Phosphosite Mapping of P-type Plasma Membrane H+-ATPase in Homologous and Heterologous Environments*

Background: Protein phosphorylation is an important posttranslational modification. Results: Both in planta and when expressed in yeast, the P-type proton pump is phosphorylated at multiple new positions at its terminal regulatory domains. Conclusion: Multiple methods for phosphopeptide enrichment are required for complete phosphosite mapping. Significance: This work provides a surprising example of functional conservation of protein kinase action between plants and yeast. Phosphorylation is an important posttranslational modification of proteins in living cells and primarily serves regulatory purposes. Several methods were employed for isolating phosphopeptides from proteolytically digested plasma membranes of Arabidopsis thaliana. After a mass spectrometric analysis of the resulting peptides we could identify 10 different phosphorylation sites in plasma membrane H+-ATPases AHA1, AHA2, AHA3, and AHA4/11, five of which have not been reported before, bringing the total number of phosphosites up to 11, which is substantially higher than reported so far for any other P-type ATPase. Phosphosites were almost exclusively (9 of 10) in the terminal regulatory domains of the pumps. The AHA2 isoform was subsequently expressed in the yeast Saccharomyces cerevisiae. The plant protein was phosphorylated at multiple sites in yeast, and surprisingly, seven of nine of the phosphosites identified in AHA2 were identical in the plant and fungal systems even though none of the target sequences in AHA2 show homology to proteins of the fungal host. These findings suggest an unexpected accessibility of the terminal regulatory domain of plasma membrane H+-ATPase to protein kinase action.

Post-translational modifications play important roles in a wide range of cellular functions. It is estimated that about onethird of all proteins in eukaryotic cells are phosphorylated at any given time (1). Reversible phosphorylation regulates the activity, stability, and spatial organization of large number of proteins (2). Post-translational regulation by phosphorylation of proteins has a key role in signal transduction cascades in cells (1,3,4). Furthermore, protein kinases influence protein-protein binding properties by regulating the phosphorylation-dependent binding of target motifs to modular protein domains (5,6). Examples of domains that specifically bind to phosphorylated targets are the Src homology 2 (SH2) domain (7), the BRCA1 C-terminal (BRCT) domain (8), and the 14-3-3 protein (9).
Proteins are phosphorylated by protein kinases, a large family of highly related enzymes (10). In the yeast Saccharomyces cerevisiae, 122 protein kinases are present (11), humans have 518 putative protein kinases (12), and 836 and 1386 protein kinases have been identified in the genomes of the plants Arabidopsis thaliana and Oryza sativa (rice), respectively (13).
Characteristic for the catalytic function of protein kinases is that they recognize and phosphorylate linear motifs (11, 14 -16). Linear motifs are short regions of proteins (typically of less than 10 residues) that often reside in regions without an ordered structure (17,18). The short length of protein kinase target sequences and often poorly conserved sequences makes them difficult to predict with certainty (16). Furthermore, phosphorylation motifs are likely to arise/disappear spontaneously via mutations, for which reason they are evolutionarily labile. Unlike protein domains, which are conserved over long evolutionary distances, phosphorylation motifs often reside in fast-evolving regions (15,17,19,20). These properties render phosphorylation sites difficult to align and trace evolutionarily (21)(22)(23)(24)(25). Given the presence of short linear and poorly conserved phosphorylation motifs, the number of potentially different phosphorylation sites becomes relatively small. With a large number of protein kinases it is, therefore, not surprising that target sequences often can be phosphorylated by more than one protein kinase, and single protein sequences can have several phosphorylation sites (26).
There are 11 isoforms of plasma membrane proton pumps in A. thaliana that are expressed throughout the plant. The closely related isoforms AHA1 3 and AHA2 together are essential for plant growth (42,43). Plasma membrane proton pumps are phosphorylated at multiple sites in vivo (44), and six phosphosites have been identified so far. By combining several methods for extraction and enrichment of phosphopeptides, we could confirm the presence of five of these phosphosites and identified five new phosphosites. A commonly used tool for studying protein phosphorylation is expression of proteins in heterologous hosts, this being bacteria, yeast, or mammalian cell lines. We compared the phosphorylation profile of AHA2 in its natural environment with that of the plant pump expressed heterologously in the fungal host S. cerevisiae. The plant proton pump was phosphorylated to a large extent in yeast and, to our surprise, at several of the same sites as in Arabidopsis even though phosphorylated sequences did not share homology to any predicted yeast gene product.
