A Darier disease mutation relieves kinetic constraints imposed by the tail of sarco(endo)plasmic reticulum Ca2+-ATPase 2b

The sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA) 2b isoform possesses an extended C terminus (SERCA2b tail) forming an 11th transmembrane (TM) helix, which slows conformational changes of the Ca2+-pump reaction cycle. Here, we report that a Darier disease (DD) mutation of SERCA2b that changes a glutamate to a lysine in the cytoplasmic loop between TM8 and TM9 (E917K) relieves these kinetic constraints. We analyzed the effects of this mutation on the overall reaction and the individual partial reactions of the Ca2+ pump compared with the corresponding mutations of the SERCA2a and SERCA1a isoforms, lacking the SERCA2b tail. In addition to a reduced affinity for Ca2+, caused by the mutation in all three isoforms examined, we observed a unique enhancing effect on the turnover rates of ATPase activity and Ca2+ transport for the SERCA2b E917K mutation. This relief of kinetic constraints contrasted with inhibitory effects observed for the corresponding SERCA2a and SERCA1a (E918K) mutations. These observations indicated that the E917K/E918K mutations affect the rate-limiting conformational change in isoform-specific ways and that the SERCA2b mutation perturbs the interactions of TM11 with other SERCA2b regions. Mutational analysis of an arginine in TM7 that interacts with the glutamate in SERCA1a crystal structures suggested that in wildtype SERCA2b, the corresponding arginine (Arg-835) may be involved in mediating the conformational restriction by TM11. Moreover, the E917K mutation may disturb TM11 through the cytoplasmic loop between TM10 and TM11. In conclusion, our findings have identified structural elements of importance for the kinetic constraints imposed by TM11.

enzymes exhibit lower Ca 2ϩ affinity than SERCA1a (14), whereas SERCA2b exhibits higher Ca 2ϩ affinity than both SERCA1a and SERCA2a (15, 18 -20). SERCA2b is furthermore characterized by significantly lower rates of the major rate-limiting conformational changes of the enzyme cycle compared with the other isoforms, which may be ascribed to the presence of the SERCA2b tail (15,20).
SERCA2b is the predominant isoform in the human skin (25,26). A variety of mutations in SERCA2b have been found to result in Darier disease (DD), an autosomal dominant skin disorder characterized by dyskeratosis and acantholysis, often associated with neuropsychiatric abnormality (26 -30). The skin manifestations seem to be caused by abnormal differentiation of the keratinocytes and lack of desmosomes between keratinocytes due to depletion of endoplasmic reticulum Ca 2ϩ stores with resulting apoptosis and change of the Ca 2ϩ gradient normally existing across the epidermal layers (26,31).
In this study, we have analyzed the functional consequences of the DD mutation E917K, which targets a glutamate located in the cytoplasmic loop between TM8 and TM9 (32). We unexpectedly found this mutation to relieve the kinetic constraints normally imposed by the SERCA2b tail. To understand the role of the SERCA2b tail in the functional effects of the SERCA2b E917K mutation, we replaced the corresponding residue in SERCA2a (E917K) and SERCA1a (E918K and E918A), which do not possess the tail characteristic of SERCA2b. Furthermore, we studied the effects of alanine mutations of Arg-835 in SERCA2a/2b and Arg-836 in SERCA1a, as this arginine residue, located in TM7, interacts with Glu-918 in the crystal structures of SERCA1a. The results demonstrate the importance of both Glu-917 and Arg-835 of SERCA2b in mediating the kinetic effects of the SERCA2b tail.

