A Proteomics Dissection of Arabidopsis thaliana Vacuoles Isolated from Cell Culture*S

To better understand the mechanisms governing cellular traffic, storage of various metabolites, and their ultimate degradation, Arabidopsis thaliana vacuole proteomes were established. To this aim, a procedure was developed to prepare highly purified vacuoles from protoplasts isolated from Arabidopsis cell cultures using Ficoll density gradients. Based on the specific activity of the vacuolar marker α-mannosidase, the enrichment factor of the vacuoles was estimated at ∼42-fold with an average yield of 2.1%. Absence of significant contamination by other cellular compartments was validated by Western blot using antibodies raised against specific markers of chloroplasts, mitochondria, plasma membrane, and endoplasmic reticulum. Based on these results, vacuole preparations showed the necessary degree of purity for proteomics study. Therefore, a proteomics approach was developed to identify the protein components present in both the membrane and soluble fractions of the Arabidopsis cell vacuoles. This approach includes the following: (i) a mild oxidation step leading to the transformation of cysteine residues into cysteic acid and methionine to methionine sulfoxide, (ii) an in-solution proteolytic digestion of very hydrophobic proteins, and (iii) a prefractionation of proteins by short migration by SDS-PAGE followed by analysis by liquid chromatography coupled to tandem mass spectrometry. This procedure allowed the identification of more than 650 proteins, two-thirds of which copurify with the membrane hydrophobic fraction and one-third of which copurifies with the soluble fraction. Among the 416 proteins identified from the membrane fraction, 195 were considered integral membrane proteins based on the presence of one or more predicted transmembrane domains, and 110 transporters and related proteins were identified (91 putative transporters and 19 proteins related to the V-ATPase pump). With regard to function, about 20% of the proteins identified were known previously to be associated with vacuolar activities. The proteins identified are involved in ion and metabolite transport (26%), stress response (9%), signal transduction (7%), and metabolism (6%) or have been described to be involved in typical vacuolar activities, such as protein and sugar hydrolysis. The subcellular localization of several putative vacuolar proteins was confirmed by transient expression of green fluorescent protein fusion constructs.

the yeast vacuole or animal lysosome. PSVs are particularly prominent in developing seeds (4). PSVs contain vacuolar storage proteins to be used for anabolism during seedling growth. The function of the different vacuoles seems to be correlated with the presence of specific tonoplast intrinsic protein (TIP) isoforms (5,6). Current knowledge of the cellular traffic of higher eukaryotes indicates that vacuole biogenesis is closely related to the traffic of proteins resident in these compartments (7)(8)(9)(10). Resident vacuolar proteins, as well as proteins intended for degradation, are delivered to the vacuole via the secretory pathway that includes the biosynthetic, autophagic, and endocytotic transport routes (2, 10 -14). Although some aspects have been studied in detail (13,15), many questions remain unanswered concerning autophagy, transport, and fusion with small vacuoles. The analogy with the processes set up by yeast or higher eukaryotes seems to indicate that these mechanisms may be common to all kingdoms (16).
Most of the compounds present in vacuoles have to be transported in a passive or active manner across the tonoplast (the vacuolar membrane) for storage or degradation, but they also need to be exported in response to plant cell demands. Surprisingly the number of transporters that have been identified on the tonoplast is quite low (17). Two proton pumps known as primary active H ϩ transport systems are present in this membrane: the vacuolar-type H ϩ -pumping ATP hydrolase (H ϩ -ATPase, VHA) (18,19) and the H ϩ -pumping pyrophosphatase (pyrophosphate-energized H ϩ -PPase, AVP1) (17,20). They are responsible for the acidification of the vacuolar lumen, thus creating proton concentration and electrical gradients across the tonoplast. It was recently shown that H ϩ -PPase also controls auxin transport and consequently auxin-dependent development (21). The tonoplast also contains secondary active transporters energized by the proton motive force and several other pumps (1,17). A Na ϩ /H ϩ antiporter (AtNHX1) is present in the tonoplast and mediates Na ϩ sequestration in the vacuole. This transporter contributes to the plant salt tolerance of transgenic Arabidopsis overexpressing AtNHX1 (22,23) and was recently shown to be regulated by calmodulin (24). The free cytosolic Ca 2ϩ concentration must also be strictly regulated as it controls many essential cellular responses (25). The tonoplast contains Ca 2ϩ /H ϩ antiporters (CAX1 and CAX2) (26 -28) that are re-sponsible, in conjunction with a Ca 2ϩ pump (P 2B -type ATPase, ACA4) (29), for the sequestration of Ca 2ϩ in the vacuolar sap (30). It was proposed recently that CAX1 regulates several plant processes including ion homeostasis, development, and hormonal responses (28). Other metal transporters have also been identified in the tonoplast. These include a Mg 2ϩ /H ϩ exchanger (AtMHX); a cation diffusion facilitator family member, MTP1 (zinc transporter of Arabidopsis (ZAT)); and the AtNramp3 and AtNramp4 transporters. AtMHX functions as an electrogenic exchanger of protons with Mg 2ϩ and Zn 2ϩ ions (31). By sequestering excess cellular zinc in the Arabidopsis thaliana vacuole, MTP1 is involved in zinc homeostasis and detoxification (32)(33)(34). This transporter is probably involved in zinc tolerance in the zinc hyperaccumulator Arabidopsis halleri (35). AtNramp3 and AtNramp4 have been shown recently to be present in the tonoplast and to participate specifically in iron mobilization from vacuolar metal stores during seed germination (36,37). Some ATPbinding cassette (ABC) transporters are also present in the tonoplast, such as MRP2 that has been shown to be not only competent in the transport of glutathione conjugates but also in the transport of glucuronate conjugates following its heterologous expression in yeast (38). AtMRP1 is also localized to the vacuolar membrane of Arabidopsis and interacts with an immunophilin-like protein (twisted Dwarf protein 1 (TWD1)) through a calmodulin-binding domain present in the C terminus of AtMRP1 (39).
