A plasma membrane syntaxin is phosphorylated in response to the bacterial elicitor flagellin.

In vivo pulse labeling of suspension-cultured Arabidopsis cells with [32P]orthophosphate allows a systematic analysis of dynamic changes in protein phosphorylation. Here, we use this technique to investigate signal transduction events at the plant plasma membrane triggered upon perception of microbial elicitors of defense responses, using as a model elicitor flg22, a peptide corresponding to the most conserved domain of bacterial flagellin. We demonstrate that two-dimensional gel electrophoresis in conjunction with mass spectrometry is a suitable tool for the identification of intrinsic membrane proteins, and we show that among them a syntaxin, AtSyp122, is phosphorylated rapidly in response to flg22. Although incorporation of radioactive phosphate into the protein only occurs significantly after elicitation, immunoblot analysis after two-dimensional gel separation indicates that the protein is also phosphorylated prior to elicitation. These results indicate that flg22 elicits either an increase in the rate of turnover of phosphate or an additional de novo phosphorylation event. In vitro, phosphorylation of AtSyp122 is calcium-dependent. In vitro phosphorylated peptides separated by two-dimensional thin layer chromatography comigrate with two of the three in vivo phosphopeptides, indicating that this calcium-dependent phosphorylation is biologically relevant. These results indicate a regulatory link between elicitor-induced calcium fluxes and the rapid phosphorylation of a syntaxin. Because syntaxins are known to be important in membrane fusion and exocytosis, we hypothesize that one of the functions of the calcium signal is to stimulate exocytosis of defense-related proteins and compounds.

Plants have the capacity to perceive a wide range of "general elicitors" from microbes, i.e. molecules that are typical of entire groups of fungi or bacteria but do not occur in plants. A number of such characteristic non-self molecules have been identified based on their ability to induce rapid biochemical changes in plant cell suspension cultures (1). While the exact role of general elicitors in disease resistance is unclear, many of the initial cellular responses are similar to those triggered by key determinants in pathogen recognition, the race-specific elicitors, or avirulence proteins (2,3). These responses include induction of ion fluxes and an oxidative burst at the plasma membrane, activation of protein phosphorylation cascades, and changes in transcription profiles.
Substantial evidence supports a pivotal role for protein phosphorylation in the transduction of the elicitor signal (4). Some of the plant resistance genes discovered through genetic screens are protein kinases (5,6), and the receptor-like kinase FLS2 is an essential component required for perception of the bacterial elicitor, flagellin (7). Stimulation by elicitors leads to a rapid increase in the incorporation of radioactive phosphate into phosphoproteins (8,9). A number of mitogen-activated protein kinase pathways (10) as well as a calcium-dependent kinase (11) are activated in response to general and racespecific elicitors, in some cases culminating in the activation of defense-related transcription (12). Strikingly, however, few phosphorylated targets of these kinases are known.
We have shown that in vivo labeling in combination with two-dimensional PAGE and mass spectrometry is suitable to identify newly phosphorylated proteins in elicited Arabidopsis cells (8). The complexity of the proteome prompted us to examine cell fractionation as a means to enrich for rare proteins relevant to signal transduction, particularly those at the plasma membrane. The fact that kinase inhibitors block the earliest known plasma membrane responses, such as ion fluxes or the oxidative burst (9), indicates that rapid, membranedelimited protein phosphorylation is part of elicitor signal transduction. Here, we demonstrate that two-dimensional gel electrophoresis is suited to analyze integral plant plasma membrane proteins, and we identify a phosphorylated membrane protein with a potential role in calcium-dependent responses.

EXPERIMENTAL PROCEDURES
Suspension cultures of Arabidopsis thaliana were maintained as described previously (3). Dextran T-500 and immobilized pH gradient strips were from Amersham Biosciences. Flagellin peptide was synthesized on site as described previously (13). All other chemicals were from Sigma or Fluka. POROS TM chromatography materials (R3 and MC) were purchased from Applied Biosystems/Applera UK (Warrington, Cheshire, UK); immobilized alkaline phosphatase was from MoBiTec (Göttingen, Germany).
