Single Point Mutations in the Small Cytoplasmic Loop of ACA8, a Plasma Membrane Ca2+-ATPase of Arabidopsis thaliana, Generate Partially Deregulated Pumps*

ACA8 is a type 2B Ca2+-ATPase having a regulatory N terminus whose auto-inhibitory action can be suppressed by binding of calmodulin (CaM) or of acidic phospholipids. ACA8 N terminus is able to interact with a region of the small cytoplasmic loop connecting transmembrane domains 2 and 3. To determine the role of this interaction in auto-inhibition we analyzed single point mutants produced by mutagenesis of ACA8 Glu252 to Asn345 sequence. Mutation to Ala of any of six tested acidic residues (Glu252, Asp273, Asp291, Asp303, Glu302, or Asp332) renders an enzyme that is less dependent on CaM for activity. These results highlight the relevance in ACA8 auto-inhibition of a negative charge of the surface area of the small cytoplasmic loop. The most deregulated of these mutants is D291A ACA8, which is less activated by controlled proteolysis or by acidic phospholipids; the D291A mutant has an apparent affinity for CaM higher than wild-type ACA8. Moreover, its phenotype is stronger than that of D291N ACA8, suggesting a more direct involvement of this residue in the mechanism of auto-inhibition. Among the other produced mutants (I284A, N286A, P289A, P322A, V344A, and N345A), only P322A ACA8 is less dependent on CaM for activity than the wild type. The results reported in this study provide the first evidence that the small cytoplasmic loop of a type 2B Ca2+-ATPase plays a role in the attainment of the auto-inhibited state.

animal members of the type 2B group of Ca 2ϩ -ATPases. In fact, while in the animal isoforms (PMCA) this domain is localized in the extended C terminus of the protein, in plant isoforms, which have a very short cytosolic C terminus, the auto-inhibitory, CaM-and phospholipid-binding domain is localized in the extended cytosolic N terminus preceding the first transmembrane domain (2-5, 7, 8). Despite this structural difference, animal and plant type 2B Ca 2ϩ -ATPases share a number of catalytic and regulatory features (2)(3)(4)(5).
Fine regulation of type 2B group of Ca 2ϩ -ATPases activity plays an essential role in cell physiology: both the extent and the rate of activation contribute to determine the size and shape of cytoplasmic Ca 2ϩ waves induced by different stimuli and thus realize cellular response (6,9,10).
Although the terminal regulatory domain of both animal and plant members of type 2B Ca 2ϩ -ATPases have been described in detail (2)(3)(4)(5)(11)(12)(13)(14), much less is known about how its autoinhibitory action is exerted. Cross-linking of PMCA with a peptide corresponding to the extended CaM-binding site allowed the identification of two putative sites of intramolecular interaction within the cytoplasmic head containing the catalytic domain: one is localized in the big cytoplasmic loop connecting transmembrane domains (TMs) 4 and 5 and the other in the small cytoplasmic loop connecting TM2 and TM3 (15,16). The latter region, which is part of the actuator domain, is highly conserved between members of the type 2B Ca 2ϩ -ATPase group, and a peptide reproducing this region of ACA8, a PMlocalized isoform of Arabidopsis thaliana type 2B Ca 2ϩ -ATPase (17), has been shown to interact with ACA8 N terminus in pulldown experiments (12).
Regulation by an auto-inhibitory terminal domain is a feature shared by other members of the P-type ATPases superfamily, such as for example the PM H ϩ -ATPase of plants, which has an extended C terminus containing an auto-inhibitory domain whose action can be suppressed by binding of 14-3-3 proteins or by lysophosphatidylcholine (for a review see Ref. 18). Several single point mutations resulting in pump activation have been identified in Np-PMA2, an isoform of Nicotiana plumbaginifolia H ϩ -ATPase: many of the residues identified in this way cluster in the C-terminal regulatory domain, but, in addition, several residues have been identified in other regions of the pump molecule, including the small cytoplasmic loop (19,20).