Expression level, maximal activity, and Ca 2؉ dependence of SERCA2a and SERCA2b E917K and SERCA1a E918K and E918A mutants
Because COS-1 cells contain little endogenous SERCA, they are ideal for expression and functional studies of SERCA iso-forms and mutants (8,11,15,18,21) and were used here. First, we examined DD mutation E917K in SERCA2b (human) and the corresponding mutations E917K in SERCA2a (human) and E918K in SERCA1a (rabbit). The expression levels of these mutants in the COS-1 cells were significantly lower than the expression levels of the respective wildtypes, and more so for the SERCA2 mutants (ϳ5-and 10-fold reduced for SERCA2b and SERCA2a, respectively) than for the SERCA1a mutant (2-fold reduced). ATPase activity and Ca 2ϩ transport measurements presented below were corrected for expression level by calculating the turnover rates (molecular rates, i.e. rate of ATP hydrolysis or Ca 2ϩ transport per active enzyme). In agreement with previous findings (11,15,20), the present data show that the maximal turnover rate of ATPase activity of wildtype SERCA2b is only about one-third that of the SERCA1a wildtype, and the same is the case for the maximal turnover rate of the ATP-dependent active Ca 2ϩ uptake in the microsomal vesicles determined under conditions of saturation of the highaffinity Ca 2ϩ -binding sites (5 M free Ca 2ϩ , Fig. 1, and Table 1). SERCA2a showed maximal rates intermediate between those of SERCA2b and SERCA1a, also in good agreement with the previous findings (15,20). This difference between the wildtypes illustrates the kinetic constraints imposed on SERCA2b by its tail. Unexpectedly, the present results with the mutants revealed that the maximal turnover rate of the ATPase activity of DD mutant SERCA2b E917K is 1.3-fold higher than that of the SERCA2b wildtype, whereas the corresponding SERCA1a E918K mutant showed a significant 1.6-fold reduction of the maximal turnover rate of ATPase activity relative to wildtype SERCA1a. A similar difference between the effects of the corresponding mutations in SERCA2b and SERCA1a was noted for the turnover rate of Ca 2ϩ transport, being 1.5-fold enhanced for E917K in SERCA2b and reduced as much as 2.5-fold for E918K in SERCA1a, relative to the corresponding wildtype (Table 1). Thus, the inhibition of the SERCA2b wildtype relative to the SERCA1a wildtype is relieved by the E917K mutation SCHEME 1. SERCA reaction cycle. The major enzyme conformations are indicated as E 1 and E 2 . P indicates phosphorylation at the conserved P-domain aspartic acid residue. Two Ca 2ϩ ions are transported in each enzyme cycle. The proton counter transport is indicated here and throughout the text by H n , where n most likely is 2 at physiological pH, but may vary dependent on pH (16). ionophore A23187, and the CaCl 2 concentration required to obtain maximal activity (cf. Fig. 2). The turnover rate was calculated relative to the active-site concentration determined by phosphorylation under stoichiometric conditions. All the individual data points collected are shown as circles, and the height of the columns show the average values. The vertical distance between the top of the column and the horizontal error bars indicates the S.D.; for further information on statistics, see Table 1.

Darier disease mutation E917K of SERCA2b
in SERCA2b. A relation of this finding to the presence of the SERCA2b tail (TM11 plus LE) was revealed by the data obtained with the SERCA2a E917K mutant and wildtype, which do not possess the SERCA2b tail but are otherwise identical to mutant and wildtype SERCA2b. In the SERCA2a protein environment, the mutation was inhibitory similar to the finding with SERCA1a, although to a lesser extent ( Fig. 1 and Table 1), thus indicating that the activation seen for SERCA2b E917K is the consequence of a change of the interaction of the SERCA2b tail, relieving its inhibitory influence.
In contrast to the differential effects of the E917K/E918K mutation on the maximal turnover rates in the three isoforms, the mutation exerted similar effects on the Ca 2ϩ activation profiles of the three isoforms studied, reducing the apparent Ca 2ϩ affinity in all cases (Fig. 2 and Table 1, K 0.5 3.7-, 2.6-, and 4.8fold increased for SERCA2b E917K, SERCA1a E918K, and SERCA2a E917K, respectively, relative to wildtype, where K 0.5 is the ligand concentration giving half-maximum effect).
The effects of the SERCA1a E918A mutation were also characterized. Like mutant E918K, E918A showed a significantly reduced maximal rate and reduced apparent Ca 2ϩ affinity, although the affinity decrease was less pronounced compared with E918K (Table 1).

Vanadate sensitivity of SERCA2a and SERCA2b E917K and SERCA1a E918K and E918A mutants
Vanadate is an analog of the phosphoryl group in the transition state occurring during dephosphorylation of H n E 2 P (cf. Scheme 1). Therefore, it binds strongly to P-type ATPases in the E 2 conformation, inhibiting the ATPase activity (33,34). SERCA2a and SERCA2b E917K showed 2-3-fold increase of the apparent vanadate affinity for inhibition of steady-state ATPase activity (reduced K 0.5 ) compared with the corresponding wildtypes. The SERCA1a E918K and E918A mutants likewise showed increased apparent vanadate affinity, as much as 6 -7-fold for E918K ( Fig. 3 and Table 1). Because vanadate binds to E 2 states of the enzyme and not to E 1 states, the increase of the apparent vanadate affinity may result from accumulation of H n E 2 during enzyme turnover, which could occur as a consequence of an increased rate of formation of this intermediate through the phosphoenzyme processing steps Ca 2 E 1 P 3 Ca 2 E 2 P 3 E 2 P 3 H n E 2 P 3 H n E 2 and/or a slowing of the Ca 2ϩ -binding reaction sequence H n E 2 3 H n E 1 3 E 1 3 Ca 2 E 1 (cf. Scheme 1). Studies of the partial reactions described below indicate that both kinds of effects are involved here and that their relative contributions depend on the isoform.

