Residues in Internal Repeats of the Rice Cation/H+ Exchanger Are Involved in the Transport and Selection of Cations*

In plants, the cation/H+ exchanger (CAX) translocates Ca2+ and other metal ions into vacuoles using the H+ gradient formed by H+-ATPase and H+-pyrophosphatase. Such exchangers carrying 11 transmembrane domains (TMs) have been isolated from plants, yeast, and bacteria. In this study, multiple sequence alignment of several CAXs revealed the presence of highly conserved 36-residue regions between TM3 and TM4 and between TM8 and TM9. These two repetitive motifs are designated repeats c-1 and c-2. Using site-directed mutagenesis, we generated 31 mutations in the repeats of the Oryza sativa CAX, which translocates Ca2+ and Mn2+. Mutant exchangers were expressed in a Saccharomyces cerevisiae strain that is sensitive to Ca2+ and Mn2+ because of the absence of vacuolar Ca2+-ATPase and the Ca2+/H+ exchanger. Mutant exchangers were classified into six classes according to their tolerance for Ca2+ and Mn2+. For example, the class III mutants had no tolerance for either ion, and the class IV mutants had tolerance only for Ca2+. The biochemical function of each residue was estimated. We investigated the membrane topology of the repeats using a method combining cysteine mutagenesis and sulfhydryl reagents. Our results suggest that repeat c-1 re-enters the membrane from the vacuolar luminal side and forms a solution-accessible region. Furthermore, several residues in repeats c-1 and c-2 were found to be conserved in animal Na+/Ca2+ exchangers. Finally, we suggest that these re-entrant repeats may form a vestibule or filter for cation selection.

In plants, the CAX is driven by the pH gradient generated by vacuolar H ϩ -ATPase and H ϩ -pyrophosphatase. cDNAs for CAXs have been cloned from plants (7,8) and Saccharomyces cerevisiae (9). Sequence information for CAX DNAs from Neurospora crassa, Synecocystis sp. PCC6803, and Bacillus subtilis is also available. The exchangers have 11 predicted transmembrane domains (TMs) and an acidic residue-rich region between TM6 and TM7. Recent studies on Arabidopsis thaliana CAX isoforms using heterologous expression in yeast have shown that the exchangers have three characteristic domains: the N-terminal regulatory region, the calcium domain, and the C domain. The N-terminal regulatory region has been shown to suppress Ca 2ϩ transport activity by interacting with its neighboring N-terminal sequence. This regulatory region was found in A. thaliana CAX1 and CAX3 and in mung bean (Vigna radiata) VCAX1 (10), but not in other exchangers. In CAX1, the 9-amino acid calcium domain exists in the hydrophilic loop between TM1 and TM2. This domain is thought to be involved in the selection of Ca 2ϩ ; however, the sequence has not been found in other CAXs. The C domain located in TM4 may be involved in the selection of Mn 2ϩ by Arabidopsis CAX2, which is the only plant CAX known to be capable of Mn 2ϩ transport (11).
Information on the domains that recognize and translocate Ca 2ϩ , H ϩ , and other metal ions is needed to elucidate the functional mechanism of the CAX transporter. In this study, we cloned a cDNA for the rice CAX (OsCAX1a) and determined the Ca 2ϩ and Mn 2ϩ transport activity. Multiple amino acid sequence alignment of CAXs of various organisms revealed two highly conserved regions with sequences similar to each other. The repeat motifs found in this study may be the most likely candidates for the essential domain for ion transport activity. We investigated the conserved repeats of the rice CAX using a combination of site-directed mutagenesis and heterologous expression in yeast. Furthermore, we performed cysteine-scanning mutagenesis against the repeats and their neighboring regions. This study provides information on the function and structure of common sequences among CAXs. The similarity of the conserved motifs of CAXs and the animal Na ϩ /Ca 2ϩ exchanger is discussed.

