Importance of Thr-353 of the Conserved Phosphorylation Loop of the Sarcoplasmic Reticulum Ca2+-ATPase in MgATP Binding and Catalytic Activity*

Mutants in which Thr-353 of the Ca2+-ATPase of sarcoplasmic reticulum had been replaced with alanine, serine, glutamine, cysteine, valine, aspartate, or tyrosine were analyzed functionally. All the mutations severely affected MgATP binding, whereas ATP binding was close to normal in the alanine, serine, glutamine, and valine mutants. In the serine and valine mutants, the maximum rate of phosphorylation from MgATP was 8- and 600-fold lower, respectively, compared with wild type. Replacement of Mg2+ with Mn2+ led to a 1.5-fold enhancement of the maximum phosphorylation rate in the valine mutant and a 5-fold reduction in the wild type. The turnover of the phosphoenzyme formed from MgATP was slowed 1–2 orders of magnitude relative to wild type in the alanine, serine, and valine mutants, but was close to normal in the aspartate and cysteine mutants. Only the serine mutant formed a phosphoenzyme in the backward reaction with Pi, and the hydrolysis of this intermediate was greatly enhanced. Analysis of the functional changes in the mutants in the light of the recent high resolution structure of the Ca2+-ATPase crystallized without the MgATP substrate suggests that, in the native activated state of the enzyme, the side chain hydroxyl of Thr-353 participates in important interactions with nucleotide and phosphate, possibly in catalysis, whereas the main chain carbonyl of Thr-353, but not the side chain, may coordinate the catalytic Mg2+.

A fundamental property of the P-type ion transporting AT-Pases is their ability to bind the substrate MgATP with high affinity and catalyze the phosphorylation of a conserved aspartic acid residue in the presence of activating ions (1). The formation as well as the further processing of the aspartyl phosphorylated intermediate is associated with protein conformational changes that couple the events in the catalytic site with changes in the ion binding sites leading to ion translocation across the membrane (Fig. 1). The recent 2.6-Å resolution structure of the sarcoplasmic reticulum Ca 2ϩ -ATPase 1 crystal-lized with bound Ca 2ϩ , but without the MgATP substrate (2), has revealed that the cytoplasmic portion of the protein is made up from three distinct domains (A, P, and N) that are rather loosely attached to each other and must move considerably to accomplish substrate binding and energy transduction (2,3). The molecular nature of the binding sites for ATP and the catalytic magnesium ion, as well as the conformational changes involved in energy transduction, are not well understood, and more information is clearly needed about the functions of the individual amino acid residues in the catalytic site.
Thr-353 of the sarcoplasmic reticulum Ca 2ϩ -ATPase is located in the catalytic region (domain P (2)) close to the phosphorylated residue, Asp-351, and is highly conserved within the family of P-type cation pumps. Maruyama and co-workers (4) reported that mutation of Thr-353 to serine or alanine resulted in pumps with reduced Ca 2ϩ transport activity but with preserved ability to undergo phosphorylation. These observations led to the proposal that the main role of Thr-353 is in the events that take place following phosphorylation, i.e. the conformational changes that are required for release of Ca 2ϩ to the lumen or the dephosphorylation (4). More recently, it was shown that Thr-353 is the residue at the phosphorylation site that becomes oxidized by the phosphate transition-state analogue monovanadate upon UV radiation under conditions favoring accumulation of the Ca 2ϩ bound E 1 form (5).
The catalytic site of the P-type ATPases seems to be homologous to that of haloacid dehalogenase, phosphoserine phosphatase, phosphonatase, phosphomutase, and CheY, the response regulator protein of bacterial chemotaxis (6 -9). Upon mutagenesis of the residue at the position equivalent to Thr-353, these proteins show considerable changes in their activity (10 -12). The atomic structures of phosphoserine phosphatase, phosphonatase, and CheY have demonstrated that the main chain carbonyl at the position corresponding to Thr-353 contributes to coordination of the catalytic Mg 2ϩ (7,9,13,14), and a similar role in Mg 2ϩ coordination has been suggested for the main chain carbonyl of Thr-353 in the P-type ATPases (8).
In the present study, we have subjected Thr-353 of the Ca 2ϩ -ATPase to mutational analysis to elucidate its role in the catalytic cycle of the pump. Thr-353 has been replaced by serine, alanine, valine, glutamine, aspartate, cysteine, and tyrosine to study the effects of variation in the size, polarity, and charge of the side chain. Several of the reaction steps shown in Fig. 1 were examined, and the results reveal a multiplicity of important functions of the threonine.

Mutagenesis, Expression, and Assays of Overall Function-Oligonu-
cleotide-directed mutagenesis of cDNA encoding the rabbit fast twitch muscle Ca 2ϩ -ATPase (SERCA1a isoform) was carried out as described previously (15). For expression, the wild type or mutant cDNA, inserted in the pMT2 vector (16), was transfected into COS-1 cells using the calcium phosphate precipitation method (17). The microsomal fraction containing expressed wild type or mutant Ca 2ϩ -ATPase was isolated by differential centrifugation (18). The concentration of expressed Ca 2ϩ -ATPase was quantified by a specific enzyme-linked immunosorbent assay (19) and by immunoblotting (15). Expressed wild type SERCA1a, for which the concentration had been determined by measurement of the maximum capacity for phosphorylation by inorganic phosphate in the presence of 30% (v/v) dimethyl sulfoxide (20), was used as standard. For evaluation of the expression level, the concentration of expressed Ca 2ϩ -ATPase was related to the total microsomal protein concentration determined by the dye-binding method of Bradford (21). The ATPdriven transport of 45 Ca 2ϩ into the microsomal vesicles was measured by filtration, and the ATPase activity was measured by determining the amount of P i liberated, as described previously (20), under conditions corresponding to maximal activity for the wild type at 37°C, pH 7.0, and 5 mM MgATP. Following the subtraction of the background activity determined with control microsomes isolated from mock-transfected COS-1 cells, the specific activity was calculated by relating the rate of Ca 2ϩ transport or ATP hydrolysis to the expression level.
