Autoinhibition of a Calmodulin-dependent Calcium Pump Involves a Structure in the Stalk That Connects the Transmembrane Domain to the ATPase Catalytic Domain*

The regulation of Ca2+-pumps is important for controlling [Ca2+] in the cytosol and organelles of all eukaryotes. Here, we report a genetic strategy to identify residues that function in autoinhibition of a novel calmodulin-activated Ca2+-pump with an N-terminal regulatory domain (isoform ACA2 from Arabidopsis). Mutant pumps with constitutive activity were identified by complementation of a yeast (K616) deficient in two Ca2+-pumps. Fifteen mutations were found that disrupted a segment of the N-terminal autoinhibitor located between Lys23 and Arg54. Three mutations (E167K, D219N, and E341K) were found associated with the stalk that connects the ATPase catalytic domain (head) and with the transmembrane domain. Enzyme assays indicated that the stalk mutations resulted in calmodulin-independent activity, withV max, K mATP, andK mCa2+ similar to that of a pump in which the N-terminal autoinhibitor had been deleted. A highly conservative substitution at Asp219 (D219E) still produced a deregulated pump, indicating that the autoinhibitory structure in the stalk is highly sensitive to perturbation. In plasma membrane H+-ATPases from yeast and plants, similarly positioned mutations resulted in hyperactive pumps. Together, these results suggest that a structural feature of the stalk is of general importance in regulating diverse P-type ATPases.

The regulation of ion transport across membranes is an essential feature of all living cells. In many cases ions are transported against their concentration gradient (i.e. "uphill") via energy-dependent pumps or cotransporters. The ion pumps belonging to the P-type ATPase superfamily all share a similar enzymatic mechanism with an aspartyl-phosphate intermediate (1). Members include proton pumps, which function to energize the plasma membrane of fungi and plants with an electrochemical potential, and Na ϩ /K ϩ -ATPases, which provide an analogous function for animal cell plasma membranes. Eukaryotic cells also use P-type ATPases to transport Ca 2ϩ , heavy metals, and lipids across different membranes. Many P-type ATPases have been shown to be regulated by autoinhibitors or interacting proteins, with their activation involving a "release of inhibition" (2)(3)(4)(5). However, there are no examples in which the mechanisms of inhibition and activation are understood at a structural level (4, 6 -9).
The regulation of Ca 2ϩ -pumps is important because their activity can alter the magnitude, duration, and frequency of a Ca 2ϩ signal (10,11). Here, we investigate the autoinhibition of ACA2, a calmodulin-activated pump that belongs to the type IIB family of Ca 2ϩ -pumps (1,12). Type IIB pumps include the plasma membrane-localized Ca 2ϩ -ATPases (PMCAs), 1 first characterized in animals, and a sub-group of unique Ca 2ϩpumps recently identified in plants, including ACA2, which was cloned from a model plant system, Arabidopsis. There are two features that distinguish these plant pumps from an animal PMCA. First, their regulatory domains are located at the N-instead of C-terminal end of the pump. Second, some plant isoforms have been found in non-plasma membrane locations, such as ACA2 (endoplasmic reticulum) (13), and BCA1 (tonoplast) (14). Despite these differences, biochemical and genetic studies on ACA2 and BCA1 suggest mechanisms of autoinhibition and calmodulin activation that are analogous to a PMCA (12,(15)(16)(17).
For a PMCA, autoinhibition was proposed to occur through an interaction of a C-terminal autoinhibitor sequence with the first and second cytoplasmic loops (18,19). Together, these two loops form the ATPase catalytic domain (or head), which is connected via a stalk structure to the transmembrane domain through which Ca 2ϩ ions are translocated. A synthetic peptide, C28W, corresponding to the autoinhibitor of the human plasma membrane Ca 2ϩ -ATPase (hPMCA4), was found to inhibit hP-MCA4 with an IC 50 of 20 M. This peptide was found to interact with residues Ile 206 -Val 271 and Cys 537 -Thr 544 located in the first and second cytoplasmic loops, respectively, as revealed by cross-linking studies. This cross-linking experiment has led to a model in which autoinhibition of a type IIB pump involves intramolecular binding between the autoinhibitor and the ATPase catalytic domain.