To investigate the role of the small cytoplasmic loop connecting TM2 and TM3 in ACA8 auto-inhibition we have started a site-directed mutagenesis project. Because alaninescanning mutagenesis of the N-terminal auto-inhibitory domain of plant isoforms of type 2B Ca 2ϩ -ATPases suggests that a positively charged surface area of this domain could be an important feature of the auto-inhibitory interaction (11,13), we started by mutating acidic residues within the small cytoplasmic loop. The reported results show that in all the tested positions substitution of Asp or Glu residues with Ala generates a partially deregulated Ca 2ϩ -ATPase. Moreover, we also found that substitution of residue Pro 322 with Ala generates a partially deregulated enzyme.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-Site-directed mutagenesis of ACA8 was conducted according to the manufacturer's protocol (Quik-Change site-directed mutagenesis, Stratagene, catalog no. 200518) using primers listed in supplemental Table 1s. Introduction of the correct mutations and absence of errors were confirmed by sequencing. The DNA coding for WT and mutant ACA8 was inserted between KpnI and XhoI in the pYES2 vector (Invitrogen), under the control of a galactose-inducible promoter. The resulting plasmids were used for yeast transformation.
Yeast Transformation, Complementation, and Growth Media-Saccharomyces cerevisiae strain K616 (MAT␣ pmr1::HIS3 pmc1::TRP1 cnb1::LEU2, ade2, ura3) was used for expression of ACA8 and of the specified single point mutants (21). Yeast cells were transformed using a lithium acetate/polyethylene glycol method (22). The transformants were selected for uracil prototrophy on synthetic complete medium lacking uracil as described before (22). Complementation tests were performed as previously described (22) except for the EGTA concentration, which was 5 mM. For negative and positive controls yeast K616 was transformed, respectively, with the empty pYES2 vector and with pYES2-⌬74 ACA8, the latter coding for a truncated form of ACA8 (22).
Electrophoresis and Immunoblotting Analysis-SDS-PAGE, Western blotting, and immunodecoration with polyclonal antibodies against ACA8 sequences Glu 268 -Trp 348 and Val 17 -Thr 31 were performed as described (12). Signal quantification was performed using the Fluor-Chem TM SP Imaging System and AlphaEaseFC software by Alpha Innotech.
ACA8 Purification by CaM-affinity Chromatography-Yeast microsomes expressing WT or D291A ACA8 mutant were incubated with n-dodecyl ␤-D-maltoside (4 mg of detergent ml Ϫ1 :4 mg of protein ml Ϫ1 ) at 25°C for 30 min in a solubilization medium reported in Ref. 17 with the addition of 5 g ml Ϫ1 pepstatin and 5 g ml Ϫ1 chymostatin. The supernatant was recovered after centrifugation at 110,000 ϫ g for 1 h.
The solubilized proteins (ϳ150 l) were incubated overnight under gentle rotation at 4°C on 0.3 ml of CaM-agarose gel (Sigma catalog no. P4385), equilibrated with solubilization medium added with 37.5 g ml Ϫ1 Brij 58. After removal of the unbound fraction the solid phase was washed with 0.6 ml of washing medium containing 10% (v/v) glycerol, 20 mM Mops-KOH, pH 7.0, 1 mM p-aminobenzamidine, 2 mM dithiothreitol, 0.25 mM NaBr, 37.5 g ml Ϫ1 Brij 58, 1 mM ITP, 100 M CaCl 2 , 100 M MgSO 4 ; a second wash was performed in the same medium but in the absence of CaCl 2 and MgSO 4 . The third wash was carried out with 0.15 ml of a solution containing 10% (v/v) glycerol, 1 mM Mops-KOH, pH 7.0, 37.5 g ml Ϫ1 Brij 58, 0.1 mM EGTA. CaM-bound proteins were eluted in the same solution of the last wash except for the presence of 5 mM EDTA instead of EGTA. The eluted fraction was added with stoichiometric CaCl 2 to neutralize EDTA, and immediately used for assay of Ca 2ϩ -ATPase activity. An aliquot (ϳ1 l) of the eluate was solubilized (17) and loaded onto a precast Tris-glycine polyacrylamide gel (4 -20% linear gradient, Anamed, catalog no. TG42010); after electrophoresis the gel was stained with a silver impregnation method (Sigma, cat PROTSIL-1KT).