Phosphorylation from [␥-32 P]ATP and 32 P i -Manual mixing experiments at various buffer and temperature conditions (detailed in the figure legends) were carried out according to the principles described previously (15,19,20). Transient kinetic experiments at 25°C were performed using a Bio-Logic quench-flow module QFM-5 (Bio-Logic Science Instruments, Claix, France) as described previously (22). In all phosphorylation experiments, acid quenching was performed with 0.5-2 volumes of 25% (w/v) trichloroacetic acid containing 100 mM H 3 PO 4 . The acid-precipitated protein was washed by centrifugation and subjected to SDS-polyacrylamide gel electrophoresis in a 7% polyacrylamide gel at pH 6.0 (23), and the radioactivity associated with the separated Ca 2ϩ -ATPase band was quantified by imaging using a Packard Cyclone Storage Phosphor System. Background phosphorylation levels were subtracted from all data points. The background was usually determined in parallel experiments with control microsomes isolated from mock-transfected COS-1 cells. In some of the dephosphorylation experiments, the constant phosphorylation level reached after the exponential decay was taken as background (maximally 10% of the initial phosphorylation level).
[␥-32 P]TNP-8N 3 -ATP Labeling-The synthesis of [␥-32 P]TNP-8N 3 -ATP, the photolabeling of COS-1 cell microsomes containing wild type or mutant Ca 2ϩ -ATPase, the inhibition by ATP, and the quantification of labeled bands by electronic autoradiography following SDS-polyacrylamide gel electrophoresis were carried out as described previously (24,25). Generally, the concentration of [␥-32 P]TNP-8N 3 -ATP was 3⅐K 0.5 in the inhibition experiments with ATP. Details of the buffer composition are given in the figure legends.
Calculations and Data Analysis-The ion concentrations in the reaction buffers were calculated using the program WEBMAXC, available on the World Wide Web, and the stability constants therein (26). The phosphorylation data were analyzed by nonlinear regression using the SigmaPlot program (SPSS, Inc.). Monoexponential functions were fitted to the phosphorylation and dephosphorylation time courses. Initial phosphorylation rates per ATPase molecule in units of s Ϫ1 were obtained from the slopes of the fitted curves at time zero, following normalization of the phosphorylation levels to the enzyme concentration. The analysis of ligand concentration dependences was based on the Hill equation, For analysis of the [␥-32 P]TNP-8N 3 -ATP labeling data, a constant or linear component was added to represent nonspecific labeling as described (24,25), and the Hill coefficient was set to 1. The "true" dissociation constant for ATP and MgATP binding was calculated using the previously validated equation for competitive inhibition of the [␥-32 P]TNP-8N 3 -ATP labeling (25).

RESULTS
Expression and Overall Function of Mutants-Seven mutants with changes to Thr-353 were examined in the present study, Thr-353 3 Ala, Thr-353 3 Ser, Thr-353 3 Gln, Thr-353 3 Cys, Thr-353 3 Val, Thr-353 3 Asp, and Thr-353 3 Tyr, all of which were expressed to levels similar to that of the wild type. Under conditions corresponding to maximal activity for the wild type at 37°C, pH 7.0, 10 -100 M Ca 2ϩ , and 5 mM MgATP, mutant Thr-353 3 Ser was able to transport Ca 2ϩ and hydrolyze ATP at a rate of ϳ20% of the wild type rate, in good agreement with previously published data (4). For all the other Thr-353 mutants, the Ca 2ϩ transport and ATPase activity determined under these conditions constituted less than 5% of the activity corresponding to wild type (data not shown).
Overview of the Phosphorylation Properties of the Thr-353 Mutants-In Fig. 2 is shown the results of experiments carried out to obtain an initial overview of the phosphorylation properties of the Thr-353 mutants. The wild type Ca 2ϩ -ATPase in the Ca 2 E 1 state is able to form a phosphoenzyme intermediate by reaction with MgATP (cf. Fig. 1). It can be seen in Fig. 2A that only the wild type and the mutant Thr-353 3 Ser formed significant amounts of phosphoenzyme upon incubation with 2 M [␥-32 P]ATP for 15 s at 0°C and pH 7.0 in the presence of 100 M Ca 2ϩ and 5 mM Mg 2ϩ (saturating concentrations for the wild type). When, however, the incubation was carried out for 5 s at 25°C in the presence of 5 M [␥-32 P]ATP, under otherwise identical conditions, a markedly increased level of phosphorylation was seen for mutants Thr-353 3 Ala, Thr-353 3 Cys, Thr-353 3 Asp, and, in particular, Thr-353 3 Val (Fig. 2B).
It has previously been demonstrated that the catalytic Mg 2ϩ can be replaced by various other divalent cations. Among these, Mn 2ϩ seems to be the most efficient co-substrate, both for phosphorylation from ATP (27) and for the backward phosphorylation from P i (28). Titration of the Mn 2ϩ dependence of the ATPase activity has shown that the concentration of Mn 2ϩ required for maximum activity is lower than that of Mg 2ϩ (29). Fig. 2C shows the results of experiments performed under conditions similar to those in Fig. 2B except for the replacement of the 5 mM Mg 2ϩ with 0.5 mM Mn 2ϩ . This caused a further 2-to 5-fold increase in the phosphorylation level of mutants Thr-353 3 Ala, Thr-353 3 Cys, Thr-353 3 Val, and Thr-353 3 Asp, and even mutant Thr-353 3 Gln now showed significant phosphorylation. Although Mn 2ϩ forms a tighter complex with sulfur than Mg 2ϩ , allowing the use of "Mn 2ϩ rescue" to locate metal ion binding sites (30,31), the replacement of Mg 2ϩ with Mn 2ϩ did not increase the phosphorylation level more in Thr-353 3 Cys than in Thr-353 3 Ala, Thr-353 3 Val, or Thr-353 3 Asp, suggesting that the side chain does not interact directly with the metal ion.
In the absence of Ca 2ϩ , the Ca 2ϩ -ATPase in the E 2 form can also be phosphorylated in the backward direction of normal turnover with inorganic phosphate, resulting in the accumulation of E 2 P, cf. Fig. 1. As seen in Fig. 2D, all the Thr-353 mutants displayed a severely reduced ability to undergo phosphorylation from 32 P i . Under conditions optimal for phosphorylation from P i in the wild type, mutant Thr-353 3 Ser was phosphorylated to a level of 24% that of the wild type, whereas the remaining six Thr-353 mutants showed no phosphorylation above the level of control microsomes. Contrary to the situation with phosphorylation from ATP, replacement of Mg 2ϩ with Mn 2ϩ did not increase the P i phosphorylation levels of the mutants. Even Thr-353 3 Ser showed no phosphorylation under these conditions, whereas the wild type phosphorylation level remained practically unaltered (data not shown).
