Short and Long Range Functions of Amino Acids in the Transmembrane Region of the Sarcoplasmic Reticulum ATPase A MUTATIONAL STUDY*

Mutational analysis of several amino acids in the transmembrane region of the sarcoplasmic reticulum ATPase was performed by expressing wild type ATPase and 32 site-directed mutants in COS-1 cells followed by functional characterization of the microsomal fraction. Four different phenotype characteristics were observed in the mutants: ( a ) functions similar to those sustained by the wild type ATPase; ( b ) Ca 2 (cid:49) transport inhibited to a greater extent than ATPase hydrolytic activity; ( c ) inhibition of transport and hydrolytic activity in the presence of high levels of phosphorylated enzyme inter- mediate; and ( d ) total inhibition of ATP utilization by the enzyme while retaining the ability to form phos- phoenzyme by utilization of P i . Analysis of experimental observations and molecular models revealed short and long range functions of several amino acids within the transmembrane region. Short range functions include: ( a ) direct involvement of five amino acids in Ca 2 (cid:49) binding within a channel formed by clustered transmem- brane helices M4, M5, M6, and M8; ( b ) roles of several amino acids in structural stabilization of the helical cluster for optimal channel function; and ( c ) a specific role of Lys 297 in sealing the distal end of the channel, suggesting that the M4 helix rotates to allow vectorial flux of Ca 2 (cid:49) upon enzyme phosphorylation. Long range functions are related to the influence of several trans- membrane amino acids on phosphorylation reactions with ATP or P i , transmitted to the extramembranous region of the ATPase in the presence or in the absence of Ca 2 (cid:49) . incubated at 37 °C with 50 (cid:109) M 32 P i as described above and then adding with rapid mixing 1.0 ml of 1.0 m M (ice-cold) nonradioactive P i . The reaction was then quenched at serial times by the addition of 0.11 ml of 10.0 M perchloric acid. A zero time sample was obtained by quenching before the addition of nonradioactive P i . Addition of carrier protein, centrifugations, washings, elec- trophoresis, and autoradiography were carried out as described above. Functional consequences of the six mutations originally reported by Clarke et al. (1), who suggested participation of the corresponding residues in Ca 2 binding The original characterization is extended to evaluate the equilibrium levels of phosphoenzyme obtained by utilization of P i , the Ca 2 concentrations required to produce a half-maximal inhibition of phosphorylation under the original conditions, and the phosphoenzyme decay rates. WT, wild type.

gested that six residues originating from four transmembrane helices (M4, Glu 309 ; M5, Glu 771 ; M6, Asn 796 , Thr 799 , and Asp 800 ; M8, Glu 908 ) are involved in Ca 2ϩ binding (Fig. 1). We have now performed a mutational analysis of several amino acids in the transmembrane region to evaluate in detail their roles in Ca 2ϩ binding and, more generally, in the catalytic and transport functions of the Ca 2ϩ ATPase.

EXPERIMENTAL PROCEDURES
Oligonucleotide-directed Mutagenesis and cDNA Expression in COS-1 Cells-The chicken fast muscle SR ATPase cDNA (3) was inserted into the pUC19 plasmid for amplification and then subcloned into the pSELECT-1 vector for site-directed mutagenesis by the Altered Sites in Vitro Mutagenesis System (Promega; Madison, WI) or by overlap extension using the polymerase chain reaction (4). Eleven unique restriction sites were introduced in the cDNA to allow cassette exchange, and a c-myc tag was added to the 3Ј terminus to monitor ATPase expression as described by Zhang et al. (5). The cDNA constructs were finally transferred into the pCDL-SR␣296 plasmid (6) for transfection of COS-1 as described by Sumbilla et al. (7).
Microsome Preparation and Immunodetection of Expressed Protein-Four days after transfection the confluent COS-1 cells from 20 plates (150 ϫ 25 mm; 4 ϫ 10 7 cells/plate) were rinsed twice with cold phosphate-buffered saline, scraped with a Teflon spatula, and collected in 80 ml of phosphate-buffered saline. The microsomal fraction was then obtained as described by Zhang et al. (5).
