Dissection of the Functional Differences between Sarco(endo)plasmic Reticulum Ca2+-ATPase (SERCA) 1 and 2 Isoforms and Characterization of Darier Disease (SERCA2) Mutants by Steady-state and Transient Kinetic Analyses*

Steady-state and rapid kinetic studies were conducted to functionally characterize the overall and partial reactions of the Ca2+ transport cycle mediated by the human sarco(endo)plasmic reticulum Ca2+-ATPase 2 (SERCA2) isoforms, SERCA2a and SERCA2b, and 10 Darier disease (DD) mutants upon heterologous expression in HEK-293 cells. SERCA2b displayed a 10-fold decrease in the rate of Ca2+ dissociation from E1Ca2 relative to SERCA2a (i.e. SERCA2b enzyme manifests true high affinity at cytosolic Ca2+ sites) and a lower rate of dephosphorylation. These fundamental kinetic differences explain the increased apparent affinity for activation by cytosolic Ca2+ and the reduced catalytic turnover rate in SERCA2b. Relative to SERCA1a, both SERCA2 isoforms displayed a 2-fold decrease of the rate of E2 to E1Ca2 transition. Furthermore, seven DD mutants were expressed at similar levels as wild type. The expression level was 2-fold reduced for Gly23 → Glu and Ser920 → Tyr and 10-fold reduced for Gly749 → Arg. Uncoupling between Ca2+ translocation and ATP hydrolysis and/or changes in the rates of partial reactions account for lack of function for 7 of 10 mutants: Gly23 → Glu (uncoupling), Ser186 → Phe, Pro602 → Leu, and Asp702 → Asn (block of E1∼P(Ca2) to E2-P transition), Cys318 → Arg (uncoupling and 3-fold reduction of E2-P to E2 transition rate), and Thr357 → Lys and Gly769 → Arg (lack of phosphorylation). A 2-fold decrease in the E1∼P(Ca2) to E2-P transition rate is responsible for the 2-fold decrease in activity for Pro895 → Leu. Ser920 → Tyr is a unique DD mutant showing an enhanced molecular Ca2+ transport activity relative to wild-type SERCA2b. In this case, the disease may be a consequence of the low expression level and/or reduction of Ca2+ affinity and sensitivity to inhibition by lumenal Ca2+.

In this case, the disease may be a consequence of the low expression level and/or reduction of Ca 2؉ affinity and sensitivity to inhibition by lumenal Ca 2؉ .
Sarco(endo)plasmic reticulum Ca 2ϩ -ATPases (SERCAs) 1 are single-subunit integral membrane P-type ATPases that mediate the ATP-driven transport of cytosolic Ca 2ϩ against a concentration gradient into the lumen of intracellular Ca 2ϩ -releasable stores such as sarcoplasmic and endoplasmic reticulum (1)(2)(3)(4). SERCAs belong to the P-type ATPase family distinguished by the obligatory formation of an aspartyl-phosphorylated intermediate as part of their catalytic cycle. The enzyme cycles reversibly between several states (Scheme 1), of which at least E 1 Ca 2 , E 1 ϳP(Ca 2 ), E 2 -P, and E 2 can be experimentally distinguished. The transfer of the ␥-phosphoryl group of ATP to the aspartate (Asp 351 ) of the phosphorylation domain, leading to the ADP-sensitive high energy phosphoenzyme intermediate E 1 ϳP(Ca 2 ), is activated by conformational changes associated with the binding of two calcium ions in exchange for protons (E 2 to E 1 Ca 2 transition). The conversion of this intermediate to the ADP-insensitive low-energy E 2 -P phosphoenzyme intermediate constitutes a crucial rate-limiting step in Ca 2ϩ translocation (1)(2)(3). High resolution models for the atomic structure, generated by x-ray crystallography of crystals in E 1 Ca 2 (5) and Ca 2ϩ -free E 2 (6) forms, and extensive mutational studies (1,(7)(8)(9)(10)(11) of the 110-kDa SERCA1a enzyme (994 amino acids) have shown that Ca 2ϩ translocation and ATP utilization are coupled through long range intramolecular interactions between the 10-helix membrane-spanning domain, harboring the Ca 2ϩ binding sites, and the large cytosolic head consisting of actuator (A), phosphorylation (P), and nucleotide-binding (N) domains (Fig. 1).
The three human SERCA genes (ATP2A1, ATP2A2, and ATP2A3) encode up to a total of 10 isoforms as a result of the alternative splicing of their pre-mRNA (12)(13)(14)(15)(16). Mutations in SERCA genes have been recently detected in human diseases such as Brody disease (muscle disorder) for SERCA1 (17) and non-insulin dependent type-II diabetes mellitus for SERCA3 (18). Previous clinical and genetical investigations have established that the SERCA2 gene (ATP2A2) represents the gene defective in Darier disease (DD; OMIM 124200), a severe autosomal dominant skin disorder characterized by loss of adhesion between epidermal cells (acantholysis) and by abnormal keratinization (dyskeratosis) and association with a wide range of neuropsychiatric problems, such as epilepsy and depression (19). It has been proposed that mutations in the SERCA2 gene produce a dominant DD phenotype through haploinsufficiency, i.e. mutation in one allele would result in complete or partial loss of function in the mutated pumps (19). Over 100 nonsense and missense mutations and in-frame deletions spanning all regions of the SERCA2 gene have been detected in DD patients (19, 20 -26), but no clear genotype-phenotype correlations have been drawn. We recently documented that acrokeratosis verruciformis of Hopf (localized disorder of keratinization affecting the distal extremities) is also caused by mutation in the SERCA2 gene, and, hence, it is allelic to DD (27). The SERCA2 gene encodes the cardiac muscle SERCA2a protein (997 amino acids) as well as the ubiquitously expressed SERCA2b (1042 amino acids), but no cardiac manifestations have been observed in DD patients. As a result of the alternative splicing, the SERCA2a-specific C terminus (AILE 997 ) is replaced by a variant sequence of 49 amino acids in SERCA2b (Fig. 1) and by a 6-amino acid stretch (VLSSEL 999 ) in SERCA2c, a very recently reported SERCA2 splice variant (16). The extended tail of the SERCA2b isoform probably contains an additional transmembrane domain (M11) as part of a SERCA2b-specific 11-helix transmembrane structure with the extreme C terminus protruding into the endoplasmic reticulum lumen (28). A tentative planar model is shown in Fig. 1, in which the indicated secondary structures correspond to the model for the atomic structure of SERCA1a in the E 1 Ca 2 state (5). The divergence in the C-terminal part is responsible for the known functional differences between SERCA2a and SERCA2b: SERCA2b has a 2-fold higher apparent affinity for Ca 2ϩ and a 2-fold lower turnover rate for Ca 2ϩ -uptake relative to SERCA2a (29,30). Upon cell stimulation, the SERCA enzymes play a critical role in restoring the cytosolic Ca 2ϩ concentration to its resting levels. SERCA2b is the major SERCA isoform in non-muscle cells, including epidermal cells, and the specific functional properties of SERCA2b have direct consequences for modulation of the frequency and amplitude of the generated Ca 2ϩ waves (31,32). Comparison of Ca 2ϩ wave properties of oocytes overexpressing either SERCA2a or SERCA2b has demonstrated that both the width and the period of the Ca 2ϩ waves are larger for SERCA2b than for SERCA2a, and, consistent with a higher Ca 2ϩ affinity of SERCA2b, the wave fronts are more sharply delineated for SERCA2b (31). Replenishment of the agonist-releasable endo-plasmic reticulum Ca 2ϩ stores is, furthermore, required for proper translation and folding of resident and secreted proteins (33,34), and lumenal Ca 2ϩ appears to modulate cell sensitivity to apoptosis (35). It is possible that both in Darier disease and in the related Hailey-Hailey disease caused by mutations in the gene encoding the Golgi Ca 2ϩ -ATPase (36,37), the major pathogenic mechanism is the presence of a low lumenal Ca 2ϩ concentration in endoplasmic reticulum and Golgi, which could result in defective processing of the newly synthesized proteins required for normal adhesion between epithelial cells in the adult skin. A very recent study, analyzing the effect of 12 mutations associated with DD, has documented that some mutants inhibited the activity of the co-expressed wild-type through protein interactions between wild-type and mutant SERCA2b monomers (38). So far, the functional effects of DD mutations on overall and partial reactions of the catalytic cycle mediated by mutant SERCA2b proteins have not been documented in any detail. In addition, the partial reaction steps responsible for the SERCA2bspecific slow catalytic cycle (29,30) have not yet been identified. The present study addresses these issues by investigating the overall and partial reactions catalyzed by wild-type SERCA2a and SERCA2b isoforms as well as 10 DD SERCA2b mutants (Fig. 1). Because this represents the first detailed study of the partial reaction steps of SERCA2 isoforms, a comparison with the well characterized SERCA1a is also made.