Nucleotide Binding-To investigate the consequences of mutations to Thr-353 for the binding of ATP and MgATP (the latter often being considered the "true substrate" (32)), a specific photolabeling assay was used, in which the affinities can be measured by competitive inhibition of [␥-32 P]TNP-8N 3 -ATP photolabeling (24,25,33). The photolabeling is carried out in the absence of Ca 2ϩ (to prevent phosphoryl transfer to the enzyme) and at pH 8.5 (to reduce nonspecific labeling to minimum and ensure that the predominant enzyme conformation is E 1 (33)), and ATP is added at various concentrations either in the presence or in the absence of Mg 2ϩ . Even though the photolabel may bind at a site that differs slightly from that of ATP or MgATP, there is sufficient overlap between the sites to ensure efficient competition, and we previously demonstrated that the competition assay produces highly accurate values for the ATP and MgATP binding affinities of wild type and mutant Ca 2ϩ -ATPase expressed in COS-1 cells (25,33). Results obtained with the Thr-353 mutants are summarized in Table I and exemplified by Thr-353 3 Ala and Thr-353 3 Ser in Fig. 3. For the wild type Ca 2ϩ -ATPase, the K 0.5 for TNP-8N 3 -MgATP photolabeling and the dissociation constant, K D , for MgATP determined in the presence of Mg 2ϩ have previously been shown to be close to 1.0 and 0.5 M, respectively (25,33). Under the same conditions, all the Thr-353 mutants, apart from Thr-353 3 Asp, showed concentration dependences for TNP-8N 3 -MgATP labeling quite similar to that of the wild type, indicating that Thr-353 does not contribute significantly to the binding of the photolabel ( Fig. 3A and Table I). On the other hand, the affinity for MgATP was, for all Thr-353 mutants, severalfold lower (K D increased) than that of the wild type: 6-to 10-fold for mutants Thr-353 3 Ser, Thr-353 3 Cys, and Thr-353 3 Val, 65-to 187-fold for Thr-353 3 Ala, Thr-353 3 Gln, and Thr-353 3 Tyr, and more than 1000-fold for Thr-353 3 Asp ( Fig. 3B and Table I).
In the absence of Mg 2ϩ (i.e. presence of EDTA), the wild type displays a 10-fold higher affinity (decrease in K 0.5 ) for TNP-8N 3 -ATP labeling as compared with the affinity in the presence of Mg 2ϩ . By contrast, the affinity for ATP is 33-fold lower than the affinity for MgATP (33). Like the wild type, all mutants showed higher affinity for the photolabel in the absence of Mg 2ϩ than in its presence ( Fig. 3A and Table I). The affinity for ATP determined by competition with the photolabel was for mutant Thr-353 3 Cys 35-fold lower than the affinity for MgATP (Table I), whereas mutants Thr-353 3 Ser and Thr-353 3 Val showed only a 4-to 7-fold decrease in the affinity upon removal of Mg 2ϩ (compare with 33-fold for the wild type), and mutants Thr-353 3 Ala, Thr-353 3 Gln, Thr-353 3 Asp, and Thr-353 3 Tyr displayed an affinity for ATP higher than   c The "true" K D calculated under the assumption of competitive inhibition as previously described (24,25). In the inhibition experiments, the concentration of TNP-8N 3 -ATP was 3 ⅐ K 0. 5 Fig. 2, B and C. In the presence of Mg 2ϩ , the wild type reaches the steady-state level of phosphorylation within a few hundred milliseconds, with a slight initial overshoot (Fig. 4A, cf. also Ref. 22). For simplicity, a monoexponential function was fitted to the data, giving an apparent rate constant k obs ϭ 63 s Ϫ1 for the approach to steady state. Under the same conditions, the k obs was 7.3 s Ϫ1 for Thr-353 3 Ser, 0.9 s Ϫ1 for Thr-353 3 Cys, and only 0.16 s Ϫ1 for Thr-353 3 Val (Fig. 4, A and B). A very slow rise in the phosphorylation level was also seen for mutant Thr-353 3 Ala (k obs ϭ 0.11 s Ϫ1 , data not shown). It should be noted that the k obs , like the steadystate level of phosphoenzyme, is a complex function of several rate constants and depends not only on the rate of phosphorylation but also on the rate of dephosphorylation. The k obs reflects the "true" rate constant for phosphorylation only in those cases where the dephosphorylation is slow relative to the phosphorylation (wild type, Thr-353 3 Ser, Thr-353 3 Val, and Thr-353 3 Ala, see below). Because the dephosphorylation of Thr-353 3 Cys is relatively fast (see below), the k obs value for this mutant most likely overestimates the "true" rate constant for phosphorylation.
Replacement of Mg 2ϩ with Mn 2ϩ resulted in a lowering of the k obs value for phosphorylation to 27 s Ϫ1 in the wild type (Fig. 4A). For mutant Thr-353 3 Ser, a k obs value of 8.0 s Ϫ1 was observed in the presence of Mn 2ϩ , i.e. indistinguishable from that observed with Mg 2ϩ , whereas for mutant Thr-353 3 Val the k obs value increased 2-fold (to 0.32 s Ϫ1 ) upon the substitution of Mn 2ϩ for Mg 2ϩ (Fig. 4B), in accordance with the increased phosphorylation level seen in Fig. 2C.
To investigate whether the reduced phosphorylation rates The data for Thr-353 3 Ser (squares) and Thr-353 3 Ala (triangles) are shown together with the best fits to the data for the expressed wild type (broken lines) obtained previously (25,33). A, binding of TNP-8N 3 -MgATP or TNP-8N 3 -ATP. Photolabeling was performed in 25 mM EPPS/tetramethyl ammonium hydroxide (pH 8.5), 20% (v/v) glycerol, the indicated concentrations of [␥-32 P]TNP-8N 3 -ATP, and either 1 mM MgCl 2 and 0.5 mM EGTA (presence of Mg 2ϩ ) or 2 mM EDTA (absence of Mg 2ϩ ). B and C, binding of MgATP and ATP, respectively. Photolabeling was performed as described above (at a [␥-32 P]TNP-8N 3 -ATP concentration of 3⅐K 0.5 ) in the presence of either 1 mM MgCl 2 and 0.5 mM EGTA or 2 mM EDTA, and ATP was included in the buffer at the concentrations indicated. The lines show the best fits to the data of the equations described under "Experimental Procedures", and the derived K 0.5 and K D values are listed in Table I were sensitive to an increase in the concentration of free Mg 2ϩ , the time dependence of phosphorylation was examined also in the presence of 20 mM Mg 2ϩ . The resulting data (not shown) gave k obs values of 57, 7.6, and 0.14 s Ϫ1 for the wild type, Thr-353 3 Ser, and Thr-353 3 Val, respectively, i.e. indistinguishable from those obtained at 5 mM Mg 2ϩ . Thus, the reduced rates of phosphorylation seen with the Thr-353 mutants cannot be explained by lack of saturation of a site binding free Mg 2ϩ . On the other hand, this result does not contradict an effect on the affinity for the MgATP complex, because the 4-fold increase in the concentration of free Mg 2ϩ does not change the concentration of MgATP significantly (4.95 M MgATP at 5 mM Mg 2ϩ and 4.99 M at 20 mM Mg 2ϩ ).
