Activation of Plant Plasma Membrane H+-ATPase by 14-3-3 Proteins Is Negatively Controlled by Two Phosphorylation Sites within the H+-ATPase C-terminal Region*

The proton pump ATPase (H+-ATPase) of the plant plasma membrane is regulated by an autoinhibitory C-terminal domain, which can be displaced by phosphorylation of the penultimate Thr residue and the subsequent binding of 14-3-3 proteins. We performed a mass spectrometric analysis of PMA2 (plasma membrane H+-ATPase isoform 2) isolated from Nicotiana tabacum suspension cells and identified two new phosphorylated residues in the enzyme 14-3-3 protein binding site: Thr931 and Ser938. When PMA2 was expressed in Saccharomyces cerevisiae, mutagenesis of each of these two residues into Asp prevented growth of a yeast strain devoid of its own H+-ATPases. When the Asp mutations were individually introduced in a constitutively activated mutant of PMA2 (E14D), they still allowed yeast growth but at a reduced rate. Purification of His-tagged PMA2 showed that the T931D or S938D mutation prevented 14-3-3 protein binding, although the penultimate Thr955 was still phosphorylated, indicating that Thr955 phosphorylation is not sufficient for full enzyme activation. Expression of PMA2 in an N. tabacum cell line also showed an absence of 14-3-3 protein binding resulting from the T931D or S938D mutation. Together, the data show that activation of H+-ATPase by the binding of 14-3-3 proteins is negatively controlled by phosphorylation of two residues in the H+-ATPase 14-3-3 protein binding site. The data also show that phosphorylation of the penultimate Thr and 14-3-3 binding each contribute in part to H+-ATPase activation.

various physiological roles, such as control of the stomatal aperture, cell elongation, plant development, organ movement, and intracellular pH homeostasis, although evidence for the direct involvement of H ϩ -ATPases in some of these roles is scarce (for reviews, see Refs. 1 and 2). Considering the high levels of H ϩ -ATPases in the plasma membrane and the large variety of physiological roles, one would expect this enzyme to be tightly regulated. These proteins are encoded by a gene family of about 10 members in Arabidopsis thaliana, Oryza sativa, and Nicotiana plumbaginifolia (3,4). Depending on the gene, expression is either restricted to particular cell types or widespread in the plant, with the possibility of more than one gene being expressed in a given cell type at the same developmental stage, thus precluding the characterization of a single isoform from plant material (3).
One case of H ϩ -ATPase post-translational regulation that has been extensively described at the molecular level involves the C-terminal autoinhibitory domain (5). Phosphorylation of the penultimate residue of H ϩ -ATPase, a Thr, triggers the binding of regulatory 14-3-3 proteins, resulting in the formation of an activated complex (6 -10). This complex was recently shown to be a dodecamer of six H ϩ -ATPases and six 14-3-3 proteins (11,12). A still unresolved question concerns the respective roles of Thr phosphorylation and 14-3-3 binding. Are they both required before the enzyme is activated, or does 14-3-3 binding further activate an enzyme already partly activated by Thr phosphorylation? Another possibility would be to consider that Thr phosphorylation fully activates H ϩ -ATPase and that 14-3-3 binding stabilizes the activated form.
H ϩ -ATPase activation by Thr phosphorylation and 14-3-3 binding has been observed in guard cells upon blue light activation (8,13); in various organs treated with fusicoccin, a fungal toxin, which stabilizes the H ϩ -ATPase⅐14-3-3 complex (7,14,15); and in plant cells upon metabolic activation (11,16,17). However, H ϩ -ATPase regulation by phosphorylation is not limited to the penultimate Thr residue. Evidence has been accumulating that phosphorylation of other unidentified residues is linked to H ϩ -ATPase inhibition in beet and oat root cells (18 -20) and cultured tobacco and tomato cells (21)(22)(23)(24). Proteomic analysis of Arabidopsis plasma membranes identified three phosphorylated H ϩ -ATPase residues corresponding to Ser 899 , Ser 940 , (25), and Thr 881 (26) of AHA2 (Arabidopsis H ϩ -ATPase isoform 2) in addition to the penultimate Thr 947 (25,26). Treatment with the bacterial elicitor flagellin results in decreased phosphorylation of AHA2 Thr 881 and Thr 947 and increased phosphorylation of Ser 899 (26). Recently, a Ser/Thr protein kinase, PKS5, was shown to phosphorylate Ser 931 of AHA2 both in vitro and in a yeast expression system, and this resulted in prevention of 14-3-3 protein binding and lower enzyme activity; however, attempts to confirm the phosphorylation of this residue in vivo by mass spectrometry (MS) 3 failed (27). Proteomics analysis of plasma membrane proteins from Arabidopsis seedlings supplied with sucrose showed that phosphorylation of Thr 881 in the AHA2 C-terminal region resulted in enzyme activation by a phosphorylation event outside the 14-3-3 binding site (16).
