Importance of transmembrane segment M1 of the sarcoplasmic reticulum Ca2+-ATPase in Ca2+ occlusion and phosphoenzyme processing.

The functional consequences of a series of point mutations in transmembrane segment M1 of sarcoplasmic reticulum Ca2+-ATPase were analyzed in steady-state and transient kinetic experiments examining the partial reaction steps involved in Ca2+ interaction and phosphoenzyme turnover. Arginine or leucine substitution of Glu51, Glu55, or Glu58, located in the N-terminal third of M1, did not affect these functions. Arginine or leucine substitution of Asp59, located right at the bend of M1 seen in the crystal structure of the thapsigargin-bound form, caused a 10-fold increase of the rate of Ca2+ dissociation toward the cytoplasmic side. Mutation of Leu60 to alanine or proline and of Val62 to alanine also enhanced Ca2+ dissociation, whereas an 11-fold reduction of the rate of Ca2+ dissociation was observed upon alanine substitution of Leu65, thus providing evidence for a relation of the middle part of M1 to a gating mechanism controlling the dissociation of occluded Ca2+ from its membranous binding sites. Moreover, phosphoenzyme processing was affected by some of the latter mutations, in particular leucine substitution of Asp59, and alanine substitution of Leu65 accelerated the transition to ADP-insensitive phosphoenzyme and blocked its dephosphorylation, thus demonstrating that this part of M1, besides being important in Ca2+ interaction, furthermore, is a critical element in the long range signaling between the transmembrane domain and the cytoplasmic catalytic site.

The functional consequences of a series of point mutations in transmembrane segment M1 of sarcoplasmic reticulum Ca 2؉ -ATPase were analyzed in steady-state and transient kinetic experiments examining the partial reaction steps involved in Ca 2؉ interaction and phosphoenzyme turnover. Arginine or leucine substitution of Glu 51 , Glu 55 , or Glu 58 , located in the N-terminal third of M1, did not affect these functions. Arginine or leucine substitution of Asp 59 , located right at the bend of M1 seen in the crystal structure of the thapsigargin-bound form, caused a 10-fold increase of the rate of Ca 2؉ dissociation toward the cytoplasmic side. Mutation of Leu 60 to alanine or proline and of Val 62 to alanine also enhanced Ca 2؉ dissociation, whereas an 11-fold reduction of the rate of Ca 2؉ dissociation was observed upon alanine substitution of Leu 65 , thus providing evidence for a relation of the middle part of M1 to a gating mechanism controlling the dissociation of occluded Ca 2؉ from its membranous binding sites. Moreover, phosphoenzyme processing was affected by some of the latter mutations, in particular leucine substitution of Asp 59 , and alanine substitution of Leu 65 accelerated the transition to ADP-insensitive phosphoenzyme and blocked its dephosphorylation, thus demonstrating that this part of M1, besides being important in Ca 2؉ interaction, furthermore, is a critical element in the long range signaling between the transmembrane domain and the cytoplasmic catalytic site.
The Ca 2ϩ -ATPase 1 of sarcoplasmic reticulum is a membrane-bound enzyme that pumps Ca 2ϩ from the cytosol to the lumen of the sarcoplasmic reticulum at the expense of chemical energy being released by ATP hydrolysis (1)(2)(3)(4). It belongs to the family of P-type ATPases characterized by the formation during the enzymatic cycle of a phosphorylated enzyme intermediate, in which the ␥-phosphoryl group of ATP has been transferred to a conserved aspartic acid side chain in the enzyme. The phosphoenzyme intermediate is fundamental to the coupling between the spatially well separated processes of ATP hydrolysis in the cytoplasmic part of the molecule and vectorial ion transport across the membrane. Phosphorylation of the enzyme by ATP is activated by Ca 2ϩ binding at cytoplasmically facing high affinity sites of the E 1 form. During the subsequent conformational change, [Ca 2 ]E 1 P 3 Ca 2 E 2 P, the Ca 2ϩ binding sites are reoriented toward the lumen and their affinity markedly lowered, resulting in release of the bound ions (Scheme 1). It is believed that, following their binding at the high affinity sites, the Ca 2ϩ ions are contained within an "occluded" state (here indicated by brackets around the Ca 2ϩ ions) with no access to the medium on either side of the membrane (5)(6)(7).
The overall tertiary structure of the Ca 2ϩ -ATPase comprises 10 transmembrane ␣-helices (M1-M10) and a large cytoplasmic headpiece divided into three distinct subdomains denoted "actuator" (A), "nucleotide-binding" (N), and "phosphorylation" (P) domain (8,9). The two Ca 2ϩ binding sites, located in the membranous part of the molecule more than 40 Å away from the cytoplasmic catalytic site, are formed by amino acid residues in transmembrane segments M4 (Val 304 , Ala 305 , Ile 307 , and Glu 309 ), M5 (Asn 768 and Glu 771 ), M6 (Asn 796 , Thr 799 , and Asp 800 ), and M8 (Glu 908 ) (8, 10 -12). Although the nature of the intramolecular signaling between the distantly located catalytic site and the Ca 2ϩ binding sites in the membrane is yet incompletely understood, the propagation of conformational changes seems to be part of this long range communication. The crystal structure has been determined for two forms of the Ca 2ϩ -ATPase, a Ca 2ϩ -bound form presumably corresponding to [Ca 2 ]E 1 (8) and a Ca 2ϩ -free form with thapsigargin bound, believed to resemble E 2 ("E 2 -Tg") (9). Comparison of the two structures reveals major conformational differences. In particular, the A-, N-, and P-domains move from a spatially well separated organization in the Ca 2ϩ -bound form to a more compact entity in E 2 -Tg (9). Interestingly, it appears that this rearrangement of the cytoplasmic domains is accompanied by a partial unfolding and bending (at Asp 59 ) of the helical structure of transmembrane segment M1, to which the A-domain is physically connected through a flexible hinge. The bending may be aided by the amphipathic nature of the most N-terminal part of M1, allowing one side of the bent part to remain associated with membrane lipid while the other side comes into contact with the cytoplasmic phase (9). An important challenge is to reveal whether these rearrangements of the cytoplasmic domains and M1 are coupled and occur in the native state during the transport cycle and to understand the functional implications. Although it was recently reported that the length and, hence, the flexibility of the hinge linking M1 and the A-domain is critical to phosphoenzyme processing (i.e. Reactions 4 -6 in Scheme 1) (13), the functional importance of M1 has so far not been characterized in any detail. A most striking aspect of M1 of the SERCA family of Ca 2ϩ -ATPases is the presence of four negatively charged amino acid residues (Glu 51 , Glu 55 , Glu 58 , and Asp 59 ), which have been suggested to provide the cytoplasmic entry pathway to the membranous Ca 2ϩ sites (14). This proposal was mainly based on the crystal structure of the Ca 2ϩ -ATPase in the Ca 2ϩ -bound form (8), but, since then, the determination of the crystal structure of the Ca 2ϩ -ATPase in the E 2 -Tg form has attracted further attention to M1 by demonstrating that the above-described bending of M1 opens a water-accessible channel leading between M1 and M3 to Glu 309 at Ca 2ϩ site II, a channel with the potential of a Ca 2ϩ entry port (9). Of Glu 51 , Glu 55 , Glu 58 , and Asp 59 in M1, the first three contribute to define the polar/charged surface of the amphipathic part near its N terminus, whereas Asp 59 is located right at the bend formed in the E 2 form. Furthermore, in the Ca 2ϩ -bound crystal structure, Glu 58 appears to be hydrogenbonded to Glu 309 at Ca 2ϩ binding site II, making a direct conformational effect on M1 possible upon Ca 2ϩ occupancy of the binding sites (14).
