Functional Comparison between Secretory Pathway Ca2+/Mn2+-ATPase (SPCA) 1 and Sarcoplasmic Reticulum Ca2+-ATPase (SERCA) 1 Isoforms by Steady-state and Transient Kinetic Analyses*

Steady-state and transient kinetic studies were performed to functionally analyze the overall and partial reactions of the Ca2+ transport cycle of the human secretory pathway Ca2+/Mn2+-ATPase 1 (SPCA1) isoforms: SPCA1a, SPCA1b, SPCA1c, and SPCA1d (encoded by ATP2C1, the gene defective in Hailey-Hailey disease) upon heterologous expression in mammalian cells. The expression levels of SPCA1 isoforms were 200–350-fold higher than in control cells except for SPCA1c, whose low expression level appears to be the effect of rapid degradation because of protein misfolding. Relative to SERCA1a, the active SPCA1a, SPCA1b, and SPCA1d enzymes displayed extremely high apparent affinities for cytosolic Ca2+ in activation of the overall ATPase and phosphorylation activities. The maximal turnover rates of the ATPase activity for SPCA1 isoforms were 4.7–6.4-fold lower than that of SERCA1a (lowest for the shortest SPCA1a isoform). The kinetic analysis traced these differences to a decreased rate of the E1∼P(Ca) to E2-P transition. The apparent affinity for inorganic phosphate was reduced in the SPCA1 enzymes. This could be accounted for by an enhanced rate of the E2-P hydrolysis, which showed constitutive activation, lacking the SERCA1a-specific dependence on pH and K+

SPCAs (encoded by ATP2C1-2 genes) as well as sarco(endo)plasmic reticulum Ca 2ϩ -ATPases (SERCAs; encoded by ATP2A1-3 genes) and plasma membrane Ca 2ϩ -ATPases (PMCAs; encoded by ATP2B1-4 genes) are characterized by the obligatory autophosphorylation during their reaction cycle (5,6). Analysis of the deduced amino acid sequences of human (7), rat (8), worm (9), insect (9), and yeast (2) SPCA proteins indicates that despite the relatively low amino acid sequence homology (30 -35%) with SERCAs, the SPCAs retain the key architectural features revealed in the recently elucidated high resolution crystal structures of the rabbit SERCA1a (994 amino acids) in Ca 2ϩ -free thapsigargin-bound E 2 and Ca 2ϩ -bound E 1 forms either with or without bound nucleotide (10 -15). Relative to SERCA1a, SPCAs appear to be similarly composed of 10 hydrophobic helices (M1-M10) displaying strong membranespanning propensity. Other SERCA1a-specific structural determinants conserved in SPCAs are the phosphorylation (P, containing the phosphorylatable Asp 350 in human SPCA1 and Asp 351 in SERCA1a), nucleotide-binding (N) and actuator (A) domains present in the large "headpiece" emerging into the cytosol. The direct structural information obtained from SERCA1a crystals (10) has confirmed that two Ca 2ϩ ions are bound at two high affinity binding sites (site I and site II) located in the transmembrane region, as previously proposed on the basis of extensive mutational studies (16 -18). The sequence alignment of SPCAs from different species (9) together with functional studies involving mutants derived from yeast (19) and human (20) SPCA proteins have shown that only the amino acid residues contributing to site II (corresponding to Glu 308 of M4 and Asn 738 and Asp 742 of M6 in human SPCA1) are conserved in SPCAs. Hence, SPCAs are most likely able to couple the chemical energy derived from the hydrolysis of one ATP molecule to the vectorial transport of only one divalent ion (Ca 2ϩ or Mn 2ϩ ). SPCAs owe their selectivity for Mn 2ϩ binding and transport to the critical packing interaction between Gln 783 in M6 and Val 335 in M4 in yeast SPCA (21); Gln 747 and Val 314 are the corresponding residues in human SPCA1. Our recent functional study on human SPCA1 mutants (20) has revealed that Gly 309 in M4 (Gly 330 in yeast SPCA) may also play an important role in conferring human SPCA1 its selectivity for Mn 2ϩ transport. Similar to SERCA1a (except for the number of transported Ca 2ϩ per cycle), it may be suggested that the binding of Ca 2ϩ (or Mn 2ϩ ) in the transmembrane region of SPCA1 (E 2 to E 1 Ca transition) activates the phosphorylation of Asp 350 by ATP, thus forming the E 1 ϳP(Ca) (ADP-sensitive high energy) phosphoenzyme in which Ca 2ϩ is occluded (Scheme 1). The next reaction step involves the transformation of this intermediate to the E 2 -P (ADP-insensitive low energy) phosphoenzyme with the concomitant release of Ca 2ϩ into the lumen and represents a known crucial rate-limiting step in the Ca 2ϩ transport cycle of SERCA1a (6). Hydrolysis of E 2 -P completes the reversible cycle. Whether SPCAs also mediate countertransport of proton(s) in association with the latter step, as in SERCA enzymes, has not been determined.
The critical importance of SPCAs in supplying the different Golgi compartments with sufficient Ca 2ϩ (and Mn 2ϩ ) for the correct execution of diverse luminal functions was recently revealed when clinical, genetic, and functional investigations established that the human SPCA1 gene (ATP2C1) is indeed the gene defective in the Hailey-Hailey skin disease (HHD; OMIM 16960) (7,20,22).
We recently elucidated the mechanism governing the alternative processing of the 3Ј-end of the human ATP2C1 primary transcripts in keratinocytes and showed that up to four SPCA1 isoforms with distinct C termini can be generated: SPCA1a (919 amino acids), SPCA1b (939 amino acids), SPCA1c (888 amino acids), and SPCA1d (949 amino acids) splice variants (20). The shortest isoform (SPCA1c) appears to be missing part of the predicted transmembrane segment M10 and, therefore, its ability to function as a calcium pump is rather questionable (20). The steady-state functional analyses of human SPCA1d and SPCA1a upon their respective heterologous expression in COS-1 cells (20) and yeast (23) have demonstrated their ability to transport both Ca 2ϩ and Mn 2ϩ with high apparent affinities, although the apparent affinities for Ca 2ϩ of SPCA1a (K 0.5 ϭ 0.26 M) and SPCA1d (K 0.5 ϭ 0.20 M) were severalfold lower than that of yeast SPCA (K 0.5 ϭ 0.07 M; Ref. 24), but quite similar to that of Caenorhabditis elegans SPCA (K 0.5 ϭ 0. 25 M,Ref. 9). None of these SPCAs are sensitive to the classical SERCA inhibitors thapsigargin, 2,5-di-(tert-butyl)-1,4-benzohydroquinone, and cyclopiazonic acid (25). So far, the overall (e.g. Ca 2ϩ or Mn 2ϩ transport and Ca 2ϩ -or Mn 2ϩ -activated ATP hydrolysis) and partial reactions of the catalytic cycle (Scheme 1) mediated by SPCAs have not been studied in any detail. Furthermore, the kinetic properties of the different human SPCA1 isoforms have not been yet compared, and, therefore, the advantage of having such isoform diversity is not obvious.
