If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This work was supported in part by grants from the Danish Medical Research Council, the Novo Nordisk Foundation, Denmark, the Lundbeck Foundation, Denmark, the Research Foundation of Aarhus University (to J. P. A.), the Carlsberg Foundation, Denmark (to J. D. C.), and the National Research Foundation of South Africa (to D. B. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1-S8. 1 Supported by the National Health Laboratory Service of South Africa.
ATP binds to sarcoplasmic reticulum Ca2+-ATPase both in a phosphorylating (catalytic) mode and in a nonphosphorylating (modulatory) mode, the latter leading to acceleration of phosphoenzyme turnover (Ca2E1P → E2P and E2P → E2 reactions) and Ca2+ binding (E2 → Ca2E1). In some of the Ca2+-ATPase crystal structures, Arg678 and Glu439 seem to be involved in the binding of nucleotide or an associated Mg2+ ion. We have replaced Arg678, Glu439, and Gly438 with alanine to examine their importance for the enzyme cycle and the modulatory effects of ATP and MgATP. The results point to the key role of Arg678 in nucleotide binding and to the importance of interdomain bonds Glu439-Ser186 and Arg678-Asp203 in stabilizing the E2P and E2 intermediates, respectively. Mutation of Arg678 had conspicuous effects on ATP/MgATP binding to the E1 form and ADP binding to Ca2E1P, as well as ATP/MgATP binding in modulatory modes to E2P and E2, whereas the effects on ATP/MgATP acceleration of the Ca2E1P → E2P transition were small, suggesting that the nucleotide that accelerates Ca2E1P → E2P binds differently from that modulating the E2P → E2 and E2 → Ca2E1 reactions. Mutation of Glu439 hardly affected nucleotide binding to E1, Ca2E1P, and E2, but it led to disruption of the modulatory effect of ATP on E2P → E2 and acceleration of the latter reaction, indicating that ATP normally modulates E2P → E2 by interfering with the interaction between Glu439 and Ser186. Gly438 seems to be important for this interaction as well as for nucleotide binding, probably because of its role in formation of the helix containing Glu439 and Thr441.
) is a membrane-bound energy transducer (“nanomotor”) that couples ATP hydrolysis with Ca2+ translocation against a concentration gradient by means of a reaction cycle (Scheme 1) in which the ATPase enzyme is transiently phosphorylated at a conserved aspartic acid residue and undergoes major conformational transitions between Ca2E1/Ca2E1P and E2/E2P forms (
). The Ca2+-ATPase consists of a membrane-spanning domain of 10 mostly helical segments and a large cytoplasmic headpiece, comprising three distinct domains, named N (nucleotide binding), P (phosphorylation), and A (actuator) (
). In the Ca2E1 and Ca2E1P conformations, the catalytic ATP-binding site is made up by residues in the N- and P-domains, and during the Ca2E1P → E2P transition the departing ADP molecule is replaced by the TGES loop of the A-domain, which subsequently assists in catalysis of E2P dephosphorylation (
). In addition to being the phosphorylating substrate in the E1 state, ATP exerts modulatory effects on various steps of the Ca2+-ATPase cycle (boxed ATP in Scheme 1). Hence, the Ca2E1P → E2P, E2P → E2, and E2 → Ca2E1 transitions are all accelerated by the binding of ATP or MgATP in a nonphosphorylating mode (
). The apparent affinity for the nucleotide is generally lower when it binds in the nonphosphorylating modulatory mode as compared with the phosphorylating mode, but it varies depending on the step being modulated. A subject of much controversy is the question whether the phosphorylating and modulatory ATP molecules are at the same locus, exhibiting variable affinity during the transport cycle depending on conformational state, or whether a separate low affinity allosteric site exists on the same Ca2+-ATPase polypeptide chain (
In previous site-directed mutagenesis studies aimed at locating the amino acid side chains involved in ATP binding, we have examined the functional roles of several conserved residues of the N- and P-domains (
) and found that the N-domain residues Thr441, Glu442, Phe487, Arg489, Lys515, Arg560, and Leu562, as well as the P-domain residues Asp351 (the phosphorylated aspartic acid), Lys352, Thr353, Gly626, Lys684, and Asp703, are important for the binding of ATP together with Mg2+ at the catalytic site. These results are consistent with the more recently published crystal structures of the Ca2+-ATPase with bound nucleotide. On the basis of the crystal structures in conjunction with the information from mutagenesis studies, it is concluded that the binding of ATP together with associated Mg2+ ions at the catalytic site of the Ca2E1 form closes the site by directly bridging the N- and P-domains and mediates the formation of an extensive network of hydrogen bonds (
). One Mg2+ ion is coordinated by P-domain residues together with the γ-phosphoryl group of ATP (“canonical Mg2+ site” or “Mg2+ site 1”). Furthermore, an Mg2+ ion is also found associated with the α,β-phosphates of the nucleotide in some of the crystal structures (“Mg2+ site 2”) (
), see Fig. 1. The Mg2+ ion at site 1 remains bound together with the phosphoryl group in the E2P state after ADP has left and is released from E2 together with Pi later in the cycle (Scheme 1).
In this study, we have addressed the functional roles of P-domain residue Arg678 and N-domain residues Gly438 and Glu439. Arg678 and Glu439 have both been suggested to be involved in the binding of nucleotide. The first indication associating Arg678 with nucleotide binding came from studies showing that chemical cross-linking of the side chains of Arg678 and Lys492 by glutaraldehyde is prevented by ATP or ADP (
The abbreviations used are: AMPPCP, adenosine 5′-(β,γ-methylenetriphosphate); CPA, cyclopiazonic acid; EPPS, N-2-hydroxyethylpiperazine-N′-3-propanesulfonic acid; MES, 2-[N-morpholino]ethanesulfonic acid; MOPS, 3-[N-morpholino]propanesulfonic acid; Tg, thapsigargin; TNP-8N3-ATP, 2′,3′-O-(2,4,6-trinitrophenyl)-8-azidoadenosine 5′-triphosphate; PDB, Protein Data Bank.