In this study, we took advantage of a transgenic N. tabacum BY2 cell line expressing His-tagged PMA2 (plasma membrane H ϩ -ATPase isoform 2) (28), one of the two most widely expressed H ϩ -ATPase isoforms in N. plumbaginifolia (3), and performed MS analysis on the purified His-tagged PMA2. Two phosphorylated sites, Thr 931 and Ser 938 , were identified. Expression of PMA2 mutated at these sites in yeast and tobacco cells strongly suggested that their phosphorylation interferes with the binding of 14-3-3 proteins and therefore H ϩ -ATPase activation.

EXPERIMENTAL PROCEDURES
Yeast Strains and Growth Conditions-Yeast strains were grown on rich medium containing 2% glucose, 2% yeast extract (YD medium) or on minimal medium containing 2% galactose, 0.7% yeast nitrogen base without amino acids (Difco), 0.115% drop mix composed of all amino acids required for growth (DOGal medium) (29). The pH was adjusted to 4.0 or 6.5 with HCl or KOH, and 2% agar was added to obtain solid medium. 5Ј-Fluoroorotic acid (5Ј-FOA) medium was prepared as described in Ref. 29.
Genetic Constructions-The yeast plasmid 2p(PMA1)6hispma2 contains the pma2 gene, with six His codons between residues 3 and 4, under the control of the S. cerevisiae PMA1 promoter, the LEU2 gene for selection, and the 2 plasmid-derived sequence for high copy number replication of the plasmid (30). The generation of the PMA2-E14D, PMA2-P154R, and PMA2-N510L activating mutants has been described previously (31). PMA2-P154R and PMA2-N510K were tagged with 6 His residues by replacing the NheI/BglII DNA fragment of 2p(PMA1)6hispma2 by that from 2p(PMA1)pma2P154R or 2p(PMA1)pma2N510K, respectively. PMA2-E14D was Histagged by PCR as described in Ref. 32. The mutation of Thr 931 and Ser 938 to Ala and Asp residues was achieved by amplifying a modified fragment between the BglII and XbaI restriction sites of 2p(PMA1)6hispma2 plasmid by triple PCR and inserting it into 2p(PMA1)6hispma2 opened by BglII and XbaI restriction. The Thr 931 -Ser 938 double mutants were obtained using the same PCR strategy, except that the starting plasmids were 2p(PMA1)6hispma2S938D and 2p(PMA1)6hispma2S938A. Insertion of the Ser 938 and Thr 931 mutations into 2p(PMA1)6hispma2E14D was performed by exchanging the BglII/XbaI DNA fragments. The PMA2-P154R and PMA2-N510L mutants carrying the Ser 938 mutations were obtained by replacing the NheI/BglII DNA fragment from 2p(PMA1)6hispma2S938D or 2p(PMA1)6hispma2S938D with that from 2p(PMA1)pma2P154R or 2p(PMA1)-pma2N510K, respectively. The YAK2 yeast strain has disrupted yeast PMA1 and PMA2 genes and contains a centromeric plasmid carrying the yeast PMA1 gene under the control of the GAL1-10 promoter and the URA3 gene for selection (30).
The plant binary vectors used for the N. tabacum BY2 cell transformation were obtained by first inserting a SmaI and XbaII fragment from the yeast 2p(PMA1)6hispma2, -S938A, S938D, T931A, and T931D plasmid, corresponding, respectively, to the PMA2-S938A, PMA2-S938D, PMA2-T931A, and T931D coding sequence into the pAUX3131 vector (33) between the N. plumbaginifolia PMA4 promoter reinforced with two copies of the cauliflower mosaic virus 35 S enhancer (34) and the nos terminator. In the second step, PMA2 expression cassettes were transferred as an I-SceI fragment into the plant binary vector pPZP-RCS2 (33) containing the nptII marker gene.
Microsomal membranes were prepared from BY2 cells as in Ref. 11. The homogenization and suspension buffers were supplemented with phosphatase inhibitors and had the same composition as those used for yeast microsomal membrane preparation, except that 0.6% polyvinylpolypyrolidone was added to the homogenization buffer.