Expression of AHA2 in Yeast-The S. cerevisiae strain RS-72 (46) was transformed and cultured essentially as described previously (47). In RS-72 (MATa ade1-100 his4-519 leu2-3,112), the natural constitutive promoter of the endogenous yeast plasma membrane H ϩ -ATPase PMA1 was replaced by the galactose-dependent promoter of GAL1. A centromeric yeast expression vector (pMP1733) was used for expression of different versions of the AHA2 gene placed under control of the PMA1 promoter. Depending on the experiment, wild-type AHA2 (pMP 1625 or 1745).
Site-directed Mutagenesis-The construction of the wild -type H ϩ -ATPase vector for heterologous expression in the yeast S. cerevisiae has been described (48). Mutants were constructed by site-directed mutagenesis using an overlap extension polymerase chain reaction. All mutated sequences were verified by DNA sequencing.
Yeast Complementation Assays-The yeast strain RS-72 was employed for functional complementation growth analysis. Plasmid-borne plant H ϩ -ATPases carrying point mutations were tested for their ability to rescue a pma1 mutant on glucose medium. Each experiment was replicated independently three times, each time with cells from independent transformation events.
Isolation of Membrane from Transformed Yeast Cells-Cells were harvested, and membranes were isolated as described (49,50) with the following modifications. Yeast cells were homogenized in the presence of phosphatase inhibitors: 50 mM Tris HCl, pH 7.5, 10% glycerol, 1 mM EDTA, 1 mM DTT, 25 mM NaF, 25 mM sodium pyrophosphate (Na 4 O 7 P 2 ), and 10 mM sodium molybdate (Na 2 MoO 4 ). Isolated membranes from transformed yeast cells were routinely subjected to Western blotting using anti-AHA2 antibodies to confirm equal expression of wild-type AHA2 and mutants derived from this protein. Protein concentration was determined by Bradford assay (51) using bovine serum albumin as reference In Solution Digestion of Proteins-For determination of H ϩ -ATPase phosphorylation in planta, plant plasma membrane vesicles (100 g protein) were inverted by Brij58 (0.01%) or treated with urea (7 M) and thiourea (2 M) and sonicated for 5 min. The plasma membrane proteins were digested with trypsin overnight at 37°C or with Lys-C for 4 h at room temperature followed with trypsin for overnight at 37°C. Yeast microsomal membrane proteins containing recombinant AHA2 or His-tag purified AHA2-His 6 were subjected to reduction by DTT and alkylation by IAM and subsequently digested individually with three proteases (trypsin, Lys-C, and Glu-C).
Phosphopeptide Enrichment by TiO 2 Microcolumn-Peptide mixtures from 100 g of plant plasma membrane proteins and 100 -200 g of yeast microsomal membrane proteins containing recombinant AHA2 were desalted with a Poros R3 column as described (52), and phosphopeptide enrichment was performed using titanium dioxide (TiO 2 ) microcolumns as described (53).
Phosphopeptide Enrichment by IMAC-Peptide mixtures from 100 g of plant plasma membrane proteins were desalted with a Poros R3 column, and phosphopeptide enrichment was performed using IMAC as described (54).
Phosphopeptide Enrichment by Calcium Phosphate Precipitation-Peptide mixtures from 100 g of plant plasma membrane proteins were subjected to phosphopeptide enrichment by calcium phosphate precipitation as described (55). The pellet was dissolved with 5% FA, the peptide mixture was desalted with a Poros R3 column, and the bound peptides were eluted with TiO 2 loading buffer (1 M glycolic acid in 80% acetonitrile and 5% TFA). Further phosphopeptide enrichment was performed by TiO 2 microcolumn.