3The abbreviations used are: AMPPCP, adenosine 5′-(β,γ-methylenetriphosphate); CPA, cyclopiazonic acid; EPPS, N-2-hydroxyethylpiperazine-N′-3-propanesulfonic acid; MES, 2-[N-morpholino]ethanesulfonic acid; MOPS, 3-[N-morpholino]propanesulfonic acid; Tg, thapsigargin; TNP-8N3-ATP, 2′,3′-O-(2,4,6-trinitrophenyl)-8-azidoadenosine 5′-triphosphate; PDB, Protein Data Bank.
(nonhydrolyzable ATP analog) or ADP, the side chain of Arg678 is sufficiently close to the nucleotide for bond formation with the ribose part (Fig. 1, upper and lower left panels, and Table 1) or even with the adenine ring (Fig. 1, lower right panel, and Table 1). The involvement of Arg678 in nucleotide binding and the binding mode may vary depending on conformational state. A role for Glu439 in nucleotide binding has been inferred from studies showing that mutation of Glu439 interferes with the protection by AMPPCP against proteolytic cleavage (
), these residues have been suggested to be part of the catalytic site and to be in contact with the nucleotide via the Mg2+ ion at site 2. Although an intervening water molecule seems to be needed to mediate interaction between this Mg2+ ion and Glu439 in the Ca2E1 crystal structures (the distance between Mg2+ and Glu439 being 4.9-5.3 Å; see Table 1 and Fig. 1, upper left panel), the crystal structure of the E2 form with bound AMPPCP suggests a closer interaction between the Mg2+ at site 2 and Glu439 (cf.Table 1 and Fig. 1, lower left panel), which might indicate that Glu439 is of importance for the binding of MgATP in the modulatory mode to the E2 form (
it was also speculated that the Mg2+ ion bound at site 2 by Glu439 might play an important role following phosphorylation, first by facilitating ADP dissociation, through attracting the Mg2+ and thereby the phosphoryl groups of ADP away from their binding in a catalytic configuration in Ca2E1P and acting as a flexible helping arm, and second through binding the modulatory ATP promoting the Ca2E1P → E2P transition (cf.Scheme 1). To clarify the roles of Arg678 and Glu439, we have determined the functional properties of Ca2+-ATPase mutants in which Arg678 or Glu439 is replaced by alanine. The function of the adjacent glycine residue Gly438, which appears too far away for direct interaction with the nucleotide or its associated Mg2+ (cf.Table 1 and Fig. 1), but is well conserved, was also examined by replacement with alanine. All the partial reactions of the Ca2+-ATPase reaction cycle indicated in Scheme 1 and their ATP dependences were analyzed in the mutants to address the effects of the mutations on the conformational changes of the enzyme as well as the interaction with ATP in catalytic as well as modulatory modes.
TABLE 1Distances in published crystal structures between nucleotide/Mg and amino acid residues examined in the present study
Distance nucleotide/Mg-amino acid residue (closest distance and other relevant distances)
) was carried out using the QuikChange site-directed mutagenesis kit, and the mutant cDNA was sequenced throughout. To express wild type or mutant cDNA, COS-1 cells were transfected using the calcium phosphate precipitation method (
). The determination of the phosphorylation level by acid quenching followed by acid SDS-PAGE and quantification of the radioactivity associated with the Ca2+-ATPase band as well as the photolabeling with [γ-32P]TNP-8N3-ATP, the inhibition of photolabeling by ATP/MgATP, and the quantification of the label bound specifically to the Ca2+-ATPase were carried out using the previously established procedures (
The experiments were generally conducted at least twice, and average values are shown in the figures. The data were analyzed by nonlinear regression using the SigmaPlot program (SPSS, Inc.) or by computation using the SimZyme program (
). Monoexponential functions were fitted to the phosphorylation and dephosphorylation time courses, except when the phosphorylation time course exhibited an overshoot. In the latter case the SimZyme program was used as described (
). A hyperbolic function plus a constant or linear component was fitted to the [γ-32P]TNP-8N3-ATP labeling data, and the “true” dissociation constant, KD, for ATP/MgATP binding was calculated from the measured K0.5 values using the validated equation for competitive inhibition (
). For analysis of the modulatory effect of nucleotide (“N”) on the rates of the partial reaction steps, the nucleotide concentration dependence of the rate constant was fitted according to Equation 1.
Here, kobs is the rate constant observed at the indicated nucleotide concentration; k0 is the rate constant in the absence of nucleotide (“basic rate”), and kmax is the extrapolated value of the rate constant corresponding to infinite nucleotide concentration. The enhancement factor kmax/k0 describes the extent of the modulatory effect, and the K0.5 value describes the affinity for the modulatory nucleotide.
Expression and Ca2+Affinity—The expression levels of the three mutants, G438A, E439A, and R678A, were similar to that obtained with wild type Ca2+-ATPase, as evaluated by immunoreactivity in a specific enzyme-linked immunosorbent assay. Like the wild type, all the mutants were able to form a phosphoenzyme in the presence of [γ-32P]ATP and Ca2+, and the mutants showed an apparent Ca2+ affinity for activation similar to that of the wild type or slightly higher (supplemental Fig. S1).