Cell Treatment and in Vivo Labeling-Cells were treated with 100 nM flg22 peptide (13) for 4 (for in vivo labeling) or 8 -12 min (for unlabeled bulk extract). Labeling with 32 P and phosphopeptide analysis were performed as described (3), except that typically 20 -40 MBq were used per sample. For homogenization of in vivo labeled cells, the homogenization buffer (see below) was supplemented with 5 M leupeptin, 1 M K252a, and 100 nM calyculin A. Cells were broken in a glass potter on ice, and cell debris was pelleted by centrifugation at 3,000 ϫ g for 5 min. The supernatant was removed and centrifuged at 120,000 ϫ g for 30 min at 4°C. The microsomal pellets were resuspended in a small volume of buffer R (see below) and added to a complete phase mixture with microsomes from a large nonradioactive preparation, and the phase was separation performed as described below.
Cell Fractionation-Suspension-cultured cells were collected by filtration and resuspended in ice-cold homogenization buffer (250 mM sucrose, 100 mM HEPES/KOH, pH 7.5, 15 mM EGTA, 5% glycerol, 0.5% polyvinylpyrrolidone K25, 3 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride; with addition of 50 mM sodium pyrophosphate, 25 mM * 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. sodium fluoride, and 1 mM sodium molybdate for phosphoprotein analysis) at 2 ml/g of fresh weight.
The slurry was enclosed in a Parr bomb (Parr Instruments) and stirred for 45 min at 4°C after addition of nitrogen gas to a pressure of 70 bar/1000 p.s.i. Cells were broken by release of the pressure, and the homogenate centrifuged for 10 min at 1,500 ϫ g (GSA rotor, 3500 rpm Ϫ1 ). The supernatant ("total protein") was centrifuged for 30 min at 120,000 ϫ g (Ti45 rotor, 33,000 rpm Ϫ1 ) to yield soluble and microsomal fractions.
For plasma membrane purification, the microsomal pellets were resuspended in buffer R (250 mM sucrose, 5 mM potassium phosphate, pH 7.5, 6 mM KCl) and subjected to phase partitioning (14) in 6.0% each of Dextran T-500 and polyethylene glycol 3350 in buffer R. The U3 phase was diluted about 5-fold with buffer R, and plasma membranes were harvested by centrifugation for 60 min at 150,000 ϫ g. Endomembranes were obtained by 2-fold extraction of L1 with fresh upper phase. For carbonate washing, the membrane pellets were resuspended in a small volume of buffer R plus 0.05% Brij 58, diluted 10-fold in 100 mM ice-cold Na 2 CO 3 , and incubated on ice for 15 min with occasional vortexing. Washed membranes were harvested by centrifugation (30 min at 100,000 ϫ g).
Marker Assays-Enzyme assays were adapted from Ref. 15. Phosphate determination (16) was scaled down to a total 1-ml volume.
Two-dimensional Electrophoresis and Immunoblotting-Precipitation of samples, two-dimensional separation, and immunoblotting was as described in Ref. 8. The two-dimensional loading buffer was changed to 7 M urea, 2 M thiourea, 4% CHAPS, 1 2% Triton X-100, 20 mM dithiothreitol, 1% immobilized pH gradient buffer, and a trace of bromphenol blue.
For affinity purification of the antibody, GST-AtSyp122-(1-41) was expressed and purified with standard methods and coupled to Affi-Gel 10 (Bio-Rad) according to the manufacturer's instructions. Anti-AtSyp122-(1-41) was purified on the affinity column as described in Ref. 17.
Immunoprecipitation-500-l aliquots of cell suspension were labeled with 20 MBq of [ 32 P]orthophosphate before and after 4 min elicitation with 100 nM flg-22. The medium and excess label were removed and 1 ml of ice-cold homogenization buffer (as above) with 1% Triton X-100 was added. Cells were homogenized in a glass potter, insoluble debris was removed by centrifugation (20 min at 20,000 ϫ g, 4°C), and the supernatant incubated with anti-Syp122 bound to protein A-Sepharose for 4 h at 4°C on a spinning wheel. After four washes with homogenization buffer plus 1% Triton X-100 and one wash with ddH 2 O, the beads were incubated with two-dimensional loading buffer (see above) on a shaker, and the solubilized proteins were analyzed by two-dimensional PAGE.