Here, we present a genetic selection to identify mutations that disrupt autoinhibition in a representative member of the plant subfamily of type IIB pumps, ACA2. Our genetic approach was made possible by the discovery that only a deregulated version of ACA2 complemented a yeast host K616 (12) grown on calcium-deficient media. K616 lacks a functional Golgi Ca 2ϩ -ATPase (PMR1), a vacuolar Ca 2ϩ -ATPase (PMC1), and calcineurin B (20). Complementation appears to require a constitutively active Ca 2ϩ -pump in the endoplasmic reticulum/ Golgi in order to scavenge trace Ca 2ϩ for proper functioning of the secretory pathway. Of seven mutants obtained through random mutagenesis, three had mutations predicted to alter the stalk that connects the ATPase to the transmembrane domains. Our results suggest that the stalk may provide a structure that functions in autoinhibition by interacting directly or indirectly with the N-terminal autoinhibitor of ACA2. This regulatory structure may also be important for other P-type ATPases because similarly positioned mutations were also found associated with hyper-activation of proton pumps from plants and yeast (21)(22)(23).
Random Mutagenesis-Yeast harboring a wild-type ACA2 cDNA were mutagenized by resuspending a cell pellet in water plus 1% ethyl methane sulfonate (EMS) (v/v) and incubating for 30 min. Cells were washed in 5% sodium thiosulfate and plated on selection media consisting of YNB U Ϫ E 10 . Plasmids were isolated from the yeast colonies that grew on E 10 , used to transform Escherichia coli (25), and were subsequently sequenced at The Scripps Biochemistry Core Facility using an automated sequencer (Prism 373XL, ABI, Foster City, CA).
Site-directed Mutagenesis-Portions of ACA2 were amplified by a two-step PCR process to obtain the desired mutations in a clonable fragment. First, pYX-ACA2-1 (wild-type ACA2 under the control of a constitutive promoter from triose-phosphate isomerase in a vector con-taining a Ura ϩ marker (12)) was used as a template in a PCR reaction using TaKaRa Taq polymerase, per the manufacturer's instructions, with primers 207-213 and primer 153 (5Ј-GACCCAGTGAGTCCTCGT-TGA-3Ј) to obtain the desired point mutation. Table I shows the sequence of the primers used in these reactions. The reaction mix was then cleared of unused dNTPs and salts using the PCR cleanup kit (Qiagen). The entire "cleaned" mix was used in a subsequent PCR step using the products of the first PCR reaction as a mega-primer and primer 159 (5Ј-GTTCCATGGTGGAGTTGATGG-3Ј) as an upstream primer to amplify an easily subcloned fragment of ACA2 (i.e. from NcoI to HindIII). This reaction was cleaned as before, and the amplified fragment was digested with NcoI and HindIII and subcloned into the NcoI/HindIII site of pYX-ACA2-1. Potential site-directed ACA2 mutant plasmids were digested with SpeI, and those plasmids that were cut with SpeI were sequenced to ensure that only the desired mutation was present and that the remaining cloned portion was free of PCR-generated errors.
A directed alanine-scanning mutagenesis of the basic residues in the N terminus (Lys 23 -Arg 54 ) was conducted according to the manufacturer's protocol (QuickChange site-directed mutagenesis kit, Stratagene) using primers listed in Table I.
To verify the absence of unintended mutations, all regions of clones derived from PCR modifications were sequenced by automated sequencing procedures at Core Facilities located at The Scripps Research Institute or the Center for Agricultural Biotechnology (University of Maryland).
Membrane Fractionation-Yeast membranes were isolated and fractionated as described previously (12). Briefly, yeast was grown in 400 ml of YNB Ura Ϫ C 10 media to an approximate A 600 ϭ 0.8. Cells were pelleted, washed with 20 ml of water, and resuspended in 7.5 ml of homogenization buffer (100 mM Tris, pH 7.5, 20% glycerol, 20 mM EDTA, 1 mM dithiothreitol), 40 g/ml pepstatin, 20 g/ml chymostatin, and 0.4 mM phenylmethylsulfonyl fluoride; protease inhibitors were added fresh). 23 g of glass beads (0.5-mm diameter) were added to the resuspended cells and vortexed for 6 ϫ 1 min, alternating vortexing and incubation on ice. Glass beads and large cell debris were pelleted, and the supernatant was diluted 4-fold in GTED (20% glycerol, 50 mM Tris, pH 7.5, 10 mM EDTA, 1 mM dithiothreitol) and spun at 28,000 rpm (141,000 ϫ g) in a SW28 rotor (Beckman) for 2 h. Membrane pellets were homogenized in GTED ϩ protease inhibitors (50 g/ml pepstatin, 25 g/ml chymostatin, 90 g/ml phenylmethylsulfonyl fluoride) in a glass homogenizer, layered onto a 20 -60% continuous sucrose gradient (20 -60% sucrose, 50 mM Tris, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol), and spun for 19 h at 30,000 rpm (150,000 ϫ g) in a SW41Ti rotor (Beckman). Fractions (1ml) were taken from the top of the gradients, frozen in liquid nitrogen, and stored at Ϫ70°C. The peak fraction of GCCGCGGCGATGGCTCGCACCAACCAGGAG a Primers are written 5Ј to 3Ј, bold letters indicate bases that have been mutated in the ACA2 cDNA, and underlined bases indicate where a silent SpeI restriction site was introduced.
