Comprehensive Analysis of Expression and Function of 51 Sarco(endo)plasmic Reticulum Ca2+-ATPase Mutants Associated with Darier Disease*

We examined possible defects of sarco(endo)plasmic reticulum Ca2+-ATPase 2b (SERCA2b) associated with its 51 mutations found in Darier disease (DD) pedigrees, i.e. most of the substitution and deletion mutations of residues reported so far. COS-1 cells were transfected with each of the mutant cDNAs, and the expression and function of the SERCA2b protein was analyzed with microsomes prepared from the cells and compared with those of the wild type. Fifteen mutants showed markedly reduced expression. Among the other 36, 29 mutants exhibited completely abolished or strongly inhibited Ca2+-ATPase activity, whereas the other seven possessed fairly high or normal ATPase activity. In four of the aforementioned seven mutants, Ca2+ transport activity was significantly reduced or almost completely lost, therefore uncoupled from ATP hydrolysis. The other three were exceptional cases as they were seemingly normal in protein expression and Ca2+ transport function, but were found to have abnormalities in the kinetic properties altered by the three mutations, which happened to be in the three DD pedigrees found by us previously (Sato, K., Yamasaki, K., Daiho, T., Miyauchi, Y., Takahashi, H., Ishida-Yamamoto, A., Nakamura, S., Iizuka, H., and Suzuki, H. (2004) J. Biol. Chem. 279, 35595-35603). Collectively, our results indicated that in most cases (48 of 51) DD mutations cause severe disruption of Ca2+ homeostasis by the defects in protein expression and/or transport function and hence DD, but even a slight disturbance of the homeostasis will result in the disease. Our results also provided further insight into the structure-function relationship of SERCAs and revealed critical regions and residues of the enzyme.

Sarco(endo)plasmic reticulum Ca 2ϩ -ATPases (SERCAs) 2 catalyze Ca 2ϩ transport coupled with ATP hydrolysis (Fig. 1) and play an essential role in maintaining Ca 2ϩ homeostasis in the cytoplasm and endoplasmic reticulum lumen of cells (1)(2)(3)(4)(5)(6)(7). SERCAs have three cytoplasmic domains: phosphorylation (P), nucleotide binding (N), and actuator (A) and 10 transmembrane helices (M1-M10 or 11 in the SERCA2b isoform, M11). In the Ca 2ϩ transport cycle, the ATPase is activated by the binding of two Ca 2ϩ ions from the cytoplasm to the transport sites composed of M4, M5, M6, and M8 (E2 3 E1Ca 2 , step 1). Asp 351 in the P domain is then phosphorylated with MgATP to form the phosphorylated intermediate (EP) (step 2). During dephosphorylation of EP, the Ca 2ϩ ions are released into the lumen. In the detailed mechanism, the dephosphorylation process includes the conformational transition of EP associated with Ca 2ϩ release (step 3) and the subsequent hydrolysis of the acylphosphate bond (step 4).
The three human SERCA genes encode SERCA isoforms (8 -10). Mutations in the SERCA2 gene (ATP2A2) and the resulting defects in the SERCA2b housekeeping isoform cause an autosomal dominant genetic skin disease, Darier disease (DD) (11,12). Over 100 mutations have been found with the DD pedigrees (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24). They include many nonsense mutations, and also substitution and deletion mutations of amino acid residues. The mutations are located throughout the SERCA2b molecule and show no "hot spots" on the primary sequence. To understand how each of the substitution and deletion mutations affects SERCA2b protein, a limited number of mutations had been explored (9 by Ahn et al. (25), 10 by Dode et al. (23), 3 by Sato et al. (26) (a total of 20 because of overlap in Refs. 23 and 25)). To provide a comprehensive insight into the molecular basis of DD, as well as to understand the basis for each case of the DD pedigrees, it is necessary to analyze further the many unexplored substitution and deletion mutations. We therefore carried out in this study a comprehensive analysis of the expression and function of most of the DD causing substitution and deletion mutations reported, i.e. the 51 mutations shown in Fig.  2. Our results showed that most of the mutations (48 of the 51) cause severe defects in protein expression and/or Ca 2ϩ transport function. The loss of the transport function was ascribed to markedly reduced ATP hydrolysis or uncoupling from ATP hydrolysis. The remaining three mutations were exceptional in that they exhibited seemingly normal protein expression and Ca 2ϩ transport function but with altered kinetic properties. Results therefore indicated diverse molecular defects as the cause of DD in the 51 pedigrees. On the basis of the atomic structures of SERCA1a, our results also provided further insight into the structure-function relationship of SERCAs.

EXPERIMENTAL PROCEDURES
Mutagenesis and Expression-The Stratagene QuikChange TM site-directed mutagenesis method (Stratagene, La Jolla, CA) was utilized for the substitution and deletions of residues in human SERCA2b cDNA in plasmid pGEM7-Zf(ϩ) (Promega). Appropriate restriction fragments with the desired mutation were excised and ligated back into the corresponding region of the full-length SERCA2b cDNA in the plasmid. The full-length SERCA2b cDNA was then excised and ligated into the pMT2 expression vector (27). The pMT2 DNA was transfected into COS-1 cells using the liposome-mediated transfection method. Microsomes were prepared from the cells as described previously (28). The "control microsomes" were prepared from COS-1 cells transfected with the pMT2 vector containing no SERCA2b cDNA. The amount of expressed SERCA2b protein was quantified by an enzyme-linked immunosorbent assay with SERCA2b-specific monoclonal antibody IID8 (Affinity Bioreagents, Golden, CO), and the expression level of the mutant in the microsomes was obtained as a value relative to that of the wild type. The amount of intrinsic SERCA2b in the control microsomes was less than 1% of the amount of SERCA2b in the microsomes from the cells transfected with the wild-type cDNA.