For enzymatic dephosphorylation of unlabeled syntaxin, microsomes were prepared from 10 g of suspension-cultured cells as described above, solubilized with 1% Triton X-100, and the syntaxin was immunoprecipitated. After several washes in homogenization buffer plus Triton, the beads were incubated in phosphatase buffer (50 mM Tris/ HCl, pH 8.0, 0.1 mM ZnCl 2 , 1 mM MgCl 2 ) and dephosphorylated with 40 units of alkaline phosphatase (Roche Applied Science) at 37°C for 90 min.
Generation of Fusion Proteins-A cDNA of AtSyp122 (accession number AJ245407) in pSPORT was kindly provided by Dr. Liliane Sticher, Fribourg, Switzerland.
In Vitro Kinase Assay-Five micrograms of recombinant renatured AtSyp122-(1-289) and 2 g of crude plant cell extract (prepared as above) were mixed in kinase buffer (50 mM HEPES/KOH, pH 7.5, 20 mM MgCl 2 , 1 mM dithiothreitol, 10 mM EGTA, and 5.8 to 9.75 mM CaCl 2 to give the indicated free calcium concentrations, as calculated with EQ-CAL) in a total volume of 25 l. Kinasing was started by addition of 200 M/40 kBq of [␥-32 P]ATP and stopped after 15 min by addition of SDS sample buffer and heating to 95°C for 5 min. For two-dimensional phosphopeptide analysis, the in vivo or in vitro phosphorylated bands/ spots were digested with trypsin and separated by two-dimensional-TLC/TLE as described in Ref. 18.
For MALDI analysis, 10 g of recombinant Syp122 was phosphorylated with 2 g of crude soluble extract in kinase buffer containing 1 M free Ca 2ϩ and 1 mM ATP for 4 h at room temperature. The proteins were precipitated with 80% acetone and digested with 0.2 g of trypsin.
Peptides were adjusted to 5% formic acid and desalted on a POROS R3 microcolumn (19), eluted with 50% acetonitrile, 100 mM acetic acid and incubated for 5 min at room temperature with Fe 3ϩ -charged POROS MC. Washing and elution with dilute ammonia, pH 11, was done as described in Ref. 20. Part of the eluate was re-acidified, bound to a new R3 microcolumn, and eluted onto the MALDI target with saturated 2,5-dihydroxybenzoic acid in 50% acetonitrile; the other part was diluted into phosphatase buffer and dephosphorylated with immobilized alkaline phosphatase. MALDI spectra were acquired on a Bruker Reflex III MALDI-TOF mass spectrometer.

RESULTS
Proteomics for Plasma Membrane Proteins-We have used aqueous two-phase partitioning with polyethylene glycol 3350/ Dextran T500 to obtain plasma membranes of high purity from suspension-cultured cells of Arabidopsis. Typical preparations were enriched by a factor of 7 to 9 in the plasma membrane (PM) marker, vanadate-sensitive ATPase activity, compared with the starting material consisting of a crude microsomal fraction (Table I). Mitochondrial and chloroplast membranes were virtually absent from the PM-enriched preparation, whereas other endomembranes (tonoplast, Golgi, and endoplasmic reticulum) were depleted by a factor of at least four to six.
For two-dimensional PAGE analysis, plasma membranes were washed with sodium carbonate. This step removed a large proportion of peripheral proteins and was essential to recover integral membrane proteins. High concentrations of detergents (CHAPS and Triton X-100) and thiourea (21) during the focusing step were important to achieve a good resolution (Fig. 1a).
To determine whether integral membrane proteins were being resolved on the two-dimensional gels under these conditions, 30 spots were isolated and analyzed by MALDI-MS. Of 19 identified proteins, five were predicted to be integral membrane proteins containing a single transmembrane helix (Table  II), and four of these are predicted to be located in the PM (PSORT). 2 In addition, several peripheral membrane proteins were identified (Table II). The other soluble proteins were predicted to be cytoplasmic (4), secreted (4), nuclear (1), or in the chloroplast (1) (data not shown). Thus, this approach was successful in resolving plasma membrane proteins with at least one transmembrane helix, but it was not possible to eliminate all contaminating proteins from other organelles and the cytoplasm.