Phosphorylation Rate and Affinity for MgATP and ATP—The physiological substrate for phosphorylation of the enzyme in the Ca2E1 form is MgATP (cf.Scheme 1). Fig. 2 shows the time course of phosphorylation with 5 μm [γ-32P]MgATP of enzyme pre-equilibrated with Ca2+. The measured phosphorylation rate depends on the rate constant for transfer of the γ-phosphoryl group of the bound MgATP to Asp351 as well as the saturation of the catalytic site with MgATP. Because the MgATP concentration used is subsaturating in the wild type, any change of MgATP affinity induced by the mutations should be revealed in this experiment. Compared with the wild type, the phosphorylation rate was significantly reduced for G438A (to 80%) and R678A (to 50%), whereas it was wild type-like in E439A. The nucleotide binding properties of the mutants were examined in more detail by using a previously described and validated assay in which the ATPase is photolabeled with [γ-32P]TNP-8N3-ATP (
). The [γ-32P]TNP-8N3-ATP concentration dependence of the photolabeling and the competitive inhibition with ATP/MgATP were studied. This assay was carried out with the enzyme being either in a Ca2+-free E1 state at pH 8.5 (
), the absence of Ca2+ ensuring that the ATP is not being utilized, or in a thapsigargin-bound E2 state, with and without Mg2+ present. Besides allowing studies of nucleotide binding to an E2 form, the thapsigargin-bound state is pertinent as the E2 crystal structures have either thapsigargin or another inhibitor (CPA in Ref.
) bound. Fig. 3 depicts the ATP and MgATP concentration dependences of inhibition of photolabeling, and Table 2 summarizes the resulting KD values for ATP/MgATP binding as well as the K0.5 values obtained from the TNP-8N3-ATP/TNP-8N3-MgATP concentration dependence of photolabeling (see the corresponding data in supplemental Fig. S2). As documented previously (
), Mg2+ and thapsigargin have only moderate effects on the affinity of the wild type for the TNP-8N3-ATP photolabel. The affinity of the wild type for MgATP is, on the other hand, about 40-fold higher than the affinity for metal-free ATP, and the presence of thapsigargin reduces the affinity for MgATP about 250-fold. We did not find any effect of thapsigargin on the affinity for ATP in the absence of Mg2+, thus suggesting that ATP binds differently from MgATP as described previously (
). As is shown in Fig. 3 and Table 2, mutant G438A showed reduced binding affinity for ATP as well as MgATP, relative to the wild type (4-5-fold in the absence of thapsigargin and 2-3-fold in its presence). There was no significant effect of mutation G438A on the affinity for the TNP-8N3-ATP photolabel under any of the conditions tested (Table 2). R678A, on the other hand, had significant effects on the binding of the photolabel as well as the binding of ATP, in the presence or absence of Mg2+ or thapsigargin. A very large reduction of the nucleotide affinity (20-50-fold) was seen with R678A in the absence of thapsigargin, whereas the effect was more modest in its presence (2.4-fold reduced affinity for MgATP, the affinity for metal-free ATP not being measurable because of poor photolabeling; see Fig. 3 and Table 2). The effects of E439A on nucleotide binding were generally much smaller than the effects of the other two mutations. Only a 2-fold reduction of MgATP affinity, relative to wild type, was seen with E439A in the absence of thapsigargin, and in its presence there was no significant difference from wild type. In the absence of thapsigargin, E439A increased the affinity for the photolabel or metal-free ATP about 2-fold. In the presence of thapsigargin, the affinity for metal-free ATP was reduced 2.5-fold. The difference between the three mutants with respect to the magnitude of the mutational effect on MgATP affinity in the E1 form (absence of thapsigargin) matches well the difference between the mutational effects on the phosphorylation rate described above (R678A > G438A > E439A ≅ wild type), indicating that the major reason for the reduced phosphorylation rate in R678A and G438A is the reduced MgATP affinity.
TABLE 2Nucleotide binding affinities determined by TNP-8N3-ATP photolabeling and ATP inhibition thereof
Reaction of the Phosphoenzyme with ADP—The phosphoenzyme intermediate formed immediately by phosphorylation with ATP is the Ca2E1P form, which subsequently is converted into E2P in association with the translocation of Ca2+ (Scheme 1). Ca2E1P is ADP-sensitive, i.e. able to donate the phosphoryl group back to ADP forming ATP, whereas E2P is ADP-insensitive. The reaction of the phosphoenzyme with ADP was examined by addition of 1, 0.1, or 0.02 mm ADP (Fig. 4). It is seen that 1 mm ADP was able to rapidly dephosphorylate more than 90% of the phosphoenzyme in the wild type and all three mutants, thus indicating that Ca2E1P is the major phosphoenzyme intermediate accumulated. However, at the lower ADP concentrations the R678A mutant behaved distinctly different from the wild type and the other mutants by showing a significantly reduced reactivity with ADP. Hence, in R678A 92% of the phosphoenzyme remained 2 s after addition of 0.02 mm ADP, which should be compared with 20% for the wild type, 14% for G438A, and 11% for E439A (Fig. 4). This shows that the affinity of the Ca2E1P form for ADP is reduced in R678A, thus matching the reduction of ATP/MgATP affinity in this mutant described above. On the other hand, the affinity of Ca2E1P for ADP does not appear to be reduced by mutations G438A and E439A.
