INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Sarcoplasmic reticulum (SR) vesicles (Ebashi and Lippman, 1962; Hasselbach and Makinose, 1961) have provided an advantageous system for characterization of the catalytic and transport cycle (Hasselbach, 1964; Tada et al., 1978; de Meis and Vianna, 1979) and of the protein structure (MacLennan et al., 1985; Toyoshima et al., 1993, 2000) of the Ca²⁺ ATPase. The cycle begins with enzyme activation by cooperative binding of two Ca²⁺ (Inesi et al., 1980), followed by phosphoryl transfer from ATP to a catalytic site residue (Asp351), vectorial translocation of the bound Ca²⁺, and hydrolytic cleavage of the phosphorylated enzyme intermediate. The mechanism whereby Ca²⁺ binding produces catalytic activation and enzyme phosphorylation causes vectorial translocation of Ca²⁺ is a fundamental question of general interest, pertinent to the role of proteins in energy transduction through biochemical reactions. It was originally proposed that phosphorylation and Ca²⁺ translocation may be directly coupled through a symport mechanism (Hasselbach, 1964; Mitchell and Koppenol, 1982), based on spectroscopic evidence of cation binding within the catalytic site (Grisham and Mildvan, 1974). However, it was later found that single mutations of residues within the membrane-bound region of recombinant ATPase (i.e., E309, E771, N796, T799, and D800) produce enzyme inactivation (Clarke et al., 1989). This effect was attributed to interference with Ca²⁺ binding, suggesting that the Ca²⁺-binding domain resides within the membrane-bound region, as later established by structural analysis (Toyoshima et al., 2000). The location of the Ca²⁺ sites is ~50 Å away from the catalytic site in the cytosolic region, and therefore a long-range intramolecular linkage is required for catalytic activation (Bigelow and Inesi, 1992; Inesi et al., 1992). We have now characterized these mutants, with the aim of defining the mechanism of binding and demonstrating unambiguously whether occupancy of the first and/or the second Ca²⁺ site is required not only for catalytic activation in the forward direction of the cycle but also for synthesis of ATP in the reverse direction.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Materials

Recombinant ATPase protein was obtained in the microsomal fraction of COS-1 cells infected with adenovirus vectors carrying wild-type (WT) or mutated cDNA encoding the chicken fast muscle Ca²⁺ ATPase (SERCA1). The methods used for construction of vectors, cultures, and preparation of microsomes were previously described in detail (Zhang et al., 2000). The total microsomal protein was determined using bicinchoninic acid with the biuret reaction (Pierce, Rockford, IL). In all experiments, the total protein concentration was adjusted to yield comparable SERCA concentration as determined by Western blotting.

Ca²⁺ binding in the absence of ATP

Microsomal samples were suspended in 2.0 ml of a medium containing 20 mM 3-(N-morpholino)propane sulfonic acid (MOPS), pH 7.0, 80 mM KCl, 5mM MgCl₂, and variable EGTA, to yield 0.8 mg of protein/ml, and 8 µl of 2 mM thapsigargin (TG) in dimethyl sulfoxide (Me₂SO), or 8 µl of Me₂SO, was also included. After a 10-min incubation in ice, an equal volume of a medium containing 20 mM MOPS, pH 7.0, 80 mM KCl, 5 mM MgCl₂, 80 µM ⁴⁵CaCl₂ was added. The total calcium concentration, after mixing, was 52.5 µM, including added and contaminant (12.5 µM) calcium. After a 10-min incubation, the suspension was sonicated three times for 10 s each, and 0.750-ml samples (corresponding to 0.3 mg of protein) were vacuum filtered (0.45 µm; Millipore, Bedford, MA). The filters were then blotted, and the radioactivity was determined by scintillation counting. The free Ca²⁺ concentration was calculated from total calcium and EGTA concentration, according to Fabiato and Fabiato (1978). The difference between binding obtained in the absence and in the presence of TG was considered to be specific binding. The TG-independent binding varied from 95% of total binding at pCa 7.0 to 40% at pCa 5.0.