Analysis of PM Phosphoproteins-To analyze phosphoproteins in the PM-enriched fraction, the preparative-scale twophase system was supplemented with microsomes from in vivo 32 P-labeled cells. After partitioning, PM proteins were separated by two-dimensional PAGE as before, and the gels analyzed by PhosphorImaging. Relatively few phosphorylated protein spots were visible on pI 3-10 non-linear two-dimensional gels of in vivo labeled PM samples (Fig. 1b). Judged from the position and pattern of labeled spots, the PM-specific set was undetectable in total labeled extracts, whereas the dominant soluble phosphoproteins (8) were absent from the PM gels (data not shown). This result demonstrates the advantage of cell fractionation for enrichment of rare proteins.
Several spots showed increased phosphate incorporation in the elicited sample (see boxed region in Fig. 1b, especially the marked "doublet"), whereas one doublet was less strongly labeled after elicitation. The six spots in the marked box were the most reproducible changes in numerous experiments and, thus, c, phosphorimage and silver stain of the boxed region (b, elicited sample) on an expanded gel (pI 5.5-6.7), indicating the position of the 41/44-kDa doublet. d, masses and peptide sequence tags (sequence reads from the C to N terminus) of three tryptic peptides that lead to the identification of PM41 and PM44 as the syntaxin AtSyp122. Peptide sequence tags (22) comprise a partial amino acid sequence, the total mass of the amino acids C-terminal to that sequence, and the total mass of that C terminus plus the mass of the sequenced amino acids. were the targets of further experiments. On an expanded gradient resolving pI 5.5-6.7 (Fig. 1c), better spot separation was achieved and allowed matching of the marked 32 P-labeled spots with silver-stained spots. Identification by electronanospray mass spectrometry was so far successful for the induced doublet marked with arrowheads. Virtually the same set of tryptic peptide masses was found for both spots, and sequence tags from three peptides (Fig. 1d) were identical. The peptide sequence tags (22) identified both proteins as a putative syntaxin (At3g53400), named AtSyp122 according to the nomenclature of A. Sanderfoot (23). Among the over 20 Arabidopsis syntaxins, AtSyp122 belongs to a group of five putative plasma membrane/phragmoplast syntaxins. 3 The ClustalW analysis in Fig.  2 shows the high degree of sequence conservation in the H3 domain that is in contact with the other SNAREs. All three identified peptides (enlarged) are from this region, but the sequence information (in bold type) and/or masses of the peptides were sufficient to identify the phosphoproteins unambiguously as AtSyp122. The sequence of the closest paralog, At-Syp121, and the equivalent tryptic peptides with their masses are included for comparison. It should be noted that the 41/44-kDa doublet is unlikely to arise from proteolytic processing because all possible cleavages would affect the pI of the protein. The larger form may be glycosylated, as has been reported for human syntaxin 1a (24).
Confirmation That AtSyp122 Is Phosphorylated in Response to Microbial Elicitation-To confirm that AtSyp122 is the differentially phosphorylated protein and not just a more abundant, co-migrating protein, we performed immunoblots and immunoprecipitations with an antibody raised against the recombinant, hexahistidine-tagged cytoplasmic domain of At-Syp122. Immunoblotting showed a characteristic 34/38-kDa doublet and confirmed the membrane localization and plasma membrane enrichment of AtSyp122 (Fig. 3a). To ensure specificity of the antibody against AtSyp122, we affinity purified an antibody fraction recognizing the unique N terminus of the protein (see sequence comparison in Fig. 2). Affinity purification completely abolished cross-reactivity against the most sim-ilar paralog, AtSyp121 (Fig. 3b). Using this batch of antibody, we analyzed one-and two-dimensional immunoblots of plasma membranes from control and elicitor-treated cells (Fig. 3, c-e). The characteristic 41/44-kDa double spot (Fig. 3c, marked with an arrow) appears only in the elicited sample. Strikingly, however, this induced doublet is only a very minor species relative to the three other doublets found in both control and treated membranes. The presence of signals with four different pI values indicates three phosphorylation sites, with the most basic doublet corresponding to the unphosphorylated protein.