In the present study, we have investigated in depth the Ca 2ϩ transport cycle of the human SPCA1 isoforms transiently expressed in the mammalian HEK-293 cells. We document for the first time the kinetic differences between SERCA1a and the fully active SPCA1a, SPCA1b, and SPCA1d isoforms, which have dramatic consequences for the Ca 2ϩ and ATP dependences, the maximum turnover rate, and vanadate inhibition. A quite unexpected finding is that the dephosphorylation of the E 2 -P phosphoenzyme intermediate is insensitive to regulation by pH and K ϩ concentration in the SPCA1 isoforms.

EXPERIMENTAL PROCEDURES
cDNA Cloning-The cDNA clone encoding the human SPCA1d isoform in expression vector pMT2 (under the control of the SV40 pro-moter) was obtained as previously described (20). The other clones encoding human SPCA1a, SPCA1b, and SPCA1c isoforms were obtained by PCR amplification of the SPCA1-specific coding regions using as template DNA the first strand cDNA reverse-transcribed from total RNA isolated from human keratinocytes. PCR primer pairs consisted of a common SPCA1-specific 5Ј-primer with a 5Ј-overhang sequence containing the EcoRI site (in italics below) and an isoformspecific 3Ј-primer with a 5Ј-overhang sequence containing the XbaI site (in italics below) and the respective stop codon. Because the human SPCA1 gene (ATP2C1) contains two closely located ATG codons, additional care was taken to ensure that the 5Ј-primer contains only the second ATG codon (a short non-functional peptide would otherwise be generated from the first ATG site). The 5Ј-primer used is 5Ј-CAG-GAATTCAATGAAGGTTGCACGTTTTC-3Ј, where the underlined part with the ATG site in bold corresponds to nucleotides 235-254 in human SPCA1a and SPCA1c nucleotide sequences deposited, respectively, under accession numbers AF181120 and AF181121 (7), and to nucleotides 129 -148 in human SPCA1b and SPCA1d nucleotide sequences deposited, respectively, under accession numbers AY268374 and AY268375 (20). The 3Ј-primers used for human SPCA1a and SPCA1c, respectively, are 5Ј-GCTCTAGAAATGCAATATGCAT-CATACTTC-3Ј and 5Ј-GCTCTAGATGGGGATGTATCTCAAC-3Ј, where the underlined part corresponds to the respective inverse complement of nucleotide stretches 2987-3007 and 2898 -2917 with the stop codon in bold. The 3Ј-primer used for human SPCA1b is 5Ј-GCTCTAGAACTGGACTTAGACACAGCTC-3Ј, which corresponds to the inverse complement of nucleotide stretch 2937-2956 with the stop codon in bold. (Note that both human SPCA1b and SPCA1d PCR fragments are amplified when this 3Ј-primer is used.) PCR amplifications with a hot start were carried out initially for 10 cycles with each cycle consisting of 30 s at 94°C, 30 s at 50°C, and 3 min at 68°C followed by 20 cycles with each cycle consisting of 30 s at 94°C, 30 s at 62°C, and 3 min at 68°C for all primer pairs using a mixture of Pwo (proofreading activity) and Taq polymerases from Roche Applied Science. To obtain the corresponding cDNA clones coding for human SPCA1a and SPCA1c isoforms, the respective amplified PCR fragments were gel-purified and subcloned in the pCA␤ vector (under the control of the ␤-globin gene promoter), whereas the pMT2 vector was employed for human SPCA1b. Rabbit SERCA1a cDNA was also used (26).
Transfection, Microsome Preparation, and Protein Characterization-Transfection of HEK-293 and COS-1 cells with each of the cDNAs in the mammalian expression vectors pMT2 or pCA␤ was performed using the transfection reagent FuGENE TM 6 (Roche Applied Science). The microsomal fraction containing the transiently overexpressed Ca 2ϩ -ATPase was isolated from transfected cells by differential centrifugation (27) and used for immunochemical and enzymatic assays. Protein concentration determination, denaturing gel electrophoresis, semi-dry blotting, and blot immunostaining were performed as reported earlier (28,29). The specific polyclonal antibody able to recognize human SPCA1 proteins was prepared as we previously described (20,30).
Functional Assays-45 Ca 2ϩ fluxes were carried out on saponin-permeabilized COS-1 cells as described earlier (9,20). The methods used for the analysis of the catalytic cycle in steady-state and transient-kinetic conditions have been previously established from studies with SERCA1a mutants (31,32) and used very recently to characterize SERCA2 (33) and SERCA3 isoforms (34,35) as well as Darier disease (SERCA2b) mutants (33). These methods were directly applicable to expressed human SPCA1 enzymes. The rapid quench-flow methodol-SCHEME 1. SPCA reaction cycle. E 1 , enzyme form with cytoplasmically facing high affinity Ca 2ϩ site; E 2 , enzyme form with low affinity for Ca 2ϩ ; E 1 ϳP(Ca), high energy ADPsensitive phosphoenzyme with occluded calcium ion (shown in parentheses); E 2 -P, low energy ADP-insensitive phosphoenzyme with luminally facing low affinity Ca 2ϩ site. The proposed cycle is analogous to that of SERCA pumps, except that the latter transport two Ca 2ϩ per hydrolyzed ATP. ogy, which allows kinetic measurements to be performed on a millisecond scale, was employed as described (31,33). Additional details of the functional assays are given in the figure legends. The background phosphorylation levels built up in reactions with [␥-32 P]ATP or 32 P i were determined by adding, respectively, excess EGTA or Ca 2ϩ prior to phosphorylation. The constant level of phosphorylation remaining after the exponential decay of phosphoenzyme was used as background in some of the dephosphorylation experiments.
Data Analysis-All data presented are average values corresponding to two or more experiments. Standard errors (S.E.) larger than the symbols are shown as error bars in the figures. Experimental data were fitted by linear and non-linear regression analysis using the SigmaPlot program (SPSS Inc.) or by means of the kinetic simulation software SimZyme (31). The values extracted for V max , K m , K 0.5 , Hill coefficient, and different rate constants are listed in TABLES ONE to THREE.