Mass Spectrometry-Samples were analyzed by an EASY nanoflow LC system (Proxeon Biosystems)-coupled LTQ-Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). Samples resuspended in Solvent A (0.5% FA) were loaded onto a custom-made 15-cm analytical column (100-m inner diameter, 375-m outer diameter, packed with Reprosil C18, 3-m reversed-phase particles (Dr. Maisch GmbH, Germany)) at a high flow rate of 550 nl/min. The pep-tides bound to the reversed phase material were eluted with a 50-min gradient of 0 -34% Solvent B (90% acetonitrile, 0.5% FA). The instrument was operated in a data-dependent mode. The peptides were detected in the Orbitrap, and up to five of the most intense peptides were selected and subjected to fragmentation using Multistage Activation (MSA) method in the linear ion trap.
Data Analysis-Raw data from LTQ-Orbitrap MS were processed using Proteome Discoverer 1.0 (Thermo) and searched in an in-house Mascot server (Version 2.2.04, Matrix Science Ltd., London, UK). The NCBI nr data base was used as the searching data base, and A. thaliana and S. cerevisiae was used as taxons when appropriate.
Searching parameters were set as: tryptic (or Lys-C and Glu-C when appropriate) peptides with up to two missed cleavages sites; carbamidomethyl cysteine as a fixed modification; protein N-acetylation-oxidized methionine and phospho_STY, permitted as variable modifications. The results were searched with a peptide mass tolerance of 5 ppm and a fragment mass tolerance of Ϯ0.6 Da. A decoy data base search was performed. Only peptides that identified as peptide rank 1 and with an expected value lower than 0.05 were considered as candidates. All phosphopeptides and phosphorylation sites presented in this work were validated manually.

Attempts to Get Complete Coverage of Phosphosites in Mem-
brane Protein-By employing a systematic approach involving several different methods for enrichment of phosphopeptides, we succeeded in identifying five new in vivo phosphosites in the plant plasma membrane H ϩ -ATPase ( Fig. 1; Table 1). Our results demonstrate that a number of complementary methods are required to get a close to complete coverage of phosphosites present in the membrane-bound transporter.
AHA proteins represented less than 3% of all proteins in the Arabidopsis plasma membranes employed. In previous proteomic studies of H ϩ -ATPase phosphorylation, Brij58 was used to prepare inside-out vesicles before digestion (56). To increase the number of phosphorylated sites detected, we treated the membranes with urea/thiourea and employed different phosphopeptide enrichment methods such as IMAC, TiO 2 , and CaPP followed by TiO 2 . These different approaches resulted in identification of the phosphorylation sites presented in Fig. 1 and Table 1.
The phosphopeptides identified belong to four or five different isoforms of the plasma membrane H ϩ -ATPases in Arabidopsis (AHA1, AHA2, AHA3 and AHA4/11; the last two isoforms could not be distinguished). We detected all phosphorylation sites previously identified in the C-terminal cytoplasmic domain (57,58) with the numbering of AHA2: Thr-881, Ser-899, Ser-904, Ser-931, Ser-944, Tyr-946, and Thr-947, except for one, Ser-904 (Table 1). Furthermore, we identified new phosphorylation sites at the N terminus (Ser-2 and Ser-3), in the P-domain (Ser-544 in AHA1), and in C terminus of (Thr-889 in AHA3 and Ser-931 in AHA2). Interestingly, among the phosphorylation sites identified in planta all but one, Ser-544, were in the N-or C-terminal domains. Phosphorylation of Ser-544 was only observed in AHA1, and this residue is not conserved in other AHAs. As AHA2 is the best characterized plasma membrane H ϩ -ATPase and a crystal structure is available for this pump (59), phosphorylated residues identified in AHA2 were, therefore, analyzed in more detail.
Functional Analysis of Phosphosites in N and C termini of AHA2-A well characterized function of the N-and C-terminal domains of plasma membrane H ϩ -ATPases is to regulate pump function, possibly by restricting domain movements during catalysis and/or blocking access to the H ϩ entry pathway (60). This would suggest that the identified phosphorylation sites in these termini could play a role in regulation of enzyme activity. A tempting hypothesis is that phosphorylation of at least some of these sites abolishes the negative interaction between the termini and the rest of the pump.