With respect to the specific Ca 2ϩ selectivity of the Ca 2ϩ -ATPase, it is interesting to note that the negatively charged M1 residues Glu 55 and Glu 58 are replaced by residues with positively charged side chains in the Na ϩ ,K ϩ -ATPase. Because transmembrane segment M3 of both ATPases contains additional negatively charged residues, which appear to be located rather close to M1 in the three-dimensional structure, the situation could be somewhat similar to voltage-gated Ca 2ϩ and Na ϩ channels, where the cation selectivity filter is formed by four negatively charged side chains in the former and two negative plus one positive side chain in the latter (15,16). Thus, the above-mentioned charged residues within the cytoplasmic part of M1 of the Ca 2ϩ -ATPase could possibly contribute to a Ca 2ϩ selectivity filter.
These considerations on the possible function of M1 of the Ca 2ϩ -ATPase in Ca 2ϩ migration and in long range signal transmission have prompted a detailed study of the structural and functional aspects of M1. To test the above-mentioned hypothesis about Glu 51 , Glu 55 , Glu 58 , and Asp 59 as part of a possible Ca 2ϩ entry pathway and/or selectivity filter, we have mutated these amino acids individually to the neutral leucine and to the oppositely charged and rather bulky arginine, which would be expected to impose electrostatic and steric effects on Ca 2ϩ migration, if positioned near the migration pathway. Moreover, we have substituted the hydrophobic amino acids Phe 57 , Leu 60 , Val 62 , and Leu 65 to investigate the demands for bulkiness and hydrophobicity of these side chains. Phe 57 is located on the hydrophobic surface of the bent part of M1 in the E 2 -Tg structure and could be instrumental in attaching this part to the lipid bilayer. The other hydrophobic residues are embedded in the membrane, Leu 65 below the level of the Ca 2ϩ ions in the Ca 2ϩ -bound structure. The effects of these mutations on the various partial reaction steps involved in Ca 2ϩ binding and catalysis have been analyzed, using among other assays also transient kinetic methods that allow measurement of the rates of conformational changes and Ca 2ϩ dissociation toward the cytoplasmic side.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis and Expression-The QuikChange sitedirected mutagenesis kit (Stratagene) was used to introduce the desired mutations directly into full-length cDNA encoding the rabbit fast twitch muscle Ca 2ϩ -ATPase (SERCA1a isoform). The template in the mutagenesis reaction consisted of SERCA1a cDNA inserted into the expression vector pMT2 (17). For expression of recombinant protein, pMT2 containing wild-type or mutant cDNA was transfected into mammalian COS-1 cells using the calcium phosphate precipitation method (18). The microsomal fraction containing expressed wild-type or mutant Ca 2ϩ -ATPase was isolated by differential centrifugation (19), and the expression level was quantitated by a specific enzyme-linked immunosorbent assay (20).
Functional Studies-The ability of wild-type or mutant Ca 2ϩ -ATPase to transport Ca 2ϩ and hydrolyze ATP was determined by studying the 45 Ca 2ϩ uptake into microsomal membranes and the P i liberation, respectively (21). The molecular turnover rate was calculated as the ratio between the specific Ca 2ϩ -ATPase activity and the active-site concentration. The latter was determined by quantitation of phosphorylation from [␥-32 P]ATP at 0°C in the presence of a saturating Ca 2ϩ concentration. Samples of wild-type and mutant Ca 2ϩ -ATPase were examined in parallel, and the molecular turnover rate corresponding to the mutants was expressed relative to that of the wild type. Partial reactions were studied according to previously described principles (20 -23), taking advantage of the acid stability of the phosphoenzyme formed from [␥-32 P]ATP or 32 P i , and details are given in the figure legends. A Bio-Logic quenched flow module QFM-5 or SFM-400/Q (Bio-Logic Science Instruments, Claix, France) was used for rapid mixing in experiments where the reaction time was below 1 s, applying the previously described protocols (22,23). In all phosphorylation experiments, the reaction was quenched by addition of 0.5-2 volumes of 25% (w/v) trichloroacetic acid containing 100 mM H 3 PO 4 . Subsequently, the acidprecipitated enzyme was washed by centrifugation, dissolved in loading buffer and subjected to separation by SDS-PAGE at pH 6.0 (24). The distribution of radioactivity associated with the gel was quantitated by electronic autoradiography of the dried gel using a Packard Cyclone TM Storage Phosphor System.
Data Analysis-Using the Sigmaplot program (SPSS Inc.), time courses of phosphorylation and dephosphorylation were fitted according to first-order kinetics, whereas ligand concentration dependences were fitted by use of the Hill equation. Generally, the experiments were performed at least twice on different enzyme preparations, and average values are shown in the figures. Standard errors are shown in the tables except for a few cases, in which the experiment (giving a wild type-like result) was performed only once.

