conciseness, at the request of the Editor. They are available from the authors.
FIG. 2. The low fluorescence FITC-ATPase species formed from P i in the presence of a Ca 2؉ gradient has properties similar to those of the species formed from AcP. FITC-labeled SR vesicles that had been passively loaded with Ca 2ϩ were diluted to 20 g/ml in buffer A. As an internal control for each experiment, EGTA was initially added (first upside-down triangle; its concentration was 2 mM for traces A-I). In the experiment illustrated by traces A-D, P i was then added at various concentrations (as indicated), followed by 2 mM Ca 2ϩ (second upside-down triangle). In the experiment illustrated by trace E, 0.25 mM orthovanadate (VO 4 ) was added instead of P i . In the other experiments, P i was added at 10 mM; in those illustrated by traces F and G, additions of EGTA, Ca 2ϩ , P i , and ionophore (2 M ionomycin) were made in various orders. In the experiment illustrated by traces H and I, 1 g/ml TG, was added, either after EGTA and before P i (trace H), or after formation of the low fluorescence species, which here had been formed by adding EGTA after P i (trace I); in the presence of TG, addition of 2 mM Ca 2ϩ at the end (triangle) was no longer efficient. In the experiment illustrated by traces J-N, various concentrations of EGTA were initially added, resulting in calculated final free Ca 2ϩ concentrations corresponding to the pCa values indicated. pation of the Ca 2ϩ gradient. Just like the low fluorescence species, the phosphoenzyme formed from FITC-ATPase and [ 32 P]P i was resistant to ionophore-induced (or detergent-induced) collapse of the Ca 2ϩ gradient, the EDTA-induced removal of Mg 2ϩ , addition of thapsigargin (with or without decavanadate) and dilution with unlabeled P i , whereas it was completely abolished by a rise in the Ca 2ϩ concentration ( Fig.  3C and data not shown). 2 These properties are at variance with the conventional ones of gradient-dependent phosphoenzyme formed from unmodified vesicles incubated with [ 32 P]P i (Fig.  3D), as well as with the properties observed with either FITCmodified or unmodified vesicles previously made leaky to Ca 2ϩ (Fig. 3, E and F).
Ca 2ϩ Binds to the Phosphorylated Low Fluorescence FITC-ATPase Species from the Cytosolic Side, and Then Gets Transported into the SR Lumen-The fact that, in the experiment illustrated in Fig. 3C, the addition of cold P i had no effect on the [ 32 P]EP level of FITC-ATPase implies that the phosphorylated low fluorescence species is not in rapid equilibrium with unphosphorylated ATPase and P i in the medium. However, when the external Ca 2ϩ concentration was raised, the low fluores-cence FITC-ATPase species was back-converted to a species with higher fluorescence (Figs. 1 and 2). This implies that Ca 2ϩ interacts directly with the phosphorylated low fluorescence species. In view of the fast effect of Ca 2ϩ (Figs. 1, 2, and 3C) compared with its relatively slow permeation through the membrane (Fig. 3A), this also suggests that Ca 2ϩ binds to the phosphorylated low fluorescence species from the external, cytosolic side. This would be an unconventional conclusion, because phosphoenzyme formation is generally thought to be associated with the reorientation of Ca 2ϩ sites toward the lumen, or at least toward the interior of the membrane for ion occlusion (17,18). In the next experiments, we therefore studied in some detail the kinetics of the events associated with Ca 2ϩ binding to the low fluorescence species.