To test whether introduction of a negative charge at these positions could result in an activated pump, we performed sitedirected mutagenesis of all Ser (four residues), Thr (five residues), and Tyr (two residues) in the C terminus of AHA2 ( Fig.  2A). Furthermore, we mutagenized Ser-2 and Ser-3 in the N-terminal domain of AHA2 (Fig. 2B). All residues were changed to either a negatively charged residue (Asp or Glu), to mimic the charge effect of phosphorylation, or to a neutral residue (Ala) and expressed in the yeast strain RS-72 (46). Yeast is equipped with its own plasma membrane H ϩ -ATPase, Pma1p, but in the yeast strain RS-72 the endogenous PMA1 gene has been put under the control of a galactose dependent promoter. This implies that introduced plant plasma membrane H ϩ -AT-Pase, the gene of which had been brought under control of the constitutive PMA1 promoter, will be produced and replace Pma1p when the cells are grown on glucose medium.
Single-point mutations at the C terminus had profound effects. In the complementation test, transformed cells expressing mutants of Thr-924, Ser-931, Thr-942, Ser-944, Tyr-946, and Thr-947 exhibited growth that was considerably impaired compared with the wild-type AHA2 no matter whether they were mutated to Ala or Asp/Glu. This suggests a functional importance of these residues. In planta phosphorylation of Thr-947 is known to create a binding site for 14-3-3 protein that activates the pump after its interaction with the pump C terminus (61). When expressed in yeast, endogenous protein kinase(s) phosphorylates Thr-947, which is then allowed to form a complex with yeast 14-3-3 protein with pump activation as the result (62). This provides an explanation why both the T947A and the T947D mutations reduce the ability of AHA2 to complement pma1; T947A because it cannot get phosphorylated in yeast, and T947D because the introduction of a negative charge at this position is not sufficient to create a 14-3-3 protein binding site.
In the case of five AHA2 residues (Thr-881, Ser-899, Tyr-900, and Ser-904), the Ala substitution resulted in a pump with less ability to complement pma1, whereas Asp or Glu substitutions produced a pump that was far better in supporting yeast growth and even better than the wild-type AHA2. This suggests that phosphorylation of these residues results in an activated pump due to neutralization of the C-terminal constraint. Thr-881 has previously been demonstrated as important for regulation of AHA2 activity (44). It is located in the regulatory region R-I (49) and appears to be homologous to Thr-912 in yeast Pma1p, which is phosphorylated in yeast in response to glucose in the medium (63). Glucose activation of Pma1p results in an activated pump with strict coupling between ATP hydrolysis and proton pumping (64).
Substitution of Ser-899 with an Asp had a clear positive effect on the activity of the pump, but this residue is not conserved in plant plasma membrane H ϩ -ATPases other than AHA1 and AHA2. This could indicate an isoform-spe-cific regulation of AHA1/AHA2. aha2 S904D showed the most pronounced activation compared with wild-type AHA2 and aha2 S904A. Ser-904 has previously been identified as a phosphorylated residue in planta (56). Ser-904 is located in the regulatory region R-II (49), a part of the C terminus that is thought to form intramolecular binding to

. Phosphosites identified in plasma membrane H ؉ -ATPases homologously expressed in A thaliana membranes. A, phosphosites were determined in
A. thaliana plasma membrane H ϩ -ATPases using different approaches for phosphopeptide enrichment. Plasma membranes purified from A. thaliana seedlings were exposed for trypsin digestion directly, after treatment with Brij58, or after solubilization by urea/thiourea. IMAC, TiO 2 , and calcium phosphate precipitation followed by TiO 2 (CaPPϩTiO 2 ) were used for phosphopeptide enrichment. Phosphorylation sites in the phosphopeptides were validated manually and are unique for the isoform of AHA mentioned in this figure with following exceptions: Site Ser-931 of AHA2 was determined in the peptide GHVEpSVVK, which can derive also from AHA2, AHA3, AHA5, AHA6, AHA8, AHA9, and Thr-959/955 in the peptide GLDIETIQQAYpTV that is shared by AHA4 and AHA11 isoforms. All the phosphopeptides determined are presented in the supplemental material. B, shown is ClustalW analysis of parts of the N terminus (N), P domain (P), and C terminus (C) of the ATPase isoforms of Arabidopsis. The identified phosphopeptides are underlined, and phosphorylated residues printed in bold and red. Novel residues are boxed. the core of the H ϩ pump protein as well as being important for stabilization of 14-3-3 binding (65, 66).
