Mechanism of N-terminal autoinhibition in the Arabidopsis Ca(2+)/H(+) antiporter CAX1.

Regulation of Ca(2+)/H(+) antiporters may be an important function in determining the duration and amplitude of cytosolic Ca(2+) oscillations. Previously the Arabidopsis Ca(2+)/H(+) transporter, CAX1 (cation exchanger 1), was identified by its ability to suppress yeast mutants defective in vacuolar Ca(2+) transport. Recently, a 36-amino acid N-terminal regulatory region on CAX1 has been identified that inhibits CAX1-mediated Ca(2+)/H(+) antiport. Here we show that a synthetic peptide designed against the CAX1 36 amino acids inhibited Ca(2+)/H(+) transport mediated by an N-terminal-truncated CAX1 but did not inhibit Ca(2+) transport by other Ca(2+)/H(+) antiporters. Ca(2+)/H(+) antiport activity measured from vacuolar-enriched membranes of Arabidopsis root was also inhibited by the CAX1 peptide. Through analyzing CAX chimeric constructs the region of interaction of the N-terminal regulatory region was mapped to include 7 amino acids (residues 56-62) within CAX1. The CAX1 N-terminal regulatory region was shown to physically interact with this 7-amino acid region by yeast two-hybrid analysis. Mutagenesis of amino acids within the N-terminal regulatory region implicated several residues as being essential for regulation. These findings describe a unique mode of antiporter autoinhibition and demonstrate the first detailed mechanisms for the regulation of a Ca(2+)/H(+) antiporter from any organism.

Transient elevations in cytosolic Ca 2ϩ are elicited in response to a range of environmental and internal stimuli, and these Ca 2ϩ elevations are then translated into a physiological response (1). The ability of a cell to generate specific Ca 2ϩ oscillations requires tight regulation of the influx, via Ca 2ϩ channels, and efflux, via Ca-ATPases (2). Ca 2ϩ /Na ϩ antiporters in animal cells and Ca 2ϩ /H ϩ antiporters in bacteria, yeast, and plants also mediate efflux of Ca 2ϩ (3,4). The antiporters are high capacity, low affinity Ca 2ϩ transporters that are very efficient at removing the transient cytosolic Ca 2ϩ spikes that occur during a signaling event (5). It is therefore important to determine the mechanisms of regulation of all of these Ca 2ϩ transporters. Many animal and plant Ca-ATPases are regu-lated by calmodulin binding to a C-terminal or N-terminal autoinhibitor (2,6,7). Ca 2ϩ /Na ϩ antiporters are regulated by intracellular concentrations of Ca 2ϩ and Na ϩ and are activated by a variety of modulators such as protein kinase C (3). However, we still know little about the regulation of Ca 2ϩ /H ϩ antiporters (4,8).
Ca 2ϩ /H ϩ antiporters have been characterized in plant, yeast, and microbial species (9 -12) and also exist in some animal tissues (13,14). These include chaA from Escherichia coli and VCX1 from Saccharomyces cerevisiae (11,12). Two Ca 2ϩ /H ϩ antiporters of Arabidopsis thaliana, CAX1 and CAX2, were identified by their ability to suppress the Ca 2ϩ sensitivity of a S. cerevisiae mutant lacking the vacuolar Ca-ATPase PMC1 and the Ca 2ϩ /H ϩ antiporter VCX1 (9). In addition, Arabidopsis has other putative cation exchanger (CAX) 1 proteins (15), including CAX3 and CAX4 (16,17). As recent studies demonstrate with CAX2, these antiporters may also transport other metals (18). A central question is how these transporters are modulated to affect ion homeostasis.