Site-directed Mutagenesis and Expression-Several
Ca 2ϩ -ATPase point mutants were designed to study the functional importance of transmembrane segment M1. An overview of these mutations is given in Table I. Two different strategies were used: (i) charge reversal or removal (Glu 51 , Glu 55 , Glu 58 , and Asp 59 ) and (ii) hydrophobic size reduction (Phe 57 , Leu 60 , Val 62 , and Leu 65 ). The carboxylic acid residues were replaced by the positively charged and rather bulky arginine and with leucine, removing the negative charge. Phe 57 , Leu 60 , Val 62 , and Leu 65 were replaced by the smaller hydrophobic alanine to test the importance of the bulkiness of these side chains. In addition, Phe 57 was replaced by the more polar glutamine, and Leu 60 by proline. The expression level of all these Ca 2ϩ -ATPase constructs in COS cells was rather similar to that of the wild type (data not shown).
Effect on Overall Reaction-To characterize the effect of the M1 mutations on the overall function of the Ca 2ϩ -ATPase, the ability to hydrolyze ATP was determined at 37°C in the pres-SCHEME 1. Ca 2؉ -ATPase reaction cycle. Simplified scheme illustrating the partial reactions of the Ca 2ϩ -ATPase reaction cycle. Occluded Ca 2ϩ ions are shown in brackets.
ence and absence of the calcium ionophore A23187. Data obtained in the presence of ionophore are shown in Table I. Mutant Leu 65 3 Ala displayed the most pronounced reduction of maximal ATPase activity (to 3% of wild type). The maximal ATPase activity of Asp 59 3 Leu was reduced to 18%. In contrast, the maximal ATPase activity was increased in Val 62 3 Ala relative to wild type (by 30%). Upon addition of the calcium ionophore, passive efflux of Ca 2ϩ from the vesicles is facilitated, thereby relieving "back inhibition" caused by accumulation of a high concentration of Ca 2ϩ inside the vesicles (21). In the absence of calcium ionophore, the ATPase activity of wild type and all mutants except Leu 65 3 Ala was 2-to 3-fold lower than in its presence (data not shown), indicating that the ability to pump Ca 2ϩ and the luminal sensitivity to Ca 2ϩ were preserved. In accordance with this conclusion, direct measurement of ATP-dependent 45 Ca 2ϩ transport at 37°C in the presence of oxalate to trap Ca 2ϩ in the lumen of the microsomal vesicles demonstrated transport activity in all mutants except Leu 65 3 Ala (Table I). The mutational effects on ATP hydrolysis and Ca 2ϩ transport are not sufficiently different to signify any change in the coupling between the catalytic site in the cytoplasmic domain and the Ca 2ϩ binding sites in the transmembrane part of the molecule. Although the 30% increase of ATPase activity seen for Val 62 3 Ala is not accompanied by an increase of the Ca 2ϩ transport rate, this relatively minor difference may well be accounted for by the higher Ca 2ϩ concentration present in the vesicles during the Ca 2ϩ transport assay, which through Ca 2ϩ binding at the luminal sites slows the turnover of the phosphoenzyme, thereby tending to mask any accelerating mutational effect. Having established the impact of the various mutations in M1 on the overall functional performance of the Ca 2ϩ -ATPase, we proceeded with a more detailed analysis of the partial reactions of the enzyme cycle.
Ca 2ϩ Dependence of Phosphorylation from [␥-32 P]ATP-The Ca 2ϩ -mediated activation of phosphorylation from ATP requires the binding of both calcium ions at the high affinity cytoplasmically facing sites (cf. Reactions 2 and 3 in Scheme 1).
Examination of the mutational effect on Ca 2ϩ titration of phosphorylation therefore provides information about the Ca 2ϩbinding properties. The Ca 2ϩ dependence of steady-state phosphorylation was studied at 0°C, where phosphoenzyme accumulation is favored. Fig. 1 shows the titration curves for some of the mutants, and Table I summarizes the data for all mutants. Substitution of Glu 51 , Glu 55 , or Glu 58 by either leucine or arginine had no significant effect on the apparent Ca 2ϩ affinity, as exemplified by Glu 58 3 Arg in Fig. 1 (cf. Fig.  1 and Table I). In contrast, arginine substitution of Asp 59 caused a 5-fold reduction of apparent Ca 2ϩ affinity (increase of K 0.5 ) compared with wild type, whereas the apparent affinity for Ca 2ϩ was unaffected by leucine or alanine substitution of this residue. These results are in accordance with a previous study in which the Ca 2ϩ dependence of Ca 2ϩ transport was found normal in a cluster mutant with alanine substitutions of Glu 55 , Gln 56 , Glu 58 , and Asp 59 (25). The mutants Phe 57 3 Ala, Leu 60 3 Ala/Pro, and Val 62 3 Ala, with alterations to hydrophobic residues, displayed moderate reductions of apparent Ca 2ϩ affinity (1.5-to 3-fold increase of K 0.5 ). On the contrary, Leu 65 3 Ala displayed higher affinity for Ca 2ϩ than wild type, the K 0.5 being reduced to 61% of the corresponding value for the wild type ( Fig. 1 and Table I).