EXPERIMENTAL PROCEDURES
Yeast Strain and Materials-S. cerevisiae strain K665 (MATa, vcx1⌬, pmc1::TRP1, ade2-1, can1-100, his3 -11,15, leu2-3,112, trp1-1, ura3-1) was used (9). This strain lacks genes for the vacuolar Ca 2ϩ /H ϩ ex-* This work was supported by Grants-in-aid for Scientific Research 13142203, 13CE2005, and 10219203 (to M. M.) from the Ministry of Education, Science, Sports, and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the DDBJ/GenBank TM /EBI Data Bank with accession number(s) AB112656.
cDNA Cloning of the CAX-Total RNA was isolated from rice seedlings (Oryza sativa japonica L. cv. Nipponbare) grown under constant light for 2 weeks at 30°C. The resulting RNA was used for the construction of cDNA. The RNAs were converted into cDNAs using Superscript TM II RNase H Ϫ reverse transcriptase (Invitrogen) and an oligo(dT) adapter primer (5Ј-CGGGATCCACTAGTTCTAGAGCGC(T) 17 -3Ј). An OsCAX1-specific cDNA was then directly amplified by PCR with a pair of primers (5Ј-GAGGGATCCCTCTCGCGACTCGACTCTC-3Ј (forward) and 5Ј-AGGAATTCTATTGAGAGCTGCATAGATG-3Ј (reverse); BamHI and EcoRI sites are underlined). These were designed using information from the EST Database (accession numbers D39665 and AU056400). The cDNA was then inserted into the BamHI-EcoRI site of pBluescript KS(ϩ) (Stratagene).
Heterologous Expression of the CAX in Yeast Cells-An OsCAX1a⌬27 cDNA encoding OsCAX1a truncated in the N-terminal autoinhibitory domain (27 residues) was generated by PCR with specific primers (5Ј-GAGGAATTCATGTCGTCGTCGTCGCTGCG (forward; EcoRI site is underlined) and 5Ј-GAGGTCGACTCATGCGGCCTGAACACTCA-3Ј (reverse; SalI site is underlined)). The DNA fragment was inserted into the EcoRI-SalI site of the yeast expression vector pKT10 (12). The construct was introduced into S. cerevisiae using the LiOAc/polyethylene glycol method and selected in medium A (50 mM potassium phosphate buffer (pH 5.5), 0.002% adenine sulfate, 0.002% tryptophan, 2% glucose, 1% casein hydrolysate, and 0.67% yeast nitrogen base without amino acids).
Mutagenesis of the Exchanger-Site-directed mutagenesis of OsCAX1a⌬27 was performed using a QuikChange site-directed mutagenesis kit (Stratagene) following the method of Kirsch and Joly (13). The identity of the mutated nucleotides was confirmed by DNA sequencing. A cysteine-less mutant of OsCAX1a⌬27 was generated by replacing 8 endogenous cysteine residues with serines. Single cysteine mutants were generated using the cysteine-less mutant as a template. The mutated nucleotides were confirmed by DNA sequencing.
Membrane Preparation from Yeast-The vacuolar membrane-enriched fraction was prepared as described previously (14,15). Yeast cells were grown to the exponential phase in medium A and harvested. Cells were then treated with zymolyase, and spheroplasts were obtained. The spheroplast suspension was homogenized using a Dounce glass homogenizer with a tight-fitting pestle. Vacuolar membranes were isolated from the spheroplast homogenate by sucrose discontinuous gradient centrifugation. The membranes were suspended in 5 mM Tris-HCl (pH 7.6), 0.1 M sorbitol, 1 mM dithiothreitol, 5 mM MgCl 2 , 1 mM phenylmethanesulfonyl fluoride, 1 mg/liter leupeptin, and 2 mg/liter pepstatin and stored at Ϫ80°C until used.
Ca 2ϩ Uptake and Metal Ion Tolerance Assay-Ca 2ϩ transport activity in membrane vesicles was measured using the filtration method with 45 CaCl 2 (8). Yeast strains transformed with OsCAX1a⌬27 and its derivatives were inoculated in medium A and then grown overnight at 30°C. Cell suspensions were diluted 100-fold with YPD medium (yeast extract/peptone/dextrose) supplemented with 50 mM Mes/Tris (pH 5.5), 0.5 g/ml FK506 (Fujisawa Pharmaceuticals, Osaka, Japan), and an appropriate concentration of CaCl 2 (0 -200 mM). Suspensions were then cultured in a microtiter plate at 30°C. The reagent FK506 was added as a potent inhibitor of calcineurin (4). The absorbance at 600 nm was measured using a Vient Model BT-MQX200 multi-spectrophotometer (Bio-Tek Instruments, Inc., Winooski, VT). For plate assays, cell cultures were diluted 10-fold with distilled water, and 5-l aliquots were spotted on YPD plates containing 0.5 g/ml FK506 and CaCl 2 or MnCl 2 . After incubation for 2-3 days at 30°C, the plates were photographed.