RESULTS
Expression of Human SPCA1 Isoforms-The expression levels of human SPCA1a, SPCA1b, SPCA1c, and SPCA1d isoforms in HEK-293 cells transfected with the corresponding cDNA were qualitatively examined and compared by Western blotting and immunocytochemistry analyses. The human SPCA1-specific polyclonal antibody used in these studies was raised against the large cytosolic loop connecting membrane-spanning helices M4 and M5 in human SPCA1 isoforms as described (20,30). Fig. 1A shows a typical Western blot analysis performed on microsomal proteins isolated from transfected HEK-293 cells. The immunoblot analysis demonstrates that, with the exception of SPCA1c, all SPCA1 isoforms were clearly expressed in HEK-293 cells at levels much higher than that of the endogenous SPCA1 in control cells transfected with empty vector. SPCA1c appeared to be expressed at a level just slightly higher than that found in control microsomes. Human SPCA1 proteins were well resolved according to their sizes by denaturing gel electrophoresis: SPCA1a (919 amino acids), SPCA1b (939 amino acids), and SPCA1d (949 amino acids). The faint band corresponding to SPCA1c (888 amino acids) showed a slower electrophoretic mobility than expected (Fig. 1A). Additionally, the upper bands in Fig. 1A detected just below the 188-kDa molecular mass standard most likely represent SPCA1a-, SPCA1b-, and SPCA1d-derived dimers. Finally, the immunocytochemical analysis showed that the expression of SPCA1a, SPCA1b, and SPCA1d appeared to be largely confined to a juxtanuclear Golgi-like compartment, whereas SPCA1c was expressed in only a few cells and showed a more ER-like distribution (data not shown). Furthermore, our immunoblot analysis using a SERCA2b-specific antibody showed that the expression level of the endogenous SERCA2b did not change following the transient overexpression of the human SPCA1 isoforms in HEK-293 cells (data not shown).
The expression levels of SPCA1 isoforms in HEK-293 cells were further quantified by measuring their maximum capacity for phosphorylation with [␥-32 P]ATP ("active site concentration") in the presence of activating Ca 2ϩ (forward reaction 3 in Scheme 1) and at 0°C and neutral pH, i.e. under conditions known to preserve the stability of the phosphorylated intermediate E 1 ϳP(Ca). The phosphorylation reactions were performed in the presence of thapsigargin (0.1 M), which inhibits the endogenous SERCA2b activity in both transfected and control HEK-293 cells. Fig. 1B illustrates the results of this functional approach. SPCA1-specific phosphorylation levels ranging from 100 to 200 pmol of active enzyme/mg microsomal protein, i.e. 1-2% of the total amount of microsomal protein, were obtained for the human SPCA1a, SPCA1b, and SPCA1d isoforms. These phosphorylation levels are comparable to those documented for SERCA pumps (33,34). For direct comparison the expression level reached for SERCA1a in the present series of experiments was ϳ200 pmol/mg (data not shown). Importantly, the expression levels of the human SPCA1 isoforms were 200 -350-fold higher than that of the endogenous SPCA1 pump in control HEK-293 cells. Therefore, the contribution of the endogenous SPCA1 enzyme to the enzymatic measurements described below is negligible. Even in the absence of thapsigargin, the background phosphorylation level generated by the combined action of SERCA2b and SPCA1 in control HEK-293 cells does not rise above more than 0.5-1.0 pmol of Ca 2ϩ -ATPase/mg microsomal protein (33). Hence, the overexpression of exogenous SPCA1a, SPCA1b, and SPCA1d and the low background level allowed the reliable use of the heterologous mammalian cell expression system for the functional characterization of these SPCA enzymes. On the other hand, the phosphorylation level of SPCA1c (Fig.  1B) is as low as that obtained with control microsomes. 45 Ca 2ϩ Transport Activity-The thapsigargin-insensitive ability of each of the transiently overexpressed SPCA1 isoforms to actively transport Ca 2ϩ into a membrane-delineated Ca 2ϩ store was assessed following expression in COS-1 cells as previously described (9,20). Using this whole cell approach the competence of the overexpressed enzyme is evaluated within a well preserved cellular environment. The filling state of the internal Ca 2ϩ stores was studied in saponin-permeabilized cells, following a 90-min loading with a 45 Ca 2ϩ -containing solution in the presence or absence of oxalate (Fig. 1C). Oxalate precipitates Ca 2ϩ within intracellular compartments, thus reducing the luminal free Ca 2ϩ concentration and, thereby, relieving the back inhibition exerted by Ca 2ϩ at the luminally facing low affinity Ca 2ϩ binding site of Ca 2ϩ -ATPase. COS-1 cells were preferred to HEK-293 cells in these experiments, because of their increased adherence to the gelatin-coated wells, thus being less prone to detaching during permeabilization, loading, and washing steps (9). At the end of the loading period (corresponding to the zero time point in each panel of Fig. 1C), the level of 45 Ca 2ϩ accumulated in the presence of oxalate by SPCA1a, SPCA1b, and SPCA1d, respectively, was 2.8-, 2.9-, and 4.0-fold increased relative to that of control cells. Furthermore, when the calcium ionophore A23187 was included in the efflux medium, the rate of 45 Ca 2ϩ release increased, thus demonstrating that during the loading period the overexpressed SPCA1 isoforms had indeed accumulated Ca 2ϩ within intracellular membranous Ca 2ϩ stores. No 45 Ca 2ϩ transport activity specific for the human SPCA1c isoform could be measured (Fig. 1C). Because of the low level of expression ( Fig. 1A) and lack of phosphorylation (Fig. 1B), no other functional measurements were further performed on SPCA1c. These results suggest altogether that SPCA1c represents a rapidly degradable non-functional SPCA1 protein.
ATPase Activity-Steady-state ATPase activity at 5 mM MgATP was determined for each of the well expressed SPCA1a, SPCA1b, and SPCA1d isoforms by monitoring the release of inorganic phosphate. For each SPCA1 isoform, Fig. 2 shows the Ca 2ϩ dependence of the ATPase turnover rate determined at pH 7.0 and 37°C, in the presence and absence of the calcium ionophore A23187. The maximum turnover rate for ATP hydrolysis, the effect of the calcium ionophore on the turnover rate, and the Ca 2ϩ titration data obtained for SPCA1 and SERCA1a enzymes are summarized in TABLE ONE. For SERCA1a, the E 1 ϳP(Ca 2 ) to E 2 -P transition (step 4 in Scheme 1) is inhibited by the binding of accumulated Ca 2ϩ at luminal low affinity sites. When the Ca 2ϩ concentration of the medium is within micromolar range, the addition of calcium ionophore relieves the inhibition by allowing rapid equilibration of Ca 2ϩ between the two sides of the microsomal membrane through passive Ca 2ϩ efflux. Similar effects of ionophore addition were observed for the SPCA1 enzymes. Surprisingly, the maximum ATPase turnover rates determined for SPCA1a, SPCA1b, and SPCA1d, respectively, were 6.4-, 5.7-, and 4.7-fold lower than that of SERCA1a. Nevertheless, the apparent affinities for Ca 2ϩ activation of the ATPase activity displayed by SPCA1a, SPCA1b, and SPCA1d were much higher (K 0.5 , 7.5-18.2-fold lower in TABLE ONE) than that of SERCA1a. Finally, the Hill coefficient values (listed in TABLE ONE) of ϳ1.0 for SPCA1 isoforms indicate that the cooperativity seen for SERCA1a is absent for SPCA1 isoforms. The latter finding is consistent with the proposal that the active transport of Ca 2ϩ performed by each human SPCA1 isoform takes place via a single high affinity Ca 2ϩ binding site (i.e. one Ca 2ϩ is transported per each ATP-driven catalytic cycle) instead of the two cooperatively interacting Ca 2ϩ binding sites of SERCA pumps.