The expressed SERCA-1 ATPase was detected by Western blotting, using monoclonal antibody 9E10 to the c-myc epitope (8) and, in parallel, a monoclonal antibody to the chicken Ca-ATPase, CaF3-5C3 (3). In addition to Western blots, quantitation of SERCA-1 expression was also obtained by enzyme-linked immunosorbent assays in microtiter plates.
Ca 2ϩ Transport and ATPase Activity-ATP-dependent Ca 2ϩ transport was measured by following the accumulation of radioactive calcium tracer in microsomal vesicles separated from the reaction mixture by filtration (0.45-m Millipore filters), as described by Zhang et al. (5).
ATPase activity was measured by determination of P i (Lanzetta et al. (9), also as described by Zhang et al. (5)). The Ca 2ϩ -dependent activity was calculated by subtracting the Ca 2ϩ -independent ATPase from the total ATPase. Both Ca 2ϩ uptake and Ca 2ϩ -dependent ATPase activity were corrected to account for the level of expressed protein in each microsomal preparation as revealed by immunoreactivity and with reference to microsomes obtained from COS-1 cells transfected with wild type cDNA.
Enzyme Phosphorylation with ATP-Steady-state levels of phosphorylated enzyme intermediate were obtained by adding 0.1 ml of 10 M [␥-32 P]ATP to 0.4 ml of a reaction mixture containing 20 mM MOPS, pH 7.0, 80 mM KCl, 5 mM MgCl 2 , 0.1 mM CaCl 2 , and 30 -100 g of microsomal protein (the amount of microsomal protein was varied to approximate the amount of expressed ATPase contained by wild type controls, as revealed by immunoreactivity). The components of the reaction mixture were preincubated in ice, and vortex mixing was carried out in the cold room. The reaction was quenched after 10 s by the addition (vortex mixing) of 0.105 ml of 10.0 M perchloric acid. The suspension was transferred immediately onto an Eppendorf tube containing 100 g of carrier protein and allowed to sit in ice for 5-10 min. After centrifugation at 5,000 rpm for 10 min, the sedimented protein was washed twice with 0.125 M perchloric acid and once with water. An aliquot of the solubilized sample was then subjected to gel electrophoresis (10) at pH 6.3, and the radioactive phosphoenzyme was detected by autoradiography.
The rates of phosphoenzyme decay were determined by first obtaining steady-state levels of phosphoenzyme as explained above. Ten s after the addition of radioactive ATP, 0.5 ml of 1.0 mM nonradioactive ATP was added with rapid mixing, and samples were acid quenched at serial times. A zero time base line was obtained by acid quenching before the chase. The entire procedure was carried out in ice and in the cold room. Washings, electrophoresis, and autoradiography were performed as described above.
Enzyme Phosphorylation with P i -Equilibrium levels of phosphorylated enzyme intermediate were obtained by adding 50 g of microsomal protein (or an amount adjusted to match the ATPase content of the reference wild type preparation) to 0.2 ml of a medium containing 50 mM MES-Tris, pH 6.2, 10 mM MgCl 2 , 20% (v:v) dimethyl sulfoxide, and 2 mM EGTA-Tris (and/or various concentrations of CaCl 2 according to experimental schedule). The reaction was started by adding 10 l of 1 mM 32 P i (10 7 dpm/nmol) to reach a 50 M final concentration. After a 10-min incubation at 37°C, the reaction was quenched by the addition of 0.11 ml of 10.0 M perchloric acid (ice cold). The suspension was transferred onto an Eppendorf tube containing 100 g of carrier protein. After a 10-min incubation in ice, the denatured protein was sedimented by centrifugation (5,000 rpm for 10 min) and washed four times with 1.0 ml of 0.125 M perchloric acid and once with water. The final sediment was dissolved and subjected to electrophoresis and autoradiography as explained above.