Trypsin Treatment-Endoplasmic reticulum-enriched fraction (1 mg of protein ml Ϫ1 ) was incubated for 10 min at 25°C in 0.1 mM EDTA, 0.5 mM ITP, 40 mM BTP-Hepes pH 7.0, in the presence or absence of 150 g ml Ϫ1 trypsin. The reaction was stopped by addition of 100-fold excess of soybean trypsin inhibitor.
Assays of ACA8 Activity-ACA8 activity was measured either as Ca 2ϩ -dependent MgITP or MgATP hydrolysis as previously described (23) or as eosin-sensitive MgATP or MgITP hydrolysis, taking advantage of the high sensitivity of plant PM Ca 2ϩ -ATPase to this inhibitor (22, 24). The latter method was routinely used for assays performed on microsomes to minimize artifacts due to the inhibiting effect of Ca 2ϩ on yeast endogenous ATPase activities: in fact (supplemental Fig. 1s), 10 M free Ca 2ϩ slightly, but significantly inhibited ATPase activity in microsomes from K616 yeast transformed with the empty vector, which was hardly affected by 0.2 M eosin Y. The inhibiting effect of Ca 2ϩ on yeast endogenous ATPase led to an underestimation of ACA8 activity measured as Ca 2ϩ -dependent MgITP or MgATP hydrolysis and overestimate its stimulation by CaM, specially in the case of poor expression or of low activity mutants.
In all cases free Ca 2ϩ concentration was buffered at the specified concentration with 1 mM EGTA; ITP was supplied at 1 mM, in the presence of 3 mM MgSO 4 , whereas ATP at 0.2 mM was supplied in the presence of 2.2 mM MgSO 4 ; unless otherwise specified, bovine testes CaM (Sigma, catalog no. P1431) was supplied at 1 M. Ca 2ϩ -dependent ATPase/ITPase activity was evaluated as the difference between activity measured in the presence of the specified free Ca 2ϩ concentration and that measured in the absence of added Ca 2ϩ ; eosyn-sensitive ATPase/ITPase activity at the specified free Ca 2ϩ concentrations was evaluated as the difference between activity measured in the absence of inhibitor and that measured in the presence of 0.2 M eosyn Y. When the effect of acidic phospholipids was tested, assays were performed in the absence of Brij 58. Samples (3-6 g of membrane protein or 5 l of purified enzyme) were incubated at 25°C for 60 min, during which the reaction proceeds linearly. All the assays were performed at least three times with three replicates.

Site-directed Mutagenesis of Acidic Residues in the Small
Cytoplasmic Loop of ACA8-Alignment of the segment of ACA8 small cytoplasmic loop, which has been shown to interact with the N-terminal regulatory domain of the pump (12), with the corresponding region of PMCA1 and ACA2, reveals the presence of several conserved acidic residues (Fig. 1, top  panel), some of which, such as Asp 278 , Asp 291 , and Glu 310 (numbers refer to ACA8 sequence), are part of P-type ATPase signature sequences (1). We produced single point mutants in which most of the conserved acidic residues were mutated to Ala and, for Asp 278 and Asp 291 , also to Asn. No mutant was produced for Glu 310 , because this residue has been shown to play an essential role in the catalytic cycle of the related sarcoplasmic reticulum Ca 2ϩ -ATPase, and even its mutation to Gln generates a drastically inactivated pump (25,26). We also produced ACA8 D239A mutant, because in ACA2 or PMCA4b mutation of the corresponding conserved residue, localized at the cytoplasmic end of TM2 generates a deregulated pump almost insensitive to further activation by CaM (11,27).
