The Protein Kinase SOS2 Activates the Arabidopsis H+/Ca2+ Antiporter CAX1 to Integrate Calcium Transport and Salt Tolerance*

The regulation of ions within cells is an indispensable component of growth and adaptation. The plant SOS2 protein kinase and its associated Ca2+ sensor, SOS3, have been demonstrated to modulate the plasma membrane H+/Na+ antiporter SOS1; however, how these regulators modulate Ca2+ levels within cells is poorly understood. Here we demonstrate that SOS2 regulates the vacuolar H+/Ca2+ antiporter CAX1. Using a yeast growth assay, co-expression of SOS2 specifically activated CAX1, whereas SOS3 did not. CAX1-like chimeric transporters were activated by SOS2 if the chimeric proteins contained the N terminus of CAX1. Vacuolar membranes from CAX1-expressing cells were made to be H+/Ca2+-competent by the addition of SOS2 protein in a dose-dependent manner. Using a yeast two-hybrid assay, SOS2 interacted with the N terminus of CAX1. In each of these yeast assays, the activation of CAX1 by SOS2 was SOS3-independent. In planta, the high level of expression of a deregulated version of CAX1 caused salt sensitivity. These findings suggest multiple functions for SOS2 and provide a mechanistic link between Ca2+ and Na+ homeostasis in plants.

In eukaryotes, transporters on both the plasma membrane and endomembranes play an important role in specifying the precise concentration of various ions within the cytosol. However, how organisms balance the concentration of ions and how multiple transporters are coordinately regulated on different membranes is poorly understood.
In plants, the balance of Ca 2ϩ and Na ϩ ions is an important component in salt tolerance (1). It has been well documented that Ca 2ϩ has at least two roles in salt tolerance, a signaling function leading to adaptation and a direct inhibitory effect on Na ϩ entry into the cell (2). Three genetically linked Arabidopsis loci (SOS1, SOS2, and SOS3) are components of a pathway that controls ion homeostasis and salt tolerance (3). SOS3 encodes a Ca 2ϩ -binding protein with sequence similarity to the regulatory subunit of the protein phosphatase calcineurin (4).
SOS2 encodes a serine/threonine kinase whose function is essential for salt tolerance (5,6). SOS2 physically interacts with SOS3 in a Ca 2ϩ -dependent manner to modulate activity of the plasma membrane H ϩ /Na ϩ antiporter SOS1 (7,8). Although SOS2 has been well studied as a regulator of Na ϩ extrusion from the cells, it is not known whether SOS2 modulates cytosolic Ca 2ϩ levels, and if so, nothing is known regarding the mechanism by which it does this.
In yeast, calcineurin is a key intermediate in signaling salt stress, controlling both cytosolic Na ϩ and Ca 2ϩ levels (9). For example, calcineurin regulates the transcription of ENA1, which encodes a plasma membrane Na ϩ efflux transporter. Calcineurin also negatively regulates the Saccharomyces cerevisiae vacuolar H ϩ /Ca 2ϩ antiporter VCX1 (10). VCX1 is an important transporter that regulates the amplitude and duration of cytosolic Ca 2ϩ levels (11). The Arabidopsis orthologs of VCX1 are the cation exchanger (CAX) 1 transporters. Arabidopsis thaliana appears to have up to 12 putative CAX transporters (CAX1-CAX11, and MHX) (12). Initially, CAX genes were cloned by their ability to suppress the Ca 2ϩ hypersensitive phenotype of a S. cerevisiae mutant (13). The activity of CAX1 appears to be regulated by an N-terminal autoinhibitory domain that was absent from the initial clone characterized by heterologous expression in yeast (14). Ectopic expression of deregulated CAX1 (termed sCAX1, missing the N-terminal autoinhibitor) in tobacco causes numerous stress sensitivity phenotypes (15); however, little is known regarding how any CAX transporter is regulated and how this regulation is synchronized with the mechanisms that mediate salt tolerance (16).
Plant Materials and Growth Conditions-Conditions for growth of A. thaliana wild-type (Columbia ecotype), cax1-1, cax1-1-expressing sCAX1, sos2-1, and sos3-1 seeds on agar media were as described (21). The ion sensitivity assay was carried out as described (7). Results presented are for growth after 7 days on the ion-containing media.