Two domains that control CAX1 activity have been identified (8,17,19). The first domain that regulates Ca 2ϩ /H ϩ antiport function has been termed the regulatory or autoinhibitory domain (8). We recently identified a CAX1 open reading frame (previously termed long-CAX1) that contains an additional 36 amino acids at the N terminus (8). These amino acids were not found in the original clone identified by suppression of the yeast vacuolar Ca 2ϩ transport mutant (9). This longer version of CAX1 localizes to the yeast vacuole, but does not transport Ca 2ϩ even in the presence of calmodulin (8). Sequence analysis suggests that an N-terminal regulatory region (NRR) may be present in all Arabidopsis CAX transporters. Using a series of N-terminal-truncated CAX (sCAX) chimeric constructs, a second domain, the Ca 2ϩ domain (CaD), has been identified that appears to modulate Ca 2ϩ transport (19). CAX3 and CAX4 can strongly suppress yeast vacuolar Ca 2ϩ transport mutants if the 9-amino acid CaD of CAX1 is inserted into N-terminal-truncated versions of these transporters (sCAX3-9 and sCAX4-9; Refs. 17 and 19). Alternatively, both CAX3 and CAX4 can suppress yeast vacuolar Ca 2ϩ transport mutants if N-terminal additions are made to these transporters (17). These findings imply that CAX-dependent Ca 2ϩ transport depends on N-terminal modifications; however, the nature of the N-terminal regulation remains unclear.
Here we demonstrate that a peptide corresponding to the 36 amino acids of the CAX1 N terminus can specifically inhibit CAX1-mediated Ca 2ϩ /H ϩ transport in yeast cells and Ca 2ϩ -treated Arabidopsis roots. We identify individual amino acids within the N terminus that are important for this autoinhibition and map a region of CAX1 to which the NRR interacts. This novel autoinhibitory mechanism suggests that plants judiciously regulate Ca 2ϩ /H ϩ transport and suggest a means to engineer plant Ca 2ϩ transport.
Preparation of Endomembrane Vesicles and Ca 2ϩ Transport Assay-Yeast vacuolar-enriched membrane vesicles were prepared as previously described (8). Arabidopsis membrane vesicles were prepared from root tissue obtained from 4-week-old plants cultured in Gamborg's B5 medium (Invitrogen, Carlsbad, CA) and pretreated with 100 mM CaCl 2 18 h prior to harvest. Microsomal membranes were isolated as described (22). Then, vacuolar-enriched vesicles were prepared as for yeast. Measurements of time-dependent 45 Ca 2ϩ /H ϩ transport into endomembrane vesicles were performed as previously described (8,19), except 5 M carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (FCCP) was used instead of gramicidin for the Arabidopsis transport assay. Synthetic peptides were synthesized (Dept. of Immunology, Baylor College of Medicine) by Fmoc (N-(9-fluorenyl)methoxycarbonyl) solid phase synthesis and purified by reverse phase high pressure liquid chromatography (HPLC). To measure Ca 2ϩ transport in the presence of a synthetic peptide, membrane vesicles were incubated with or without the peptide for 10 min at 25°C prior to the addition of 1 mM MgATP.
Yeast Two-hybrid Interaction Assay-The CAX1 NRR was amplified by PCR and cloned in-frame into the EcoRI site of pACT2. Partial fragments of sCAX1, sCAX3, and sCAX3-9-7aa (residues 44 -69) and a partial fragment of sCAX1 (residues 438 -463) were amplified by PCR and cloned in-frame into the NcoI and BamHI sites of pAS2. Plasmids were transformed into the yeast strain Y190 and selected on synthetic complete (SC) agar medium lacking Leu, Trp, and His (23). Expression of constructs was confirmed by Western blotting using an anti-hemagglutinin antibody (8). Protein interaction was determined by assaying for LacZ expression using the ␤-galactoside filter assay (23).

