Distinct Roles of the C-terminal 11th Transmembrane Helix and Luminal Extension in the Partial Reactions Determining the High Ca2+ Affinity of Sarco(endo)plasmic Reticulum Ca2+-ATPase Isoform 2b (SERCA2b)*

Background: Little is known about the mechanism by which the luminal extension (LE) and 11th transmembrane helix (TM11) of SERCA2b regulate function. Results: Mutant and chimeric SERCA constructs were analyzed kinetically. Conclusion: The LE affects Ca2+ interaction in E1 and E1P, whereas TM11 affects Ca2+-free E2/E2P states. Significance: Distinct roles of LE and TM11 in controlling the enzyme cycle are revealed. The molecular mechanism underlying the characteristic high apparent Ca2+ affinity of SERCA2b relative to SERCA1a and SERCA2a isoforms was studied. The C-terminal tail of SERCA2b consists of an 11th transmembrane helix (TM11) with an associated 11-amino acid luminal extension (LE). The effects of each of these parts and their interactions with the SERCA environment were examined by transient kinetic analysis of the partial reaction steps in the Ca2+ transport cycle in mutant and chimeric Ca2+-ATPase constructs. Manipulations to the LE of SERCA2b markedly increased the rate of Ca2+ dissociation from Ca2E1. Addition of the SERCA2b tail to SERCA1a slowed Ca2+ dissociation, but only when the luminal L7/8 loop of SERCA1 was simultaneously replaced with that of SERCA2, thus suggesting that the LE interacts with L7/8 in Ca2E1. The interaction of LE with L7/8 is also important for the low rate of the Ca2E1P → E2P conformational transition. These findings can be rationalized in terms of stabilization of the Ca2E1 and Ca2E1P forms by docking of the LE near L7/8. By contrast, low rates of E2P dephosphorylation and E2 → E1 transition in SERCA2b depend critically on TM11, particularly in a SERCA2 environment, but do not at all depend on the LE or L7/8. This indicates that interaction of TM11 with SERCA2-specific sequence element(s) elsewhere in the structure is critical in the Ca2+-free E2/E2P states. Collectively these properties ensure a higher Ca2+ affinity of SERCA2b relative to other SERCA isoforms, not only on the cytosolic side, but also on the luminal side.

Active removal of Ca 2ϩ from the cytosol is a critical aspect of signal transduction pathways in animal cells, facilitating the rapid cytosolic Ca 2ϩ oscillations required for a wide range of physiological functions such as muscle contraction, neuronal transmission, cellular motility, and cell growth. By coupling the hydrolysis of ATP with the transport of Ca 2ϩ ions across the endoplasmic reticulum membrane, the Ca 2ϩ -ATPase of sarco-(endo)plasmic reticulum (SERCA) 3 is an important player in the control of the intracellular Ca 2ϩ concentration.
The human genome contains three SERCA genes, ATP2A1, ATP2A2, and ATP2A3, encoding SERCA1, SERCA2, and SERCA3 proteins, respectively. Through tissue-specific and developmentally regulated alternative splicing, each of the three genes gives rise to 2-6 splice variants with differing carboxyl termini. The SERCAs are thought to accomplish Ca 2ϩ transport by a reaction cycle (Scheme 1) involving the sequential formation and decay of an aspartyl-phosphorylated intermediate, coupled to changes in the Ca 2ϩ binding sites, altering between states that are either directed toward the cytosol, tightly occluded inside the protein, or directed toward the endoplasmic reticulum lumen. A change in affinity, being high toward the cytosolic side and low toward the luminal side, facilitates the unidirectional translocation of Ca 2ϩ against a large Ca 2ϩ gradient. Crystal structures of the SERCA1a isoform have demonstrated that the cytoplasmic N ("nucleotide binding"), P ("phosphorylation"), and A ("actuator") domains undergo large rearrangements during the phosphoenzyme processing steps, which are transformed into both translational and rotational movements of transmembrane helices containing Ca 2ϩ binding amino acid residues (1,2).