ACA2p (typically, fraction 5), as indicated by Western blots, was used for ATPase assays.
Western Blotting and Protein Quantification-Protein concentration of the individual sucrose fractions was determined by the method of Bradford (26). For SDS-polyacrylamide gel electrophoresis, protein samples were mixed with 3ϫ loading buffer (100 mM Tris, pH 6.8, 3.7% (w/v) SDS, 5% (w/v) dithiothreitol, 20% glycerol, 0.3% (w/v) bromphenol blue) and incubated for 10 min at 37°C. Samples were then electrophoresed through an 8% polyacrylamide gel (37.5:1, acrylamide:bisacrylamide, Fisher) and transferred to nitrocellulose using a Bio-Rad transfer apparatus. The transfer buffer consisted of 25 mM Tris, pH 8.3, 192 mM glycine, 20% (v/v) methanol, and 0.02% SDS. After overnight transfer at 130 mA, blots were blocked in blocking buffer (10 mM Tris, pH 7.6, 137 mM NaCl, 0.5% (v/v) Tween 20 (TBS-T), with 5% (w/v) nonfat dry milk) for at least 1 h at room temperature with shaking. The blots were then incubated for 1 h with the primary antibody 1371 raised against the ACA2 protein (12) diluted 1:2000 in blocking buffer. The blots were then washed 4 ϫ 10 min in TBS-T and incubated for 1 h with secondary antibody diluted 1:5000 in blocking buffer. The secondary antibody used for detection was a donkey anti-rabbit IgG conjugated with horseradish peroxidase (Amersham Pharmacia Biotech). Following secondary antibody incubation, the blots were washed 4 times for 10 min in TBS-T, and detection was made using ECL (Amersham Pharmacia Biotech).
ATPase Assays-ATPase assays were performed using a modified method of Lanzetta et al. (27). Sucrose fractions enriched in endomembranes (typically fraction 5 from the top, approximately 30% (w/w) sucrose (28)) were diluted to 0.125 g/l in GTED, and 10 l of a diluted fraction was added to the reaction mix to start the reaction. The reaction mix consisted of 90 l of 20 mM MOPS, pH 7.0, 8 mM MgSO 4 , 50 mM KNO 3 (which inhibits V-ATPase), 1 mM NaMoO 4 (inhibits phosphatase), 1 mM EGTA, 1 mM NaN 3 (inhibits mitochondrial ATPase), Ϯ 1 mM CaCl 2 and Ϯ 100 nM calmodulin. For each assay, a plus vanadate (90 M sodium orthovanadate) reaction was included and used to determine the vanadate-sensitive ATPase activity. A minus CaCl 2 reaction was also included to determine Ca 2ϩ -dependent, vanadate-sensitive ATPase activity. Ca 2ϩ -stimulated, vanadate-sensitive ATPase activity was determined in the presence and absence of calmodulin. Each reaction was incubated at 22°C for 40 min and then stopped by the addition of 800 l of stop solution (8.5 mM ((NH 4 ) 6 Mo 7 ) 24 -H 2 O (99.98%, Aldrich), 0.88 mM malachite green, 1 N HCl, and 0.04% tergitol (tergitol was added fresh)). After an 8 -12-min incubation at room temperature, 100 l of quenching solution (34% (w/v) trisodium citrate) was added to stabilize the color development. Quenched reactions were incubated for 1 h at 22°C before measuring the A 660 . The amount of phosphate produced in each reaction was determined by comparing the A 660 to a standard curve of known phosphate concentrations. A Western blot analysis comparing the amount of mutant enzyme to a dilution series of wildtype enzyme was used to correct for the amount of enzyme added to each ATPase reaction. Therefore, ATPase activity is expressed as the nmol Pi produced/min/unit of enzyme.