Rate of Ca 2ϩ Dissociation toward the Cytoplasmic Side-To determine more directly the effects of the mutations on Ca 2ϩ binding, we examined the rate of Ca 2ϩ dissociation toward the cytoplasmic side from the [Ca 2 ]E 1 form at pH 6.0, 25°C (Fig. 2), using a previously described quenched flow technique (23). Mutant or wild-type enzyme pre-equilibrated with saturating amounts of Ca 2ϩ was treated with an excess of the Ca 2ϩ chelator EGTA for various time intervals prior to a 34-ms incubation with [␥-32 P]ATP testing the ability to phosphorylate. Because occupancy of the Ca 2ϩ binding sites is required to render the enzyme phosphorylatable from ATP, the time course of disappearance of the ability to phosphorylate reflects the dissociation of Ca 2ϩ . Because the enzyme is not phosphorylated at the time when Ca 2ϩ dissociates, the dissociation must occur  (21) at 37°C. The medium contained 50 mM TES/Tris (pH 7.0), 100 mM KCl, 5 mM ATP, 7 mM MgCl 2 , 1 M Ca 2ϩ ionophore A23187, 1 mM EGTA, and either no added CaCl 2 (for background) or 0.9 mM CaCl 2 (giving a free Ca 2ϩ concentration of 3 M). Following subtraction of background, the molecular turnover rate was calculated as the ratio between the rate of ATP hydrolysis and the active-site concentration, using as active-site concentration the maximum amount of phosphoenzyme that can be formed from ATP (cf. Fig. 1), and the resulting data are shown relative to wild type (turnover rate 110 s Ϫ1 ).
b Measurement of ATP-driven Ca 2ϩ transport was carried out by filtration following incubation for 5 min at 37°C in a medium containing 20 mM MOPS/Tris (pH 6.8), 100 mM KCl, 5 mM ATP, 5 mM MgCl 2 , 0.5 mM EGTA, 0.45 mM 45 CaCl 2 , and 5 mM potassium oxalate to trap Ca 2ϩ in the lumen of the microsomal vesicles. To correct for expression level, the data were related to the active-site concentration as described above for ATP hydrolysis, and the results are shown relative to wild type.
c Data obtained as for Fig. 1. toward the cytoplasmic side of the membrane. It is believed that the ability to phosphorylate disappears with the dissociation of the first Ca 2ϩ ion in a sequential mechanism (26). Fig.  2 shows such time courses for some of the mutants, and Table  I summarizes the results for all mutants in terms of the halflife of the Ca 2ϩ -bound enzyme form capable of undergoing phosphorylation. The half-lives corresponding to mutants Glu 51 3 Arg/Leu, Glu 55 3 Arg/Leu, Glu 58 3 Arg/Leu, and Asp 59 3 Ala were indistinguishable from that of the wild type, whereas a decrease of around 10-fold was determined for Asp 59 3 Arg/Leu, corresponding to a 10-fold higher rate of Ca 2ϩ dissociation ( Fig. 2 and Table I). Furthermore, the Ca 2ϩ dissociation kinetics was quite sensitive to the size of the hydrophobic side chains in the middle part of M1. Hence, Leu 60 3 Ala/Pro and Val 62 3 Ala displayed markedly reduced half-lives relative to wild type, i.e. corresponding to a 2-to 3-fold increase of the Ca 2ϩ dissociation rate. Most conspicuously, alanine substitution of Leu 65 resulted in an 11-fold increase of the half-life (reduction of the rate of Ca 2ϩ dissociation). Mutation of Phe 57 (to glutamine or alanine) did not affect the Ca 2ϩ dissociation rate to any significant extent ( Fig. 2 and Table I).
Time Course of the E 2 3 [Ca 2 ]E 1 P Transition-To further examine the Ca 2ϩ binding properties of the mutants, the E 2 3 [Ca 2 ]E 1 P transition, consisting of Reactions 1-3 in Scheme 1, was studied by determining the rate of phosphoenzyme formation following simultaneous addition of Ca 2ϩ and [␥-32 P]ATP to enzyme initially present in the Ca 2ϩ -deprived E 2 form at pH 6.0, 25°C. Under these conditions, the E 2 3 [Ca 2 ]E 1 transition is rate-limiting for the phosphorylation, as the reaction with ATP (Reaction 3 in Scheme 1) is much faster (cf. Refs. 22 and 27, also confirmed for the mutants studied here, data not shown). Fig. 3 shows the time course for the wild type and Asp 59 3 Arg, and Table I summarizes the results for all mutants in terms of the rate constants obtained by fitting a monoexponential function to the data. For Asp 59 3 Arg, the observed rate constant was 1.8-fold higher than that corresponding to wild type, suggesting an enhanced rate of the E 2 3 [Ca 2 ]E 1 transition. For the other mutants the data were indistinguishable from wild type.
Vanadate Dependence of Inhibition of Phosphorylation from [␥-32 P]ATP-Vanadate (VO 4 3Ϫ ), acting as an analogue of the phosphoryl group in the transition state during dephosphorylation, binds preferentially to the E 2 form (28,29). Hence, the apparent affinity for vanadate can be used to quantify mutational effects on the equilibrium between E 2 and E 1 conformations, provided that the intrinsic affinity of the E 2 form for vanadate is not affected by the mutations. We have previously devised an assay in which the enzyme is equilibrated with various concentrations of vanadate at 25°C in the absence of  Table I together with the corresponding values for other mutants obtained in the same way. Ca 2ϩ and ATP, followed by determination of the amount of phosphoenzyme formed upon the addition of Ca 2ϩ and [␥-32 P]ATP at 0°C (23,30). Because vanadate binding and phosphorylation are competitive, and vanadate dissociation is very slow at 0°C, the amount of phosphoenzyme formed from ATP under these conditions reflects the proportion of enzyme present in the vanadate-free form before the addition of Ca 2ϩ and ATP. Using this assay, the apparent affinity constants for vanadate were determined for mutants with altered Ca 2ϩ binding characteristics and for the Phe 57 mutants, and are listed in Table II. The titration data are shown in Fig. 4 for wild type and Asp 59 3 Arg, which displayed a strikingly reduced appar-ent vanadate affinity compared with that of the wild type. This is consistent with a displacement of the E 2 -E 1 equilibrium in favor of E 1 and suggests that the above-described enhancement of the E 2 3 [Ca 2 ]E 1 transition rate in Asp 59 3 Arg is caused by an enhanced rate of the E 2 3 E 1 conformational change of Ca 2ϩ -free enzyme (Reaction 1 in Scheme 1). Among the other mutants, Leu 60 3 Ala and Val 62 3 Ala showed a slight (2-fold) increase in vanadate affinity, suggesting that there could be a displacement of the E 2 -E 1 equilibrium in favor of E 2 .