Immunoblotting-We synthesized a peptide corresponding to the N-terminal region (positions 22-44, NH 2 -SRTAHNMSSSSLRKKS-DAALVRKC-COOH). For preparation of antibodies, the peptide was linked with a carrier protein (keyhole limpet hemocyanin). The conjugate was homogenized with Freund's complete adjuvant for the initial injection and with Freund's incomplete adjuvant for the booster injections. The conjugate was then injected into rabbits. The antibody obtained was used as the anti-OsCAX1a antibody. For immunoblot analysis, proteins were electroblotted from the SDS-polyacrylamide gel onto a polyvinylidene difluoride membrane (Millipore Corp.) using a semidry blotting apparatus. Immunoblotting was performed using horseradish peroxidase-linked protein A and ECL Western blotting detection reagents (Amersham Biosciences).
Cysteine-scanning Mutagenesis-Vacuolar membrane-enriched fractions with right side-out orientation were prepared as described previ-ously (14). Cysteine-scanning analysis with sulfhydryl reagents was conducted as described previously (16,17). Membranes (30 g) were suspended in labeling buffer (5 mM Mes/Tris (pH 7.2), 0.3 M sorbitol, 25 mM KCl, 5 mM MgCl 2 , 1 mM phenylmethanesulfonyl fluoride, 1 mg/liter leupeptin, and 2 mg/liter pepstatin) to a final volume of 100 l. The sulfhydryl reagent AMS was added to a final concentration of 400 M and then incubated on ice for 5 min. To stop the AMS reaction and to start the BM labeling reaction, 400 l of labeling buffer supplemented with 1 mM BM and 0.02% Triton X-100 was added. After incubation on ice for 30 min, the BM reaction was stopped with 2% ␤-mercaptoethanol. After centrifugation at 100,000 ϫ g for 30 min, the pellet was resuspended in 1 ml of labeling buffer and recentrifuged. The resulting pellet was suspended in 200 l of solubilization buffer (10 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 1 mM phenylmethanesulfonyl fluoride, 1 mg/liter leupeptin, and 2 mg/liter pepstatin). This mixture was then subjected to sonic oscillation for 10 min using a bath sonicator. The suspension was added to solubilization buffer (800 l) and incubated on ice for 5 min. After centrifugation at 100,000 ϫ g for 30 min, the supernatant (1 ml) was mixed with 15 l of streptavidinagarose beads (Pierce) and then incubated with gentle agitation for 1 h at 4°C. The agarose beads were washed five times with 500 l of solubilization buffer. The resultant beads were suspended in 15 l of sample buffer for SDS-PAGE and then boiled for 10 min. SDS-PAGE was performed on 12.5% polyacrylamide gel. The OsCAX1a protein was detected by immunoblotting with the anti-OsCAX1a antibody.

FIG. 1. Heterologous expression of OsCAX1a in yeast.
A, yeast strain K665 was transformed with OsCAX1a⌬27 cDNA. Vacuolar membrane fractions (5 g) were subjected to immunoblot analysis using the anti-OsCAX1a antibody. The arrowhead and asterisk indicate the positions of OsCAX1a⌬27 (WT exchanger) and a lower band, respectively. B, cells cultured for 16 h were diluted to 10-fold, and aliquots (5 l) were spotted on YPD plates containing CaCl 2 or MnCl 2 . Growth was recorded after 2 days (CaCl 2 ) or 3 days (MnCl 2 ) of incubation at 30°C. C, transformants (WT exchanger (ࡗ) and vector (E)) precultured for 16 h were diluted to 100-fold and then cultured in YPD liquid medium containing CaCl 2 for 16 h at 30°C. The absorbance at 600 nm was measured. D, membrane vesicles (10 g) were assayed for Ca 2ϩ uptake activity. Aliquots (100 g) of membrane suspension were preincubated in reaction medium (1 ml) with (f and E) or without (ࡗ) 0.2 M bafilomycin A 1 (baf.) for 5 min at 25°C. The reaction was started by addition of 100 M 45 CaCl 2 at 0 min. The calcium ionophore A23187 was added at 11 min at a final concentration of 5 M. Solid and dashed lines in C and D indicate the strain transformed with OsCAX1a⌬27 and vector, respectively. The data represent the means Ϯ S.D. for four independent experiments.