The residues making up the Ca 2ϩ sites in the transmembrane region are conserved among the SERCA isoforms, being located in transmembrane segments M4, M5, M6, and M8. All of the SERCA isoforms, with the exception of the ubiquitously expressed SERCA2b, possess a total of 10 transmembrane segments. The SERCA2b enzyme contains a 49-amino acid C-terminal extension ("the SERCA2b tail," Fig. 1) that makes up an 11th transmembrane helix (TM11) continuing into an 11amino acid luminal extension (LE) (3)(4)(5)(6)(7). Analysis of the Ca 2ϩ dependence of Ca 2ϩ transport and ATPase activity at steady state has demonstrated that SERCA2b exhibits markedly higher apparent affinity for cytosolic Ca 2ϩ and lower maximal transport activity than any other SERCA isoform (4, 6, 8 -12). Both TM11 and the LE of SERCA2b have in steady-state measurements of the Ca 2ϩ dependence of ATPase activity been found critical to the higher apparent Ca 2ϩ affinity of SERCA2b. Hence, deletion of the LE of SERCA2b resulted in a significant lowering of the apparent affinity for cytosolic Ca 2ϩ (6,8), and supplementation of SERCA1a with a synthetic peptide corresponding to either the LE (6) or TM11 (7) increased the apparent Ca 2ϩ affinity. In the LE region, particularly the last four residues 1039 MFWS appeared important (6).
Little is, however, known about the mechanisms by which the various parts of the SERCA2b tail exert a regulatory effect. Dode et al. (10) showed that the SERCA2b tail as a whole is critical for the rates of several partial reaction steps in the enzyme cycle, including the dissociation of Ca 2ϩ toward the cytoplasmic side from Ca 2 E1, the phosphoenzyme processing reaction sequence Ca 2 E1P 3 E2P 3 E2, and the Ca 2ϩ binding transition, E2 3 E1 3 Ca 2 E1. The apparent affinity for Ca 2ϩ activation of the overall reaction depends not only on the "true" Ca 2ϩ affinity (i.e. the on and off rates of Ca 2ϩ at the E1 sites), but in addition rate constants of the Ca 2 E1P 3 E2P and E2 3 E1 conformational transitions that limit the overall rate of the cycle are critical determinants of the apparent Ca 2ϩ affinity, as has been demonstrated both experimentally and in a series of computer simulations of the cycle (13). This raises the question whether TM11 and the LE affect the same or distinct partial reactions of the enzyme cycle, and which ones? This question was not addressed in the previous studies focusing on the functional regions of the SERCA2b tail in studies of the overall reaction and its Ca 2ϩ dependence. Information about effects on the individual partial reactions by the various parts of the tail is crucial to the understanding of when in the cycle, and how, physical interaction takes place between the tail and the protein, and is thus a key point in the elucidation of the mechanism(s) underlying the regulatory effects exerted by the tail.
Hence, in the present study we set out to investigate these matters in transient kinetic studies of the differential effects of the TM11 and LE on individual partial reactions of the enzyme cycle. On the one hand, we studied mutants with alterations to either the TM11 or LE, which were previously reported to display a loss of SERCA2b tail function with respect to the apparent Ca 2ϩ affinity determined for the overall reaction (6). On the other hand, we followed a gain of function approach by characterizing the kinetics of SERCA1a-2b chimeras in which the SERCA2b tail is gradually reconstructed in the SERCA1a background. Using this combined approach we show that the LE is the most important determinant of the dissociation of Ca 2ϩ from the Ca 2 E1 form, whereas TM11 is of particular importance for the rates of the E2P 3 E2 and E2 3 E1 partial reaction steps. A critical interaction of TM11 with SERCA2 specific component(s) appears to be involved in the E2/E2P states.

EXPERIMENTAL PROCEDURES
The cDNA encoding the wild types or various SERCA constructs studied in the present work was either the same as in previous studies (6,10) or was newly prepared using the methods described there. For expression, COS-1 cells were transfected with cDNA inserted into expression vectors pMT2 (14), pcDNA3.1 (Invitrogen), or pcDNA6 (Invitrogen), using the calcium phosphate precipitation method (15). Microsomal vesicles containing the expressed SERCA construct were isolated by differential centrifugation (16). The concentration of expressed Ca 2ϩ -ATPase was determined by measuring the maximum capacity for phosphorylation with [␥-32 P]ATP or 32 P i ("active site concentration") (17).
Measurements of phosphorylation and dephosphorylation were generally carried out by manual mixing at 0°C (17,18). Transient state kinetics at 25°C was analyzed using the Bio-Logic quench-flow module QFM-5 (Bio-Logic Science Instruments, Claix, France) with mixing protocols as previously described (19,20). The determination of the phosphorylation level by acid quenching followed by acid SDS-polyacrylamide gel electrophoresis and quantification of the radioactivity associated with the Ca 2ϩ -ATPase band was carried out using previously established procedures (17,18,20).