The rates of phosphoenzyme decay were determined by first obtaining equilibrium levels of phosphoenzyme incubated at 37°C with 50 M 32 P i as described above and then adding with rapid mixing 1.0 ml of 1.0 mM (ice-cold) nonradioactive P i . The reaction was then quenched at serial times by the addition of 0.11 ml of 10.0 M perchloric acid. A zero time sample was obtained by quenching before the addition of nonradioactive P i . Addition of carrier protein, centrifugations, washings, electrophoresis, and autoradiography were carried out as described above.
Molecular Modeling-Quanta and Hyperchem molecular graphics programs were used for molecular modeling. The valence mapping program was kindly supplied by Prof. E. Di Cera, Washington University, St. Louis.

ATPase Expression in COS-1 Cells-
The studies reported here were performed with wild type and mutated cDNAs encoding the SR ATPase and the mutants listed in Tables I-III. After disruption of the transfected cells and fractionation, the expressed ATPase is recovered with microsomal vesicles and can be detected by Western blots and enzyme-linked immunosorbent assay titrations. The wild type ATPase and the mutants studied in these experiments were expressed at similar, but not identical levels (Fig. 2). Therefore, the relative level of expressed ATPase in each preparation, with reference to a selected wild type ATPase preparation, was used to obtain corrected values for the functional activity of 1 mg of microso-  (Tables I-III).
A special case was the lack of expression of the Lys 297 3 Gly mutant (see Fig. 2), which we failed to recover with the microsomal fraction of COS-1 cells after repeated transfections with two batches of plasmid amplified, banded, and sequenced at different times. Northern blot analysis confirmed the presence of an mRNA transcript for this mutant consistent with that of wild type (results not shown). It is of interest that mutation of the same residue to Phe did yield expression, although at low level, and the Lys 297 3 Met, Lys 297 3 Arg, and Lys 297 3 Glu mutants were expressed at normal levels ( Fig. 2).
Ca 2ϩ Uptake and ATPase Activity-It is shown in Fig. 3 and Table I that several mutants within the M4, M5, M6, and M8 helices sustain levels of Ca 2ϩ transport and ATPase activity which are comparable to those of the wild type ATPase. Therefore, it is possible to introduce point mutations within the transmembrane region without interfering with function. These findings confer a character of specificity to several other mutations that do interfere with function. In fact, not all mutants retaining high ATPase hydrolytic activity retained a correspondingly high Ca 2ϩ uptake. Among the mutants retaining high hydrolytic activity, this differential effect is pronounced in Val 314 3 Ala, Ile 315 3 Ala, Trp 794 3 Ala, Val 795 3 Ala, and Leu 802 3 Ala mutants (Table I).
A group of mutations producing inhibition of ATPase hydrolytic activity is shown in Table II. It is of interest that in all of these mutations, Ca 2ϩ transport was inhibited significantly more than hydrolytic activity. Since this uncoupling effect is produced by mutations which do (Table II) or do not (Table I) inhibit hydrolytic activity, it is apparent that a specific structural perturbation is involved in its onset, independent of other perturbations that may affect catalytic activity.
Characterization of the six mutations originally reported by Clarke et al. (1), who proposed that the six native residues participate in Ca 2ϩ binding, is shown in Table III. We confirmed that these mutations produce total inhibition of Ca 2ϩ transport and hydrolytic activity, and then we extended the functional characterization as explained below.
It is noteworthy that the effects of mutations are very specific and dependent both on the residues that are mutated and the side chains that are introduced. An example of the high specificity of mutational effects is found in the total inhibition resulting from the Glu 908 3 Ala mutation compared with the full activity retained by the Glu 908 3 Gln mutant (Fig. 3). Of great interest are also the different and specific effects produced by mutating Lys 297 to Gly (no expression), to Phe (low expression and low activity), to Met (low activity), and to Arg or Glu (nearly normal ATPase and slight inhibition of Ca 2ϩ uptake).

FIG. 3. Examples of ATP-dependent Ca 2؉ uptake and ATPase activity by microsomal vesicles obtained from transfected COS-1 cells.