Transient kinetics of phosphorylation of SERCA2a and SERCA2b E917K and SERCA1a E918K and E918A mutants
The transient kinetics of phosphorylation was studied at pH 7.0 and 25°C using a Bio-Logic QFM-5 quenched-flow module for rapid mixing. Phosphorylation was initiated by mixing Ca 2ϩ -saturated enzyme with 2.5 M [␥-32 P]ATP. The observed phosphorylation time courses show a phosphorylation overshoot of variable size, before steady state is reached (Fig. 4). The overshoot appears because of accumulation of dephosphoen- a Expression levels (shown relative to the corresponding wildtype) were determined by measuring the level of phosphorylation under stoichiometric conditions (see "Experimental procedures"). b Maximal turnover rate of ATPase activity (mol of phosphate liberated/s/mol of enzyme expressed). c Maximal turnover rate of Ca 2ϩ transport (mol of Ca 2ϩ taken up in microsomes/min/mol of enzyme expressed). Note that turnover rates of ATPase activity and Ca 2ϩ transport were determined under different conditions (see "Experimental procedures") and therefore are comparable only on a relative basis relating the values to that of the corresponding wildtype. d The K 0.5 value for Ca 2ϩ activation was determined by curve fitting as described in the legend to Fig. 2. e The K 0.5 value for vanadate inhibition was determined by curve fitting as described in the legend to where h is the Hill coefficient (here between 1 and 2); V is the ATPase activity at the given Ca 2ϩ concentration, and V max is the maximum ATPase activity, in the graph used to normalize the data such that the maximum is 100%. The dotted line represents the SERCA1a wildtype from the right panel for direct comparison. The extracted K 0.5 values with statistics are presented in Table 1 ("Ca 2ϩ affinity").

Darier disease mutation E917K of SERCA2b
zyme at steady state. The overshoot was minimal for the SERCA1a wildtype. The SERCA2b wildtype showed a larger overshoot, and the overshoot of SERCA2a was intermediate. For all mutants, the overshoot was larger than that of the corresponding wildtype, which is in line with the accumulation of H n E 2 predicted from the increased vanadate sensitivity, as described above. The enzyme kinetics simulation computer software SimZyme (35) was used to estimate relevant rate constants by comparison of computed time courses with the experimental data points (Fig. 4). The computation is based on the simplified reaction cycle shown in Table 2, in which the first rate constant, k A , represents the phosphorylation of Ca 2 E 1 , forming Ca 2 E 1 P; the second rate constant, k B , represents the sequence of phosphoenzyme processing steps Ca 2 E 1 P 3 Ca 2 E 2 P 3 E 2 P 3 H n E 2 P 3 H n E 2 , and the last rate constant, k C , represents the Ca 2ϩ -binding reaction sequence H n E 2 3 H n E 1 3 E 1 3 Ca 2 E 1 (cf. Scheme 1). A lowering of k A reduces the steepness of the rise of the phosphorylation overshoot as well as the magnitude of the overshoot, whereas the overshoot increases if k C is reduced. The k B, as well, has impact on the size of the overshoot, which becomes smaller with reduced k B and larger if k B is increased. Furthermore, the steepness of the decay phase of the overshoot increases with an increase of k B . Because of the strong dependence of the size and shape of the overshoot on the three rate constants, there is generally not much latitude in the determination of the rate constants by curve fitting when an overshoot of significant magnitude is present. As seen in Fig.  4, good fits to the data could be obtained with a phosphorylation rate constant k A ϭ 27 s Ϫ1 for SERCA1a, SERCA2a, and SERCA2b wildtypes as well as mutants (note that due to the ATP concentration of 2.5 M present during this experiment, the k A value is lower than that previously determined for SERCA1a at 5 M ATP (35 s Ϫ1 (35)). SERCA2a showed a small increase of k B and a lower k C compared with SERCA1a wildtype, and the k B and k C values determined for the SERCA2b wildtype were both lower than that of SERCA1a, in agreement with previous findings (15). To obtain a good fit to the data for SERCA2a E917K and SERCA2b E917K, k B had to be increased, as much as 2.3-fold for SERCA2b E917K and 1.5-fold for SERCA2a E917K. Furthermore, k C had to be reduced 2-fold, relative to the respective wildtypes. Hence, the above described increase of the maximal turnover rate of ATPase activity and Ca 2ϩ transport induced by E917K in SERCA2b may be attributed to an increased rate of the sequence of phosphoenzyme processing steps For the SERCA1a E918K mutant, we found a decrease of both k B and k C relative to wildtype SERCA1a, which is reflected in the reduced maximal turnover rate of ATPase activity and Ca 2ϩ transport of this mutant. The reduction of k C (5.7-fold) required to fit the SERCA1a E918K data is larger than the 2-fold reduction required for E917K in SERCA2a and SERCA2b. SERCA1a E918A showed the same tendency as E918K although not as pronounced.