Preparation of Yeast Vacuolar Membrane Vesicles and Ca 2ϩ Transport Assay-Vacuolar-enriched membrane vesicles were prepared from vector alone, CAX1-expressing, and sCAX1-expressing K667 yeast as described previously (16,22). Measurements of H ϩ / 45 Ca 2ϩ antiport activity were determined as described previously (22). To measure Ca 2ϩ transport in the presence of GST-SOS2 or GST-SOS3 fusion proteins, membrane vesicles were incubated with or without the fusion protein for 15 min at 25°C prior to the addition of 1 mM Mg 2ϩ -ATP. The reactions were equilibrated at 25°C for a further 5 min then uptake was initiated by the addition of 10 M 45 CaCl 2 .

RESULTS
SOS2 Activates CAX1-Full-length CAX1 expressed in a pmc1vcx1cnb1 S. cerevisiae strain (K667), which is hypersensitive to Ca 2ϩ , grows well in standard media but not in media containing high levels (200 mM) of Ca 2ϩ because of the absence of endogenous vacuolar Ca 2ϩ transporters and the lack of activity of full-length CAX1 in yeast ( Fig. 1A) (22). H ϩ /Ca 2ϩ antiport activity of CAX1 is regulated through a hydrophilic N-terminal tail (14,16,22). Only deregulated CAX1 (sCAX1), lacking the first 36 amino acids at the N terminus, can suppress the yeast vacuolar Ca 2ϩ transport defect ( Fig. 1A) (13,14). We tested whether SOS2 could allow full-length CAX1expressing pmc1vcx1cnb strains to grow in media containing high levels of Ca 2ϩ . Indeed, strains expressing both CAX1 and SOS2 grew in a manner similar to sCAX1 (Fig. 1A). Given that SOS2 activated CAX1, we were interested in determining if SOS2 could activate other plant CAX transporters. As shown in Fig. 1A, in this yeast assay, SOS2 could not activate full-length CAX2, CAX3, or CAX4. We were also interested in determining whether SOS3, one of the other components in the SOS pathway that is required for the activation of the H ϩ /Na ϩ antiporter SOS1, could also activate CAX1 in yeast. As shown in Fig. 2, co-expression of SOS3 and CAX1 was unable to suppress the Ca 2ϩ hypersensitivity phenotype.
A purified GST-SOS2 fusion protein added directly to vacuolar membrane preparations from CAX1-expressing yeast cells caused a dose-dependent activation of H ϩ /Ca 2ϩ antiport (Fig.  1B). This activation was dependent on CAX1 because the addition of GST-SOS2 to membrane vesicles isolated from yeast expressing vector alone could not induce Ca 2ϩ transport (data not shown). This activation was also independent of SOS3 or the GST fusion protein, because GST-SOS2/GST-SOS3 activated CAX1 in a manner similar to the addition of only GST-SOS2, and purified GST-SOS3 alone did not activate CAX1mediated H ϩ /Ca 2ϩ transport (data not shown). Moreover, other CAX transporters could not be activated by purified GST-SOS2. The kinase activity of SOS2 has a preference for Mn 2ϩ (24), but the addition of various concentrations of Mn 2ϩ had no effect on the GST-SOS2 activation of H ϩ /Ca 2ϩ antiport activity mediated by CAX1 (data not shown).
Properties of SOS2 Activation of CAX1-As an initial step toward deducing the specificity and mechanism of SOS2 activation of full-length CAX1, we tested whether SOS2 could activate chimeric sCAX constructs. A CAX1 homolog called CAX3 is unable to suppress the K667 Ca 2ϩ sensitivity phenotype, even if its N terminus is truncated to give sCAX3 (17,25). Neither sCAX3 alone nor sCAX3 co-expressed with SOS2 was able to suppress the yeast Ca 2ϩ sensitivity (Fig. 3A). We have previously generated various chimeric constructs between sCAX1 and sCAX3, some of which retain the sCAX1-like property of being able to transport Ca 2ϩ whereas others, such as sCAX3, are inactive despite all chimeras having equivalent protein stability (17). SOS2 co-expressed with an inactive sCAX3/sCAX1 chimeric construct that contained the central region and C terminus of CAX1 (construct sCAX1-␣) was also unable to suppress the Ca 2ϩ -sensitive phenotype (Fig. 3B). However, when another inactive chimeric construct was used that contains 36 amino acids of the N-terminal region of CAX1 (Met 37 -Leu 73 ) fused to the central region and C terminus of CAX3 (construct sCAX3-␣1), SOS2 could activate this chimeric construct (Fig. 3B).