Phosphorylation by [␥-32 P]ATP-The Ca 2ϩ dependence of phosphorylation with [␥-32 P]ATP was studied at pH 7.0 and 25°C (Fig. 3). This analysis allows a very accurate determination of the apparent affinity (K 0.5 ) for Ca 2ϩ and Hill coefficient. Relative to SERCA1a, the K 0.5 values a Maximum turnover rate for ATP hydrolysis at 37°C in the presence of calcium ionophore A23187 at pH 7.0 (Fig. 2). b The ratio between the maximal ATPase activities with and without calcium ionophore A23187 at pH 7.0 (Fig. 2). c The Ca 2ϩ concentration giving half-maximal activation of ATPase activity at 37°C in the presence of calcium ionophore A23187 (Fig. 2). d The Ca 2ϩ concentration giving half-maximal activation of phosphorylation by ͓␥-32 P͔ATP at 25°C (Fig. 3). e Rate constants corresponding to appearance of ability to phosphorylate with ͓␥-32 P͔ATP at 25°C upon addition of 100 M CaCl 2 to the Ca 2ϩ -deprived enzyme and the initial phosphorylation levels at time 0 (Fig. 4). f Rate constants corresponding to disappearance of ability to phosphorylate with ͓␥-32 P͔ATP at 25°C upon addition of EGTA (Fig. 5). The ordinate shows the molecular turnover rate for ATP hydrolysis estimated as the ratio between the ATP hydrolysis rate/mg of microsomal protein and the maximum capacity for phosphorylation with [␥-32 P]ATP/mg of microsomal protein (active site concentration). The lines show the best fits of a function with two Hill components (one with a negative sign) to the data. By fitting a single Hill component to the rising phase, the K 0.5 (M Ϯ S.E.) and Hill coefficients (n H ) were obtained and listed in TABLE ONE together with the maximum turnover rates. NOVEMBER 25, 2005 • VOLUME 280 • NUMBER 47 obtained for SPCA1 enzymes (TABLE ONE) were ϳ30-fold decreased, consistent with the data from the Ca 2ϩ titrations of the ATPase activities presented above. Furthermore, the Hill coefficient values for the human SPCA1 isoforms (Ͻ1 in contrast to the value of 1.8 for SERCA1a, see TABLE ONE) are also in agreement with the data from Fig. 2 supporting the notion of a single cytoplasmically facing high affinity Ca 2ϩ binding site per human SPCA1 molecule.

Kinetics of Human SPCA1 Isoforms
Rapid Kinetic Analyses of Ca 2ϩ Binding and Phosphorylation-To obtain further information on the Ca 2ϩ binding transition (forward reactions 1 and 2 in Scheme 1), Ca 2ϩ dissociation (reverse reaction 2 in Scheme 1), and phosphorylation (reaction 3 in Scheme 1) properties of the human SPCA1 enzymes at 25°C, rapid kinetic measurements of the phosphorylation with [␥-32 P]ATP were conducted using the quenchflow methodology previously described and validated for SERCA1, SERCA2, and SERCA3 isoforms (31)(32)(33)(34).
The rate constant of the Ca 2ϩ binding transition was determined according to the double mixing protocol shown as a diagram at the top of Fig. 4. This procedure takes advantage of the fact that only Ca 2ϩbound enzyme is able to be phosphorylated from ATP. Accordingly, the Ca 2ϩ -deprived enzyme is incubated with saturating Ca 2ϩ (100 M final free Ca 2ϩ concentration) for a variable period of time (t in Fig. 4) followed by a 50-ms incubation with 5 M [␥-32 P]ATP before acid quenching. The results obtained for SPCA1 and SERCA1a isoforms are shown in Fig. 4. The rate constants corresponding to the Ca 2ϩ binding transition (determined by fitting a monoexponential function with an initial offset) are presented in TABLE ONE together with the initial offset, which depends on the initial amount of enzyme present as E 1 and the amount of E 1 formed from E 2 during the 50-ms incubation with Ca 2ϩ and [␥-32 P]ATP. SPCA1a and SERCA1a displayed rather similar rate constants (8.64 s Ϫ1 versus 11.44 s Ϫ1 ). By contrast, the rate constants determined for human SPCA1b and SPCA1d were ϳ2-fold enhanced. The initial offset values for the SPCA1 isoforms were only slightly higher than that of SERCA1a (TABLE ONE).
The Ca 2ϩ interaction of the SPCA1 pumps in the E 1 form was further investigated by determining the rate of Ca 2ϩ dissociation (k ϪCa ) from the Ca 2ϩ -bound E 1 form, again taking advantage of the fact that only Ca 2ϩ -bound enzyme is able to be phosphorylated from ATP. In this procedure (31)(32)(33)(34), the level of phosphoenzyme (EP ATPϩEGTA ) is measured 34 ms after the simultaneous addition of excess EGTA (to initiate Ca 2ϩ dissociation) and [␥-32 P]ATP to the enzyme pre-equilibrated with saturating Ca 2ϩ , and compared with the phosphoenzyme level (EP ATP ) measured after 34 ms incubation with [␥-32 P]ATP in the presence of Ca 2ϩ , i.e. without allowing Ca 2ϩ to dissociate. The data obtained at 25°C and pH 7.0 with the SPCA1 and SERCA1a enzymes are shown in Fig. 5, and TABLE ONE displays the rate constants for Ca 2ϩ dissociation derived as previously explained (31), taking into consideration the difference in phosphorylation rate constant (see below). Because both Ca 2ϩ sites must be occupied for activation of phosphorylation in   SERCA1a, the Ca 2ϩ dissociation assay monitors only the first Ca 2ϩ dissociation step in this enzyme, corresponding to the Ca 2ϩ ion bound last, at site II (31). Because this is the site equivalent to that present in SPCA1, it is reasonable to compare directly the Ca 2ϩ dissociation rates determined for SPCA1 isoforms with that of SERCA1a. The rates of Ca 2ϩ dissociation for SPCA1a, SPCA1b, and SPCA1d were, respectively, 1.61-, 1.27-, and 1.36-fold enhanced relative to SERCA1a. These results clearly demonstrate that the high apparent affinity of any functional SPCA1 isoform for cytosolic Ca 2ϩ (low K 0.5 value) manifested during the Ca 2ϩ titration of ATPase (cf. Fig. 2) and phosphorylation (cf. Fig. 3) activities does not reflect an increase of the intrinsic (true) affinity of the E 1 form for Ca 2ϩ .
