Loss of halophytism by interference with SOS1 expression.

The contribution of SOS1 (for Salt Overly Sensitive 1), encoding a sodium/proton antiporter, to plant salinity tolerance was analyzed in wild-type and RNA interference (RNAi) lines of the halophytic Arabidopsis (Arabidopsis thaliana)-relative Thellungiella salsuginea. Under all conditions, SOS1 mRNA abundance was higher in Thellungiella than in Arabidopsis. Ectopic expression of the Thellungiella homolog ThSOS1 suppressed the salt-sensitive phenotype of a Saccharomyces cerevisiae strain lacking sodium ion (Na(+)) efflux transporters and increased salt tolerance of wild-type Arabidopsis. thsos1-RNAi lines of Thellungiella were highly salt sensitive. A representative line, thsos1-4, showed faster Na(+) accumulation, more severe water loss in shoots under salt stress, and slower removal of Na(+) from the root after removal of stress compared with the wild type. thsos1-4 showed drastically higher sodium-specific fluorescence visualized by CoroNa-Green, a sodium-specific fluorophore, than the wild type, inhibition of endocytosis in root tip cells, and cell death in the adjacent elongation zone. After prolonged stress, Na(+) accumulated inside the pericycle in thsos1-4, while sodium was confined in vacuoles of epidermis and cortex cells in the wild type. RNAi-based interference of SOS1 caused cell death in the root elongation zone, accompanied by fragmentation of vacuoles, inhibition of endocytosis, and apoplastic sodium influx into the stele and hence the shoot. Reduction in SOS1 expression changed Thellungiella that normally can grow in seawater-strength sodium chloride solutions into a plant as sensitive to Na(+) as Arabidopsis.

Accompanying the production and accumulation of osmolytes and other protective molecules, an important aspect of plant responses leading to salt stress tolerance is the regulation of uptake, reexport, and control over the distribution of sodium ions (Na + ; Hasegawa et al., 2000;Tester and Davenport, 2003). Na + appear to enter the root by several pathways (Essah et al., 2003;Pardo et al., 2006), although the nature of participating genes and their interaction in pathways require further investigation. Once Na + has entered the root endodermis, a tissue that represents a barrier to ions (Peng et al., 2004), it is generally assumed that the ion enters the xylem following the movement of water to aerial parts of the plant. Despite substantial efflux of Na + across the plasma membrane of root cells, the net flux of Na + is unidirectional from soil to roots and then to the shoot, except for possible recirculation via the phloem . In a range of species, the severity of damaging symptoms is positively correlated with the content of Na + reaching photosynthetic tissues (Davenport et al., 2005;Ren et al., 2005;Munns et al., 2006). However, halophytic species can accumulate very high amounts of Na + in vacuoles, such that Na + may account for most of the total cellular osmotic potential , and the presence of Na + accelerates growth in euhalophytes to some degree (Adams et al., 1998). Emerging as the major advantage of halophytes appears to be their exceptional control over Na + influx combined with export mechanisms, the ability to coordinate its distribution to various tissues, and efficient sequestration of Na + into vacuoles. These characteristics are of particular advantage when plants are subjected to a sudden increase of Na + salts in their environment (Hasegawa et al., 2000), whereas gradual increases in Na + may be tolerated even by plants that are not halophytic in nature.
Na + -ATPases, major Na + export systems in organisms such as fungi and the moss Physcomitrella patens, have not been found in higher plants (Lunde et al., 2007). In Arabidopsis (Arabidopsis thaliana), transporters of monovalent (alkali) cations, such as HKT1 (Berthomieu et al., 2003;Rus et al., 2004), members of the NHX family (Yamaguchi et al., 2005;Pardo et al., 2006), and SOS1 (for Salt Overly Sensitive 1; Shi et al., 2000Shi et al., , 2002Shi et al., , 2003, have been shown to play roles in the movement and distribution of Na + ions. Studies have shown the involvement of nonselective ion channels with roles in the transport of Na + ions, but the genes encoding such function(s) have not been identified (Demidchik and Maathuis, 2007). SOS1, whose deletion resulted in a strong salt-sensitivity phenotype in Arabidopsis, encodes a plasma membrane Na + /H + antiporter involved in removing Na + ions from cells (Shi et al., 2000). This efflux strategy, which may be sufficient for the survival of unicellular organisms, must be accompanied by other means of Na + confinement to avoid carryover of Na + between cells in futile cycles. Hence, the physiological role of a plasma membrane Na + /H + antiporter must be embedded in the context of tissue, organ, and whole plant distribution of ions and their transporters. A recent discovery on cell layer-specific differential responses to the salt stress of root cells supported this notion (Dinneny et al., 2008).
In Arabidopsis, the SOS1 gene is most strongly expressed in the epidermis of the root tip region and in cells adjacent to vascular tissues . Based on the salt concentration in shoot, root, and xylem sap of wild-type Arabidopsis and its sos1 knockout mutants, the SOS1 antiporter is assumed to function in Na + export under severe salt stress conditions . However, detailed knowledge about how a Na + excluder achieves salt tolerance in a multicellular eukaryote is still missing. Significantly also, even though SOS1 has been an intensely studied component of the ion homeostasis mechanism, its involvement in the exceptional salt tolerance of halophytes is not known.
Thellungiella salsuginea (salt cress), which had before been called T. halophila by us, is a close relative of Arabidopsis, which has become a model to study the genetic basis of this plant's extreme tolerance to a variety of abiotic stress factors, including salinity (Inan et al., 2004;Gong et al., 2005;Vera-Estrella et al., 2005;Volkov and Amtmann, 2006;Amtmann, 2009). Thellungiella lacks specialized morphological structures, such as salt glands or large sodium storage cells found in other halophytes, making it a useful model for studying stress tolerance mechanisms that could be applicable to further understanding or to embark on engineering of conventional crops (Inan et al., 2004). Recently, it has been reported that Thellungiella had lower net Na + uptake compared with Arabidopsis. The unidirectional influx of Na + ions to roots appeared to be more restricted and/or tightly controlled in Thellungiella than in Arabidopsis. To compensate for greater influx, Arabidopsis roots showed higher Na + efflux (Wang et al., 2006).
Here, we wished to explore the role(s) by which ThSOS1, the SOS1 homolog in Thellungiella, could be involved in shaping the halophytic character of the species using ectopic expression of the gene in yeast and in Arabidopsis and Thellungiella SOS1-RNA interference (RNAi) lines. The results identified ThSOS1 as a genetic element whose activity limits Na + accumulation and affects the distribution of Na + ions at high concentration, thus acting as a major tolerance determinant.