The DNA coding for WT or mutant ACA8 was used to transform Saccharomyces cerevisiae strain K616, which lacks endogenous Ca 2ϩ -ATPases (21) and is unable to grow in Ca 2ϩ -deprived media, unless it expresses a fully deregulated Ca 2ϩ pump (11,13,21,22). When the ability of the produced ACA8 mutants to complement K616 phenotype was tested ( Fig. 1, middle panel), only mutant D239A allowed K616 growth in the Ca 2ϩ -depleted medium, similar to the N-deleted mutant ⌬74 ACA8 (22). This result confirms the relevance of this conserved residue localized in the stalk of the small cytoplasmic loop in the mechanism of auto-inhibition of type 2B Ca 2ϩ -ATPases (11,27).
Protein expression was checked by Western blot of the microsomal fraction with an antiserum against the enzyme N terminus (17) (Fig. 1, bottom panel). No signal could be detected for mutants D278A and D278N suggesting that this residue is essential for ACA8 folding and stability. All of the other mutants were expressed: quantification of signals from three different blots indicated that the expression level varied ϳ2-fold between the least expressed mutant and the WT (Table 1).
To check whether the introduced mutations affect the enzyme catalytic activity, hydrolytic activity of WT and mutated ACA8 was measured in the presence of CaM (Table 1). Also CaM-stimulated Ca 2ϩ -ATPase activity varied among different mutants, being lowest in the least expressed mutants. Molecular activity of the mutants was evaluated, as a percentage of that of WT ACA8, by the ratio between activity in the presence of CaM and signal intensity in the immunoblot ( Table  1). All of the expressed mutants have a molecular activity at least half that of the WT, indicating that the mutations have no dramatic effect on the enzyme catalytic activity.
To test the degree of auto-inhibition of the mutants, we evaluated ACA8 basal activity in the absence of added CaM: for each mutant results are expressed as a percentage of activity measured in the presence of CaM. Fig. 2 shows that under the applied experimental conditions basal activity of WT ACA8 was ϳ20% of that measured in the presence of CaM (i.e. CaM ; the sequence of the peptide that has been shown to interact with ACA8 N terminus is underlined, and the amino acids selected for mutation are in bold. Middle panel, complementation of the K616 phenotype by WT and mutant ACA8. The Ca 2ϩ -ATPase-deficient yeast strain K616 transformed with WT or mutant ACA8 was grown in synthetic complete medium lacking uracil, 2% (w/v) glucose, and 10 mM CaCl 2 . 5-l drops of yeast culture (A 600 ϭ 1) were spotted on plated synthetic complete medium lacking uracil, 2% (w/v) galactose, 1% (w/v) raffinose, 5 mM EGTA and incubated for 3-4 days at 30°C. Results are from one experiment representative of four. Bottom panel, expression of WT and mutant ACA8 in yeast strain K616 as detected by immunodecoration of microsomal proteins (4 g per lane) with an antiserum against ACA8 N terminus following SDS-PAGE and blotting. Results are from one experiment representative of three.

TABLE 1 Relative molecular activities of ACA8 mutants
Quantification of ACA8 protein in microsomes from yeast expressing WT or mutant ACA8 was performed by densitometric scanning analysis of Western blots immunodecorated with an antiserum against the ACA8 N terminus. ACA8 activity was measured as eosin-sensitive ATPase activity (nanomoles of P i min Ϫ1 mg Ϫ1 protein) in the presence of 40 M free Ca 2ϩ and 1 M CaM. Molecular activities, evaluated as the ratio between ACA8 activity and ACA8 protein level, are expressed as a percentage of that of WT ACA8. Results are the mean Ϯ S.E. of three experiments.