Expression of the Rice CAX in Yeast as a Functional
Enzyme-To characterize a constitutively active form of OsCAX1a, we truncated a nucleotide sequence corresponding to the N-terminal autoinhibitory domain (first 27 residues) from the original cDNA. Such domains in the Ca 2ϩ /H ϩ exchangers of A. thaliana (CAX1) and mung bean (VCAX1) are known to inhibit ion transport activity (19,20). Residue 28 is the third methionine of OsCAX1a. The truncated form (OsCAX1a⌬27) was expressed in and accumulated in yeast strain K665, which lacks genes for the vacuolar Ca 2ϩ -ATPase (PMC1) and Ca 2ϩ /H ϩ exchanger (VCX1). Immunoblotting of the crude membrane fraction prepared from the transformant with the anti-OsCAX1a antibody, which was prepared against the N-terminal region from positions 22 to 44, yielded a major band of 44 kDa and a minor band of 42 kDa (Fig. 1A). Here, OsCAX1a⌬27 was designated as a wild-type (WT) exchanger. The reaction of the antibody was completely suppressed with the authentic peptide, which was used for preparation of the antibody (data not shown). Thus, we conclude that the lower band is likely a degradation product of the normal translation product (OsCAX1a⌬27); however, we cannot exclude other possibilities such as glycosylation.
Plate assay of the yeast transformant showed Ca 2ϩ and Mn 2ϩ transport activity (Fig. 1B). Strain K665, which lacks PMC1 and VCX1, was sensitive to high concentrations of Ca 2ϩ (200 mM) and Mn 2ϩ (4 mM) (Fig. 1B, Vector). Introduction of OsCAX1a⌬27 into strain K665 rendered the strain tolerant to Ca 2ϩ and Mn 2ϩ . In the liquid medium assay, the transformant with OsCAX1a⌬27 showed tolerance for Ca 2ϩ at concentrations up to 200 mM. In contrast, the control yeast cells transformed with an empty vector did not grow in medium containing Ca 2ϩ even at concentrations of 50 mM (Fig. 1C). Thus, OsCAX1a⌬27 complements the yeast vacuolar Ca 2ϩ /H ϩ exchanger in the K665 strain. These results strongly suggest that there is Ca 2ϩ and Mn 2ϩ transport activity in OsCAX1a⌬27 in yeast vacuoles.
We confirmed the presence of 45 Ca 2ϩ uptake activity in membrane vesicles (Fig. 1D). In this system, the pH gradient across vacuolar membranes was generated by vacuolar H ϩ -ATPase. The membrane vesicles took up 45 Ca 2ϩ from the medium in a time-dependent manner for up to 10 min. The accumulated 45 Ca 2ϩ was released immediately after the addition of the Ca 2ϩ ionophore A23187. In the absence of ATP or in the presence of bafilomycin A 1 , a potent inhibitor of vacuolar H ϩ -ATPase, no such activity was detected (Fig. 1D). Membrane vesicles of yeast cells transformed with an empty vector had no activity. These results indicate that OsCAX1a⌬27 expressed in yeast cells functions as a Ca 2ϩ /H ϩ exchanger.
The apparent K m of the WT exchanger for Ca 2ϩ was 9.5 M. Thus, OsCAX1a has a high affinity for Ca 2ϩ , the K m value of which is comparable with that of oat CAX (K m ϭ 10 M) (21), A. thaliana CAX1 (K m ϭ 13.1 M) (7), and V. radiata VCAX1 (K m ϭ 25 M) (16). The V max value of OsCAX1a in yeast membranes was 3.8 nmol/min/mg.
Structural Features of the Rice CAX-Based on results from the TMpred program, 2 we predict that the 451-amino acid protein OsCAX1a may, like other CAX proteins, have 11 TMs (Fig. 2A). The N-terminal region, which has a 59-residue hydrophilic sequence, is likely exposed to the cytosol. OsCAX1a has an acidic motif in the cytosolic loop between TM6 and TM7; similar findings have been described for V. radiata, A. thaliana, and S. cerevisiae CAXs (7)(8)(9). In this study, multiple sequence alignment of nine CAXs from plants and yeast revealed two novel highly conserved regions. Both regions have 36 residues, 14 of which are identical. Furthermore, 7 residues are conserved in both motifs. The sequences of repeats c-1 and c-2 are GNX 2 EXIX 4 AX 8 VX 4 LGSXLSNXLXV and GNAA-EHX 6 AX 5 DX 2 LGX 3 GSX 2 QX 3 FX, respectively (Fig. 2B). The consensus sequence is GNX 2 EX 21 GSX 8 (Fig. 2B). Repeat c-1 is located between TM3 and TM4 and faces the vacuolar lumen in this topological model ( Fig. 2A). Repeat c-2 is located between TM8 and TM9 and faces the cytosol.