The two upper panels show Ca 2ϩ uptake; the two lower panels show Ca 2ϩ -dependent ATPase activity. Note that in some mutants (e.g. Val 314 3 Ala) Ca 2ϩ uptake undergoes a relatively greater inhibition than ATPase activity. The reaction mixture for Ca 2ϩ uptake contained 20 mM MOPS, pH 7.0, 80 mM KCl, 5 mM MgCl 2 , 0.2 mM CaCl 2 , 0.26 mM EGTA to yield 1.4 M free Ca 2ϩ (11), 5-10 g of microsomal protein/ml, 5 mM potassium oxalate, and 3 mM ATP. The reaction was started (30°C) by the addition of oxalate and ATP and was terminated at sequential times by vacuum filtration. The reaction mixture for ATPase activity contained 20 mM MOPS, pH 7.0, 80 mM KCl, 3 mM MgCl 2 , 0.2 mM EGTA, 0.2 mM CaCl 2 , 5 mM azide, 30 g of microsomal protein/ml, 3 M ionophore A23187, and 3 mM ATP. Ca 2ϩ -independent ATPase activity was assayed in the presence of 2 mM EGTA and no added Ca 2ϩ . The reaction was started (37°C) by the addition of ATP, and samples were taken at serial times for P i determination. WT, wild type.

TABLE I
Mutants exhibiting no significant reduction of Ca 2ϩ -dependent ATPase activity and normal or reduced Ca 2ϩ uptake Ca 2ϩ uptake and ATPase activity are given in percentage of the wild type (WT) values which were 38.4 Ϯ 9.1 (20) and 134.2 Ϯ 39.3 (23) nmol/mg of microsomal protein/min at 30 and 37°C, respectively (in parentheses are the numbers of determinations in two to four microsomal preparations obtained from different transfections). The approximate levels of phosphoenzyme obtained by utilization of ATP in the presence of Ca 2ϩ , or of P i in the absence of Ca 2ϩ , as well as their relative decay rates are indicated. Phosphoenzyme formation by utilization of P i was inhibited by 10 M Ca 2ϩ in all mutants. The pertinent transmembrane helix (M) and the orientations of side chains with respect to the lumen, interfacing sides, or exterior of the helical cluster (as revealed by molecular modeling) are also indicated.  Fig. 4). The phosphoenzyme levels obtained with all other mutants were within the range obtained with wild type enzyme. Minor variations of phosphoenzyme levels on the gels were not related meaningfully to the ATPase activities of the corresponding mutants ( Fig. 5 and Tables I and II). We then conducted isotopic chase experiments to evaluate the effect of mutations on the decay of the phosphorylated enzyme intermediate (reactions 3 and 4 in Fig. 4). We found   (20) and 134.2 Ϯ 39.3 (23) nmol/mg of microsomal protein/min at 30 and 37°C, respectively (in parentheses are the numbers of determinations, in two to four microsomal preparations, obtained from different transfections). The approximate levels of phosphoenzyme obtained by utilization of ATP in the presence of Ca 2ϩ , or of P i in the absence of Ca 2ϩ , as well as their relative decay rates are indicated. Phosphoenzyme formation by utilization of P i was inhibited by 10 M Ca 2ϩ in all these mutants. The pertinent transmembrane helix (M) and the orientations of side chains with respect to the lumen, interfacing sides, or exterior of the helical cluster (as revealed by molecular modeling) are also indicated.  that mutants sustaining ATPase activity comparable to that of the wild type enzyme exhibited phosphoenzyme turnover that was similar to that of the wild type enzyme. On the other hand, the rate of phosphoenzyme decay was reduced in mutants exhibiting inhibition of steady-state ATPase activity (e.g. see Lys 297 3 Met, Cys 774 3 Ala, and Ile 775 3 Ala in Fig. 6 and Table II).
Observations of mutational effects on phosphoenzyme turnover are very important for two reasons. (a) They confirm the inhibitory effects of mutations under conditions that are not dependent on the enzyme concentration (since phosphoenzyme decay is a first order phenomenon); this dispels doubts on whether observed inhibitions of steady-state ATPase activity (which is dependent on enzyme concentration) may in fact be related to inaccuracies in the immunological determination of expressed ATPase. (b) They indicate that the mutational effect is on ATPase partial reactions that follow formation of the phosphorylated intermediate (step 3 and/or step 4 in Fig. 4), and these reactions are rate-limiting for completion of the catalytic and transport cycle.