Maximum Phosphorylation Rate and ATP Dependence of Phosphorylation-The very large difference between the k obs values for phosphorylation of mutants Thr-353 3 Ser and Thr-353 3 Val seen in Fig. 4 is surprising in the light of the similar affinities of these two mutants for MgATP (Table I) and suggests that the V max for phosphorylation differs between Thr-353 3 Ser and Thr-353 3 Val. To address this question, the substrate concentration dependence of the initial phosphorylation rate was examined for the wild type and mutants Thr-353 3 Ser and Thr-353 3 Val under conditions otherwise identical to those described for Fig. 4. The results are shown in Fig. 5 as double-reciprocal plots of the initial phosphorylation rate per ATPase molecule against the MgATP concentration. Because these plots were linear, the V max for phosphorylation as well as the Michaelis constant (K m ) could be extracted (see legend to Fig. 5). The V max was found 8-and 600-fold reduced in mutants Thr-353 3 Ser and Thr-353 3 Val, respectively, relative to wild type. The 75-fold difference between the V max values of the two mutants explains the difference between the k obs values for phosphorylation seen in Fig. 4. On the other hand, for both mutants the K m for phosphorylation by MgATP is similar to that of the wild type. It should be noted in this connection that, because the K m is a function of several kinetic constants, the K m should not necessarily be expected to correlate with the "true" dissociation constant, K D , for the enzymesubstrate complex (34), which was found to be 10-fold increased for either mutant, relative to wild type (Table I). Even for a simplified reaction scheme such as Scheme 1 as follows, where K m ϭ (k Ϫ1 ϩ k 2 )/k 1 , V max ϭ k 2 , and K D ϭ k Ϫ1 /k 1 (V max measured in units of rate per ATPase molecule, so that it equals the turnover number "k cat "), it can be seen that the K m could stay rather constant when the k Ϫ1 and the k 2 change in opposite directions, as appears to be the case for the Thr-353 3 Ser and Thr-353 3 Val mutants, the K D being increased and the V max decreased, relative to wild type. Furthermore, the apparent lack of correlation of the K m with the K D could reflect an effect of Ca 2ϩ binding on the K D (cf. Ref. 33), because K D was determined in the absence of Ca 2ϩ (Fig. 3 and Table I) to avoid activation of the phosphoryl transfer reaction, whereas the determination of the K m for phosphorylation requires the presence of activating Ca 2ϩ . As further seen in Fig. 5, replacement of Mg 2ϩ with Mn 2ϩ resulted in a 5-fold decrease in the V max as well as the K m for the wild type (Fig. 5A). A similar 3.5-fold decrease in the K m was observed for mutant Thr-353 3 Val, but in this mutant, contrary to the wild type, the V max was increased 1.5-fold by the replacement of Mg 2ϩ with Mn 2ϩ (Fig. 5B). From the opposite changes in the K m and the V max , it may be deduced that in mutant Thr-353 3 Val the K D for the enzyme-substrate complex decreases upon replacement of Mg 2ϩ with Mn 2ϩ (cf. the equations given above in connection with Scheme 1). Thus, the positive effect of Mn 2ϩ on the ability of Thr-353 3 Val to phosphorylate seems to be exerted both on the binding of nucleotide and on the catalysis. For the wild type, the reduced V max demonstrates a negative effect on catalysis, and the lower K m could be explained solely by a reduction of the rate constant(s) determining V max (k 2 in Scheme 1). Unfortunately, direct nucleotide binding measurements similar to those described in relation to Fig. 3 and Table I could not be performed in the presence of Mn 2ϩ , because this divalent cation, like Ca 2ϩ , is able to activate the phosphoryl transfer reaction (29). However, a useful indication of changes to the K D can be obtained by calculating the ratio K m /V max (34), which hardly changes for the wild type but decreases 5.2-fold for Thr-353 3 Val upon the replacement of Mg 2ϩ with Mn 2ϩ .
Ca 2ϩ Affinity-All the studies of the phosphoenzyme formed from ATP described above were carried out at a Ca 2ϩ concentration of 100 M, i.e. 1-2 orders of magnitude higher than required for saturation of the activating sites on the wild type enzyme. To examine whether the effects of the mutations presented above could have resulted from lack of saturation of the Ca 2ϩ sites, the apparent Ca 2ϩ affinity was examined for mutants Thr-353 3 Ser, Thr-353 3 Ala, and Thr-353 3 Val by titrating the steady-state phosphorylation level as a function of the Ca 2ϩ concentration under conditions otherwise similar to those corresponding to Fig. 2B. The Ca 2ϩ concentration resulting in half-maximum phosphorylation was 0.40, 0.31, 0.71, and 0.95 M for the wild type and mutants Thr-353 3 Ser, Thr-353 3 Ala, and Thr-353 3 Val, respectively (data not shown). Although this result indicates that there could be an effect of the alanine and valine substitutions on the properties of the Ca 2ϩ sites (attesting to the existence of a long-range interaction between the catalytic site and the Ca 2ϩ sites), the change in apparent affinity is clearly too small to significantly reduce the fraction of enzyme with Ca 2ϩ bound under the experimental conditions used in the phosphorylation studies.