ACA8
ACA8 protein ACA8 activity plus CaM Molecular activity stimulated the activity of WT ACA8 ϳ5-fold). In agreement with its behavior in the complementation test and with reported data on other isoforms of type 2B Ca 2ϩ -ATPase (11,27), mutant D239A had high basal activity, which was almost insensitive to CaM. All of the other mutants had basal activities significantly (p Ͻ 0.01) higher than that of the WT, although none were completely CaM-insensitive: basal activity of the least auto-inhibited mutants, as for example D291A, was about half of that measured in the presence of CaM. These results suggest that the negative charge of acidic residues plays a role in ACA8 auto-inhibition. Interestingly, mutant D291A was significantly (p Ͻ 0.05) less auto-inhibited than mutant D291N. Characterization of D291A ACA8 Mutant-To characterize mutant D291A, we purified by sucrose density gradient the endoplasmic reticulum-enriched fraction, in which overexpressed ACA8 accumulates (22): D291A ACA8 distributed as the WT, ruling out the possibility of altered mutant localization (data not shown). Basal ACA8 activity was somewhat lower than in the microsomal fraction, but the difference between WT and mutant D291A persisted (Fig. 3). Fig. 3 also shows the response of WT and D291A ACA8 to controlled proteolysis with trypsin, which selectively cleaves the N terminus of plant PM Ca 2ϩ -ATPase (12,28,29). Under the applied conditions tryptic treatment of the membranes effectively, albeit not completely, cleaved the N terminus of WT and D291A ACA8 (Fig. 3,  bottom panel), increasing the activity of both enzymes nearly as much as CaM addition. As a consequence of its higher basal activity, mutant D291A was much less stimulated than WT ACA8 also by tryptic cleavage of the N terminus.
The auto-inhibitory action of ACA8 N terminus can be suppressed, besides by CaM, also by acidic phospholipids (APLs) such as phosphatidylserine (PS) or phosphatidylinositol-4P (PI-4P). As for PMCA, APLs activate ACA8 via two distinct mech-anisms involving their binding to different sites (2,5,7,8): APL binding to a site in the protein N terminus, overlapping the auto-inhibitory and CaM-binding domain, stimulates ACA8 activity similar to CaM or to cleavage of the N terminus, whereas binding to a second, as yet unidentified, site further stimulates ACA8 activity by lowering its K 0.5 for free Ca 2ϩ . When ACA8 activity is assayed in the absence of CaM, the strong activation observed mainly reflects the effect of APL bound to the site in the N terminus, whereas activity assays performed in the presence of CaM at low free Ca 2ϩ concentration highlight the effect of APL bound to the second site (7).   21%). Under these conditions D291A ACA8 mutant was also less stimulated by 50 M PI-4P (inset). Conversely, in the presence of CaM (Fig. 4, bottom panel) WT and D291A ACA8 were similarly stimulated both by PS (half-maximal stimulation at 82 Ϯ 6 M and 97 Ϯ 11 M; maximal stimulation 91 Ϯ 5% and 83 Ϯ 6%) and by PI-4P (inset). Thus, mutation D291A only decreases the extent of ACA8 activation due to APL binding to the enzyme N terminus.
The fact that ACA8 mutant D291A has high basal activity and is less stimulated by all treatments known to suppress the auto-inhibitory action of the N terminus, CaM binding, tryptic cleavage, and acidic phospholipids, suggests that the mutation may loosen the auto-inhibitory interaction of the N terminus with the catalytic head and thus possibly make the CaM-binding site more accessible. To test this hypothesis we analyzed the dependence of stimulation of WT and D291A ACA8 activity on CaM concentration. The results reported in Fig. 5 show that mutant D291A required lower CaM concentrations to be activated: analysis of results from three independent experiments indicates that half-maximal activation of WT and D291A ACA8 is attained at 33 Ϯ 2 and 11 Ϯ 2 nM CaM, respectively.
In a final set of experiments we checked the auto-inhibition state of D291A ACA8 mutant following enzyme solubilization and purification by CaM-affinity chromatography (17,30). The high affinity of ACA8 for CaM, together with the high expression level of ACA8 in yeast, allowed single step enzyme purification from microsomes to virtual homogeneity (Fig. 6, left panel) both for the   WT enzyme and for the D291A mutant. Both enzymes, purified in the absence of added phospholipids, had very low basal activity, but also in this case basal activity of the D291A mutant was about twice that of the WT (Fig. 6, right panel).