We initially investigated them by forming a low fluorescence The low fluorescence species, initially formed (at 80 g protein/ml) by 2.5-min incubation with 2 mM AcP, followed by Ca 2ϩ chelation, was mixed with various EGTA-or calcium-containing solutions (final pCa values are indicated). Panel C, kinetics of 45 Ca 2ϩ binding during rapid perfusion. FITC-ATPase was prepared either in its low fluorescence phosphorylated state (as shown in panels A and B, but now at 0.3 mg of protein/ml, triangles), or in its control nonphosphorylated but Ca 2ϩ -deprived state (circles); control unmodified ATPase was also prepared (squares). In all cases, 0.3 mg of protein was adsorbed onto a Millipore HA filter, manually rinsed for a few seconds with 100 M EGTA, and perfused with 50 M 45 Ca 2ϩ for various periods (see abscissa). Experiments were repeated in the absence of membranes for control (diamonds). Panel D, kinetics of Ca 2ϩ -induced dephosphorylation. 32 P-labeled phosphoenzyme was prepared as in . Panels E and F, same experiment as that illustrated in panels C and D, but with vesicles made leaky with ionomycin before phosphorylation (ionophore/protein was 1% w/w). Circles, control experiments. Squares in panel E, EDTA was added after 1.5 min. species of FITC-ATPase (by adding EGTA after AcP-supported Ca 2ϩ uptake) and then monitoring the rate at which subsequent addition of Ca 2ϩ reversed the previous drop in FITC fluorescence. In the range of submicromolar Ca 2ϩ concentrations, this reversal was slow enough to be monitored with a regular fluorometer, but it accelerated when the free Ca 2ϩ concentration was raised (Fig. 4A). For micromolar Ca 2ϩ concentrations, we had to use stopped-flow detection (Fig. 4B). We found that the rate constant of the fluorescence rise was about 10 s Ϫ1 at 30 M free Ca 2ϩ (pCa 4.5), and this was almost the maximal value (the rates measured at 300 M and 2 mM free Ca 2ϩ were 13 and 15 s Ϫ1 , respectively). This rate was of the same order of magnitude as that for Ca 2ϩ binding to dephosphorylated native ATPase in the absence of nucleotides under similar conditions (19,20). The Ca 2ϩ dependence of the stability of the low fluorescence species was qualitatively similar after collapsing the calcium gradient by either ionophore or detergent (data not shown).
We also directly measured, with rapid filtration equipment at a free Ca 2ϩ concentration of 50 M, the kinetics of 45 Ca 2ϩ binding to the low fluorescence phosphorylated species, as well as, in control experiments, the kinetics of 45 Ca 2ϩ binding to either nonphosphorylated FITC-labeled ATPase or unlabeled SR-ATPase. For short perfusion periods of up to 1 s, the binding patterns were essentially similar (Fig. 4C), and biphasic as found previously (21,22); together with the above stopped-flow fluorescence data, this confirms that the Ca 2ϩ binding sites on the low fluorescence species are as accessible from the external compartment of the vesicles as the binding sites on the control nonphosphorylated ATPase with or without FITC. However, for longer periods of 45 Ca 2ϩ perfusion onto the FITC-ATPase species of initially low fluorescence, the amount of 45 Ca 2ϩ associated with the vesicles appeared to slowly rise to levels higher than those required for saturation of the two high affinity binding sites (triangles in Fig. 4C; see below for interpretation).
In the final experiment of this series, we then measured the rate of Ca 2ϩ -induced dephosphorylation of the low fluorescence species formed from [ 32 P]P i and Ca 2ϩ -loaded FITC-labeled SR. This rate turned out to be slower than the rate at which the fluorescence rose and the rate at which Ca 2ϩ binding took place (open triangles in Fig. 4D, compare with panels B and C). The slow dephosphorylation was accelerated (nevertheless remaining slower than Ca 2ϩ binding) when the low fluorescence species was treated with ionophore before Ca 2ϩ was added (closed triangles in Fig. 4D), while the phosphoenzyme remained stable when no Ca 2ϩ was added, as expected (diamonds in Fig.  4D). Our interpretation is that addition of Ca 2ϩ to the Ca 2ϩfree low fluorescence phosphorylated species allows the ATPase to re-enter the catalytic cycle and permits ATPase dephosphorylation through the normal forward pathway, in which dephosphorylation after Ca 2ϩ internalization is slow compared with Ca 2ϩ binding, especially in the presence of lumenal Ca 2ϩ . In Ca 2ϩ -tight vesicles, the internalization of those 45 Ca 2ϩ ions that have initially interacted with the low fluorescence species combines with the passive rebinding of 45 Ca 2ϩ (resulting from continuous perfusion) to newly available dephosphorylated ATPase to explain why bound 45 Ca 2ϩ slowly rises to a final level higher than that required for saturation of the ATPase high affinity binding sites (as shown by the open triangles in Fig. 4C).