Understanding the key steps involved in the transport process of substrates to the vacuole and their storage depends on the identification of additional membrane proteins. Recently proteomics analyses of the tonoplast have been published (40 -42). Shimaoka et al. (40) identified a large number of mostly soluble proteins within their vacuolar fractions. Fortytwo of the 163 proteins were annotated with one or more transmembrane domains, and 39 proteins were predicted to have more than two transmembrane domains, 17 of which were putative transporters. The procedure of Szponarski et al. (41) allowed characterization of 70 proteins from an Arabidopsis tonoplast-enriched fraction, including only a small number of transporters. The most complete study published so far identified 402 proteins (42). However, almost half of the proteins listed were identified by a single peptide hit, which is often insufficient for certain identification. From these proteins, 29 were putative or known transporters, and 17 were related to the H ϩ -ATPase complex. Taken together, all these previously published results indicated the need to extend the knowledge of the vacuolar proteome of Arabidopsis.
In the present study, intact vacuoles were isolated from Arabidopsis suspension cells. Potential cross-contaminations were examined by Western blot analyses, and the quality of the vacuole preparations led us to a proteomics investigation. A proteomics approach was developed to identify the protein components present in the membrane and the soluble fractions of the Arabidopsis cell vacuoles. This approach includes p22HBP, heme-binding protein; PTR2-B, peptide transporter 2B; PDR, pleiotropic drug resistance; PMA, plasma membrane H ϩ -ATPase; POT, proton-dependent oligopeptide transporter; SPFH, stomatin prohibitin flotillin Hbc; TAP, transporter associated with antigen processing; TIP, tonoplast intrinsic protein; TOM, translocase of the mitochondrial outer membrane; TRH, tiny root hair protein (AtTRH1 ϭ KuP4); VHA, vacuolar-type H ϩ -pumping ATP hydrolase; YS1, Yellow Stripe 1; YSL, Yellow Stripe-like; GFP, green fluorescent protein; ER, endoplasmic reticulum; FA, formic acid; TMD, transmembrane domain; FMA, false mass assignment; HDEL, His-Asp-Glu-Leu peptide. a mild oxidation step leading to the transformation of the cysteinyl residues into cysteic acid and methionine into methionine sulfoxide, which facilitates peptide assignment; an in-solution proteolytic digestion of membrane proteins; and/or a prefractionation of proteins by SDS-PAGE. Peptides were identified using liquid chromatography coupled to tandem mass spectrometry. The combined results of these approaches allowed the identification of over 650 proteins. Among these, 195 were considered as integral membrane proteins based on the presence of one or more predicted transmembrane domains, and 91 transporters were identified. The subcellular localization of several putative vacuolar proteins was confirmed by transient expression in Arabidopsis protoplasts overexpressing GFP fusion proteins.

EXPERIMENTAL PROCEDURES
Cell Material and Growth Conditions-A. thaliana cells (var. Columbia) were cultivated at 22°C under constant light (80 mol of photons⅐m Ϫ2 ⅐s Ϫ1 ) and shaking (135 rpm) in a Murashige and Skoog medium (MS basal medium, Sigma, catalog number M5519, pH 5.5) containing 88 mM sucrose and 0.02 g⅐liter Ϫ1 2,4-dichlorophenoxyacetic acid. Every 7 days, 100 ml of fresh medium were inoculated with an aliquot of the culture of 7-day-old cells (5 ml of packed cells after centrifugation at 110 ϫ g for 5 min).
Vacuole Isolation-Protoplasts were obtained after digestion of 5-day-old cells in 0.6 M mannitol, 2% (w/v) cellulase, 0.5% (w/v) pectolyase, 25 mM MES, pH 5.5. After 2 h, protoplasts were filtered through a 50-m nylon net, and the suspension was centrifuged for 1 min at 1270 ϫ g. The protoplast pellet was washed with a rinsing medium (0.7 M mannitol, 10 mM Tris, 15 mM MES, pH 7.0) and adjusted to a final concentration of 52 Ϯ 4 ϫ 10 6 protoplasts⅐ml Ϫ1 . Vacuoles were then purified following a protocol adapted from Frangne et al. (43). The protoplast suspension was diluted 4-fold in a lysis medium prewarmed to 42°C (medium A: 0.2 M mannitol, 10% (w/v) Ficoll 400, 20 mM EDTA, 5 mM HEPES-KOH, pH 7.5). After a 15-min incubation, vacuoles were isolated using a three-step gradient. The suspension of protoplasts was loaded at the bottom of a centrifuge tube and covered with 2 volumes of a mixture (1:2, medium A:medium B) (medium B: 0.4 M betaine, 30 mM KCl, 20 mM HEPES-KOH, pH 7.5) and 1 volume of medium B. The gradient was then centrifuged at 1500 ϫ g for 20 min, and vacuoles were collected at the interface of the first and second layers corresponding to 0 and 3.3% Ficoll. The vacuoles were then concentrated by centrifugation (1800 ϫ g for 20 min).