EXPERIMENTAL PROCEDURES
The cDNA clones used encode the human SERCA2a and SERCA2b isoforms (13), and rabbit SERCA1a (39). All SERCA2b mutants were generated using the QuikChange TM site-directed mutagenesis kit from Stratagene (La Jolla, CA). Transfection of HEK-293 cells with the cDNA in expression vector pMT2 or pcDNA3.1(ϩ) (Invitrogen) was performed using the calcium phosphate precipitation method (40). The microsomal fraction containing expressed wild-type or mutant Ca 2ϩ -ATPase was isolated by differential centrifugation (41). Protein concentration determination, denaturing gel electrophoresis, semi-dry blotting, and blot immunostaining were performed as reported earlier (42,43). All methods used for the analysis of the catalytic cycle in steadystate and transient-kinetic conditions have been previously established from studies with SERCA1a mutants (7-10, 44 -46) and used very recently to characterize several SERCA3 isoforms (47). These procedures including the quench-flow methodology (8,10), which allows rapid kinetic measurements to be performed on a millisecond scale, were directly applicable to expressed wild-type and mutant SERCA2 enzymes. Further details of the functional assays are given in the figure legends. All data presented are average values corresponding to two to seven experiments and standard errors larger than the symbols are shown as error bars in the figures. Experimental data were fitted by nonlinear regression analysis using the SigmaPlot program (SPSS Inc.) or by means of the kinetic simulation software SimZyme developed in the Department of Physiology, University of Aarhus, Denmark, as described (8). All values extracted for K 0.5 , Hill coefficient, and different rate constants are listed in Tables I and II. Generally, the best fits are shown as lines in the figures only for non-overlapping curves. Several figures, additional explanations, and discussion are deposited as Supplemental Materials.

RESULTS
Expression and Phosphorylation Levels-In the present study, the human wild-type SERCA2a and SERCA2b isoforms as well as SERCA2b mutants Gly 23 3 Glu, Ser 186 3 Phe, Cys 318 3 Arg, Thr 357 3 Lys, Pro 602 3 Leu, Asp 702 3 Asn, Gly 749 3 Arg, Gly 769 3 Arg, Pro 895 3 Leu, and Ser 920 3 Tyr were examined. These SERCA2b mutations have been previously documented in DD patients (19 -21, 23, 27), except for Pro 895 3 Leu, which is a novel DD mutation. 2 For comparison, the well characterized rabbit wild-type SERCA1a isoform was SCHEME 1. Ca 2؉ -ATPase reaction cycle. E 1 , enzyme form with cytoplasmically facing high affinity Ca 2ϩ sites; E 2 , enzyme form with low affinity for Ca 2ϩ ; E 1 ϳP(Ca 2 ), phosphoenzyme with high energy phosphoryl group (transferable to ADP) and occluded calcium ions (shown in parentheses); E 2 -P, phosphoenzyme with low energy phosphoryl group and lumenally facing low affinity Ca 2ϩ sites. included in some of the experiments of the present study. Furthermore, the mutation Ser 920 3 Tyr was also introduced in SERCA2a, to examine its effect in the presence of the shorter C terminus (cf. Fig. 1). Below, the two Ser 920 3 Tyr mutants will be denoted Ser 920 3 Tyr-2a and Ser 920 3 Tyr-2b. Fig. 2A compares the expression levels of wild-type SERCA2b and its corresponding DD mutants by Western blot analysis of microsomal proteins from transfected HEK-293 cells, using a polyclonal antibody raised against an epitope corresponding to the last 12 amino acids of the pig SERCA2b C terminus (42). In the HEK-293 cells, most mutants appeared to be expressed at levels similar to wild-type SERCA2b, but a consistent reduction in expression was found for Gly 23 3 Glu, Gly 749 3 Arg, and Ser 920 3 Tyr-2b in several experiments. This was further confirmed by the functional approach illustrated in Fig. 2B, showing results of measurement of phosphorylation from [␥-32 P]ATP in the presence of Ca 2ϩ (forward reaction 3 in Scheme 1), or from 32 P i in the absence of Ca 2ϩ (reverse reaction 6 in Scheme 1). Maximal phosphorylation levels of 150 -250 pmol of Ca 2ϩ -ATPase/mg of protein were measured for wildtype SERCA2b as well as for the SERCA1a and SERCA2a isoforms, i.e. several hundred-fold higher than that of endogenous HEK-293 cell SERCA2b protein (0.5 pmol of enzyme/mg of protein). Similar levels were obtained for the DD mutants Ser 186 3 Phe, Cys 318 3 Arg, Pro 602 3 Leu, Asp 702 3 Asn, and Pro 895 3 Leu. Relative to wild-type SERCA2b, the maximum phosphorylation level for Gly 23 3 Glu, Gly 749 3 Arg, and Ser 920 3 Tyr-2b, was reduced to 54, 12, and 57%, respectively, FIG. 1. Schematic presentation of the primary and putative secondary structure of the human SERCA2 isoforms highlighting the Darier disease mutants studied. The model is based on the E 1 Ca 2 crystal structure of SERCA1a (5). Each circle corresponds to an amino acid residue indicated by the single-letter code inside the circle, ␣-helical structures are shown as stacked diagonal rows of three or four residues, ␤-strands are displayed as ladder-type residue arrangements, and linear sections represent loops. M1-M10 denote the membrane-spanning helices in SERCA2a. SERCA2a and SERCA2b isoforms are identical in the first 993 amino acid residues (Pro 993 represents the divergence point), but differ in the C terminus: the SERCA2a-specific sequence AILE 997 (purple circles with white letters) is replaced in SERCA2b by a 49-amino acid tail (green circles with white letters), which could form an additional transmembrane segment (M11). Amino acid substitutions in Darier disease patients analyzed in the present study are highlighted (red circles with white letters) and include Gly 23  which seems to correspond to the relative expression levels seen by Western blot analysis in Fig. 2A. Both Thr 357 3 Lys and Gly 769 3 Arg are well expressed, but unable to phosphorylate with either [␥-32 P]ATP or 32 P i . The protein expression level and phosphorylation properties of Ser 920 3 Tyr-2a were indistinguishable from those of Ser 920 3 Tyr-2b (data not shown).