RESULTS
Thellungiella Expresses SOS1 at Higher Levels Than Arabidopsis ThSOS1 (EF207775) is most closely related to Arabidopsis SOS1 (AtSOS1; At2g01980) in the deduced amino acid sequences (83% identity) among SOS1 coding regions from other plants (Supplemental Fig.  S1). ThSOS1 exists as a single-copy gene in the Thellungiella genome (data not shown). Transcript abundance of SOS1 was compared between Arabidopsis and Thellungiella by reverse transcription (RT)-PCR. Reference genes were selected among genes more than 90% identical in their sequence identities between both species, while they showed unaltered expression levels under stress conditions (Czechowski et al., 2005). All primers were designed to be identical in both species. In Thellungiella, SOS1 mRNA was significantly more abundant than in Arabidopsis in both shoot and root, more prominently expressed in roots in the absence of salt stress, while the levels of reference gene expression were indistinguishable between the two species (Fig. 1A). SOS1 mRNA was quantified by real-time RT-PCR (Fig. 1B). Compared with wild-type Arabidopsis, ThSOS1 mRNA abundance was 2.9-and 7.6-fold higher under normal conditions in shoot and root, respectively, and up to 5.7-fold higher compared with salt-stressed Arabidopsis roots under stressed conditions. This difference in SOS1 mRNAs in both species persisted in older plants of similar mass and growth stages grown on artificial soil as described (Gong et al., 2005) under either nonsalinized or highly saline conditions (Supplemental Fig. S2). The Saccharomyces cerevisiae strain AXT3K (Dena1-4 Dnha1 Dnhx1), lacking major Na + transporters essential for tolerance of yeast, showed growth inhibition at Na + concentrations higher than 50 mM, while the wild type was not affected . Expression of ThSOS1 partially suppressed the salt sensitivity of AXT3K ( Fig. 2A). In Arabidopsis, AtSOS1 is activated by the SOS2/SOS3 protein kinase complex (Qiu et al., 2002;Quintero et al., 2002). Coexpression of Arabidopsis SOS2 and SOS3 in AXT3K together with ThSOS1 dramatically increased the salt tolerance of the transformed cells, leading to growth in medium with 400 mM sodium (Fig. 2B). Expression of ThSOS1 conferred salt tolerance to the yeast mutant at concentrations higher than those sustained by the expression of AtSOS1, both alone and after activation by SOS2/ SOS3 (Fig. 2, A and B). To test whether the unequal Na + tolerance was related to differential activity of AtSOS1 and ThSOS1, the Na + /H + exchange activity was measured in plasma membrane vesicles purified by twophase partitioning from yeast transformants (Fig. 2C). The strong inhibition (approximately 84%) of ATP hydrolysis by vanadate, an inhibitor of plasma membrane H + -ATPases, demonstrated that vesicle preparations were enriched in plasma membranes (data not shown). Cells were grown on selective Arg-phosphate medium containing 1 mM KCl and transferred to the same medium supplemented with 100 mM NaCl for 1 h to ensure activation of the SOS2/SOS3 kinase complex when present. Maximal Na + /H + exchange activity was observed in cells coexpressing SOS1 proteins and the Arabidopsis SOS2/SOS3 kinase complex (Fig. 2D). No significant differences were found in the Na + /H + exchange activity of plasma membrane vesicles containing ThSOS1 or AtSOS1 with and without coexpression of the SOS2/SOS3 kinase complex ( Fig.  2D; data not shown).
Transgenic Arabidopsis expressing ThSOS1 under control of the cauliflower mosaic virus-35S promoter were analyzed together with AtSOS1-overexpressing lines (Shi et al., 2003) for SOS1 expression and salt tolerance (Fig. 3). The transgenic lines showed various levels of SOS1 expression, highest in a ThSOS1-overexpressing line (Fig. 3A, line 30). The survival rates of the wild type and SOS1-overexpressing lines were quantified by rescuing the seedlings onto nonsaline medium after a brief stress. Seedlings that resumed growth were counted as survivors (Fig. 3B). The survival rates of SOS1-overexpressing lines were proportional to the level of SOS1 under stress (r 2 = 0.95), regardless of SOS1 origin, indicating that expression strength/mRNA stability or protein amount/activity are determining factors in tolerance acquisition. However, even the highest tolerance observed in line 30 ( Fig. 3B) did not approach that shown by wild-type Thellungiella (Inan et al., 2004).