RESULTS
A Synthetic Peptide Corresponding to the NRR of CAX1 Specifically Inhibits sCAX1 Ca 2ϩ /H ϩ Antiport Activity-To determine whether the CAX1 NRR functions as an autoinhibitory domain, a synthetic peptide was generated that corresponds to all 36 amino acids (CAX1-NRR peptide, MAGIVTEPWSVA-ENGNPSITAKGSSRELRLGRTAHNM, net charge ϩ1). Ca 2ϩ /H ϩ antiport activity by sCAX1, which lacks the NRR, was inhibited in a concentration-dependent manner. A peptide concentration of 5 Ϯ 1.5 M was sufficient to inhibit 50% of sCAX1 transport activity with 10 M Ca 2ϩ (Fig. 1, A and B). A control peptide (LKIKTEGNFEQANEELRAIIKKIWKRT-SMKLLDQV, net charge ϩ3) of similar size and charge to CAX1-NRR did not inhibit sCAX1-dependent transport at concentrations of as much as 20 M (Fig. 1A). The K m of sCAX1 for Ca 2ϩ increased ϳ3-fold in the presence of 6 M peptide (Fig.  1C). Inhibition of Ca 2ϩ /H ϩ antiport activity was specific to sCAX1, as the peptide was unable to inhibit Ca 2ϩ transport by other antiporters sCAX2, VCX1, and sCAX3-9 (Fig. 1A).
We wanted to verify that the inhibitory effect of the CAX1-NRR peptide on CAX1-dependent Ca 2ϩ transport was not an artifact of the yeast expression system. MgATP-dependent H-ATPase-driven Ca 2ϩ transport, in the presence of the Ca-ATPase inhibitor vanadate, was determined in vacuolar-enriched membrane vesicles from wild-type Arabidopsis roots ( Fig. 2A). Given that CAX1 expression is highly induced by exogenous Ca 2ϩ (24), we isolated vacuolar-enriched membrane vesicles from Arabidopsis plants treated with 100 mM Ca 2ϩ for 18 h. Significant Ca 2ϩ /H ϩ antiport activity could be measured only from Ca 2ϩ -treated plants ( Fig. 2A). Ca 2ϩ /H ϩ antiport activity was inhibited in a concentration-dependent manner by the CAX1-NRR peptide (Fig. 2B). A peptide concentration of ϳ9 -10 M was required for 50% inhibition of Ca 2ϩ transport.
The CAX1 NRR Interacts with the N Terminus of sCAX1-To deduce which residues were important for interaction of the NRR with sCAX1, the CAX1-NRR peptide was tested on sCAX1 chimeras that contain various domains of sCAX3 and sCAX2. These constructs were previously used to identify the CaD within the N terminus of sCAX1 that is critical for Ca 2ϩ transport (19). sCAX3-9 contains an in-frame fusion of the CAX1 CaD (residues 87-95), allowing sCAX3 to transport Ca 2ϩ (19). As the CAX1-NRR peptide interacts with sCAX1, but not sCAX3-9 (Fig. 1A), sCAX1/CAX3 chimeras were used to determine the amino acids that are required for binding the NRR peptide. A chimera (sCAX3-A) that contains the first quarter of sCAX1 (residues 37-149) fused in-frame to the CAX3 open reading frame was inhibited by the peptide (Fig. 3A). We then tested whether the NRR interacts with the CAX1 CaD. Ca 2ϩ transport by sCAX3-9 and by sCAX2 containing the CAX1 CaD (sCAX2-9) was not inhibited by the NRR peptide ( Fig. 3A; data not shown), whereas transport by sCAX1 containing the equivalent CaD of CAX2 (sCAX1-9) was still inhibited. The effect of the peptide on Ca 2ϩ transport by sCAX3-A2 and sCAX1-A1 was then tested. In these constructs, the first quarter domain (residues 37-149) was further divided in half and comprised the immediate N terminus of sCAX3 (residues 37-73); only the second half of this domain was CAX1 (residues 74 -149). No inhibition of Ca 2ϩ transport occurred for either of these constructs (Fig. 3A).