In a similar manner, we analyzed how various SOS2 mutants (Fig. 4A) altered the activation of CAX1 by observing the growth of K667 yeast containing CAX1 and SOS2 (Fig. 4B). In these assays, we have increased the level of CaCl 2 in the media to accentuate differences between the levels of activation mediated by the SOS2 variants (250 mM CaCl 2 versus 200 mM CaCl 2 ) (Fig. 4B). An SOS2 mutant that is inactive for kinase activity due to a point mutation Lys 40 -Asn (construct SOS2A) (5) could activate CAX1, although to a slightly weaker degree to wild-type SOS2 (Fig. 4B); however, a deletion in the entire kinase domain of SOS2 (construct SOS2T1) is not capable of strongly activating CAX1 but gives only very weak yeast growth. A constitutively active SOS2 kinase can be generated by a Thr 168 -Asp change in the activation loop of SOS2 or by a deletion of the C-terminal portion of the protein (18). Combining the Thr 168 -Asp mutation with an SOS2 truncation in the regulatory domain resulted in a kinase (construct T/DSOS2/ 308) that is more active than the previous single mutant version (18); however, the activation of CAX1 by this kinase mutant was much weaker than full-length SOS2 (Fig. 4B). An SOS2 mutation that is without the domain for SOS3 binding, the FISL domain (construct SOS2D1), also is capable of activating CAX1, as strongly as wild-type SOS2.

FIG. 3. SOS2 activates CAX chimeras and CAX1 mutants. A-C,
suppression of the yeast vacuolar Ca 2ϩ transport defect mutant by CAX chimeras or CAX1 mutants activated by SOS2. K667 yeast cells coexpressing various plasmids as indicated were grown and assayed as described in Fig. 1A. The CAX1 open reading frame is shown as a solid bar whereas CAX3 is shown as an open bar (not to scale).

FIG. 4. Activation of CAX1 by SOS2 variants.
A, schematic diagram of the domain structure for SOS2 and SOS2 deletion variants. JD, junction domain; FISL, FISL motif; SOS2A, a mutant harboring Lys 40 -Asn mutation, which abolishes SOS2 kinase activity; SOS2T1, a mutant lacking the kinase domain; SOS2D1, a mutant in which the 21amino acid FISL motif was deleted; T/DSOS2/308, a mutant lacking the regulatory domain and having a Thr 168 -Asp change within the kinase activation loop. B, K667 yeast cells co-expressing various plasmids as indicated were grown and assayed on YPD media and YPD media supplemented with 250 mM CaCl 2 . Plasmids used for expression of the SOS2 mutant proteins were: p2UGpd-SOS2A, p2UGpd-SOS2D1, p2UGpd-SOS2T1, and p2UGpd-T/DSOS2/308 (18).

SOS2 Associates with the N Terminus of CAX1 in Yeast-
To determine whether this activation was caused by a physical interaction between SOS2 and the CAX1 N terminus, a yeast two-hybrid experiment was performed. SOS2 co-expressed with the N terminus of CAX1 (Met 37 -Leu 73 or Met 1 -Asn 65 ) activated the ␤-galactosidase reporter, suggesting a direct interaction between SOS2 and the CAX1 N terminus (Fig. 5).
Expression of CAX1 and Salt Stress-Disruption of CAX1 by a transposon insertion (cax1-1) did not significantly affect sensitivity to salt stress (21) (Fig. 6). When sCAX1 was expressed at high levels under the control of the cauliflower mosaic virus (CaMV) 35S promoter in a cax1 mutant, the plants were saltsensitive (Fig. 6). However, this sensitivity is not as severe as the growth defects seen in sos2 and sos3 mutants.

DISCUSSION
When plants are challenged with salinity stress, an increase in the concentration of Ca 2ϩ often can ameliorate the inhibitory effect on growth (1,2). However, the underlying mechanisms have remained largely unexplained. Using a yeast-based approach, we have identified SOS2, an important regulator of Na ϩ transporters, also as a regulator of cytosolic Ca 2ϩ levels ( Fig. 1). Previously, the components of the SOS pathway (SOS1, SOS2, and SOS3) were shown to be Ca 2ϩ -activated (3). When the regulatory protein SOS3 binds Ca 2ϩ , SOS3 binds SOS2 and activates the kinase to modulate the plasma membrane H ϩ /Na ϩ transporter SOS1. An important conclusion from this work is that SOS2 may have functions in plant signaling that are independent of SOS3 and SOS1 because we demonstrate that SOS2 may also act in an SOS3-independent manner to directly regulate the vacuolar H ϩ /Ca 2ϩ antiporter CAX1.