RNAi-Based Reduction of ThSOS1 Expression Resulted in Decreased Salt Tolerance with Faster Na + Accumulation in Shoots
We have developed transgenic Thellungiella transformed with a ThSOS1 RNAi vector (Fig. 4A). Two lines, designated thsos1-4 and thsos1-6, were used for further analyses. Both showed 3:1 segregation in BASTA resistance and salt sensitivity, indicating a single insertion locus of the RNAi construct (data not shown).
Seedlings on plates (Fig. 4B) and mature plants grown in artificial soil (Supplemental Fig. S3A) showed a salt-sensitive phenotype in the RNAi lines. Shoots of RNAi line plants showed partial bleaching and root growth stopped at 200 mM NaCl, while the wild type and vector control (thsos1-11) continued to grow without symptoms in NaCl solutions up to 300 mM (Fig. 4, B and C). The abundance of ThSOS1 mRNA was determined by quantitative real-time Thellungiella (Th). A, Transcript abundance was compared by semiquantitative RT-PCR between Arabidopsis and Thellungiella for SOS1. Actin (ACT2/8), ribulose-bisP carboxylase small chain 1A (RBCS1A), and elongation factor 1-a (EF1a) were used as references (Czechowski et al., 2005). Treatment of 350 mM NaCl to Arabidopsis was not included because this stress condition is lethal. B, Quantification of SOS1 transcripts by real-time RT-PCR with error bars indicating SD from six repeats. All primers were designed against regions where the genes of the two species showed perfect identity to generate amplicons of identical length with more than 95% sequence identity between the two species.
PCR in roots and shoots with or without salt stress. The decline of SOS1 mRNA by RNAi was more prominent under salt stress, where thsos1-4 showed 30% and 84% and thsos1-6 showed 17% and 40% of wild-type expression in the shoot and root, respectively ( Fig. 4D). A similar reduction in SOS1 transcript amount persisted throughout development and at different strengths of NaCl (Supplemental Figs. S3B and S5). The vector control (thsos1-11) showed no significant difference in ThSOS1 expression from the wild type (data not shown).
To further characterize the role of ThSOS1 in the stress response, 2-week-old plants of wild type and thsos1-4 were transferred to 200 mM NaCl and fresh weight and water and ion contents of shoots were measured, using inductively coupled plasma-optical emission spectrometry. Wild-type shoots continued to gain weight, but thsos1-4 shoots showed no growth after day 4 (Fig. 5A). Water content was deduced from a comparison of the fresh and dry weights of seedling shoots. Salt-stressed wild-type shoots maintained water content slightly lower than the unstressed control throughout the experiments. However, a more significant decrease in water content in thsos1-4 shoots became apparent after 7 d (Fig. 5B). In contrast to the wild type, which showed a linear, gradual increase of Na + in the leaves, sodium content in thsos1-4 leaves peaked by day 4 and later declined, most probably by leakage due to the loss of water (Fig. 5C). Loss of potassium was observed in both the wild type and thsos1-4, with a seemingly insignificant higher decrease in thsos1-4 during later times of the experiment (Fig. 5D). These results already pointed to ThSOS1 as important and required for the ability to tolerate extremely high levels of NaCl.
The changes in ion content of stressed plants after removal of the stress were compared between the wild type and thsos1-4 ( Fig. 6). Hydroponically grown plants were treated by increasing NaCl stepwise to the nonlethal concentration of 150 mM within 2 weeks, transferred to medium without NaCl, and harvested to determine Na + and K + concentrations. Both shoots and roots of thsos1-4 showed higher Na + (Fig. 6, A and C) and lower K + (Fig. 6, B and D) contents compared with the wild type. In the wild type, shoot ion contents did not change over a 72-h period after removal of the plants from NaCl solutions, while thsos1-4 showed gradual decreases in Na + and increases in K + , eventually converging on the Na + contents observed in the wild type (Fig. 6, A and B). The roots of wild-type plants exhibited a sharp drop in Na + ion content within 1 h after removal of NaCl, while the efflux of Na + ions was much slower in thsos1-4 roots, which failed to reach the level of Na + content in the wild type at 72 h (Fig. 6C). To assess the contribution of ThSOS1 to the reduction of net sodium uptake in the short term, wild-type and thsos1-4 plants were transferred to hydroponic medium supplemented with 0.1, 1, and 10 Figure 2. Complementation of the yeast AXT3K mutant with ThSOS1. A, The yeast strain AXT3K (Dena1-4 Dnha1 Dnhx1) transformed with either AtSOS1 or ThSOS1 was grown on medium supplemented with the designated concentrations of salt. B, Coexpression of AtSOS1 or ThSOS1 with Arabidopsis SOS2 and SOS3. C, Na + /H + exchange activity in plasma membrane vesicles. Formation of pH gradient, acidic inside, was initiated with ATP. Addition of NaCl started Na + /H + exchange and fluorescence recovery. The reaction was terminated by adding 25 mM (NH 4 ) 2 SO 4 , which dissipated the pH gradient. One representative experiment is shown. D, Initial rates of Na + /H + exchange (means and SE; n = 3); units are percentage change of fluorescence (DF) per minute and per milligram of protein.