The Forward Processing of the Ca2E1P Phosphoenzyme and Its Modulation by MgATP and ATP—The dephosphorylation of the Ca2E1P phosphoenzyme was also examined in the absence of ADP, using a chase with EGTA to terminate phosphorylation (removing Ca2+ but not Mg2+). Under these conditions Ca2E1P is converted into E2P, which subsequently undergoes hydrolysis, liberating Pi (cf.Scheme 1). Various concentrations of non-radioactive MgATP were added with the chase (0, 1, and 5 mm MgATP shown in supplemental Fig. S3). For each MgATP concentration, a monoexponential decay function was fitted to the time course of phosphoenzyme decay, and the dependence of the rate constant on the MgATP concentration is displayed in Fig. 5. The data could be fitted satisfactorily by the hyperbolic function described under “Experimental Procedures” (see Equation 1), and the extracted values for k0, kmax, and K0.5 are shown in Table 3. In all three mutants the basic dephosphorylation rate without added MgATP (k0) was found significantly reduced (4-6-fold) relative to wild type. The reduced rate of dephosphorylation of Ca2E1P reflects inhibition of the Ca2E1P → E2P transition, because the latter is the rate-limiting step in the reaction sequence Ca2E1P → E2P → E2 under the conditions applied here (see supplemental Fig. S4). MgATP is known to enhance the Ca2E1P → E2P transition in the wild type by binding with low affinity in a nonphosphorylating modulatory mode, and as seen in Table 3 the analysis in Fig. 5 showed that MgATP, acting with a K0.5 of 1.44 mm, caused a 5-fold increase of the rate of Ca2E1P turnover in the wild type. In E439A, MgATP induced a 14-fold increase of the dephosphorylation rate with a K0.5 of 1.45 mm, thus indicating that the modulatory effect of MgATP on the Ca2E1P → E2P transition is fully intact, and even enhanced, in E439A. In G438A and R678A, MgATP enhanced the rate of Ca2E1P turnover 5- and 9-fold, with K0.5 values of 1.51 and 3.55 mm, respectively, i.e. a wild type-like apparent affinity for MgATP in G438A and a 2.5-fold reduced affinity in R678A (Table 3).
TABLE 3Apparent affinities and kinetic parameters for the modulatory effects of ATP/MgATP on the partial reaction steps
To examine whether the modulatory effect of metal-free ATP is different from that of MgATP, experiments were conducted in which the phosphoenzyme was chased with an excess of EDTA (removing Ca2+ as well as Mg2+) together with various concentrations of metal-free ATP (see the data for 0, 0.2, and 3 mm ATP in supplemental Fig. S5). In this case, the phosphoenzyme decay was not strictly monoexponential, because a small but significant fraction (<30%) decayed much more slowly than the major part. The slow phase may represent phosphoenzyme from which the Mg2+ bound at site 1 (needed for rapid dephosphorylation of E2P) dissociated before the dephosphorylation had taken place. It is also possible that because the Mg2+ concentration present during the phosphorylation was reduced to 1 mm in these experiments, to allow the EDTA to remove Mg2+ efficiently, a significant fraction of the enzyme was phosphorylated by CaATP instead of MgATP. This would result in a biphasic phosphoenzyme decay, because the turnover of phosphoenzyme formed from CaATP (presumably having Mg2+ site 1 occupied by Ca2+) is much slower than that of phosphoenzyme formed from MgATP (
). The two phases were well resolved, thus allowing extraction of the rate constants, and Fig. 6 shows the ATP concentration dependence of the rate constant corresponding to the fast phase (the majority of the decay). Hence, this rate constant is supposed to reflect the rate-limiting part of the Ca2E1P → E2P → E2 reaction sequence, i.e. the Ca2E1P → E2P transition, of enzyme that has Mg2+ bound at site 1 and is being modulated by the added metal-free ATP. It is seen in Fig. 6 and Table 3 that metal-free ATP enhanced the Ca2E1P → E2P transition with an apparent affinity that was quite similar in the wild type and E439A (K0.5, 0.35 and 0.31 mm, respectively) and less than 2-fold reduced in G438 and R678A (K0.5, 0.58 and 0.53 mm, respectively). The apparent affinities for metal-free ATP were 3-7-fold higher than for MgATP. Again the basic dephosphorylation rate without added nucleotide (k0) was found 4-6-fold reduced in the mutants relative to wild type, and the enhancement induced by ATP was most pronounced in E439A, the enhancement factor being as high as 11-fold for E439A (compare with 4-fold for the wild type). The k0 and the kmax values were generally higher than those determined in the presence of free Mg2+, 2-3-fold for the wild type and R678A and somewhat more for G438A, whereas E439 showed only a slight difference. This could be a consequence of removing the Mg2+ ion at site 2, which is supposed to be more loosely bound than the Mg2+ at site 1.