When the Rapid AcP-dependent Turnover of FITC-ATPase Results in Ca 2ϩ Depletion, a Low Fluorescence Species Forms Spontaneously; FITC Absorbance Also Changes-In addition, we found that the low fluorescence FITC-ATPase species can form spontaneously after AcP-dependent Ca 2ϩ pumping, before re-addition of Ca 2ϩ leads to another round of pumping. This was shown by the fact that in the presence of oxalate and a high enough concentration of vesicles, i.e. when these vesicles are capable of depleting the free Ca 2ϩ concentration in the medium down to submicromolar values, FITC-ATPase was to a large extent converted spontaneously into a low fluorescence species, after a lag corresponding to the completion of Ca 2ϩ withdrawal from the medium (thin top trace in Fig. 5A). Subsequent sequential Ca 2ϩ additions, which triggered renewed Ca 2ϩ uptake, drove the fluorescence level back toward a higher value until the added Ca 2ϩ was pumped into the vesicles again. Note that, in such experiments, the slight increase in turbidity concomitant with the lumenal precipitation of calcium oxalate (23, 24) may serve as a marker of the completion of Ca 2ϩ uptake, even in the absence of any Ca 2ϩ -sensitive dye (bottom thick trace in Fig. 5A). In parallel experiments, we confirmed that the rates of AcP hydrolysis and AcP-dependent Ca 2ϩ uptake after formation of the first microcrystals of calcium oxalate were only marginally slower for FITC-modified ATPase than for unmodified ATPase (in agreement with Ref. 5); coupling ratios were also unaltered by FITC modification. However, FITC-modified ATPase had become completely unable to hydrolyze ATP, as expected (data not shown). 2 Optical density recordings under the exact same conditions allowed us to conclude that changes in FITC fluorescence were in fact due, at least in part, to changes in FITC absorbance (thick top trace in Fig. 5A). Using a diode-array spectrophotometer, it was possible to reveal the changes in the entire FITC absorption spectrum, as they were large enough to show up on top of the light scattering by the vesicles (Fig. 5B, spectra 1-3). These changes were visualized even more easily after detergent-induced solubilization of the vesicles (spectra 4 and 5), which was shown previously (Fig. 1F) to leave the low fluorescence species stable until Ca 2ϩ was re-added. Similar changes FIG. 5. Spontaneous conversion of FITC-ATPase into a low fluorescence species, with a different absorption spectrum, after Ca 2؉ depletion in the presence of oxalate. Panel A, buffer A was supplemented with 8 mM oxalate, 40 M Ca 2ϩ , and 0.4 mg/ml SR vesicles that had just been incubated with FITC (2 mg/ml SR and 16 M FITC for 60 min). 10 mM AcP was added to trigger Ca 2ϩ uptake. 100 M Ca 2ϩ was subsequently added twice, followed by 1 mM EGTA. FITC fluorescence at this high protein concentration was recorded (thin bottom line, plotted after appropriate normalization). In a parallel measurement, optical densities at 495 nm (top thick trace) and 545 nm (bottom trace) were recorded. The 545-nm trace reflects changes in turbidity mainly due to (initially delayed) precipitation of Ca 2ϩ -oxalate. Panel B, absorption spectra recorded at various times during such an experiment. As indicated in panel A, spectra were recorded: 1, in the initial state; 2, after AcP-dependent withdrawal of Ca 2ϩ from the medium, and 3, after EGTA addition. At this point, 0.5 mg/ml C 12 E 8 was added, resulting in spectrum 4. Readdition of Ca 2ϩ to the now solubilized sample resulted in spectrum 5 (see Fig. 1F for the related fluorescence recovery).
in FITC absorbance were also seen after Ca 2ϩ depletion in the presence of 25 mM P i instead of oxalate, or in the absence of Ca 2ϩ -precipitating anions (data not shown). FITC is known to have an absorption spectrum (and not only a fluorescence spectrum) highly sensitive to protonation, polarity, or interactions (25,26). FITC spectral absorbance changes for phosphorylated FITC-ATPase after Ca 2ϩ depletion are qualitatively similar to those occurring for FITC in either an acidic or an apolar medium.