The C-terminal residue Ser-931 was consistently phosphorylated in planta and showed unusual properties. An Ala substitution at this position resulted in a pump more efficient in complementing pma1, whereas an Asp substitution gave rise to a mutant pump that could not support yeast growth in the absence of Pma1p, i.e. the opposite effect of the corresponding Thr-881 and Ser-904 mutations. Phosphorylation at this position by the Arabidopsis protein kinase PKS5/CIPK11/ SnRK3.22, which is related to Snf1p in S. cerevisiae, has been shown to impair binding of 14-3-3 protein no matter whether the penultimate Thr is phosphorylated or not (48,58).
Mutations of Ser-2 and Ser-3 at the N terminus did not affect the ability of the plant pump mutants to complement pma1 ( Fig. 2A). A functional role of phosphorylation of these residues is, therefore, difficult to deduce. However, it remains a possibility that these events are important for interaction with regulatory proteins(s) present in the natural hosts but absent in the yeast system. In the future it will be important to separate those phosphorylation events that regulate activity from those that might regulate protein-protein interactions, trafficking, etc. in planta.
Phosphorylated Residues in Recombinant AHA2 Heterologously Expressed in Yeast Membranes-A number of Ala substitutions resulted in a plant pump less effective in supporting yeast growth, and this effect could be reversed by an Asp or Glu substitution (see above and Fig. 2). This suggested to us that the involved residues might already be subject to phosphorylation in the heterologous host, which would produce an activated pump. One such example, phosphorylation of AHA2 Thr-947 in yeast, has previously been reported (62,67). We, therefore, decided to determine whether other phosphorylation sites could be identified in recombinant AHA2 produced in yeast.
As a starting point, we found that AHA2 protein represented around 5.5-10% of protein in the microsomal membrane fraction obtained from transformed yeast cells. When membrane fractions containing recombinant AHA2 were subjected to mass spectrometry, several phosphorylation sites were identified in the plant pump (Fig. 3). Surprisingly, seven of nine phos-phorylation sites of AHA2 that we had identified in planta were also phosphorylated in yeast ( Fig. 3 and Table 2: Ser-2, Ser-3, Ser-899, Ser-931, Ser-944, Tyr-946, and Thr-947). The two in planta phosphorylation sites that could not be identified in AHA2 produced in yeast were Ser-544 and Thr-881. In the reverse, in fungus, two phosphorylation sites identified in recombinant AHA2 were not observed in the plant material, namely Thr-511 and Thr-942.
Thr-511 is situated in the cytoplasmic P-domain and was the only residue in yeast expressed AHA2 that was phosphorylated outside the N and C termini. The corresponding phosphopeptide was only observed once in yeast, and phosphorylation of this residue could not be detected in planta. Nevertheless, Thr-511 was analyzed in more detail (Fig. 4), as it is a highly conservative residue in P-type ATPases (68) and is situated in the conserved motif MXTGD and close to the Asp-329, which plays an essential role in the catalytic mechanism of the pump (Fig. 4B). Both Ala and Asp mutants completely abolished yeast growth (Fig. 4A) even though they were expressed at the same levels as wild-type AHA2 (Fig. 4C). This suggests a crucial role of this residue in the functioning of the enzyme. Obviously, in planta phosphorylation at this position, should it occur, would result in effective inhibition of pump activity.

New in Planta Phosphosites Identified in Arabidopsis Plasma
Membrane H ϩ -ATPases-By employing complementary phosphopeptide enrichment methods, we have in this work identified 10 different in vivo phosphosites in different isoforms of the plasma membrane H ϩ -ATPase (Table 1), five of which have not been reported before, bringing the total number of phosphosites in this pump up to 11, which is substantially higher than reported for any other P-type ATPase.