Enzyme phosphorylation with Pi

Enzyme phosphorylation with Pi was obtained by incubating 60 µg of microsomal protein in 1 ml of a medium containing 50 mM 2-[N-morpholino]ethane sulfonic acid (MES), pH 6.2, 10 mM MgCl₂, 100 µM ³²P_i, 500 µM EGTA, 20% (v/v) Me₂SO, and various concentrations of Ca²⁺ as required by the experimental schedule. After 10 min of incubation at 25°C, the reaction was quenched by the addition of 0.5 ml of 3 M perchloric acid (PCA), and 100 µg of carrier microsomal protein previously quenched in 0.5 ml of 1 M PCA was then added. The quenched samples were cooled in ice, sedimented by low-speed centrifugation, and resuspended in 1 ml of cold 0.125 M PCA. This washing procedure was repeated three times, resuspending twice in PCA and once with H₂O. The final sediment was dissolved in 0.2 ml of a medium containing 5% lithium dodecyl sulfate in 50 mM phosphate buffer, pH 6.3. The residual protein concentration was measured, and 60 µg per sample was placed on 6.5% acrylamide gels and subjected to electrophoresis by the method of Weber and Osborn (1969). The gels were then dried and the radioactivity determined by phosphoimaging.

ATP synthesis by reversal of the catalytic cycle

Microsomal samples were added to a medium containing 5 mM MES, pH 6.2, 10 mM MgCl₂, 0.1 mM ³²P_i, 0.5 mM EGTA, 20% (v/v) Me₂SO, and 2 µM A23187 (Ca²⁺ ionophore), to yield 50-100 µg of protein/ml. After a 5-min incubation at 25°C, 0.2-ml samples were rapidly mixed with 1.5 ml of ice-cold 50 mM HEPES, pH 8.0, 10 mM MgCl₂, 1 mM CaCl₂, and 0.1 mM ADP. Such samples were quenched at serial times by the addition of 1.0 ml of ice-cold 3 M PCA. After a brief centrifugation, the supernatant was collected for extraction of ³²P_i by addition of 1.0 ml of acetone, 0.5 ml of 5% ammonium molybdate in 2.5 N H₂SO₄ and 10 µl of 100 mM P_i (carrier). When the solution was not completely clear, more acetone was added to obtain a clear yellowish solution. The phosphomolybdate complex was extracted by vortexing with 2 ml of isobutanol-benzene (1:1) and discarding the upper phase. Two additional extractions were carried out with 1 ml of acetone, 10 µl of 100 mM P_i, and 2 ml of isobutanol-benzene. A fourth extraction was performed to eliminate any remnant ammonium molybdate, with 1 ml of acetone, 0.2 ml of 100 mM P_i, and 2.0 ml of isobutanol-benzene. A final extraction was performed with 1.0 ml of acetone and 2.0 ml of isobutanol-benzene, and an aliquot of the remaining solution was then processed for determination of radioactivity. The correspondence of residual radioactivity to newly synthesized ATP was independently demonstrated by chromatography.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The functional consequences of E771, T799, D800, E309, or N796 mutations, i.e., interference with Ca²⁺ activation of ATP use and Ca²⁺ inhibition of enzyme phosphorylation by P_i, suggest mutational interference with Ca²⁺ binding (Clarke et al., 1989). Direct measurements of Ca²⁺ binding to recombinant enzyme, however, are quite difficult. Nevertheless, preliminary measurements of binding at a single Ca²⁺ concentration suggested that E771Q, T799A, and D800N mutations interfere with binding of both Ca²⁺ required for enzyme activation, whereas mutations of E309 or N796 allow binding of only one Ca²⁺ (Skerjanc et al., 1993; Zhang et al., 2000). We have now obtained direct measurements of Ca²⁺ binding with recombinant ATPase at various Ca²⁺ concentrations. Fitting the WT binding data requires a cooperative binding equation with two interdependent constants (7e+5 and 2e+6 M-1) for two sites exhibiting positive cooperativity, whereas the mutant data can be fitted with an independent binding equation and a single constant (2e+6 M-1). The resulting equilibrium binding isotherms (Fig. 1 A) demonstrate that, as opposed to the positive cooperativity of the WT enzyme, the E309Q, N796A, or E309Q/N796A mutants sustain noncooperative binding with a maximal stoichiometric ratio of one Ca²⁺ per ATPase within the pCa 7.0-5.0 concentration range. It is likely that additional, noncooperative binding occurs at higher concentrations that preclude reliable measurements. However, such an additional binding has no specific functional consequences, because no ATP hydrolysis or ATP synthesis is obtained by addition of Ca²⁺ concentrations as high as millimolar (see below).