Formation of Phosphoenzyme by Utilization of P i -In addition to enzyme phosphorylation by utilization of ATP in the presence of Ca 2ϩ , phosphoenzyme can be formed in the reverse direction of the ATPase cycle by utilization of P i in the absence of Ca 2ϩ (reverse of reaction 4 in Fig. 4; see Ref. 13). We studied this reaction by equilibrating the enzyme with P i in the absence of Ca 2ϩ , at mildly acid pH, and in the presence of 20% (v/v) dimethyl sulfoxide to facilitate phosphorylation, as was also done in the experiments by Clarke et al. (1).
When we used samples of wild type enzyme reacted separately with P i under these conditions, we obtained identical phosphoenzyme levels, indicating that our technique of detection was highly reproducible (Fig. 7). On the other hand, the equilibrium levels of phosphoenzyme obtained with various mutants were quite different. These differences were not related to whether the mutants exhibited or did not exhibit inhibition of steady-state ATPase activity and/or enzyme phosphorylation with ATP. In some cases ( Fig. 7 and Tables I-III), high levels of phosphoenzyme were obtained through the P i reaction with mutants yielding low ATPase activity (Val 300 3 Ala, Cys 774 3 Ala, Ile 775 3 Ala), or no ATPase activity and no phosphoenzyme in the presence of ATP and Ca 2ϩ (Glu 309 3 Gln, Glu 771 3 Gln). On the other hand, hardly detectable levels of phosphoenzyme were obtained through the P i reaction with some mutants (Val 314 3 Ala and Leu 319 3 Ala, for instance) retaining relatively high ATPase activity and enzyme phosphorylation with ATP ( Fig. 7 and Table I).
As opposed to the steady-state levels (resulting from all four steps of the diagram in Fig. 4) obtained in the presence of ATP and Ca 2ϩ , the P i reaction (reverse of step 4 of the diagram in Fig. 4) may be considered to occur as where E is the enzyme equilibrated with P i . We then explored the possibility that the observed variations of E-P may be due to altered affinity of the mutants for P i . We found, however, that the P i concentration used in our experiments was saturating in all cases (not shown). Therefore, the observed variations of phosphoenzyme levels were likely due to mutational effects  on the E⅐P i 7 E-P equilibrium constant, which is dependent on the ratio of the two bidirectional rate constants. In fact when we followed the phosphoenzyme decay after isotopic chase, we found that high levels of phosphoenzyme were in most cases accompanied by slow decays, whereas low levels of phosphoenzyme were accompanied by fast decays (Fig. 8 and Tables I-III).
Ca 2ϩ Inhibition of Enzyme Phosphorylation with P i -The six mutants Glu 309 3 Gln, Glu 771 3 Gln, Asn 796 3 Ala, Thr 799 3 Ala, Asp 800 3 Asn, and Glu 908 3 Ala display total inhibition of Ca 2ϩ uptake, ATPase activity, and phosphoenzyme formation in the presence of ATP and Ca 2ϩ while still able to be phosphorylated with P i (Table III). Occurrence of phosphorylation with P i in the presence of 0.1 mM Ca 2ϩ was taken as an indication of mutational interference with Ca 2ϩ binding (1).
We extended the functional characterization of these mutants by measuring the levels of EP formed with P i in the presence of various concentrations of Ca 2ϩ . We performed these titrations at mildly acid pH to lower the affinity of the enzyme for Ca 2ϩ and to maximize any effect of mutational perturbations on Ca 2ϩ binding.