N. tabacum BY2 Cell Culture and Transformation-BY2 cells were maintained on MS medium (catalog number 2610024; MP Biomedicals, LLC) supplemented with 3% sucrose, 0.2 mg/liter 2,4-dichlorophenoxyacetic acid, 0.2 g/liter KH 2 PO 4 , pH 5.8, 50 mg/liter myo-inositol, 5 mg/liter thiamine-HCl. BY2 cells were diluted 20-fold each week in fresh medium and grown as described previously (35). Transient expression of the mutated PMA2 isoforms was performed by co-cultivating for 68 h 24 ml of 4-day-old BY2 cell culture with 7 ml of Agrobacterium tumefaciens strain LBA4404virG (36) transformed with the different pPZP-PMA2 and grown to an A 600 of 1.4. A. tumefaciens and BY2 cells were washed and co-cultured in BY2 cell medium at pH 5.3 supplemented with 10 mM glucose and 50 M acetosyringone.
Mass Spectrometric Analysis-The band of interest was excised from the SDS gel and subjected to overnight digestion at 37°C with 250 ng of trypsin in 100 -200 l of 200 mM ammonium bicarbonate.
For the neutral loss analysis, the phosphorylated peptides were purified using immobilized metal (Fe 3ϩ ) affinity chromatography (POROS MC20, Applied Biosystems) according to the protocol supplied by the manufacturer, except for the elution buffer (1 N NH 4 OH). Neutral loss scanning was executed on an Applied Biosystems API 3000 mass spectrometer, equipped with an off-line nanospray source.
For the multiple reaction monitoring analysis, the peptide mixture was enriched for phosphopeptides on PhosTrap (PerkinElmer Life Sciences) beads as described by the manufacturer. Phosphopeptides were analyzed by nano-LC-MS/MS on a Dionex Ultimate capillary liquid chromatography system coupled to an Applied Biosystems 4000 QTRAP mass spectrometer. Peptides were separated on a PepMap C18 column developed with a 30-min linear gradient (0.1% formic acid, 6% acetonitrile/water to 0.1% formic acid, 40% acetonitrile/water). Multiple reaction monitoring was then used to induce product (ϩ) ion scanning to determine the peptide sequence and to localize the phosphorylated residue(s).
ATPase Assays-ATPase assays were performed in a 96-hole microplate at 30°C in reaction medium (4 mM MgATP, 1 mM free Mg 2ϩ (MgCl 2 ), 50 mM Mes (pH 6.5), 20 mM KNO 3 , 0.2 mM NaMoO 4 , 0.05% Brij, and 10 mM sodium azide). The reaction was started by adding 10 l of 4.34 mM ATP (the total amounts of ATP and Mg 2ϩ to be added to the reaction mixture at the indicated pH were calculated in order to obtain 4 mM MgATP with 1 mM free Mg 2ϩ (37)) to 40 l of reaction medium containing 1.5 g of protein in the presence or absence of 0.2 mM orthovanadate. After 3, 6, and 9 min, 60 l of 1% SDS was added to stop the reaction, and 30 l of 50% (w/v) (NH 4 ) 6 Mo 7 O 24 , 4 N H 2 SO 4 and 30 l of 1% 4-methylaminophenol sulfate, 3% NaHSO 3 were added successively.
Measurements of the External Acidification by Yeast Cells-Cells were grown in YD medium, harvested at the late exponential phase, and washed three times in ice-cold water. Then 2 10 8 (pH 6.5) or 10 9 (pH 4.0) cells were added to 10 ml of 200 mM glucose, 20 mM KCl in either 2 mM Mes, pH 6.5, or 0.2 mM citrate, pH 4.0, in a vial with a magnetic stirrer and a pH electrode at 30°C. pH was recorded for 20 min, and the pumping activity was calculated as pH units min Ϫ1 (10 9 cells) Ϫ1 .