Rates of individual partial reactions of SERCA2b E917K and SERCA1a E918K and E918A mutants
To study more directly the mutational effect on the rate of the where V is the ATPase activity at the given vanadate concentration, and V 0 is the ATPase activity in the absence of vanadate, in the graph used to normalize the data such that the maximum is 100%. The dotted line represents the SERCA1a wildtype from the right panel for direct comparison. The extracted K 0.5 values with statistics are presented in Table 1 ("Vanadate affinity").  Table  2, using the SimZyme program with the rate constants indicated in Table 2.

Table 2 Rate constants of partial reactions determined by computational fitting of the experimental time courses of phosphorylation
The enzyme kinetics simulation computer software "SimZyme" (35) was used to compute time courses fitting the experimental data points as shown in Fig. 4 and thereby estimate the rate constants k A , k B , and k C . The simplified reaction cycle used as the basis for the computation is shown.

Darier disease mutation E917K of SERCA2b
Scheme 1), the enzyme was phosphorylated from [␥-32 P]ATP in the presence of 10 M Ca 2ϩ under conditions (0°C, presence of 80 mM K ϩ , neutral pH) known to make the Ca 2 E 1 P 3 E 2 P transition rate-limiting for the dephosphorylation in the SERCA1a and SERCA2b wildtypes, thus leading to accumulation of Ca 2 E 1 P (15,20). To follow the dephosphorylation of E 1 PCa 2 through the Ca 2 E 1 P 3 Ca 2 E 2 P 3 E 2 P 3 H n E 2 P 3 H n E 2 reaction sequence (cf. Scheme 1), the phosphoenzyme was chased by removing Ca 2ϩ with excess EGTA, thereby preventing rephosphorylation (Fig. 5, upper panels, circles, and Table 3 column designated "Ca 2 E 1 P 3 E 2 P"). The observed dephosphorylation rate was ϳ1.7-fold enhanced for SERCA2b E917K compared with the wildtype SERCA2b, whereas it was about 2-fold slower in SERCA1a E918K and E918A than in the SERCA1a wildtype. These data confirm the isoform-specific mutational effects on k B deduced by the computational analysis of the phosphorylation time course described above. Because of the rate limitation of the dephosphorylation imposed by the Ca 2 E 1 P 3 E 2 P transition, the observed dephosphorylation time course reflects Ca 2 E 1 P 3 E 2 P and not H n E 2 P 3 H n E 2 . To verify that prior to the EGTA chase all of the accumulated phosphoenzyme really was the Ca 2 E 1 P species, which is ADP sensitive, unlike the ADP-insensitive E 2 P, an experiment was carried out in which ADP was added together with EGTA allowing Ca 2 E 1 P to react backwards with ADP to form ATP and dephosphorylated Ca 2 E 1 , which resulted in almost complete dephosphorylation within 5 s both in wildtypes and mutants (Fig. 5, upper panels, triangles).
One of the factors ensuring that Ca 2 E 1 P 3 E 2 P is rate-limiting for the dephosphorylation just described is the presence of 80 mM K ϩ , which accelerates the H n E 2 P 3 H n E 2 step (36). H n E 2 P 3 H n E 2 is slower at low K ϩ concentration, allowing analysis of this reaction, as well. Incubation of the enzyme with 32 P i in the absence of Ca 2ϩ at acid pH and in the presence of the organic solvent dimethyl sulfoxide leads to accumulation of radioactively labeled H n E 2 P formed backwards from H n E 2 . The enzyme phosphorylated in this way was chased by addition of non-radioactive P i and dilution in dimethyl sulfoxide-free medium of neutral pH and a low K ϩ concentration (Fig. 5, lower  panels). Under these conditions, SERCA2b E917K exhibited a 3-fold faster dephosphorylation, i.e. rate of H n E 2 P 3 H n E 2 , than the SERCA2b wildtype, whereas SERCA1a E918K and E918A showed dephosphorylation rates similar to that of the SERCA1a wildtype (Fig. 5, lower panels, and Table 3 column designated "H n E 2 P 3 H n E 2 ").
The Ca 2ϩ -binding transition of the dephosphoenzyme H n E 2 3 Ca 2 E 1 (i.e. H n E 2 3 H n E 1 3 E 1 3 Ca 2 E 1 in Scheme 1) was studied at 25°C using the quenched-flow module. The enzyme was preincubated in the absence of Ca 2ϩ at pH 6.0, leading to accumulation of H n E 2 , followed by addition of Ca 2ϩ to initiate the transition to Ca 2 E 1 . The time course of formation of Ca 2 E 1 was observed by adding [␥-32 P]ATP at various time intervals followed in each case by quenching 34 ms later, thus taking advantage of the exclusive ability of Ca 2 E 1 to become phosphorylated by ATP. Under these conditions the appearance of phosphorylated Ca 2 E 1 P enzyme reflects the appearance of Ca 2 E 1 (35). Using the same method, we have previously found that the Ca 2ϩ -binding transition is markedly slower in the SERCA2b wildtype as compared with SERCA2a and SERCA1a wildtypes (20), and the lower rate of the SERCA2b wildtype was confirmed here. The rate observed for SERCA2b E917K was slightly enhanced relative to the SERCA2b wildtype (0.38 Ϯ 0.03 versus 0.30 Ϯ 0.01 s Ϫ1 ), whereas SERCA1a E918K was slower by 2-fold than the SERCA1a wildtype, thus approaching the rate of the SERCA2b wildtype (Fig. 6, upper panels, and Table 3 column designated "H n E 2 3 Ca 2 E 1 ").
The pH of 6.0 applied in the experiment just described ensures that in the wildtypes the proton release step H n E 2 3 H n E 1 3 E 1 is rate-limiting for the reaction sequence leading to Ca 2 E 1 . The evidence that such a proton release step exists and becomes rate-limiting at low pH (37) is in line with the proton counter transport steps shown in Scheme 1. To obtain information on the step involved directly in the binding of Ca 2ϩ after the release of protons, i.e. E 1 3 Ca 2 E 1 , we increased the pH to 7.0, thereby allowing E 1 to accumulate during the pre-incubation in the absence of Ca 2ϩ (37). Under these conditions both SERCA2b E917K and SERCA1a E918K displayed a reduced rate of phosphoenzyme appearance relative to the wildtype, 4-and 2.7-fold, respectively (Fig. 6, lower panels, and Table 3 column designated "E 1 3 Ca 2 E 1 "). This finding is in accordance with the transient kinetic analysis of the phosphorylation time course, indicating a reduced k C for both SERCA2b E917K and SERCA1a E918K relative to the corresponding wildtype at pH To follow the dephosphorylation, the phosphoenzyme was chased by removing free Ca 2ϩ with EGTA added at a final concentration of 9 mM, thus preventing further phosphorylation. ADP sensitivity was demonstrated by including 1 mM ADP in the chase medium (triangle in the lower left corner). Acid quenching was performed at the indicated time intervals after the chase. For H n E 2 P dephosphorylation, phosphorylation was carried out at 25°C for 10 min in medium containing 0.5 mM 32 P i , 100 mM Mes/Tris (pH 6.0), 10 mM MgCl 2, 2 mM EGTA, and 30% (v/v) dimethyl sulfoxide. Following cooling at 0°C, the phosphoenzyme was chased by a 19-fold dilution with 40 mM ice-cold MOPS/Tris (pH 7.0), 10 mM KCl, 2 mM EGTA, 2 mM MgCl 2 , and 0.5 mM non-radioactive P i , and acid quenching was performed at the indicated time intervals after the chase. Error bars (seen only when larger than the size of the symbols) indicate S.D. The lines show the best non-linear regression fits of a monoexponential decay function EP ϭ EP⅐e Ϫk⅐t , giving the rate constants listed with statistics in Table 3 ("Ca 2 E 1 P 3 E 2 P" corresponding to the upper panels of this figure and "H n E 2 P 3 H n E 2 " corresponding to the lower panels). The initial level of phosphorylation corresponding to the regression line was taken as 100%.