Dephosphorylation of the Phosphoenzyme Formed from ATP-In the wild type Ca 2ϩ -ATPase, the phosphoenzyme formed by the reaction with MgATP turns over at a rate of about 5 s Ϫ1 under the conditions corresponding to Figs. 2B and 4 (22). Hence, the finding that even the very low phosphorylation rate of 0.16 s Ϫ1 seen for mutant Thr-353 3 Val can lead to stoichiometric phosphorylation on prolonged incubation (Fig.  4B) shows that the dephosphorylation must be considerably slower in this mutant than in the wild type. To further address this issue, we investigated the dephosphorylation kinetics following chase with non-radioactive ATP of the phosphoenzyme formed by incubation with [␥-32 P]ATP (Fig. 6). Fig. 6A depicts results obtained at 0°C. In the presence of 5 mM Mg 2ϩ , the dephosphorylation rates of mutants Thr-353 3 Val, Thr-353 3 Ala, Thr-353 3 Ser, Thr-353 3 Cys, and Thr-353 3 Asp were found to be reduced ϳ35-, 28-, 9-, 2-, and 3-fold, respectively, relative to wild type. For mutants Thr-353 3 Gln and Thr-353 3 Tyr, reliable measurements could not be performed due to the low level of phosphoenzyme (cf. Fig. 2). For Thr-353 3 Val, Thr-353 3 Ala, and Thr-353 3 Ser, the dephosphorylation at 0°C was studied in the presence of 20 mM Mg 2ϩ , as well, to examine whether the dephosphorylation block is caused by lack of saturation of a site binding free Mg 2ϩ . As shown in Fig. 6A, the increase in Mg 2ϩ concentration did not increase the dephosphorylation rate of Thr-353 3 Ala, and the same was found for Thr-353 3 Val and Thr-353 3 Ser (data not shown). Fig. 6B shows the dephosphorylation at 25°C in the presence of 0.5 mM Mn 2ϩ and, for Thr-353 3 Val and Thr-353 3 Ser, also in the presence of 5 mM Mg 2ϩ . In the case of Thr-353 3 Val, the rate constant for dephosphorylation in the presence of Mg 2ϩ was only 0.02 s Ϫ1 , which should be compared with the value of about 5 s Ϫ1 for the wild type in the presence of Mg 2ϩ determined previously in rapid quench experiments (22). For Thr-353 3 Ser, the dephosphorylation rate in the presence of Mg 2ϩ was also considerably lower (0.27 s Ϫ1 ) than that of the wild type. For these mutants, there was no significant difference between the dephosphorylation rates determined in the presence of Mn 2ϩ and Mg 2ϩ , supporting the notion that the higher phosphorylation level seen for Thr-353 3 Val in Fig. 2C relative to Fig. 2B is caused by a higher rate of phosphorylation in Mn 2ϩ relative to Mg 2ϩ , rather than by a lower rate of dephosphorylation. For Thr-353 3 Cys and Thr-353 3 Asp, the results shown in Fig. 6B, like those in Fig. 6A, indicate a rate of dephosphorylation much closer to that of the wild type than seen for the other Thr-353 mutants.
ADP Sensitivity of the Phosphoenzyme Formed from ATP-In the wild type Ca 2ϩ -ATPase, the turnover of the phosphoenzyme occurs through the conversion of E 1 P to E 2 P and subsequent hydrolysis of E 2 P. E 1 P is thought to contain Ca 2ϩ in an occluded state, whereas in E 2 P the Ca 2ϩ sites face the lumen and display low affinity (Fig. 1). E 1 P and E 2 P can normally be distinguished by their different reactivities toward ADP (1). E 1 P is ADP-sensitive, i.e. able to donate the phosphoryl group back to ADP, forming ATP, whereas E 2 P is insensitive to ADP and dephosphorylates only by hydrolysis. The ADP sensitivity of the phosphoenzyme intermediate accumulated at neutral pH and in the presence of K ϩ was assessed by comparing the phosphorylation levels determined following 5-s dephosphorylation either in the presence or absence of ADP (Table II). In the wild type, the presence of ADP leads to almost complete dephosphorylation within 5 s, because the ADP-sensitive E 1 P form prevails as a consequence of the relatively high rate of hydrolysis of E 2 P. A similar high reactivity toward ADP was observed for mutants Thr-353 3 Ala, Thr-353 3 Ser, and Thr-353 3 Cys. For Thr-353 3 Ala and Thr-353 3 Ser, the high dephosphorylation rate in ADP contrasts sharply with the low rate seen in the absence of ADP. Because practically all the accumulated phosphoenzyme was ADP-sensitive, the reason for the low dephosphorylation rate observed in the absence of ADP must be a block of the E 1 P to E 2 P transformation rather than a block of the subsequent hydrolysis of E 2 P, which would have led to accumulation of E 2 P and, thus, to ADP insensitivity. For Thr-353 3 Val, the major part of the phosphoenzyme likewise disappeared during the 5-s incubation with ADP, but the remaining phosphoenzyme pool was clearly higher than for Thr-353 3 Ala, Thr-353 3 Ser, and Thr-353 3 Cys. For mu- To observe the dephosphorylation, the phosphoenzyme was chased by addition of 0.6 mM non-radioactive ATP, and acid quenching was performed at the indicated time intervals. A, the dephosphorylation was carried out at 0°C. For the wild type and mutant Thr-353 3 Ser, formation of phosphoenzyme took place at 0°C, as well, but because mutants Thr-353 3 Val, Thr-353 3 Ala, Thr-353 3 Cys, and Thr-353 3 Asp were unable to phosphorylate at 0°C (cf. Fig. 2), they were phosphorylated at 25°C followed by chilling on ice water for 30 s, before the addition of non-radioactive ATP. B, the phosphorylation as well as the dephosphorylation was carried out at 25°C. The lines show the best fits of a monoexponential decay function, giving the rate constants tant Thr-353 3 Asp, most of the phosphoenzyme appeared to be ADP-insensitive, because there was little difference between the amounts of phosphoenzyme remaining following the 5-s dephosphorylation with and without ADP (Table II). Because the rate of dephosphorylation of Thr-353 3 Asp was rather high also in the forward direction of the reaction cycle in the absence of ADP, the accumulation of ADP-insensitive phosphoenzyme in Thr-353 3 Asp cannot be explained by a low rate of hydrolysis of E 2 P. Conceivably, the insensitivity of Thr-353 3 Asp toward ADP reflects a reduced affinity of the E 1 P form for ADP, because the nucleotide affinity was generally much lower in Thr-353 3 Asp than in any of the other Thr-353 mutants (Table I). Fig. 7 shows the results of experiments in which the conversion of E 1 P to E 2 P was examined more directly for Thr-353 3 Ser and Thr-353 3 Val. Here, the phosphorylation from ATP was carried out at alkaline pH and in the absence of K ϩ . Under these conditions, the hydrolysis of E 2 P is rather slow in the wild type Ca 2ϩ -ATPase and E 2 P accumulates as the major steady-state intermediate (Fig. 7, closed circles). Because the E 2 P form is insensitive to ADP, the addition of ADP to the wild type phosphoenzyme does not lead to a rapid dephosphorylation under these conditions (Fig. 7, open circles). In contrast, the addition of ADP to mutant Thr-353 3 Ser incubated under the same conditions resulted in a rapid and almost complete dephosphorylation (Fig. 7, open squares), demonstrating that the conformational transition from E 1 P to E 2 P is severely inhibited in this mutant. Because no dephosphorylation was seen for mutant Thr-353 3 Ser when EGTA was added to terminate the phosphorylation in the absence of ADP (Fig. 7, closed  squares), the dephosphorylation observed following the addition of ADP could not have been caused by an increased rate of hydrolysis of E 2 P. For mutant Thr-353 3 Val, an intermediate reactivity toward ADP was observed (Fig. 7, open triangles) in line with the result described above for pH 7.0 in the presence of K ϩ (Table II). This shows on the one hand that the E 1 P to E 2 P transition is inhibited in Thr-353 3 Val, because a large part of the phosphoenzyme was able to react with ADP but, on the other hand, that some ADP-insensitive phosphoenzyme or phosphoenzyme with reduced ADP sensitivity accumulated. This suggests that, in addition to the inhibition of the E 1 P to E 2 P transformation, there is a block of E 2 P hydrolysis or the reactivity of the E 1 P form toward ADP is reduced, e.g. due to defective binding of ADP and/or Mg 2ϩ .