To determine a time course of phosphorylation or dephosphorylation, at least two independent series of kinetic experiments were carried out (see n values in Tables 1-3), with the earliest time points in duplicate in each series. Here independent means: carried out on different days, using enzyme preparations from different transfections. The complete set of data points collected (all shown in the graphs) was analyzed by nonlinear regression using the SigmaPlot program (SPSS, Inc.), and the extracted parameters with the S.E. values from the regression are reported in the tables. Monoexponential functions were fitted to the time courses of dephosphorylation, Ca 2ϩ binding, and Ca 2ϩ dissociation. The analysis of the Ca 2ϩ -concentration dependence of phosphorylation was performed using the modified Hill equations given in the corresponding legends.

Design and Expression of the Chimeric and Mutant
Ca 2ϩ -ATPases-Nine wild type or modified Ca 2ϩ -ATPase constructs were included in the present study ( Fig. 1 and Table  1), designed with the purpose of elucidating the distinct influences of TM11 and LE on the partial reaction steps of the pump cycle (Scheme 1). The proteins were expressed in COS-1 cells to similar high levels, thus facilitating the study of the transient kinetics of the individual reactions. The results presented in the figures are summarized in Tables 2 and 3.
The wild types studied were the rabbit SERCA1a ("S1a"), human SERCA2a ("S2a"), and human SERCA2b ("S2b"). A group of loss of SERCA2b tail function mutants with alteration to the LE or TM11 of the SERCA2b tail were analyzed. "S2b-T1032*" is SERCA2b truncated just prior to Thr 1032 by substituting the codon corresponding to Thr 1032 with a STOP codon indicated by an asterisk (*), thereby deleting the entire 11-a-mino acid LE of the SERCA2b tail ( Fig. 1 and Table 1). In "S2b-MFWS-AAAA," the last four amino acid residues of the SERCA2b LE ( 1039 MFWS) are replaced by alanines.
A set of SERCA1a-2b chimeras were also studied, in which either the complete 49-amino acid C-terminal tail of SERCA2b was attached to SERCA1a ("S1a2b"), or only the part of the tail consisting of TM11 without the LE was attached to SERCA1a ("S1a2b-T1032*," i.e. S1a2b truncated as described for S2b-T1032*). The effect of simultaneously replacing the luminal loop connecting transmembrane helices TM7 and TM8 (L7/8) with the corresponding loop of SERCA2a/SERCA2b was also examined ("S1a2b-L7/8" with "S1a-L7/8" as reference), because the large L7/8 loop contains several amino acid differences between SERCA1 and SERCA2 isoforms and is likely to be in contact with the LE (Fig. 1B).
Rate of Ca 2ϩ Dissociation Toward the Cytosolic Side-Ca 2ϩ dissociation from Ca 2 E1, i.e. toward the cytosolic side, was In A, each circle corresponds to an amino acid residue indicated by the single-letter code inside the circle. The phosphorylated aspartate (Asp 351 ) is indicated by white lettering on black background with a large encircled P attached. TM1-TM11 denotes the transmembrane helices, of which TM11 is found only in SERCA2b. In the transmembrane helices, circles with white lettering on black background indicate the calcium binding residues. The unique C-terminal tail of SERCA2b is colored red, whereas the corresponding part of SERCA2a is shown in blue. SERCA2a and SERCA2b are identical in the first 993 amino acid residues. SERCA1a (994 residues, not shown) has a topology similar to SERCA2a with 84% amino acid identity. The SERCA2b-specific tail comprises in addition to TM11 also an 11-amino acid luminal extension (residues 1032-1042). The cytoplasmic domains A, N, and P, are indicated using different color codes, as are the luminal loops L1/2, L3/4, L5/6, L7/8, and L9/10. B, the C-terminal tail of SERCA2b modeled on a SERCA2a homology model derived from the crystal structure of SERCA1a in Ca 2 E1 form (6). The modeled TM11 is in proximity to TM7 and TM10. As it descends from TM11 the luminal extension is in close contact with the luminal loop L7/8. The last four residues, MFWS, occupy a luminal docking site formed by luminal loops L1/2, L3/4, and L7/8 (6). The luminal loops are indicated by the same color code as in A. ARNYLEPGKECVQPATKSCSFSACTDG ISWPFVLLIMPLVIWVYSTDTNFSDAAAA