In addition to changing the pI of the protein, phosphorylation also increases the apparent molecular mass by 3 kDa. The apparent high molecular weight form appears transiently after elicitation (Fig. 3d) and is sensitive to phosphatase treatment (Fig. 3e), demonstrating transient, elicitor-induced phosphorylation of Syp122.
The affinity purified antibody proved unsatisfactory for immunoprecipitation (data not shown), possibly because of steric hindrance or phosphorylation site(s) in this domain (25) (see below). Fig. 3d shows a two-dimensional analysis of immunoprecipitated syntaxin from in vivo labeled cell extracts using the unpurified antibody. Although the minor spots may arise from copurifying and/or cross-reacting proteins, the major labeled doublets migrate to similar positions as the confirmed AtSyp122 spots (Fig. 3c). The most basic doublet in the immunoblots (Fig. 3c) is not labeled, confirming that it represents the unphosphorylated protein. We conclude from these results that the protein identified as AtSyp122 is phosphorylated during responses to microbial elicitation.
Calcium-dependent Kinase Activity Phosphorylates At-Syp122-To characterize the kinase activity responsible for AtSyp122 phosphorylation, in vitro assays with recombinant AtSyp122-(1-289)-His 6 and cell extracts as kinase source were performed. In vitro phosphorylation was strongly enhanced by low micromolar concentrations of calcium (Fig. 4a). Although calcium-dependent kinases are often associated with membranes (26,27), we found that the kinase activity was more strongly associated with the soluble than the membrane fraction (data not shown). A comparison of tryptic phosphopeptides separated by two-dimensional thin-layer chromatography/elec-3 tc.umn.edu/ϳsande099/atsnare.htm.

FIG. 2. Schematic diagram of AtSyp122 domain organization and ClustalW analysis of N-terminal and H3 domains of syntaxin 1a
from Drosophila melanogaster (Dm, accession number AAB34841), Caenorhabditis elegans (Ce, T37265), Homo sapiens (Hm, JN0466), and Arabidopsis syntaxins AtSyp121 (At3g11820) and AtSyp122 (At3g53400). Sequences of the tryptic peptides that allowed the identification of AtSyp122 (Fig. 1d) are enlarged, with the MS/MS peptide sequence tags (see Fig. 1d) in bold type, and their predicted masses annotated. The corresponding peptides of the closest paralogue AtSyp121 are included to show that the identification was unambiguous.
trophoresis revealed that two of the three phosphopeptides generated in vitro by a calcium-dependent kinase activity comigrate with authentic in vivo phosphopeptides (Fig. 4b). This result indicates that the calcium-dependent kinase activity mediates the AtSyp122 phosphorylation observed after elicitation in vivo.
The in vitro kinase assay allowed us to prepare sufficient phosphorylated syntaxin to characterize the phosphopeptide(s) by mass spectrometry. Using immobilized metal ion affinity chromatography to enrich for phosphopeptides (IMAC) (20), we found two peptides that showed an 80-Da mass shift upon treatment with phosphatase and hence are phosphopeptides (Fig. 4c). The 1191.49-Da peptide corresponds to the phosphorylated extreme N terminus. The smaller peptide does not match any of the predicted peptides, but because of the mass difference to the bigger peptide (131.01 Da in the phosphorylated peptides), it probably represents the same peptide after cleavage of the start methionine. Calcium-dependent phosphorylation of the N terminus was confirmed in an in vitro assay using the GST-Syp122-(1-41) fusion (data not shown). The N-terminal phosphopeptide contains two potential phosphorylation sites (Ser-6 and Ser-8), of which the latter is conserved among the Syp1 family. Comigration of both in vitro phosphorylated peptides with the in vivo peptides would suggest that the syntaxin is also partially cleaved in vivo (i.e. in Arabidopsis). Therefore, the three spots in the in vivo map may repre- (2), GST-AtSyp122-(1-41) (3), and AtSyp121-(1-290)-His 6 (4) was loaded. c, two-dimensional immunoblots of plasma membrane proteins, probed with affinity purified anti-AtSyp122. The arrowheads in c-f points at the elicitor-induced, apparent high molecular weight spot/ band. d, one-dimensional immunoblot of microsomal proteins, probed with affinity purified anti-AtSyp122; time course after elicitor treatment is shown. e, immunoblot of immunoprecipitated Syp122 (6 min after elicitor treatment), mock-(Ϫ), or alkaline phosphatase-treated (AP). The filled arrowhead marks the dephosphorylated Syp122 with apparent low molecular