RNA Preparation and Northern Blot Analysis-Total RNA was extracted by the RNeasy Mini Kit (Qiagen) from COS-1 cells transfected with the pMT2 vector containing wild-type or mutant SERCA2b cDNA, and 0.5 g of total RNA was electrophoresed and blotted onto Hybond Nϩ nylon membrane (Amersham Biosciences). Hybridization was performed with a digoxigenin-labeled RNA probe for the region (nucleotides 2249 -2620) of SERCA2b cDNA. Digoxigenin labeling of the probe was performed with the DIG Northern Starter Kit (Roche Applied Science). Digoxigenin-labeled actin RNA probe (Roche Applied Science) was used as the control probe. After washing the blotted membrane, digoxigenin-labeled RNA was detected with anti-digoxigenin Fab fragments and visualized by the chemiluminescence technique with CDP-Star (Roche Applied Science).
Quantitative Real Time Reverse Transcriptase-PCR-The SERCA2b mRNA level in 0.02 g of the total RNA was determined by the real time reverse transcriptase-PCR with the LightCycler TM system and the LightCycler-RNA Master SYBR Green I (Roche Molecular Biochemicals). Melting curve analysis was performed to enhance specificity of the amplification reaction, and LightCycler software version 3.5 was used to evaluate the amplification efficiency and thus quantify the relative mRNA level in comparison with the internal standard curve obtained with the cells transfected with wild-type SERCA2b. The mRNA levels of the mutants relative to the wildtype level thus obtained were corrected by the mRNA level of glucose-6-phosphate dehydrogenase. The primers used for SERCA2b were GGCAATCTACAACAACATGAAAC (forward) and GTAGGAAATGACTCAGCTTGG (reverse), and for glyceraldehyde-3-phosphate dehydrogenase, CATGTTC-GTCATGGGTGTGA (forward) and AGTGAGCTTCCCGT-TCAGCTC (reverse).
ATPase Activity-The rate of ATP hydrolysis was determined at 37°C in a mixture containing 20 g/ml microsomal protein, 1 mM ATP, 1 M A23187, 7 mM MgCl 2 , 0.1 M KCl, 5 mM NaN 3 , 50 mM MOPS/Tris (pH 7.0), and 1.84 mM CaCl 2 with 2 mM EGTA (3.2 M free Ca 2ϩ (pCa 5.5)). When the Ca 2ϩ concentration dependence was determined, the CaCl 2 concentration was varied in the presence of 2 mM EGTA. The reaction was terminated by the addition of ice-cold trichloroacetic acid, and the amount of P i released was quantified by the Youngburg and Youngburg method (29). Total Ca 2ϩ -ATPase activity of the microsomes was obtained as above, but by subtracting the Ca 2ϩ -independent ATPase activity, determined in the presence of 5 mM EGTA without added CaCl 2 . The Ca 2ϩ -ATPase activity of the expressed SERCA2b was then obtained by subtracting the Ca 2ϩ -ATPase activity of the control microsomes (background level) from that of the microsomes expressing SERCA2b. This background level was as low as 3% of the activity of those microsomes expressing the wild-type SERCA2b from its cDNA. The Ca 2ϩ -ATPase activity of the expressed wild type in the microsomes thus obtained was 122.6 Ϯ 3.7 nmol/ min/mg of microsomal protein (n ϭ 5). The Ca 2ϩ -ATPase activity of each of the mutants was normalized to its protein expression level relative to that of the wild type (determined as above), thus the specific Ca 2ϩ -ATPase activity of each of the mutants relative to that of the wild type was obtained.
Ca 2ϩ Transport Activity-Oxalate-dependent and thapsigargin-sensitive Ca 2ϩ transport was assayed as described previously (30) at 25°C in the presence and absence of 0.5 M thapsigargin in a mixture containing 20 g/ml microsomal protein, 1 mM ATP, 7 mM MgCl 2 , 0.1 M KCl, 20 mM MOPS/ Tris (pH 7.0), 5 mM potassium oxalate, and 0.462 mM 45 CaCl 2 with 0.5 mM EGTA (3.2 M free 45 Ca 2ϩ ( pCa 5.5)). The Ca 2ϩ transport activity of the SERCA2b expressed from its cDNA in the microsomes was obtained by subtracting the thapsigarginsensitive activity of the control microsomes (background level) from that of the microsomes expressing SERCA2b from its cDNA. This background level was as low as 1% of the activity of microsomes expressing the wild-type SERCA2b from its cDNA. The Ca 2ϩ transport activity of the expressed wild type in the microsomes thus obtained was 68.1 Ϯ 4 nmol/min/mg of microsomal protein (n ϭ 4). The Ca 2ϩ transport activity of each of the mutants was normalized to its protein-expression level relative to that of the wild type (determined as above), thus the specific Ca 2ϩ transport activity of each of the mutants relative to that of the wild type was obtained.