Each exchanger protein was clearly detected at 44 kDa upon immunoblotting (Fig. 3). In many cases, a minor band of 42 kDa was detected, although no minor band was found in mutants K144A, K149A, S151A, G154A, and K341A. As discussed above, this 42-kDa protein may be a degradation product. It is likely that a 2-kDa C-terminal region is removed by protease-mediated cleavage, as the anti-OsCAX1a antibody recognizes the N-terminal region (residues 22-44). Differences in the extent of proteol-2 Available at www/ch.embnet.org/software/TMPRED-form.html. ysis among mutants and the appearance of an addition band of 40 kDa in the N340A mutant may be due to the alteration of fine tertiary structure in the mutant exchangers.
The ion transport activities of the K665 strains with the WT (OsCAX1a⌬27) or 31 mutant exchangers were determined by a growth test on agar plates supplemented with CaCl 2 or MnCl 2 (Fig. 4). Tolerant cells grew normally and generated clear colonies on these plates. In this assay, mutants N129A, E132Q, G154A, E329Q, E329D, H330A, and H330R were sensitive to 200 mM CaCl 2 . Sensitivity to a high concentration of Mn 2ϩ varied with the mutant. Sixteen mutants, including G128A and N129A, were not tolerant to Mn 2ϩ . Mutant T354A was more tolerant to Mn 2ϩ compared with the WT exchanger.
Classification of Mutants-Quantitative characterization of the Ca 2ϩ tolerance of yeast strains with mutant exchangers was performed in liquid culture medium containing 50 or 200 mM CaCl 2 (Fig. 5). Strains with mutants that did not form colonies on agar plates (N129A, E132Q, G154A, E329Q, E329D, H330A, and H330R) did not grow in the liquid medium even at a Ca 2ϩ concentration of 50 mM. We classified the 31 mutants into six categories based on their tolerance for Ca 2ϩ and Mn 2ϩ ( Table I).
The class I mutants (T131A, K142A, E146Q, E146D, N340A, D343N, D343E and T345A) had the same tolerance for both Mn 2ϩ and Ca 2ϩ as the WT exchanger. The class II mutants (G128A, E132D, K144A, K149A, N326A, K341A, G351A, S352A and Q355A) had a lower tolerance for both ions. The class III mutants (N129A, E132Q, G154A, E329Q, E329D, H330A, and H330R) were sensitive to both ions, indicating that these residues are essential for ion transport. The class IV mutants (S155A, S158A, N159A, G325A, and K339A) were tolerant to Ca 2ϩ , but sensitive to Mn 2ϩ . The class V mutant (T354A) had the same tolerance for Ca 2ϩ as the WT exchanger and a higher tolerance for Mn 2ϩ . The class VI mutant (S151A) showed the same tolerance for Mn 2ϩ as the WT exchanger, but a lower tolerance for Ca 2ϩ . The amino acid residues mutated in classes IV-VI may be involved in ion selectivity. In total, ion selectivity and transport activity were affected in 23 of the 31 mutant exchangers.

Mutagenesis of Cysteine
Residues-To map the TM topology of the repeat motifs, we carried out cysteine replacement/biotinylation analysis. First, we prepared a cysteine-less OsCAX1a⌬27 construct in which all 8 endogenous cysteine residues were substituted with serine by site-directed mutagenesis. This construct was then expressed in strain K665. The cysteine-less mutant protein accumulated in yeast membranes as shown on an immunoblot (Fig. 6A). In membranes containing a cysteine-less mutant, Ca 2ϩ uptake activity, which is sensitive to bafilomycin A 1 , was 36% of that in membranes containing the WT exchanger (Fig. 6B). We therefore conclude that the cysteine-less mutant retains its functional structure in yeast and can be used for cysteine-scanning mutagenesis.