Darier disease mutation E917K of SERCA2b
7.0 ( Fig. 4 and Table 2). Because the rate of the E 1 3 Ca 2 E 1 step was only 0.34 s Ϫ1 in SERCA2b E917K, it is clear that in this mutant the E 1 3 Ca 2 E 1 step contributes very significantly to rate limitation of the Ca 2ϩ -binding transition H n E 2 3 H n E 1 3 E 1 3 Ca 2 E 1 at pH 7.0. The slightly increased rate of the complete transition described above for SERCA2b E917K at pH 6.0 seems therefore to be due to a balance between the reduced rate of E 1 3 Ca 2 E 1 and an acceleration of the H n E 2 3 H n E 1 3 E 1 step. Hence, the acceleration is likely more prominent than immediately apparent from the rate constant determined at pH 6.0, and it may be concluded from this analysis that the effects of E917K/E918K on the conformational transition H n E 2 3 E 1 in SERCA2b and SERCA1a are opposite, relieving inhibition in the former isoform and causing inhibition in the latter, whereas in both isoforms the E 1 3 Ca 2 E 1 step is slowed by the mutation.

Analysis of SERCA2a and SERCA2b R835A and SERCA1a R836A mutants
The functional consequences of replacing Arg-835 in SERCA2a and SERCA2b and the corresponding Arg-836 in SERCA1a with alanine were studied as well, because Arg-835/ 836 seems to interact with Glu-917/918 (Fig. 7). The same strategy as described for the other mutants above was applied, characterizing first the overall reaction and then the rates of the partial reaction steps. SERCA2b R835A greatly enhanced the maximal turnover rate of ATPase activity and Ca 2ϩ transport relative to wildtype SERCA2b, in fact more than seen for E917K. For SERCA2a and SERCA1a, this mutation also enhanced the turnover rate, but to a lesser extent. Like E917K/ E918K, the R835/836A mutation reduced the apparent Ca 2ϩ affinity and increased the apparent affinity for vanadate significantly in all three isoforms (Table 1). These results were supported by analysis of the transient kinetics of phosphorylation of the R835/836A mutants performed as for Fig. 3, showing for all three mutants a significant enhancement of k B , which was most pronounced for SERCA2b, and reduction of k C , relative to the corresponding wildtype (Table 2), which should lead to accumulation of the vanadate-reactive form, H n E 2 , during enzyme turnover at steady state, thus explaining the increased apparent affinity for vanadate.
Using the same procedures as for Figs. 5 and 6, the rates of the partial reactions were determined directly for SERCA1a R836A and SERCA2b R835A ( Table 3). The Ca 2 E 1 P 3 E 2 P transition was wildtype-like in SERCA1a R836A, but was markedly enhanced in SERCA2b R835A relative to the SERCA2b wildtype. The rate of the H n E 2 P 3 H n E 2 dephosphorylation step showed 4-and 6-fold increases in SERCA1a R836A and SERCA2b R835A, respectively. Furthermore, R836A/R835A reduced the rate of the H n E 2 3 Ca 2 E 1 transition markedly in both isoforms.

Discussion
Prompted by the finding that a sporadic mutation changing Glu-917 of SERCA2b to lysine causes DD (32), we have studied the functional consequences of this mutation and the corresponding mutations in SERCA2a and SERCA1a. In DD, the epidermal keratinocytes are prone to develop endoplasmic Table 3 Rate

constants of partial reactions determined directly by measuring kinetics of phosphorylation and dephosphorylation
Rate constants Ϯ S.D. were extracted from the data in Figs. 5 and 6 and data from similar experiments with E918A, R836A, and R835A by non-linear regression curve fitting as described in the figure legends. The number of independent experiments from which the data are derived is indicated as n. ND means not determined.   Table 1 ("H n E 2 3 Ca 2 E 1 " corresponding to the upper panels of this figure and "E 1 3 Ca 2 E 1 " corresponding to the lower panels). See a previous description of this analysis (43) for the rationale of the offset. In each case, the phosphorylation level at infinite time extracted from the fit was taken as 100%.

Darier disease mutation E917K of SERCA2b