RESULTS
Identification of New PMA2 Phosphorylated Sites-Our aim was to identify phosphorylation sites in PMA2, one of the two most widely expressed H ϩ -ATPases in N. plumbaginifolia, using an N. tabacum cell line (BY2) expressing His 6 -tagged PMA2 (28). PMA2 was solubilized from the microsomal fraction, purified by Ni 2ϩ -NTA chromatography, and electrophoresed on SDS gels. The band corresponding to PMA2 was excised from the gel and digested with trypsin and phosphorylated tryptic peptides enriched by PhosTrap titanium beads (PerkinElmer Life Sciences) or by immobilized metal (Fe 3ϩ ) affinity chromatography (POROS MC20; Applied Biosystems). Peptide analysis by neutral loss of 32.4 and 49 Da, corresponding to the loss of a phosphate residue (98 Da) in peptides bearing three (98/3) or two (98/2) positive charges, respectively, indicated the presence of several phosphorylated PMA2 peptides located in the C-terminal region or in the large loop of the enzyme (see supplemental Fig. 1 and Table 1). Among these peptides, we sought to confirm the identification of the phosphorylated residues in the C-terminal region, because they were observed by both neutral loss of 32.4 and 49 and also because they belong to the 14-3-3 binding site. Using multiple reaction monitoring, we confirmed the presence of one phosphopeptide containing phosphorylated Ser 938 and another containing phosphorylated Thr 931 and Ser 938 (Fig. 1). Both residues are located within the C-terminal region interacting with 14-3-3 proteins (12) and are well conserved in the H ϩ -ATPase family. Plant H ؉ -ATPase Phosphorylation FEBRUARY 13, 2009 • VOLUME 284 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4215
Characterization of PMA2 Thr 931 and Ser 938 Mutants Expressed in Yeast-For a more detailed characterization, Thr 931 and Ser 938 were mutagenized singly or in combination to Ala (to prevent phosphorylation) or Asp (to mimic the negative charge of a phosphorylated residue), and the resulting His 6 -tagged PMA2 mutants were expressed from a yeast plasmid in the S. cerevisiae YAK2 strain in which the cell's own two H ϩ -ATPase genes are disrupted and which is maintained by the presence of yeast PMA1 under the control of the GAL1 promoter on a centromeric plasmid carrying the selective URA3 gene (30). In order to test the ability of the PMA2 mutants to sustain yeast growth, the yeast transformants were shifted to glucose medium in order to repress yeast PMA1 expression. As shown in Fig. 2A, only wild-type PMA2 and PMA2-S938A were able to sustain yeast growth (left), and none of the combined Thr 931 and Ser 938 mutations allowed yeast growth (right), demonstrating a dominant effect of the Thr 931 mutations. Western blotting confirmed the presence of the PMA2 mutants (data not shown), ruling out an indirect effect of the mutations on H ϩ -ATPase expression. To confirm the results, we shifted the transformed strains to glucose medium containing 5Ј-FOA, a chemical that is converted into a toxic compound by the URA3 gene product. Since this gene is present on the plasmid bearing the yeast PMA1, growth in the presence of 5Ј-FOA selects for strains that have lost this plasmid and thus the yeast PMA1 but contain a functional plant H ϩ -ATPase gene present on the other plasmid. Colonies were obtained for the strains expressing wild-type PMA2 and PMA2-S938A, the latter growing faster (Fig. 2B). No colony was ever obtained with the other constructs, confirming that the T931A, T931D, or S938D mutation reduces H ϩ -ATPase activity to a level that does not allow yeast growth (data not shown).
To further investigate the role of these phosphorylated residues, we introduced the Thr 931 and Ser 938 single mutations into the constitutively activated PMA2-E14D mutant, in which the activating mutation results in higher H ϩ -ATPase activity and better yeast growth, probably by decreasing the interaction of the C-terminal autoinhibitory domain with the rest of the protein and making Thr 955 more accessible to phosphorylation and 14-3-3 binding. This in turn might reduce access of phosphatases to phosphorylated Thr 955 (31). As shown in Fig. 3, in this PMA2-activated form, the T931A, T931D, or S938D mutation did not prevent the appearance of colonies after 5Ј-FOA selection and did not affect yeast growth at pH 6.5 on solid medium (left) and in liquid culture (duplication time between 2.6 and 2.9 h for PMA2-E14D and E14D-mutated strains). However, when grown under more demanding conditions (i.e. at pH 4.0 (right)), PMA2-E14D/T931D and PMA2-E14D/T931A cells did not grow (more than 20 h duplication time in liquid culture), whereas PMA2-E14D/S938D (6.5 h duplication time) grew more slowly than those expressing PMA2-E14D (5 h) or PMA2-E14D/S938A (4.8 h), showing that the activating E14D mutation only partly overruled the inhibitory effect of the other mutations.