The Dephosphorylation of the E2P Phosphoenzyme and Its Modulation by ATP—All the mutants could be phosphorylated in the reverse direction (“backdoor”) by 32Pi under favorable conditions (absence of Ca2+, acid pH, presence of dimethyl sulfoxide, and absence of K+). To examine the modulatory effect of metal-free ATP on the E2P → E2 step, the dephosphorylation of the E2P phosphoenzyme intermediate formed backdoor from 32Pi was examined following dilution of the phosphorylated sample in a chase medium containing excess of EDTA together with various concentrations of ATP (see supplemental Fig. S6). The ATP dependence of the rate constant for E2P → E2 derived from these data is shown in Fig. 7. When possible, the data were analyzed according to Equation 1 (see “Experimental Procedures”), and the results of the analysis are indicated in Fig. 7 and Table 3. The basic rate observed in the absence of ATP (k0) was found markedly higher in G438A (5-fold) and E439A (11-fold), whereas it was 2-fold reduced in R678A, compared with the wild type rate. ATP, acting with a K0.5 of 19 μm, enhanced the rate of E2P dephosphorylation 3-fold in the wild type, which is in accordance with results obtained with sarcoplasmic reticulum Ca2+-ATPase isolated from rabbit skeletal muscle (
). It is of note that the apparent affinity of E2P of the wild type enzyme for metal-free ATP is more than 10-fold higher than the apparent affinity for modulation of Ca2E1P → E2P discussed above (cf.Table 3). It has been demonstrated previously that only metal-free ATP, and not MgATP, binds to E2P with reasonable affinity (
). The latter observation was confirmed in the present study, as we found the rate of dephosphorylation of E2P insensitive to ATP added in millimolar concentration in the presence of excess Mg2+ (data not shown). Fig. 7 and Table 3 indicate that the modulatory effect of metal-free ATP on the dephosphorylation of E2P was conserved in mutant G438A, albeit with a 2-fold reduction of the apparent affinity relative to wild type. By contrast, E439A showed no enhancement of the dephosphorylation rate upon addition of ATP, neither in the absence (Fig. 7) nor in the presence (not shown) of Mg2+. In fact, in this mutant the dephosphorylation rate was slightly reduced by ATP in the millimolar concentration range. In R678A, there was a slight, hardly significant enhancement of the dephosphorylation rate in the millimolar ATP concentration range, but there was no effect of ATP in the concentration range where the dephosphorylation of the wild type and G438A was accelerated (Fig. 7). Hence, E439A as well as R678A interfered profoundly with the ATP modulation of E2P dephosphorylation. A concern was whether the lack of ATP modulation of the dephosphorylation of E2P in E439A could in some way be a consequence of the high basic rate of dephosphorylation in this mutant (one possibility simply being that the time resolution of the manual mixing technique used for this type of experiment is too low to allow detection of the ATP-induced acceleration of a reaction whose rate is already very high). Therefore, we also examined the ATP modulation of E2P dephosphorylation in another mutant, P248A, for which we previously had demonstrated a very rapid dephosphorylation of E2P in the absence of ATP (
). The data obtained with this mutant are also shown in supplemental Fig. S6 and are summarized in Fig. 7, from which it appears that even though the basic rate of dephosphorylation was about as high for P248A as for E439A, ATP induced an easily measurable 3-fold acceleration of the dephosphorylation in P248A with a K0.5 of 14 μm, i.e. very similar to that of the wild type. Supplemental Fig. S6 and Fig. 7 also show results obtained with three other mutants, T441A, F487S, and R560L, which we previously found defective with respect to ATP binding at the catalytic site in the E1 conformation (
). F487S and R560L displayed conspicuous reductions of the apparent affinity for ATP modulation of E2P dephosphorylation (>50- and 30-fold, respectively). T441A showed a less pronounced, but significant, reduction of the apparent affinity, similar to that described above for G438A.
Rate of the Ca2+Binding Transition and Its Modulation by MgATP—Figs. 8 and 9 show the results of rapid mixing experiments carried out to determine the rate of the Ca2+ binding transition of the dephosphoenzyme (i.e. the E2 → E1 conformational change and accompanying Ca2+ binding leading to Ca2E1) and the modulatory effect of MgATP on this transition. The enzyme was pre-equilibrated in the absence of Ca2+ (presence of EGTA) at acid pH (pH 6.0) to accumulate E2, and the transition to Ca2E1 was followed by taking advantage of the fact that only the Ca2E1 form is phosphorylated by MgATP (
). To follow the transition to Ca2E1 in the absence of MgATP, the Ca2+-depleted enzyme was mixed with Ca2+ without nucleotide and incubated for the time intervals indicated in the figure. The amount of phosphorylatable Ca2E1 accumulated during the Ca2+ incubation step was determined for each time interval by a further 34-ms incubation with [γ-32P]MgATP prior to acid quenching. To follow the transition to Ca2E1 in the presence of MgATP, [γ-32P]MgATP was added together with Ca2+ to the Ca2+-depleted enzyme, and the phosphorylation level was determined by acid quenching at the time intervals indicated in the figure. These two protocols have been previously validated and the resulting rate constants shown to reflect the Ca2+ binding transition (
). A requirement is that the reaction of MgATP with Ca2E1 is relatively rapid compared with the Ca2+ binding transition, and this was the case even with R678A (see supplemental Fig. S7). The results of determining the rate of the Ca2+ binding transition in the absence of MgATP and in the presence of 10 or 50 μm MgATP are displayed in Fig. 8, and Fig. 9 shows the analysis of the MgATP dependence according to Equation 1 under “Experimental Procedures”; the results are summarized in Table 3. The basic rate of the Ca2+ binding transition observed in the absence of MgATP (k0) was 1.6- and 3.4-fold higher in G438A and R678A, respectively, as compared with wild type, whereas it was wild type-like in E439A. In the wild type, MgATP, acting with a K0.5 of 45 μm, enhanced the rate of the Ca2+ binding transition 19-fold, which is in accordance with results obtained with sarcoplasmic reticulum Ca2+-ATPase isolated from rabbit skeletal muscle using a similar method as described here or measurements of the fluorescence change associated with the Ca2+ binding transition (
). In G438A, the K0.5 value was 37 μm, i.e. similar to that of the wild type, and in E439A it was 25 μm, corresponding to a less than 2-fold increase of affinity for the modulatory MgATP. The enhancement factors were somewhat lower than that of the wild type (5- and 9-fold for G438A and E439A, respectively). R678A showed only a very weak MgATP dependence of the Ca2+ binding transition, consistent with a strongly reduced affinity of the E2 form for MgATP. The effect of metal-free ATP was not examined, as the method applied here requires Mg2+ to be present to catalyze phosphorylation.