Ca 2ϩ Transport Activity-The ability to transport Ca 2ϩ actively in the presence of ATP was assessed by measurement of oxalate-supported 45 Ca 2ϩ accumulation in the microsomal vesicles. Mutants Pro 895 3 Leu and Ser 920 3 Tyr-2b were able to accumulate Ca 2ϩ , but the remaining DD mutants completely lacked Ca 2ϩ uptake activity ( Fig. 2C and Table I). In Table I, the amount of Ca 2ϩ accumulated per second during a 10-min incubation at 27°C at saturating Ca 2ϩ concentration is shown relative to the maximum phosphorylation level from Fig. 2B, thus providing an estimate of the maximum rate of Ca 2ϩ uptake per enzyme molecule ("molecular Ca 2ϩ transport activity" or "turnover rate"). SERCA2a, like SERCA1a, showed a 2-fold higher maximum molecular Ca 2ϩ transport activity relative to SERCA2b, in good agreement with previous studies (29,30). Interestingly, Ser 920 3 Tyr-2b also displayed 2-fold higher maximum molecular Ca 2ϩ transport activity relative to wildtype SERCA2b. Ser 920 3 Tyr-2a showed a maximum molecular Ca 2ϩ transport activity close to that of Ser 920 3 Tyr-2b and wild-type SERCA2a. Pro 895 3 Leu displayed a decrease to about half of wild-type SERCA2b. Analysis of the Ca 2ϩ dependence of Ca 2ϩ transport (Table I and Fig. A of Supplemental Materials) showed a 1.6-fold increase of the apparent affinity for Ca 2ϩ (decrease of K 0.5 ) for SERCA2b relative to SERCA2a. Ser 920 3 Tyr-2b showed a 3-fold lower apparent affinity for Ca 2ϩ (increase of K 0.5 ) compared with wild-type SERCA2b. For Ser 920 3 Tyr-2a, there was likewise a substantial 3.5-fold lowering of the apparent affinity for Ca 2ϩ relative to the corresponding wild-type SERCA2a.
ATPase Activity-Steady-state ATPase activity was determined for SERCA1a, SERCA2a, SERCA2b, and 7 of the 10 DD mutants (the low expression level of Gly 749 3 Arg and lack of phosphorylation of Thr 357 3 Lys and Gly 769 3 Arg precluded this measurement). For Ser 186 3 Phe, Pro 602 3 Leu, and Asp 702 3 Asn, the level of ATPase activity was indistinguishable from background under all conditions tested (Table I). A significant ATPase activity was found for each of the SERCA2 isoforms and the remaining 4 DD mutants, and Ca 2ϩ titration data obtained in the presence and absence of the calcium ionophore A23187 are shown in Fig. B of Supplemental Materials and summarized in Table I. Consistent with the Ca 2ϩ transport data presented above, SERCA2b displayed higher apparent affinity for Ca 2ϩ at the activating sites than SERCA2a. The addition of calcium ionophore relieves the "back inhibition" of the E 1 ϳP(Ca 2 ) to E 2 -P transition (steps 4 through 5 in Scheme 1) brought about by binding of accumulated Ca 2ϩ at lumenal low affinity sites, because the ionophore allows passive efflux of the Ca 2ϩ actively transported into the microsomal vesicles, thereby maintaining the Ca 2ϩ concentration inside the microsomes at a relatively low level. In this non-inhibited state, SERCA2b showed a 2-fold lower catalytic turnover rate than SERCA2a (compare 35 s Ϫ1 with 70 s Ϫ1 in Table I). Surprisingly, the maximal catalytic turnover rate measured with mutant Gly 23 3 Glu was similar to that of wild-type SERCA2b, even though the Ca 2ϩ accumulation was below the detection limit for this mutant (cf. Fig. 2 and Table I). Consistent with the lack of significant Ca 2ϩ accumulation, there was little stimulation of the maximal rate of ATP hydrolysis upon ionophore addition in mutant Gly 23 3 Glu. Therefore, it appears that the ATP hydrolysis to a large extent is uncoupled from Ca 2ϩ translocation in mutant Gly 23 3 Glu. The apparent affinity for Ca 2ϩ displayed by this mutant was, furthermore, 2-fold lower (K 0.5 higher) than that of wild-type SERCA2b. Cys 318 3 Arg is another DD mutant showing noticeable uncoupling properties. Compared with wild-type SERCA2b, the maximum catalytic turnover rate reached with Cys 318 3 Arg was reduced by 37%, but again no Ca 2ϩ uptake could be measured, even though in this case a significant, although reduced, ionophore stimulation of ATPase activity was noted (Table I). This mutant also showed a 2-fold lower apparent affinity for Ca 2ϩ compared with wild-type SERCA2b. For mutant Pro 895 3 Leu, a reduced catalytic turnover rate and similar apparent affinity for Ca 2ϩ were observed in comparison with SERCA2b, which corresponds with the Ca 2ϩ transport data shown above. For mutant Ser 920 3 Tyr-2b, a 2-fold increase in the maximal catalytic turnover rate relative to wild-type SERCA2b was seen (but similar to that of SERCA2a), and disappearance of the activating effect of calcium ionophore. Like Ser 920 3 Tyr-2b, Ser 920 3 Tyr-2a was insensitive to ionophore, and its turnover rate was similar to that of wildtype SERCA2a (Table I).
Ligand Dependence of Phosphorylation Levels-The phosphoryl transfer from [␥-32 P]ATP requires the binding of two calcium ions at the cytoplasmically facing high affinity sites. For the wild-type enzymes and mutants capable of forming a phosphoenzyme by reaction with [␥-32 P]ATP (cf. Fig. 2B), Ca 2ϩ titration of the phosphorylation reaction was carried out (  Table II shows a summary of the data. SERCA2b displayed an almost 2-fold higher apparent Ca 2ϩ affinity than both SERCA2a and SERCA1a isoforms. Mutants Gly 23 3 Glu, Cys 318 3 Arg, and Asp 702 3 Asn showed a somewhat reduced apparent affinity for Ca 2ϩ relative to wild-type SERCA2b. Mutants Ser 920 3 Tyr-2b and Ser 920 3 Tyr-2a showed as much as 3-fold reduction in apparent Ca 2ϩ affinity compared with their corresponding wild-type isoforms. Ser 186 3 Phe, Pro 602 3 Leu, and Pro 895 3 Leu were wild-typelike in this respect. Table II also shows the results of titrating the Ca 2ϩ inhibition of 32 P i phosphorylation ( Fig. D in Supplemental Materials). Again, SERCA2b showed a significantly (2-fold) higher apparent affinity for Ca 2ϩ as compared with SERCA2a, and the apparent Ca 2ϩ affinities displayed by mutants Ser 920 3 Tyr-2a and Ser 920 3 Tyr-2b in this assay were 2-and 3-fold reduced relative to wild-type SERCA2a and SERCA2b, respectively. The other mutants were SERCA2blike in this respect. It is, furthermore, shown in Table II that the apparent affinity for 32 P i displayed by SERCA2a and SERCA2b was similar to that of SERCA1a (Fig. E in Supplemental Materials). Wild-type-like apparent 32 P i affinity was observed for mutants Cys 318 3 Arg, Pro 895 3 Leu, and Ser 920 3 Tyr, but significant 3-, 7-, 3-, and Ͼ10-fold reductions were found for mutants Gly 23 3 Glu, Ser 186 3 Phe, Pro 602 3 Leu, and Asp 702 3 Asn, respectively.