S1a2b S1a with the C terminus of S2b 988 ARNYLEPGKECVQPATKSCSFSACTDG ISWPFVLLIMPLVIWVYSTDTNFSDMFWS S1a2b-T1032* S1a with the C terminus of S2b, terminated before Thr 1032 a 988 ARNYLEPGKECVQPATKSCSFSACTDG ISWPFVLLIMPLVIWVYS S1a2b-L7/8 S1a with the C terminus and the L7/8 luminal loop of S2b Same as S1a2b S1a-L7/8 S1a with the L7/8 luminal loop of S2b Same as S1a a Note that "1032" here refers to S2a/S2b numbering. Due to the presence of an extra amino acid in the nucleotide binding domain of S1a relative to S2a/S2b the S1a2b-T1032* chimera is actually one residue longer than S2b-T1032*. SCHEME 1. Ca 2؉ -ATPase reaction cycle. Major conformational changes and substrate binding and dissociation steps are shown. examined in rapid kinetic measurements taking advantage of the requirement for occupancy of the two Ca 2ϩ sites to allow phosphorylation from ATP. The mixing protocol is illustrated at the top of Fig. 2 (for a more detailed description of this method see Ref. 20). Upon addition of the Ca 2ϩ chelator EGTA the ability to become phosphorylated by [␥-32 P]ATP disappears at a rate corresponding to the rate of Ca 2ϩ dissociation (actually the rate corresponding to the Ca 2ϩ ion dissociating first in the sequential mechanism, because both Ca 2ϩ sites must be filled for phosphorylation to occur). The Ca 2ϩ dissociation was markedly slower for SERCA2b than for SERCA1a (3.2-fold) or SERCA2a (5.5-fold). Much of this effect seems attributable to the LE of the tail, as the truncation of SERCA2b at Thr 1032 led to a marked 3.4-fold increase of the rate of Ca 2ϩ dissociation ( Fig. 2 and Table 2, S2b-T1032*). A conspicuous increase in the Ca 2ϩ dissociation rate (2.6-fold) was also obtained by the less extensive modification in which only the last four amino acid residues ( 1039 MFWS) of the SERCA2b LE were replaced by alanines ( Fig. 2 and Table 2, S2b-MFWS-AAAA), implying that these last four amino acids are critical for Ca 2ϩ dissociation. TM11 may also be of some minor importance in determining the slow Ca 2ϩ dissociation of SERCA2b, because the removal of TM11 resulted in a 1.6-fold increase of the Ca 2ϩ dissociation rate (compare SERCA2a with S2b-T1032*). Similar conclusions can be drawn from the gain of function approach. Hence, although addition of the SERCA2b tail to SERCA1a, with or without the LE (S1a2b and S1a2b-T1032*, respectively) had only a slight effect on the rate of Ca 2ϩ dissociation (1.2-fold reduction), suggesting a modest role for TM11, combining the addition of the SERCA2b tail to SERCA1a with substituting the L7/8 loop of SERCA1a with that of SERCA2b (S1a2b-L7/8) led to a marked 2.6-fold reduction of the rate of Ca 2ϩ dissociation (Fig. 2). This yields a rate not far from that of wild type SERCA2b (3.2-fold lower than that of wild type SERCA1a), indicating that the L7/8 loop of SERCA2b is required for the tail, in particular LE, to fully exert its effect on Ca 2ϩ dissociation. A control experiment in which the L7/8 loop of SERCA1a was replaced with that of SERCA2b in the absence of the SERCA2b tail (S1a-L7/8), to reveal any effect of the L7/8 substitution per se, showed a less pronounced reduction of the Ca 2ϩ dissociation rate by only 1.8-fold ( Fig. 2 and Table 2).    These findings are supportive of a model where the LE docks near L7/8 (Fig. 1B).
Rate of Ca 2 E1P 3 E2P-To study this conformational transition, the enzyme was first phosphorylated by [␥-32 P]ATP, and the phosphoenzyme was then chased at 0°C by addition of excess EGTA (to terminate phosphorylation by removal of Ca 2ϩ ) with or without 1 mM ADP (Fig. 3). As shown by the closed symbols in Fig. 3, the accumulated phosphoenzyme disappeared rapidly in the presence of ADP for both the wild type pumps and the mutants, indicating that the major phosphoenzyme form accumulated is ADP-sensitive Ca 2 E1P, i.e. for all constructs the transition of this form to the ADP-insensitive E2P phosphoenzyme intermediate is the rate-limiting step in the Ca 2 E1P 3 E2P 3 E2 reaction sequence under the specific conditions applied here. Hence, the rate of dephosphorylation observed when phosphorylation was terminated by adding EGTA in the absence of ADP (open symbols in Fig. 3) reflects the Ca 2 E1P 3 E2P reaction step.