reticulum stress response with resulting apoptosis and characteristic dyskeratotic features. As part of this complex process, defective SERCA2b function leads to endoplasmic reticulum Ca 2ϩ store depletion of the keratinocytes (26) with consequences for Ca 2ϩ signaling at various levels, including changes in the expression levels of various proteins involved, such as store-operated Ca 2ϩ channels (38). Using the mammalian COS-1 cell expression systems, we have found both a 5-fold reduced expression level and a reduced apparent Ca 2ϩ affinity of the DD mutant E917K as compared with the wildtype SERCA2b. Assuming that in the patients the mutation reduces the expression level of SERCA2b in the keratinocytes to a similar extent as seen for the COS-1 cells, the net effect of the mutation on the endoplasmic reticulum Ca 2ϩ uptake may well be a decrease, in line with current views on the pathophysiological mechanism (26,38). However, we also found that the E917K mutation leads to a 30 -50% increase of the maximal turnover rates of ATPase activity and Ca 2ϩ transport determined at saturating Ca 2ϩ concentration. Because the Ca 2ϩ levels can become quite high in restricted subcellular microdomains, and the spatial extent and temporal persistence of Ca 2ϩ signals may be significantly shaped by subpopulations of SERCA operating at high or even saturating Ca 2ϩ , it cannot be excluded that the increased maximal turnover rate contributes to the pathophysiology. The increased maximal turnover rate observed for SERCA2b E917K, relative to the wildtype SERCA2b, is isoform-specific and dependent on the presence of the SERCA2b tail, because the equivalent SERCA2a and SERCA1a mutants instead showed reduced maximal turnover rates relative to the corresponding wildtypes. Therefore, our observation sheds new light on the structure-function relationship that determines the kinetic constraints imposed by the SERCA2b tail. Our analysis of the partial reaction steps showed for SERCA2b E917K increased rate of both the Ca 2 E 1 P 3 E 2 P transition and the H n E 2 P 3 H n E 2 dephosphorylation step ( Table 3). Either of these steps contributes to rate limitation of the overall reaction at the conditions applied for measurement of maximal turnover of ATPase activity and Ca 2ϩ transport (15,39). On the contrary, the E918K mutation in SERCA1a markedly slowed the Ca 2 E 1 P 3 E 2 P transition, thus explaining the reduced maximal turnover rate of the SERCA1a mutant. In SERCA1a, the mutation was without effect on the rate of H n E 2 P 3 H n E 2 (Table 3). Another isoform-specific difference that appears from our analysis is that in SERCA2b the mutation E917K caused acceleration of the H n E 2 3 E 1 step of the Ca 2ϩ -binding transition of the dephosphoenzyme, whereas in SERCA1a E918K, we found H n E 2 3 E 1 slower than in the corresponding wildtype. This step is not likely to contribute to rate limitation of the overall reaction at the neutral pH and millimolar ATP concentration present during measurements of ATPase activity and Ca 2ϩ transport (39), but the acceleration of H n E 2 P 3 H n E 2 by E917K constitutes yet another example of relief of the inhibitory influence of the SERCA2b tail caused by the E917K mutation. It was previously found that the presence of the SERCA2b tail slows the phosphoenzyme-processing steps Ca 2 E 1 P 3 E 2 P and H n E 2 P 3 H n E 2 as well as the H n E 2 3 E 1 step of the Ca 2ϩbinding transition, and that the two main parts of the SERCA2b tail, the 11th transmembrane helix (TM11) and the associated 11-amino acid LE, play differential roles in the inhibition, with the TM11 contributing to the slowing of all the three steps mentioned and the LE contributing only to the slowing of Ca 2 E 1 P 3 E 2 P (20). Hence, it is clear that the relief of the inhibition by the SERCA2b tail caused by the E917K mutation can be ascribed to an influence on the interaction of TM11, whereas the interaction of the LE may or may not be influenced by the mutation.
In the absence of X-ray crystallographic evidence, there are two possibilities for the location of TM11 in SERCA2b, as judged from the known SERCA1a structure (Fig. 7). Either