Enzyme Activities and Western Blot Analyses-Purification yield of vacuole preparations was followed by enzymatic assay of the vacuole-specific marker ␣-mannosidase according to Boller and Kende (44). Protein (10 -100 g) was added to a medium containing 50 mM citric acid-NaOH (pH 4.5) and 1 mM p-nitrophenyl-␣-D-mannopyranoside. After 20, 40, and 60 min of incubation at 37°C, the reaction was stopped by adding 0.8 ml of 1 M Na 2 CO 3 /0.5 ml of assay medium. After centrifugation (16,000 ϫ g for 15 min), enzyme activity was evaluated by detecting the product, p-nitrophenol, at 405 nm (⑀ pNP(405 nm) ϭ 18.5 mM Ϫ1 ⅐cm Ϫ1 ).
Purity of the vacuole was controlled by Western blot analyses. Fifteen micrograms of proteins from protoplast or vacuole extracts were loaded on a 12% SDS-polyacrylamide gel and, after migration at 200 V, transferred to a nitrocellulose membrane. The presence of different characteristic proteins was assessed using Western blot analyses and primary polyclonal antibodies raised against the outer envelop protein 21 (45) and the light-harvesting complex b (O. Vallon, Institut de Biologie Physico-Chimique, Paris, France) (plastid markers), the preprotein translocase of the mitochondrial outer membrane (TOM40, channel-forming subunit (46)), the plasma membrane P-type H ϩ -ATPase (PMA2 (47)), and the HDEL domain of the endoplasmic reticulum (ER) proteins (48,49). Antibodies raised against tonoplastic markers such as the tobacco ␣-TIP (50) and the cauliflower ␥-TIP (51) were used to control for vacuole enrichment.
Vacuolar Protein Preparation and Trypsin Digestion for Mass Spectrometry Analyses-Vacuoles were disrupted by a freeze/thaw cycle in nitrogen. The suspension was then centrifuged for 60 min at 100,000 ϫ g. The pellet containing the membranes was resuspended in 100 l of 100 mM ammonium bicarbonate (pH 8.2). The mixture was then adjusted to 1500 l with 0.55 M NaCl and incubated at 4°C for 60 min with constant shaking.
For in-gel digestion, 20 g of proteins from either the soluble or the membrane fractions were separated by a short electrophoresis (2.5 cm) by 12% SDS-PAGE. After coloration with Coomassie Blue R250, the gel was cut into 1.5-mm slices. Each band was further cut and washed twice in 100 l of destaining solution (50 mM NH 4 HCO 3 / CH 3 CN, 50:50 (v/v)) at room temperature for 30 min before dehydration with 100 l of pure ACN. The solution was then removed, and the gel pieces were dried in a speed vacuum and rehydrated in 100 l of 7% H 2 O 2 at room temperature for 15 min in the dark. The oxidizing solution was removed, and the gel slices were rinsed in water as previously described above. After complete drying, the bands were rehydrated in 20 l of digestion buffer (150 mM Tris-HCl, 10 mM CaCl 2 , 100 mM urea, pH 8.1/CH 3 CN, 95:5 (v/v) containing 150 ng of sequencing grade modified trypsin (Promega, Madison, WI)). After 15 min of incubation at 4°C, 30 l of digestion buffer were added, and the digestion reaction was carried out at 37°C for 5 h with constant shaking. The digestion solution was then collected, and peptides were extracted from the gel by diffusion in 50 l of 0.3 M urea, 90% (v/v) CH 3 CN for 30 min with sonication. Digestion and extraction solutions were pooled and dried in a speed vacuum. Peptide mixtures were redissolved in 25 l of water/CH 3 CN (95:5, v/v) containing 0.2% formic acid (FA) prior to LC-MS/MS analysis.
For in-solution digestion, the vacuolar membrane fraction was resuspended in 80 l of 25 mM NH 4 HCO 3 , pH 8.1, and heated at 90°C for 10 min. Denaturation was stopped by adding 120 l of cold (Ϫ20°C) methanol, thus avoiding protein refolding, which can occur during slow cooling. Digestion was carried out overnight at 35°C with an enzyme/substrate ratio of 1:100 (w/w). Finally tris(2-carboxyethyl)phosphine hydrochloride was added to a final concentration of 10 mM, and reduction was carried out at 35°C for 30 min. The digestion solution was then dried in a speed vacuum, and peptide mixtures were redissolved in 25 l of water/CH 3 CN (95:5, v/v) containing 0.2% FA prior to LC-MS/MS analysis.