Absence of Special Interactions between Chains in the Low Fluorescence FITC-ATPase-We asked whether the unusual properties of the phosphorylated FITC-ATPase species were due to the appearance of unrecognized ATPase-ATPase interactions. The answer, however, was no: first, because similar spectroscopic properties were observed with ATPases that had been labeled with FITC only partially (1 nmol of FITC molecule bound/mg of protein instead of 5-7 nmol/mg) (data not shown); second, because measurements of the fluorescence polarization of bound FITC (an index of ATPase-ATPase proximity) showed that most of the homotransfer-induced depolarization of the bound FITC in FITC-labeled vesicles was lost upon solubilization at low Ca 2ϩ concentrations, whereas the low fluorescence species remained stable (data not shown); and third, because size exclusion chromatography experiments indicated that this detergent-solubilized low fluorescence FITC-ATPase species was still essentially monomeric (data not shown). DISCUSSION AcP-dependent Ca 2ϩ uptake followed by chelation of external Ca 2ϩ , as well as phosphorylation from P i in the presence of a Ca 2ϩ gradient, both permit accumulation of a phosphorylated FITC-ATPase species with unusual properties. The species has an extremely low fluorescein fluorescence, much lower than any other catalytic intermediate of the ATPase cycle. Once it has been formed, it no longer depends on the persistence of a Ca 2ϩ gradient or a membranous state. The phosphorylated and low fluorescence species has vacant, outwardly oriented, high affinity Ca 2ϩ binding sites. It is very stable, as long as the free Ca 2ϩ concentration is kept close to zero, but re-addition of Ca 2ϩ causes inward Ca 2ϩ transport followed by dephosphorylation. All these properties point to a novel species of Ca 2ϩ -ATPase, which appears to be of special interest both for the present functional description of the mechanism of ion transport and for the future structural studies of phosphorylated forms of the pump.
Relationship of the Low Fluorescence Species to the Usual Catalytic Intermediates-It was suggested that the Ca 2ϩ -ATPase catalytic scheme comprises four major enzyme intermediate species, namely phosphorylated or nonphosphorylated ATPase with or without bound Ca 2ϩ (Ref. 27; reviewed in Ref. 28). In terms of a simple four-species scheme, the Ca 2ϩ -free nonphosphorylated form of ATPase must expose its Ca 2ϩ -binding sites toward the cytosolic side of the SR, whereas the Ca 2ϩ -bound phosphorylated ATPase must expose its Ca 2ϩ binding sites toward the lumenal side (29). The same rationale is valid for ATP-supported or AcP-supported activity, since both catalytic cycles appear to be similar (30 -32). In Ca 2ϩaccumulating vesicles, phosphoenzyme hydrolysis slowing down by the high lumenal concentration Ca 2ϩ should thus lead to steady-state accumulation of the Ca 2ϩ -bound phosphoenzyme; if excess EGTA is then added, even more of this phosphoenzyme should form as a result of the reaction of P i (derived from AcP hydrolysis) with residual nonphosphorylated enzyme, by the reverse reaction. The same Ca 2ϩ -bound phosphoenzyme should also be formed from P i (again by the reverse reaction) in experiments performed with passively Ca 2ϩ -loaded vesicles like those illustrated in Fig. 2. However, in the context of such a four-species scheme, a phosphoenzyme species with outwardly oriented Ca 2ϩ binding sites cannot be generated upon addition of EGTA to Ca 2ϩ -loaded SR vesicles, and this phosphoenzyme cannot remain stable after disruption of the initial Ca 2ϩ gradient. Consideration of the more recent hypothesis that during transport, Ca 2ϩ ions move from a first, cytosolically oriented pair of sites to a second, lumenally oriented pair of sites (e.g. Ref. 33) does not help much, since in that alternative view the cytosolically oriented pair of sites is no longer accessible after ATPase phosphorylation (see Fig. 1 in Ref. 33). In addition, no evidence for Ca 2ϩ binding to any lumenal site was found in the atomic structure of Ca 2ϩ -ATPase derived from protein crystallized in the presence of 10 mM Ca 2ϩ (34).