In yeast, hyperosmotic stress caused by NaCl and other stresses induce an immediate increase in cytosolic Ca 2ϩ levels that activate the phosphatase calcineurin to mediate ion homeostasis and salt tolerance (26,27). Calcineurin is composed of a regulatory and catalytic subunit that is stimulated by Ca 2ϩ / calmodulin (28) to regulate many transporters including the vacuolar H ϩ /Ca 2ϩ transporter VCX1 (10). VCX1 functions in Ca 2ϩ sequestration much more efficiently when calcineurin is inactivated, which may be important for the production of Ca 2ϩ signals (10). This work helps further establish that the SOS pathway of Arabidopsis is the functional equivalent of the yeast calcineurin pathway. Like calcineurin, SOS2 may directly participate in the regulation of both Na ϩ transporters (SOS1) and Ca 2ϩ transporters (CAX1).
The inability of SOS2 to activate CAX2, CAX3, and CAX4 suggests specificity in the SOS2/CAX1 interaction. Three different experimental approaches suggest this specificity depends on multiple domains within the N terminus of CAX1. The first approach utilized chimeric sCAX3/sCAX1 constructs. Constructs that contain the N terminus of CAX1 (Met 37 -Leu 73 ) fused to CAX3 fail to transport Ca 2ϩ when expressed in yeast, despite this construct lacking amino acids 1-36 encoding the putative autoinhibitory domain, but the co-expression of SOS2 in these strains allowed some growth on high Ca 2ϩ -containing media (Fig. 3B). This indicates that the N terminus of CAX1, at least amino acids 37-73 but not the equivalent region of CAX3, is required for the activation of Ca 2ϩ transport by SOS2. In the second approach, alterations in the N terminus of CAX1, which do not perturb autoinhibition, were still activated by expression of SOS2 (Fig. 3C). Finally, two-hybrid analysis demonstrated an interaction between SOS2 and the N terminus of CAX1, demonstrating that amino acids 1-73 of CAX1 are involved in the interaction with SOS2 (Fig. 5). Further work will be needed to address the complexities of the SOS2/CAX1 interaction. For example, further yeast two-hybrid analysis with the CAX1 N terminus and different regions of SOS2 will identify the domains of SOS2 involved in this interaction. Our initial studies suggest multiple regions of the CAX1 N terminus interact with SOS2. For example, the significance of Ser 10 and Ser 25 in the CAX1 N terminus for SOS2 activation is evident by the inability of SOS2 to activate CAX1 when these residues have been mutated (Fig. 3C). At the same time, the two-hybrid data demonstrate the potential for a physical interaction between these two proteins when the region containing both Ser 10 and Ser 25 were deleted (Fig. 5).
The independence for the SOS2 kinase activity in this interaction was reinforced by the ability of mutant forms of SOS2 to activate CAX1 (Fig. 4). The 446 amino acid kinase contains a 267-amino acid N-terminal catalytic domain that is similar in sequence to the yeast sucrose nonfermenting 1 kinase and the FIG. 5. Interaction of SOS2 with the N terminus of CAX1. Interaction of SOS2 with the CAX1 N terminus was examined by yeast two-hybrid analysis. Y190 yeast cells co-expressing SOS2 in pACT2 and the CAX1 N-terminal fragments CAX1-N(37-73aa) and CAX1-N (1-65aa) in pAS2 were grown on synthetic complete medium lacking His, Trp, and Leu and assayed for LacZ expression. Results are shown from ␤-galactosidase assays on a filter. mammalian AMP-activated protein kinase (5,6). The C-terminal regulatory domain interacts with the kinase domain to cause autoinhibition. Mutants that abolish SOS2 kinase activity (SOS2A) still activated CAX1 (Fig. 4). Likewise, SOS2 mutants that do not contain the kinase domain (SOS2T1) also activated CAX1, albeit very weakly. That the activation of CAX1 Ca 2ϩ transport activity by GST-SOS2 did not require Mn 2ϩ , previously demonstrated to be preferential for kinase activity by SOS2, further suggests kinase independence. Further studies are required to determine whether the SOS2/ CAX1 interaction is Ca 2ϩ -dependent. This work also demonstrates that SOS2 is not always dependent on SOS3 for all of its functions. The SOS3 independence of this SOS2 activation of CAX1 was demonstrated with regulatory domain mutants that do not interact with SOS3 (SOS2D1) but still activated CAX1. Similarly, the GST-SOS2 activation of Ca 2ϩ transport did not require GST-SOS3 (Fig. 1B, data not shown). The addition of a fusion protein to SOS2 and SOS3 does not affect SOS2/SOS3 interaction or function (7); therefore the apparent lack of further activation or inhibition by GST-SOS3 was not due to GST. Because SOS3 targets SOS2 to the plasma membrane to allow activation of the SOS1 H ϩ /Na ϩ antiporter (29) and SOS2 must be recruited to the tonoplast to activate CAX1, it is not unexpected that this CAX1 activation is SOS3-independent. Arabidopsis has a number of SOS3 homologs, and it is conceivable that one of these may interact with SOS2 in this CAX1 pathway (see below).