The CAX1 NRR Interacts with a Region of 7 Amino Acids within the First Quarter of sCAX1-The amino acid sequences for CAX1 and CAX3 within the region of residues 37-73 were compared. Two regions of heterogeneity were identified within this sequence, a region of 2 amino acids and a region of 7 amino acids. Mutagenesis was performed on sCAX3-9 and sCAX1-A1 to change residues LV to VI (residues 50 -51) or residues CK-TLKNI to YKGLKDF (residues 56 -62). Following mutagenesis, both constructs were able to transport Ca 2ϩ . The CAX1-NRR peptide was unable to inhibit Ca 2ϩ transport by sCAX3-9 and sCAX1-A1 (Fig. 3A). This peptide was able to inhibit Ca 2ϩ transport by both constructs when they contained the 7-amino acid substitution (sCAX3-9-7aa and sCAX1-A1-7aa; Fig. 3A). However, no inhibition was found when they contained the 2-amino acid substitution. Furthermore, when mutagenesis was performed on CAX1 so that the 7-amino acid region was swapped with that of CAX3 (to give CAX1-7aa), CAX1-7aa was now able to suppress the Ca 2ϩ -sensitive phenotype of K667, although not as strongly as sCAX1 (Fig. 3B). sCAX1 containing the 7-amino acid region of CAX3 (sCAX1-7aa) could suppress the Ca 2ϩ -sensitive phenotype as strongly as sCAX1 (Fig. 3B). However, Ca 2ϩ transport by sCAX1-7aa could not be inhibited by the CAX1-NRR peptide (Fig. 3C).
A physical interaction between the NRR and the 7-amino acid region of sCAX1 was tested using a yeast two-hybrid assay. When the CAX1 NRR in pACT2 was coexpressed with a hydrophilic N-terminal fragment of sCAX1 (residues 44 -69 including the 7-amino acid region) in pAS2, the lacZ gene was activated, indicating that an interaction had occurred (Fig. 3D). No interaction occurred with a similar N-terminal fragment of sCAX3 or a hydrophilic C-terminal fragment (residues 438 -463) of CAX1. Similar results were obtained with other twohybrid constructs; specifically, the NRR of CAX1 only interacted with constructs containing residues 44 -69 of CAX1 (data not shown). No interaction was observed when CAX1 NRR was coexpressed with an N-terminal fragment of sCAX3 that included the 7-amino acid region of CAX1, although we did occasionally observe a weak interaction with some of the replicates (data not shown).
Modification of Amino Acids in the NRR of CAX1-We previously noted that the NRR of CAX1 contains putative phosphorylation sites (8). To investigate whether any of the five residues Thr 6 , Ser 10 , Ser 24 , Ser 25 , and Thr 33 may be phosphorylated, they were each mutated to Ala. This change was expected to block any phosphorylation event required to deactivate CAX1. All five CAX1 mutants were expressed at equal levels in yeast (data not shown), but only the T33A change strongly suppressed the K667 Ca 2ϩ -sensitive phenotype (Fig.  4A). Thr 33 was further mutated to Asp to mimic continuous phosphorylation to see whether activation could be blocked. The T33D change and all other mutations of this residue (T33S and T33E) allowed yeast growth on high Ca 2ϩ , indicating CAX1 activation (Fig. 4A). Because the S25A mutation had , or CAX3 open reading frames, respectively. All constructs are truncated and start at Met 37 . Ca 2ϩ transport activity was determined by comparing ⌬pH-dependent 10 M Ca 2ϩ uptake activity (nmol of Ca 2ϩ mg of protein Ϫ1 ) after 10 min in the absence of peptide by each CAX construct. Activities are scaled from "ϩϩϩϩ" (high activity) to "ϩ" (low activity). For all of the CAX constructs that were inhibited, inhibition of transport activity was greater than 60%. Ca 2ϩ transport activity in the absence and presence of peptide was determined using 2-4 independent membrane preparations. B, suppression of Ca 2ϩ sensitivity of K667 by sCAX1 and CAX1 containing a 7-amino acid region of CAX3 (residues 56 -62). Saturated liquid cultures of K667 expressing various plasmids, as indicated, were diluted to an A 600 of 1.5, then spotted onto SC medium lacking His (-His) and yeast extract-peptone-dextrose (YPD) medium containing 200 mM CaCl 2 (ϩCa). Yeast growth is shown after 3 days. C, the effects of 8.5 M CAX1-NRR peptide on ⌬pH-dependent 10 M Ca 2ϩ transport by sCAX1 and sCAX1-7aa into yeast endomembrane vesicles measured at a 10-min time point. Results are the mean (ϮS.E.) of two experiments. D, interaction of the CAX1 NRR with fragments of sCAX1 or sCAX3 by yeast two-hybrid analysis. Yeast strains expressing fragments of sCAX1 (residues 44 -69), sCAX3 (residues 44 -69), or sCAX1 (residues 438 -463) in pAS2 with NRR in pACT2 and grown on SC medium lacking His, Trp, and Leu were assayed for LacZ expression using the ␤-galactoside filter assay. Filters are shown following 4 h of development.