Formation of EP-Phosphorylation of SERCA2b in microsomes with [␥-32 P]ATP or 32 P i was performed under conditions described in the figure legends and Table 1. The reaction was quenched with ice-cold trichloroacetic acid containing P i . The precipitated proteins were separated by 5% SDS-polyacryl-amide gel electrophoresis at pH 6.0 according to the Weber and Osborn method (31). The radioactivity associated with the separated SERCA2b was quantitated by digital autoradiography as described previously (32). The amount of EP formed with the expressed SERCA2b was obtained by subtracting the background radioactivity with the control microsomes. This background level was less than 5% of the radioactivity of EP formed with the expressed wild-type SERCA2b.
Miscellaneous-Protein concentrations were determined using the method of Lowry et al. (33) with bovine serum albumin as the standard. Free Ca 2ϩ concentrations were calculated as described previously (34). Data were analyzed using Origin software (Microcal Software, Inc., Northampton, MA). Three-FIGURE 2. Locations of DD mutations examined in this study on the secondary structure model of SERCA2b (a) and on the atomic structure of SERCA1a (b). a, the residues with the 51 DD mutations examined in this study are indicated by red circles (substitutions) and dotted red circles (deletions) on the secondary structure model of SERCA2b depicted on the basis of the SERCA1a model, the sequence of human SERCA2b, and its hydropathy profile (5,8). Asp 351 , autophosphorylation site. Blue circles, the side chain oxygen atoms of Glu 309 (M4), Asn 767 , Glu 770 (M5), Asn 795 , Thr 798 , Asp 799 (M6), and Glu 907 (M8) (in the numbering of SERCA2b) contribute as the Ca 2ϩ ligands in the high affinity Ca 2ϩ -binding sites. The residue Gly 509 in SERCA1a is absent in SERCA2b, therefore the numbering of residues after this position in SERCA2b is lower by one than in SERCA1a. SERCA2b possesses the 11th transmembrane helix (M11) in its long extra C terminus region, i.e. 1040 residues in SERCA2b versus 994 residues in SERCA1a (8, 11, 64 -66). The amino acid sequence of SERCA2b is otherwise highly homologous to that of SERCA1a; actually the residues with the 51 DD mutations examined here and the residues described under "Discussion" and in the supplemental discussion in relation to these DD mutations for the structure/function relationship are all conserved in SERCA1a and 2b. b, residues with the DD mutations examined are shown with the main chain ␣-carbons (blue, green, and red) on the atomic structure E1Ca 2 of SERCA1a (PDB accession code 1SU4 (39,44,53,63). The colors of the ␣-carbons indicate the DD mutants with largely reduced protein expression (blue; less than 30% of the wild-type level, see Fig. 3), those with largely reduced Ca 2ϩ -ATPase activity (green; less than 30% of the wild-type activity, see Fig. 4), and those with high or normal Ca 2ϩ -ATPase activity (red; higher than 50% of the wild-type activity, see Fig. 4). Asp 351 , Ca 2ϩ ions (yellow spheres) at the high affinity sites, three cytoplasmic domains P (pink), N (blue), and A (yellow), 10 transmembrane helices (M1-M10), L6 -7 (loop connecting M6 and M7), and L7-8 are indicated. The color for M1-M10 changes gradually from red to blue. dimensional models of the enzyme were reproduced by the program VMD (35).

Effects of DD Mutations on Protein Expression and ATPase
Activity-Each of the 51 DD-causing substitution and deletion mutations was introduced into SERCA2b cDNA, and the mutant cDNA or wild-type cDNA was transfected into COS-1 cells. The expression level of the mutant protein in microsomes prepared from the cells was determined and compared with that of the wild-type protein (Fig. 3a). The amount of intrinsic wild-type SERCA2b protein in the control microsomes (prepared from the control cells transfected with the vector without having the SERCA2b cDNA) was less than 1% of the wild-type protein expressed with the cDNA. Depending on the mutations introduced, the expression of mutants varied significantly from undetectable levels to those comparable with the wild-type level. Expression levels of the 15 mutants were less than 30% of the wild-type level, thus very low. There were no hot spots on the primary and tertiary structures for the markedly reduced expression (see Figs. 2b and 3b). Such mark-edly reduced expression occurred with the specific mutations in the cytoplasmic, transmembrane, and lumenal regions of the enzyme.
The transcription levels were checked by the Northern blot analysis and more quantitatively by the real time reverse transcriptase-PCR method for these 15 mutants and three more mutants of the relatively low protein expression (30 -40% of the wild-type level). We found, first of all, that the mRNA levels of all these mutants as well as of the wild type were ϳ500 to 800 times higher than the intrinsic wild-type SERCA2b mRNA level of the control cells, showing their extremely high transcription levels and suggesting high transfection efficiency. In the Northern blot analysis, the size of the mRNA of all these mutants and that of the wild type in the expression vector were shown to be exactly the same and their expression levels were very similar with no major reduction in the mutants (data not shown). The more quantitative comparison by the real time reverse transcriptase-PCR method in Fig. 3c actually showed no major reduction, although in some mutants, the mRNA levels were somewhat lower or moderately reduced as compared with the wild type. In any case, the mRNA reduction was not comparable or accountable to the marked reduction of the protein expression. The markedly reduced protein expression, likely due to the quality control of the misfolded SERCA (30) with some contribution of the possible moderate reduction in mRNA level, could be responsible for the development of DD in the pedigrees with these mutations.