Nano-LC-ESI-MS/MS-Injected samples (6 l) from in-gel digestion were first trapped and desalted isocratically on a PepMap C 18 65-mm precolumn cartridge (300-m inner diameter, 5 m, and 100 Å; Dionex, Sunnyvale, CA). Chromatographic separation was accomplished by loading peptide samples onto a 15-cm fused silica C 18 column (75-m inner diameter, 3 m, 100 Å, and 360-m outer diameter; Dionex) using an autosampler. Sequential peptide elution was achieved using the following linear gradient: (i) from 10 to 40% solvent B (CH 3 CN/water (90:10, v/v) containing 0.1% FA) for 40 min and (ii) from 40 to 90% solvent B for 5 min. (iii) The remaining percentage of the elution solvent was made of solvent A (water/ CH 3 CN (95:5, v/v) containing 0.1% FA). Flow rate through the nano-LC column was set to 200 -300 nl/min. The mass spectrometer was calibrated using the product ions generated from fragmentation of the doubly charged molecular ion of Glu-fibrinopeptide B. Raw data were processed using PeptideAuto (ProteinLynx, MassLynx 4.0) using smooth 3/2, Savitzky Golay. The mass spectrometer was op-erated in the positive ion electrospray ionization mode with a resolution of 9,000 -11,000 full-width half-maximum. For automatic LC-MS/MS analysis the Q-TOF Ultima instrument was run in datadependent mode with the following parameters: 1-s scan time and 0.1-s interscan delay for MS survey scans; 400 -1400 and 50 -2000 m/z mass ranges for the survey and the MS/MS scans, respectively; five components; MS/MS to MS switch after 5 s; switchback threshold, 30 counts/s; include charge states 2, 3, and 4 with the corresponding optimized collision energy profiles. A list of the m/z corresponding to the most intense peptides of trypsin was set as an exclude list. Peptide identification from the resulting MS/MS dataset was achieved using an in-house Mascot server (version 2.0) (Matrix Sciences, London, UK). Chromatographic separation of in-solution digested proteins was accomplished by loading 0.15 g of peptide mixture on the column. Sequential elution of peptides was performed using the following linear gradient: (i) from 15 to 60% solvent B for 90 min and (ii) from 60 to 90% solvent B for 5 min. The mass spectrometer was set as described above.
The acquired data were postprocessed to generate peak lists (.pkl) using PeptideAuto, which is a part of ProteinLynx from MassLynx 4.0. The following parameters were used: quality assurance threshold, 10; smooth window, 3 (two times in Savitzky Golay mode); centroid at minimum peak width at half-height, 4; centroid top, 80%. The peak lists are appended as a single file.
Each sample was submitted to consecutive searches against the Swiss-Prot TrEMBL database and the specific A. thaliana database using Mascot 2.0. Mascot search parameters used with MS/MS data were: database ϭ A. thaliana (nuclear, mitochondrial, and chloroplastic genome), enzyme ϭ trypsin/proline, one missed cleavage allowed, peptide tolerance ϭ 0.25 Da, MS/MS tolerance ϭ 0.25 Da, peptide charge ϭ 2ϩ/3ϩ, and variable modifications. For the in-gel digestion procedure, variable modifications were N-terminal acetylation/oxidized methionine under sulfone and sulfoxide form/FMA ϩ 1/FMA Ϫ 1/cysteic acid. For the in-solution digestion procedure, variable modifications were acetyl-Nter/oxidized methionine under sulfone form/ FMA ϩ 1/FMA Ϫ 1. Proteins identified by at least two peptides with a Mascot MOWSE score higher than 50 were automatically validated. When this criterion was not met, the fragmentation spectrum from each peptide was manually interpreted using the conventional fragmentation rules. In particular, we looked for a succession of at least five y-and/or b-ions, specific immonium ions, specific fragment ions (proline and glycine), and signatures of any modifications carried by the peptides. In cases where the protein was mainly identified by a single peptide match, the MS/MS ions were manually examined for notable sequence tag and independently verified using the PEAKS studio program (Bioinformatics Solutions Inc.). The algorithm can efficiently choose the best amino acid sequence, from all possible amino acid combinations, to interpret the MS/MS spectrum according to the same chemical modifications defined in Mascot. An additional search was carried out for the identification of the possible N-terminal peptide of proteins using semitrypsin and N-terminal acetylation as Mascot parameters (see Supplemental Table III).

Expression of Protein Fusion in Plants and
Protoplasts-GFP-protein fusions were generated using Gateway technology (Invitrogen). cDNA encoding the protein to be tested for subcellular localization (At3g19820, At1g19450, At3g63520, At5g58070, At3g16240, and At1g69840) were provided in entry clones (pENTR/SD/D-TOPO U15125, U16253, U16861, U17005, U17252, and U25581, respectively) by the Arabidopsis Biological Resource Center stock center. Cloning reactions were performed following the manufacturer's instructions using the destination vector pK7WGF2 (55), which was kindly provided by the Flanders Interuniversity Institute for Biotechnology. The resulting expression vectors were introduced in Agrobacterium tumefaciens C58 strain. For transient expression in tobacco leaves, a 24-h-old culture of A. tumefaciens C58 was diluted five times in an induction medium as described previously (56). After 6 h, the bacterial suspension was adjusted to an A 600 of 0.5 and leached into leaves of 9-week-old tobacco (Nicotiana benthamiana). In parallel and as a positive control for tonoplastic localization, tobacco plants were also transformed with the vector pNB96 containing the Nramp3-GFP construction (kindly provided by S. Thomine). For transient protoplast transformation, the SmaI-DraI cassettes from expression vectors, containing the cauliflower mosaic virus 35S, the GFP-protein fusion, and the nopaline synthase terminator were subcloned into pCR-BLUNT II-TOPO (Invitrogen) for more efficient transformation. Protoplasts were prepared as described for vacuole isolation except that the rinsing medium was replaced by W5 medium (154 mM NaCl, 125 mM CaCl 2 , 5 mM KCl, 5 mM glucose, 1.5 mM MES-KOH, pH 5.6). In a microcentrifuge tube, 100 l of protoplasts (corresponding to 10 5 protoplasts) were added to 50 g of plasmid DNA in 10 l of water. PEG solution (110 l) (0.4 M mannitol, 100 mM CaCl 2 , 40% (w/v) polyethylene glycol 4000) was added to the mixture. After 25 min of incubation at room temperature, protoplasts were diluted in 440 l of W5 medium, centrifuged for 1 min at 100 ϫ g, resuspended in 1 ml of W5 buffer, and incubated at 22°C for 4 days before observation. Subcellular localization was assessed by scanning confocal microscopy (TCS SP2, Leica), GFP was excited at a wavelength of 488 nm, and fluorescence measurements were collected between 500 and 550 nm.