FIG. 4. Effects of site-directed mutagenesis on selected Ser and
Thr residues in the N terminus of CAX1. A, growth of K667 expressing vector alone, sCAX1, and wild-type and mutant CAX1 with Ser or Thr substitutions as indicated. All yeast strains were grown on SC medium lacking His (ϪHis) and YPD medium containing 200 mM CaCl 2 (ϩCa). Yeast growth is shown after 2 days. B, amino acid sequence of CAX1 from residues 1-37 with the results of the mutagenesis summarized below. Mutated residues are highlighted in bold type.
caused very weak suppression of Ca 2ϩ sensitivity, we further mutated this residue. A S25T change gave no suppression ( Fig.  4B; data not shown) whereas a S25D change caused strong suppression of the K667 phenotype (Fig. 4A). The yeast growth was caused by Ca 2ϩ /H ϩ transport activity conferred by CAX1S25D as demonstrated by the 45 Ca 2ϩ uptake into endomembrane-enriched vesicles from CAX1S25D expressing yeast (Fig. 5). No Ca 2ϩ uptake could be measured for CAX1 and CAX1S25A. When other Ser residues in the NRR were mutated to Asp (S10D or S24D) no suppression of the K667 phenotype was observed ( Fig. 4B; data not shown).
The CAX1 NRR contains a number of charged residues, particularly Glu and Arg (Fig. 6). To investigate whether any of these residues have a role in the binding mechanism of the NRR, three sets of double substitutions were created to remove the charge of these residues: E7A/E13A, R26A/E27A, and R29A/R32A. Only the R29A/R32A mutation allowed CAX1 to strongly suppress the Ca 2ϩ -sensitive phenotype of K667, comparable to the suppression by sCAX1 (Fig. 6). DISCUSSION The regulation of Ca 2ϩ efflux mediated by high capacity Ca 2ϩ /H ϩ exchange activity (and by Ca 2ϩ pumping directly energized by ATP hydrolysis) is likely to be an important com-ponent in Ca 2ϩ signaling (2). Vacuolar Ca 2ϩ /H ϩ antiporters can sequester a sudden burst of cytosolic Ca 2ϩ more efficiently than the higher affinity Ca-ATPase (5). However, the regulation of Ca 2ϩ /H ϩ antiporters is poorly understood. For example, the yeast Ca 2ϩ /H ϩ antiporter VCX1 appears to be negatively regulated at the post-translational level by the Ca 2ϩ /calmodulindependent phosphatase calcineurin; however, the nature of this regulation remains unclear (12). Furthermore, nothing is known about the regulation of the E. coli Ca 2ϩ /H ϩ antiporter chaA or the mung bean antiporter VCAX1 (10,11). We recently described the presence of an NRR on Arabidopsis CAX1 (8).
Here we demonstrate that the NRR of CAX1 regulates Ca 2ϩ /H ϩ antiport activity by a unique autoinhibitory mechanism.
A synthetic peptide composed of the 36 amino acids of the CAX1 NRR inhibited Ca 2ϩ transport by the truncated sCAX1 at micromolar concentration, but did not inhibit transport by other Ca 2ϩ /H ϩ antiporters (Fig. 1A). Furthermore, a control peptide with similar charge and size properties did not inhibit Ca 2ϩ transport by sCAX1 (Fig. 1A). The CAX1-NRR peptide also substantially increased the K m for Ca 2ϩ of sCAX1 (Fig.  1C). The inhibition by the CAX1-NRR peptide was not caused by a reduction of the proton motive force across the membrane

FIG. 6. Effects of site-directed mutagenesis on selected Glu and Arg residues in the N terminus of CAX1.