Since both Thr 931 and Ser 938 are in the 14-3-3 protein binding domain, we wondered whether the growth reduction was   related to a lower activation of the enzyme due to reduced phosphorylation of the penultimate Thr 955 and/or reduced 14-3-3 protein binding. We and others previously measured phosphorylation of the Thr 955 using commercial anti-phospho-Thr antibodies (27,32), but the identification of other phosphorylated Thr residues (e.g. PMA2 Thr 931 ) calls into question the specificity of the signal observed with these antibodies. In the present study, we therefore used a monoclonal antibody, PMA2pT, raised against a synthetic peptide mimicking the phosphorylated PMA2 C-terminal peptide 949 TIQQSTpTV (where pT represents phosphothreonine), which has been demonstrated to be specific for Thr 955 -phosphorylated PMA2. 4 His 6 -tagged wild type or mutant PMA2 was purified from the different yeast strains and analyzed by Western blotting (see Fig. 4, A and B, for quantification). As expected, no Thr 955 phosphorylation or co-purified 14-3-3 proteins were seen with PMA2-E14D/T955A.
In agreement with the reduction of yeast growth at low pH, S938D mutation of PMA2-E14D resulted in a large reduction in Thr 955 phosphorylation levels (down to 27%) and almost no co-purified 14-3-3 proteins (down to 2.2%), showing that this mutation has a dramatic effect on 14-3-3 protein binding. This was also supported by the fact that, at the similar level of Thr 955 phosphorylation seen with wild-type PMA2 and PMA2-E14D/ S938D, almost no 14-3-3 proteins were bound to the latter compared with the amount bound to the wild-type PMA2.
In contrast, introduction of the S938A mutation into PMA2 did not influence the level of Thr 955 phosphorylation compared with wild-type PMA2 but significantly increased the binding of 14-3-3 proteins by 44%, in agreement with the greater capacity of this mutant to sustain yeast growth at pH 6.5 compared with the wild-type PMA2. Introduction of the S938A mutation into  . Comparative growth at pH 6.5 and 4.0 of yeast strains expressing PMA2 mutants. After selection on 5Ј-FOA-containing medium, YAK2 yeast cells expressing the indicated PMA2 mutant were grown overnight in YD medium, diluted to an A 600 of 1, 0.1, 0.01, or 0.001; plated on solid YD medium; and allowed to grow for 2 days on YD medium at pH 6.5 or for 4 days on YD medium at pH 4.0. The untransformed YAK2 strain was plated as a negative control. isolated from a 36-h culture (YD, pH 6.5) of yeast strains expressing wild-type or mutant PMA2 was solubilized, and the His 6 -tagged PMA2 was purified by Ni 2ϩ -NTA chromatography and analyzed by SDS-PAGE and Western blotting using antibodies against PMA2 (top), PMA2 phospho-Thr 955 (pPMA2; middle), or 14-3-3 proteins (bottom). Note that the double band identified by anti-14-3-3 protein antibodies corresponds to the two S. cerevisiae 14-3-3 proteins. B, PMA2 phospho-Thr 955 /PMA2 signal ratio (dark gray bars) and 14-3-3 proteins/PMA2 signal ratio (light gray bars) from A. The signal for each band was normalized to that for PMA2-E14D (set at 100%), and then the signal ratios were calculated. Quantification was performed using Image Station 4000R and Molecular Imaging Software from Eastman Kodak Co. The data are the mean and the S.E. for the results from three independent experiments.

P M A 2 E 1 4 D S 9 3 8 A P M A 2 E 1 4 D S 9 3 8 D P M A 2 E 1 4 D P M A 2 E 1 4 D P M A 2 E 1 4 D T 9 5 5 A P M A 2 E 1 4 D T 9 3 1 A P M A 2 E 1 4 D T 9 3 1 D P M A 2 S 9 3 8 A P M
the PMA2-E14D strain did not result in significant modification of phosphorylation or binding of 14-3-3 proteins, agreeing with the lack of growth modification resulting from this additional mutation. These results indicate that Ser 938 phosphorylation has a much stronger impact on 14-3-3 protein binding than on Thr 955 phosphorylation.
Modifying Thr 931 to Ala and Asp reduced Thr 955 phosphorylation to 56 and 32%, respectively, compared with the control PMA2-E14D, whereas the 14-3-3 protein binding was abolished in both cases, indicating that Thr 931 phosphorylation affects the 14-3-3 protein binding more than Thr 955 phosphorylation and that a high level of Thr 955 phosphorylation together with 14-3-3 protein binding are required to allow yeast growth at low pH.