The 36 mutants showed significant protein expression, over 30 -100% of the wild-type level, and therefore we determined the Ca 2ϩ -ATPase activity of these mutants at the optimum pCa 5.5. The specific activity of each of the mutants relative to that of the wild type was calculated and plotted versus the relative protein expression level (Fig. 4). Seven mutants (L321F, I274V, M719I, N767S, G807R, A803T, and V843F) among the 36 showed fairly high activity being 50 -120% of the wild type activity. In these seven mutants, the expression levels were also fairly high. By contrast, the other 29 mutants showed markedly reduced or no activity regardless of whether the expression level was high or low. It should be noted that there was no mutant displaying reduced expression with high ATPase activity (such mutant types, if present, would be plotted on the area (far) above the dotted line at the low expression range in Fig. 4). The findings may be clinically reasonable, because the high functional activity of the protein would compensate for the reduced protein expression.
Ca 2ϩ Transport Activity and Its Coupling with ATP Hydrolysis-Ca 2ϩ transport activity was determined with the seven mutants that were shown above to possess fairly high Ca 2ϩ -ATPase activity (L321F, I274V, M719I, N767S, G807R, A803T, and V843F) and also with eight other mutants that exhibited significantly reduced but still some activity (M699I, S916Y, S920Y, Q108H, A838P, S186P, P680L, and G310V, see Fig. 4). In Fig. 5, the specific Ca 2ϩ transport activity of each mutant relative to that of the wild type was plotted versus the relative specific Ca 2ϩ -ATPase activity. In the mutants L321F, M719I, and I274V, the transport activity was as high as or only somewhat lower than the wild-type activity. In each of these three mutants, the relative Ca 2ϩ transport activity was comparable with the relative Ca 2ϩ -ATPase activity, therefore Ca 2ϩ transport is coupled with ATP hydrolysis as in the wild type. By contrast, in the other 12 mutants, Ca 2ϩ transport activity was significantly reduced or almost completely lost. The most notable mutants were A803T and V843F, because they possess fairly high ATPase activity but no transport activity. Thus Ca 2ϩ transport is strongly uncoupled from ATP hydrolysis. Mutants N767S and G807R have Ca 2ϩ -transport activity only 28% of the wild-type activity with high Ca 2ϩ -ATPase activity (63-67% of the wild-type activity), thus in these mutants, the tight coupling for Ca 2ϩ transport is disrupted, exhibiting partial uncoupling. Three other mutants, S916Y, S920Y, and A838P, exhibited no detectable Ca 2ϩ transport and thus appear to be strongly uncoupled from their markedly reduced but still remaining Ca 2ϩ -ATPase activity (24,19, and 19% of the wild-type activity, see Table 1).
EP Formation from ATP and from P i -To examine the phosphorylation ability at the catalytic site of the DD mutants and to understand the possible causes for the inhibition of ATP hydrolysis observed in Fig. 4, we determined EP formation from ATP and P i with the 36 DD mutants that were expressed high enough for the analysis (more than 30% compared with the wild type, see Fig. 3). The affinity for Ca 2ϩ in the ATP-induced EP formation was also determined in some cases and summarized in Table 1. In Fig. 6, the amount of EP formed from ATP at 0°C and pCa 5.5 at steady state was plotted versus Ca 2ϩ -ATPase activity. The mutants L321F, I274V, M719I, G807R, and A803T, which possess high ATPase activity, formed a fairly high amount of EP. Many of the other mutants, which exhibited the strongly or completely reduced ATPase activity, formed high levels of EP from ATP at a steady state, therefore the dephosphorylation process was probably inhibited in these mutants. These mutants include P680L, M699I, and C344Y in the P domain; P160H, P160L, G211D, N39D, N39T, S186P, and S186F in the A domain; and ⌬L41 and ⌬P42 on the A domain/M1 linker. In the transmembrane domain, they are L65S on M1, C318R on M4, A838P on M7, and Q108H on M2. On the other hand, EP formation from ATP and P i were almost completely inhibited with the mutants K683E at the catalytic site, D149N and G23E on the A domain, S765L and G769R on M5, and N809I at top of M6 (see Table 1 for the amounts of EP from P i ). The mutant F487S showed strongly inhibited EP formation from ATP (thus no ATPase activity) but only a slight reduction in E2P formation from P i , being consistent with the predicted function of Phe 487 in ring stacking for the adenine moiety (36 -39).
During the examination of EP formation from ATP, we further found that mutants G310V, N767S, and V843F at or near FIGURE 4. Ca 2؉ -ATPase activity of DD mutants and its relation with the protein expression level. The Ca 2ϩ -ATPase activity of the expressed SERCA2b in the microsomes was determined at pCa 5.5 and the specific activity of each of the mutants relative to that of the wild type was calculated, as described under "Experimental Procedures." The relative specific activity of the mutant (the mean Ϯ S.D. (n ϭ 3)) thus obtained was plotted versus its relative protein expression level determined in Fig. 3. The dotted line was drawn from 0 through the wild-type level (100%). The values are also summarized in Table 1. FIGURE 5. Ca 2؉ transport activity of DD mutants and its relation with Ca 2؉ -ATPase activity. The Ca 2ϩ transport activity of the expressed SERCA2b in the microsomes was determined at pCa 5.5 and the specific activity of each of the mutants relative to that of the wild type was calculated, as described under "Experimental Procedures." The relative specific Ca 2ϩ transport activity of the mutant (the mean Ϯ S.D. (n ϭ 3)) thus obtained was plotted versus its relative specific Ca 2ϩ -ATPase activity obtained in Fig. 4. The dotted line was drawn from 0 through the wild-type level (100%). The values are also summarized in Table 1. AUGUST 11, 2006 • VOLUME 281 • NUMBER 32 the Ca 2ϩ -binding sites in the transmembrane domain show nearly no EP at 0°C and pCa 5.5 but can form high levels of EP at a high temperature of 25°C (see Fig. 7 for the Ca 2ϩ dependence of EP formation from ATP). Results suggested the possible importance of the thermal motions of these regions for the high affinity Ca 2ϩ binding and Ca 2ϩ activation of the phosphorylation site. It should be noted that N767S and V843F possessed high ATPase activity at the high temperature (but uncoupling in the Ca 2ϩ transport), whereas G310V showed the completely inhibited Ca 2ϩ -ATPase activity probably because of a blockage in the dephosphorylation process.