Isolation and Evaluation of Purity of the Vacuole Preparations Isolated from Cultured Arabidopsis Cells-Vacuoles
from A. thaliana suspension cultures grown in the light were purified after protoplast preparation as described previously (43) with some modifications. An average of 120 Ϯ 35 g of vacuole proteins could be obtained from 18 g of 5-day-old Arabidopsis cells. The specific activity of the vacuolar marker ␣-mannosidase was 4.6 Ϯ 0.5 mol⅐h Ϫ1 ⅐mg Ϫ1 of proteins in purified vacuoles. This represented an enrichment factor of 42-fold when compared with the specific activity of the same marker in crude protoplast extract (0.11 Ϯ 0.02 mol⅐h Ϫ1 ⅐ mg Ϫ1 of proteins) and an average yield of 2.1% based on this activity. To estimate cross-contamination, vacuole preparations were visually checked using epifluorescence microscopy (set for 4Ј,6-diamidino-2-phenylindole coloration). In these conditions very few red-fluorescing chloroplasts or protoplasts were seen among the blue (due to the presence of phenolic compounds in the lumen) vacuoles. This low contamination by chloroplasts and protoplasts was confirmed by the absence of detectable chlorophyll when analyzed after acetone extraction (not shown).
Protoplast and vacuolar proteins were separated by SDS-PAGE (Fig. 1A), and Western blots were performed to analyze cross-contaminations by other subcellular compartments including plastids but also mitochondria, plasmalemma, and endoplasmic reticulum (Fig. 1B). Western blotting experiments using antibodies raised against the tobacco ␣-TIP (50) and the cauliflower ␥-TIP (51) were also performed and showed that these TIP proteins were both highly enriched in purified vacuoles compared with the protoplast protein extract in which they were hardly visible under our experimental conditions (Fig. 1B). In contrast, the plasmalemma, the plastid, and the mitochondrial protein markers were not detectable in the vacuolar protein extract (Fig. 1B). In the case of the ER marker, a thin band of 70 kDa (corresponding to the most abundant HDEL protein) (49) was detectable in the vacuole fraction, reflecting minor ER contaminations. Taken together these results showed that our vacuole preparations were of high quality and that contaminations by other organelle membranes were very low, allowing further proteomics investigation.
Identification of the Major Vacuolar Proteins Isolated from Arabidopsis Suspension Cells: Use of an In-organic-aqueous Digestion Method to Identify Membrane Proteins-We first undertook the identification of the most abundant vacuolar proteins. The most intensely stained bands of the total vacuolar proteins separated by SDS-PAGE presented in Fig. 1A were trypsin-digested, and peptide mixtures were submitted to nano-LC-ESI-MS/MS analysis (see also Fig. 2A). This proteomics analysis led essentially to the identification of soluble proteins (annotated in Fig. 1A). Among them, we identified the tripeptidyl peptidase II (At4g20850, a subtilase family protein); the A and B subunits of the vacuolar type H ϩ -ATPase (VHA-A, At1g78900; VHA-B1, At1g76030); a protein similar to the bacterial TolB protein, a member of the Tol system (At1g21680) (57); a putative leucine aminopeptidase (At2g24200); a putative pectin methylesterase (At1g11580); the glycosyl hydrolase family 17 (At4g16260), similar to the glucan endo-1,3-␤glucosidase vacuolar isoform from Hevea brasiliensis; three band 7 family proteins (At5g62740, At5g51570, and At1g69840), bearing strong similarities to hypersensitive-induced response protein from Zea mays; a basic endochitinase (At3g12500) involved in the ethylene/jasmonic acid-mediated signaling pathway during systemic acquired resistance; a glutathione S-transferase (At2g30870 or AtGSTF10) induced by dehydration (58); and a SOUL heme-binding family protein (At1g17100). Interestingly the latter protein, which has never been described in plants, is similar to the mammalian and chicken members of the SOUL/p22HBP family. p22HBP, a cytosolic protein, has been suggested to be involved in heme utilization for hemoprotein synthesis (59).