Time Course of Dephosphorylation of Phosphoenzyme Formed from [␥-32 P]ATP-To study the effects of the mutations on the processing of the phosphoenzyme, we first followed the decay of phosphoenzyme formed from [␥-32 P]ATP at pH 7.0, 0°C, presence of K ϩ , i.e. conditions where it is well recognized that the [Ca 2 ]E 1 P 3 Ca 2 E 2 P transition (cf. Reaction 4 in Scheme 1) is the major rate-limiting step during phosphoenzyme turnover of the wild-type enzyme (31,32). To observe the dephosphorylation, phosphorylation from [␥-32 P]ATP was terminated by simultaneous addition of EGTA and excess non-labeled ATP. Fig. 5 shows the phosphoenzyme decay for some of the mutants, and Table II summarizes the results for all mutants in terms of the rate constants obtained by fitting a monoexponential decay function to the data. The rate of disappearance of phosphoenzyme was wild type-like in mutants Glu 51 3 Arg/Leu, Glu 55 3 Arg/Leu, and Glu 58 3 Arg/Leu, as well as Asp 59 3 Arg. In contrast, a reduction of the rate of phosphoenzyme decay to only 17% was seen for Asp 59 3 Leu, whereas alanine substitution of this residue resulted in a much smaller decrease (to 68%). Among the mutants with alteration to hydrophobic residues, the most pronounced reduction of the rate of dephosphorylation (to 9%) was observed for Leu 65 3 Ala, whereas Leu 60 3 Ala and Val 62 3 Ala were slightly affected in the opposite direction (1.7-fold increase). To distinguish between effects on Reactions 4 -6 in Scheme 1, we determined the ADP sensitivity of the phosphoenzyme intermediate accumulated after phosphorylation with [␥-32 P]ATP, i.e. its ability to react with ADP and donate the phosphoryl group back to ADP, forming ATP. Only [Ca 2 ]E 1 P is able to undergo this were performed using Bio-Logic quenched flow module QFM-5 or SFM-400/Q according to previously described mixing protocols (22). Wildtype or mutant enzyme preincubated in 40 mM MES/Tris (pH 6.0), 80 mM KCl, and 2 mM EGTA, was mixed with an equal volume of 40 mM MES/Tris (pH 6.0), 80 mM KCl, 10 mM MgCl 2 , 2.2 mM CaCl 2 , and 10 M [␥-32 P]ATP, followed by acid quenching at the indicted times. A monoexponential function was fitted to the data, and the results were normalized separately for wild type and mutant, taking the phosphorylation corresponding to infinite time as 100%. The (relative) rate constant is shown in Table I together with the corresponding values for other mutants obtained in the same way.  reaction. Following phosphorylation of wild-type and mutant Ca 2ϩ -ATPase under conditions identical to those described for Fig. 5 5 and 6 in Scheme 1). The reduced rate of phosphoenzyme decay observed for these mutants in Fig. 5 suggests the latter possibility. To clarify this issue further, we measured directly the dephosphorylation of the E 2 P form (Figs. 6 and 7), as well as the rate of 3Ϫ ] n / (K 0.5 n ϩ [VO 4 3Ϫ ] n )) was fitted to the data, and the results were normalized separately for wild type and mutant, taking EP max as 100%. The (relative) K 0.5 value is shown in Table II  To follow the dephosphorylation, the phosphoenzyme was chased by addition of 6.7 mM EGTA with 1 mM non-radioactive MgATP followed by acid quenching at the time intervals indicated. A monoexponential decay function was fitted to the data, and the maximal phosphorylation was taken as 100%. The (relative) rate constants are listed in Table II together with the corresponding values for other mutants obtained in the same way.
FIG. 6. Time course of dephosphorylation of E 2 P phosphoenzyme at pH 6.0. Phosphorylation was carried out at 25°C for 10 min in 100 mM MES/Tris (pH 6.0), 10 mM MgCl 2 , 2 mM EGTA, 30% (v/v) Me 2 SO, and 0.5 mM 32 P i . Subsequently, the phosphoenzyme was chased by a 19-fold dilution of an aliquot into a medium (kept at 25°C) containing 100 mM MES/Tris (pH 6.0), 2 mM EGTA, 10 mM EDTA, 15% Me 2 SO, and 0.5 mM non-radioactive P i , and acid quenching was performed at the indicated time intervals. A monoexponential decay function was fitted to the data, and the maximal phosphorylation was taken as 100%. Each mutant is shown together with wild-type data obtained in the same series of experiments, and the (relative) rate constants are listed in Table II Fig. 6, except that the chase medium was kept at 0°C and contained 40 mM MOPS/Tris (pH 7.0), 5 mM MgCl 2 , 2 mM EGTA, 80 mM KCl, and 0.5 mM non-radioactive P i . accumulation of ADP-insensitive phosphoenzyme upon phosphorylation with ATP ( Fig. 8 and Table II).
Phosphoenzyme Formed from 32 P i -Under favorable conditions (absence of Ca 2ϩ , pH 6.0, 25°C, presence of the organic solvent dimethyl sulfoxide, absence of alkali metal ions), it is possible to form the phosphoenzyme intermediate E 2 P by "backward" reaction of E 2 with inorganic phosphate (cf. Reaction 6 in Scheme 1), even if the latter is present only at submillimolar concentration. The P i concentration dependence of the amount of E 2 P formed this way showed characteristics similar to wild type for Glu 51 3 Arg/Leu, Glu 55 3 Arg/Leu, and Glu 58 3 Arg/Leu (Table II). Arginine and leucine substitution of Asp 59 affected K 0.5 for P i in opposite directions, the former causing a 6-fold increase and the latter a decrease relative to wild type. No significant difference was seen between Asp 59 3 Ala and the wild type. A 2-fold increase of K 0.5 for P i was seen with Phe 57 3 Gln/Ala and Val 62 3 Ala, and in Leu 60 3 Pro the K 0.5 was 4-fold increased. Remarkably, the K 0.5 was lowered around 3-fold (apparent P i affinity increased) in Leu 65 3 Ala.