SOS1-Based Halophytism in Thellungiella
Plant Physiol. Vol. 151,2009 mM and the Na + content was determined after 15 and 30 min and 2 and 48 h. Roots of line thsos1-4 started to show significantly greater Na + contents than the wild type after 2 h of salt imposition (Supplemental Fig. S4).

Down-Regulation of ThSOS1 Resulted in Increased Sodium Accumulation in the Root Tip and Cell Death in the Elongation Zone
To determine the role of ThSOS1 in specific cell types in the roots, the distribution of sodium was imaged using the fluorescent, sodium-specific dye CoroNa-Green AM (Invitrogen), as outlined in "Materials and Methods," comparing wild-type and thsos1-4 seedling roots by confocal microscopy. Incubations in 150 mM NaCl did not result in rapid differences between the wild type and thsos1-4, but within 24 h thsos1-4 roots showed higher fluorescence in the meristematic region than the wild type (Fig. 7, A and B). When the vacuolar fluorescence intensities of root cortex cells from five individual plants from each line were quantified, the wild type showed a mean relative value of fluorescence intensity of 29.68 (SD = 9.43; arbitrary units), while the mean was 68.80 (SD = 16.89) for thsos1-4 (Fig. 7C).
Cellular events in the root tip region at this time point were observed by CoroNa Green AM and FM4-64 (Invitrogen). In the wild type, CoroNa Green specifically stained the prevacuolar compartment of irregular shape in the meristematic and expanding cells (Fig. 7D). Cells of thsos1-4 showed significantly stronger CoroNa-Green fluorescence in small round intracellular vacuoles, which converged into one or two large bodies within 24 h in medium with 150 mM NaCl (Fig. 7E). Interestingly, the endocytotic inclusion of the FM4-64 dye, which was apparent in the wild type (Fig. 7D, arrows in inset), was absent in thsos1-4 root cells (Fig. 7E, inset). Neither the shape of vacuoles nor endocytosis was affected under normal conditions in the wild type or thsos1-4 (data not shown).
Whereas wild-type plants did not show any symptoms even during longer term exposure to salt (Fig.  7F), increasing Na + -specific fluorescence in thsos1-4 was followed by cell death within and adjacent to the elongation zone, visualized by intracellular staining of propidium iodide (Fig. 7G, arrow). The number of thsos1-4 seedlings showing higher fluorescence and/or cell death at the root tip and elongation zone increased to two-third of the tested seedlings (10 of 15) within 36 h of stress, and the primary root of all thsos1-4 seedlings (10 of 10) had died within 48 h.

thsos1-4 Accumulated Sodium in the Root Stele after Long-Term Stress
To visualize Na + distribution in the stele, wild-type and thsos1-4 seedlings were incubated for longer times at low concentrations of CoroNa Green (see "Materials and Methods"). Fluorescein diacetate was used as a control in separate experiments to ensure that dyes penetrated to the vasculature (data not shown). In the wild type, Na + was confined in the vacuoles of epidermis, cortex, and, less pronounced, endodermis cells, and fluorescence was absent from pericycle cells and cells within the vasculature after treatment with 150 mM NaCl for up to 4 d (Fig. 8, A, B, E, G, and I). In contrast, cell damage in the elongation zone in thsos1-4 extended to the older sections of the root as the stress continued (Fig. 8, C and D), resulting in compromised cells with intensive intracellular staining of propidium iodide (Fig. 8F, arrow 1). Strongest CoroNa Green fluorescence was observed in cells adjacent to structurally compromised cells (Fig. 8F, arrow 2). While epidermis and cortex cells were either terminally damaged or failed to confine Na + to vacuoles, the Na + -specific fluorescence was also found in cells inside the pericycle (Fig. 8H, arrows). Within 4 d of incubation, pericycle cells revealed damage and CoroNa Green stained xylem vessels (Fig. 8J) in thsos1-4 at the low concentration of NaCl (150 mM), which was not recognized in wildtype plants as a stressful condition, as shown in Figure  8I and also suggested by microarray experiments (Gong et al., 2005).
The higher accumulation of Na + ions inside the endodermis of thsos1-4 was confirmed by scanning electron microscopy and energy-dispersive x-ray microanalysis (SEM-EDX) in the roots of mature plants (Fig. 9). After treatment with 250 mM NaCl for 2 d, thsos1-4 root accumulated more than twice the amount of Na + than wild-type root, concurrent with a dramatic decrease of K + in the vacuole of xylem parenchyma that resulted in a more than 12 times higher Na + -K + ratio. In contrast, cortex cells did not show a significant difference between the wild type and thsos1-4.