Dephosphorylation of Phosphoenzyme Intermediates formed from ATP and P i -The decay of phosphoenzyme formed with [␥-32 P]ATP was examined as illustrated in Fig. 3A and Fig. F of Supplemental Materials, and a summary of the data is given in Table II. As previously described, the phosphorylation conditions applied (0°C, pH 7.0, and presence of K ϩ ) result in accumulation of the ADP-sensitive E 1 ϳP(Ca 2 ) intermediate for SERCA1a and SERCA3 isoforms (44,47). As illustrated by the dotted line in Fig. 3A (showing a very rapid dephosphorylation upon ADP addition), this was also the case for wild-type SERCA2 isoforms and mutant SERCA2 proteins other than Cys 318 3 Arg. Addition of excess EGTA (Fig. 3A) or nonradioactive ATP (Fig. F Fig. 2B, was used as an estimate of enzyme concentration).
b Turnover rate for ATP hydrolysis at 37°C in the presence of calcium ionophore A23187 (Fig. B of Supplemental Materials). c The ratio between the maximum ATPase activities with and without calcium ionophore at pH 7.0 (Fig. B    Low expression levels for Gly 749 3 Arg (Fig. 2) and inability to phosphorylate with either [␥-32 P]ATP or 32 P i (Fig. 2)  phorylation by [␥-32 P]ATP and allowed observation of dephosphorylation in the forward direction (reactions 4 -6 in Scheme 1). The differences observed between the various isoforms and mutants were similar for EGTA chase and nonradioactive ATP chase (Table II). SERCA2b displayed half the dephosphorylation rate of SERCA2a, the latter being similar to that of SERCA1a. Mutant Pro 895 3 Leu showed a reduction of the dephosphorylation rate relative to wild-type SERCA2b. Dephosphorylation was almost completely blocked in mutants Ser 186 3 Phe, Pro 602 3 Leu, and Asp 702 3 Asn. By contrast, the dephosphorylation rate was wild-type-like for Gly 23 3 Glu. For Ser 920 3 Tyr-2b, the dephosphorylation rate was actually enhanced relative to wild-type SERCA2b. Likewise, the dephosphorylation rate of Ser 920 3 Tyr-2a was significantly higher than that of its corresponding wild-type SERCA2a (Table II). These variations of the dephosphorylation rate explain to a large extent the observed differences in the maximal cat-alytic turnover rate described above. For the remaining mutant, Cys 318 3 Arg, the dephosphorylation rate was significantly reduced both in the forward direction as observed upon chase with EGTA or nonradioactive ATP, as well as in the reverse direction upon addition of ADP (more than 80% of the initial phosphoenzyme was left after 5 s of chase with ADP, Fig.  3A, solid hexagon). The lack of ADP sensitivity indicates that during the phosphorylation period the E 2 -P intermediate actually accumulated, suggesting a slow E 2 -P 3 E 2 reaction. The decay of the E 2 -P phosphoenzyme intermediate was further examined (Fig. 3B and Table II) by diluting E 2 -P formed "backwards" from 32 P i in the dephosphorylation medium. The decay rate of SERCA2b was 3-fold reduced relative to SERCA2a. Fig.  3B, furthermore, demonstrates that the E 2 -P decay rate for Cys 318 3 Arg was 3-fold reduced relative to wild-type SERCA2b, thus confirming that a slow E 2 -P 3 E 2 reaction constitutes the major reason for the low dephosphorylation rate observed for this mutant (cf. Fig. 3A). A wild-type-like E 2 -P decay was observed for Pro 895 Fig. 3A is caused by inhibition of the E 1 ϳP(Ca 2 ) 3 E 2 -P transition. Ser 920 3 Tyr-2b displayed an increased rate of the E 2 -P decay relative to wild-type SERCA2b, but similar to that of Ser 920 3 Tyr-2a and its corresponding wild-type SERCA2a ( Fig. 3B and Table II).

Rapid Kinetic Analysis of Phosphorylation and Ca 2ϩ
Binding-For SERCA2 isoforms and DD mutants, rapid kinetic measurements were performed at 25°C as previously reported for SERCA1a and SERCA3 isoforms (8,47). The results, some of which are summarized in the last three columns of Table  II, provide information on the Ca 2ϩ binding transition (reactions 1 and 2 in Scheme 1), phosphorylation (reaction 3 in Scheme 1), Ca 2ϩ dissociation from E 1 Ca 2 (reverse reaction 2 in Scheme 1), and dephosphorylation (reactions 4 -6 in Scheme 1) at 25°C.
When the time course of phosphorylation was studied after the simultaneous addition of [␥-32 P]ATP and excess Ca 2ϩ to Ca 2ϩ -deprived enzyme (Fig. 4, examination of the sequence consisting of reactions 1-3 in Scheme 1), the observed rate constants for SERCA2a and SERCA2b were, respectively, 1.7and 2.2-fold lower than that of SERCA1a (Table II). Rate constants similar to wild-type SERCA2b were obtained for mutants Gly 23 3 Glu, Ser 186 3 Phe, Pro 602 3 Leu, and Ser 920 3 Tyr-2b. Relative to wild-type SERCA2b, the rate constant for mutant Cys 318 3 Arg was more than 2-fold enhanced, whereas that of Asp 702 3 Asn was 2-fold reduced.