For K mATP measurements, the ATP concentration was varied between 0.3 and 3 mM ATP. For each ATP concentration, a ϮCaCl 2 and a Ϯvanadate reaction was included, and the Ca 2ϩ -stimulated, vanadatesensitive activity was determined by subtracting vanadate-sensitive activity in the absence of the Ca 2ϩ from that in the presence of Ca 2ϩ . For all K mATP measurements, the reaction mix included 2 mM phosphoenolpyruvate and 50 g/ml pyruvate kinase as an ATP regeneration system. For K mCa 2ϩ measurements, the concentration of Ca 2ϩ added to the reactions was 0 -1 mM. For each Ca 2ϩ concentration, a Ϯvanadate reaction was included, and the Ca 2ϩ -stimulated, vanadate-sensitive activity was calculated as described above. In the K mCa 2ϩ measurements, [Ca 2ϩ ] free was determined using fluo-3, pentapotassium salt, and the calcium calibration concentrated buffer kit per the manufacturer's instructions (Molecular Probes). In brief, a 1 mM fluo-3 stock was made in distilled water, aliquoted, and stored in the dark at Ϫ20°C. 1 M fluo-3 was used to construct a standard curve of known [Ca 2ϩ ] free by exciting the sample at 425 nM and reading the emission at 520 nM in a spectrofluorometer. The emission at 520 nM of 425 nM-excited samples containing the ATPase reaction mix and various [Ca 2ϩ ] were determined, and the [Ca 2ϩ ] free value was calculated using the standard curve.

Mutagenesis Reveals Two Regions of the Pump Involved in
Autoinhibition-To select for mutations that deregulate ACA2 activity, yeast K616 expressing a wild-type ACA2 were mu-tagenized with EMS and plated on Ca 2ϩ -deficient media. From more than 10,000 colonies that survived an initial selection (E 10 -resistant), an estimated 2,000 (20%) harbored a deregulated ACA2 mutant. Most of the initial survivors failed to grow after continued passage on E 10 media. Of those that survived multiple selections, more than 50% were found to have mutations linked to a plasmid encoding an ACA2 pump, as shown by the isolation of the plasmid and its reintroduction into a K616 host. A sequence analysis of 81 independent ACA2 mutants revealed only seven different point mutations (Fig. 1, Table II), four of which occurred in the putative autoinhibitor domain (R36H, R37W, R39P, R47P), whereas the remaining three occurred in the stalk region (E167K, D219N, and E341K).
The D219N mutation was found in 80% of the sequenced clones (Table II). This mutation was always observed as a nucleotide change from G to A in the context of dACT AGC (GϾA)AT TAC CGC. Because this mutation occurred with such high frequency, we set up multiple screens to look selectively for mutations downstream of this Asp 219 . For example, from a pool of approximately 2000 potential mutants, plasmid DNA was isolated, and DNA fragments containing potential mutations in the C-terminal portion of the pump downstream of Asp 219 were subcloned into a wild-type ACA2 background. This subcloned DNA was transformed back into yeast, and the yeast were subjected to selection on E 10 media. Of thousands of transformed yeast, only one survived selection and was subsequently shown to harbor a mutant pump with an E341K mutation. Three independent searches with this strategy confirmed that mutations that are located downstream of Asp 219 and confer deregulation were generated at a very low frequency by EMS mutagenesis.
Because the N-terminal domain was previously shown to contain an autoinhibitor, it was not surprising to find four mutations in this region produced by random EMS mutagenesis. To obtain more mutations in this region, we selectively mutagenized the N-terminal autoinhibitory region through mutagenic PCR and site-directed Ala scanning mutagenesis (Table II). Mutagenic PCR yielded 3 additional point mutations that conferred E 10 resistance, N33D, T41P, and A42D. The directed alanine substitutions of all 9 basic residues between Lys 23 and Arg 54 revealed that all but 2 (K35A and K46A) conferred growth to K616 on Ca 2ϩ -deficient media. In total, we found 15 mutations in the N-terminal domain between Lys 23 -Arg 54 that appear to generate a deregulated pump.