The data in Fig. 3 show that SERCA2b undergoes the Ca 2 E1P 3 E2P reaction step ϳ2-fold slower than SERCA1a and SERCA2a (see also Table 2), a characteristic that contributes to determine the high apparent Ca 2ϩ affinity of SERCA2b relative to SERCA1a and SERCA2a, because less Ca 2ϩ is required for a certain amount of phosphoenzyme to accumulate, when Ca 2 E1P 3 E2P is slow (10,13). Truncation of the SERCA2b LE increased the Ca 2 E1P 3 E2P rate 1.4-fold ( Fig. 3 and Table 2, S2b-T1032*), indicating a role of the LE in determining the low Ca 2 E1P 3 E2P rate in wild type SERCA2b. Adding the  Table 2. The broken lines represent the curves corresponding to wild type SERCA1a and SERCA2b from the upper left panel.  Table 2. The closed symbols represent similar experiments in which 1 mM ADP was added together with the EGTA chase medium. In each case, the 100% value corresponds to the steady-state level of phosphoenzyme present just prior to the initiation of dephosphorylation. The broken lines represent the curves corresponding to wild type SERCA1a and SERCA2b from the upper left panel.
SERCA2b C-terminal tail to SERCA1a, with (S1a2b) or even without (S1a2b-T1032*) the LE, reduced the rate of Ca 2 E1P 3 E2P to a level close to that of SERCA2b, thus indicating that TM11 is important as well. Simultaneous addition of the entire SERCA2b C-terminal tail to SERCA1a and substitution of the L7/8 loop (S1a2b-L7/8) further strengthened the effect, whereas substitution of the L7/8 loop of SERCA1a with that of SERCA2b in the absence of the SERCA2b tail (S1a-L7/8) had little effect on the rate of Ca 2 E1P 3 E2P. Hence, both the LE and the TM11 contribute to the low rate of the Ca 2 E1P 3 E2P transition in wild type SERCA2b, and the effect depends on the L7/8 loop, likely because it interacts with the LE. However, the last four residues of the LE are not much involved in this effect, because their replacement by alanines (S2b-MFWS-AAAA) had almost no impact, if any, on the rate of Ca 2 E1P 3 E2P.
Rate of E2P 3 E2 Dephosphorylation-The dephosphorylation of the E2P phosphoenzyme was examined by first phosphorylating the Ca 2ϩ -deprived enzyme by 32 P i in the backward direction of the normal reaction cycle to accumulate E2P (cf. Scheme 1), and then chasing the E2P phosphoenzyme by dilution into a dephosphorylation buffer containing nonradioactive P i at pH 7 and only little (10 mM) K ϩ to ensure a relatively slow dephosphorylation thereby allowing a very accurate measurement, cf. Ref. 21).
As seen in Fig. 4 and Table 2, E2P dephosphorylation was markedly slower for SERCA2b than for SERCA1a (3-fold) and SERCA2a (1.5-fold). Neither substitution of the last four residues of SERCA2b by alanines (S2b-MFWS-AAAA) nor truncation of the entire SERCA2b LE (S2b-T1032*) had any impact on E2P dephosphorylation. A ϳ2-fold slowing of E2P dephosphorylation was seen upon adding the C-terminal tail to SERCA1a, but this effect was not dependent on simultaneous substitution of the L7/8 loop (compare S1a2b with S1a2b-L7/8 in Fig. 4 and Table 2). These findings indicate that TM11, but not the LE, is critical for the slower E2P dephosphorylation in SERCA2b. It is also noteworthy that the efficiency of TM11 in slowing the dephosphorylation is less when added to SERCA1a as compared with the SERCA2 environment (compare S1a2b with S2b and S1a2b-T1032* with S2b-T1032* in Fig. 4 and Table 2). This suggests that direct or indirect interaction of the TM11 with SERCA2-specific sequence element(s) elsewhere in the protein is involved.
Rate of E2 3 E1-Another possible determinant of the apparent affinity for Ca 2ϩ activation of the overall reaction is the rate of the E2 3 E1 transition of the dephosphoenzyme, i.e. the step immediately preceding Ca 2ϩ binding at the cytoplasmically facing high affinity Ca 2ϩ sites. Fig. 5 shows the kinetics of the E2 3 E1 3 Ca 2 E1 reaction sequence under conditions where the E2 3 E1 transition is rate-limiting because of the low pH (pH 6) (22). It is apparent from Fig. 5 and Table 2 that the SERCA2b tail slows the rate of E2 3 E1, SERCA2b displaying 2.3-fold lower rate than SERCA2a and a 3.8-fold lower rate than SERCA1a. Because the rate of the E2 3 E1 transition influences the amount of E1 available for Ca 2ϩ binding at steady-state, a slow E2 3 E1 transition should in principle tend to reduce the apparent Ca 2ϩ affinity observed in steady-state measurements of the overall ATPase reaction, however, because the slow phosphoenzyme processing as well as the slow Ca 2ϩ dissocia-tion exert the opposite effect, tending to increase the apparent Ca 2ϩ affinity, the net result is that a high apparent Ca 2ϩ affinity prevails in SERCA2b relative to the other isoforms.
