Deletions in the Acidic Lipid-binding Region of the Plasma Membrane Ca2+ Pump

The C-terminal segment of the loop between transmembrane helices 2 and 3 (AL region) of the plasma membrane Ca2+ pump (PMCA) is not conserved in other P-ATPases. Part of this region, just upstream from the third transmembrane domain, has been associated with activation of the PMCA by acidic lipids. cDNAs coding for mutants of the Ca2+pump isoform h4xb with deletions in the AL region were constructed, and the proteins were successfully expressed in either COS or Chinese hamster ovary cells. Mutants with deletions in the segment 296–349 had full Ca2+ transport activity, but deletions involving the segment of amino acids 350–356 were inactive suggesting that these residues are required for a functional PMCA. In the absence of calmodulin the V max of mutant d296–349 was similar to that of the recombinant wild type pump, but its K 0.5 for Ca2+ was about 5-fold lower. The addition of calmodulin increased theV max and the apparent Ca2+ affinity of both the wild type and d296–349 enzymes indicating that the activating effects of calmodulin were not affected by the deletion. At low concentrations of Ca2+ and in the presence of saturating amounts of calmodulin, the addition of phosphatidic acid increased about 2-fold the activity of the recombinant wild type pump. In contrast, under these conditions phosphatidic acid did not significantly change the activity of mutant d296–349. Taken together these results suggest that (a) deletion of residues 296–349 recreates a form of PMCA similar to that resulting from the binding of acidic lipids at the AL region; (b) the AL region acts as an acidic lipid-binding inhibitory domain capable of adjusting the Ca2+ affinity of the PMCA to the lipid composition of the membrane; and (c) the function of the AL region is independent of the autoinhibition by the C-terminal calmodulin-binding region.

The plasma membrane Ca 2ϩ pump (PMCA) 1 extrudes Ca 2ϩ from the cytosol to the extracellular space playing an important role in the maintenance of the resting level of intracellular Ca 2ϩ and in the control of the Ca 2ϩ transients. The overall topology of the PMCA is similar to that of other P-type ion motive ATPases with about 10 transmembrane segments (M1-M10) and most of the protein exposed to the cytosol (1). Four genes for the PMCA have been identified in humans, and each of these genes produces additional isoforms by alternative splicing of primary transcripts (2).
The molecular mass of the PMCA (ϳ135 kDa) is slightly higher than that of other P-ATPases, and this is mainly due to the extended C-terminal region (C region) and also to the insertion of a highly charged segment of about 60 amino acids in the loop between M2 and M3. In this report we call the latter the A L region.
The C region includes a calmodulin-binding autoinhibitory domain, and the binding of calmodulin enhances both the Ca 2ϩ sensitivity and the turnover of the pump. Alternative splicing of mRNA modifies region C leading to isoforms with different responsiveness toward calmodulin activation (3). The removal by limited proteolysis or deletion mutagenesis of the C-terminal 120 amino acids including the calmodulin-binding autoinhibitory domain suffices for a calmodulin-like activation (4 -7).
It has long been known that the PMCA is also activated by acidic lipids (5, 6, 8 -11), and this effect has been in part accounted for by the finding that these lipids interact with the calmodulin-binding site (12,13). However, the fact that acidic lipids are capable of enhancing the Ca 2ϩ sensitivity of the PMCA to a greater extent than calmodulin suggested the existence of an independent acidic lipid-responsive region (5,6). Studies of controlled proteolysis with trypsin of the PMCA from human erythrocytes (mostly isoform h4xb) have shown that the activating effects of acidic lipids is lost concomitantly with the appearance of a 76-kDa peptide lacking the N-terminal portion of the molecule and the C-terminal regulatory region (5,6). Because the proteolytic fragment of 76 kDa and its precursor, a lipid-responsive peptide of 81 kDa, have identical C termini, it was inferred that the portion of the A L region cleaved during the 81-76-kDa conversion (amino acids 315-358) is involved in lipid activation (14). In addition, it was later shown that synthetic peptides made with the sequence of segment 339 -360 specifically bind acidic lipids (3,13). As with the C region, alternative splicing at the so-called site "A" can generate different versions of the A L region. These changes were proposed to affect the sensibility of the PMCA toward acidic lipids (15,16). However, no functional consequences of the changes in this region have been reported.