View larger version (16K): [in this window] [in a new window]   FIGURE 1   Ca2+ binding by WT, E309Q, T796A, and E309Q/T796A mutants (A) and molecular graphics representation of the two Ca2+-binding sites in WT ATPase (B). (A) Recombinant ATPase protein was obtained from Cos1 cells infected with adenovirus vectors, and Ca2+ binding was measured in the absence of ATP as described by Zhang et al. (2000). , WT; the mutants are E309Q (), N796A (), and E309Q/N796A (). Each point is the average of 30-35 samples. The errors bars correspond to standard deviations of the average of all experimental values obtained with the mutants at any Ca2+ concentration, which were then used for fitting. The experimental points (Ca2+-bound/E) obtained with WT ATPase required fitting with a cooperative two-site equation (K1[Ca2+] + 2K1K2[Ca2+]2/1 + K1[Ca2+] + K1K2[Ca2+]2), whereas those for the E309Q, T796A, and E309Q/T796A mutants could be fitted simply with an independent binding equation (K[Etot][Ca2+]/1 + K[Ca2+]). The dashed line shows the poor fitting of the WT data using an independent binding equation. No significant binding was observed with the E771Q, T799A, or D800N mutants (not shown). (B) Representation of the residues involved in Ca2+ binding was obtained directly from the crystallographic structure of the SR ATPase with Ca2+ bound (Toyoshima et al., 2000) using a silicon graphics (SGI) system with Turbo-FRODO software. M4, M5, M6, and M8 refer to the transmembrane segments originating the binding residues. The section is along the plane of the membrane, viewed from the cytosolic side.

Binding by any of the E309Q, N796A, or E309Q/N796A mutants is not significantly different, indicating that the E309 and N796 mutations eliminate completely binding at one site, while allowing stoichiometric occupancy of the other site. On the other hand, we found that the E771Q, T799A, and D800N mutations interfere totally with binding at either site, within the 7.0-5.0 pCa range (Table 1). This indicates that binding of a first Ca²⁺, involving E771, T799, and D800, is required to reposition the N796 and E309 side chains for high-affinity binding of a second Ca²⁺. In the final cooperative complex, the D800 side chain contributes one oxygen to the first Ca²⁺ complex and the other oxygen to the second, as shown by crystallographic analysis (Fig. 1 B). A most important feature of this system is that even though the E309Q and N796A mutants retain Ca²⁺ binding at the first site, the ATPase cannot be phosphorylated by ATP, and catalytic activity is totally inhibited (Table 1). Therefore, activation of the catalytic site in the cytosolic region requires a long-range signal that is triggered by cooperative occupancy of the second site. It is noteworthy that the plasma membrane ATPase (Strehler et al., 1990), which handles only one Ca²⁺ per cycle, retains a strong homology corresponding to site II, whose occupancy is strictly required for enzyme activation. On the other hand, divergence in the sequence corresponding to site I (such as M for T799, S for N768, and A for E771) do not allow Ca²⁺ binding but are likely to prime the conformation of site II for high-affinity binding.