Although SOS2 clearly interacts with CAX1 at the N terminus in our yeast assays, we cannot discriminate between gene products that activate or repress CAX1 transport in plants. Previously, we have shown that minor modification to the N terminus of CAX1 or CAX4 activates these Ca 2ϩ transporters in yeast (30). Thus, it remains a formal possibility that SOS2 negatively regulates CAX1 in planta. The physiological significance of the interaction between SOS2 and CAX1 thus requires further investigation.
Here we have demonstrated that if CAX1 is deregulated, such as by ectopic expression of sCAX1, the elevation of salt stress induced in cytosolic Ca 2ϩ may be buffered by sCAX1, thereby partially disrupting the SOS salt tolerance pathway (Fig. 6). This observation suggests that CAX1 may be autoinhibited during salt stress. In sos2 lines, there does not appear to be a significant alteration in CAX1-mediated Ca 2ϩ transport (data not shown); however, recent work suggests multiple factors regulate CAX1, and other CAX1 activators (16) may be able to compensate for loss of SOS2.
An important question is whether the various roles of SOS2 are coordinately regulated. One possibility is that the Ca 2ϩ -dependent binding of SOS2 to another SOS3-like Ca 2ϩ -binding protein (SCaBP) different from SOS3 could counteract the interaction of SOS2 and CAX1 in planta, thereby permitting autoinhibition of the transporter during the salt-induced Ca 2ϩ transient. Multiple SCaBPs (SOS3 and SCaBP1-SCaBP6) have been identified in Arabidopsis (18), and future work will focus on the biological significance of the interactions between various SCaBPs and SOS2.
The application of purified GST-SOS2 fusion protein activated H ϩ /Ca 2ϩ transport in tonoplast preparations from CAX1expressing yeast cells but not vesicles expressing other CAX transporters (Fig. 1B, data not shown). We have previously identified other proteins that can interact with the N terminus of CAX1 and alter CAX1 function, CAX interacting proteins (16). The dose-dependent nature of SOS2 activation portends the ability to quantify the amount of H ϩ /Ca 2ϩ transport per g of protein among different CAX interacting proteins and to more precisely determine which domains within SOS2 mediate this activation.
SOS2 is a member of a large gene family in Arabidopsis composed of 25 protein kinases (SOS2 and PKS1 to PKS24) that are related to sucrose nonfermenting 1 from yeast (SnRK3 family) (31). SOS2 is not by itself a membrane-associated protein and may be localized to the plasma membrane through its interaction with SOS3 (29). Our model is that SOS2 or a closely related SnRK interacts with an SOS3-like protein (SCaBP) to target the kinase to the tonoplast to regulate CAX1 activity. Our working hypothesis is that the kinase-independent SOS2 regulation of CAX transport may be occurring in the absence of a salt stress.
In summary, these studies have identified SOS2 as a direct regulator of the vacuolar H ϩ /Ca 2ϩ transporter CAX1. We postulate that SOS2 responds to environmental cues to differentially regulate both Na ϩ transporters (SOS1) and Ca 2ϩ transporters (CAX1). These findings therefore provide a mechanistic link between Na ϩ and Ca 2ϩ homeostasis in plants. In the future, we plan to further characterize the nature of CAX regulation to elucidate the role of vacuolar Ca 2ϩ sequestration in stress responses.