Growth of K667 expressing vector alone, sCAX1, and wild-type and mutant CAX1 with double Glu or Arg substitutions as indicated. All yeast strains were grown on SC medium lacking His (ϪHis) and YPD medium containing 200 mM CaCl 2 (ϩCa). Yeast growth is shown after 2 days. Below, amino acid sequence of CAX1 from residues 1 to 37 with the results of the mutagenesis summarized below. Mutated residues are highlighted in bold. Amino acid charge is denoted by "ϩ" or "Ϫ". through a nonspecific interaction with lipids or proteins, because the peptide did not disrupt ⌬pH-dependent Ca 2ϩ transport by sCAX2, VCX1, or sCAX3-9 (Fig. 1A). It could also be argued that the peptide was directly binding Ca 2ϩ , as this would also raise the apparent K m for Ca 2ϩ . However, we believe that this was not the case, because Ca 2ϩ binding by the peptide would cause a reduction in free Ca 2ϩ concentration that would reduce the level of Ca 2ϩ transport by all antiporters, not just sCAX1. In addition, we observed a reduction in transport only for sCAX1. These results indicate that the 36-residue sequence does function like an inhibitory domain confirming that CAX1 is regulated by autoinhibition. We wished to verify that CAX1 is similarly regulated in Arabidopsis and that our observations were not an artifact of the yeast expression system. In very few studies has Ca 2ϩ /H ϩ antiport activity been measured in Arabidopsis tissue (25). We were able to measure Ca 2ϩ /H ϩ antiport activity in vacuolar membranes from Arabidopsis roots (Fig. 2A). These plants were pretreated with 100 mM Ca 2ϩ that has been shown to strongly induce expression of CAX1 (24) but not CAX2 (18). This antiport activity was strongly inhibited by the CAX1-NRR peptide in a manner similar to the inhibition of sCAX1-dependent Ca 2ϩ transport in yeast (Fig. 2B). The ability of the peptide to inhibit the antiport activity suggests that under certain conditions (such as when high levels of exogenous Ca 2ϩ are present) the majority of Ca 2ϩ /H ϩ antiport activity at the root vacuole is due to CAX1. We speculate that CAX1 only exists as a full-length version in planta; therefore, the antiport activity measured was mediated by an activated CAX1. Therefore, this result indicates that the CAX1-NRR peptide can inhibit full-length CAX1. We may hypothesize that an activator protein interacts with the NRR and prevents autoinhibition, allowing the NRR peptide to bind to CAX1 instead and inhibit activity. Alternatively, the peptide may compete directly for the activator protein, preventing the activator from interacting with CAX1, which remains autoinhibited.