When the time course of phosphorylation was monitored with the enzyme initially present in the Ca 2ϩ -saturated E 1 Ca 2 form (Fig. 5), the rise of the phosphorylation level occurred with similar rates in the wild-type SERCA2 isoforms as in SERCA1a, but for the wild-type SERCA2 isoforms there was a larger overshoot, most pronounced for SERCA2a. The overshoot is indicative of rapid dephosphorylation and/or a relatively slow step intervening between the dephosphorylation and rephosphorylation reactions. As indicated by the lines in Fig. 5, the experimental data could be reproduced rather accurately by simulation of a simplified reaction cycle (Scheme 1 in Ref. 8) using the SimZyme program as described (8). The simplified cycle consisted of three reactions: phosphorylation (reaction 3 in Scheme 1) with the rate constant k A , dephosphorylation (reactions 4 -6 in Scheme 1, combined in one step) with  Table II. The solid hexagon represents the ADP chase for Cys 318 3 Arg, whereas the dotted lines represent ADP chases for wild-type SERCA2 isoforms and mutant SERCA2 proteins other than Cys 318 3 Arg. B, phosphorylation was performed at 25°C for 10 min in the presence of 100 mM MES/Tris, pH 6.0, 10 mM MgCl 2 , 2 mM EGTA, 30% (v/v) dimethyl sulfoxide, and 0.5 mM 32 P i . 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 MOPS/Tris, pH 7.0, 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. The phosphorylation levels are shown relative to the initial level determined after 10 min phosphorylation without dephosphorylation. The lines represent the best fits of a monoexponential decay function and the extracted rate constants (s Ϫ1 Ϯ S.E.) are shown in Table II. the rate constant k B , and the Ca 2ϩ binding transition (combined reactions 1 and 2 in Scheme 1) with the rate constant k C . Consistent with previous reports (8,47), a good fit for SERCA1a was obtained using the rate constants k A ϭ 35 s Ϫ1 , k B ϭ 5 s Ϫ1 , and k C ϭ 25 s Ϫ1 , but for SERCA2a it was necessary to reduce k C to 13 s Ϫ1 and increase k B slightly to 6 s Ϫ1 . To obtain a good fit for SERCA2b, both k B and k C had to be reduced to 3 s Ϫ1 and 10 s Ϫ1 , respectively. The SERCA1a-like k B value for SERCA2a and the much lower k B value for SERCA2b, as well as the reduced k C values are in good agreement with the data presented above for the dephosphorylation of E 1 ϳP(Ca 2 ) and E 2 -P (Fig. 3) and the phosphorylation starting from the Ca 2ϩ -deprived enzyme (Fig. 4), k C supposedly being rate-limiting for the latter reaction sequence (8,47). To fit the much larger overshoot obtained with mutant Ser 920 3 Tyr-2b (Fig.  5), it was necessary to increase k B from 3 s Ϫ1 (wild-type SERCA2b value) to 6 s Ϫ1 , i.e. a 2-fold increase in the dephosphorylation rate, making it similar to that of SERCA2a. This is also in perfect agreement with the dephosphorylation data show the best fits of a monoexponential function, giving the rate constants (s Ϫ1 ) listed in Table II ("Rate of E 2 to E 1 ϳP(Ca 2 )"). For direct comparison, the dotted lines in the wild-type SERCA2 panels reproduce the line corresponding to SERCA1a, whereas the dashed lines in mutant panels reproduce the line for wild-type SERCA2b.

FIG. 5. Time course of phosphorylation by [␥-32 P]ATP of enzyme preincubated with Ca 2؉ .
Quench-flow experiments were carried out at 25°C using the QFM-5 module as illustrated in the diagram above the panels, by mixing microsomes preincubated in a medium containing 40 mM MOPS/Tris, pH 7.0, 80 mM KCl, 5 mM MgCl 2 , and 100 M CaCl 2 with an equal volume of the same medium containing in addition 10 M [␥-32 P]ATP, followed by acid quenching at the indicated time intervals. In each panel, the phosphorylation is shown relative to the maximum level reached. For direct comparison, the dotted lines in the wild-type SERCA2 panels reproduce the line corresponding to SERCA1a and the dashed lines in mutant panels reproduce the line corresponding to wild-type SERCA2b. The lines were generated by computer simulation as described (8), using the following rate constants for SERCA1a (k A ϭ 35 s Ϫ1 , k B ϭ 5 s Ϫ1 , k C ϭ 25 s Ϫ1 ), SERCA2a (k A ϭ 35 s Ϫ1 , k B ϭ 6 s Ϫ1 , k C ϭ 13 s Ϫ1 ), SERCA2b (k A ϭ 35 s Ϫ1 , k B ϭ 3 s Ϫ1 , k C ϭ 10 s Ϫ1 ), and S920Y-2b (k A ϭ 35 s Ϫ1 , k B ϭ 6 s Ϫ1 , k C ϭ 8 s Ϫ1 ). For mutants G23E, S186F, C318R, P602L, and D702N, the solid lines show the best fits of a monoexponential function giving the following rate constants: 11.4, 34, 33, 19, and 8 s Ϫ1 . described above in relation to Fig. 3. The experimental data obtained for the other DD mutants studied could be fitted by a monoexponential function, the extracted rate constant being quite similar to k A for wild-type SERCA2b in the case of Ser 186 3 Phe and Cys 318 3 Arg, but lower for Gly 23 3 Glu, Pro 602 3 Leu, and Asp 702 3 Asn (Fig. 5 and Table II). For Ser 186 3 Phe, the absence of an overshoot is explained by the low dephosphorylation rate (Fig. 3), corresponding to a markedly reduced value for k B . For Cys 318 3 Arg, the dephosphorylation was also found markedly reduced, and in addition the Ca 2ϩ binding E 2 3 E 1 Ca 2 transition seems to be enhanced, corresponding to an increased k C value, as indicated by the enhanced phosphorylation rate seen for this mutant in Fig. 4 under conditions where the E 2 3 E 1 Ca 2 transition is ratelimiting in the wild-type. The latter effect should contribute to remove the overshoot. For Gly 23 3 Glu, the lack of an overshoot is clearly because of the reduced rate of phosphorylation (k A ). Fig. 5, furthermore, indicates a reduced phosphorylation rate for Pro 602 3 Leu and, in particular, Asp 702 3 Asn (k A only 8 s Ϫ1 , see Table II). The latter mutants also showed a very low rate of the E 1 ϳP(Ca 2 ) to E 2 -P transition (Fig. 3). These effects cause the absence of an overshoot. Because of the low rate of phosphorylation of Gly 23 3 Glu and Asp 702 3 Asn, this reaction is probably also rate-limiting for the reaction sequence studied in Fig. 4, which proceeded at the same rate as the phosphorylation studied in Fig. 5.
By adding an excess of EGTA together with [␥-32 P]ATP in experiments otherwise similar to those illustrated in Fig. 5, it is possible to monitor a single turnover cycle, because the removal of Ca 2ϩ with EGTA prevents rephosphorylation after the dephosphorylation has occurred. The results of such experiments are shown in Fig. G of Supplemental Materials. This analysis confirms the conclusions derived from the data in Figs. 3-5. It is particularly noteworthy that this precise determination of k B at 25°C shows that phosphoenzyme processing (i) occurs with similar rates in SERCA1a and SERCA2a, (ii) is slower for SERCA2b, (iii) is slower for Cys 318 3 Arg, relative to wild-type SERCA2b, and faster for Ser 920 3 Tyr-2b, the latter behaving very similar to SERCA2a, and (iv) is completely blocked for Ser 186 3 Phe and Pro 602 3 Leu.