To identify the major membrane proteins present in the tonoplast, vacuoles were first frozen/thawed as described under "Experimental Procedures." The membrane fraction was then separated from the soluble material by ultracentrifugation and salt-washed to eliminate membrane-associated soluble proteins. At this step, to prevent the use of detergent to solubilize hydrophobic proteins, we used an alternative method that consists of an in-solution trypsin digestion in the presence of methanol. Membrane proteins were first denatured at a high temperature (90°C) followed by rapid cooling by the addition of cold (Ϫ20°C) MeOH to 60% (v/v) final concentration, and tryptic digestion was then carried out in this solution. The resulting peptide mixture was analyzed by nano-LC-ESI-MS/MS (Fig. 2B), and 122 proteins were identified from a very low amount of proteins (2 ϫ 0.15 g). Peptides identified with a Mascot score sufficient for protein identification are listed in Supplemental Table I. This procedure allowed the identification of the most hydrophobic tonoplast proteins. Thirty of the most abundant membrane proteins identified in the tonoplast are presented in Table I (the  others are presented in Supplemental Table I). This indicative classification is based on a simplified exponentially modified protein abundance index (emPAI) proposed by Ishihama et al. (60). Relative protein abundance was estimated by normalizing the number of peptides per protein by the theoretical number of peptides per protein. A caveat applies to this estimation as extremely abundant proteins may affect the efficiency of protein identification because of ionization suppression and detector saturation. Highly hydrophobic proteins may also be under-represented because they generate a limited number of peptides due to their sequence characteristics. Despite these limitations, approximate comparison of the relative abundance of tonoplastic membrane proteins is possible because of the common characteristics they share. It is noteworthy that a total of 70% of the proteins were identified by two peptide hits or more. The majority of the proteins identified are transporters, and the most representative proteins, as expected, are ␦-TIP, subunits of the vacuolar H ϩpumping ATPase, and the H ϩ -pumping pyrophosphatase (H ϩ -PPase AtVP1). Among the 30 most abundant proteins, there are also three ABC transporters (MRP1, MRP4, and MRP10), five multidrug and toxin extrusion (MATE) efflux family proteins, a sodium/calcium exchanger family protein and a calcium-transporting ATPase (ACAc), a peptide transporter (PTR2-B), a copper transporter family protein (COPT5), and a cationic amino acid transporter (CAT2) ( Table I).
This preliminary proteomics work not only confirmed the quality of our vacuole preparations but also revealed the identification of several proteins that had never been identified before by classical approaches (40 -42). On the other hand, it also showed the lack of proteins, such as the thioglucosidases (At1g54010 and At1g54000), that are known to be the main components of other vacuolar systems (42). Taken together, these results indicated the need to extend the knowledge of the vacuolar proteome of cultured Arabidopsis cells.
Toward an Extended Vacuolar Proteome Study-To increase the number of proteins identified in the tonoplast, prefractionation of the membrane and soluble proteins was performed by SDS-PAGE (Fig. 2, C and D). A short migration (2.5 cm) was carried out to obtain efficient separation and to avoid a diffusion effect across the gel and keep the total number of gel bands to be analyzed reasonable. The gel was cut into 15 bands. A peroxide treatment was performed, and proteins were then in-gel digested with trypsin and identified by nano-LC-ESI-MS/MS analysis. Extended coverage of the FIG. 2. Strategy used for the identification of vacuolar proteins purified from Arabidopsis cell culture. Vacuoles were purified from A. thaliana suspension cultures. After vacuole purification, proteins were either separated by 12% SDS-PAGE and stained with Coomassie Blue (A) or subjected to centrifugation to obtain two different enriched fractions: membranes (B and C) and vacuolar sap (D). Major bands from the whole vacuole SDS-PAGE migration were cut out and subjected to in-gel trypsin digestion (A; Fig. 1A). Proteins from the membrane fraction were either subjected to in-solution trypsin digestion (B) (see text and "Experimental Procedures") or separated by a short SDS-PAGE migration and in-gel digested (C). Proteins from the soluble fraction were in-gel digested following a short SDS-PAGE migration (D). Peptides from A to D were separated by liquid chromatography prior to MS/MS analysis. protein sequences was obtained using both (i) dynamic exclusion during the MS/MS process and (ii) a second nano-LC-ESI-MS/MS analysis using an exclusion list to limit refragmentation of peptides fragmented during the first run. The peroxide treatment applied before trypsin digestion is a mild oxidation procedure that transforms the cysteinyl residues into cysteic acids (ϩ48 Da). The advantage of this method lies not only in the increased peptide coverage (not shown) but also in the speed of sample preparation compared with classical reduction/alkylation approaches. This treatment also converts all the methionines to the maximal oxidation state (sulfone, ϩ32 Da), providing better quality MS/MS spectra, and prevents the apparition of the intermediate fragmentation pattern of Ϫ64 Da that complicates the mass attributions obtained with the Mascot software.
Using this procedure, 387 proteins were identified from the The pI and molecular mass values shown are theoretical. Score, the Mascot score of in-solution digestion protocol. Cov., the protein sequence coverage (%). No. peptide represents the number of peptides assigned. N obsbl is the number of observable peptides per protein in our experimental conditions. Rel ab is an estimation of relative protein abundance obtained using the number of peptides detected per protein (No. peptide) normalized by the theoretical number of detectable peptides (N obsbl ). This classification is based on a simplified exponentially modified protein abundance index (emPAI) proposed by Ishihama et al. (60). TMD, the number of transmembrane domains predicted by ARAMEMNON (crombec.botanik.uni-koeln.de/) (52). Function, the predicted protein function as assigned by the Munich Information Center for Protein Sequences (MIPS) Functional Catalogue (FunCat) (mips.gsf.de/proj/funcatDB/search_main_frame.html). Myr, N-myristoylation; MB, membrane protein.

Name
Gene  Table II). Using both the in-gel and in-solution digestion procedures, a total of 416 non-redundant proteins were identified from the tonoplast. Among them, 50 had been demonstrated previously to be localized to the vacuole, 195 were integral membrane proteins based on the presence of one or more predicted transmembrane domains (TMHMM2.0) (Supplemental Tables I and II), five had transmembrane ␤ barrel structures such as porins, 29 did not have any transmembrane domain but were known to be part of membrane complexes such as H ϩ -ATPase, and 31 were predicted or known to have covalent lipid modifications leading to their insertion into the membrane (myristoylation and prenylation sites were predicted using the Plant-Specific Myristoylation Predictor (93) and Psort II (54). Among the latter, we found several band 7 family proteins, a putative glycosylphosphatidylinositol-anchored protein, a multicopper oxidase, and Ras-related GTP-binding proteins. The 351 soluble proteins identified from the vacuolar sap will not be discussed here. The whole proteomics work presented here represents the identification of more than 650 non-redundant proteins; this is the most complete study done to date. The proteins found both in our study and in previously published studies (40 -42) are listed in Tables I and II and Supplemental Tables I and II.