In this study, we have investigated the effects of deletions in the segment 296 -356 of the human PMCA4xb. This segment encompasses the variable region between M2 and M3 (A L region) close to the site of tryptic cleavage associated with the acidic lipid activation (Fig. 1). The mutant proteins were successfully expressed either transiently in COS-1 cells or in a stable form in CHO cells. Contrary to our expectations, the recombinant PMCAs with deletions involving residues 350 -356 were completely inactive suggesting that, in addition to its proposed regulatory role, this segment is also necessary for the biogenesis of a functional enzyme. In contrast, deletion of residues 296 -349 resulted in a fully active pump with a high affinity for Ca 2ϩ characteristic of the activated form of the PMCA that would result from binding of acidic lipids at the lipid-binding site of region A L .

EXPERIMENTAL PROCEDURES
Materials-Reagents were purchased from the following companies: enzymes used in DNA manipulations, New England Biolabs; oligonucleotide primers, DNAgency, Malvern, PA; columns for DNA purification, Qiagen; 45 Ca and [␥-32 P]ATP, PerkinElmer Life Sciences; Immobilon transfer membranes and nitrocellulose filters, Millipore; immunochemicals, Vector Laboratories and Amersham Biosciences; and reagents for cell culture, thapsigargin, phosphatidic acid, and other chemicals, Sigma. The expression vector pED was a generous gift of Dr. Randall J. Kaufman, Genetics Institute, Boston.
Primer A1 has a SalI site incorporated at its 5Ј position, whereas primer 5EGE-3EGE contains a restriction site for nuclease MluI. The PCR products were digested either with SalI and MluI or MluI and BspEI (internal site naturally occurring in the wild type h4xb DNA at position 1911) to produce cohesive fragments coding for residues 1-299 and 357-719, respectively, and cloned into the corresponding position of pED-h4xb or pED-ct120. Mutant d296 -349 was obtained using oligonucleotides A1 and 887 5Ј-cttcttacgcgtcccttcgtcatcctcattgacc-3Ј. A similar strategy was used to obtain the other mutants. The nucleotide sequence of the mutant cDNAs extending between the SalI and BspEI sites was verified by dideoxy chain termination sequencing.
Protein Expression and Isolation of Cellular Membranes-COS-1 or CHO(dhfrϪ) cells were lipofected using either the ESCORT transfection reagent (Sigma) or polyfect (Qiagen) according to the manufacturer's protocol. COS-1 cells were harvested 48 h after transfection for the preparation of microsomes. Stable CHO cell lines expressing the recom-binant wild type h4xb was described previously (19). To express in a stable form the d300 -356 and d296 -349 proteins, the transfected CHO cells were split into dishes of 15 cm in diameter 48 h post-transfection and cultured in a selective Eagle's ␣-minimum essential medium without ribonucleosides and deoxyribonucleosides, supplemented with antibiotics and 10% of dialyzed fetal calf serum. After 3 weeks about 3-5 of the resulting colonies were cloned and expanded, and the expression of the pump was investigated by immunoblotting. The crude microsomal membrane fractions were prepared by the procedure of Enyedi et al. (7). Protein concentration was estimated by means of the Bio-Rad protein assay, with bovine serum albumin as a standard.
Detection of the Ca 2ϩ Pump Protein-SDS-PAGE and immunoblotting were carried out as described previously (20). Proteins were electrophoresed on a 7.5% acrylamide gel according to Laemmli (21) and subsequently transferred to Millipore Immobilon membranes. The membranes were incubated at 37°C for 1 h with 5F10 monoclonal antibody (22,23) from ascites fluid (dilution, 1:2000). For staining, biotinylated anti-mouse immunoglobulin G and avidin-streptoavidin peroxidase conjugate were used.