Three Stalk Mutations Result in Constitutively Active Pumps-All three stalk mutants (E167K, D219N, and E341K) resulted in approximately equal growth of K616 on Ca 2ϩ -deficient media ( Fig. 2A). Their expression levels in K616 were slightly lower than a wild-type ACA2 control (Fig. 2B), indicating that complementation was not an artifact of a pump simply being over-expressed. To determine whether the mutant pumps were deregulated, membranes were partially purified from yeast and their Ca 2ϩ -stimulated ATPase activities determined in the presence and absence of calmodulin (Fig. 2C). All three mutant pumps showed a high constitutive level of ATPase activity that was insensitive to calmodulin stimulation. Their specific activities, when normalized per unit of wild-type pump, were comparable with the deregulated pump created by an N-terminal truncation of the autoinhibitor (ACA2-2 ϭ ⌬N) (12,16).
The K mATP , K mCa 2ϩ and catalytic efficiencies (V max /K m ) of these mutant enzymes were measured and compared with truncated (⌬N) and wild-type pump controls (Table III). In all cases, the stalk mutants, like the truncated enzyme, showed kinetic parameters most similar to the fully activated state of a wild-type enzyme. Interestingly, each of the stalk mutants showed a different catalytic efficiency with respect to ATP, with D219N being the highest at "287," whereas E341K was more than two-fold lower at "120"; the mutant E167K at "226" was most similar to the truncated deregulated enzyme. In contrast, the catalytic efficiency with respect to Ca 2ϩ was similar for all three stalk mutants, but at a level 30% lower than the truncated pump. Thus, all three stalk mutations resulted in deregulated pumps but were nevertheless distinguished from each other by different kinetic parameters.
Mutation D219E Results in a Deregulated Pump-The remainder of this study focused on mutations in the immediate vicinity of the Asp 219 position. This position was considered of the highest priority for further investigation because alignments of different P-type ATPases indicated that this position was the most highly conserved among all pumps transporting Ca 2ϩ , H ϩ , or Na ϩ /K ϩ (28). To determine whether the D219N mutation resulted in a deregulated pump because of the loss of the Asp or the gain of an Asn, site directed mutagenesis was performed to replace Asp 219 with mutations D219A, D219E, and D219K. These mutant pumps all provided complementation when expressed in K616 (data not shown). In vitro ATPase assays confirmed that each mutant pump showed calmodulinindependent ATPase activity comparable with the activity of the calmodulin-stimulated wild-type enzyme (Fig. 3). Since even a conservative substitution of D219E resulted in a deregulated pump, autoinhibition of ACA2 appears to have a strict requirement for an Asp at the Asp 219 position.
Autoinhibition Is Not Disrupted by Three Ala Substitutions Downstream of Asp 219 -To determine whether the stalk region immediately downstream of Asp 219 is also involved in autoinhibition, the three neighboring residues were independently substituted (Y220A, R221A, Q222A). All three mutations failed to generate pumps conferring E 10 resistance in yeast host K616 (data not shown). A control showed that the same mutations introduced into a truncated pump (⌬N) still allowed these pumps to function (i.e. provide E 10 resistance), indicating that the mutations did not merely create "dead pumps". A normal ATPase activity was subsequently confirmed by in vitro assays on each full-length mutant (Fig. 3). Thus, the Asp 219 position alone at the beginning of stalk segment 2 represents a highly sensitive position in a regulatory structure.

Selection of Deregulated Ca 2ϩ -ATPase Mutants in Yeast-We
have used a molecular genetic approach to examine the structure and function of ACA2, a representative member of a unique subfamily of calmodulin-dependent Ca 2ϩ -pumps with an N-terminal autoinhibitor. The results presented here demonstrate that complementation of the yeast strain K616, deficient in two endogenous Ca 2ϩ -pumps, can be used to select for random mutations that constitutively activate ACA2. Using this strategy, we identified three residues in the stalk domain that play a critical role in autoinhibition. The power of this genetic approach is illustrated by the fact that despite an extensive literature on the structure and function of Ca 2ϩpumps, there was no a priori reason to suspect the identified stalk residues as part of a regulatory structure.