The most complete study previously published identified 402 proteins (42). In general, our respective fractions are quite different, and the overlap rate of the two studies is around 26%. This weak overlap, in terms of identification, is almost certainly related to differences in starting material. Carter et al. (42) used Arabidopsis leaves, whereas we used Arabidopsis suspension cultures. It is noteworthy that in our study around 70% of the 650 proteins were identified by two or more peptide hits. Contrary to the analysis of the total vacuolar fraction by Carter et al. (42), our identification did not reveal the presence of a 42-kDa protein representing the bulk of vacuolar protein content (identified as a myrosinase-associated protein, At3g14210, and the corresponding myrosinase gene products, At5g25980 and At5g26000). This difference is probably due to the presence of myrosin cells in the Arabidopsis leaves that accumulate the thioglucoside glucohydrolase (61). Our membrane protein data were much more complete because the transporters identified in the previous study were well overlapped, and our analysis identified a larger number of transporters with better coverage rates (and associated Mascot scores). To be complete, our membrane protein tonoplastic set must be compared with the results of Shimaoka et al. (40) where 163 proteins including well char-

TABLE II Compilation of the H ϩ -ATPase subunits identified in the vacuole proteome
The table gives the protein acronym (Subunit), gene designation (Gene), number of amino acid (AA) residues in the deduced translation product and the molecular mass, the predicted pI, and the number of transmembrane domains (TMD) predicted by ARAMEMNON (52). Score gel and cov. (%) represents the Mascot score and protein sequence coverage in parentheses using the in-gel digestion protocol. No. gel peptide is the number of peptides assigned using the in-gel digestion protocol, and (Cys) represents the number of peptides found with cysteic acid. Score solution is the Mascot score using the in-solution digestion protocol. Score Carter et al. is the Mascot score found by Carter et al. (42). Cov. Shimaoka et al. is the sequence coverage found by Shimaoka et al. (40). PTM represents post-translational modifications found using independent Mascot searches with semitrypsin and N-terminal acetylation (AcNt) as additional parameters.

Subunit
Gene AA/molecular mass (kDa) pI TMD Score gel and cov. acterized tonoplast proteins such as V-type H ϩ -ATPases and V-type H ϩ -PPases were identified. Both identifications were obtained from suspension-cultured Arabidopsis cells. However, the data of Shimaoka et al. (40) led to the identification of 163 proteins of which 42 were membrane proteins (annotated with one or more transmembrane domains). Of these 42, 39 were predicted to have more than two transmembrane domains, and 17 are possible transporters. Shimaoka et al. (40) and Szponarski et al. (41) identified a large number of mostly soluble proteins within their vacuolar fractions. This may be because of a lack of thorough washing of the vacuolar membranes or due to different LC or MS analysis protocols. Fig. 3 shows the cross-correlation of the different proteome analyses of the vacuolar membrane system. Proteins were categorized into 13 major groups: transporters, stress response, signal transduction, metabolism, cellular transport, protein synthesis and degradation, cytoskeleton, glycosyl hydrolase, RNA degradation, unclassified, and contaminants. It is of note that the main contamination of our preparations seems to be cytosolic proteins, in particular with ribosomal proteins. Few possible contaminating mitochondrial or chloroplast proteins were detected. Among the classes of proteins identified, transporters are of particular interest because few vacuolar transporters have been identified and fully characterized so far.
Pumps and Transporters-Import and export of metabolites and ions require their corresponding transport systems across the tonoplast, and many transporters and channels in this membrane remain to be identified. In the present work, we identified 91 proteins with demonstrated or predicted transporter activities. In addition, the two major components of the tonoplast, the vacuolar H ϩ -ATPase and the H ϩ -PPase were identified. The vacuolar H ϩ -ATPase is a large oligomeric protein complex (Ͼ700 kDa) composed of 12 subunits that form a hydrophilic V 1 subdomain (VHA-A to VHA-H) localized in the cytoplasm and an integral membrane V 0 subdomain (VHA-a, VHA-c, VHA-d, and VHA-e) (19,62). In Arabidopsis, five subunits are encoded by alternative splicing of a single gene (VHA-A, VHA-C, VHA-D, VHA-F, and VHA-H); the others are encoded by multiple genes. Among the 28 possible subunits, 19 proteins were clearly identified. Table II summarizes the H ϩ -ATPase subunits identified in the present work alongside those identified in other published works (40,42). The high quality of the vacuole preparation and the analytical methods we used have allowed coverage of a high percentage of the protein sequences (12-75%). Corroborating this, very high Mascot score values, reaching for example 2038 in the case of the VHA-A subunit identification (versus 222 in the published data of Carter et al. (42)), were obtained. The identification of the other vacuolar pump, the H ϩ -PPase (AVP3) (63), was also achieved with a high Mascot score value (1949 and 1229 obtained from in-solution and in-gel digestion, respectively, versus 88 in previously published data (Carter et al. (42)).