Nucleotide Dependence of the Steady-state Rate of ATP Hydrolysis—The ATP/MgATP modulation of the various partial reaction steps is reflected in the nucleotide dependence of the ATPase activity, which is rather complex in the wild type Ca2+-ATPase, consisting of at least three activation phases (
). The K0.5 values corresponding to these phases depend on the affinities of the various intermediate states for the nucleotide as well as the rate constants of the partial reaction steps. The ATPase activation profiles shown in Fig. 10 were determined at 37 °C (highest possible temperature, to obtain an activity significantly higher than background even at the lowest nucleotide concentrations) and in the presence of 2 mm free Mg2+, thereby allowing ∼4% of the nucleotide to be metal-free ATP, which can modulate Ca2E1P → E2P and E2P → E2 as described above, and possibly also E2 → Ca2E1, whereas the major part of the nucleotide is MgATP, which is the substrate of the phosphorylation reaction and modulates only the Ca2E1P → E2P and E2 → Ca2E1 transitions. For the wild type, the activation below 100 μm nucleotide likely reflects the role of MgATP as substrate of the phosphorylation reaction in combination with the modulation by MgATP of the Ca2+ binding transition, i.e. the MgATP activation of the reaction sequence E2 → Ca2E1 → Ca2E1P. Two more activation phases may be distinguished for the wild type, one between 0.1 and 1 mm nucleotide and one above 1 mm. The latter two activation phases reflect the ATP/MgATP modulation of the two steps in phosphoenzyme processing, Ca2E1P → E2P and E2P → E2. A characteristic feature of mutant E439A is that it shows a very distinct activation phase between 0.1 and 1 mm nucleotide and lacks the activation phase above 1 mm. From the finding that in mutant E439A the Ca2E1P → E2P transition is very sensitive to nucleotide (cf. Figs. 5 and 6, activation factors 14 and 11 for MgATP and ATP, respectively), whereas the dephosphorylation of E2P is insensitive to nucleotide (Fig. 7), it may thus be deduced that the activation seen in Fig. 10 for E439A between 0.1 and 1 mm nucleotide reflects the Ca2E1P → E2P transition. The activation phase above 1 mm nucleotide seen for the wild type, but not for E439A, may reflect the modulatory effect of metal-free ATP on E2P dephosphorylation, the apparent affinity being low, because most of the nucleotide is present as MgATP rather than metal-free ATP. For R678A, separate activation phases could not be distinguished in Fig. 10, likely because of the marked reduction of the nucleotide affinity of the intermediates that in the wild type exhibit high affinity.
All three mutations studied here interfered profoundly (4-6-fold reduction) with the basic rate of the Ca2E1P → E2P transition observed in the absence of added ATP/MgATP (Figs. 5 and 6 and Table 3). The Ca2E1P → E2P transition is a global protein conformational change that leads to disruption of the high affinity Ca2+-binding sites in the membrane and opening of the Ca2+ release pathway toward the lumen simultaneously with a rotation of the A-domain that causes the replacement of the leaving ADP molecule at the catalytic site with the TGES loop of the A-domain, thereby altering the catalytic specificity from “kinase activity” to “phosphatase activity” (
). For such a large rearrangement to occur at a reasonable rate, new side chain interactions must successively replace those being broken during the course of the transition. No crystal structure corresponding to the E2P intermediate has yet been determined, and we can therefore only guess about the possible interactions of the side chains of Glu439 and Arg678 in E2P. Significantly however, Glu439 and Arg678 are within bonding distance of the respective A-domain residues Ser186 and Asp203 in the crystal structures thought to bear the closest resemblance to E2P (
), cf. supplemental Fig. S8. It is clear from supplemental Fig. S8 that if the interactions of Glu439 and Arg678 in the
structure also exist in genuine E2P, and possibly are formed during the transition from Ca2E1Pto E2P (certainly feasible for the Glu439 and Ser186 pair as the residues are close to the pivot point at Glu486 and Thr171), then they could be very important in guiding or stabilizing the rearrangements of the A-, N-, and P-domains, thus explaining the slowing of the Ca2E1P → E2P transition in mutants E439A and R678A.
Of the three mutants studied here, R678A showed the most conspicuous effects on ATP/MgATP binding in the phosphorylating as well as the modulatory modes. Thus, in mutant R678A the affinity of the E1 form for MgATP was 20-fold reduced relative to wild type (Fig. 3 and Table 2). The phosphorylation rate (Fig. 2) and the affinity of Ca2E1P for ADP (Fig. 4) were likewise reduced significantly in R678A, and a modulatory effect of ATP/MgATP on the E2P → E2 and E2 → Ca2E1 transitions was hardly discernible in R678A (Figs. 7 and 9 and Table 3). The most obvious conclusion seems to be that Arg678 must be directly involved in the ligation of the phosphorylating nucleotide in Ca2E1 and the leaving of ADP in Ca2E1P, as well as in the ligation of the modulatory nucleotide binding to E2P and E2. By contrast, the role played by Arg678 in the ligation of the MgATP/ATP-modulating Ca2E1P → E2P seems less crucial, as the R678A mutant showed only 2.5/2-fold reduced apparent affinity for MgATP/ATP activation of the latter transition (Figs. 5 and 6 and Table 3).