DISCUSSION
In Arabidopsis, the SOS pathway has been documented as an essential component of the ion homeostasis system. The known signal transduction components of the pathway, a complex of SOS2 and SOS3, control the activity of SOS1, a plasma membrane-localized Na + /H + antiporter (Qiu et al., 2002;Quintero et al., 2002;Chinnusamy et al., 2006). The importance of this mechanism in the glycophyte Arabidopsis notwithstanding (Shi et al., , 2003, questions remained about the relative impact of this pathway in a naturally salt-tolerant species. We used the halophytic Arabidopsis relative Thellungiella (Bressan et al., 2001) to address this question.

SOS1 Abundance Determines the Extent of Shoot Na + Accumulation
While the sequence of SOS1 is highly conserved between Thellungiella and Arabidopsis (Oh et al., 2007), a conspicuous difference emerged from comparisons of expression strength: transcript levels in Thellungiella were eight to 10 times higher in both nonsalinized and stressed states, and the saltdependent induction of expression or stabilization SOS1 mRNA (Chung et al., 2008) known for Arabidopsis was significantly higher in Thellungiella. Consistent with earlier reports (Kawasaki et al., 2001;Taji et al., 2004), prestress-elevated or constitutive high expression of stress-relevant genes could be the basis that determines relative abiotic stress tolerance differences between plants.
Complementation by SOS1 cDNAs of a yeast mutant lacking Na + transporters suggested that ThSOS1 can function in Na + exclusion more efficiently than AtSOS1, especially at higher levels of NaCl (Fig. 2, A  and B). Coexpression of AtSOS2/3 indicated that ThSOS1 was activated by Arabidopsis SOS signaling components (Fig. 2B), indicating conservation of the pathway in Thellungiella. However, no significant differences in Na + /H + exchange rates were found in plasma membrane vesicles from cells expressing AtSOS1 or ThSOS1 (Fig. 2D). These results indicate that the two highly conserved SOS1 proteins are substantially equivalent, and the long-term effect of differential specific activity or protein abundance, too subtle to be discriminated in transport ion assays, may become amplified over time to render cells with improved salt tolerance. This is in agreement with the correlation between the salt tolerance of Arabidopsis transgenic lines and the expression level of SOS1, regardless of the origin of the protein (Fig. 3).
Sodium fluxes into and out of Thellungiella roots have been studied in comparison with Arabidopsis (Wang et al., 2006). A higher Na + efflux in the roots of Arabidopsis was reported, which partly compensated for Na + influx. Still, Arabidopsis plants accumulated more Na + than Thellungiella (Wang et al., 2006). Consequently, it was proposed that limitation of Na + influx, not higher efflux, should be the main mechanism by which Thellungiella could achieve lower net Na + accumulation under salinized conditions in com- Figure 6. Comparison of Na + efflux characteristics between the wild type (WT) and thsos1-4. Hydroponically grown plants (30 d old) were subjected to stepwise increases of NaCl in growth medium to 150 mM over 2 weeks. After the removal of sodium from the medium, ion concentration in shoots and roots was monitored for 72 h. ANOVA results probing all measurements were significantly variable over time, except for the thsos1-4 root K + (P = 0.065). A, Na + in shoots. B, K + in shoots. C, Na + in roots. D, K + in roots. Figure 5. Growth, water, and ion contents of wild-type (WT) and thsos1-4 seedling shoots. Shoots of seven seedlings grown in the presence of 200 mM NaCl were pooled. The control (C) was from seedlings incubated for 4 d without salt stress. Error bars indicate SD of four independent replicates. Bars with different letters are significantly different at P , 0.05 (Tukey test). A, Average fresh weight of the shoot. B, Water content calculated as follows: (fresh weight 2 dry weight)/fresh weight 3 100 (%). C and D, Sodium (C) and potassium (D) contents of shoots. DW, Dry weight. parison with Arabidopsis. The staining by CoroNa-Green indicates Na + exclusion by the activity of SOS1 in specific regions of the root, rather than in all regions ( Figs. 7 and 8). Thus, SOS1 may not be directly involved in plant-level Na + efflux but rather may function in protecting the particularly vulnerable cells of the root elongation zone (Fig. 7, F and G). Indeed, older regions of the root are not characterized by higher Na + content in thsos1-4 roots, apart from the CoroNa-Green signal that indicated higher content in the xylem at later stages of salt stress (Fig. 8). The protection of the young root cells by a more abundant SOS1 in the halophyte may help to counteract Na + influx and, in turn, net Na + accumulation.
The RNAi-induced reduction of SOS1 led to faster leaf senescence accompanied by severe shoot water loss during salt stress (Figs. 4B and 5B; Supplemental  Fig. S3A). This water loss in RNAi lines was not based on impaired stomatal conductance, as the RNAi plants showed severe stress symptoms even at 100% humid-ity (data not shown). The phenotype was strictly Na + specific; the RNAi lines did not show differences compared with the wild type at K + concentrations up to 300 mM (Fig. 4B). The phenotype rather appeared related to the rate of Na + accumulation during the initial stages of exposure to high Na + and not to the absolute amount of Na + in the shoots after long-term exposure. Generally, tolerant species that have been categorized as Na + excluders accumulate large amounts of the ion over time, but this accumulation proceeds more slowly than in sensitive "sodiumincluding" species . Indeed, Na + accumulation in thsos1-4 was faster and appeared less controlled than in the wild type, which eventually contained more Na + than the RNAi line without adverse effects on growth at the moderate concentrations of NaCl used (Fig. 4C). Considering the positive correlation between SOS1 transcript abundance, the long-term overall high accumulation of Na + , and the control over the rate of accumulation during early stages of salt stress, Thellungiella behaves like a true halophyte and SOS1 expression appears to constitute an essential trait at the basis of halophytic growth. Observations of Na + efflux from plant tissues after removal from nonlethal concentrations of Na + confirmed this halophytic nature (Fig. 6). Thellungiella wild-type plants maintained nearly 300 mM of Na + in the shoot tissue even after removal of Na + from the medium, indicating that this halophyte might utilize ions as an osmoticum (Fig. 6A). In contrast, the higher Na + contents in the shoot of thsos1-4 converged over time to the levels in wild-type shoots (Fig. 6A), indicating accumulation of Na + ions in thsos1-4 shoots as an uncontrolled process requiring redistribution after the removal of external Na + . Under very low external Na + (0.1, 1, and 10 mM NaCl; Supplemental  Fig. S4), thsos1-4 roots took up more Na + than wildtype roots after 2 h of treatment. This, together with the slower Na + exclusion from roots in the RNAi lines (Fig. 6C), emphasized the conserved function of ThSOS1 in Na + exclusion/export under nonlethal stress conditions.