Comparison of Phosphosites in Homologously and Heterologously Produced Protein-Our data demonstrate that a heterologously expressed protein can undergo phosphorylation to a large extent. The phosphorylation patterns of AHA2 expressed in planta and in yeast membranes are strikingly similar ( Table  2). Phosphosites covering Thr-881 and Ser-544 were the only exceptions being observed in material derived from plants only and not in recombinant AHA2 produced in yeast. Common for phosphorylation sites observed in AHA2 whether expressed in planta or in fungus is that they are situated in the N-and C-terminal domain, which in AHA2 play regulatory roles (60). These extensions of the plant plasma membrane H ϩ -ATPase are not conserved in its fungal counterpart Pma1p (supplemental Figs.  1 and 2). For this reason it was unexpected to find a similar phosphorylation pattern between AHA2 expressed in planta and in yeast (Fig. 5B).   FEBRUARY 10, 2012 • VOLUME 287 • NUMBER 7
Several reasons for the observed high similarity between phosphorylation sites in the two distantly related organisms can be proposed. The most obvious possibility is perhaps that the structures of the terminal domains of plasma membrane proton pumps are in extended loose or helical forms that make them extremely accessible to protein kinase action. An extended structure is not unlikely given the fact that at least the C-terminal domain is predicted to make interactions with residues in several cytoplasmic domains (42).
More speculative models could be the following: (a) specific protein kinases phosphorylating plant plasma membrane H ϩ -ATPases are conserved in yeast; (b) both organisms express nonspecific protein kinases that phosphorylate targets at all surface-accessible phosphorylation sites; (c) both organisms have their own palette of protein kinase(s) recognizing essentially the same phosphorylation sites.
Model (a) is unlikely be correct, as there is no sequence similarity between the terminal domains of AHA2 and its yeast counterpart Pma1p (supplemental Fig. 1), and there is no homology between the phosphorylation motifs of AHA2 and any other yeast gene product (supplemental Fig. 2). Still, we cannot entirely rule out the possibility that protein kinases in FIGURE 4. Phosphorylation of highly conservative Thr-551 situated in the catalytic center inhibits pump activity. A, shown is abolishment of yeast growth complementation when mutant in Thr-511 AHA2 was expressed. B,inthe crystal structure of AHA2, Thr-511 is situated close to Asp-329, the conserved Asp phosphorylated during catalysis (59). C, shown is a Western blot analysis of expression level of the mutant AHA2 in yeast cells. the two organisms have the same structure and/or substrate specificities even though not all possible substrates are present in each organism. Model (b) above is not likely to be true as in vivo phosphosites are exclusively located in the terminal domains, whereas Ser and Thr residues, which according to the crystal structure of AHA2 (supplemental Fig. 3; Ref. 59) are exposed in cytosolic domains, fail to become phosphorylated. However, we cannot rule out the possibility that other conformations of AHA2 associated with transport or regulation allow for nonspecific kinases to phosphorylate the cytosolic domains.
According to the Model (c), protein kinases do not evolve according to their protein targets but rather as a group express a variety of potential binding preferences. If this is true, protein targets may preferentially evolve to become the substrate of preexisting protein kinases, which even they have the same specificity, need not be phylogenetically related.
All protein kinases have a very conserved active site that catalyzes phosphoryl transfer from ATP to a protein substrate (69,70). Flexibility is achieved by separating the region of catalysis from that of molecular recognition, the docking groove (71)(72)(73). However, this separation is not absolute, and docking grooves of protein kinases are still intimately connected to the active sites, for which reason protein kinase docking grooves show a limited degree of modularity and evolvability (74 -76). The fact that protein kinase recognition motifs can easily be copied by new targets, whereas docking grooves in protein kinases show less flexibility, provides a molecular rationale for Model (c) above.
AHA2 is so far the first example of a protein being targeted by yeast protein kinases in a similar manner as by protein kinases in its natural host. Whether this is a unique feature of AHA2 or a common phenomenon across kingdoms still needs to be demonstrated.
Conclusion-In this work we have demonstrated the power of combining different methods for enrichment of phosphopeptides when mapping phosphosites in a membrane protein. We identified five new phosphosites in the plant plasma membrane H ϩ -ATPase, bringing the total number of phosphosites mapped in this pump up to 11. Most of the phosphosites are in terminal regulatory domains. When the AHA2 pump isoform is expressed in yeast it is phosphorylated at seven of these sites and at two more not observed in planta. The fact that phosphosites are almost identical in these two homologous and heterologous systems is surprising and could indicate that the terminal domains of proton pumps are unusually accessible for protein kinases.