Additional intermediate forms of ATPase have been suggested to exist within the ATPase catalytic cycle (35)(36)(37)(38)(39)(40). As a result, more elaborate schemes now explicitly include different forms for Ca 2ϩ -bound phosphorylated ATPase (permitting interconversion between the outside and inside orientations of occupied Ca 2ϩ sites), different forms for Ca 2ϩ -free nonphosphorylated ATPase (permitting interconversion between the outside and inside orientations of free Ca 2ϩ sites), and distinct enzyme forms permitting ion "occlusion" (either for Ca 2ϩ or for the counter-transported protons) (40,41); this is illustrated by the main cycle in Scheme 1. Of course, some of the various intermediate forms postulated by such schemes may be very transient during turnover. For instance, with unmodified vesicles, since dissociation of Ca 2ϩ ions to the outside of the vesicles was found to become impossible almost concomitantly with ATPase phosphorylation (17,18), the early phosphoenzyme form with outwardly oriented sites in Scheme 1, Ca 2 EP, must be present in only small amounts. A transient species, however, can conceivably be made more stable by manipulating experimental conditions or modifying the enzyme.
We have shown here that when FITC-modified vesicles are suddenly depleted of external calcium after AcP-dependent loading (Fig. 1), or when Ca 2ϩ -loaded FITC-labeled vesicles are phosphorylated from P i in the absence of cytosolic Ca 2ϩ (Fig. 2), a Ca 2ϩ -free phosphorylated form accumulates (Fig. 3), with outwardly oriented (cytosolically oriented) high affinity sites (Fig. 4). In the context of the above discussion about the ATPase catalytic scheme, the simplest explanation for these results is that, in FITC-labeled Ca 2ϩ -loaded vesicles, the early phosphorylated ATPase with outwardly oriented Ca 2ϩ sites, Ca 2  EP-FITC in Scheme 1, is stabilized and now constitutes a very significant fraction of total phosphoenzyme, from which Ca 2ϩ can dissociate toward the cytosolic side. Note that stabilization by FITC of the total level of EP formed from P i in the presence of lumenal Ca 2ϩ is not an interpretation, but a fact (see Fig. 3, C and D).
Stabilization of phosphorylated ATPase with Ca 2ϩ sites not SCHEME 1. Post-Albers-type model with occluded states and transported protons. yet occluded might be derived from changes in either the forward or the reverse rate of occlusion in Scheme 1. Since it is known that this occlusion reaction is not rate-limiting in the normal cycle, such FITC-dependent alterations would not greatly reduce the overall AcP-dependent turnover in leaky vesicles or in the presence of oxalate, as actually observed. However, in tight vesicles with high lumenal Ca 2ϩ , the Ca 2 EP-FITC species could accumulate, and at a low enough external free Ca 2ϩ concentration a phosphorylated species with unoccupied Ca 2ϩ sites facing the cytosol,  EP-FITC in Scheme 1, could finally be formed from the previous one. A scheme similar to Scheme 1 (except for omission of the occluded states) was in fact proposed previously as an alternative to an overly simple four-species scheme, to explain AcP-dependent 45 Ca 2ϩ -40 Ca 2ϩ exchange, an exchange that is slow but nevertheless measurable in normal SR (42). Note that, even though the early phosphoenzyme form might be very transient in the absence of FITC, it implies that phosphorylation actually precedes ion occlusion itself and does not occur simultaneously with it. Note also that the absence of Mg 2ϩ at the catalytic site of the "E1P" phosphoenzyme has also been shown to stabilize a phosphoenzyme form with open Ca 2ϩ sites, thereby permitting Ca 2ϩ release toward the cytosolic side (43). cedes ion occlusion, it is worth mentioning that previous experiments with a photoactivatable analog of ATP (TNP-8-azido-ATP) revealed that, when this analog was covalently tethered to the ATPase active site at Lys-492, most of its slow hydrolysis was uncoupled from Ca 2ϩ transport, again as if the transport sites had remained open to the cytosolic medium (44). This did not occur with the free TNP nucleotide, which exhibited tight coupling of calcium transport and phosphoenzyme hydrolysis. In the phosphorylated FITC-labeled ATPase, as in the phosphorylated ATPase with a tethered nucleotide analog, the nucleotide site remains permanently occupied, perhaps mimicking partly a state in which ADP has not yet dissociated itself from the unmodified ATPase (45). It might thus be speculated that, following phosphorylation, ADP dissociation from the phosphoenzyme plays a significant role for fast closure of the cytosolic gate of the ion transport site. With unmodified ATPase, from which ADP dissociates rapidly, the early phosphoenzyme form with transport sites still open toward the external side of the SR vesicles would be very transient, while bound FITC or the tethered analog might stabilize this form by mimicking to some extent bound ADP. To our knowledge, the impact of ADP dissociation on ion dissociation from other Ptype phosphorylated ATPases has not much been studied. The possibility we suggest should perhaps be kept in mind as an appealing speculation, although it is fair to say that it is not immediately reconciled with the fact that the tight binding of Cr.ATP to unphosphorylated Ca 2ϩ -ATPase favors Ca 2ϩ occlusion (39,46).