ThSOS1 Activity Leads to the Exclusion of Sodium from the Root Meristematic Region and Protects Cells of the Elongation Zone
A main objective was to observe the accumulation and distribution of Na + ions in cell lineages of the root, considering that we lack information on the genetic makeup that allows halophytic multicellular plants to achieve control over the rate of Na + accumulation. Sodium uptake was followed using the membranepermeable fluorescent dye CoroNa-Green that binds Na + ions only after it has been confined within cells (Meier et al., 2006), while propidium iodide staining permitted observations of membrane integrity.
Early during salt stress, thsos1-4 roots showed little differences when compared with the wild type in intensity or distribution of the Na + -specific fluorescence (data not shown). Observable changes eventually originated at the root tip region, which began to show stronger fluorescence signals in thsos1-4, indicating SOS1's function in Na + exclusion (Fig. 7, A-C). Propidium iodide, which stained cell walls and dead cells, revealed that increased Na + fluorescence was accompanied by a gradual loss of membrane integrity, initially confined to cells of the elongation zone (Fig.  7G), spreading to cells of the root hair zone over time (Fig. 8, C and F). It appears that Na + exclusion by SOS1 is most critical in cells that expand and consequently take up water, whereas cells closer to the quiescent center seem to be protected by the absence of large vacuoles. These observations correspond with results from a recent study that identified the beginning of the elongation zone as the most responsive to salt stress along the longitudinal axis of the primary root (Dinneny et al., 2008). Compromised membranes may result in increased apoplastic Na + flux and deposition of excess (compared with the wild type) Na + into vacuoles, which then appeared to initiate cell death in adjacent cells, where the strongest Na + -specific signals were typically observed (Fig. 7G). A chain reaction of cell death events accompanied the influx of Na + into the older part of the root as the stele began to accumulate more Na + . This behavior of thsos1-4 was in sharp contrast to that of wild-type plants, which confined Na + ions nearly exclusively to epidermis (and root cap) cells (Figs. 7F and 8, A and B). Only at concentrations higher than 350 mM NaCl in the medium did we observe the beginning deterioration of the root elongation zone in wild-type Thellungiella, while wild-type Arabidopsis showed this phenotype at greater than 180 mM NaCl (data not shown).
In roots incubated for an extended time (18 h) with the dye, wild-type plants showed significantly lower Na + -specific fluorescence in cells of the pericycle and the stele (Fig. 8, E, G, and I). This visual observation matched studies using x-ray microanalysis for the localization of Na + in roots and confirmed the root endodermis as a major barrier controlling ion influx into the stele (Peng et al., 2004;Ottow et al., 2005). This barrier function may be important under severe stress conditions, as it has recently been reported that under moderate stress (50 mM NaCl) this role appears to be satisfied by cells of the epidermis and cortex of wheat (Triticum aestivum;Läuchli et al., 2008). In strong contrast, the localization of Na + and cell viability staining in thsos1-4 roots clearly revealed higher uptake into the root vasculature (Fig. 8, F, H, and J) and greater movement most likely to the shoot under transpiring conditions. This identifies the crucial barrier at the pericycle/endodermis boundary, where indeed Arabidopsis showed strong expression of SOS1 in a construct expressing GUS under control of the AtSOS1 promoter . Interestingly, the appearance of Na + ion-specific fluorescence inside the pericycle of thsos1-4 occurred at the same time as root tip damage, suggesting that the protection of root tips by SOS1 may also contribute to the protection of the root stele from the intrusion of Na + ions. SEM-EDX analysis on roots of mature plants confirmed higher accumulation of Na + in the root stele and cells of the photosynthetic tissues of thsos1-4 ( Fig. 9), as observed by measuring ion contents (Figs. 5C and 6A) and monitoring Na + -specific fluorescence (Fig. 8) in younger plants.