The dephosphorylation kinetics of E 2 P phosphoenzyme (cf. Reaction 6 in Scheme 1), formed under the conditions just described, was studied for some of the mutants by a 19-fold dilution into a chase medium, which terminated the phosphorylation from 32 P i . Fig. 6 shows the time course of E 2 P dephosphorylation at 25°C in medium containing 15% dimethyl sulfoxide at pH 6.0 in the absence of K ϩ , i.e. conditions where this reaction is rather slow and, therefore, easy to follow, and which are rather similar to those under which the K 0.5 values for P i were obtained. The rate constants determined by fitting a monoexponential function are listed in Table II. The variation of the k obs for E 2 P dephosphorylation paralleled to a large extent the change in K 0.5 for P i . Hence, the variation of the K 0.5 for P i is largely due to the mutational effects on E 2 P dephosphorylation. In particular, an 8-fold increase of k obs was determined for Asp 59 3 Arg, whereas Asp 59 3 Leu showed a 4-fold decrease (Fig. 6). Leu 60 3 Pro showed a 5-fold enhanced k obs , whereas Phe 57 3 Gln/Ala, Leu 60 3 Ala, and Val 62 3 Ala showed minor changes relative to wild type (up to 2.3-fold). A most conspicuous, almost complete block of E 2 P dephosphorylation was seen for Leu 65 3 Ala (Fig. 6 and Table II).
The markedly reduced k obs values for E 2 P dephosphorylation of Asp 59 3 Leu and Leu 65 3 Ala seem to explain perfectly well the accumulation of ADP-insensitive phosphoenzyme (E 2 P) observed for these mutants in the phosphorylation experiments with ATP described above. These experiments were, however, performed at pH 7.0, 0°C, and in the presence of K ϩ , whereas the results presented in Fig. 6 were obtained under conditions that deliberately slowed the dephosphorylation of E 2 P. This led us to study the dephosphorylation of E 2 P further for selected mutants at pH 7.0, 0°C, presence of K ϩ , and the results are seen in Fig. 7 and Table II. Clearly, the dephosphorylation of E 2 P was found markedly inhibited in Asp 59 3 Leu and Leu 65 3 Ala also under these conditions (ϳ10-and 33-fold, respectively, relative to wild type).
Time Course of the [Ca 2 ]E 1 P 3 Ca 2 E 2 P Transition-Because the data in Fig. 7 show that E 2 P 3 E 2 is the major rate-limiting step in the dephosphorylation of mutants Asp 59 3 Leu and Leu 65 3 Ala, even at pH 7.0, 0°C, presence of K ϩ , the results in Fig. 5 do not provide information about the rate of the [Ca 2 ]E 1 P 3 Ca 2 E 2 P transition (cf. Reaction 4, Scheme 1) in these mutants. Such information was instead obtained using an experimental design that takes advantage of the ADP-sensitivity of [Ca 2 ]E 1 P. The enzyme was phosphorylated with [␥-32 P]ATP under conditions known to slow down dephosphorylation of E 2 P, such that E 2 P accumulates even in the wild type (0°C, pH 8.0, presence of Li ϩ instead of K ϩ , high Mg 2ϩ / Ca 2ϩ ratio). After various phosphorylation intervals, the amount of ADP-insensitive phosphoenzyme was determined as the phosphoenzyme remaining after a 4-s incubation with ADP to remove [Ca 2 ]E 1 P. Because the [Ca 2 ]E 1 3 [Ca 2 ]E 1 P reaction is very rapid, the rate of appearance of ADP-insensitive phosphoenzyme corresponds to that of the [Ca 2 ]E 1 P 3 Ca 2 E 2 P transition under these conditions. Interestingly, the rates determined for Asp 59 3 Leu and Leu 65 3 Ala in this way were, respectively, 3.4-and 4.2-fold higher than that of the wild type ( Fig. 8 and Table II). This constitutes a second factor, besides the block of E 2 P dephosphorylation, contributing to the high steady-state concentration of ADP-insensitive phosphoenzyme in these mutants. DISCUSSION The present results provide the first functional evidence that transmembrane segment M1 of the Ca 2ϩ -ATPase is critical to Ca 2ϩ interaction and to phosphoenzyme turnover. During active Ca 2ϩ transport, the Ca 2ϩ -ATPase binds Ca 2ϩ from the cytoplasmic side of the membrane, transfers the ions across the membrane, and releases them on the luminal side. It is generally believed that the translocation of Ca 2ϩ across the membrane involves an "occluded" state, where the ions are inaccessible to the medium on either side of the membrane and their dissociation restricted by structural components acting as gates. In the occluded state, Ca 2ϩ is bound by residues in M4, M5, M6, and M8 (8, 10 -12). The Ca 2ϩ ions may become occluded within the protein concomitantly with phosphorylation from ATP (5,33). Evidence has, however, been presented that the Ca 2ϩ ions are not only occluded in the phosphoenzyme, but also most of the time in the non-phosphorylated Ca 2ϩ -bound form (6), as indicated by the brackets in Scheme 1. This seems to agree with the crystal structure of the Ca 2ϩ -ATPase with bound Ca 2ϩ , revealing no obvious Ca 2ϩ entry or exit pathways (8). Furthermore, the fact that Ca 2ϩ can dissociate only very slowly from the enzyme complex with CrATP, even though the ␥-phosphoryl group is not transferred to the enzyme in this complex, indicates that phosphorylation is not required for occlusion of Ca 2ϩ (7). The present findings are consistent with the existence of a Ca 2ϩ -occluded non-phosphorylated state, whose stability depends on the structural properties of M1. Our measurements indicate a 10-fold increase of the rate of Ca 2ϩ mM EGTA, and 2 mM ADP was added to remove ADP-sensitive phosphoenzyme, followed by acid quenching 4 s later. A monoexponential function was fitted to the data, and the results were normalized separately for wild type and mutant, taking the phosphorylation corresponding to infinite time as 100%. The (relative) rate constant is shown in Table II together with the corresponding value for Asp 59 3 Leu obtained in the same series of experiments. dissociation from this state toward the cytoplasmic side in mutants Asp 59 3 Arg and Asp 59 3 Leu, whereas Ca 2ϩ dissociation was wild type-like in the mutants with arginine or leucine substitution of Glu 51 , Glu 55 , or Glu 58 (Fig. 2 and Table  I). The Phe 57 mutants also displayed wild type-like Ca 2ϩ dissociation, whereas Leu 60 3 Ala/Pro and Val 62 3 Ala caused an acceleration of Ca 2ϩ dissociation, although to a lesser extent than the Asp 59 mutations. Moreover, a remarkable 11-fold reduction of the rate of Ca 2ϩ dissociation toward the cytoplasmic side was observed for Leu 65 3 Ala. This demonstration of the role of M1 in Ca 2ϩ occlusion was made feasible by the quenched flow technique that allows measurement of Ca 2ϩ dissociation from the [Ca 2 ]E 1 form toward the cytoplasmic side, even with the relatively small amounts of enzyme that can be harvested from the cell culture. Thus, the rate of Ca 2ϩ dissociation from the [Ca 2 ]E 1 form can be both increased and decreased by mutation within the middle part of M1 (C-terminal to the bend of M1 seen in the E 2 -Tg crystal structure), and our results suggest that the middle part of M1, but not the most N-terminal part containing Glu 51 , Glu 55 , and Glu 58 , is important in control of the gates at the Ca 2ϩ occlusion sites.