Truncation of the SERCA2b LE (S2b-T1032*) did not increase the E2 3 E1 rate, indicating that TM11 is responsible for the low rate in SERCA2b. The addition of the SERCA2b tail to SERCA1a without or with concomitant substitution of the L7/8 loop with that of SERCA2b (S1a2b and S1a2b-L7/8, respectively) did not reduce the rate of E2 3 E1 relative to that of the SERCA1a wild type. Hence, SERCA2-specific element(s) elsewhere in the protein must be involved in causing the slow E2 3 E1 imposed by TM11, in line with the findings in relationship to E2P dephosphorylation. However, because the mere substitution of the L7/8 loop in SERCA1a with that of SERCA2 (S1a-L7/8) resulted in a rate of E2 3 E1 somewhat slower than that of SERCA1a and similar to that of SERCA2a, a SERCA2specific element of the L7/8 loop might also in some way contribute to the slowing of E2 3 E1. Kinetics of E2P dephosphorylation. The microsomal enzyme was phosphorylated for 10 min at 25°C in 100 mM MES/Tris (pH 6.0), 10 mM MgCl 2 , 2 mM EGTA, 30% (v/v) dimethyl sulfoxide (the organic solvent ensuring maximal amount of phosphoenzyme), and 0.5 mM 32 P i . Dephosphorylation was then studied at 0°C by a 19-fold dilution of pre-chilled phosphorylated microsomes into ice-cold medium containing 40 mM MOPS/Tris (pH 7.0), 10 mM KCl, 2 mM MgCl 2 , 2 mM EGTA, and 0.5 mM nonradioactive P i , followed by acid quenching at the indicated time intervals. All data points are shown. The lines show the best nonlinear regression fits of a monoexponential decay function, giving the rate constants listed in Table 2. In each case, the 100% value corresponds to the level of phosphoenzyme present just prior to the initiation of dephosphorylation. The broken lines represent the curves corresponding to wild type SERCA1a and SERCA2b from the upper left panel.

Comparison of Apparent Ca 2ϩ Affinities at Cytoplasmic and
Luminal Sites-The Ca 2ϩ concentration dependence of steadystate phosphorylation of Ca 2 E1 from [␥-32 P]ATP was studied to evaluate the combined effects of the changes to the partial reaction steps on apparent Ca 2ϩ affinity of the cytoplasmically facing sites (Table 3, first column). A higher apparent Ca 2ϩ affinity is seen for SERCA2b, relative to SERCA1a and SERCA2a. The apparent Ca 2ϩ affinity of SERCA2b was lowered by manipulations to the tail and increased by addition of the SERCA2b tail to SERCA1a, in particular in combination with the L7/8 loop substitution. Both the LE and TM11 appear to contribute. These results are in good agreement with our findings for the rate constants of the individual partial reactions and with previous analysis of the overall reaction by measurement of the Ca 2ϩ dependence of steady-state ATPase activity (6).
The effect of the entire SERCA2b tail and its separate LE and TM11 parts on the apparent Ca 2ϩ affinity of the luminally facing low-affinity Ca 2ϩ sites has not previously been determined. To this end, we studied the dependence of phosphorylation of E2 from inorganic phosphate on luminal Ca 2ϩ (cf. Scheme 1 and Refs. [23][24][25]. Microsomal vesicles containing wild type or mutant Ca 2ϩ pumps were loaded passively with various Ca 2ϩ concentrations overnight and then diluted into a medium containing 32 P i and EGTA to observe the phosphorylation from 32 P i occurring in the absence of free Ca 2ϩ in the medium outside the vesicles.
The data obtained are shown in Fig. 6 and Table 3, second column. Increasing the luminal Ca 2ϩ concentration stimulated the phosphorylation from 32 P i with apparent Ca 2ϩ affinities of the SERCA1a, SERCA2a, and SERCA2b wild type pumps corresponding to K 0.5 values of 7.3, 8.4, and 4.8 mM Ca 2ϩ , respectively, thus indicating that the apparent Ca 2ϩ affinity is highest for SERCA2b, not only on the cytoplasmic side, but on the luminal side as well.