Double Mutants in the Small Cytoplasmic Loop of ACA8-The results reported so far indicate that mutation of acidic residues within the small cytoplasmic loop of ACA8 generates partially deregulated pumps. However, with the exception of mutant D239A, which confirms the relevance of this conserved acidic residue in the stalk of the small cytoplasmic loop in autoinhibition of type 2B Ca 2ϩ -ATPases (11,27), none of the generated mutants were fully deregulated. To further investigate the role of the small cytoplasmic loop in ACA8 auto-inhibition, a new set of mutants was generated. To identify residues, which may contribute to generate a binding site for the N terminus, we used the structure of rabbit sarcoplasmic reticulum Ca 2ϩ -ATPase isoform 1a (SERCA1a) determined at 3.1-Å resolution in a Ca 2ϩ -free (E2) state (31) to build an approximate homology model (32,33) of ACA8 small cytoplasmic loop. Despite the low similarity of the two primary sequences SERCA is at present the best available structural model, which has been widely used to model the structure of other, even less related, P-type ATPases (e.g. Ref. 18). Moreover, SERCA-type ATPases of both plant and animal origin can interact with peptides reproducing the autoinhibitory CaM-binding domain of type 2B Ca 2ϩ -ATPases attaining an auto-inhibited conformation (34 -36): thus it is conceivable that the auto-inhibitory CaM-binding domain of type 2B pumps interacts with regions of the pump that are structurally conserved in distantly related type 2A pumps. Residues Ile 284 , Asn 286 , Pro 289 , and Pro 322 , putatively exposed to the external environment and proximal to Asp 291 in the model shown in Fig. 7 (top panel), were selected and mutated to Ala both in WT ACA8 and in the D291A mutant. Moreover, mutants V344A and N345A were generated, because these residues align with two residues of N. plumbaginifolia H ϩ -ATPase isoform PMA2, which when mutated give rise to pump activation (19,20).
All of the newly generated mutants were expressed in S. cerevisiae strain K616, albeit to variable extent (Fig. 7, middle  panel). Most of the single mutants had molecular activities in the presence of CaM (Fig. 7, middle panel) similar to that of the WT enzyme, indicating that the mutations had not seriously compromised enzyme activity. Conversely, all of the double mutants, albeit expressed as much as or more than the corresponding single mutants, had much lower molecular activities in the presence of CaM. None of the mutants were able to complement the phenotype of S. cerevisiae strain K616 restoring its growth in a calcium-depleted medium (data not shown).
To test the degree of auto-inhibition of the mutants, we measured basal ACA8 activity in the microsomal fraction. The results reported in Fig. 7 (bottom panel) show that among the newly generated single mutants only mutant P322A had a basal activity higher than that of the WT (44% of that measured in the presence of CaM, versus 20% for WT, p Ͻ 0.01). The phenotype of the double mutants was roughly similar to that of mutant D291A, as expected from the lack of effect of the newly introduced mutation, with the possible exception of mutant P322A/ D291A: this might have a basal activity higher than that of each of the corresponding single mutants, but its low molecular activity hampers a precise analysis of its activation state.

DISCUSSION
Different subfamilies of P-type ATPases are characterized by extended cytosolic terminal domains with regulatory role, which can exert an auto-inhibitory action by interacting with the catalytic head of the enzyme (1)(2)(3)(4)(5)17). However, although the regions of these terminal domains involved in auto-inhibition as well as modifications modulating or suppressing their auto-inhibitory action have been described in detail for different members of both the type 2B Ca 2ϩ -ATPase subfamily and the type 3A H ϩ -ATPase subfamily (2-5, 11-14, 17-19), little is known about how the auto-inhibitory action is exerted. Evidence has been presented that the terminal auto-inhibitory domain of both animal and plant type 2B Ca 2ϩ -ATPases is able to interact with a region of the small cytoplasmic loop connecting TM domains 2 and 3 (12,15,16), but the role of such an interaction in auto-inhibition remains to be elucidated.