The E 2 P Phosphoenzyme of Thr-353 3 Ser-As demonstrated in Fig. 2D, the only mutant capable of undergoing backward phosphorylation from 32 P i was Thr-353 3 Ser and only to an extent of 24% that of the wild type. Titration of the phosphorylation at varying concentrations of 32 P i and 10 mM Mg 2ϩ under conditions optimal for formation of the E 2 P phosphoenzyme intermediate at equilibrium showed a 50-fold reduced apparent affinity (K 0.5 increased) for P i in mutant Thr-353 3 Ser, relative to wild type (Fig. 8). As in the case of phosphorylation from ATP, the "true" substrate in the phosphorylation from P i could be the Mg 2ϩ complex (MgP i ) (35). To investigate the role of free Mg 2ϩ , the experiments were carried out in the presence of 40 mM Mg 2ϩ as well. No significant change in the apparent affinity for P i was, however, seen as a result of this 4-fold increase in Mg 2ϩ concentration for either wild type or mutant (Fig. 8). Thus, the reduced apparent P i affinity seen for mutant Thr-353 3 Ser does not seem to be caused by a reduced affinity for free Mg 2ϩ . On the basis of Scheme 2, it can be deduced that the P i concentration giving half-maximum phosphorylation is K 0.5 ϭ K D ⅐k Ϫ2 /(k 2 ϩ k Ϫ2 ). The dissociation constant for the enzyme-MgP i complex is K D ϭ k Ϫ1 /k 1 . Because k 2 is much larger than k Ϫ2 for the wild type under the phosphorylation conditions applied here (low pH, presence of dimethyl sulfoxide, absence of alkali metal ions), the K 0.5 will be dependent on k Ϫ2 , as well as k 2 and K D . To study k Ϫ2 , the wild type and the Thr-353 3 Ser mutant were phosphorylated under optimal conditions for formation of the E 2 P phosphoenzyme, followed by dilution at 25°C into a medium of the same composition except for the absence of radioactively labeled phosphate, a reduction of the dimethyl sulfoxide concentration from 30% to 15%, and the presence of excess EDTA to remove free Mg 2ϩ and thus terminate the phosphorylation (Fig. 9A). It is seen in the figure that the rate of dephosphorylation was more than 10-fold higher in the mutant than in the wild type under these conditions. To examine whether the observed difference depended on the specific composition of the medium, the dephosphorylation was studied in two other media as well. In the experiment corresponding to Fig. 9B, the phosphorylated enzyme was diluted into a pH 7.0 medium containing Mg 2ϩ and K ϩ , but lacking dimethyl sulfoxide (usually added to promote the phosphorylation (35)). Under these conditions, the dephos-TABLE II ADP sensitivity of wild type and mutant Ca 2ϩ -ATPases Phosphorylation was carried out for 5 s at 25°C in a buffer containing 40 mM MOPS/Tris (pH 7.0), 80 mM KCl, 5 mM MgCl 2 , 100 M CaCl 2 , and 5 M [␥-32 P]ATP, followed by chilling on 0°C ice water for 30 s. Dephosphorylation was studied at 0°C by addition of either 1 mM EGTA or 1 mM EGTA with 1 mM ADP to the phosphorylated samples, followed by acid quenching 5 s later. To determine the phosphorylation level corresponding to 100%, parallel experiments were conducted where the samples were acid-quenched at the time corresponding to addition of the chase solution. phorylation is so rapid that it was necessary to lower the temperature to 0°C. Again, the dephosphorylation rate of the mutant was found substantially higher than that of the wild type (at least 4-fold, but probably more, because the rate constant for the mutant was outside the measurable range). The experiment corresponding to Fig. 9C was also conducted at pH 7.0 and 0°C in the absence of dimethyl sulfoxide, but now Mg 2ϩ and K ϩ were omitted. This reduced the dephosphorylation rates of both wild type and mutant to such an extent that an accurate value for either rate constant could be determined, revealing a 22-fold difference. Thus, it is clear that the rate of E 2 P hydrolysis is markedly increased in the mutant, relative to wild type, both at pH 6.0 and 7.0, in the presence and absence of dimethyl sulfoxide, free Mg 2ϩ , or K ϩ , and at 0°C as well as 25°C. The increased rate of E 2 P hydrolysis seems to account for most of the reduction of the apparent affinity for P i seen in Fig. 8.

DISCUSSION
The present data demonstrate that mutations to Thr-353 of the Ca 2ϩ -ATPase have severe consequences for several of the partial reaction steps in the pump cycle. All seven Thr-353 mutants investigated here showed impaired phosphorylation from MgATP. To some extent this is explained by a decrease in the binding affinity for MgATP, ranging from 6-to 10-fold in Thr-353 3 Cys, Thr-353 3 Val, and Thr-353 3 Ser to more than 1000-fold in Thr-353 3 Asp, but the kinetic studies of Thr-353 3 Ser and Thr-353 3 Val furthermore revealed a low V max for the phosphoryl transfer reaction (8-and 600-fold reduced, respectively, relative to wild type). The turnover of the phosphoenzyme formed from ATP was likewise slowed, as much as 1-2 orders of magnitude in Thr-353 3 Ser, Thr-353 3 Ala, and Thr-353 3 Val, due to a block of the E 1 P to E 2 P conformational change, whereas mutants Thr-353 3 Cys and Thr-353 3 Asp behaved more similar to the wild type in this respect. Furthermore, the ability to form a phosphoenzyme in the backward reaction with P i was impaired in the Thr-353 mutants, Thr-353 3 Ser displaying a 50-fold reduced apparent affinity for P i that could be accounted for by an abnormally high rate of hydrolysis of E 2 P, and the other Thr-353 mutants showing no phosphorylation at all from P i . As discussed below, these effects can be understood on the basis of a central posi-tion of Thr-353 in the catalytic site. Both the side-chain and the main-chain carbonyl of Thr-353 may participate in important interactions.