Ca 2ϩ Transport Assay-Ca 2ϩ uptake assays were performed as described previously (19). The reaction mixture contained 100 mM KCl, 50 mM Tris-HCl (pH 7.3 at 37°C), 5 mM NaN 3 , 0.1 M thapsigargin, 4 g/ml oligomycin, 20 mM sodium phosphate, 1.5 mM ATP, 95 M EGTA, 2.5 mM MgCl 2 and CaCl 2 (labeled with 45 Ca) to give the desired concentration of free Ca 2ϩ . The free concentrations of Ca 2ϩ were calculated using the program of Fabiato and Fabiato (24). Vesicles (10 g of protein) were preincubated at 37°C for 5 min, and the reaction was initiated by the addition of ATP. The reaction was finished after 5 min by filtering the samples through a 0.45-m filter. The 45 Ca taken up by the vesicles was determined by counting in a scintillation counter. Uptake activities were expressed per mg of membrane protein. For each data point the activity of the recombinant PMCA was estimated by subtracting the activity of the endogenous PMCA (V e ) from the overall transport activity. V e was obtained as V c ϫ (I e /I c ), where V c is the rate of Ca 2ϩ uptake by microsomal vesicles made from cells transfected with vector without insert; I e is the intensity of the endogenous PMCA band in microsomes containing endogenous and recombinant enzymes as determined by immunoblotting, and I c is the intensity of the endogenous PMCA band in microsomes made from cells transfected with vector without insert. The values of V e were less than 20% of the total Ca 2ϩ uptake.
To measure the effect of PA on the activity of the recombinant PMCA, PA was dissolved in methanol:chloroform (2:1) at a concentration of 17 mg/ml. Immediately before use, the methanol:chloroform was dried off, and the lipid was resuspended in 5 mM Tris-HCl (pH 7.3 at 37°C), 160 mM KCl. The suspension was sonicated for 5-10 min until clear, and an aliquot was added to the reaction media. The microsomes were preincubated at 37°C with PA for 5 min in the reaction medium before initiating calcium uptake by the addition of ATP.
Detection of the Phosphorylated Intermediate-The phosphorylation reaction was carried out at 4°C in a medium containing 30 g of microsomal protein, 160 mM KCl, 25 mM Tris-HCl (pH 7.0 at 4°C), 4 M thapsigargin, 0.2 mM CaCl 2 , and 0.06 mM LaCl 3 in a reaction volume of FIG. 1. Alignment of wild type h4xb Ca 2؉ pump and deletion mutants. The 1st line shows the segment of amino acids 290 -375 of the wild type h4xb pump. The residues that are missing in the mutants are indicated by a dash. Threonine and arginine residues that were introduced in the sequence of the mutants as a result of the construction of the deletions are in lowercase. The sites for tryptic cleavage in the N-terminal portion of the molecule leading to major proteolytic products of 81 and 76 kDa are indicated. Residues 301-312 shown in italics are absent in the alternatively spliced variant h4zb. The sequence of the peptide G25, which has been shown to interact with acidic lipids (15), is shown in bold letters. Residues 365-389 that according to hydropathy plots are part of the membrane-embedded portion of transmembrane helix M3 are underlined. Following the crystal structure of SERCA the residues from the M3 including those exposed to the cytosol are indicated by #.
0.25 ml. La 3ϩ was added because it is known to stabilize the phosphoenzyme of the plasma membrane Ca 2ϩ pump (25). The reaction was initiated by the addition of 1 M [␥-32 P]ATP and terminated after 1 min with 15 l of a solution containing 100% trichloroacetic acid. The precipitated proteins were dissolved in sample buffer and separated by SDS-PAGE in a 7% acrylamide gel according to Sarkadi et al. (4). After drying the gel, autoradiographs were produced with 24 -72 h of exposure at Ϫ70°C using X-Omat x-ray films.