Membrane Topology of Repeats-In additional experiments, we treated right side-out vacuolar membrane vesicles containing mutant exchangers with AMS and then with BM. AMS is a membrane-impermeable sulfhydryl reagent that reacts with The mutants with a significant decreased in calcium tolerance compared with the WT exchanger are indicated by asterisks (p Ͻ 0.05 versus the WT exchanger, Student's t test). cysteine residues that face the cytosol, but not with residues in vacuoles. BM is conditionally membrane-permeable and reacts with residues in hydrophilic regions. After adding AMS, we added 0.02% Triton X-100 together with BM to render membranes permeable. In this system, BM is not able to react with cytosolic cysteine residues that have already reacted with AMS. Conversely, cysteine residues in the vacuolar lumen or the membrane are not labeled with AMS and therefore are able to react with BM. The BM-linked exchangers were solubilized and selected using streptavidin-agarose, which strongly binds to the biotinyl moiety of BM. The isolated exchangers were then subjected to immunoblotting with the anti-OsCAX1a antibody.
All of the mutant exchangers were labeled with BM in the absence of AMS, except for S158C, S170C, S314C, A331C, L338C, S357C, and S364C (e.g. see S158C in Fig. 7A). Four mutants of repeat c-1 (T118C, I145T, A178C, and S187C) and four mutants of repeat c-2 (S308C, A333C, A349C, and T354C) were not detected after treatment with AMS (Fig. 7, B and C), suggesting that these residues are exposed to the cytosol. Six mutants of repeat c-1 (S105C, L113C, A130C, A136C, V148C, and S150C) and two mutants of repeat c-2 (G300C and G325C) were clearly labeled with BM even after treatment with AMS. We conclude that these residues are embedded in the membrane or are exposed to the vacuolar lumen.

DISCUSSION
The aim of this study was to identify the functional domains for the recognition and translocation of metal ions in the rice CAX. We focused our attention on the conserved repeats c-1 and c-2. In preliminary experiments using immunoblotting with the anti-OsCAX1a antibody, we found that OsCAX1a is a 46-kDa protein detectable in vacuolar membranes isolated from rice seedlings. Thus, it appears that OsCAX1a is expressed and functions in the vacuolar membrane.
OsCAX1a and a Truncated Form-The CAXs of various organisms have a common motif of an acidic residue-rich region between TM6 and TM7 (7)(8)(9)22). OsCAX1a also contains such an acidic region (residues 268 -272) (Fig. 2). Although the detailed amino acid sequence is not conserved among CAXs, this acidic region is thought to be involved in the capture and selection of metal ions (7,18).
Recently, an N-terminal regulatory region has been found in A. thaliana CAX1 and CAX3 and mung bean VCAX1 (19,20). It should be noted that not all CAXs have an extensive N-terminal region. This region has been demonstrated to regulate the activity negatively by intramolecular interaction with a specific site in the N-terminal part (10,22). OsCAX1a also contains a long hydrophilic region at the N terminus and has 6 residues in the tail (residues 21-26, RXRTAH), which are identical to those of three other exchangers (CAX1, CAX3, and VCAX1). The full-length OsCAX1a protein showed reduced activity of the Ca 2ϩ /H ϩ exchanger when expressed in yeast (data not shown). We suggest that the N-terminal sequence (27 residues) of   Fig. 4 by comparison with the WT exchanger. Ca 2ϩ tolerance is indicated by ϩϩ (same tolerance as the WT exchanger; growth rate of 80 -120% of the WT exchanger), ϩ (reduced tolerance compared with the WT exchanger; p Ͻ 0.05 versus the WT exchanger, Student's t test), and Ϫ (no tolerance; absence of colony formation on the plates in Figs. 4 and Fig. 5B). For assessment of Mn 2ϩ tolerance (Fig. 4), growth of yeast transformants in the presence of 4, 6, and 10 mM MnCl 2 was individually quantified using the NIH Image program. The values were summed up for each mutant and compared with those of the WT exchanger. Tolerance is expressed by ϩϩϩ (enhanced tolerance; Ͼ120% of the WT exchanger), ϩϩ (same tolerance as the WT exchanger; 70 -120%), ϩ (reduced tolerance; 30 -70%), and Ϫ (no tolerance; 0 -30%).