The manipulations to the SERCA2b LE (S2b-T1032* and S2b-MFWS-AAAA) slightly increased the K 0.5 for the luminal Ca 2ϩ activation of phosphorylation from 32 P i relative to the SERCA2b wild type. Adding the SERCA2b tail to SERCA1a with (S1a2b) or without (S1a2b-T1032*) the LE reduced the K 0.5 for luminal Ca 2ϩ to a value similar to that of SERCA2b. Replacement of the L7/8 loop of SERCA1a with that of SERCA2 (S1a-L7/8) reduced the K 0.5 for Ca 2ϩ , however, the simultaneous substitution of the L7/8 loop and addition of the SERCA2b tail to SERCA1a (S1a2b-L7/8) resulted in the lowest K 0.5 for Ca 2ϩ of all constructs. Collectively, these results suggest that TM11 is the main determinant of the higher apparent luminal Ca 2ϩ affinity of SERCA2b relative to the other isoforms, but the LE also contributes to some extent via its interaction with the L7/8 loop. Despite the more than 1000-fold difference in the actual affinities for cytoplasmic and luminal Ca 2ϩ , there is a remarkable correlation between the apparent affinities for Ca 2ϩ on the two sides of the membrane, as illustrated in Fig. 7 by the plot of the K 0.5 values observed for cytoplasmic Ca 2ϩ against the K 0.5 values observed for luminal Ca 2ϩ .

DISCUSSION
The SERCA2b tail changes the kinetics of the Ca 2ϩ pump at various partial reaction steps, inducing (i) a slower Ca 2ϩ dissociation from the high-affinity sites in Ca 2 E1 toward the cytosol, (ii) a slower Ca 2 E1P 3 E2P transition, (iii) slower E2P dephosphorylation, and (iv) a reduced rate of E2 3 E1. The present kinetic analysis of the partial reaction steps of mutants and chimeric constructs has demonstrated that TM11 and the LE differentially affect these steps that determine the high apparent Ca 2ϩ affinity and low maximal turnover rate of SERCA2b. The major conclusions are indicated below the rate constants in Table 2. The lines represent the best nonlinear regression fits of a monoexponential "rise to max" function with an initial offset, giving the rate constants listed in Table  2. In each case, the phosphorylation level at infinite time extracted from the fit was taken as 100%. The broken lines represent the curves corresponding to wild type SERCA1a and SERCA2b from the upper left panel.
Role of the LE-The LE (particularly the last four residues, MFWS) was shown here to be critical to the slow Ca 2ϩ dissociation from Ca 2 E1 toward the cytosolic side in SERCA2b. The gain of SERCA2b tail function with respect to the slow Ca 2ϩ dissociation from Ca 2 E1 observed upon addition of the SERCA2b tail to SERCA1a depends on simultaneous substitution of the L7/8 loop with that of SERCA2, thus suggesting that the LE interacts with L7/8 to stabilize the Ca 2ϩ bound Ca 2 E1 state. Such a scenario appears realistic from a structural point of view, if the LE docks between the luminal L3/4 and L7/8 loops at a site available only in the Ca 2 E1 state and not in E2 as proposed by the model in Fig. 1B (6), because M4, which is connected to L3/4, plays a central role in the function of the Ca 2ϩ sites (26,27). The structural change underlying the stabilizing effect of the LE on Ca 2 E1 might be related to the change in the proteinase K cleavage pattern observed upon addition of synthetic LE peptide to SERCA2a, showing the appearance of a 90-kDa cleavage product (6).
The LE and L7/8 are also important for the low rate of the Ca 2ϩ translocating conformational change Ca 2 E1P 3 E2P in SERCA2b, which likewise may owe to a stabilization of a Ca 2ϩ bound form, in this case the Ca 2 E1P phosphoenzyme, which has Ca 2ϩ in the occluded state. However, it is noteworthy that the data (cf. Table 2) shows that the interaction of the LE in Ca 2 E1P involves the four C-terminal MFWS residues to a much lesser extent than in Ca 2 E1, thus indicating that the occlusion of the Ca 2ϩ ions affects the docking of the C terminus.
Role of TM11-Although the TM11 seems to affect all the partial reaction steps studied to some extent, TM11 is most critical to the E2P dephosphorylation step, E2P 3 E2, and the conformational transition of the dephosphoenzyme, E2 3 E1. The latter two steps are not at all dependent on the LE, and in contrast to the Ca 2 E1P 3 E2P transformation, the E2P 3 E2 and E2 3 E1 steps are highly dependent on the presence of TM11 in a SERCA2 environment. This applies with or without the LE present and irrespective of simultaneous exchange of L7/8, thus indicating that interaction of the TM11 with a SERCA2-specific sequence element(s) elsewhere in the structure is critical in the E2/E2P states.