Implications for the Catalytic Cycle of Both FITC-Modified and Unmodified ATPase: A Role for ADP Dissociation in the
At this point, it is also worth saying a word about the relative fluorescence level of the two phosphoenzyme forms with outwardly oriented Ca 2ϩ sites, Ca 2 EP-FITC and  EP-FITC. The latter obviously has a low fluorescence level. The former, if it indeed accumulates at steady state in the presence of external Ca 2ϩ , must have intermediate fluorescence (Fig. 1). The difference between the low and intermediate levels of fluorescence (and absorbance, see Fig. 5) reflects rearrangement of the FITC environment upon Ca 2ϩ dissociation or binding, and illustrates the long distance coupling between the transport sites and the catalytic domain. Differences in fluorescence level have functional counterparts, since the Ca 2ϩ -free  EP-FITC species is unusually stable and hardly reacts with water, while rapid re-binding of Ca 2ϩ allows the enzyme to re-enter the cycle for normal handling of the phosphoenzyme (Fig. 4). The absolute requirement for Ca 2ϩ at the transport site for normal phosphoenzyme processing highlights the extraordinary degree to which partial reactions at the catalytic site can be coupled to changes at distant transport sites by long range transmission of information (47).
By Which Molecular Mechanism Does Fluorescein Stabilize the Phosphorylated Form of FITC-ATPase?-FITC reacts with lysine 515 at the high affinity nucleotide binding site (2,5,48), probably as an affinity label, mimicking nucleotides. Although under our conditions FITC derivatization increased 3-4-fold the amount of phosphoenzyme formed from 1 mM P i in the presence of a calcium gradient, and made it unusually stable (Fig. 3, C and D), FITC seems to have only slight effects on most individual steps of the cycle, including dephosphorylation and Ca 2ϩ binding (Refs. 5 and 7; see also Fig. 4B); Ca 2ϩ occlusion can also occur (although with a modified rate, as discussed above), since Ca 2ϩ is taken up efficiently in the presence of AcP. Thus, FITC does not sterically block any essential conformational change. For understanding the effect of FITC on the stability of the intermediate and low fluorescence species, Ca 2 EP-FITC and  EP-FITC, respectively, a clue might come from the spectroscopic changes experienced by the fluorescein moiety under these conditions.
The absorbance (and therefore fluorescence) of fluorescein is critically dependent on its protonation state and/or the hydrogen bonding power of the environment (25,26). Fluorescein has two ionizing groups, the xanthene phenolic group, 3-OH, with a pK a of about 6.7, and the benzoate carboxyl group, with a pK a of about 4.5. The dianionic species is strongly absorbant at 495 nm, whereas the monoanion is much less so. Therefore, one explanation for the unusually low fluorescence and absorption of the  EP-FITC species is an increase in the pK a of the phenolic 3-OH, which would stabilize the protonated form. This could arise either from an increase in hydrophobicity around this group or from the close approach of a negatively charged residue. Alternatively the fluorescence may be quenched by a salt linkage of the negatively charged 3-O Ϫ to a positively charged residue. Lys-515, to which the fluorescein moiety is covalently attached in FITC-modified Ca 2ϩ -ATPase, is located deep in the nucleotide binding pocket, while the rest of this pocket as well as the region surrounding the phosphorylatable aspartate contains a large number of charged residues (34). Since the ATPase turnover probably involves large relative movements of the various subdomains in the ATPase cytosolic head (34), we suggest that the large drop in FITC fluorescence and absorbance associated with formation of an early phosphoenzyme form might reflect the fact that, at this step, the FITC moiety is brought toward the walls of the cavity, which could easily result in a charged residue being very close to the 3-O Ϫ of the fluorescein. This could result in salt type bonding if a positive charge approaches. It is also possible that if a negative charge approaches and results in protonation (change in pK a ) for the fluorescein 3-OH, the hydroxyl group could now hydrogen-bond to neighboring residues. In both cases this could be viewed as a form of cross-linkage between subdomains. Of course, the carboxyl group of the fluorescein may also contribute interaction energy. The link resulting from these interactions could be of moderate strength in the Ca 2 EP-FITC intermediate, but reorganization of the cytosolic head after the dissociation of Ca 2ϩ could stabilize it further, result-ing in the unusually stable, low fluorescence, and Ca 2ϩ -free  EP-FITC species.