Inhibition of Ca 2ϩ -activated EP Formation at High Ca 2ϩ
Concentrations in K683R, D702N, and G703S-During the examination of EP formation from ATP and its Ca 2ϩ dependence, we also found the unique behavior of mutants K683R, G703S, and D702N at the catalytic site on the P domain. As shown in Fig. 8, at 25°C, the ATP-induced EP formation was activated by the high affinity Ca 2ϩ binding but markedly reduced when the Ca 2ϩ concentration was increased to over 10 M. This biphasic behavior was more profound at 0°C especially in K683R and G703S, and actually Ca 2ϩ even at 1 M was inhibitory. The behavior was in sharp contrast to that of the

Summary for functional properties of DD-causing mutants
The 36 DD mutants, whose protein expression levels were higher than 30% of that of the wild type, were subjected to functional analyses, and the data obtained are summarized. The specific Ca 2ϩ -ATPase activity of each of the mutants relative to that of the wild type (WT) was determined as described in the legend to Fig. 4, and shown in the fourth column. The mutants are placed from the top row in the table according to their activities thus obtained (i.e. from highest to lowest). The specific Ca 2ϩ transport activity of each of the mutants relative to that of the wild type was determined in Fig. 5. The amount of EP formed from ͓␥-32 P͔ATP of the mutant was determined at steady state and at 0°C, pCa 5.5, and normalized to its expression level relative to that of the wild type as described in the legend to Fig. 6. K 0.5 and the Hill coefficient (n) were determined in the Ca 2ϩ dependence of this EP formation at 0°C by fitting the data to the Hill equation. EP formation from ͓␥-32 P͔ATP was also performed at 25°C, 10 mM Ca 2ϩ for 10 s in a mixture containing 40 g/ml of microsomal protein, 0.1 mM ͓␥-32 P͔ATP, 5 mM MgCl 2 , 80 mM KCl, 30 mM Tris-HCl (pH 7.5), and 10 mM CaCl 2 , and the amount of EP in the mutant was calculated as above. The amount of EP formed with the wild type was 126 Ϯ 6 (n ϭ 6) pmol/mg of microsomal protein, and nearly the same as that at 0°C and pCa 5.5. EP formation from 32 P i was performed in the absence of Ca 2ϩ at 25°C for 10 min in a mixture containing 40 g/ml of microsomal protein, 0.1 mM 32 P i , 1 M A23187, 10 mM MgCl 2 , 35% (v/v) Me 2 SO, 50 mM MES/Tris (pH 6.0), and 5 mM EGTA, i.e. under the conditions previously demonstrated by Barrabin et al. (70) with SR Ca 2ϩ -ATPase to phosphorylate virtually all the phosphorylation sites. The amount of EP formed with the mutant was calculated as above. The amount of EP formed with the wild type was 134 Ϯ 9 (n ϭ 4) pmol/mg of microsomal protein. All the values are the mean values obtained in three to eight independent experiments (shown without the standard deviations for simplicity, but see figures for the deviations). In the second and third columns, the locations of the DD mutation in the gene and in the enzyme protein are indicated with the exon number and the region in the tertiary structure, respectively. The line below the mutant V843F was drawn to distinguish the first seven mutants (L321F-V843F) that possess the high specific ATPase activity (higher than 50% of that of wild type). It should be noted that mutants K683R, G703S, and D702N showed completely abolished ATPase activity (see Fig. 8). The loss of activity is probably due to blocking of the dephosphorylation process because EP was formed from ATP at high levels at steady state under the almost same conditions as those for ATPase assay, i.e. the high temperature, pCa 5.5. Actually, inhibition of the dephosphorylation process (and also the observed inhibition of EP formation from P i in the mutants for Lys 683 and Asp 702 , see Table 1) is consistent with previous mutation studies for these residues with SERCA1a (Lys 684 and Asp 703 in SERCA1a) (36,40,41). The importance of Gly 703 (in the D 702 GVND loop) in the ATPase was revealed here for the first time. It should also be noted that EP formation from ATP was completely lost in K683E (in contrast to K683R), being consistent with the critical role of the positive charge of this residue for the phosphoryl transfer from ATP (36).