Since many other activating mutants of PMA2 have been identified (38), we extended our analysis by combining the Ser 938 mutations with two other activating mutations, P154R and N510K, which are localized in the small and large loop, respectively, unlike E14D, which is in the N-terminal region. These three activated mutants result in reduced constraint of the C-terminal inhibitory domain and have been proposed to take part into the enzyme domain interacting with the C-terminal inhibitory domain (38,39). The single P154R and N510K activating mutations resulted in yeast growth at pH 4.0 similar to that obtained with E14D (Fig. 3). Combining these mutations with either S938A or S938D had the same effects as the same combination with the E14D mutation (i.e. no major modifications with S938A but slower growth at pH 4.0 (Fig. 3), reduced Thr 955 phosphorylation, and absence of 14-3-3 protein binding (Fig. 5) with S938D). We conclude that the activation process of each of these three activating mutations, localized in three different regions, was partly reduced by mutation of Ser 938 into Asp, indicating that full activation depends on 14-3-3 protein binding.
To link the above data to H ϩ -ATPase activity, ATPase assays were performed on the plasma membrane fraction from the Ser 938 mutants, which were chosen because Ser 938 mutations into Ala and Asp residue resulted in different phenotypes. As shown in Fig. 6A, PMA2-S938A had a higher activity than PMA2, in agreement with the better growth and increased 14-3-3 protein binding. As expected, all of the mutants on the PMA2-E14D background displayed higher ATPase activity than the wild-type PMA2. No statistically significant difference was observed between PMA2-E14D and PMA2-E14D combined with the T955A or S938A mutation. Unexpectedly, the PMA2-E14D/S938D mutant displayed slightly higher activity compared with the PMA2-E14D isoform instead of lower activity, as expected from the slower growth at pH 4.0. Since the PMA2 pPMA2 14-3-3 H ϩ -ATPase expression levels of the different mutants were similar, we wondered whether the in vitro ATPase activity reflected the in vivo proton pumping activity of these enzymes. Acidification of the external medium by intact cells was monitored as previously (38,40) to assess the in vivo H ϩ -ATPase activity. As shown in Fig. 6B, at external pH 6.5, no significant difference was observed between the strains expressing PMA2-E14D or PMA2-E14D/S938D, whereas, at external pH 4.0, the latter had a 28% reduced acidification rate compared with the former (p Ͻ 0.03), in agreement with the slower growth. Characterization of PMA2 Thr 931 and Ser 938 Mutants Expressed in Tobacco Cells-All of the above data were obtained using yeast. Although 14-3-3 proteins are well conserved, we cannot rule out the possibility that, in some cases, yeast 14-3-3 proteins might behave differently from plant 14-3-3 proteins. To examine whether the marked differences in 14-3-3 protein binding of the S938D and T931D mutations seen in yeast were recapitulated in plants, we performed a transient expression of wild-type PMA2 and the S938A, S388D, T931A, and T931D PMA2 mutants in N. tabacum BY2 suspension cells by co-cultivation with transformed A. tumefaciens.
In a previous study (11), we showed that, under standard culture conditions, little phosphorylation of PMA2 Thr 955 and little 14-3-3 protein binding are seen in N. tabacum BY2 cells, but the addition of fusicoccin, a fungal toxin known to stabilize the H ϩ -ATPase⅐14-3-3 complex, results in a dramatic increase in the amount of this complex and of Thr 955 phosphorylation. In the present study, 10 M fusicoccin was added to the BY2 cells for 30 min before ectopic wild-type or mutant His 6 -tagged PMA2 was purified from the microsomal fraction of BY2 cells. Western blotting (Fig. 7) showed that PMA2-S938D as well as PMA2-T931A and PMA2-T931D had a lower level of phosphorylated Thr 955 and of bound 14-3-3 proteins than wild-type PMA2 or PMA2-S938A, confirming the data obtained in yeast.

DISCUSSION
A well characterized activation mechanism of H ϩ -ATPase consists of the phosphorylation of its penultimate residue and the binding of 14-3-3 protein dimers. Eight additional phosphorylation sites have been identified in the C-terminal region of A. thaliana or O. sativa H ϩ -ATPases through proteomics analysis of plasma membranes or through more detailed analysis (16, 25-27, 41, 42). In the present work, we focused on the phosphorylation sites of PMA2 from N. plumbaginifolia, which belongs to a different H ϩ -ATPase subfamily than AHA2, an Arabidopsis isoform in which the phosphorylation sites have been well studied. Two phosphorylated residues, Thr 931 and Ser 938 , were identified by mass spectrometry in the C-terminal region of PMA2 in N. tabacum culture cells.