As regards the proposed role of Glu 51 , Glu 55 , and Glu 58 in the Ca 2ϩ entry pathway (14), the results reported here clearly argue against any involvement of these glutamate side chains in Ca 2ϩ recognition and binding, and it is not likely that they contribute to a Ca 2ϩ selectivity filter. Furthermore, the finding that substitution of Glu 58 by arginine left the Ca 2ϩ binding properties of the Ca 2ϩ -ATPase unaffected questions the existence of a close interaction between Glu 58 in M1 and Glu 309 at Ca 2ϩ binding site II, such as seen in the published crystal structure of the Ca 2ϩ -bound enzyme (8). A possible reason for this apparent discrepancy could be a high degree of thermal mobility of the Glu 58 side chain. It is also possible that the [Ca 2 ]E 1 form adopted in the native state differs more profoundly from the crystal structure.
The reduced apparent affinity of mutant Asp 59 3 Arg for Ca 2ϩ activation of phosphorylation ( Fig. 1) reflects the enhanced rate of Ca 2ϩ dissociation from [Ca 2 ]E 1 and not a displacement of the E 2 -E 1 equilibrium in favor of the low affinity E 2 form. In fact, the data in Figs. 3 and 4 suggest that mutation Asp 59 3 Arg enhances the rate of the E 2 3 E 1 conformational change, leading to accumulation of E 1 with resulting low sensitivity to vanadate inhibition of phosphoenzyme formation. Because release of counter-transported protons from E 2 may be rate-limiting for the E 2 3 E 1 transition (27), the enhancement of this transition suggests that proton dissociation from the transport sites, like Ca 2ϩ dissociation, is facilitated by the Asp 59 3 Arg mutation.
For mutants Leu 60 3 Ala/Pro and Val 62 3 Ala, the enhanced rate of Ca 2ϩ dissociation likewise seems to result in a reduced apparent affinity for Ca 2ϩ activation of phosphorylation. In addition, there could be a contribution to the reduced apparent affinity for Ca 2ϩ from displacement of the E 2 -E 1 equilibrium in favor of the low affinity E 2 form in Leu 60 3 Ala and Val 62 3 Ala, because the vanadate affinity was slightly increased in these mutants (Table II). The apparent affinity for Ca 2ϩ in activation of phosphorylation was normal in Asp 59 3 Leu, despite the increased Ca 2ϩ dissociation rate (Figs. 1 and 2). This may be accounted for by a considerably reduced rate of dephosphorylation (Figs. 5-7). As previously reported, the effect of a low rate of phosphoenzyme turnover on the apparent affinity for Ca 2ϩ activation of phosphorylation can be understood on the basis of computer simulations of the Ca 2ϩ -ATPase reaction cycle (34). According to the computational analysis, an increased apparent affinity for Ca 2ϩ is actually expected when the rate of phosphoenzyme turnover is reduced, because a lower phosphorylation rate (i.e. Ca 2ϩ saturation) is required to maintain a certain level of phosphorylation under these conditions. Hence, for Asp 59 3 Leu, the increased rate of Ca 2ϩ dissociation and the decreased rate of phosphoenzyme turnover act in opposite directions, thereby masking the effects on the apparent Ca 2ϩ affinity.
The importance of the middle part of M1 in the gating at the Ca 2ϩ occlusion sites may be related to the presence of a wateraccessible channel leading between M1 and M3 in the crystal structure of Ca 2ϩ -ATPase in the Ca 2ϩ -free E 2 -Tg form. This channel has the potential of a Ca 2ϩ entry port and is apparently opened by the bending and partial unfolding of the helical structure of M1 at Asp 59 (9). A similar channel may exist in the native non-crystalline protein and function as a migration pathway for Ca 2ϩ in one or more of the conformations that precede the occluded [Ca 2 ]E 1 form in the process of Ca 2ϩ binding (E 2 , E 1 , and CaE 1 ). Because the channel leads to Glu 309 at Ca 2ϩ site II, it may provide a passage for the Ca 2ϩ ion that binds at site II (i.e. the one that binds last in the sequential mechanism). In line with this assignment, the presently observed mutational effects on the rate of disappearance of ability to phosphorylate from ATP upon addition of EGTA (Fig. 2) reflect changes of the rate of dissociation of the Ca 2ϩ ion that leaves first in the sequential mechanism, i.e. the one that was bound last (26,35). The Ca 2ϩ ion that binds first (at site I, cf. Ref. 12) might enter (and leave upon dissociation) by a different route. Some evidence has actually been presented suggesting the existence of a pre-binding site for Ca 2ϩ in the loop between transmembrane segments M6 and M7 ("L6 -7"), which could be related to the entry port for the Ca 2ϩ ion binding at site I (36).