The time course of phosphorylation was monitored with the enzyme initially present in the Ca 2ϩ -saturated form (Fig. 6). As previously described (31,33), the phosphorylation profile for SERCA1a displayed a slight overshoot that could be accurately reproduced by computation (shown as a line in Fig. 6). As indicated in Fig. 6, no phosphorylation overshoot was present for the SPCA1 enzymes, as the data could be fitted by monoexponential functions (lines in Fig. 6), and the rate constants were 3.4-, 2.7-, and 2.8-fold lower for SPCA1a, SPCA1b, and SPCA1d, respectively, relative to the phosphorylation rate constant of SERCA1a, see TABLE TWO. The lack of phosphorylation overshoot can be accounted for by the reduced phosphorylation rate. In addition, the reduced phosphoenzyme turnover rate (see below, Fig. 8) may also be instrumental in removing the overshoot.
ATP Dependence of Phosphorylation-To further examine the reduced rate of phosphorylation of the SPCA isoforms, the ATP concentration dependence of phosphorylation with [␥-32 P]ATP was studied, both at steady-state (Fig. 7A) and by determination of initial rates (Fig. 7B). As shown in Fig. 7A, SPCA1 isoforms displayed 2-4.1-fold lower apparent affinities for ATP (increased K 0.5 values in TABLE TWO) than SERCA1a, thus in good agreement with their reduced phosphorylation rates (cf. Fig. 6). The K 0.5 values obtained at steadystate depend on the affinity for ATP, the maximal phosphorylation rate corresponding to saturating ATP concentration (V max ), as well as the phosphoenzyme turnover rate (K 0.5 increases with increasing phosphoenzyme turnover rate). Because the phosphoenzyme turnover rate of SPCA1 isoforms is markedly reduced relative to SERCA1a (see below, Fig. 8), only a reduced affinity for ATP or a reduced V max for phosphorylation can account for the data in Fig. 7A. For SPCA1b and SPCA1d the V max for phosphorylation was determined by examining the ATP concentration dependence of the initial phosphorylation rate at 25°C under the same conditions as those described for Fig. 6. The double reciprocal (Lineweaver-Burk) plots of the initial phosphorylation rate per Ca 2ϩ -ATPase molecule versus the ATP concentration shown in Fig.  7B and the values extracted for K m and V max listed in TABLE TWO showed that: (i) the V max for SPCA1b and SPCA1d was, respectively, 7and 4.9-fold lower than that of SERCA1a, and (ii) the respective K m values were 1.8-and 1.3-fold reduced relative to SERCA1a. Both effects can be explained by a reduced k 2 in the simplified Michaelis-Menten reaction Scheme 2,  (Fig. 6). The phosphorylation reaction denotes the transition from E 1 Ca to E 1 ϳP (Ca) for SPCA1 enzymes and E 1 Ca 2 to E 1 ϳP(Ca 2 ) for SERCA1a. b The ATP concentration giving half-maximal activation of phosphorylation by ͓␥-32 P͔ATP at 0°C when n H is set to 1 (Fig. 7A). c The maximal phosphorylation rate (V max ) and the Michaelis-Menten constant (K m ) were calculated by linear regression from the double reciprocal plots of the initial phosphorylation rate versus ATP concentration (Fig. 7B). d The 32 P i concentration giving half-maximal activation of phosphorylation by 32 P i (Fig. 9A). e The vanadate concentration giving half-maximal inhibition of phosphorylation by ͓␥-32 P͔ATP at 0°C (Fig. 9B). f ND, not determined. where K m ϭ (k Ϫ1 ϩ k 2 )/k 1 , and k 2 ϭ V max per Ca 2ϩ -ATPase molecule. Because the observed K m values for SPCA1b and SPCA1d are reduced (as expected because of the reduced k 2 ) rather than increased, relative to SERCA1a, there is no evidence for a reduction in SPCA1b and SPCA1d of the true affinity for ATP (increase of the dissociation constant for the enzyme-ATP complex, K D ϭ k Ϫ1 /k 1 ).
Dephosphorylation of the Phosphoenzyme Intermediate Formed from ATP-As illustrated in Fig. 8, the decomposition of phosphoenzyme intermediate was studied after phosphorylation with [␥-32 P]ATP. Phosphorylation was carried out at neutral pH, 0°C, and presence of K ϩ , i.e. under conditions known for SERCA1a, SERCA2, and SERCA3 isoforms to lead to the accumulation of the high energy ADP-sensitive E 1 ϳP(Ca 2 ) intermediate (33,34). For SPCA1, it is expected that the phosphoenzyme intermediate E 1 ϳP(Ca) contains only a single occluded Ca 2ϩ . These intermediates were chased either with excess ATP and EGTA (Fig. 8A) leading to dephosphorylation in the forward direction of the reaction cycle (via reactions 4 and 5 in Scheme 1) or with excess ADP and EGTA (Fig. 8B) leading to dephosphorylation in the backward direction with formation of ATP via the transfer of the phosphoryl group to ADP (reverse reaction 3 in Scheme 1). The rate constants are shown in TABLE THREE. Following the ATP/EGTA chase, SPCA1a, SPCA1b, and SPCA1d isoforms displayed dephosphorylation rates that were, respectively, 14.2-, 12.0-, and 9.0-fold slower than that obtained for SERCA1a. Such pronounced reductions of the rate of dephosphorylation can explain the observed differences in the maximal ATPase turnover rate (cf. Fig. 2 and TABLE ONE) and the high apparent affinity for Ca 2ϩ (cf. Figs. 2 and 3 and TABLE ONE) described above. It is furthermore interesting that the slightly different dephosphorylation rates of the SPCA1 isoforms appeared to correlate quite well with the length of their alternatively spliced C-terminal part: the isoform with the shortest C terminus (SPCA1a) dephosphorylated slower than the other SPCA1 isoforms and SPCA1d (the longest) was the fastest. It is also noteworthy that the dephosphorylation rates obtained following the ADP/EGTA chase of the SPCA1 E 1 ϳP(Ca) intermediates were 2.3-4.1-fold reduced relative to SERCA1a (Fig. 8B). This represents significant reductions in the rate of reverse partial reaction 3 (Scheme 1), thus matching the decreased rates observed for the phosphorylation step (forward reaction 3 in Scheme 1, cf. Fig. 6).