The importance of Arg678 in the binding of MgATP in Ca2E1 and ADP in Ca2E1P is consistent with the E1 crystal structures, where the side chain of Arg678 is within hydrogen bonding distance to the ribose O-3′ atom of AMPPCP or ADP and within van der Waals interaction distance to ribose O-2′ (Fig. 1, upper left panel, and Table 1). To understand the crucial role of Arg678 in MgATP modulation of the E2 → Ca2E1 transition, the most relevant crystal structures to use for reference are E2·AMPPCP and E2·ADP (Fig. 1, lower panels). In addition to the nucleotide, the E2·AMPPCP structure contains Mg2+ bound at site 2, as well as thapsigargin for stabilization of the membrane part (
). There is no Mg2+ at site 1, in agreement with the view that the latter Mg2+ leaves with the phosphate (Scheme 1). In the E2·AMPPCP structure, the side chain of Arg678 is within van der Waals interaction distance of the ribose O-3′ atom of AMPPCP (cf.Table 1). Consistent with such interaction, mutation R678A was found to reduce the affinity of the thapsigargin-bound enzyme for MgATP 2.4-fold (Table 2). The almost complete disruption of the modulatory effect of MgATP on the E2 → Ca2E1 transition by mutation R678A (Fig. 9) indicates, however, that Arg678 is even more important for MgATP binding in the functioning enzyme in the E2 state than it is for MgATP binding to the thapsigargin-inhibited enzyme. Likewise, the difference between the MgATP affinity constants of the E2 conformation of the thapsigargin-inhibited wild type enzyme (K0.5 130 μm; cf.Table 2) and the functioning wild type enzyme (45 μm, cf.Table 3) suggests some perturbing influence of thapsigargin. It is possible that in the E2 state of the functioning Ca2+-ATPase, Arg678 is able to approach the ribose of MgATP somewhat closer than in the thapsigargin-bound state. Moreover, an alternative relevant possibility for the binding of the modulatory MgATP to the E2 form has appeared from the recently published crystal structure of E2·ADP (
). In this structure the adenine ring of ADP is sandwiched between Arg678 and Arg489, the guanidino group of Arg678 being involved in cation-π interaction (Fig. 1, lower right panel);aMg2+ ion binds near the canonical site 1, and the β-phosphate of ADP takes the place of the enzyme-bound phosphoryl group as Mg2+ ligand, and the membrane domain is stabilized by the inhibitor CPA. A similar central role of Arg678 in the binding of MgATP in the E2 form of the native enzyme would clearly be in good accordance with the conspicuous effect of the R678A mutation on the modulatory effect of MgATP seen in Fig. 9. Both of the above mentioned modes of interaction of the MgATP with Arg678 would interfere with the ion pairing between Arg678 and Asp203 that attaches the A-domain to the P-domain in the E2·MgF2-4 structure (cf.Fig. 1). Because the basic rate of the E2 → Ca2E1 transition was found 3.4-fold higher in R678A as compared with the wild type (cf.Table 3), it seems that E2 is stabilized by the ion pairing. Hence, it is likely that the competition between MgATP and Asp203 for bond formation with Arg678 destabilizes E2 and promotes the E2 → Ca2E1 transition by helping release the A-domain.
By contrast, the basic rate of the E2P → E2 step was found 2-fold reduced in R678A. Hence, in this case the mutation exerts an effect opposite that of the nucleotide, and it is therefore not likely that interference of the nucleotide with the ion pairing between Arg678 and Asp203 plays any significant role in the strong effect of mutation R678A on the modulation of E2P → E2 by ATP. In
, the closest analog of E2P with bound nucleotide for which a structure is available, the ADP is bound such that the side chain of Arg678 is too far away to interact with the ribose or adenine ring. The β-phosphate of the ADP is, however, only 5.0 Å from the side chain of Arg678 (Fig. 1, upper right panel, and Table 1), and it is therefore reasonable to speculate that in E2P Arg678 interacts with the γ-phosphate of ATP. Importantly ADP, unlike ATP, does not accelerate the dephosphorylation of E2P (
). Hence, the interaction of Arg678 with the γ-phosphate of ATP may be required for the modulatory effect on E2P → E2. This hypothesis could also explain that the modulatory ATP binding to E2 P needs to be Mg2+-free, because a Mg2+ ion near the γ-phosphate might prevent the interaction with Arg678. Because the α- and β-phosphates of the ADP bound in the
structure seem to interact with residues in the A-domain, ATP bound at the same position, and with the γ-phosphate near Arg678, might actually stabilize the attachment of the A-domain to the P-domain even more than the ion pairing between Arg678 and Asp203, and this might lead to the acceleration by ATP of E2P → E2 through stabilization of the transition state of the reaction.
Considering the conspicuous effects of mutation R678A on the modulation of E2P → E2 and E2 → Ca2E1 by ATP/MgATP, it is remarkable that the affinity for ATP/MgATP modulation of the Ca2E1P → E2P transition was little affected by mutation R678A. Importantly, a similar situation exists with Phe487 and Arg560, which are key residues in the binding of MgATP at the catalytic site (
), and like Arg678 were found critical to ATP modulation of E2P → E2 in the present study (Fig. 7), whereas they could be substituted without significant effect on the MgATP modulation of Ca2E1P → E2P (
). Thus, the picture that emerges is one in which the nucleotide that modulates the E2P → E2 and E2 → Ca2E1 steps is ligated by some of the same residues (including Arg678) that form the catalytic site in Ca2E1 and Ca2E1P, whereas the ATP/MgATP that accelerates the Ca2E1P → E2P transition binds differently, perhaps at a completely different locus, which would be in agreement with a previous proposal based on the finding that ATP does not compete with ADP for binding to Ca2E1P (
). The higher apparent affinity of ATP compared with MgATP for modulation of Ca2E1P → E2P (Table 3) would be consistent with a critical role of electrostatic forces in the interaction involved here (note that the situation is opposite at the catalytic site, MgATP binding with higher affinity than ATP; cf.Table 2). The need for a nucleotide site for enhancement of the Ca2E1P → E2P transition differing from the nucleotide site involved in the modulation of the rate of E2 → Ca2E1 may arise from the fact that the rearrangements of the contacts between the A-, N-, and P-domains occurring in relation to the Ca2E1P → E2P transition are opposite those associated with E2 → Ca2E1. The binding of MgATP to E2 at the site containing Phe487, Arg560, and Arg678 as key residues promotes the E2 → Ca2E1 transition by tying the N- and P-domains together and loosening the contact with the A-domain, thereby turning this modulatory site into the catalytic site by a sort of “induced fit” mechanism. Because the opposite rearrangements of the cytoplasmic domains take place during the Ca2E1P → E2P transition, interactions of the N- and P-domains with the A-domain being established and the TGES loop of domain A being inserted into the catalytic site, it may seem logical to find that the nucleotide interaction needed to promote Ca2E1P → E2P is very different from that needed to promote E2 → Ca2E1.