To obtain direct information on the properties of the cytoplasmically facing Ca 2ϩ sites for the wild-type SERCA2 isoforms and selected mutants, we determined the rate of Ca 2ϩ dissociation from E 1 Ca 2 at 25°C by the previously described procedure (8 -10, 47), which compares the amount of phosphoenzyme measured 34 ms after the simultaneous addition of [␥-32 P]ATP and excess EGTA to Ca 2ϩ -saturated enzyme ("EP ATPϩEGTA ") with the amount of phosphoenzyme measured after 34 ms incubation with [␥-32 P]ATP and Ca 2ϩ ("EP ATP "). The data are shown in Fig. 6, and the derived rate constants for Ca 2ϩ dissociation are listed in the last column of Table II. Importantly, Ca 2ϩ dissociation was about 10-fold slower in SERCA2b relative to SERCA1a and SERCA2a, thus explaining the high affinity of SERCA2b for Ca 2ϩ seen in the above described measurements. Both Ser 186 3 Phe and Cys 318 3 Arg displayed a Ca 2ϩ dissociation rate similar to that of wild-type SERCA2b or perhaps even lower (in this range the method does not allow a very accurate differentiation). On the other hand, the Ca 2ϩ dissociation rate was about 8-fold increased in Ser 920 3 Tyr-2b relative to wild-type SERCA2b, whereas in Gly 23 3 Glu it was as much as 13-fold increased. DISCUSSION In the present study, we investigated the overall and partial reactions of the catalytic cycle mediated by the human SERCA2a and SERCA2b isoforms, as well as 10 SERCA2b mutants, each containing a missense mutation detected in patients with DD. The importance of the results, which have been summarized in Tables I and II, is 3-fold: (i) the partial reaction steps responsible for the long known differences in the Ca 2ϩ activation at the cytosolic sites and catalytic turnover rates between SERCA1a and SERCA2 isoforms are now identified; (ii) abnormal enzymatic behavior and Ca 2ϩ transport properties of DD mutants are demonstrated, underscoring the crucial role played by the wild-type SERCA2b isoform as part of the cellular Ca 2ϩ signaling machinery involved in maintaining epidermal integrity; (iii) the analysis of the overall and partial reactions of DD mutants provides new insight regarding structure-function relationships of SERCA enzymes. First we will discuss the new information about the wild-type SERCA2 enzymes.
SERCA2a and SERCA2b Isoforms-Contrary to SERCA2a, SERCA2b is characterized by the inclusion of an additional 49-amino acid stretch (amino acids 994 to 1042) containing a hydrophobic sequence, which most likely represents an extra transmembrane domain (M11 in Fig. 1) (28, 48, 49). The extended tail in SERCA2b must be responsible for the observed differences between the partial reaction steps of SERCA2a and SERCA2b: (i) a ϳ10-fold lower rate of Ca 2ϩ dissociation from E 1 Ca 2 in SERCA2b; (ii) a ϳ2-fold lower rate of conversion of the ADP-sensitive E 1 ϳP(Ca 2 ) phosphoenzyme intermediate to ADP-insensitive E 2 -P in SERCA2b; and (iii) a 3-fold lower rate of dephosphorylation of E 2 -P to E 2 in SERCA2b. SERCA2 isoforms showed no difference from SERCA1a with respect to the rate of the phosphorylation reaction E 1 Ca 2 3 E 1 ϳP(Ca 2 ). Both SERCA2a and SERCA2b isoforms displayed a ϳ2-fold reduced rate of the E 2 to E 1 Ca 2 transition relative to SERCA1a. The latter effect must be ascribed to some (or only one) of the ϳ160 amino acid differences between the SERCA1 and SERCA2  Table II for the indicated wild-type and mutant SERCA proteins) were determined at 25°C as previously described (8)  enzymes (50). It is noteworthy in this connection that the 35-residue lumenal loop connecting M7 and M8 contains as much as 15 amino acid differences between SERCA1 and SERCA2.
The observed differences among the isoforms with respect to the partial reaction steps can account for the characteristic isoform differences described here and in previous studies (29,30) with respect to the apparent affinity for Ca 2ϩ and the turnover rate of the overall reaction. Relative to SERCA2a, an ϳ2-fold increase in apparent Ca 2ϩ affinity was observed for SERCA2b by Ca 2ϩ titration of Ca 2ϩ transport activity and ATPase activity at steady state (Table I), as well as phosphorylation from ATP (Table II). The affinity for Ca 2ϩ inhibition of phosphorylation from P i was also higher in SERCA2b relative to SERCA2a (Table II). By Ca 2ϩ titration of Ca 2ϩ transport and ATPase activity, similar observations were made for SERCA2b from rabbit and human (29), as well as pig (30). The major reason for the increased apparent affinity of SERCA2b seems to be the ϳ10-fold reduction of the rate of Ca 2ϩ dissociation from E 1 Ca 2 (Table II and Fig. 6). This represents a unique finding, indicating that the "true" affinity of the E 1 form for Ca 2ϩ is enhanced because of the presence of the C-terminal extension in SERCA2b. In addition, the slow phosphoenzyme processing might contribute to the increased apparent Ca 2ϩ affinity of SERCA2b, as more phosphoenzyme accumulates at low Ca 2ϩ concentrations when phosphoenzyme turnover is slow, thereby shifting the Ca 2ϩ activation curve to higher apparent affinity. Such a "kinetic effect" on apparent Ca 2ϩ affinity can be demonstrated by computer simulation of the reaction cycle (10). The slow phosphoenzyme processing definitely constitutes a major reason for the lower overall rates of Ca 2ϩ transport and ATP hydrolysis in SERCA2b relative to SERCA2a (Table I). Furthermore, the reduced rate of the reaction sequence E 2 3 E 1 Ca 2 in both SERCA2a and SERCA2b, relative to SERCA1a, contributes further to slowing of the SERCA2b cycle and explains why in some conditions even SERCA2a showed a reduced catalytic turnover rate relative to SERCA1a (Table I,  Site-directed mutagenesis analysis and the crystal structure of SERCA1a in the E 1 Ca 2 form have revealed that the Ca 2ϩ liganding side chains in the high affinity Ca 2ϩ binding/ occlusion sites are donated by residues Glu 309 in M4, Asn 768 and Glu 771 in M5, Asn 796 , Thr 799 , and Asn 800 in M6, and Glu 908 in M8 (1,5,7). These residues are conserved between SERCA1a and SERCA2 isoforms (Fig. 1), although the numbering in SERCA2 differs by one residue from that of SERCA1a in M5, M6, and M8, because of the deletion of a single amino acid corresponding to Gly 509 in SERCA1a. Therefore, the reduced rate of Ca 2ϩ dissociation from the E 1 Ca 2 form of SERCA2b (and implicitly its high affinity for Ca 2ϩ ) is not brought about by replacement of the Ca 2ϩ ligands themselves, but more indirectly by a fine tuning of the properties of the binding pocket through interaction involving the C-terminal extension. The two crystal structures of SERCA1a show that the transition from E 1 Ca 2 to E 2 occurring in relation to Ca 2ϩ dissociation is accompanied by dramatic rearrangements of the first six transmembrane domains, M1-M6 (5,6). It is possible that because of structural constraints imposed by the presence of the C-terminal extension, the transmembrane region is less flexible in SERCA2b relative to SERCA1a and SERCA2a. This could explain the slowing in SERCA2b of conformational changes involved in dissociation of Ca 2ϩ from E 1 Ca 2 and E 1 ϳP(Ca 2 ) 3 E 2 -P and E 2 -P 3 E 2 transitions. Because deletion of the last 12 amino acid residues of SERCA2b, thought to be lumenal, or mutagenesis of one of these, Asn 1036 in rat SERCA2b (corresponding to Asn 1035 in human SERCA2b, cf. Fig. 1), resulted in SERCA2a-like behavior with respect to the apparent Ca 2ϩ affinity in the overall reaction and Ca 2ϩ wave properties at the cellular level (30,31), it is likely that the SERCA2bspecific Ca 2ϩ binding properties owe particularly to interactions of the lumenal part of the tail, possibly with lumenal proteins such as calreticulin and calnexin (31,51), thereby reducing the conformational flexibility of the transmembrane part of SERCA.