Prior studies on Ca 2ϩ -pumps (mainly animal SERCA) have analyzed hundreds of rationally designed mutations (28,29). However, a severe limitation of these studies has always been the laborious requirement that each mutant be individually expressed and biochemically characterized, which was further complicated by the fact that suitable expression systems all contained endogenous Ca 2ϩ -pump activities. Here, we demonstrate the efficacy of (i) a genetic strategy to select for a deregulated pump from millions of potential mutants by functional expression in yeast strain K616, and (ii) a rapid biochemical assay to determine a mutant pump's activity in yeast membranes lacking endogenous Ca 2ϩ -ATPase activity.
Our genetic strategy appears to be applicable to analyzing Each circle represents a single amino acid residue. Mutations resulting in deregulated pumps are shown as black boxes, and those that retain a wild type-like activity are shown as black circles. The autoinhibitor is shown as a shaded box. The bold P 32 indicates the conserved Asp residue that forms the phosphoenzyme intermediate. The open boxes in the cytosolic domains represent sequences homologous to the two domains in hPMCA4 that interact with the autoinhibitor. Predicted ␣-helical structures are shown as groups of angled rows of three residues. In the text, the ATPase catalytic domain refers to the first two cytosolic loops. ER, endoplasmic reticulum.
other divergent members of the type IIB Ca 2ϩ -pump family. Although complementation of K616 appears to require Ca 2ϩ transport into the endoplasmic reticulum/Golgi system, it is notable that complementation has been reported for mutant (deregulated) pumps related to ACA2 that normally target to other locations, including the plasma membrane or tonoplast in plant cells (5,30). It is likely that these non-endoplasmic reticulum/Golgi pumps nevertheless accumulate to sufficient levels in the yeast secretory pathway to provide the necessary activity in this non-native location. Whether a deregulated animal PMCA will function in this system has not yet been reported. Functional complementation has also been reported for other endomembrane calcium pumps, including ECA1 from Arabidopsis (31), SERCA1a from rabbit (32), and SMA2 from Schistosoma mansoni (33). However, these pumps differ from the type IIB pumps described above in that they (i) do not have calmodulin-regulated autoinhibitors and (ii) provide complementation in yeast when expressed as wild-type enzymes.
The N-terminal Autoinhibitor-Mutations identified here provided genetic evidence that the N-terminal autoinhibitor of ACA2 includes residues from Lys 23 to Arg 54 . In this stretch of 32 residues, 15 different mutations were found that appeared to disrupt the autoinhibitor, as indicated by the mutant pumps ability to restore growth of K616 on Ca 2ϩ -deficient media. It is likely that additional residues, both within and outside this region, will be found that contribute to a functional autoinhibitor, as the random and site-directed mutagenesis studies per-formed here were limited.
As delineated by mutations, the N-terminal autoinhibitor encompasses nearly all of a peptide sequence corresponding to Val 20 -Leu 44 , which was previously shown to function as a peptide inhibitor of ACA2 (16). This peptide inhibited Ca 2ϩ -transport by a wild-type ACA2 and a constitutively active, N-terminally truncated ACA2 (⌬N). In addition, it inhibited the activity of ECA1, a type IIA pump that is more closely related to the SERCA pumps in animals. Similarly, an inhibitory peptide (C28W) derived from the analogous C-terminal autoinhibitor of a human PMCA (hPMCA4) was also shown to inhibit a SERCA pump. Together, these results suggest that both N-and C-terminal autoinhibitors of type IIB pumps function by a similar mechanism, presumably by interacting with regions of the pump that are structurally conserved in distantly related type IIA pumps (e.g. SERCA, ECA1).
It is unclear how any of the mutations found here actually disrupt autoinhibition. We suspect that some substitutions pro-  2. Mutations E167K, D219N, and E341K result in deregulated pumps. A, growth on media containing 10 mM CaCl 2 or 10 mM EGTA of selected ACA2 mutants expressed in yeast strain K616 compared with that of vector only and wild-type ACA2 controls. B, Western blot showing relative levels of wild-type and selected ACA2 mutant enzymes in the sucrose gradient fraction (fraction 5) used in the assays. Numbers indicate the g of total protein loaded onto the gel. Equivalent results were seen for unfractionated membranes. C, Ca 2ϩ -stimulated ATPase activity in the presence and absence of calmodulin of wild-type and selected ACA2 mutants expressed in K616. Values represent the mean Ϯ S.D. of at least two determinations from at least two different enzyme preparations.
vide a global disruption of the autoinhibitor's structure, whereas others disrupt specific bonds between the autoinhibitor and its target binding site. For example, three of the mutations were found as proline substitutions (T41P, R39P, and R47P), which could easily disrupt a secondary structure. However, a more subtle disruption of structure appears likely for the seven Ala substitutions that resulted in deregulated pumps. These Ala substitutions all removed a basic charge, suggesting that one important feature of the autoinhibitor's interaction may be a positively charged surface area that is easily perturbed by the loss of even a single basic residue.