The closeness of Glu439 to the Mg2+ ion at site 2 in the
and E2·AMPPCP crystal structures (Fig. 1 and Table 1) has led to the suggestion that this glutamate is important in the binding of MgATP in catalytic as well as modulatory modes, and perhaps in aiding the release of the ADP leaving from Ca2E1P (
). We found small and opposite effects of the E439A mutation on the affinities for MgATP (2-fold reduction) and metal-free ATP (2-fold increase) in the absence of thapsigargin (Fig. 3 and Table 2), which could indeed reflect a weak electrostatic interaction between Glu439 and a Mg2+ ion associated with ATP in E1. It appears from Fig. 2 that such interaction, if it exists, is too weak to influence the rate of phosphorylation. There was likewise no significant effect of E439A on the apparent affinity of Ca2E1P for ADP (Fig. 4), thus contradicting the suggestion that Glu439 interacts in a functionally important way with the leaving ADP molecule (
). Furthermore, Glu439 was not found to be critical for ATP/MgATP ligation at the site involved in the activation of the Ca2E1P → E2P transition, as the activation was retained with normal apparent affinity in E439A and a markedly increased enhancement factor relative to wild type (Figs. 5 and 6 and Table 3). The apparent affinity with which MgATP modulates the E2 → Ca2E1 transition was slightly increased (less than 2-fold) in E439A, relative to wild type (Table 3), and the affinity of the thapsigargin-bound E2 form of E439A for MgATP was wild type-like (Table 2). Hence, it seems clear that despite the closeness of Glu439 to the Mg2+ ion at site 2 in the E2·AMPPCP crystal structure, Glu439 contributes very little, if at all, to the binding of MgATP in the E2 form of the native enzyme.
A highly interesting finding with E439A was the complete disruption of the modulatory effect of metal-free ATP on the dephosphorylation of E2P. E439A furthermore showed a conspicuous 11-fold increase of the basic rate of E2P dephosphorylation in the absence of ATP, i.e. constitutive activation (Fig. 7 and Table 3). Our studies of mutant P248A, which like E439A showed a marked enhancement of the basic rate of E2P dephosphorylation, but nevertheless retained a wild type-like modulation by ATP (Fig. 7), demonstrate that the disruption of ATP modulation in E439A is not a consequence of the increased basic rate of E2P dephosphorylation. In E439A, the destabilization of E2P may be caused by the absence of the hydrogen bond between Glu439 and Ser186 mentioned above (cf.Fig. 1, upper right panel, and supplemental Fig. S8). This explanation would be consistent with a previous finding that the S186F mutation identified in patients with Darier disease leads to a block of the Ca2E1P → E2P transition as well as activation of E2P → E2, similar to what is seen for E439A (
). Hence, it may be further speculated that when ATP binds to E2P, the dephosphorylation is accelerated because the nucleotide disturbs the interaction between Glu439 and Ser186. Thus, the importance of Glu439 for the modulatory effect of ATP on E2P dephosphorylation would not arise because of involvement of Glu439 in the ligation of ATP but because ATP interferes with the function of Glu439 in the stabilization of E2P. The increased enhancement factor (with normal apparent affinity) seen for the modulatory effect of ATP/MgATP on the Ca2E1P → E2P transition in E439A (11/14-fold versus 4/6-fold in the wild type; cf. Figs. 5 and 6 and Table 3) might be the consequence of the Ca2E1P → E2P transition becoming almost irreversible, because the product state, E2P, is removed instantaneously because of the high dephosphorylation rate in the mutant.
Finally, we focus on the results obtained with the G438A mutant. Like E439A, G438A showed both a markedly reduced rate of the Ca2E1P → E2P transition and a significant increase of the basic rate of E2P dephosphorylation (Table 3). Furthermore, G438A showed moderate reductions of the affinity for ATP/MgATP in the binding assays (Table 2) and a slight 2-fold reduction of the apparent affinity for the ATP modulating E2P → E2 (Fig. 7 and Table 3), an effect very similar to that of mutation T441A (Fig. 7). Thr441 was previously shown to be an important residue for ATP/MgATP binding to E1 (
). The crystal structures show that Glu439 and Thr441 are located in the first part of an α-helix that is preceded by a loop, whose structure depends on the presence of Gly438. The bend between the α-helix and the loop at this glycine is very acute (cf.Fig. 1, upper right corner of upper left panel), and a Ramachandran plot analysis reveals dihedral angles that would be energetically very unfavorable for residues with side chains. It is therefore rather unlikely that these angles would be maintained in mutant G438A. The consequence of the mutation might well be that the loop expands to make the bend less acute, thereby partially unwinding the N-terminal part of the helix containing Glu439 and Thr441. This would clearly disturb the interactions of the Glu439 and Thr441 side chains to some extent, thus providing a neat explanation of the resemblance of the functional effects of the G438A mutation to some of those seen for the E439A and T441A mutations.
We thank Lene Jacobsen and Karin Kracht (University of Aarhus, Denmark), and Irene Mardarowicz and Joy Norman (University of Cape Town, South Africa) for expert technical assistance. Drs. Poul Nissen, Jesper V. Møller, Anne-Marie Lund Jensen, and Claus Olesen (University of Aarhus, Denmark) and Chikashi Toyoshima (University of Tokyo, Japan) are thanked for discussions and for providing information about their unpublished structural work. We thank Anja P. Einholm for participating in the preparation of mutant cDNA.