Possible Roles of the Side Chain of Thr-353- Fig. 10 shows the architecture of the catalytic site around the phosphorylated Asp-351 as it appears in the crystal structure of the Ca 2ϩ -ATPase in the Ca 2ϩ bound (but Mg 2ϩ free) E 1 conformation (2). In the structure of the native enzyme having MgATP bound, the side chain of Thr-353 is likely to be positioned closer to Asp-351 than the 7 Å in the crystal, because domain N must move toward domain P to bring the ␥-phosphate of ATP in position to phosphorylate Asp-351 (2). Because some of the residues in domain N known to bind ATP (25) are located more than 30 Å from the phosphorylated Asp-351, and because the length of a fully "stretched" ATP molecule is only about half FIG. 9. Dephosphorylation of phosphoenzyme formed from 32 P i in the wild type and Thr-353 3 Ser. Wild type (circles) and mutant Thr-353 3 Ser (squares) were phosphorylated at 25°C for 10 min in a medium containing 100 mM MES/Tris (pH 6.0), 2 mM EGTA, 30% (v/v) dimethyl sulfoxide, 10 mM MgCl 2 , and 0.5 mM 32 P i . A, dephosphorylation was studied at 25°C by a 19-fold dilution into a medium containing EDTA (to remove free Mg 2ϩ ) corresponding to a final concentration of 10 mM, 100 mM MES/Tris (pH 6.0), 2 mM EGTA, 15% (v/v) dimethyl sulfoxide, and 0.5 mM non-radioactive P i , followed by acid quenching at serial time intervals. B and C, the phosphorylated sample was chilled on 0°C ice water, and dephosphorylation was studied at 0°C by a 19-fold dilution into ice-cold medium containing either 50 mM MOPS/Tris (pH 7.0), 2 mM EGTA, 2 mM MgCl 2 , 100 mM KCl, and 5 mM non-radioactive P i (B) or 50 mM MOPS/Tris (pH 7.0), 2 mM EGTA, and 5 mM non-radioactive P i (C), followed by acid quenching at serial time intervals. The lines show the best fits of a monoexponential decay function, giving the rate constants indicated in parentheses: A, circles, wild type (0.08 s Ϫ1 ); squares, Thr-353 3 Ser (Ͼ1 s Ϫ1 ). B, circles, wild type (0.75 s Ϫ1 ); squares, Thr-353 3 Ser (Ͼ3 s Ϫ1 ). C, circles, wild type (0.021 s Ϫ1 ); squares, Thr-353 3 Ser (0.46 s Ϫ1 ). that distance, quite a large displacement of domain N is required. This movement implies that Asn-359 and Asp-601, which are located in the hinge region between domain P and domain N only 3-4 Å away from Thr-353, must push against Thr-353, so that the latter approaches Asp-351 at the center of the active site. In particular, the proximity of Asp-601 and its negative charge could swivel the side-chain hydroxyl group of Thr-353 so that it points directly toward Asp-351. The position of the hydroxyl would now be approximately a phosphate distance from Asp-351, and hydrogen bonding with the ␤or ␥-phosphate, or bridging oxygen, of ATP becomes feasible. Interference of the mutations with this interaction would lower the affinity for MgATP and could also destabilize the transition state, thereby explaining the low V max of the phosphoryl transfer reaction. The bulkiness of the tyrosine side chain would be expected to be disruptive, in agreement with the data in Fig. 2 and Table I. It is noteworthy that for the neutral substituents of a size not too different from that of the threonine, i.e. alanine, serine, glutamine, and valine, the affinity for ATP was rather similar to that of the wild type, despite the low affinity for MgATP. This suggests that the threonine does not interact very closely with the nucleotide in the absence of Mg 2ϩ , but does so in the presence of the cation. The magnesium ion apparently directs the phosphates toward Thr-353, and likely Asp-351 as well. The dramatic effect of the Thr-353 3 Asp mutation on the nucleotide binding affinity seen both in the presence and absence of Mg 2ϩ could be due to electrostatic effects between the negatively charged substituent, the phosphates of ATP, and the Asp-351 side chain, tending to expel the nucleotide. The finding that also mutant Thr-353 3 Cys showed very low nucleotide affinity in the absence of Mg 2ϩ most likely reflects ionization of the thiol group in this condition, thus causing Thr-353 3 Cys to behave in a manner similar to Thr-353 3 Asp. The ionized state of the thiol group might be stabilized by the positively charged side chain of Lys-352, which is within hydrogen bonding distance from Thr-353 (cf. Fig. 10).
The nature of the Thr-353 side chain is obviously very important for the turnover of the phosphoenzyme, which was much more strongly inhibited in Thr-353 3 Ser, Thr-353 3 Val, and Thr-353 3 Ala than in Thr-353 3 Asp and Thr-353 3 Cys. Because valine is isosteric with threonine, and alanine is smaller, it is unlikely that the inhibition is caused by steric hindrance, but interference with charge transfer or electrostatic potential migration around the phosphoryl group as part of the energy-transducing events could play a role. Again the very similar behaviors of Thr-353 3 Asp and Thr-353 3 Cys could imply that the cysteine thiol is ionized, now in the phosphoenzyme. In these mutants, electrostatic repulsion of the phosphoryl group or leaving ADP molecule might facilitate the conversion of E 1 P to E 2 P.
The impairment of the phosphorylation from P i seen for the Thr-353 mutants and the enhanced rate of E 2 P hydrolysis observed for Thr-353 3 Ser could reflect interference with hydrogen bonding to the phosphoryl group resulting in destabilization of E 2 P in analogy with the situation discussed above for the interaction with nucleotide. However, because no phosphoenzyme was formed from P i in any of the other mutants under conditions optimal for P i phosphorylation, the requirement for the side-chain hydroxyl group may be absolute, suggesting a direct role in the catalysis of P i phosphorylation. The hydroxyl could act as a general acid-base catalyst by protonating the leaving hydroxyl in the phosphorylation reaction and activating the water molecule attacking the phosphorous atom in the dephosphorylation reaction. A transiently protonated state of the threonine hydroxyl, -OH 2 ϩ , might be stabilized by the neighboring negatively charged Asp-601 or the phosphoryl group itself. The enhanced dephosphorylation of Thr-353 3 Ser could imply that the removal of the methyl group from the side chain facilitates the proton extraction activating the water molecule.