The Low Fluorescence Species: A Supercompact Form of the ATPase Cytosolic Head, Almost a Transition-like State? Potential Value for Future Structural Studies-Along this line, it is particularly appealing to further speculate that the abovementioned "cross-link" could occur between the nucleotide binding domain and the phosphorylation domain. We know that, at some stage during the catalytic cycle, the nucleotide domain and the phosphorylation domain must be able to come close together, to make phosphoryl transfer possible. The early phosphoenzyme form that is stabilized by FITC is as close as possible to that state. A further possibility that we may consider is that the low fluorescence is in fact caused by the interaction of the FITC moiety with the phosphoryl group itself. It could help to fix domain N and domain P (34) together and simultaneously alter the reactivity of the phosphoryl group. The low fluorescence species might therefore be a compact phosphoenzyme form, with the cytosolic head domains tightly associated.
Independently of the precise assignment of the residues interacting with FITC in the low fluorescence species, the structure of the low fluorescence species might mimic not only that of an early phosphoenzyme form, but also that of the transition state for phosphoryl transfer. Indeed, the phosphorylation event for unmodified ATPase can be broken into several substeps, as illustrated in Scheme 2. Phosphorylation is thought to be preceded by a rate-limiting nucleotide-induced conformational change to a species, Ca 2  a E⅐ATP, from which phosphorylation is very rapid (49 -51). In Scheme 2, the transition state for phosphoryl tranfer is placed in square brackets to indicate its transient existence. Jencks and co-workers (50,52) have found that Ca 2ϩ dissociates toward the cytosolic side of the membrane from both Ca 2 E⅐ATP and Ca 2  a E⅐ATP. The low fluorescence FITC-ATPase species, or more likely the Ca 2ϩ -bound species with intermediate fluorescence from which it is immediately derived, might to some extent mimic this transition state for phosphoryl transfer. If this idea is correct, the ATPase cytosolic head can again be expected to be in a very compact state, with restricted access of water or other phosphoryl acceptors to the active site.
At any rate, the low fluorescence phosphorylated form of FITC-ATPase can probably be safely classified as an "E1Plike" form, on the basis of its being derived from an early phosphoenzyme in the cycle and of its poor reactivity to water. The structure of its head region is likely to be substantially different from that of either the Ca 2 ⅐E1 or the E2⅐VO 4 states; the structure of the former state, deduced from Ca 2ϩ -ATPase three-dimensional crystals grown in the presence of Ca 2ϩ , shows a widely open head region, with the nucleotide binding domain and the phosphorylation domain separated by ϳ25 Å (34), whereas two-dimensional crystals of the E2⅐VO 4 species show a more closed head region in which full interaction of the nucleotide binding and phosphorylation domains is nevertheless hindered by decavanadate binding at the interface between domains (34,53). The structure of the low fluorescence species could provide critical new information on head domain interactions in a closed state and, by comparison with the other structures, on the extent and nature of domain movements.
In this direction, the fact that the low fluorescence species remained stable in a detergent-solubilized state (Fig. 1) might be of future value for three-dimensional crystallization attempts (e.g. as in Ref. 34 or 54). In addition, we have now found that it is possible to grow two-dimensional crystals from the low fluorescence and phosphorylated ATPase in the presence of thapsigargin and decavanadate (work in progress), 2 i.e. under conditions previously shown to induce the formation of twodimensional arrays of unphosphorylated ATPase (53,(55)(56)(57). The present demonstration of a stable, phosphorylated form of FITC-ATPase might thus provide a starting point toward the future crystallization and structural analysis for the first time of a phosphorylated form of the pump.