Under these experimental conditions, half-maximal inhibition of the P i reaction with wild type enzyme and several mutants is obtained at approximately 20 M Ca 2ϩ . However, the same parameter is shifted to the mM range when the Glu 309 3 Gln, Glu 771 3 Gln, Thr 799 3 Ala, Asp 800 3 Asn, and Glu 908 3 Ala mutants are used (Fig. 9, also Refs. 14 -16). These experiments are quite accurate and easy to interpret as they reflect equilibration of Ca 2ϩ and P i with the enzyme (as opposed to studies on the Ca 2ϩ concentration dependence of ATP utilization, which depends on several kinetic constants). Our titration experiments (Fig. 9) show that single mutations do not eliminate completely the effect of Ca 2ϩ but rather increase by 2-3 orders of magnitude the effective Ca 2ϩ concentration (i.e. reduce the affinity of the enzyme for Ca 2ϩ ). They also demonstrate unambiguously that the Asn 796 3 Ala mutation does not interfere with Ca 2ϩ inhibition of enzyme phosphorylation by P i , even though it does interfere with Ca 2ϩ -dependent enzyme phosphorylation with ATP. We interpret our present findings to indicate that Asn 796 does not participate in Ca 2ϩ binding under our conditions (as well as those of Clarke et al. (1)) for enzyme equilibration with P i . It is possible that Asn 796 participates in Ca 2ϩ binding under different conditions and/or in another specific conformational state of the enzyme (e.g. Ca 2ϩ occlusion in the presence of CrATP (17)).
It should be pointed out that Andersen and Vilsen (16) re-ported that the Ca 2ϩ sensitivity of the Glu 309 3 Gln mutant is in the 10 M range as opposed to the mM range observed in our experiments (Fig. 9). It is likely that this difference is due to the higher pH used by Andersen and Vilsen (16), favoring ionization of Ca 2ϩ binding acidic functions with pK near neutrality (18), thereby increasing the affinity of the enzyme for Ca 2ϩ and obscuring the effect of single mutations. Finally, we note that mutation of Glu 908 3 Ala interferes with inhibition of the P i reaction by Ca 2ϩ , whereas mutation of Glu 3 Gln leaves the enzyme perfectly functional (also noted by Clarke et al. (1)). This is of specific interest, considering that mutation Glu 309 3 Gln, Glu 771 3 Gln, or Asp 800 3 Asn interferes strongly with inhibition of the P i reaction by Ca 2ϩ (Tables  I and III and Fig. 9). The different effect of the Glu 908 mutations to Gln or Ala suggests that although the Glu 309 , Glu 771 , and Asp 800 contributions to Ca 2ϩ complexation depend on both side chain oxygens, Glu 908 contributes only one side chain oxygen, which is still present following mutation to Gln, but not to Ala. As the acidic function is lost by the Gln mutant, we conclude that the carbonyl oxygen is able to participate in coordination of Ca 2ϩ .

Critical Evaluation of a Putative Ca 2ϩ Binding Domain by
Molecular Modeling-We found it helpful to evaluate by molecular modeling whether the residues pointed out by mutational analysis can in fact generate a Ca 2ϩ binding domain. Accordingly, a model was constructed by clustering transmembrane helices M4, M5, M6, and M8 and thereby forming a channel that can admit two Ca 2ϩ in single file (19). Rotation of the helices for optimal positioning of acidic side chains within the lumen of the channel is favored by the amphiphilic character of the helices.
An important question is whether it is possible to have two closely spaced calcium ions bound within the same domain (in spite of possible charge repulsion). Crystallographic resolution of several Ca 2ϩ binding structures (for instance, the duplex Ca 2ϩ binding site of thermolysin; Ref. 20) indicates that two Ca 2ϩ can in fact reside in close proximity and yields information regarding the appropriate distances between binding oxygens and Ca 2ϩ and between the two bound Ca 2ϩ .
In previous attempts to model all six residues suggested by Clarke et al. (1), we encountered difficulty in orienting the diverging side chains of the Asn 796 , Thr 799 , and Asp 800 (which originate from the same helix M6) for simultaneous participation in Ca 2ϩ binding. For this reason we proposed exclusion of Thr 799 (2). Our present experimental observations, however, indicate clearly that Asn 796 is the residue to be excluded; and this solves the problem. However, we do not exclude that Asn 796 may participate in Ca 2ϩ binding by the enzyme in some specific conformational state that we do not see under our conditions. Andersen and Vilsen suggested (16) that Asn 796 may participate in complexation of only the proximal (closer to the cytosolic side of the membrane) Ca 2ϩ . However, modeling shows that Asn 796 is actually the most distal of the putative binding residues and is unlikely to interact with the proximal Ca 2ϩ .