Summary of Defects in 51 DD Mutants-We have examined
here the expression and function of 51 DD mutants of SERCA2b, and compared them with those of the wild type. The results showed that, among the 51 mutants, 15 are severely defective in protein expression (Fig. 3), 29 severely defective in ATP hydrolysis (Fig. 4), and four severely defective in Ca 2ϩ transport due to its uncoupling from ATP hydrolysis (Fig. 5). Thus, cellular Ca 2ϩ transport function must be considerably disrupted by these mutations (in total 48) to result in an increase in cytoplasmic Ca 2ϩ and a decrease in endoplasmic reticulum lumenal Ca 2ϩ , hence DD. This is consistent with notions (23,25) that DD is caused by haploinsufficiency in Ca 2ϩ homeostasis or by a possible intermolecular interaction of the defective mutant with the wild type, which may possibly inhibit the function of the wild type or cause its protein degradation. On the other hand, exceptions among the 51 DD mutants were found in the three mutants I274V, L321F, and M719I as they exhibited seemingly normal protein expression and Ca 2ϩ transport function coupled with ATP hydrolysis (Fig. 5). Because these three mutations happened to be in the cases of the three  5)). The amount of EP formed was determined as described under "Experimental Procedures," and was 126 Ϯ 9 (n ϭ 4) pmol/mg of microsomal protein with the expressed wild type. The amount of EP of each of the mutants was normalized to its protein expression level relative to that of the wild type determined in Fig. 3. The amount of EP of the mutant relative to that of the wild type was thus obtained and plotted versus its relative specific Ca 2ϩ -ATPase activity as determined in Fig. 4. The values presented are the mean Ϯ S.D. (n ϭ 3). The dotted line was drawn from 0 through the wild-type level (100%). The values are also summarized in Table 1.  Table 1. The maximal EP levels in pmol/mg of microsomal protein (i.e. the values without correcting by the protein expression levels) obtained at 25°C with the mutants were 85.6 (N767S), 50.8 (V834F), and 86.2 (G310V), and the levels with the wild type at 25 and 0°C were both 126. DD pedigrees found by us (16), we carried out a detailed kinetic analysis and found (26) that the three mutants possess altered kinetic properties, each exhibiting distinct types of abnormality, i.e. a slightly decreased Ca 2ϩ transport rate (I274V and M719I as also shown in Fig. 5), a slightly decreased Ca 2ϩ affinity at the transport sites (L321F and M719I, see Table 1), and a markedly reduced sensitivity to the feedback inhibition by lumenal Ca 2ϩ (L321F). The possible increase in lumenal Ca 2ϩ due to L321F abnormality was suggested to be related to the specific symptoms of this DD pedigree, neuropsychiatric disorder and behavior problems (26). Thus, for the 51 DD pedigrees, our results revealed the molecular defects of SERCA2b as the causes of disease. Taken together, our findings indicate that the DD mutations in most cases cause severe disruption of Ca 2ϩ homeostasis and hence DD but even a slight disturbance still results in DD, and further points to the possible relation between the specific molecular defects of SERCA2b and the specific symptoms associated with DD. It is also important to note that uncoupling in the Ca 2ϩ transport and the reduced sensitivity to feedback inhibition by lumenal Ca 2ϩ in the SERCA2b mutants may reduce the energy charge of cells by wasting or consuming more ATP, which then could be an additional and important pathogenic mechanism for the development of symptoms.
In the following discussions, we further inspected the functional defects of the DD mutants on the basis of the atomic structures of SERCA1a to learn more about the structure and function of SERCAs from the DD mutations. We focused on the three findings with the DD mutations in the transmembrane domain (N767S, A803T, G807R, and V843F) causing uncoupling in the Ca 2ϩ transport, those in the catalytic site (K683R, D702N, and G703S) uniquely exhibiting Ca 2ϩ -induced inhibition of EP formation from ATP, and those in the A domain revealing two critical regions (at Pro 160 and Asp 149 ) of this domain. Other interesting findings are discussed in detail under the supplemental discussion, which cited additional references (Refs. [71][72][73][74][75][76][77][78]. It is important to note that the residues of DD mutations and those described in the supplemental discussion are all conserved in SERCA2b and SERCA1a (as the two are highly homologous), enabling such inspection on the basis of the SERCA1a atomic structures.
Mutations N767S, A803T, G807R, and V843F, Causing Uncoupling-The DD mutants N767S on M5 and V843F on M7 at or near the high-affinity Ca 2ϩ -binding sites, and A803T and G807R on the cytoplasmic side of M6 have never been explored except V843F, for which Ca 2ϩ transport was previously reported to be significantly slowed (25). We found here that these mutations cause complete loss (A803T and V843F) or significant reduction (N767S and G807R) of Ca 2ϩ -transport activity due to strong or partial uncoupling from ATP hydrolysis (Fig. 5). The results indicate the importance of these residues for the formation of the proper transport pathway. Interestingly, no change in Ca 2ϩ affinity at the high-affinity binding  K683R, G703S, and D702N. EP formation from ATP was performed with the expressed wild-type SERCA2b and DD mutants K683R, G703S, and D702N at 25 and 0°C and at various concentrations of free Ca 2ϩ , otherwise as described in the legend to Fig. 6. The maximal amount of EP obtained with the wild type at pCa 5.5 and 25°C was normalized to 100%, and shown with open squares (25°C) and open triangles (0°C). The specific Ca 2ϩ -ATPase activities of the mutants and the wild type were determined and normalized to the activity of the wild type at pCa 5.5 as described in the legend to Fig. 4, and shown with closed circles. It should be noted that the decrease in ATPase activity of the wild type at high concentrations of Ca 2ϩ is known to be due to the inhibition by low-affinity Ca 2ϩ binding to the transport sites of EP (from the lumenal side) and the slowed ATP hydrolysis with CaATP formed at the high Ca 2ϩ level.
sites was observed in Ca 2ϩ activation of the enzyme with A803T and G807R, and a ϳ6-fold reduction with N767S and V843F at 25°C (Fig. 7 and Table 1). The reduction with N767S is consistent with the function of this residue as a Ca 2ϩ ligand predicted with SERCA1a (42)(43)(44). In V843F, the bulky side chain introduced in the vicinity of Ca 2ϩ ligands Asn 767 /Glu 770 and the critical residue Gly 769 of M5 likely affected the Ca 2ϩ affinity (in SERCA1a, Asn 768 /Glu 771 and Gly 770 ).