The Stalk Is Involved in Autoinhibition-The most significant insight from our genetic analysis was the discovery that the stalk domain could be altered to create a constitutively active Ca 2ϩ -pump. Here, the stalk domain refers to a structure that connects the transmembrane helices to the ATPase catalytic domain (head). The stalk is delineated in Fig. 5 by segments identified as S1 through S5, although structural studies on a SERCA would suggest that only four of the five provide helical segments in the stalk (7). The three mutations in the stalk domain (E167K, D219N, E341K) all resulted in deregulated pumps with activity similar to a truncated pump in which the N-terminal autoinhibitor was deleted, as shown by an increase in V max and an insensitivity to further stimulation by calmodulin. Interestingly, each of the three stalk mutants displayed a different catalytic efficiency with respect to ATP, although they were nearly identical with respect to Ca 2ϩ . This finding suggests that each stalk mutation caused a deregulated phenotype through a slightly different perturbation of structure, distinct from a simple deletion of the autoinhibitor. Nev-ertheless, the overall kinetic parameters of these stalk mutants are most similar to the fully activated state, as opposed to the basal state, of a wild-type pump.
Site-specific mutagenesis of Asp 219 and neighboring residues revealed three additional insights. First, the three residues downstream of Asp 219 can be replaced by alanines without a detectable disruption of autoinhibition or loss of calmodulinstimulated activity, as shown by complementation studies and enzyme assays (Fig. 3). Second, the loss of Asp at the Asp 219 position, as opposed to gain of an Asn is responsible for deregulation, is indicated by the deregulated activity of the mutant containing a D219A substitution (Fig. 3). Third, the Asp itself is uniquely required at this position and cannot be replaced by a highly conservative substitution of Glu (Fig. 3); this is significant because Asp and Glu differ by only a single carbon in the length of their acidic side chains.
To our knowledge, this is the first study to show a regulatory role for a residue corresponding to position Asp 219 in any Ca 2ϩpump. Of the three acidic residues identified here as stalk mutations, the Asp 219 position appears to be the most highly conserved in P-type ATPases, with a corresponding acidic residue present in PMCAs and SERCAs, as well as proton pumps and Na ϩ /K ϩ -ATPases (28). Although a mutation of a PMCA at this corresponding position has not been reported, the analogous position has been mutated in a rabbit fast twitch SERCA pump without a detectable effect (E109A) (34). Likewise, positions corresponding to the other two ACA2 stalk mutations were also mutated in a SERCA pump (E55Q, E243A, E244A, and D245A) without a detectable effect (34). However, these SERCA pump studies did not evaluate whether the mutations might disrupt an inhibitory interaction by phospholamban and therefore did not address the question of whether the stalk domain of a SERCA might be involved in regulation.
Similarities to Other P-type ATPases-There has been speculation that P-type ATPases may utilize similar mechanisms of inhibition regardless of whether the inhibition is accomplished by an autoinhibitor or a separate interacting protein (5,35). Nevertheless, there are many potential ways in which a pump could be inhibited, and there are no clear structural paradigms for how any P-type ATPase is regulated.
The PMCAs of animals and the plasma membrane proton pumps of plants and fungi represent the two best studied examples of autoinhibition of a P-type ATPase (2,4,(21)(22)(23). Although the autoinhibitors of these pumps are all located at the C-terminal end, domain swapping in a PMCA indicated that its autoinhibitor was still partially active in an N-terminal location, suggesting that the mechanism of autoinhibition was very tolerant of structural rearrangements (36). Thus, the unique location of an N-terminal autoinhibitor in members of a subfamily of ACA2-like pumps does not preclude the possibility that both N-and C-terminal located autoinhibitors function by  a Values represent the mean of least two independent K m measurements Ϯ the respective standard deviations. n ϭ number of independent K m measurements.
b Catalytic efficiency (V max /K m ) was determined for each enzyme using the means of V max and K m . c ND, could not be accurately determined by ATPase assay used here.
a similar mechanism.