These two residues were further studied in a yeast expression system by site-directed mutagenesis. Table 2 summarizes the most important data on yeast growth, Thr 955 phosphorylation, and 14-3-3 protein binding for the wild-type PMA2 and its different mutants. His-tagging of the mutant forms allowed their purification together with bound 14-3-3 proteins. This avoided performing Western blotting with a membrane fraction, which might have given a 14-3-3 protein signal not totally due to H ϩ -ATPase, since 14-3-3 proteins can bind to many different proteins. H ϩ -ATPase purification also made it possible to measure 14-3-3 proteins bound to the native H ϩ -ATPase instead of to a partly denatured enzyme, as in far Western blotting. Another important feature of our study was the use of monoclonal antibodies directed specifically against the Thr 955phosphorylated form of PMA2, 3 thus avoiding the use of general phosphothreonine antibodies that cannot distinguish between the penultimate Thr and other Thr residues that might be phosphorylated, such as PMA2 Thr 931 (this work) or PMA2 Thr 889 , which is homologous to AHA2 Thr 881 (16).
Phosphorylation of Thr 931 or a homologous residue had not been previously identified in a plant H ϩ -ATPase. This residue is located in the C-terminal region that binds a 14-3-3 protein P M A 2 PMA2 pPMA2 14-3-3 Twenty-four milliliters of a 4-day culture of N. tabacum BY2 cells was co-cultured for 68 h with A. tumefaciens strains transformed with the plasmid expressing the indicated wild-type or mutant His-tagged PMA2. After supplementing the cells with 10 M fusicoccin for 30 min, the cells were harvested and homogenized, and a microsomal fraction was prepared. Proteins (1.5 mg) were solubilized, and the tagged PMA2 purified by Ni 2ϩ -NTA chromatography and analyzed by SDS-PAGE and Western blotting using antibodies against PMA2 (top), PMA2 phospho-Thr 955 (pPMA2; middle), or 14-3-3 proteins (bottom). The result shown is representative of that obtained in two independent experiments on each construct.

Strain
Growth at pH 4.0 Growth at pH 6.5 Thr 955 phosphorylation

Plant H ؉ -ATPase Phosphorylation
dimer. Its mutagenesis to Ala and Asp resulted in abolition of yeast growth when introduced into wild-type PMA2 or its reduction when associated with the activating mutation E14D ( Table 2). Analysis of the purified mutants showed that the penultimate Thr 955 was still phosphorylated to some extent, but no 14-3-3 proteins were bound. How can we explain that a mutation (Asp) expected to mimic a phosphorylated residue and another one (Ala) expected to prevent phosphorylation resulted in the same effect? A three-dimensional structure of the H ϩ -ATPase C-terminal region crystallized with 14-3-3 proteins has been obtained (12). A 14-3-3 protein dimer simultaneously binds two interacting H ϩ -ATPase peptides, each of which terminates as a loop within the typical 14-3-3 protein binding groove. Thr 931 lies in a small loop between two helices embedded in this groove. The fact that an Ala mutation has the same effect as the Asp mutation of a phosphorylated residue is not surprising in this case, since an Ala residue, which is more favorable to the formation of a helical structure, is expected to disturb the loop as much as an Asp residue. Actually, the T931A mutation in a construct consisting of the last 52 residues of PMA2 fused to intein also impairs 14-3-3 protein binding in vitro (12). The homologous residue in Arabidopsis AHA2 has also been pointed out as an important residue for 14-3-3 protein binding (43). As with Thr 931 , mimicking Ser 938 phosphorylation by mutating it to Asp had a greater effect on 14-3-3 protein binding than on Thr 955 phosphorylation. Ser 938 is located in a helical structure embedded in the complex, and its phosphorylation is predicted to disturb the association of the two H ϩ -ATPase C-terminal regions within the 14-3-3 protein dimer. In particular, phosphorylation at this position introduces a negative charge close to that of Glu 928 , and the charge repulsion between the two residues could destabilize the enzyme C-terminal secondary structure (12). Characterization of a total protein extract from trichloroacetic acid-treated cells to inactivate any phosphatase activity showed that the Thr 955 phosphorylation was also affected in vivo by the S938D mutation (data not shown), suggesting that in the absence of 14-3-3 protein binding, phospho-Thr 955 is not protected and thus more susceptible to phosphatases in vivo. Mutation of