The analysis of single point mutants of ACA8, an isoform of A. thaliana PM Ca 2ϩ -ATPase (17), reported herein provides the first straightforward evidence that the small cytoplasmic loop and in particular its Glu 268 -Trp 348 sequence plays a role in the attainment of the auto-inhibited state. Mutation to Ala of any of six tested acidic residues (Glu 252 , Asp 273 , Asp 291 , Asp 303 , Glu 302 , and Asp 332 ) results in a partially deregulated enzyme with higher basal activity in the absence of added CaM. Although none of these single point mutants are fully deregulated, these results point out the relevance of acidic residues in this region in ACA8 auto-inhibition. Because alanine scanning mutagenesis of the N terminus of plant isoforms of type 2B Ca 2ϩ -ATPases reveals the involvement of basic residues in the auto-inhibitory domain (11,13), it is possible that the negative charge conferred by acidic residues to the surface area of the small cytoplasmic loop favors and/or stabilizes its auto-inhibitory interaction with the N-terminal auto-inhibitory domain. D291A ACA8, which has the highest basal activity, has been characterized in some detail: this mutant is less stimulated by all treatments known to suppress the auto-inhibitory action of the N terminus such as CaM binding, tryptic cleavage of the N terminus, and acidic phospholipids and its phenotype is maintained after enzyme purification by CaM affinity purification. Altogether these results indicate that mutation D291A generates a genuine partially deregulated enzyme.
The D291A mutant has an apparent affinity for CaM ϳ3-fold higher that that of WT ACA8: because the ACA8 CaM-binding site is localized in the N terminus of the enzyme (13, 17) and thus is not directly affected by the mutation, the simplest interpretation of this result is that mutation D291A loosens the interaction between the N terminus and the catalytic head making the CaM-binding site more accessible. Whether this effect is due to suppression of a negative charge, to structural modification of an acceptor site for the auto-inhibitory N terminus, or to a broad conformational change, which may destabilize the auto-inhibited conformation of the enzyme, remains an open question. Our attempt to further test the role of the negative charge of acidic residues by generating a double mutant in which both Asp 291 and Asp 303 were mutated to Ala failed because we could not find any expression of it in yeast (data not shown).
The finding that the basal activity of D291A ACA8 is significantly higher than that of mutant D291N suggests a possible direct involvement of this residue in generation of an intramolecular binding site for the auto-inhibitory N-terminal domain. To identify an acceptor site for the auto-inhibitory N terminus we mutated residues close to residue Asp 291 in a putative threedimensional structure of ACA8 based on the structure of rabbit SERCA in a Ca 2ϩ -free (E2) state (31): only mutation of residue Pro 322 to Ala generates an enzyme with higher basal activity, less stimulated by CaM. The lack of effect of the other single point mutations on ACA8 auto-inhibition might be an outcome of the poor fit of the model: because no better reference exists at present, we will have to wait for definition of the threedimensional structure of type 2B Ca 2ϩ -ATPases, which promises to be a difficult task. All of the double mutants of ACA8 containing mutation D291A have the partially deregulated phenotype of the D291A mutant: unfortunately, the double mutant D291A/P322A ACA8 has a very low molecular activity, which makes it difficult to determine to what extent the effect of the two mutations sums up to generate a stronger phenotype. Moreover, preliminary results indicate that the apparent affinity for CaM of ACA8 is not affected by the P322A mutation (data not shown). This result suggests that the P322A mutation might destabilize the auto-inhibited conformation of ACA8 without directly affecting the interaction between the auto-inhibitory terminal domain and the catalytic head. It has recently been shown, both by intramolecular fluorescence resonance energy transfer and by measuring phospholipid binding, that transition of PMCA between the auto-inhibited and the fully active state involves substantial conformational rearrangements of both the cytoplasmic regions and the transmembrane segments (36,37). The latter may explain why single point mutations of residues localized at the cytoplasmic end of TM2 domain, such as residue Asp 239 in this work or the corresponding Asp residue of ACA2 or PMCA4b (11,27), generate deregulated pumps almost insensitive to further activation by calmodulin, possibly without a full disengagement of the auto-inhibitory terminal domain from the catalytic core (27,37).