Darier disease mutation E917K of SERCA2b
TM11 is on "the TM7 side" or it is on "the TM8 side" of TM10 (blue and black dashed lines, respectively, in Fig. 7). Based on functional effects of mutations in TM7 and TM10, a structural model of SERCA2b was previously generated, where TM11 was docked at the TM7 side of TM10 (18,19). In this study, we found that mutation of the arginine residue in TM7 (Arg-835 in SERCA2a/2b and Arg-836 in SERCA1a) caused effects in SERCA2b reminiscent of those of E917K, further pointing to a role of TM7 in the TM11 inhibition mechanism. The acceleration of Ca 2 E 1 P 3 E 2 P by the R835A mutation, in particular, is unique for SERCA2b, not being seen for R836A in SERCA1a (Table 3). In the SERCA1a crystal structures (both E 1 and E 2 states), this arginine seems to interact closely with the Glu-918 backbone carbonyl group. Hence, the interaction of Arg-835 with Glu-917 could be instrumental in mediating the slowing of Ca 2 E 1 P 3 E 2 P by TM11 in wildtype SERCA2b, which according to our data is relieved by mutation E917K or R835A, either of which is expected to disrupt the bond between them (in case of E917K, the drastic change of side chain would probably move the backbone). Hence, the arginine side chain may work as a brake on the Ca 2 E 1 P 3 E 2 P conformational transition in the presence of the TM11. Such a mechanism would require TM7 to be functionally linked with TM11. Irrespective of whether TM11 is on the TM7 side or on the TM8 side of TM10, it is likely that the luminal extension of TM11 interacts with the luminal loop between TM7 and TM8 (18,20), an interaction that might be disturbed by changes to TM7.
It is also very likely that the E917K mutation affects the position of TM11 through interaction of the lysine side chain with residue(s) in the nearby cytoplasmic loop connecting TM10 and TM11 ("L10 -11"). Hence, in the crystal structure of SERCA1a, the C-terminal glycine of SERCA1a would clearly be within reach of the side chain of a lysine substituent at the position of Glu-918 (cf. Fig. 7, "C terminus"). Previous results obtained with another DD mutation, S920Y (15), support such a mechanism. Of the more than 50 Darier disease mutations so far analyzed functionally (15,30,39,40), only S920Y caused in addition to reduced Ca 2ϩ affinity and expression level also the acceleration of the phosphoenzyme processing steps similarly to E917K (15). Ser-920 is located only three residues from Glu-917 in the same cytoplasmic loop between TM8 and TM9 ("L8 -9"). Like E917K, the S920Y mutation has a potential for disturbing L10 -11 and thereby the TM11 path due to the introduction of the bulky tyrosine.
One of our present findings with the Glu-917/918 mutants is not isoform-specific, namely the reduced affinity for Ca 2ϩ binding to E 1 , which was observed for all three isoforms. The data suggest that the reduced Ca 2ϩ affinity is due to a reduced rate of the E 1 3 Ca 2 E 1 transition ( Table 3). Because of its lack of isoform specificity, this effect may for all isoforms be rationalized on the basis of the SERCA1a structure and mutagenesis results. The SERCA1a crystal structures show Glu-918 surrounded by several arginine and lysine residues positioned with a potential for ionic interaction or hydrogen bonding with the Glu-918 side-chain carboxyl group or backbone carbonyl oxygen (Fig. 7). Thus, in addition to Arg-836 (TM7), also Lys-758 (TM5), Arg-762 (TM5), and Arg-819 (cytoplasmic loop between TM6 and TM7) could be involved in functionally important interactions with Glu-918. Previous mutagenesis of SERCA1a Lys-758 (41), Arg-762 (42), and Arg-819 (43) has also given rise to reduced Ca 2ϩ affinity and kinetics compatible with a defective E 1 3 Ca 2 E 1 transition. It is therefore quite likely that the cause of this defect is interference with the intricate bonding network centered around Glu-918. TM5 contains residues that are directly involved in Ca 2ϩ binding, and the conformational rearrangements of the Ca 2ϩ sites responsible for the cooperative binding of Ca 2ϩ during the E 1 3 Ca 2 E 1 transition include a straightening of the bent TM5 helix (4). Hence, the linking of Glu-918 to the two TM5 residues Lys-758 and Arg-762, which is disrupted by mutation of these residues, may be important in the mechanism of Ca 2ϩ binding. In addition, the function of the Ca 2ϩ -binding residues in TM6 (4) might be influenced by a change of the position of the loop between TM6 and TM7, and indirectly by the position of TM7, thus making the links of Arg-819 and Arg-836 to Glu-918 important for Ca 2ϩ binding.