PMA2-Ser 938 into Ala did not significantly modify Thr 955 phosphorylation but increased 14-3-3 protein binding and resulted in better yeast growth. The affinity of 14-3-3 proteins for the C-terminal region of PMA2 S938A has already been analyzed in vitro and was about 3 times higher with Ala than with Ser at position 938 (12). It therefore seems that a Ser instead of an Ala residue at position 938 is the "price to pay" to allow regulation by phosphorylation. Similar effects of Asp or Ala mutation of Ser 938 on 14-3-3 protein binding have been recently reported for the homologous residue in AHA2 (Ser 931 ) (27). These authors identified a protein kinase, PKS5, which is able to phosphorylate this residue in vitro or in vivo in a yeast expression system. However, phosphorylated AHA2-Ser 931 has still to be identified in Arabidopsis. In the plant cells, both PMA2 T931D and S938D mutants were shown to be affected at the level of their Thr 955 phosphorylation and capacity to bind 14-3-3 proteins as it was observed in the yeast system. Phosphorylation of PMA2 Thr 955 was even more affected in the plant context expression. This could probably result from the fact that experiments performed in yeast were made using the PMA2-E14D activated form or that plant phosphatases are more efficient in dephosphorylating the unprotected Thr 955 . These effects of Ser 938 and Thr 931 mutagenesis observed in a plant cell context as well as the fact that both residues were found to be phosphorylated in BY2 cells support the relevance of their role in H ϩ -ATPase regulation in the plant. In addition, phosphorylation of the AHA2 residue homologous to PMA2 Ser 938 has been proposed to down-regulate the enzyme activity under high pH growth conditions, suggesting a role in vivo of this regulation (27).
The strain combining the activating mutant PMA2-E14D and the mutation T955A, which prevented phosphorylation of this residue and subsequent 14-3-3 protein binding, grew at the same rate as the strain expressing PMA2-E14D and hence faster than wild-type PMA2 at pH 6.5 but, unlike the strain expressing PMA2-E14D, did not grow at pH 4.0 ( Table 2). This suggests that the enzyme activation by the E14D mutation depends only in part on Thr 955 phosphorylation and 14-3-3 protein binding. This conclusion is corroborated by the observation that both of the Thr 931 mutants and the S938D mutant of PMA2-E14D also showed a growth intermediate between that with PMA2-E14D or wild-type PMA2.
An interesting observation was made with the S938D mutation on the PMA2-E14D background. It reduced growth and proton pumping at pH 4.0, but its in vitro ATPase activity at pH 6.5 was slightly increased. This discrepancy suggests that the in vitro ATPase activity might not reflect the actual in vivo performance of the enzyme. This might result from a partial uncoupling of ATPase hydrolysis and proton pumping as observed for AHA2-D684N (44). Determination of whether this is the case will require a detailed analysis of this mutant, combining in vivo and in vitro approaches.
A still unresolved question regarding H ϩ -ATPase activation through its C-terminal region is whether phosphorylation of the penultimate Thr (Thr 955 in PMA2) is sufficient for activation (in which case, 14-3-3 protein binding might play a role in stabilization), whether both phosphorylation and 14-3-3 protein binding are required before the enzyme is activated, or whether each process contributes partly to activation. This question can now be answered by comparing the growth of the different mutants. Yeast expressing PMA2 or PMA2-S938A showed the same level of Thr 955 phosphorylation, but the S938A form had more 14-3-3 protein bound, and the yeast strain grew faster at pH 6.5 than the wild type ( Table 2). This suggests that the binding of 14-3-3 proteins contributes to enzyme activation. What about Thr 955 phosphorylation? Neither PMA2-E14D/T955A nor PMA2-E14D/S938D mutant bound 14-3-3 proteins, whereas the latter, unlike the former, still contained phosphorylated Thr 955 , and the yeast strain still grew at pH 4.0. These data suggest that Thr 955 phosphorylation and the binding of 14-3-3 proteins each contribute in part to the activation process.
In conclusion, we have identified two new phosphorylation sites, Thr 931 and Ser 938 , in N. plumbaginifolia PMA2. Their study by directed mutagenesis suggests that phosphorylation at each position interferes with the binding of 14-3-3 proteins to the C-terminal region and thus prevents full activation of the enzyme.