Tryptic Digestion of Microsomes-Microsomes from CHO cells were suspended in a medium containing 0.05 mM EGTA, and the reaction was initiated by the rapid addition of 5 l of trypsin 0.1 g/l (ratio of microsomal protein:trypsin of 20:1). After incubation on ice for 30 s or 1, 10, or 25 min the reaction was stopped with 4 l of aprotinin 1.2 mg/ml. Controls (0 s) were done without trypsin.

RESULTS
Expression and Activity of Ca 2ϩ Pumps Lacking Residues from the Segment 296 -356 -Our initial goal was to see how much of the region A L could be removed from the PMCA molecule without changing the maximum capacity of the enzyme to transport Ca 2ϩ . To detect changes in the activity without the interference of the calmodulin-binding autoinhibitory domain, the deletions were made in the fully active ct120 pump (7). Immunoblots of microsomes from transfected COS cells showed two major bands, one of about 140 kDa corresponding to the endogenous PMCA and the other of faster migration corresponding to the recombinant ct120 enzymes (Fig. 2). Despite some variation in the amount of expressed proteins due to the efficiency of the transfection, none of the deletions significantly affected the amount of recombinant PMCA.
We began making deletions at downstream residue 300 because the amino acid sequences of different PMCA isoforms started to diverge approximately at that position. In fact, residues 301-312 were absent in the alternatively spliced option h4zb of isoform 4 (26). The deletion was extended to residue 356 because previous proteolysis work suggested that the segment 315-358 was not essential for activity (5)(6)(7)14). Table I shows the Ca 2ϩ uptake activity of the deletion mutants measured at a saturating concentration of Ca 2ϩ of 10 M. Mutants lacking residues from segment 300 -349 were as active as the ct120 pump. Moving the deletion upstream to residues 296 -349 to remove a cluster of three consecutive lysine residues still resulted in a fully active enzyme. In contrast, shifting the dele-tions downstream to residue 356 led to a near total loss of the Ca 2ϩ transport activity.
In order to characterize in more detail mutants d296 -349 and d300 -356, they were stably expressed as full-length Ca 2ϩ pumps in CHO cells. The expression levels estimated by immunoblotting were 100 and 60% of the wild type for d296 -349 and d300 -356, respectively (not shown).
The ability of the mutants to form an acyl phosphate from ATP was investigated (Fig. 3). Membranes containing the h4xb or the ct120 enzymes showed strong bands corresponding to the phosphorylated recombinant pumps. The active mutants d300 -314(ct120) and d300 -349(ct120) were also capable of forming phosphoenzyme. In contrast, only a weak band similar to that of the endogenous PMCA was observed in microsomes containing mutants with deletions involving residues 350 -356. Thus, in these mutants the lack of Ca 2ϩ transport activity was accompanied by the loss of the ability to react with Ca 2ϩ and ATP to form the phosphorylated intermediate.  in the A L region The rate of Ca 2ϩ uptake from COS cell microsomes was measured for 5 min at 37°C in the absence of calmodulin. The activity is given in percentage Ϯ S.D. of the wild type ct120 value that was 2.6 Ϯ 0.7 nmol/mg of microsomal protein/min. The free Ca 2ϩ concentration was 10 M. The activity of the endogenous PMCA (about 0.5 nmol/mg of microsomal protein/min), estimated as described under "Experimental Procedures," was subtracted. The number of determinations in two to four microsomal preparations is shown in parentheses. Partial Proteolysis of the d300 -356 and d296 -349 Mutants-The structure of the mutant proteins was investigated by examining their sensitivity to degradation by a brief exposure to trypsin (Fig. 4). The proteolytic PMCA fragments were recognized either with antibody 5F10, which reacts in the central portion of the molecule (amino acids 719 -738), or antibody JA9, which reacts near the N-terminal end (amino acids 16 -75). As reported before (4, 6, 20), a small amount of PMCA fragments that likely result from the activity of endogenous proteases are detected in the microsomal preparation before the beginning of the treatment with trypsin. As the digestion progressed the recombinant wild type protein was cleaved to fragments similar to those previously observed (5) using human erythrocyte membranes. Despite its lack of activity, the d300 -356 protein was rather stable and only became significantly degraded after 10 min of digestion. The relative resistance to degradation by trypsin was more obvious in mutant d296 -349. Particularly noticeable in the mutants was absence of a proteolytic fragment of about 35 kDa produced by cleavage after lysine 314, and the lower amounts of other fragments of smaller size that also originated from the N-terminal portion of the molecule. However, whereas the intensity of peptides of 90 -81 kDa was very weak in mutant d296 -349, in mutant d300 -356 a fragment of about 85 kDa detected by JA9 accumulated at the longest times of proteolysis.