ThSOS1, Endocytosis Protection, and Halophytic Adaptation
New evidence has revealed SOS1 and the SOS pathway with a function not only as a Na + exporter but as a mediator of intracellular Ca 2+ and pH homeostasis (Cheng et al., 2004;Shabala et al., 2005). Involve- Figure 9. SEM-EDX analysis for ion measurement in mature plants. Wild-type (Wt) and thsos1-4 plants were grown hydroponically for 3 weeks and treated with 250 mM NaCl for 2 d. Relative contents of K + and Na + in the indicated cell types were determined by EDX analysis as described in "Materials and Methods." A, Cross section of leaf. B, Close-up of the boxed region in A, showing central vein of leaf. C, Cross section of root. D, Close-up of the boxed region in C. Cell types identified are leaf palisade (pa), lagunar mesophyll (me), xylem parenchyma (xp), and root cortex (ct). E, Samples from both lines were processed simultaneously, and the quantitation was performed on four to six cells from each tissue type in two to three plants of each genotype. Values presented are percentages of total counts. Within pairwise comparisons, means followed by letters were statistically different at P , 0.1 (A), P , 0.01 (B), or P , 0.001 (C) by Fisher's LSD test. ment of SOS1 in regulating ROS metabolism through an interaction with RCD1, involved in radical-based signaling, through its long C-terminal cytoplasmic tail has been reported (Katiyar-Agarwal et al., 2006). Mutations in SOS1 and other SOS pathway components are known to affect aspects of root development, such as cortical microtubule organization and gravitropism (Sun et al., 2008) under salt stress. Considering the negative effect on endocytosis by a suppression of ThSOS1 expression, indicated by the abolishment of FM4-64 import in thsos1-4 root cells (Fig. 7E), protection of endocytosis may be suggested as an additional role of SOS1. In Arabidopsis, SOS1 affects cortical microtubule organization and endocytic vesicles are known to be transported along actin filaments and microtubules (Shoji et al., 2006;Soldati and Schliwa, 2006). In affecting endocytosis under salinity conditions, a lack of SOS1 could also interfere with other functions of endocytosis, including polar auxin transport (Dhonukshe et al., 2007), brassinosteroid signaling (Geldner et al., 2007), recycling of plasma membrane receptors and ion channels (Murphy et al., 2005;Sutter et al., 2007), and cytokinesis (Dhonukshe et al., 2006;Reichardt et al., 2007), and therefore result in the termination of root growth and cell death (Fig. 7F). SOS1 may be involved in endocytosis via the yet unknown interaction with a component or regulator of intracellular trafficking or via an indirect pathway including the regulation of cytosolic or vacuolar pH under salt stress (Shabala et al., 2005), which may then affect endomembrane and vesicle trafficking Shoji et al., 2006). SOS1 activity could protect endocytosis of cells in the root tip and elongation zone and ultimately sustain membrane integrity, thus providing an essential stopgap measure or temporary protective solution allowing for other defensive measures to become established in the plants. Supporting this view is the fact that cells in Thellungiella wild-type roots that developed after the plants had adapted to increased salinity showed a stronger fluorescence than cells that developed during the stress imposition period (Fig. 8B, compare areas 1 and 2). Apparently, the halophyte achieved adaptation in newly developed root cells within less than 2 d, which was absent from thsos1-4 roots that suffered extensive damage, cell death, and uncontrolled apoplastic Na + influx into the root stele and shoot (Fig. 8D). Stress "preparedness" by Thellungiella seems to extend to other functions, because the plant contains, compared with Arabidopsis, higher amounts of metabolites, including Pro, trehalose, inositols, and several organic acids and substantially unique yet unknown compounds (Taji et al., 2004;Gong et al., 2005;Oh et al., 2007) that are additionally salt stress inducible. Although Arabidopsis shows induction, often stronger than Thellungiella, for a number of putatively protective pathways, the ultimate accumulation of metabolites achieved by Arabidopsis is less than that seen in Thellungiella (Gong et al., 2005). Similarly, thsos1-4 shared with Arabidopsis the ab-sence of drastically increased levels of metabolites (Oh et al., 2007). By providing a temporal barrier to a sudden exposure to high Na + in the root cell elongation zone, ThSOS1 prohibited the onset of a chain reaction leading to cell death and apoplastic Na + influx into the shoots. ThSOS1 activity then resulted in adaptation of the entire plant, allowing higher levels of sodium accumulation in the shoot to be nontoxic.