weight. f, immunoprecipitation of syntaxin from in vivo labeled extracts using the crude antiserum. sent only two phosphorylation sites with the third arising from ϩ/ϪMet heterogeneity of one peptide. The two-dimensional electrophoresis separation of the syntaxin predicts three phosphorylation sites (Fig. 3C). However, assuming ordered phosphorylation, only the second and third phosphorylation sites are strongly labeled under the experimental conditions (see Fig. 3f). Thus, these may correspond to the only peptides detected in the two-dimensional TLC-TLE experiment. DISCUSSION We have performed a proteomic analysis of PM proteins phosphorylated in response to the microbial elicitor flg22. Multiple roles in growth, development, and the exchange of information and substances make the plant plasma membrane an important object of study. However, biochemical research on its protein composition is challenged by the low abundance and difficulty in handling of hydrophobic membrane proteins. Proteomic studies have identified a small number of true transmembrane proteins on two-dimensional gels (28,29). Using established protocols for the analysis of hydrophobic proteins, we obtained well resolved two-dimensional gels of carbonatewashed plasma membranes from cell cultures. Judging from a sample of 19 identified proteins, integral membrane proteins are under-represented, and some endomembrane contamination was detected. It is not possible, thus, to deduce plasma membrane localization from the fact that a protein is found in this fraction. Moreover, multispanning membrane proteins were not found and may be missing altogether. New peptidebased mass spectrometric strategies have been developed that overcome the bias against these proteins (30), but the unrivaled capacity of two-dimensional gels to detect post-translational modifications by size/IEP shifts makes it desirable to develop the technology further for membrane proteins. Even if it is limited to single-pass membrane proteins, the presented strategy strongly improves the chance to find intrinsic PM proteins "in the crowd," which was the primary aim in this study. Several proteins enriched in the PM fraction were found to be differentially phosphorylated, among them the syntaxin AtSyp122.
Syntaxins are involved in membrane trafficking in all eukaryotic cells. They represent part of the target membrane receptors for complementary structures on the docking vesicles (31). The formation of a tight complex between two target (t-) SNAREs, syntaxin and a SNAP25-like protein, and one or more vesicular (v-) SNARE(s) is necessary to drive membrane fusion. In this capacity, syntaxins appear to be a major site of regulatory control, and a large number of proteins bind to syntaxins and modulate complex formation (32).
AtSyp122 has five paralogues comprising the AtSyp1 cluster. These proteins are related to yeast and mammalian plasma membrane syntaxins, although less closely than orthologous plant, yeast, and mammalian endomembrane syntaxins (33). Two members of the AtSyp1 cluster, AtSyp111 ("KNOLLE") and AtSyp121 (AtSyr1), have been previously characterized and their subcellular localization determined. AtSyp111 resides in the phragmoblast and directs Golgi-derived vesicles to the growing cell plate during cytokinesis (34). AtSyp121 is the closest relative of the phosphoprotein described in this work, AtSyp122. Its tobacco orthologue, NtSyr1, is a PM protein (35) that is essential for abscisic acid signal transduction (36). Our data from cellular fractionation studies indicate that AtSyp122 is also located at the PM.
Based on its similarity with plasma membrane syntaxins, AtSyp122 would be expected to function in the exocytosis of secretory vesicles (37). Microbial infections generally induce the transcriptional activation and secretion of a variety of pathogenesis-related proteins that are thought to play a role in general defense (38). In a cell culture/elicitor model system, gene induction can be as rapid as 15 min after stimulation (39), but this response would still be significantly later than the phosphorylation we observed in this study. However, during a fungal attack, heavy vesicle traffic toward the site of attempted infection is induced within minutes (40). Similarly, flagellin stimulates callose deposition in plants (41). Thus, one role for exocytosis during defense responses may be the strengthening of the cell wall to prevent penetration by the microbe (40). In addition, the syntaxin also may mediate the secretion of a preformed pool of antimicrobial compounds and/or (poly)peptides.