We considered whether the two oxygen functions of each acidic residue may converge on the same Ca 2ϩ or may be shared by the two Ca 2ϩ . Indications that the proximal Ca 2ϩ may be independently influenced by the Glu 309 3 Gln mutation (21) suggest that Glu 309 contributes both oxygen functions to the proximal Ca 2ϩ as shown in model 1 of Fig. 10. Both oxygens of Glu 771 may then be approximated to the distal Ca 2ϩ as proposed by Andersen and Vilsen (16) to provide adequate distribution of charge. Approximation of Thr 799 to the distal Ca 2ϩ then places Asp 800 in a position to share the two side FIG. 9. Inhibition of the P i reaction by Ca 2؉ . Incubations were carried out as described under "Experimental Procedures," but various aliquots of CaCl 2 were added to yield the free Ca 2ϩ concentrations (11) given in the figure. chain oxygens with the proximal and distal Ca 2ϩ , thereby contributing binding cooperativity. Finally, we place one of the Glu 908 oxygens near the proximal Ca 2ϩ since the different effects of mutation of this residue to Gln or Ala indicate that only one oxygen participates in Ca 2ϩ binding (see "Results").
Model 1 in Fig. 10 is not the only one that is consistent with the experimental results. For instance, it is possible to reposition the helices longitudinally so that Glu 309 could bind the distal Ca 2ϩ and Glu 771 the proximal Ca 2ϩ , everything else remaining the same (model 2 in Fig. 10).
To accommodate the five residues while excluding Asn 796 , we found it sterically convenient to alternate the four transmembrane helices as shown in Fig. 10. In the models the five binding residues were positioned to approximate their oxygen ligands within 2.6 Å of one or both calcium ions. Additional oxygens may be contributed by coordinated water.
Although the arrangements in Fig. 10 are speculative, the important question that they answer is whether the oxygen functions singled out by mutational analysis can generate discrete areas of high binding potential for Ca 2ϩ and if such areas are unique in the transmembrane helical cluster. To this aim, we subjected the two models to analysis of structural coordinates to obtain valence maps, using the algorithm of Nayal and Di Cera (22), which is a consistent predictor of Ca 2ϩ binding sites in proteins of known structure. Analysis of the entire model generates two distinct areas with valence points of 1.4 or higher, i.e. a high probability for binding Ca 2ϩ . The two areas are shown in Fig. 10, A-D, as cluster of dots. No other site of high valence is present throughout the four clustered helices, consistent with the results of extensive mutational work on the four helices (23,24).
Short Range Functions of Amino Acid Side Chains within the Transmembrane Region-The experimental observations discussed above indicate that five amino acids in the transmembrane region sustain short range functions as they participate directly in Ca 2ϩ binding. In addition, several mutations within the four transmembrane helices M4, M5, M6, and M8 result in inhibition of Ca 2ϩ uptake with little or no inhibition of hydrolytic activity. A similar uncoupling effect was first reported by Andersen (25) regarding the mutation Tyr 763 3 Gly. We found that the effect of the Tyr 763 3 Gly mutation is not unique but is also produced by several other mutations in the transmembrane region (Tables I and II). Inspection of Tables I and II, in the light of the structural model, indicates that pronounced uncoupling is produced by mutations of amino acids with side chains exiting helices M4 or M5 from their outer and lateral surfaces, thereby interacting with the lipid bilayer and/or other components of the transmembrane cluster. Therefore, the uncoupling effects of these mutations are likely to be produced by local interference with proper packing of the helices and opti- mal flux of Ca 2ϩ during active transport. These observations provide a cogent argument for the functional role of the transmembrane channel resulting from the four clustered helices and explain the effects of various hydrophobic compounds on Ca 2ϩ fluxes through the ATPase (26).