Apparent Affinity for Ca 2ϩ and Calmodulin Activation of d296 -349 -After knowing that residues 296 -349 may be removed from the PMCA without altering the maximum Ca 2ϩ transport activity, we examined the effects of this deletion on the affinity of the PMCA for Ca 2ϩ and its regulation. Fig. 5 shows the dependence of Ca 2ϩ uptake on free Ca 2ϩ with or without added calmodulin. In the absence of calmodulin the recombinant wild type enzyme had low activity and low Ca 2ϩ affinity (K 0.5 about 1 M). Under these conditions and at low concentrations of Ca 2ϩ , d296 -349 was between 2 and 10 times more active than the recombinant wild type pump. However, as the Ca 2ϩ concentration increased the V max of d296 -349 reached a value similar to that of the recombinant wild type enzyme. Thus, the absence of residues 296 -349 activated the PMCA by enhancing its response to Ca 2ϩ . Calmodulin had a similar effect on the recombinant d296 -349 and wild type enzymes increasing the V max at saturating concentrations of Ca 2ϩ and reducing the K 0.5 for Ca 2ϩ . In addition the cooperativity for Ca 2ϩ activation, which was shown to be a characteristic of the calmodulin-activated pump (5), was less evident in the d296 -349 enzyme.
Effect of Phosphatidic Acid on d296 -349 -Because the only mechanisms that have been reported to decrease the K 0.5 for Ca 2ϩ to values lower than those obtained with calmodulin are the removal by proteolysis of region A L and the treatment of the enzyme with acidic lipids (5, 6), in the following experiment we investigated whether the activation caused by deletion of residues 296 -349 was additive to that of acidic lipids.

FIG. 4. Tryptic digestion of CHO cells membranes containing the recombinant h4xb and mutant pumps
d300 -356 or d296 -349. Membranes containing similar amounts of recombinant proteins (10 g for h4xb and d296 -249 or 16 g for d300 -356) were digested at 4°C with trypsin (20:1 protein:trypsin weight ratio) as described under "Experimental Procedures." The numbers at the top of each lane indicate the duration in minutes of the digestion. The PMCA fragments were either separated using a 12.5% SDS-PAGE and detected with antibody JA9, which reacts between amino acids 16 and 75, or separated using a 7.5% SDS-PAGE and detected with antibody 5F10, which reacts between amino acids 719 and 738.
To examine the activating effects of acidic lipids due to the putative N-terminal acidic lipid-binding site, the activity of the recombinant PMCA was measured at low concentrations of Ca 2ϩ and in the presence of a concentration of calmodulin high enough to saturate the C-terminal calmodulin acid lipid-binding site (Fig. 6). Under these conditions, increasing concentrations of PA stimulated the activity of the recombinant wild type reaching a maximum value of about 200% at 30 M PA, and then it decreased at higher concentrations of PA, probably reflecting the inhibitory effect of acidic lipids reported previously (8,13,27). In contrast, PA minimally affected the activity of mutant d296 -349 increasing its activity up to a maximum of about 120% at 75 M PA.