Phosphoenzyme Formation from 32 P i and Its Dephosphorylation-Backdoor phosphorylation with 32 P i (reverse reaction 5 in Scheme 1) was analyzed for SPCA1 isoforms under conditions (acid pH, absence of Ca 2ϩ and alkali metal ions, but presence of dimethyl sulfoxide) known for SERCA1a (36) to increase its affinity for P i and to result in the formation of a rather stable ADP-insensitive E 2 -P phosphoenzyme product. As seen in Fig. 9A, SPCA1 isoforms displayed rather similar apparent affinities for 32 P i , which were 5.5-6.0-fold reduced (larger K 0.5 values in TABLE TWO) relative to SERCA1a.
The decay of the E 2 -P intermediate formed by phosphorylation with 32 P i under conditions described in relation to Fig. 9A was monitored (Fig. 9, B-E and TABLE THREE) following its dilution in dephosphorylation media of varying pH at 0°C in the presence or absence of K ϩ . For SERCA1a, the E 2 -P dephosphorylation becomes rate limiting for the overall activity at acidic or alkaline reaction conditions, but it is very  rapid at neutral pH (34,37,38). Moreover, the dephosphorylation rate depends on the presence of monovalent alkali metal ions, being accelerated more than 10-fold by addition of K ϩ (38,39). Surprisingly, the results illustrated in Fig. 9, B-E, with the rate constants being shown in TABLE THREE, demonstrated that the E 2 -P dephosphorylation in SPCA1 enzymes was markedly accelerated relative to SERCA1a under all conditions examined. In the absence of K ϩ , three pH conditions were tested, and the acceleration relative to SERCA1a was 73-93-fold at pH 6.0, 4 -14-fold at pH 7.0, and 68 -238-fold at pH 8.5. There was little influence of pH on the dephosphorylation of SPCA1 isoforms, and only SPCA1d resembled SERCA1a in displaying a higher dephosphorylation rate at pH 7.0 compared with pH 6.0 and pH 8.5. The addition of K ϩ at pH 6.0 resulted in an 11-fold increase in the rate of E 2 -P dephosphorylation for SERCA1a (cf. Fig. 9, B and C), but K ϩ did not accelerate the E 2 -P dephosphorylation in the SPCA1 isoforms (cf. Fig. 9, B and C). In fact, when compared with the decay rates obtained in the absence of K ϩ (TABLE THREE), it seemed that K ϩ actually exerted a slightly inhibitory effect on the dephosphorylation of E 2 -P in SPCA1 isoforms, which was more pronounced in SPCA1b and SPCA1d isoforms than in SPCA1a. However, the rates with which SPCA1 enzymes dephosphorylated from E 2 -P at pH 6.0 in the presence of 80 mM K ϩ were still 3-5-fold faster than that of SERCA1a (Fig. 9C). It can be concluded that irrespective of pH and K ϩ concentration, the E 2 -P dephosphorylation in SPCA1 isoforms takes place faster relative to SERCA1a, and that the dependence of this step on pH and K ϩ is too weak to be of importance in SPCA1 isoforms.
Inhibition by Vanadate-Vanadate, an analog of phosphate, binds to the E 2 dephosphoenzyme leading to a dead end complex (stable in icecold conditions) believed to resemble the transition state between E 2 -P and E 2 ⅐P i occurring during enzyme turnover. Therefore, any change brought to this transition state or to the E 2 state reacting with vanadate would affect the apparent affinity of the enzyme for vanadate. Vanadate binding was studied under equilibrium conditions (i.e. in the absence of phosphoenzyme turnover) in which vanadate was allowed to bind E 2 in the absence of Ca 2ϩ and ATP (1 h at 25°C). After the incubation, the fraction of phosphorylatable vanadate-free enzyme (E 1 ) was measured by phosphorylation (at 0°C) following the addition of excess Ca 2ϩ and [␥-32 P]ATP as previously described (33). The results shown in Fig. 10 demonstrate that SPCA1a, SPCA1b, and SPCA1d isoforms displayed, respectively, 2.3-, 4.5-, and 4.9-fold lower apparent affinities for vanadate than SERCA1a (higher K 0.5 values in TABLE TWO), thus consistent with the data obtained from the P i titration of P i phosphorylation (cf. Fig. 9A).  i (B-E). A, phosphorylation was carried out for 10 min at 25°C in the presence of 100 mM MES/Tris, pH 6.0, 10 mm MgCl 2 , 2 mM EGTA, 30% (v/v) dimethyl sulfoxide, 0.1 M thapsigargin and the indicated concentrations of 32 P i , followed by acid quenching. Phosphorylation of SERCA1a was performed without thapsigargin. The data were normalized separately by taking as 100% the maximum phosphorylation level reached, and the lines for the indicated proteins show the best fits of the Hill equation with n H ϭ 1. The K 0.5 values (MϮ S.E.) are listed in TABLE TWO. B, phosphorylation with 0.5 mM 32 P i was performed as in A. Following cooling of the sample to 0°C, the phosphoenzyme was chased by a 19-fold dilution of an aliquot into a medium (kept at 0°C) containing 100 mM MES/Tris, pH 6.0, 10 mM MgCl 2 , 2 mM EGTA, and 0.5 mM non-radioactive P i , and acid quenching was performed at the indicated serial time intervals. C, the same as in B, but in addition the dephosphorylation medium contained 80 mM KCl. D-E, the same as in B, but the dephosphorylation buffer contained 100 mM MOPS/ Tris, pH 7.0 (D) or 100 mM Tris/HCl, pH 8.5 (E) instead of 100 mM MES/Tris, pH 6.0. The phosphorylation levels in B-E are shown relative to the level determined by acid quenching after the 10-min phosphorylation at 25°C. The lines represent the best fits of a monoexponential decay function, and the obtained rate constants (s Ϫ1 Ϯ S.E.) are shown in TABLE THREE.  Fig. 8A) or 1.0 mM ADP and 10 mM EGTA (ADP/EGTA chase; Fig. 8B). b Rate constants corresponding to dephosphorylation at 0°C and the indicated pH values (in the presence or absence of potassium ions) of phosphoenzyme formed in the presence of 32 P i (panels B-E in Fig. 9).