Apart from vesicle traffic, the regulation of ion channels is a unique feature of plasma membrane syntaxins (42) and was also confirmed for the tobacco syntaxin NtSyr1 (36). For N-type calcium channels, direct inhibitory interactions with human syntaxin 1 have been established (43), whereas for the cystic fibrosis transmembrane conductance regulator channel, both direct interaction and regulation of cell surface concentration via exocytosis are being discussed (44). In plant cells, elicitors induce rapid changes of ion fluxes in plant cells, and it is conceivable that AtSyp122 plays a role in modulating these fluxes.
The calcium-dependent phosphorylation of AtSyp122 suggests a novel way of linking an upstream calcium signal with exocytosis. In all cases investigated, rapid responses to microbial elicitors include elevation of intracellular calcium concentrations (45); and this response also has been observed after treatment with flagellin. 4 In addition, a calcium-dependent protein kinase has been shown to be an essential factor in defense signal transduction (11). Our finding that in vivo phosphorylation of AtSyp122 is most likely calcium-dependent indicates that one of the functions of calcium-dependent protein kinases in decoding calcium signals may be to regulate exocytosis. Plant cells contain pools of vesicles that can be secreted upon artificially increasing cytosolic calcium (46), reminiscent of the "readily releasable pool" of synaptic vesicles. In the context of polarized plant cell growth, the link between local calcium influx and exocytosis is well established (47). In the case of increased cell wall deposition during defense responses, however, this area will need to be examined in greater detail.
What may be the function of syntaxin phosphorylation, particularly if the levels of the phosphorylated forms barely change during the stimulus? Although well documented in vitro (25,48), only a few reports exist on stimulated phosphorylation of a syntaxin in vivo. It is interesting that the Nterminal phosphorylation site of Syp122 (upstream of the coiled-coil H ABC domain) is similar to Ser-14 phosphorylation of human syntaxin 1A (49), but no stimulus is known yet that induces the latter phosphorylation. Activation of platelets by thrombin induces phosphorylation of syntaxin 4 by protein kinase C (50), which leads to reduced binding to SNAP23, its partner t-SNARE. A weak disruptive effect on binary SNARE interactions is generally observed after in vitro phosphorylation by various kinases (51). If SNARE complex formation is regulated by phosphorylation, it has to be a cyclic process that includes dephosphorylation because SNAREs are known to be recycled after membrane fusion (31,52). We believe that our data support the hypothesis of a phosphorylation-dephosphorylation cycle, and that the elicitor stimulus increases the turnover in this cycle. The following scenario integrates our data with established knowledge. Exocytosis results in ciscomplexes of v-and t-SNAREs in the plasma membrane (31). These complexes need to be disassembled, possibly aided by phosphorylation, to allow new rounds of vesicle fusion (53). When a secretory vesicle docks at the PM, dephosphorylation is necessary to allow SNAREs to engage into new complexes (54). Our data support this model because AtSyp122 has a constitutive phosphorylation level in untreated cells (as seen on twodimensional immunoblots), but can be pulse-labeled with [ 32 P]phosphate only after elicitor stimulation, indicating that it has been dephosphorylated (and probably "in use") before. In accordance, the yeast plasma membrane syntaxin Sso1p is in vivo constitutively phosphorylated on a protein kinase A site, and dephosphorylation by a ceramide-activated type 2A phosphatase allows exocytosis (55). Notwithstanding this hypothesis, a distinct signaling function for the triply phosphorylated form of AtSyp122, which is only induced transiently in elicited cells, is conceivable as well, provided the phosphorylation is sequential (nonrandom) and a new phosphorylated site is generated. Our results demonstrate the benefit of a combination of pulse labeling and immunoblotting to gain insight into possible mechanisms. Whereas in vivo labeling reveals rapid changes of kinase activity and/or turnover of phosphorylation on proteins, only immunoblots unambiguously show steady-state levels of different phosphorylated forms.
In summary, we have developed a reproducible method for isolating plasma membrane proteins and resolving these hydrophobic proteins on two-dimensional gels. Using this technology in conjunction with a functional proteomic analysis of elicitor responses, we discovered a novel phosphoprotein in Arabidopsis, the syntaxin AtSyp122. Finally, the calcium dependence of AtSyp122 phosphorylation suggests a novel mechanism of SNARE regulation through calcium.