Identification of putative phosphorylation sites in the NRR led us to investigate whether CAX1 is regulated by a phosphorylation event. Our previous findings showed that minor deletions in the NRR can perturb inhibition (8); however, changing Thr 6 , Ser 10 , or Ser 24 to Ala residues did not disrupt N-terminal inhibition (Fig. 4). Mutagenesis of Thr 33 to Ala strongly suppressed the Ca 2ϩ sensitivity of K667 expressing CAX1, initially indicating that this residue may be phosphorylated to block CAX1 activation. However, any mutation of this residue caused yeast growth to occur, suggesting that this Thr has an important structural role in the autoinhibition rather than in being phosphorylated. The Ser 25 to Ala change caused minimal growth; however, when Ser 25 was mutated to Asp to mimic continuous phosphorylation, CAX1 was as active as sCAX1 (Figs. 4 and 5). No suppression of the K667 phenotype was observed following mutation of Ser 10 or Ser 24 to Asp. This suggests that phosphorylation of CAX1 may activate Ca 2ϩ transport. Some of the positively charged Arg residues also appear to be involved in the autoinhibition mechanism. Substitution of these residues to Ala, thereby removing their charge, activated CAX1 (Fig. 6). Binding of a regulatory protein to the CAX1 NRR may prevent autoinhibition and activate FIG. 7. Regulatory domains of CAX1. A, topology of CAX1. The positions of the NRR, the regulatory-dependent region (RDR), the CaD, and the acidic motif are shown on a hydropathy plot of CAX1. The NRR and acidic motif are predicted to be on the cytosolic side. Numbers in bold denote the predicted transmembrane spans. The N terminus of CAX1 (boxed) is shown (below) in a partial amino acid sequence alignment with CAX3 produced using ClustalW 1.8. Identical residues are black; similar residues are gray. Important residues in the NRR of CAX1 are highlighted by an asterisk (Ser 25 ), a black square (Thr 33 ), and black circles (Arg 26 , Arg 29 , and Arg 32 ). The arrow denotes the start codon of sCAX1 and sCAX3. B, model of CAX1 regulation by the NRR. Ca 2ϩ /H ϩ antiport activity of CAX1 is inhibited following autoinhibition by the NRR. The NRR interacts with an N-terminal region (residues 37-73) that includes seven residues (shaded circles) termed the RDR that are required for the interaction to occur. Five residues have been identified in the NRR, Ser 25 , Arg 26 , Arg 29 , Arg 32 , and Thr 33 (shaded squares), which may have a role in this interaction. Following an appropriate stimulus, a conformational change occurs that prevents the NRR from interacting with the RDR, thereby activating CAX1. When the N terminus is truncated (as for sCAX1) protein translation occurs from Met 37 (black circle), and the antiporter is constitutively active because of the absence of interaction with the RDR. CAX1. Some of the Ser, Thr, and Arg residues highlighted by this work may be candidates for part of a binding site for regulatory proteins. The identification of activators for CAX1 will enable us to understand how it is regulated and how it might relate to a particular Ca 2ϩ signal transduction pathway.
The mechanism of CAX1 autoinhibition appears to involve a physical interaction between two discreet domains, the NRR and residues C-terminal to this region (Fig. 7). In order to show this interaction we used two different approaches. Initially, a series of chimeric CAX constructs were used to discriminate the region of CAX1 that interacts with the CAX1 NRR. Peptide inhibition studies showed that the NRR peptide could inhibit any sCAX construct containing the N-terminal 7 amino acids from sCAX1. When N-terminal chimeras using other CAX transporters were analyzed, no peptide inhibition was seen. However, when 7 amino acids from CAX1 (residues 56 -62) were inserted into sCAX3-9 at the N terminus, the CAX1 peptide could inhibit sCAX3-9 mediated Ca 2ϩ /H ϩ transport. Furthermore, when the equivalent 7-amino acid region of CAX3 was swapped into sCAX1, peptide inhibition of Ca 2ϩ /H ϩ transport was abolished. That this construct could still transport Ca 2ϩ confirms that this region is not required for transport. CAX1 containing the CAX3 7-amino acid region (CAX1-7aa) was no longer autoinhibited, and this construct was able to suppress the Ca 2ϩ sensitivity of the K667 yeast mutant. This chimeric CAX/CAX1-NRR approach was informative but rather unconventional. We also used yeast two-hybrid analysis to confirm an interaction between the CAX1 NRR and the CAX1 7-amino acid region by using a hydrophilic construct (residues 44 -69) of sCAX1 (Fig. 3D). Similar results showed an interaction with the N terminus of sCAX1 (residues 37-222) using a fragment that contained some hydrophobic regions (data not shown). Surprisingly, no strong interaction was observed for CAX1 NRR with the construct containing the CAX1 7-amino acid region swapped into CAX3. This may be because of inappropriate folding of this protein. We term this 7-amino acid region the regulatory-dependent region. Our findings establish that these 7 amino acids are required for autoinhibition, although this analysis does not conclude that these are the only amino acids that interact with the N-terminal 36 amino acids. Given that CAX3 and CAX1 are very similar and many of the amino acids in their N termini are identical (Fig. 7A), other residues may be involved in this interaction. Further work is required to determine which residues in the NRR are required for interaction with its binding site. Residues Ser 25 , Arg 29 , Arg 32 , and Thr 33 appear to be important, indicating that the C-terminal end of the NRR could contain the autoinhibitory binding site (Fig. 7). However, we have previously shown that removal of the first ten residues of the NRR prevents inhibition (8), indicating that the structure of the entire NRR may be important for inhibition.