In genetic studies similar to those conducted here for ACA2, proton pumps from a tobacco plant (NpPMA2 (21,22)) and yeast (ScPMA1 (23)) were also found to become hyperactivated as a result of mutations located in both the autoinhibitory domain and core regions of the pump. The relative positions of mutations found in the core regions of NpPMA2 and ScPMA1 are shown in Fig. 4 in comparison with the stalk mutations found in ACA2. Six of the 20 proton pump mutations were found in stalk segments 1, 2, and 3, in the same general vicinity as the three mutations found in ACA2. Thus, at least some proton and calcium pumps appear to have a structure involved in autoinhibition that can be perturbed by mutations located in the stalk.
The stalk mutations in the tobacco proton pump were identified by a genetic screen for pumps that could complement a deletion of the yeast plasma membrane proton pump when cells were grown under low pH growth conditions (21,22). These mutant pumps were all shown subsequently to have increased specific activities, which correlated with a structural change causing the C-terminal autoinhibitory domain to be more sensitive to trypsin digestion. This structural change is consistent with the hypothesis that the mutations resulted in the displacement of the C-terminal autoinhibitor from its site of inhibition.
The mutations in the yeast proton pump PMA1 were identified by a genetic screen for second site mutations that reversed the inhibitory effect of a primary mutation in the C-terminal autoinhibitor (23). As with the tobacco pump, a small number of mutations were located in the stalk. The two mutations found nearest to a position corresponding to Asp 219 in the ACA2 calcium pump were A165V and VD170/IN, predicted to be two and six residues downstream from an acidic residue that aligns with Asp 219 (28). The effect on enzyme activity of these second site mutations was examined in the absence of the primary mutation. In membranes harboring these mutant pumps, proton ATPase activity showed a V max approximately 5-10-fold higher than membranes containing wild-type enzyme. This apparent increase in activity supports the hypothesis that growth of yeast harboring the mutant pumps was because of a dominant mutation giving rise to a deregulated hyperactive pump.
During the reaction cycle of a P-type ATPase, the stalk domain plays a critical role in the transduction of conformational changes in the ATPase catalytic domain into conformational changes in the transmembrane helices that translocate ions through the channel domain (18). It is likely that the stalk domain also contributes to the structure of a pore through which ions enter the channel domain. Thus, there are multiple features of the stalk that could provide potential points of regulation. Which of these features is regulated by the Nterminal autoinhibitor in ACA2 is not known.
We offer two models for how the stalk may be involved in the autoinhibition of ACA2 (Fig. 5). In both models the N-terminal autoinhibitor "clamps" the enzyme in a non-active conformation that reduces activity by (i) blocking substrate binding sites (i.e. ATP or Ca 2ϩ ) or by (ii) preventing the relay of a conformational change between the ATPase catalytic domain and the channel domain. The primary distinguishing feature of the two models is the hypothesis that the autoinhibitor interacts directly with the ATPase catalytic domain in Model I, as suggested by cross-linking studies with the animal PMCA, compared with Model II in which the interaction is only with the stalk, at a position corresponding to the Asp 219 mutation identified here in ACA2. If Model I is correct ("loop interaction"), the mechanism by which the ACA2 stalk mutations deregulate the pump could be explained by a long distance conformational change that disrupts the docking of the autoinhibitor with the ATPase catalytic domain. If Model II is correct ("stalk interaction"), the mechanism of deregulation could be explained by a local change in conformation that directly disrupts an autoinhibitor binding site located in the stalk. Interestingly, all three stalk mutations identified here disrupted acidic residues, whereas eight of the mutations in the N-terminal autoinhibitor neutralized basic residues, consistent with the potential importance of an electrostatic interaction between these two regions.
In conclusion, the ability to select for active Ca 2ϩ -pumps in the yeast host K616 provides a powerful approach in which random mutagenesis can be used to dissect the mechanism of autoinhibition and calmodulin activation of type-IIB Ca 2ϩpumps. The discovery here of stalk mutations that deregulate ACA2 provides the first evidence for a regulatory structure associated with this region of a Ca 2ϩ -pump. This observation, together with the findings of analogous mutations in proton pumps, provides genetic evidence supporting a similar structural basis for autoinhibition among two very distantly related P-type ATPases.