DISCUSSION
The N-terminal portion of the segment of amino acids connecting transmembrane helices M2 and M3 of the PMCA is part of the recently identified domain A of the SERCA that is highly conserved in all P-ATPases and is essential for catalytic activity (28). In contrast, the C-terminal portion of the M2-M3 loop that we called the A L region is only present in the PMCA, and it was proposed to be involved in the regulation of the PMCA by acidic lipids (6,14). Assuming that the primary function of this region would be the regulation of the activity of the enzyme, it should be possible to remove it from the molecule without altering the maximum capacity to transport Ca 2ϩ . Indeed this is exactly what results from the removal of the C-terminal calmodulin-binding region (7). Previous studies (5, 6) using limited tryptic proteolysis have shown that cleavage by trypsin after arginine 358 leads to a fully active 76-kDa fragment that is no longer responsive to activation by acidic lipids. In apparent contradiction with these findings, we found that deletions including residues 350 -356 led to the loss of Ca 2ϩ transport activity and phosphorylation from ATP. The discrepancy between the result of limited proteolysis and deletion mutagenesis may indicate that residues 350 -356 not only have a regulatory function but they are also required for the biogenesis of a functional PMCA. Because the interaction with anionic lipids has been shown to be important for the correct insertion and targeting of membrane proteins (29,30), it will be interesting to investigate whether the deletions in the A L region affect the delivery of the PMCA to the plasma membrane.
Mutants d300 -356 and d296 -349 seemed more resistant to trypsin digestion than the recombinant wild type pump. Al-though this may in part result from the removal of the proteolytic site at position 315, it also suggests that the global structure of the protein was not grossly altered by the deletions. Furthermore, the proteolysis patterns suggest that in these mutants the cleavage at position 358 was impaired. Nevertheless, at the longest times of proteolysis a fragment of about 85 kDa detected by JA9 became more intense in mutant d300 -356 than in either the recombinant wild type or mutant d296 -349, a fact that may reflect a structural change associated with the lack of activity of mutant d300 -356.
The crystal structure of SERCA has shown that the M3 and other transmembrane helices are longer than predicted previously (28). If this is also the case for the PMCA, the N-terminal end of M3 would be at valine 351, and hence the deletion of residue 350 -356 would directly affect the structure of M3. This transmembrane segment has been shown to be critical for the function of SERCA (31) and is part of the binding site for the highly specific SERCA inhibitor thapsigargin (32).
We found that at saturating concentrations of Ca 2ϩ , mutants with deletions involving amino acids from the segment 296 -349 had a Ca 2ϩ transport activity similar to that of the recombinant wild type pump. The importance of a net excess of positive charges on the cytoplasmic end of transmembrane segments for a proper membrane topology is well known (29,30). However, because the activity of mutant d296 -349 was preserved, it seems that residues 296 -349 are not critical for the correct folding or insertion of the PMCA in the membrane despite the fact that the deletion of amino acids 296 -349 effectively removed 15 positively charged residues from the segment preceding M3.
Although the removal of residues 296 -349 did not significantly affect the V max of Ca 2ϩ transport, it changed the behavior of the pump at low concentrations of Ca 2ϩ . Like acidic lipids, the deletion of residues 296 -349 increased the Ca 2ϩ affinity of the pump, and it did so more effectively than calmodulin. Moreover, when the calmodulin-binding site was saturated with calmodulin, the d296 -349 enzyme was not significantly stimulated by phosphatidic acid. These results suggest that the d296 -349 mutant resembles the activated form of the PMCA resulting from the binding of acidic lipids to their site in the A L region of the molecule.
The N-terminal acidic lipid-responsive segment of the PMCA is bracketed by two segments near the catalytic region of the pump that were proposed to form a receptor site for the Cterminal autoinhibitory sequence (33). Because calmodulin had a similar effect on the recombinant d296 -349 and wild type enzymes, it seems that the activation of the PMCA by the binding of acidic lipids to the A L region does not modify the inhibitory interaction of the C-terminal segment with its receptor site.