Exactly how TM11 exerts the functional effects and which interaction partners are involved remain unclear in the absence of rigorous structural information about the relationship of the SERCA2b tail. Modeling based on the SERCA1a crystal structure (Fig. 1B) suggests that in addition to TM10, TM7 also might be an interaction partner of TM11 in SERCA2b, and both of these transmembrane helices contain residues that are not conserved between SERCA2 and SERCA1 isoforms (7). Moreover, SERCA2-specific sequence element(s) in cytoplasmic segments might play an indirect role by influencing the positioning of transmembrane segment(s) that interact with TM11. It is possible that in SERCA2b, TM11, like the ␤-subunit of Na ϩ ,K ϩ -ATPase, impinges on TM7 and causes this helix to kink (7). The observation of such interaction in Na ϩ ,K ϩ - The phosphorylation was then determined at 25°C by a 50-fold dilution of the Ca 2ϩ -loaded vesicles into a phosphorylation medium containing 150 mM MOPS/Tris (pH 7.0), 125 mM sucrose, 77 mM KCl, 10 mM EGTA (removing Ca 2ϩ from the medium outside the vesicles), and 0.5 mM 32 P i , followed by acid quenching 1 min later. The ordinate shows the results as percentage of the maximal phosphorylation level obtained by phosphorylation from 32 P i under optimal conditions for E2P formation in the presence of dimethyl sulfoxide, as described in the legend to Fig. 4. All data points are shown. The lines show the best nonlinear regression fits of a modified Hill equation EP ϭ EP min ϩ (EP ϱ Ϫ EP min ) ⅐ [Ca 2ϩ ] n /(K 0.5 n ϩ [Ca 2ϩ ] n ), with EP min accounting for the basic phosphorylation level in the absence of a Ca 2ϩ gradient and EP ϱ representing the extrapolated phosphorylation level at infinite intravesicular Ca 2ϩ . The K 0.5 values obtained are listed in Table 3. The broken lines represent the curves corresponding to wild type SERCA1a and SERCA2b from the upper left panel. FIGURE 7. Correlation between the apparent Ca 2؉ affinities of cytoplasmically and luminally facing sites. The K 0.5 values for Ca 2ϩ activation of phosphorylation from the cytoplasmic and luminal side (data from Table 3) are represented by the abscissa and ordinate, respectively, on a logarithmic scale. The labeling is as follows: a, S1a; b, S2a; c, S2b; d, S2b-T1032*; e, S2b-MFWS-AAAA; f, S1a2b; g, S1a2b-T1032*; h, S1a2b-L7/8; i, S1a-L7/8. ATPase relates to the K ϩ -bound E2 conformation, and a role in determining K ϩ selectivity has been suggested (28). Along the same line, interaction between TM11 and TM7 specific for E2/E2P states (i.e. the states corresponding to those binding K ϩ in Na ϩ ,K ϩ -ATPase) might be part of a mechanism slowing the E2P 3 E2 and E2 3 E1 reactions in SERCA2b.
Conclusions-The present data indicate that docking of the LE near L7/8 preferentially stabilizes Ca 2 E1 and Ca 2 E1P, thus slowing Ca 2ϩ dissociation from SERCA2b and thereby increasing the "true" (intrinsic) Ca 2ϩ affinity, whereas functional interaction of the TM11 with a SERCA2-specific sequence element(s) elsewhere in the structure occurs particularly in the Ca 2ϩ -free E2/E2P states, slowing E2P 3 E2 and E2 3 E1 transitions and thereby reducing the maximal turnover rate of SERCA2b and increasing the apparent Ca 2ϩ affinity by an indirect mechanism quite different from that of the LE.
The stabilization of Ca 2 E1P as well as the slow E2P 3 E2 and E2 3 E1 transitions favor the phosphorylation of E2 by P i promoted by luminal Ca 2ϩ and the subsequent accumulation of Ca 2 E1P formed backwards from E2P. Collectively these effects therefore also explain that not only the apparent affinity for Ca 2ϩ binding from the cytoplasmic side is enhanced in SERCA2b relative to the other isoforms, but in addition the apparent affinity for Ca 2ϩ binding from the luminal side is affected in a similar manner. This previously unrecognized property of SERCA2b likely limits the capacity for Ca 2ϩ storage by SERCA2b, because the luminal Ca 2ϩ concentration at which "back inhibition" of the pump occurs is lower for SERCA2b than for other SERCA isoforms. Together with the lower maximal pump activity the limited storage capacity of SERCA2b may constitute an important reason why the housekeeping SERCA2b is replaced by/supplemented with SERCA1a, SERCA2a, or SERCA3 in cells that cope with large amounts of Ca 2ϩ .