DD Mutant Ser 920 3 Tyr-The introduction of mutation Ser 920 3 Tyr in SERCA2b accelerated Ca 2ϩ dissociation from E 1 Ca 2 as well as E 1 ϳP(Ca 2 ) 3 E 2 -P and E 2 -P 3 E 2 transitions, leading to rates similar to those of SERCA2a. As a result, the apparent Ca 2ϩ affinity and catalytic turnover rate of the overall reaction of this SERCA2b mutant were quite SERCA2a-like. Because this raised the question whether the Ser 920 3 Tyr mutation acts by interference with the effect of the C-terminal extension in SERCA2b, we tested the mutation in SERCA2a as well. When Ser 920 3 Tyr was introduced in SERCA2a, the resulting apparent Ca 2ϩ affinity at the cytoplasmically facing sites determined by Ca 2ϩ titration of Ca 2ϩ uptake and phosphorylation from ATP was considerably lower than that of the wild-type SERCA2a or Ser 920 3 Tyr-2b (Tables I and II), showing that mutation and removal of the C-terminal extension exerts additive and independent effects on Ca 2ϩ affinity. Both Ser 920 3 Tyr mutants were characterized by complete absence of the ionophore-induced activation of ATPase activity, i.e. the "back inhibition" by lumenal Ca 2ϩ was relieved by the mutation. Because the Ser 920 3 Tyr mutants showed Ca 2ϩ transport activity, the insensitivity to ionophore cannot be ascribed to lack of Ca 2ϩ accumulation in the microsomal vesicles. It is, however, possible that the sensitivity to inhibition by the accumulated lumenal Ca 2ϩ is reduced in the Ser 920 3 Tyr mutants. This could be brought about by a reduction of the true affinity of the lumenally facing Ca 2ϩ release sites on the E 2 -P form by increasing the rate of Ca 2ϩ dissociation from E 2 -PCa 2 , in line with the effect on the Ca 2ϩ sites in their cytoplasmically facing configuration. In the crystal structures of SERCA1a, the serine corresponding to Ser 920 of SERCA2 isoforms is located at the cytoplasmic surface, in the loop connecting M8 and M9 (Fig.  1). Its side chain is rather close to residues at the C-terminal end of M10, and it is likely that the replacement of the serine side chain with a bulky tyrosine causes a steric clash and thereby changes the tilt of M10. This would probably affect the lumenal loop between M9 and M10 and, in turn, M9 and possibly other lumenal loops and transmembrane segments as well, including also M11 in SERCA2b. This might increase the conformational flexibility of the transmembrane region, explaining the enhanced rates of the conformational changes associated with Ca 2ϩ dissociation, E 1 ϳP(Ca 2 ) 3 E 2 -P, and E 2 -P 3 E 2 in the mutants.
DD Mutant Gly 23 3 Glu-Gly 23 is conserved in all known SERCAs and secretory pathway/Golgi Ca 2ϩ -ATPases, as well as Na ϩ ,K ϩ -and H ϩ ,K ϩ -ATPases, but has not been previously studied by mutagenesis. Remarkably, the Gly 23 3 Glu mutant displayed Ca 2ϩ -activated ATPase activity with a maximum catalytic turnover rate similar to that of wild-type SERCA2b, but without any measurable Ca 2ϩ uptake in the microsomes. There was little activation of ATP hydrolysis when the vesicles were made permeable to Ca 2ϩ with ionophore, thus confirming that Ca 2ϩ uptake is insignificant for this mutant (Table I). The partial reactions of the Gly 23 3 Glu mutant differed from those of wild-type SERCA2b in several respects (Table II): the rate of the phosphorylation reaction E 1 Ca 2 3 E 1 ϳP(Ca 2 ) was reduced; the rate of dephosphorylation (E 2 -P 3 E 2 ) was 3-fold enhanced (resulting in a 3-fold reduced apparent affinity for P i ); and the rate of Ca 2ϩ dissociation from E 1 Ca 2 was 13-fold enhanced, relative to wild-type SERCA2b, resulting in a reduced apparent affinity for Ca 2ϩ . The defects in Ca 2ϩ binding and in the Ca 2ϩactivated phosphorylation from ATP may reflect changes to the Ca 2ϩ sites related to the inability to couple ATP hydrolysis with Ca 2ϩ transport. The highly conserved Gly 23 is located in a loop that connects the two cytoplasmic ␣-helices before M1 near the N terminus (Fig. 1). These helices are part of domain A, which undergoes a large rearrangement in relation to the transition from E 2 to E 1 Ca 2 (6). It is, therefore, possible that the flexibility of the loop introduced by the presence of the glycine is required for a proper domain movement and long range interaction with the Ca 2ϩ sites. Similar uncoupling of ATPase activity from Ca 2ϩ transport induced by mutagenesis has previously only been observed for a SERCA1a mutant where a tyrosine located much closer to the Ca 2ϩ binding structure (corresponding to Tyr 762 in Fig. 1) had been exchanged for glycine (46). DD Mutant Cys 318 3 Arg-Mutant Cys 318 3 Arg was also found more deficient with respect to Ca 2ϩ accumulation than ATPase activity. Even though Ca 2ϩ accumulation was below the detection limit, a significant, although reduced, ionophore stimulation of ATPase activity was noted. The rate of dephosphorylation of the ADP-insensitive phosphoenzyme intermediate was found 3-fold reduced relative to wild-type SERCA2b, with resulting accumulation of this intermediate at steady state. Because lumenal Ca 2ϩ binds to sites on E 2 -P, accumulation of E 2 -P is expected to increase the sensitivity to inhibition by lumenal Ca 2ϩ . The functional effects of mutation Cys 318 3 Arg are probably caused by the presence of the bulky and highly charged arginine side chain, because the substitution of Cys 318 in SERCA1a with alanine causes little effect on the rate of Ca 2ϩ transport (52). Cys 318 is located in transmembrane segment M4 near the cytoplasmic boundary, and analysis of the crystal structures of SERCA1a shows that the replacement with arginine is likely to lead to steric clash with residues located in M5 and M6 near the Ca 2ϩ binding sites. This could interfere with conformational changes involving these transmembrane segments ("a stick in the wheel" effect), thereby preventing the opening of the Ca 2ϩ binding pocket toward the lumenal side. In fact, it is possible that the E 2 -P form accumulated during the enzymatic cycle of mutant Cys 318 3 Arg contains bound Ca 2ϩ (i.e. the E 2 -PCa 2 intermediate indicated in Scheme 1). Hence, a common explanation of the increased sensitivity to inhibition by lumenal Ca 2ϩ and the reduced dephosphorylation rate could be a reduced rate of lumenal dissociation of Ca 2ϩ from E 2 -PCa 2 .
DD Mutant Ser 186 3 Phe-The major defect in mutant Ser 186 3 Phe seems to be a reduced rate of the E 1 ϳP(Ca 2 ) 3 E 2 -P transition, accounting for the lack of significant Ca 2ϩ transport and ATPase activity. In addition, a 30-fold increase of the rate of E 2 -P dephosphorylation was observed, suggesting that the underlying cause of both effects could be a destabilization of the E 2 -P state. Ser 186 is located close to the highly conserved TGES segment of domain A, which according to the SERCA1a crystal structures moves from an isolated position in E 1 Ca 2 into close contact with domain P in the E 2 conformation (6), a rearrangement that also seems to occur during the E 1 ϳP(Ca 2 ) 3 E 2 -P transition (53)(54)(55). It is conceivable that replacement of the serine with the bulky phenylalanine could interfere profoundly with the interactions at the interface between domains A and P in the E 2 P conformation, thereby creating instability of this form.