Another advantageous feature of the Ca²⁺ pump is its reversal, demonstrated by measurements of ATP synthesis coupled to Ca²⁺ efflux from loaded vesicles (Makinose and Hasselbach, 1971). The partial reactions involved in ATP synthesis can be studied by reacting P_i with the enzyme destabilized by Ca²⁺ dissociation, to yield ADP-insensitive phosphoenzyme (Masuda and de Meis, 1973; Kanazawa and Boyer, 1973). Addition of millimolar Ca²⁺ then renders the phosphoenzyme ADP sensitive, and ATP is obtained if ADP is added with Ca²⁺ (Knowles and Racker, 1975; de Meis and Tume, 1977). It is of interest that ADP-insensitive phosphoenzyme can be obtained by reacting site I and site II mutants with P_i just as well as with recombinant WT ATPase (Table 1). Although the P_i reaction with WT enzyme is readily inhibited by high-affinity Ca²⁺ binding, phosphorylation of site I mutants is hardly inhibited by millimolar Ca²⁺ (Clarke et al., 1989). On the other hand, inhibition of the P_i reaction with site II mutants is produced by Ca²⁺ within the micromolar concentration range (Andersen and Vilsen, 1992; Vilsen and Andersen, 1992). We show here that the patterns of inhibition are similar to those of Ca²⁺ binding for WT and the T796A mutant (including the difference in cooperativity; compare Figs. 1 A and 2). Inhibition of the E309Q mutant phosphorylation, however, requires a significantly higher Ca²⁺ concentration (compare Figs. 1 A and 2). It should be pointed out that inhibition of the P_i reaction requires transmission to the catalytic site in addition to Ca²⁺ binding. It is clear that this transmission requires engagement of the M4 helical segments, which normally occurs by participation of E309 in Ca²⁺ binding. It is likely that in the E309Q mutant, alternative oxygen functions participate with lesser affinity.

View larger version (21K): [in this window] [in a new window]   FIGURE 2   Ca2+ inhibition of enzyme phosphorylation by Pi. (A) Examples of phosphorylation detected by electrophoretic gels and phosphoimaging; (B) Average levels of phosphoenzyme obtained by equilibrating WT (), E309Q (), and T796A () protein with Pi (see Materials and Methods) in the presence of various Ca2+ concentrations. (pCa 8 data are repeated to establish reliably the maximal phosphorylation levels. The E771Q, T799A, and D800N exhibited only slight inhibition at pCa 3 (not shown).

A most important finding of our current experiments is that while the WT phosphoenzyme synthesizes ATP with 80% efficiency upon addition of millimolar Ca²⁺ and ADP, no ATP is obtained with either site I or site II mutants, even when millimolar Ca²⁺ is added with ADP (Fig. 3). In fact, the phosphoenzyme formed by reacting the mutant enzyme with P_i remains in its ADP-insensitive form before decaying slowly by hydolytic cleavage (Fig. 3 A). Therefore, even though site II mutants can bind Ca²⁺ to inhibit the P_i reaction, they cannot undergo the Ca²⁺ conformational change that is required to form ATP from phosphoenzyme and ADP. It is therefore clear that occupancy of the second Ca²⁺ site in its high-affinity state and the engagement of N796 and E309 are both necessary to confer high phosphorylation potential to the intermediate formed by the P_i reaction.