Plant Growth and Stress Treatment
All plants were grown under 14-h-day and 10-h-night conditions. For the assessment of salt tolerance of Arabidopsis expressing ThSOS1 and AtSOS1, seedlings were transferred to medium containing 200 mM NaCl as described (Shi et al., 2003). Except when indicated otherwise, seedlings were grown on quarter-strength Murashige and Skoog medium supplemented with 2% Suc and 0.8% Select-Agar (Invitrogen), with the petri dish sealed using porous Micropore tape. For root growth assay, 10-d-old seedlings were transferred to medium containing various concentrations of NaCl with the plates oriented vertically. Salt treatment of mature plants was performed as described (Oh et al., 2007).

Transcript Level
All analyses used pools of 10 plants for each sample. For the interspecific comparison, 2-week-old (for Arabidopsis) or 3-week-old (for Thellungiella) plants, thus accounting for identical growth stage (Gong et al., 2005), were incubated on vertical plates containing salt for 12 h, and their tissues were pooled and harvested for RNA extraction and RT-PCR analyses. Quantitative real-time PCR results were normalized to ACT2 gene expression. For the list of primers used, see Supplemental Table S1.

Ion Measurements
For measurement of ion contents in the seedling shoot, medium containing 200 mM NaCl was placed in single compartments of a half-divided petri dish. Two-week-old seedlings were grown vertically with their shoot placed in the empty compartment not contacting the medium. Harvested shoot samples were dried, dissolved in HNO 3 , and analyzed by inductively coupled plasmaoptical emission spectrometry (Optima 2000; Perkin-Elmer). For measurement of ion contents in hydoponically grown plants, atomic absorption spectrophotometry (Perkin-Elmer 1100B) was applied to saps extracted from leaf and root frozen samples as described (Gorham et al., 1994).
SEM-EDX was used on frozen sections of leaves and roots harvested from 3-week-old hydroponically grown plants. Samples were mounted in slots of copper holders, fixed with OCT Compound (BDH), and dipped into a bath of slush nitrogen prior to transfer under vacuum into the cryopreparation chamber (CT1500; Oxford Instruments) attached to the scanning electron microscope (DSM 960; Zeiss). The chamber temperature was left to rise from 2163°C to 290°C and set for 10 min for ice sublimation before sputter coating with gold (2 min). Samples were analyzed with an ATW detector interfaced with a Link ISIS analyzer (Oxford Instruments) under the following conditions: accelerating voltage, 15 kV; takeoff angle, 42°C; collecting time of x-ray counts, 100 s; working distance between sample and detector, 24 mm. Measurements were performed by focusing on exposed vacuoles of specific cells.

Ion Transport Assays
Vesicles of the yeast plasma membrane were produced by two-phase partitioning as described previously (Martinez-Atienza et al., 2007). The purity of vesicle preparations was tested by measuring ATP hydrolysis in the presence of inhibitors of mitochondrial (azide), vacuolar (nitrate), and plasmalemma (vanadate) ATPases. The relative sensitivity of total ATPase activity to these inhibitors demonstrated that vesicle preparations were highly enriched in the plasma membrane. Na + /H + exchange was monitored by the quinacrine fluorescence quenching method. An inside-acid proton gradient (DpH) across vesicle membranes was established after the addition of ATP. NaCl was added once DpH reached a steady state, and fluorescence recovery (i.e. dissipation of the DpH) was recorded with a fluorescence spectrophotometer (Hitachi F-2500). To determine initial rates of Na + /H + exchange, the change of relative fluorescence was measured 30 s after the addition of sodium salts. Specific activity was calculated by dividing the initial rate of fluorescence recovery, expressed as a ratio of the preformed pH gradient, by the mass of plasma membrane protein in the reaction and time (DF mg 21 min 21 , where DF = F 30 2 F 0 /F max 2 F min ). The change of pH value was measured at excitation and emission wavelengths of 430 and 500 nm, respectively.

Visualization of Na + Ions
One week-old seedlings were stained and observed by confocal microscopy (TCS SP2 RBB; Leica) after salt treatment on medium containing 1.1% type A agar (Sigma-Aldrich). Staining to reveal Na + content was performed as described (Mazel et al., 2004;Leshem et al., 2006;Meier et al., 2006). Roots were either stained with 20 mM CoroNa-Green AM (Invitrogen) in the presence of a final concentration of 0.02% pluronic acid (Invitrogen) for 3 h or incubated on a filter paper soaked with medium containing 10 mM CoroNa-Green for 18 h to stain the Na + in the root stele. For visualizing the stele of roots, fluorescein diacetate (Invitrogen) replaced CoroNa-Green AM in some experiments as positive controls of dye penetration. Where indicated, 2.5 mg mL 21 propidium iodide (Invitrogen) or 5 mM FM4-64 (Invitrogen) was added after incubation with CoroNa-Green AM.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Phylogenic relationships of SOS1 homologs from various species.
Supplemental Figure S2. Comparison of SOS1 mRNA abundance in mature Arabidopsis and Thellungiella plants.
Supplemental Figure S4. Comparison of sodium uptake under mildly saline conditions.
Supplemental Figure S5. Relative SOS1 mRNA levels in the seedling roots used for microscopy.
Supplemental Figure S6. Confocal planes of CoroNa-Green staining of the root tip region under salt stress.
Supplemental Table S1. List of RT-PCR primers.