Because acidic lipids also bind and activate via the C-terminal autoinhibitory domain (12,13), the independence of the effects associated with the interaction of acidic lipids at a site in region A L has been difficult to assess. This in part was due to the fact that it has not been possible to use partial proteolysis to remove selectively the region A L of the molecule without also cleaving the C-terminal segment. In this respect, the mutant d296 -349 may be useful to study in more detail the effects of acidic lipids on the enzyme not activated by disruption of the inhibitory interaction between the C-terminal domain and the catalytic region of the pump. According to a model proposed previously (5,6) the Ca 2ϩ pump may be in three distinct functional states as follows: (i) the resting state with low V max and low Ca 2ϩ affinity (K 0.5 Ͼ1 M); (ii) the calmodulin activated state of high V max and intermediate affinity for Ca 2ϩ (K 0.5 ϳ0.5 M); and (iii) acidic lipid-FIG. 6. Stimulation of d296 -349 by phosphatidic acid as compared with the recombinant wild type Ca 2؉ pump. The Ca 2ϩ uptake activity was measured as described under "Experimental Procedures." Before starting the reaction by adding ATP, 10 g of microsomal protein was preincubated at 37°C for 5 min with 240 nM of calmodulin and increasing amounts of phosphatidic acid. The free Ca 2ϩ concentration was 0.2 M. The activity of the endogenous PMCA estimated as described under "Experimental Procedures" has been subtracted. The basal activity of each enzyme at 0 M of PA was taken as 100%. Data points are averages of two to seven independent determinations. Circles, recombinant h4xb; triangles, d296 -349. activated state of high V max and high Ca 2ϩ affinity (K 0.5 ϳ0.2 M). The results of the deletion 296 -349 suggest that, through the interaction of the A L region with the acidic lipids of the membrane, the PMCA could adopt a fourth state characterized by low V max and intermediate or high affinity for Ca 2ϩ .
It has been suggested that a lower rate of Ca 2ϩ extrusion by the PMCA due to a decrease in the level of activating phosphoinositides facilitates the rapid transient elevation of intracellular Ca 2ϩ after stimulation by agonists (34,35). Because the affinity for acidic lipids at region A L would seem higher than that of the C region (9,12,13), it is tempting to speculate that under normal resting conditions when the free cytoplasmic Ca 2ϩ level decreases to a level where calmodulin activation is insignificant, the phosphoinositides keep the PMCA in an activated state similar to that exhibited by the d296 -349 enzyme, and in response to the activation induced by an agonist, the PMCA would lower its affinity for Ca 2ϩ .
Our results suggest that in the absence of acidic lipids the A L region keeps the PMCA in a state of low affinity for calcium. In the wild type enzyme the binding of part of the A L region to the anionic lipids of the membrane is expected to relieve this inhibition. However, if this region is deleted as in mutant d296 -349, the interaction with acidic lipids is no longer required for a high Ca 2ϩ affinity. Thus, according to this model we propose that, besides the well characterized C-terminal autoinhibitory domain, the PMCA would contain an additional autoinhibitory membrane-binding domain. Autoinhibitory lipid-binding domains are characteristic of amphitropic proteins such as the CCT (cytidine 5Ј-triphosphate CTP phosphocholine cytidylyltransferase). The soluble CCT becomes activated when it binds to membrane lipids (36). Like the PMCA, the CCT is highly sensitive to anionic phospholipids and fatty acids, and their effectiveness relates mainly to the negative charge of the head group. Moreover, in analogy with the d296 -349 mutant, truncation of the CCT enzyme to remove the membrane-binding site generates a constitutive active enzyme (37).
The nature of the apparent increase in Ca 2ϩ affinity of the PMCA mediated by acidic lipids is a major quest for future work. It is worth mentioning that recent studies (31) of the SERCA have led to the proposal that the N-terminal portion of M3 is involved in the control of the access to the Ca 2ϩ transport sites. In the PMCA, the occurrence of the A L region just upstream M3 may turn the function of this transmembrane segment unusually sensitive to the lipid composition of the membrane.