DD Mutants Pro 602 3 Leu, Asp 702 3 Asn, and Pro 895 3 Leu-Mutants Pro 602 3 Leu and Asp 702 3 Asn also showed a conspicuous reduction of the rate of the E 1 ϳP(Ca 2 ) 3 E 2 -P transition, explaining the lack of significant Ca 2ϩ transport and ATPase activity. For mutant Pro 895 3 Leu, the rate of the E 1 ϳP(Ca 2 ) 3 E 2 -P transition was ϳ2-fold reduced, consistent with more moderately reduced molecular Ca 2ϩ transport and ATPase activities. Pro 602 is located at the "hinge" between domains N and P, whereas Asp 702 is located in domain P, near the phosphorylated Asp 351 , and Pro 895 is located far away at the lumenal end of transmembrane helix M8 (Fig. 1). The effects on the E 1 ϳP(Ca 2 ) 3 E 2 -P transition of mutations of these residues, as well as mutation Ser 186 3 Phe in domain A discussed above, support a global nature of this conformational change, involving almost every region of the protein, similar to the transition between the E 1 Ca 2 and E 1 forms for which the SERCA1a atomic structures are known (6). The importance of Pro 602 and Asp 702 for the E 1 ϳP(Ca 2 ) 3 E 2 -P transition in SERCA2b demonstrated here is consistent with previous findings for the corresponding mutations, Pro 603 3 Leu and Asp 703 3 Asn, in SERCA1a (56), whereas Pro 895 has not been previously studied by mutagenesis. The reduced rate of phosphorylation of E 1 Ca 2 to E 1 ϳP(Ca 2 ) observed for Pro 602 3 Leu and Asp 702 3 Asn (Fig. 5) has not yet been reported for the corresponding SERCA1a mutants, but the formation of phosphoenzyme was completely abolished in SERCA1a mutants Pro 603 3 Gly and Asp 703 3 Ala (56, 57), thus indicating the crucial roles of these residues for the phosphorylation reaction. Pro 602 (SERCA2 numbering) is probably critical for a hinge bending movement involved in the approach of domains N and P in connection with the transfer of the ␥-phosphoryl group from ATP to Asp 351 . Asp 702 is conserved among all members of the haloacid dehalogenase superfamily to which P-type AT-Pases belong (58). As the corresponding aspartate in phosphoserine phosphatase, Asp 702 , through its two side chain oxygen atoms, could contribute to ligation of Mg 2ϩ required as cofactor for catalysis of the phosphorylation reaction (59,60).
DD Mutants Thr 357 3 Lys, Gly 749 3 Arg, and Gly 769 3 Arg-The lack of phosphorylation of Thr 357 3 Lys and Gly 769 3 Arg from either ATP or P i is in agreement with previous findings for SERCA1a mutants Thr 357 3 Ala and Gly 770 3 Val (61, 62). Thr 357 is located in domain P close to the phosphorylated Asp 351 and has been shown to be crucial to ATP binding in SERCA1a (63). Gly 769 is located in transmembrane helix M5 adjacent to a glutamate participating in Ca 2ϩ ligation (Glu 770 in Fig. 1). It is, therefore, not surprising that substitution of the glycine with arginine interferes with the Ca 2ϩ activated phosphorylation from ATP. The abolishment of phosphorylation from P i is, however, not a consequence of defective Ca 2ϩ binding, as the latter reaction requires a Ca 2ϩ -free enzyme (cf. Scheme 1). In fact, mutation of the Ca 2ϩ ligating glutamate in M5 leads to increased stability of the E 2 -P phosphoenzyme intermediate (64,65). It may be envisioned that the flexibility introduced by the presence of the glycine is important for conformational changes involved in the long range interaction between the transmembrane region and the phosphorylation site occurring in connection with formation of E 2 -P. Finally, mutant Gly 749 3 Arg was expressed only at a very low level. The corresponding glycine in SERCA1a has been replaced by alanine without loss of expression, but substitution with leucine reduced enzyme expression (9). For SERCA1a, it has, furthermore, been shown that the adjacent arginine (corresponding to Arg 750 in SERCA2) is crucial both to the structural and functional integrity of the enzyme (9), probably because it participates in a network of hydrogen bonds and van der Waals interactions with residues in the loop between transmembrane segments M6 and M7 (5). This function would probably be disturbed by the insertion of another arginine instead of the adjacent glycine, and the low expression level of the Gly 749 3 Arg mutant may be explained by structural instability leading to enhanced proteasome degradation.
Why Do These Mutations Cause Disease?-Eight of the 10 DD mutants studied here showed no significant Ca 2ϩ transport activity. The change in Ca 2ϩ homeostasis leading to DD can, thus, be explained by the hypothesis of haploinsufficiency, i.e. because one of the two alleles encodes a Ca 2ϩ pump with impaired Ca 2ϩ transport activity or no activity at all, the total cellular capacity for Ca 2ϩ accumulation in internal stores is markedly reduced. Ser 920 3 Tyr and Pro 895 3 Leu are the first DD mutants reported to show significant Ca 2ϩ transport activity. For Pro 895 3 Leu the molecular Ca 2ϩ transport activity was reduced, in line with the hypothesis of haploinsufficiency. For Ser 920 3 Tyr, the molecular Ca 2ϩ transport activity was in fact increased relative to wild-type SERCA2b ( Table I). The dominant DD phenotype produced by mutant Ser 920 3 Tyr could, nevertheless, be caused by haploinsufficiency, because the expression level of mutant Ser 920 3 Tyr in HEK-293 was significantly lower than its corresponding wild-type protein (Fig. 2, A and B) and its contribution to the total cellular capacity for Ca 2ϩ accumulation was markedly reduced (cf. Fig.  2C), despite its high intrinsic molecular Ca 2ϩ transport activity. The marked reduction in the protein level for Ser 920 3 Tyr in HEK-293 cells, which could be because of enhanced cellular proteasome-mediated degradation, may also reflect the situation in epidermal cells. It has earlier been proposed that Ca 2ϩ transport activity of the wild-type enzyme encoded by one of the alleles is inhibited through protein-protein interactions with mutant enzyme (38). In addition, it is possible that the wild-type enzyme present in dimeric complexes with mutant Ser 920 3 Tyr becomes more susceptible to cellular degradation than in its monomeric state. Furthermore, the observed reduction of Ca 2ϩ affinity for Ser 920 3 Tyr to a value similar to that of SERCA2a may contribute to an abnormal Ca 2ϩ homeostasis, thus implying that wild-type SERCA2b with its higher Ca 2ϩ affinity is crucial for normal functioning of epidermal cells. Finally, the observed reduced sensitivity of Ser 920 3 Tyr to inhibition by Ca 2ϩ accumulated in the lumen appears important and could contribute to abnormal Ca 2ϩ homeostasis and regulation in Darier disease.