In all the atomic models of SERCA1a, each of the four residues, Asn 767 , Ala 803 , Gly 807 , and Val 843 , is involved in interac-tions between transmembrane helices (see Fig. 9 for E1⅐AlF x ⅐ADP, and in the SERCA1a numbering they are Asn 768 , Ala 804 , Gly 808 , and Val 844 ). Asn 767 on M5 not only coordinates Ca 2ϩ at site I (44) but also forms hydrogen bonds with Ala 305 and Ala 306 on M4. Ala 803 on M6 is between the bound Ca 2ϩ and cytoplasmic surface of the membrane, and its side chain is situated at the center of the hydrophobic residues clustered from M5 (Phe 759 /Tyr 762 / Leu 763 ), M6 (Pro 802 /Leu 806 ), and M4 (Val 314 ). Therefore Ala 803 on M6 likely functions as a hydrophobic spacer for the proper packing of M6/M4/M5. 3 Gly 807 at the top of M6 forms hydrogen bonds with Cys 318 on M4 and Asn 755 on M5, which also forms hydrogen bonds with residues on L6 -7, thus Gly 807 is involved in interactions of M6 with M4/M5 at the cytoplasmic surface of the membrane. Asn 767 , Ala 803 , and Gly 807 therefore probably function to stabilize the properly packed M4/M5/M6 and to prevent the possible formation of a leakage path from the Ca 2ϩ -binding sites to the cytoplasm. The DD mutations N767S, A803T, and G807R probably disrupted the interaction networks or caused a steric clash between the helices. It might also be possible that these DD mutations disrupted the function of Glu 309 on M4, the cytoplasmic gate to occlude Ca 2ϩ (39,44,45), because the three residues of the DD mutations are close to Glu 309 .
Val 843 on M7 is situated at the slightly lower part (lumenal side) of the bound Ca 2ϩ , and very close to M5 (the Ser 766 -Val 772 region that includes the Ca 2ϩ ligands Asn 767 / Glu 770 and the critical Gly 769 ), M8 (Leu 903 /Ile 906 /Glu 907 ), and M10 (Val 976 ), e.g. within 3.4 Å to Gly 769 , Ile 906 , and Val 976 in E1⅐AlF x ⅐ADP. It is likely that the introduced bulky side chain in FIGURE 9. Detailed structure at Asn 767 , Ala 803 , Gly 807 , and Val 843 . The structures were depicted with E1⅐AlF x ⅐ADP, the E1PCa 2 ⅐ADP atomic model (PDB accession code 1WPE (53)) with the numbering of the residues for SERCA2b (in SERCA1a, they are Asn 768 , Ala 804 , Gly 808 , and Val 844 ). The color changes gradually from the N terminus (red) to the C terminus (blue). Potential hydrogen bonds are shown with dotted green lines. Residues are depicted by a ball and stick model with the atoms oxygen (red), nitrogen (blue), carbon (cyan), and sulfur (yellow). a, locations of the four residues in the transmembrane domain and the bound two Ca 2ϩ ions (yellow spheres). b-e, enlarged pictures at each of the residues. d, Asn 767 on M5 coordinates Ca 2ϩ at the binding site I (44) and also forms hydrogen bonds with the main chain carbonyls of Ala 305 (which also coordinates Ca 2ϩ at Site II (44)) and of Ala 306 on M4. c, Ala 803 on M6 is between the Ca 2ϩ -binding sites and cytoplasmic surface of the membrane (e.g. between Asn 767 and Gly 807 , see panel a), and its side chain is situated at the center of the hydrophobic residues clustered from M5 ( V843F caused a steric clash between these helices and disrupted their proper packing, and thus disrupted the formation of a proper Ca 2ϩ release pathway to the lumen. It is of interest to note that Ca 2ϩ transport in A803T and V843F was strongly uncoupled (i.e. no transport activity) but it was only partially uncoupled in N767S and G807R (i.e. the transport activity was 42-44% of the ATPase activity, see Fig. 5 and Table 1). The difference in the degree of uncoupling might be due to the structural facts described above that Ala 803 and Val 843 are extensively involved in the hydrophobic interactions between the transmembrane helices, whereas Asn 767 and Gly 807 are involved only for (a few) hydrogen bonds. It is possible that the disruption of the proper hydrophobic interactions have much more serious consequences on the proper packing of the transmembrane helices for formation of the proper transport pathway. Regarding G807R, it is also possible that the structural disruption at the cytoplasmic surface affected less seriously the transmembrane region.