Mutants
Ca 2ϩ tolerance Mn 2ϩ tolerance Class Mutants Ca 2ϩ tolerance Mn 2ϩ tolerance Class OsCAX1a is a negative regulatory region. We therefore used the truncated form (OsCAX1a⌬27) as the WT exchanger in this study.
Presence of Common Repeats in CAXs-We found two conserved repetitive sequences that are present in OsCAX1a and the CAXs of A. thaliana, V. radiata, Zea mays, and S. cerevisiae (Fig. 2). For OsCAX1a, 12 residues are identical between the two repeats. The repeat sequences are common to multiple CAX proteins and may have been generated by gene duplication at an early stage of molecular evolution.
Role of Repeats in Ion Transport and Ion Selectivity-Amino acid substitution of residues in the repeats had a negative effect on CAX activity in many cases ( Fig. 4 and Table I). Mutagenic analysis revealed that 13 residues (Gly 128 , Asn 129 , Glu 132 , Lys 144 , Lys 149 , Gly 154 , Asn 326 , Glu 329 , His 330 , Lys 341 , Gly 351 , Ser 352 , and Gln 355 ; classes II and III) are essential for the transport of Ca 2ϩ and Mn 2ϩ . These residues are likely involved in the translocation of metal ions. It should be noted that there are three pairs of conserved residues among repeats c-1 and c-2 (Ans 129 and Asn 326 , Glu 132 and Glu 329 , and Gly 154 and Gly 351 ). These residues are essential for the translocation of both metal ions.
At present, we cannot exclude the possibility that these residues are involved in the translocation of H ϩ . In the hardsoft acid-base theory, Ca 2ϩ , Mn 2ϩ , and H ϩ belong to the "hard metal" group and interact preferentially with oxygen atoms of carboxyl groups and nitrogen atoms of amino groups in amino acid side chains. Thus, several essential residues, including Asn 129 , Glu 132 , Lys 144 , Lys 149 , Asn 326 , Glu 329 , Lys 341 , and Gln 355 , may interact with Ca 2ϩ , Mn 2ϩ , and/or H ϩ during ion transport.
The class IV mutants (S155A, S158A, N159A, G325A, and K339A) were tolerant to Ca 2ϩ , but sensitive to Mn 2ϩ . The residues may recognize a difference in ion radius and select Mn 2ϩ because Mn 2ϩ (ion radius of 80 pm) is smaller than Ca 2ϩ (99 ion radius of pm). The class V mutant (T354A) showed higher tolerance for Mn 2ϩ than did the WT exchanger. Thr 354 in the WT exchanger may interfere with Mn 2ϩ translocation. Most residues in classes IV and V may be located in the Mn 2ϩ transport pathway. Recently, it has been reported that a 10amino acid region (CAFFCGGLVF), especially the Cys-Ala-Phe motif in TM4, is responsible for the Mn 2ϩ specificity of Arabidopsis CAX2 (11). This region may be specific for CAX2 because it is not conserved in OsCAX1a or other CAXs.
The mutant S151A (class VI), in which a side chain OH group was replaced with a hydrogen, was less tolerant to Ca 2ϩ compared with the WT exchanger (Fig. 5), although tolerance for Mn 2ϩ was retained. Thus, this residue may be involved in the selection of and/or affinity for Ca 2ϩ .
The class I mutations at Thr 131 , Lys 142 , Glu 146 , Asn 340 , Asp 343 , and Thr 345 did not affect tolerance for Ca 2ϩ and Mn 2ϩ , suggesting that these 6 residues make little contribution to ion transport function. Lys 142 , Asn 340 , and Thr 345 are not conserved among CAXs. There was no effect on the T131A mutant. FIG. 7. Cysteine-scanning mutagenesis and membrane topology test. A, shown is the reactivity of cysteine-less (CLO⌬), S105C, and S158C mutant exchangers with BM in the presence of 0.1% SDS and 1% Triton X-100. B and C, membranes prepared from yeast cells expressing the cysteine-less mutant and single cysteine mutants of repeats c-1 and c-2, respectively, were incubated in the absence (Ϫ) or presence (ϩ) of 40 M AMS, followed by 5-fold dilution and labeling with 100 M BM in the presence of 0.02% Triton X-100. Membranes were treated with 0.1% SDS and 1% Triton X-100, and then OsCAX1a labeled with BM was isolated using streptavidin-agarose. The proteins obtained were subjected to immunoblotting with the anti-OsCAX1a antibody.