Comparison of the two crystal structures of the Ca 2ϩ -ATPase indicates that the transition from [Ca 2 ]E 1 to E 2 -Tg is accompanied by a lateral and upward (toward the cytoplasmic side) movement of M1 in the membrane, and bending of the helix at Asp 59 , which probably is caused by steric collision with M3 (9). In E 2 -Tg, the N-terminal part of M1 containing Glu 51 , Glu 55 , and Glu 58 is bent away from M3, and their side chains define the hydrophilic surface of the amphipathic section. These residues are therefore peripherally located with respect to the proposed Ca 2ϩ entrance. A similar position of this part of M1 in the native non-crystalline protein could explain the lack of importance of the glutamate side chains for Ca 2ϩ binding. Asp 59 , on the other hand, is located right at the bending point of M1. Thus, the dramatic effect on Ca 2ϩ dissociation of substitution of Asp 59 by leucine or arginine may result from direct interference with Ca 2ϩ interaction or interference with the movement of M1 that occludes Ca 2ϩ , a movement that conceivably is facilitated by flexibility at Asp 59 . The finding that alanine substitution of Asp 59 is tolerated stresses the requirement for a small side chain that allows movement. The negative charge of the Asp 59 side chain, which might have been expected to participate in directing Ca 2ϩ to the binding sites, seems not to be required for normal Ca 2ϩ interaction.
Among the hydrophobic side chains substituted in the present study, Phe 57 seems less important for the Ca 2ϩ -binding properties and the E 2 3 E 1 conformational transition than expected if it were crucial to the attachment of the bent part of M1 to the lipid bilayer and, thereby, to the positioning of this part of M1. On the other hand, the lack of a significant effect on Ca 2ϩ binding of substitution of Phe 57 may be considered consistent with its location away from the proposed Ca 2ϩ inlet, on the hydrophobic surface of the bent part of M1, and it is conceivable that the removal of a single anchoring side chain is not enough to disrupt the attachment to the lipid bilayer. As regards Leu 60 , Val 62 , and Leu 65 , which are embedded within the membrane, their bulky hydrophobic side chains appear to form an integral part of the wall lining the water-accessible channel seen in the E 2 -Tg structure, with Leu 65 at the bottom. Hence, our finding of significant changes of the Ca 2ϩ dissociation rate upon substitution of these residues with the smaller alanine seems to be compatible with the hypothesis that this channel serves as a migration pathway for Ca 2ϩ . The effects of hydrophobic size reduction in this part of M1 could simply be due to local changes in the proportions of the channel, providing a more free passage for the dissociating Ca 2ϩ ions in case of Leu 60 3 Ala/Pro and Val 62 3 Ala and collapsing the channel in case of the Leu 65 3 Ala mutation. Of course, it cannot be excluded that changes to M1 exert more distant effects on a Ca 2ϩ migration pathway located elsewhere in the protein.
In the crystal structure of the Ca 2ϩ -bound enzyme, Leu 65 is below the level of the Ca 2ϩ ions (i.e. closer to the luminal surface), and it seems that its bulky side chain may contribute to form a luminal gate, preventing the immediate transport of the ions to the lumen upon binding from the cytoplasm. A premature transfer of the Ca 2ϩ ions to the lumen caused by a defective luminal gate in Leu 65 3 Ala (i.e. rapid formation of an enzyme form somewhat similar to Ca 2 E 2 P) could account for the rapid disappearance of ADP sensitivity of the phosphoenzyme in this mutant (Fig. 8), because the ADP sensitivity is thought to be specifically associated with the [Ca 2 ]E 1 P form. Interestingly, also Asp 59 3 Leu showed an enhanced rate of disappearance of ADP sensitivity of the phosphoenzyme (Table  II), which again might indicate premature Ca 2ϩ transfer to luminally facing sites. In addition to the enhanced rate of disappearance of ADP sensitivity of the phosphoenzyme, Leu 65 3 Ala and Asp 59 3 Leu both showed a conspicuous block of the dephosphorylation of E 2 P, the rate being reduced 33-fold and 4-to 10-fold (depending on pH), respectively (Table II). Hence, there is a preference for forms with luminally facing Ca 2ϩ sites and ADP insensitivity of the phosphoenzyme (Ca 2 E 2 P and E 2 P) in both of these mutants. A difference between these mutants is, however, that Ca 2ϩ dissociation toward the cytoplasmic side is increased in Asp 59 3 Leu, whereas it is inhibited in Leu 65 3 Ala (Fig. 2).
Some of the other mutations also affected the processing of the phosphoenzyme (Reactions 4 -6 in Scheme 1). Hence, Leu 60 3 Ala and Val 62 3 Ala showed a slightly enhanced rate of dephosphorylation of phosphoenzyme formed from ATP ( Fig. 5 and Table II), probably reflecting enhancement of the rate-limiting [Ca 2 ]E 1 P 3 Ca 2 E 2 P transition. The rate of E 2 P dephosphorylation was markedly increased in Asp 59 3 Arg (8-fold) and Leu 60 3 Pro (4-to 5-fold) and to a lesser extent in Phe 57 3 Ala/Gln and Val 62 3 Ala (about 2-fold, Table II). It is noteworthy that the replacement of Asp 59 with arginine and leucine led to opposite effects on E 2 P dephosphorylation, which may be attributable to the difference in charge/polarity, resulting in different potentials for salt bridge formation or interaction with the lipid bilayer. The leucine may fix this part of M1 in close contact with the lipid bilayer, whereas the arginine may tend to draw it into the solvent phase. Taken together, the effects of M1 mutations on phosphoenzyme processing demonstrate that M1, besides being important for Ca 2ϩ interaction, furthermore is a critical element in the long range signaling between the transmembrane domain and the catalytic site in the cytoplasmic part of the Ca 2ϩ -ATPase molecule. Several pieces of evidence suggest that the A-domain undergoes large movements during the catalytic cycle, being separated from the P-domain in [Ca 2 ]E 1 and [Ca 2 ]E 1 P and in rather close contact with this domain in E 2 and E 2 P, probably contributing to the catalytic site in these forms (9,(37)(38)(39). Because M1 is physically connected to the A-domain through a flexible hinge segment at the N-terminal end and through M2 at the other end, it is tempting to speculate that the movement of the A-domain is closely coupled with movement of M1, and that the effects of the M1 mutations on the processing of the phosphoenzyme, described here, are brought about by an altered position and/or reduced motional freedom of the A-domain, caused by changes of the position and structure of M1.
In conclusion, our results provide evidence for a relation of the middle part of M1 to a gating mechanism controlling the dissociation of Ca 2ϩ from its membranous binding sites. Our results further indicate that these residues are also critical to the proper processing of the phosphoenzyme and, thus, to the long range transmission of conformational changes to the cytoplasmic catalytic site.