View larger version (13K): [in this window] [in a new window]   FIGURE 3   Phosphoenzyme decay (A) and ATP synthesis (B) after addition of Ca2+ and ADP to phosphoenzyme formed by reacting the ATPase with Pi in the absence of Ca2+. Phosphoenzyme was obtained by reacting WT (), E309Q (), and T796A () mutants with Pi in the absence of Ca2+ (Table 1). At time 0, 1 mM Ca2+ and 0.1 mM ADP were added, and serial samples were taken for measurements of residual phosphoenzyme and newly synthesized ATP Reaction conditions and methods were as described by de Meis and Inesi (1982) (see Materials and Methods). The values are in percentage of the phosphoenzyme level at time 0.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The sequence of partial reactions comprising the catalytic and transport cycle of the Ca²⁺ ATPase is given in Fig. 4, which is based on the Post-Albers mechanism as written by de Meis and Vianna (1979) for the Ca²⁺ ATPase. Additional reactions, however, and their microscopic constants are included in the sequence, as required to fit the experimentally observed equilibrium and kinetic behavior of the Ca²⁺ ATPase (Inesi et al., 1988). In the scheme, the effect of site II occupancy by Ca²⁺ occurs after the E'Ca + Ca²⁺  E" Ca₂ reaction, allowing catalytic activation and ATP use to form the phosphorylated enzyme intermediate. A most interesting isomerization of the phosphoenzyme occurs then with the E'-PCa₂  E"~PCa₂ reaction, whereby phosphorylation and binding potentials are affected concomitantly, whereas free energy is conserved through conformational change (K_eq  1). The importance of this isomerization, which requires occupancy of both Ca²⁺ sites, is that it leads to Ca²⁺ dissociation in the forward direction of the cycle and ATP synthesis in the reverse direction. The sequential reactions spelled out in Fig. 4 were originally proposed to explain the cooperative character of Ca²⁺ binding and the kinetic behavior (Inesi et al., 1980) of the SR ATPase. Their occurrence and role in the catalytic cycle are here demonstrated unambiguously by mutational analysis. Furthermore, these phenomena can now be explained in structural terms (Fig. 5). In fact, the long-range effect of Ca²⁺ occupancy of site II is due to engagement of E309 and displacement of M4 transmembrane segment, which is directly connected to the phosphorylation site (D351) through a highly conserved and mutation-sensitive sequence (Zhang et al., 1995). Cooperative interactions within the two sites results in displacement of additional segments, especially the mutation-sensitive M5 and M6/M7 loop, which are then transmitted to the phosphorylation/catalytic site (Sorensen and Andersen, 2000; Zhang et al., 2001). Additional changes within the enzyme headpiece are expected upon nucleotide binding and use. Our findings define a specific functional role for the Ca²⁺-dependent conformational changes demonstrated by crystallographic studies (Toyoshima et al., 2000; Xu et al., 2002; Toyoshima and Nomura, 2002) and its linkage to the phosphorylation potential.

View larger version (21K): [in this window] [in a new window]   FIGURE 4   The catalytic and transport cycle of the SR ATPase. This sequence of partial reactions, with their microscopic constants, is required to fit the observed ATPase behavior, in equilibrium or kinetic experiments (Inesi et al., 1988). Most partial reactions, and their constants, were determined experimentally. The units are in s1 for first-order reactions, and s1 M1 for second-order reactions. The overall Keq is 4.9e+5, and is the result of ATP terminal phosphate hydrolytic cleavage under standard conditions. E' and E" indicate the conformations of the enzyme with one or two calcium ions bound with high affinity. E-P and E~P indicate phosphoenzyme with low or high phosphorylation potential. The constants given here were obtained in studies with rabbit (native) SR ATPase and may not be totally identical to those pertinent to chicken (recombinant) SERCA1.

View larger version (39K): [in this window] [in a new window]   FIGURE 5   Long-range linkage of Ca2+ binding and phosphorylation sites. Structural representation of the transmembrane helices involved in Ca2+ binding, and its connection to the cytosolic phosphorylation domain, was obtained directly from the crystallographic structure of the SR ATPase with Ca2+ bound (Toyoshima et al., 2000). The I298-G360 and R604-S915 sequence segments were selected, using an SGI system with Turbo-FRODO software. D351 and D800 indicate residues with a central role in phosphorylation and high-affinity Ca2+ binding, respectively. The numbers 4-8 refer to transmembrane (M) segments, and L67 refers to the cytosolic loop between M6 and M7. Refer to Fig. 1 B for the contribution of M4, -5, -6, and -8 to Ca2+ binding. The lower third of the structure corresponds to the membrane-bound region. The M1, M2, M3, M9, and M10 transmembrane segments and A (actuator) and N (nucleotide) cytoplasmic domains are not shown.