Unique Behavior of Mutants K683R, D702N, and G703S at the Catalytic Site-These mutations at the binding sites for the catalytic cofactor Mg 2ϩ and phosphate moiety (the 702 DGVND loop and Lys 683 , respectively) uniquely exhibited the inhibition of Ca 2ϩ -activated EP formation from ATP by high concentrations of Ca 2ϩ (Fig. 8). The more profound inhibition at 0°C than at 25°C suggests the importance of possible thermal motions in the conformation of the enzyme and/or the bound ATP for relieving this inhibition. It is possible that the coordination of Ca 2ϩ became more favored as compared with Mg 2ϩ (especially at the lower temperature) in the mutated catalytic site and inhibited the phosphorylation. In phosphoserine phosphatase, a member of the haloacid dehalogenase superfamily of SERCAs, the coordination geometry of Ca 2ϩ at the catalytic site was previously revealed to be different from that of the native cofactor Mg 2ϩ and therefore not suitable for catalysis (46). For another possible cause of the Ca 2ϩ -induced inhibition of phosphorylation observed here, it is of interest to note that there are two distinct binding sites for the phosphate moiety with the coordinated Mg 2ϩ (i.e. Asp 351 and Thr 441 ), and also two different conformations of ATP (i.e. extended and folded ones). In the E1⅐AMPPCP crystals (38,39), the bound AMPPCP (ATP) adopts an uncommon extended conformation (Fig. 10, and also see Fig. 4 in Ref. 39 (47,49). In this case, MgATP possibly adopts its folded conformation (see also Fig. 8 in Ref. 47). To bring ␥-phosphate and Mg 2ϩ from Thr 441 to Asp 351 , the enzyme with bound MgATP should have conformational changes to produce closure of the N and P domains and their cross-linking by MgATP in its extended form, thereby Asp 351 is phosphorylated. Actually, the structure E1⅐AMPPCP fixed in the crystals with the extended Mg-AMP-PCP (38,39) was recently interpreted as a transient conformation for phosphoryl transfer (48). Consistently, the rate-limiting step for EP formation from E1Ca 2 and MgATP was previously shown to be the conformational change in the E1Ca 2 ⅐MgATP complex (to bring ␥-phosphate to Asp 351 (50)) preceding the phosphoryl transfer (50 -52). This conformational change is largely slowed with CaATP even in the native enzyme (52). It is tempting to speculate that the binding of CaATP (formed at the high Ca 2ϩ level) to the Thr 441 site became more significant as compared with the binding to the Asp 351 site when the structure around Asp 351 is disrupted by DD mutations K683R, D702N, and G703S. In any case, the native structure of the catalytic site with Lys 683 and the FIGURE 10. Detailed structure at Lys 683 , Asp 702 , and Gly 703 in the catalytic site with bound MgAMPPCP. The enlarged picture for the P and N domain was depicted with E1⅐AMPPCP, the E1Ca 2 ⅐MgATP atomic model (PDB accession code 1VFP (39)) with the numbering of the residues for SERCA2b (in SERCA1a, Lys 684 , Asp 703 , and Gly 704 ). The bound AMPPCP in its extended and zigzag conformation (wire model), Mg 2ϩ (green sphere), water molecules (purple spheres), Asp 351 (phosphorylation site), Lys 683 , Asp 702 /Gly 703 /Asp 706 on the 702 DGVND loop (yellow loop), and Phe 487 (binding site for the adenine moiety) are shown. The residues are depicted by a ball and stick model with red (oxygen), blue (nitrogen), and cyan (carbon). Potential hydrogen bonds and coordination to Mg 2ϩ are shown with dotted green lines and dotted blue lines, respectively. Gly 703 is important as indicated here by its DD mutation, probably for stabilizing the proper structure of the 702 DGVND loop by its hydrogen-bonding network with Asp 702 and Asp 706 within the loop. Thr 441 was predicted as a binding site for Mg 2ϩ coordinated by the ␤/␥-phosphate of ATP (47,49,67), which possibly adopts the folded conformation as depicted by the thin wire model for ATP with the small green sphere for Mg 2ϩ (the PDB code 1KAX (68)) according to Ref. 47. The binding of ATP in the folded conformation was also previously inferred with Na ϩ ,K ϩ -ATPase (69) and most recently with Ca 2ϩ -ATPase (48).
domain/M3 linker, the strain of which has also been predicted to be critical as a moving force for the A domain rotation in the dephosphorylation process (53). The DD mutations of Pro 160 , Gly 211 , and Asn 39 probably disrupted these structural bases of the A domain and the connected linkers, and possibly inhibited the proper motions of the A domain required for the dephosphorylation process. It should be noted that the Pro 160 3 Ala mutant in SERCA1a was previously found to possess normal Ca 2ϩ transport function (62). The serious consequence of the DD mutations P160L and P160H observed here is therefore probably due to a steric clash caused by the introduced bulky side chains, which is consistent with the importance of the appropriate interaction network involving the Pro 160 main chain in this region.
Asp 149 is located slightly more to the outer (or upper) part of the A domain than Pro 160 (Fig. 11a). All the DD mutations of the residues located at this Asp 149 region, Asp 149 , Gly 23 , Arg 131 , and Val 223 , caused a large reduction in protein expression (Fig.  3) and markedly reduced or no EP formation from ATP and P i (thus disrupting even the structure at the phosphorylation site far from this region, see Fig. 6 and Table 1 for the analyzed mutants D149N and G23E). These residues form an extensive interaction network (Fig. 11b and the legend for details). Asp 149 , Arg 131 , and Val 223 are each located on different layers of the bent loop/␤-strand, which are gathered by interaction networks to constitute the A domain with the "distorted jellyrolllike structure" (44). The residues are also at the top of the three linkers that connect the A domain to M1 (Gly 23 /Leu 24 ), M2 (Arg 131 ), and M3 (Val 223 ), respectively. The DD mutations D149N, G23E, R131Q, and V223M probably caused a steric clash and disrupted the interactions in this region, and seriously affected the protein structure. The residues and their interaction network at the Asp 149 region therefore seem to be critical in the formation and stabilization of the fundamental structure of the A domain with the properly positioned linkers. In a detailed kinetic analysis, the importance of Gly 23 in function was also previously noted (23).