DISCUSSION
In the present study, we have performed for the first time a detailed characterization of the reaction cycle mediated by the human SPCA1 isoforms heterologously expressed in mammalian HEK-293 and COS-1 cells. The validity of the mammalian expression systems employed in this study was clearly demonstrated, as SPCA1a, SPCA1b, and SPCA1d were well expressed to levels comparable to SERCA pumps and functionally active (cf. Fig. 1, B and C). The low expression level, the anomalous electrophoretic mobility (slower than expected) and the lack of phosphorylation from [␥-32 P]ATP shown here for SPCA1c (Fig. 1) have provided the first experimental evidence favoring the hypothesis that truncation of transmembrane segment M10 (as a consequence of alternative splicing of the ATP2C1 primary transcript) may generate an improperly folded 888-amino acid protein likely liable to enhanced degradation.
The results of our functional investigations on SPCA1a, SPCA1b, and SPCA1d have been summarized in TABLES ONE to THREE. In comparison with SERCA1a, the human SPCA1 isoforms display essential differences with respect to the overall Ca 2ϩ -ATPase reaction (cf. Fig. 2): (i) much higher apparent affinities for Ca 2ϩ activation at the cytosolic site and (ii) lower catalytic turnover rates. The Hill coefficient values extracted from the data describing the Ca 2ϩ titration of ATP hydrolysis and phosphorylation (Figs. 2 and 3) moreover support the presence of only a single cytoplasmically oriented high affinity ion-binding site per SPCA1 molecule. With regard to the partial reaction steps of the Ca 2ϩ transport cycle illustrated in Scheme 1, the human SPCA1 isoforms also display kinetic properties distinct from SERCA1a as follows: (i) a markedly reduced maximum rate (V max ) of phosphorylation (E 1 Ca to E 1 ϳP(Ca) transition, cf. Figs. 6 and 7B); (ii) a markedly reduced rate of the E 1 ϳP(Ca) to E 2 transition (decreasing with the length of the alternatively spliced C terminus, cf. Fig. 8A); (iii) a markedly increased rate of E 2 -P dephosphorylation, which furthermore does not show the dependence on pH and K ϩ characteristic of SERCA1a (cf. Fig. 9, B-E); (iv) a 2-fold increase of the rate of the Ca 2ϩ binding transition in SPCA1b and SPCA1d (cf. Fig. 4); and (v) a slight increase in the rate of Ca 2ϩ dissociation from E 1 Ca (cf. Fig. 5). As will be discussed below, the characteristics of the partial reaction steps of SPCA1 isoforms can explain the distinct features of their overall reaction described here (cf. Figs. 2, 3, 7A, 9A, and 10).
Increased Apparent Affinity for Cytosolic Ca 2ϩ -Relative to SERCA1a, a 7.5-18.2-fold increase in the apparent affinity for Ca 2ϩ was observed for SPCA1 isoforms by Ca 2ϩ titration of ATPase activity at steady-state (Fig. 2). An even larger difference between SPCA1 isoforms and SERCA1a (ϳ30-fold) was observed for the apparent Ca 2ϩ affinity determined in studies of the Ca 2ϩ dependence of phosphorylation from [␥-32 P]ATP (Fig. 3). In principle, the increased apparent Ca 2ϩ affinities of SPCA1 isoforms could be caused by a shift of the equilibrium between E 2 and E 1 in favor of the E 1 form. As seen in Fig. 4, our rapid kinetic analysis indeed demonstrated a 2-fold increase of the rate of the Ca 2ϩ binding transition, thus in apparent agreement with the E 2 to E 1 equilibrium being shifted toward E 1 . However, it can be generally demonstrated by computer simulation of the Ca 2ϩ dependence of phosphorylation with ATP (32) that the 2-fold increase in the E 2 to E 1 transition rate seen here for both SPCA1b and SPCA1d is too small a change to account for the observed ϳ30-fold decrease of the K 0.5 values for Ca 2ϩ activation of phosphorylation (Fig. 3). Moreover, SPCA1a did not show the increase in the E 2 to E 1 transition rate seen for the other SPCA1 isoforms (TABLE ONE). Therefore, other effects on partial reaction steps must be involved as well. For SERCA2b and SERCA3 enzymes, we recently demonstrated significant changes of the Ca 2ϩ dissociation (E 1 Ca 2 to E 1 Ca transition) rate, which could explain the observed differences from SERCA1a with respect to the apparent affinity of the E 1 form for cytosolic Ca 2ϩ (33,34). Therefore, we expected to find a reduced Ca 2ϩ dissociation rate for SPCA1 isoforms correlating with the increased apparent Ca 2ϩ affinity. Surprisingly, the rates of Ca 2ϩ dissociation ( Fig. 5 and TABLE ONE) from the Ca 2ϩ -bound E 1 form (E 1 Ca) found in the SPCA1 isoforms were instead slightly enhanced relative to SERCA1a and, thus, did not explain the enhanced apparent Ca 2ϩ affinity of SPCA1 enzymes.
In SERCA1a, the transition from E 1 ϳP(Ca 2 ) to E 2 -P constitutes the most rate-limiting step of the cycle at physiological pH and substrate concentrations. The ensuing hydrolysis reaction E 2 -P 3 E 2 is much faster than the E 1 ϳP(Ca 2 ) to E 2 -P transition at neutral pH and presence of K ϩ . Computation based on the reaction cycle of SERCA1a demonstrated that acceleration or slowing of the rate-limiting E 1 ϳP(Ca 2 ) to E 2 -P transition causes, respectively, a decreased or increased apparent affinity for Ca 2ϩ ions (32). Hence, the observed slow processing of the ADP-sensitive E 1 ϳP(Ca) phosphoenzyme (cf. Fig. 8A and TABLE THREE) seems to be a major determinant of the high apparent affinity for Ca 2ϩ displayed by all SPCA1 isoforms (when phosphoenzyme turnover is slow, more phosphoenzyme is accumulated at low Ca 2ϩ concentrations, thus shifting the Ca 2ϩ activation curve toward lower K 0.5 values). Because the dephosphorylation of E 2 -P is rapid in SPCA1 isoforms, the E 1 ϳP(Ca) to E 2 -P transition must be the slow step in the E 1 ϳP(Ca) processing.
The finding that the high apparent affinity for Ca 2ϩ displayed by SPCA1 isoforms owes to a "kinetic effect" rather than a true increase of affinity relative to SERCA1a has implications for understanding the structure of the Ca 2ϩ binding site of the SPCA1 isoforms. Our data are consistent with the hypothesis that the Ca 2ϩ binding site of SPCA1 is similar to site II in SERCA1a, i.e. the site composed of side chains from Glu 309 , Asn 796 , and Asp 800 , which have counterparts, Glu 308 , Asn 738 , and Asp 742 , in SPCA1. Because the rate of Ca 2ϩ dissociation in SPCA1 isoforms is comparable to that observed for SERCA1a, it seems likely that residue Asp 742 in SPCA1 contributes with only one of its two oxygen ligands to the site, exactly like its counterpart (Asp 800 ) in SERCA1a.