Of special interest are the mutations of Lys 297 , including Lys 297 3 Gly, which interferes with ATPase assembly and expression, Lys 297 3 Met and Lys 297 3 Phe producing strong functional inhibition, and Lys 297 3 Arg and Lys 297 3 Glu producing little or no inhibition (Tables I and II). In the model, Lys 297 places its charge at the distal end of the M4 segment, facing the lumenal end of the channel. In fact, Lys 297 provides the only positive charge within the channel. It is apparent then that the presence of a highly polar moiety in this position is required to stabilize the channel structure. The important role of this stabilization is demonstrated by the interference with ATPase expression produced by replacement of Lys 297 with the much less restrictive Gly, which may break propagation of the M4 helix during protein assembly (27). Furthermore, the functional inhibition produced by Lys 297 replacement with Met or Phe and the lack of inhibition by Lys 297 replacement with Arg or even with Glu indicate that intrusion of a net charge (positive or negative) at the distal end of the channel seals the lumen and prevents flux of Ca 2ϩ . This suggests to us that the M4 helix may rotate (and possibly be displaced distally so that Lys 297 can reach the membrane interface with water) in synchrony with ATPase phosphorylation to allow exit of transported Ca 2ϩ into the lumen.
Long Range Functional Linkages of Amino Acids within or near the Ca 2ϩ Binding Domain-In previous studies (1,14,15,28,29) it was shown that structural perturbations in the transmembrane region interfere with Ca 2ϩ binding and prevent Ca 2ϩ activation of the enzyme reaction with ATP, as well as Ca 2ϩ inhibition of the enzyme reaction with P i . These findings suggested a long range functional linkage between the Ca 2ϩ binding site within the transmembrane region and the catalytic site in the extramembranous region (30). We now find the following.
(a) Mutations of residues that are normally involved in Ca 2ϩ binding affect the catalytic site (i.e. the P i reaction) even in the absence of Ca 2ϩ . In fact, mutations Glu 309 3 Gln, Glu 771 3 Gln, and Asp 800 3 Asn increase the equilibrium level of the phosphoenzyme obtained with P i (in the absence of Ca 2ϩ ) through a drastic reduction of its breakdown kinetics ( Fig. 8 and Table III). An analogous finding was reported by Andersen (25) regarding mutation of Glu 771 to Ala or Gly. Therefore, specific mutations in the transmembrane region produce long range conformational effects favoring phosphorylation of the catalytic site with P i . This suggests that the acidic functions of these residues sustain a very important structural role possibly through hydrogen bonding with neighboring residues in the absence of Ca 2ϩ , in addition to participating in cooperative coordination in the presence of Ca 2ϩ . Note that Glu 908 does not sustain this function since its carbonyl oxygen, but not its carboxyl function, is required for Ca 2ϩ coordination.
(b) Mutations of neighboring residues that are not involved in Ca 2ϩ binding interfere with catalytic activity both in the presence and in the absence of Ca 2ϩ . A specific case, in this regard, is Asn 796 whose mutation to Ala produced total inhibition of Ca 2ϩ uptake, ATPase activity, and phosphoenzyme formation by utilization of ATP in the presence of Ca 2ϩ , as well as slow decay of the phosphoenzyme obtained by utilization of P i in the absence of Ca 2ϩ (Table III and Fig. 8). It is then apparent that Asn 796 sustains a very important role in stabilization of the native enzyme conformation, possibly through involvement of its side chain oxygen or amino group in hydrogen bonding.
In several other mutants (Table II), an inhibitory effect is primarily manifested with a slow decay of the phosphorylated intermediate formed by utilization of ATP in the presence of Ca 2ϩ . Furthermore, in the absence of Ca 2ϩ , the inhibition manifests itself with a slow decay of the phosphoenzyme formed by utilization of P i (e.g. Val 300 3 Ala, Cys 774 3 Ala, Ile 775 3 Ala; Table II).
These experimental observations demonstrate that the long range linkage between the Ca 2ϩ binding domain and the catalytic site is not necessarily dependent on Ca 2ϩ but is rather an intrinsic feature of the protein structure. It is clear that both the energy transduction mechanism and kinetic regulation of the Ca 2ϩ ATPase are not only related to the organic chemistry of the catalytic site, but are strongly dependent on extended features of protein conformation.