Like CAX1, some plant Ca-ATPases with truncations at the N terminus are able to suppress yeast mutants that are defective in endomembrane Ca 2ϩ transport; however, the N-terminal autoinhibitory mechanisms of plant Ca-ATPases appear to drastically differ from that proposed for CAX1. Arabidopsis Ca-ATPases ACA2, ACA4, and Brassica BCA1 are regulated by calmodulin binding or phosphorylation of an N-terminal autoinhibitory domain (26 -29). A calmodulin-binding sequence is present within the N-terminal domain of ACA2, and the fulllength pump demonstrates calmodulin-stimulated Ca 2ϩ transport in yeast (29,30). CAX1, however, is not activated by calmodulin (8). A Ca 2ϩ -dependent protein kinase-binding site is present in the N terminus of ACA2 that phosphorylates a Ser near the calmodulin-binding site. This kinase activity inhibits ACA2 activity (27). However, our results suggest that CAX1 may be activated rather than repressed by phosphorylation (Fig. 4). Autoinhibitory domain peptides of Ca-ATPases have been shown to inhibit truncated and full-length calmodulinactivated pumps (30,31). The degree of inhibition in these studies is similar to what we have observed for CAX1 (Fig. 1). A peptide corresponding to the autoinhibitory domain of ACA2 inhibited 50% of activity (IC 50 ) of truncated ACA2 at 4 M and increased the pump's K m for Ca 2ϩ by 2-fold (30). Similarly, a peptide corresponding to the autoinhibitory domain of rabbit PMCA2b had an IC 50 of 1 M (31). The CAX1-NRR peptide only inhibited CAX1 (Fig. 1A). In contrast, peptides derived from some mammalian and plant Ca-ATPases were not as specific (30,32). In these studies, a Ca-ATPase autoinhibitory domain peptide could inhibit different Ca 2ϩ transporters, including the Ca 2ϩ /Na ϩ exchanger (32). Similarly, a peptide corresponding to the autoinhibitory domain of ACA2 could inhibit activity of another Arabidopsis Ca-ATPase ECA1 (30). Now that we have identified domains important for CAX1 autoinhibition, future work must assess possible conformational changes associated with autoinhibition. In animal Ca-ATPases, the autoinhibitory domain affects activity by causing self-association of the transporter (6). In some plant Ca-AT-Pases, the autoinhibitory domain binds to a stalk region and hydrophilic loops near the N terminus (7). Regardless of the specifics of inhibition, activation of the Ca-ATPases in plants and animals requires the sensing of Ca 2ϩ fluxes that then activate calmodulin.
Our previous identification of a longer open reading frame of CAX1 suggested that the original N-terminal-truncated sCAX1 is most likely encoded by a partial-length cDNA. Thus, we concluded that only a truncated and deregulated version of CAX1 could suppress the Ca 2ϩ sensitivity of the yeast mutant K667 (8). In this study we have shown how mutagenesis of single amino acids can be used to determine some of the mechanisms of autoinhibition. We have shown how to create a deregulated and activated Ca 2ϩ /H ϩ antiporter. We have also shown how a peptide corresponding to the autoinhibitory domain can specifically inhibit the activity of deregulated CAX1 both in yeast and in Arabidopsis. This is the first report of peptide inhibition of Ca 2ϩ /H ϩ antiport activity. These findings suggest that it may be possible to temporally and spatially engineer expression of CAX1 NRR peptides in plants as a means to modulate CAX1-mediated Ca 2ϩ transport.