Lower Turnover Rate-We showed here for the first time that the SPCA1 isoforms display much lower catalytic ATPase turnover rates relative to SERCA1a (cf. Fig. 2 and TABLE ONE). The increase in the catalytic turnover rate upon addition of the calcium ionophore A23187 demonstrated that the E 1 ϳP(Ca) to E 2 -P transition plays indeed a rate- limiting role in SPCA1 enzymes (TABLE ONE), because this transition should be sensitive to the back inhibition exerted by Ca 2ϩ accumulated inside the microsomal vesicles (in the millimolar concentration range). Even in the presence of the calcium ionophore, the maximal turnover rates of the ATPase activity for SPCA1 isoforms were 4.7-6.4-fold lower than that of SERCA1a (TABLE ONE). The lower overall rate of ATP hydrolysis in SPCA1 can be explained by the slow transition of E 1 ϳP(Ca) to E 2 -P. It is interesting that both the rate of ATP hydrolysis and the rate of phosphoenzyme processing increased with the length of the alternatively spliced C terminus, i.e. in the order SPCA1a, SPCA1b, and SPCA1d.
Our rapid kinetic measurements indicated that the SPCA1 enzymes are characterized by decreased phosphorylation rates (TABLE TWO). However, the rate limitation of the overall ATPase reaction imposed by the latter effect is insignificant in comparison with that imposed by slowing of the subsequent E 1 ϳP(Ca) to E 2 -P transition, because phosphorylation basically is one of the fastest steps of the cycle. Our examination of the ATP concentration dependence of the initial phosphorylation rate data showed that the reduced phosphorylation rate is a V max effect rather than a consequence of reduced affinity for the substrate. Interestingly, the reverse transfer of the phosphoryl group from E 1 ϳP(Ca) to ADP was also slower in the SPCA1 isoforms compared with SERCA1a ( Fig. 8B and TABLE THREE), indicating a difference with respect to catalytic ability in both directions.
Enhanced (pH-and K ϩ -independent) E 2 -P Dephosphorylation-The functional properties displayed by SPCA1 enzymes when present in E 2 /E 2 -P conformation were also distinct from those of SERCA1a. Hence, the apparent affinities for reaction of E 2 with inorganic phosphate and the phosphate analog vanadate were both found reduced (K 0.5 increased) relative to SERCA1a (Figs. 9A and 10, and TABLE TWO). The kinetic studies showed that the rate of E 2 -P dephosphorylation was markedly enhanced in SPCA1 isoforms, relative to SERCA1a, thus explaining the reduced apparent affinity for P i . It is conceivable that the same factors that are responsible for destabilization of E 2 -P also destabilize the vanadate-bound E 2 form. The acceleration of dephosphorylation was observed irrespective of the presence of K ϩ and pH, and, unlike SERCA1a, the SPCA1 isoforms showed little dependence of the E 2 -P dephosphorylation on K ϩ and pH. The inhibition of E 2 -P dephosphorylation seen for SERCA1a at alkaline pH may in part be ascribed to lack of binding of protons (to be countertransported) at the transport sites in their luminally facing configuration (38). Hence, it seems that the SPCA1 isoforms either bind the protons with an unusually high affinity compatible with activation at pH values as high as 8.5, or have no need for such proton binding for activation of E 2 -P dephosphorylation. Such an effect might be a consequence of the lack in SPCA1 of residues equivalent to Glu 771 and Glu 908 at Ca 2ϩ site I in SERCA1a, as these residues might be involved in the proton countertransport and/or its regulation in SERCA1a.
It is well known for SERCA1a that monovalent alkali metal ions, in particular K ϩ , accelerate the E 2 -P dephosphorylation rate (38,39). X-ray crystallography of SERCA1a recently demonstrated a binding site for a K ϩ ion located in domain P (12,40,41). Mutational analysis showed that Glu 732 (contributing to the binding site) is of major importance for the stimulatory effect of K ϩ (40). For SPCA1 isoforms, the experiments illustrated in Fig. 9, B and C clearly demonstrated that addition of K ϩ to the dephosphorylation medium did not increase the rate of E 2 -P dephosphorylation. Interestingly, the SPCA1 residue at the position equivalent to Glu 732 in SERCA1a is Asp 674 . Mutagenesis exper-iments with SERCA1a 4 indicate that this subtle difference (only the length of the side chain differs between glutamate and aspartate) is sufficient to attenuate the K ϩ regulatory effect. It is furthermore interesting that Pro 709 of SERCA1a is substituted by Val 650 in SPCA1 isoforms. In SERCA1a, this proline residue appears critical for the formation of a loop (residues 710 -715), which together with Glu 732 contributes to coordination of K ϩ via backbone carbonyls (40). Thus, it is possible that in SPCA1 isoforms K ϩ is not bound, or is bound only weakly relative to SERCA1a.
The constitutive high activation of E 2 -P dephosphorylation in SPCA1 isoforms could be a necessary mechanistic consequence associated with the slowness of the E 1 ϳP(Ca) to E 2 -P step, because the involved destabilization of the E 2 -P conformation inevitably leads to an enhanced E 2 -P dephosphorylation, thus precluding any regulation. This would be similar to certain SERCA1a mutants with alterations to residues in the linker regions connecting the cytoplasmic domains with the membrane, where a reduced rate of the E 1 ϳP(Ca 2 ) to E 2 -P transition was found associated with enhanced E 2 -P dephosphorylation (42,43).
Physiological Relevance-The biochemical properties of the SPCA1 pumps revealed in the present study appear to be compatible with its function in the secretory pathway. A high Ca 2ϩ transport rate is not actually required in the SPCA1-containing Golgi compartment, which appears to be insensitive to inositol trisphosphate-generating agonists and is depleted of Ca 2ϩ only by the probably slower process of downstream trafficking of Ca 2ϩ -rich vesicles. In these non-releasable Ca 2ϩ stores, a high affinity and low turnover rate pump such as SPCA1 can be expected to render the adjustment of the luminal Ca 2ϩ level less dependent on the cytosolic Ca 2ϩ concentration, mainly by limiting the fluctuation of pumping activity during cytosolic Ca 2ϩ transients. In contrast, the function of SERCA2b in the endoplasmic reticulum and SERCA1a in the sarcoplasmic reticulum appears to be rather well adapted to the faster Ca 2ϩ replenishment of releasable Ca 2ϩ stores and the decline of cytosolic Ca 2ϩ transients after cell stimulation.