Conclusions
The most likely reason that the SERCA2b E917K mutation reduces the endoplasmic reticulum Ca 2ϩ uptake, thereby causing DD, is a reduced expression level in keratinocytes, similar to that observed here for COS-1 cells. The unexpected SERCA2bspecific increase of the maximal turnover rate observed for the DD mutant is due to a relief of the kinetic constraints of specific partial reactions imposed by the TM11. This relief is caused by disturbance of the interaction of Glu-917 with Arg-835 of TM7 and likely also by an effect on TM11 mediated through the cytoplasmic loop L10 -11.

Experimental procedures
Mutations were introduced in cDNA encoding SERCA1a, SERCA2a, and SERCA2b (8,9) in the pMT2 expression vector (44) using the QuikChange site-directed mutagenesis kit (Agilent Technologies). The wildtypes and mutants were transiently expressed in COS-1 cells using the calcium phosphate precipitation method for transfection (45). Harvest of microsomal vesicles containing SERCA enzyme was carried out by differential centrifugation (46). The concentration of expressed active SERCA enzyme was determined by measuring the maximum phosphorylation capacity with either [␥-32 P]ATP or 32 P i under stoichiometric phosphorylation conditions (41). Ca 2ϩ -ATPase activity was determined under the conditions described in the figure legends by following the liberation of P i at 37°C using the Baginski method (47) over 10 min, which is within the linear time range (41). The rate of ATP-driven 45 Ca 2ϩ uptake in the isolated endoplasmic reticulum vesicles was determined by filtration, following incubation of the microsomal vesicles for 5 min at 37°C in medium containing 5 mM ATP, 20 mM MOPS/Tris (pH 6.8), 100 mM KCl, 5 mM MgCl 2 , 0.5 mM EGTA, 0.45 mM 45 CaCl 2 (corresponding to 5 M free Ca 2ϩ ), and 5 mM potassium oxalate (to precipitate and thereby trap Ca 2ϩ accumulated at high concentration inside the vesicles) (48).
Phosphorylation experiments using [␥-32 P]ATP or 32 P i to analyze the partial reactions of the Ca 2ϩ -ATPase reaction cycle were carried out as described previously (15,20,35,48), either

Darier disease mutation E917K of SERCA2b
by manual mixing at 0°C or at 25°C using a Bio-Logic quenchflow module QMF-5 (Bio-Logic Science Instruments, Claix, France). Following acid-quenching and acid SDS-PAGE, the radioactivity associated with the separated Ca 2ϩ -ATPase band was quantified by phosphorimaging using a Cyclone storage phosphor system (PerkinElmer Life Sciences). Further details of the experimental conditions are given in the figure legends.
All experiments were conducted at least twice on independent samples (see n values in tables), and average values are shown on the graphs, with error bars (seen only when larger than the size of the symbols) representing standard deviation (S.D.). Non-linear regression analysis was carried out using the SigmaPlot program (SPSS, Inc.), and extracted parameters with statistics are reported in the tables.
The non-monoexponential time courses of phosphorylation were analyzed using the program SimZyme to determine the rate constants by "trial and error" in kinetic simulation of the reaction cycle (35). The rate constants were varied, until an optimal fit to the data points was obtained. For any choice of reaction cycle and rate constants, the SimZyme program solves the relevant differential equations numerically and shows a graphical representation of the time dependence of the concentrations of the reaction intermediates. In this study, we compared the calculated time dependence of the concentration of phosphoenzyme with experimental data points corresponding to the phosphorylation time course.