The catalytic site of the P-type ATPases is thought to be homologous to that of other members of the haloacid dehalogenase superfamily, such as haloacid dehalogenase, phosphoserine phosphatase, phosphonatase, and phosphomutase. Furthermore, there seems to be some homology to the more remote bacterial response regulator protein CheY (6 -9). The residue corresponding to Thr-353 is an aspartate in the phosphatases and phosphomutases, an alanine in the phosphonatases, and a tyrosine in the dehalogenases, the respective residues being highly conserved among the members of each protein family (6,9,10). The variation between the haloacid dehalogenase superfamily members at the residue corresponding to Thr-353 is likely to reflect the different substrates utilized by these enzymes. Thus, the tyrosine of the dehalogenases is believed to play a role in attracting the leaving halide ion of the haloacid (36), and the aspartate of the phosphatases and phosphomutases has been proposed to be used as general acid-base catalyst, donating the proton to the leaving hydroxyl group (9,10) in much the same way as suggested above for the threonine in P-type ATPases in the E 2 state. In the P-type ATPases, the high affinity binding of the triphosphate substrate in the E 1 state may require the presence of the neutral threonine side chain instead of the negatively charged aspartate to reduce the electrostatic repulsion, as discussed above. The bacterial response regulators show less conservation at the position corresponding to Thr-353, although there seems to be a strong preference for polar residues (37). These proteins utilize autophosphorylated histidine protein kinases as their substrates.
Possible Role of the Main-chain Carbonyl of Thr-353 in Mg 2ϩ Ligation-In the phosphoserine phosphatase as well as in CheY, the main-chain carbonyl corresponding to Thr-353 contributes to coordination of Mg 2ϩ , the other five Mg 2ϩ ligands being the side chains corresponding to Asp-351 and Asp703, phosphate, and two water molecules, one of which is positioned by the side-chain corresponding to Asp-707 (7,8,13,14). A similar coordination of Mg 2ϩ , involving the main-chain carbonyl of Thr-353, is not possible in the crystal structure of the Ca 2ϩ -ATPase shown in Fig. 10, because the distances between Mg 2ϩ and its proposed protein coordination points would be 3-4 Å, whereas the distance between Mg 2ϩ and its six coordinators in an octahedral arrangement is typically around 2.1 Å (38). However, the structure of CheY has been determined both in the presence and absence of Mg 2ϩ , and upon Mg 2ϩ binding the protein coordination points move closer together around the metal ion (13). Similar changes may occur in the Ca 2ϩ -ATPase as a result of the domain shifts discussed above. We have shown previously that the electrostatic repulsion between Asp-351 and the ␥-phosphate of ATP becomes increasingly stronger as the ligands change from ATP, to MgATP, to Ca 2ϩ plus MgATP, compatible with the ␥-phosphate coming closer to Asp-351 in each conformational state on the path to phosphorylation (33). Hence, the possibility should be considered that the main-chain carbonyl corresponding to Thr-353 contributes to Mg 2ϩ coordination also in the Ca 2ϩ -ATPase and other P-type ATPases (8). Indeed, defective Mg 2ϩ binding could be responsible for, or contribute to, some of the mutational effects observed in the present study. Mg 2ϩ is involved not only in ATP binding and catalysis of phosphorylation from ATP, but remains bound to the enzyme when ADP is released, and is needed for the conversion of E 1 P to E 2 P as well as the hydrolysis of E 2 P and the backward phosphorylation from P i (1, 27-29, 32, 35, 39 -41). In some of the mutants the phosphorylation was enhanced upon the replacement of Mg 2ϩ with Mn 2ϩ (Figs. 2, 4, and 5). For Thr-353 3 Val, we found the V max for phosphorylation 1.5-fold higher in Mn 2ϩ relative to Mg 2ϩ , whereas for the wild type, the V max decreased 5-fold upon the replacement of Mg 2ϩ with Mn 2ϩ . As judged from the decrease in the ratio K m /V max seen for Thr-353 3 Val, the substitution of Mn 2ϩ for Mg 2ϩ furthermore increased the affinity for the ATPmetal ion complex. Because of the different ionic radii of Mg 2ϩ and Mn 2ϩ (0.65 and 0.80 Å, respectively), these observations most likely reflect a change in the architecture of the metal ion binding site induced by the mutation. The side-chain substitution could lead to some displacement of the main-chain carbonyl or the phosphate interacting with the metal ion, thereby creating a larger space that allows Mn 2ϩ to interact more tightly and be positioned more favorably for the catalysis than the smaller Mg 2ϩ .
The stronger effects of some of the Thr-353 mutations on the binding of MgATP as compared with ATP would be in accordance with a role for Thr-353 in the binding of the Mg 2ϩ in the MgATP complex. It has been discussed that in addition to the site binding MgATP there may exist a separate site binding free Mg 2ϩ (41,42). The present studies seem to exclude that the major functional effects of the Thr-353 mutations are caused by a decrease in the affinity of a Mg 2ϩ site of the latter type, because neither the phosphorylation rate, the rate of the E 1 P to E 2 P conversion, nor the apparent affinity for P i was increased toward a normal value by a rise in the concentration of free Mg 2ϩ .
Our results furthermore seem to exclude a role for the sidechain hydroxyl in interaction with Mg 2ϩ , because the enhancing effect on phosphorylation of replacing Mg 2ϩ with Mn 2ϩ was equally strong for Thr-353 3 Ala, Thr-353 3 Val, and Thr-353 3 Asp as for Thr-353 3 Cys (Fig. 2, B and C). Because Mn 2ϩ forms a tighter complex with sulfur than Mg 2ϩ (30,31), one would have expected the rescuing effect of Mn 2ϩ on phosphorylation to be stronger for the Thr-353 3 Cys mutant than for any of the other mutants, if the side chain were interacting with the metal ion. Another finding supporting the notion that Mg 2ϩ does not interact with the side chain of Thr-353 is the conspicuous lowering of the affinity for MgATP observed in the Thr-353 3 Asp mutant. If the side chain of Thr-353 were involved in coordination of Mg 2ϩ , one would on the contrary have expected an increased affinity, due to the effect of the negative charge.
In conclusion, the scenario depicted by the present results is one in which the side-chain hydroxyl of Thr-353 interacts with the ␤or ␥-phosphate, or bridging oxygen, of ATP in the E 1 form of the enzyme and with the phosphoryl group in E 2 P, possibly being directly involved in some of the catalytic events, whereas the main-chain carbonyl of Thr-353 may participate in the binding of Mg 2ϩ in MgATP.