The main representative superfamily of transporters that we identified was the ABC transporter family. Substrates assigned to members of this large family of transporters include compounds as diverse as peptides, sugars, lipids, heavy metal chelates, polysaccharides, alkaloids, steroids, inorganic acids, and glutathione-conjugated compounds (64). The 129 ABC proteins encoded by the Arabidopsis genome are classified into 12 subfamilies based on their size, orientation, domain organization, and resemblance to ABC proteins from other organisms (65,66). In the present proteomics work, 14 different members belonging to five different subfamilies: multidrug resistance protein (MDR), multidrug resistance-associated proteins (MRP), pleiotropic drug resistance (PDR), nonintrinsic ABC protein (NAP), and transporter associated with antigen processing (TAP) were identified (Table III). Ten were clearly identified as MRP: MRP1-8, MRP10, and MRP14. The degree of identities shared between these MRP subfamily members ranges from 33 to 87% (67) making their unambiguous identification difficult unless a high protein coverage is obtained (41,42). Only three of them, AtMRP1, AtMRP2, and AtMRP4, have been identified and localized to the tonoplast so far by classical methods (38,39,69). Our data confirmed their presence in the tonoplast membrane, and six others were identified. Table III shows once again the high Mascot score and coverage values obtained in our analyses even for these difficult-to-identify proteins. As an example, MRP10 (At3g62700) was identified with a Mascot score of 2613 and 32% sequence coverage. Based on the protein coverage found in the in-solution digestion protocol, MRP10 seems to be the most abundant ABC subclass transporter in cultured Arabidopsis cells. Analysis of AtMRP10 (At3g62700) using the ARAMEMNON database (52) predicts the presence of a strong chloroplast-targeting signal and a weaker mitochondrion signal peptide. These predictions show that the in silico subcellular localization prediction algorithms are probably not robust enough. Surprisingly we also found MRP4 in the tonoplast, whereas Klein et al. (68) using confocal microscopy analysis of onion epidermal cells transiently expressing At-MRP4-EGFP showed fluorescence at the periphery of cells, suggesting that AtMRP4 protein was located at the plasma membrane. In support of our data, previous proteomics studies confirm the presence of MRP4 in the purified vacuolar fractions (40,42), but both localizations could be possible. Moreover in a recent study, Dunkley et al. (69) have unambiguously assigned 18 transporters to specific organelles by LOPIT (localization of organelle proteins by isotope tagging) technology. The authors have shown that the vacuolar membrane class is dominated by proteins involved in membrane transport, including eight ABC transporters (MRP1-6, MRP10, and TAP2).

TABLE III List of the ABC transporters identified on the tonoplast of vacuoles isolated from Arabidopsis cell cultures
The table gives the protein acronym (Name), protein identification number (Protein ID), gene designation in TAIR (Gene), and number of amino acid residues in the deduced translation product (AA). TMD, putative transmembrane domains predicted by ARAMEMNON. Topology, number and orientation of transmembrane domain-nucleotide binding fold (TMD and NBF). Score gel and cov., Mascot score and protein sequence coverage in parentheses (%) obtained using the in-gel digestion protocol. No. gel peptides, the number of peptides assigned using the in-gel digestion protocol. Score Sol and cov., Mascot score and protein sequence coverage in parentheses (%) obtained with the in-solution digestion protocol. No. sol peptides, number of peptides assigned with the in-solution digestion protocol. Score Carter et al., Mascot score found by Carter et al. (42)
Previous studies argued that plant cells have the ability to generate and maintain separate vacuole organelles with each being marked by a different TIP subtype (5,6). For example, the fully differentiated cell types PSVs are marked by ␣-TIP plus ␦-TIP, and lytic vacuoles are marked by ␥-TIP. The presence of both markers in our proteomics work could be explained by the presence of different kinds of vacuoles in our samples or the presence of vacuoles that come from fusion of storage and lytic vacuoles combining properties of both vacuoles as mentioned previously (4).
Other transporters identified in the tonoplast proteome are listed in Table IV. Among them, transporters that mediate the efflux of a broad range of compounds have been identified: for example, 18 members of detoxifying efflux transporters, which include eight major facilitator superfamily (MFS) members, six MATE family members, three examples of drug/ metabolite transporter (DMT) superfamily proteins, and one member of the resistance/nodulation/division superfamily. Members of the MFS family are secondary transporters that represent the largest group of ion-coupled transporters involved in the symport, antiport, or uniport of various substrates such as sugars, Krebs cycle intermediates, phosphate esters, oligosaccharides, and antibiotics (70). Members of the multidrug/oligosaccharidyl-lipid/polysaccharide (MOP) family were also identified here. These are interesting because Arabidopsis expresses many members of this family not found in the animal kingdom. The MATE family, which is a MOP subtype, shows homologies with bacterial efflux transporters. The Arabidopsis genome codes for at least 54 members of this family, and only a few members of the MATE family are characterized functionally. Their contribution to drug resistance has been shown only for a few isolated cases (71). Hydropathy analysis suggests that proteins of the MATE family have a common topology consisting of 12 transmembrane domains. Six different MATE efflux transporters were characterized in the tonoplast fraction suggesting the potential contribution of the vacuole system to drug resistance (Table IV).
Five peptide transporters were identified, three members of the proton-dependent oligopeptide transporter (POT) family and two oligopeptide transporters (OPTs). Interestingly one of the best characterized plant OPT members is the maize "Yellow Stripe 1" (YS1) transporter, a Fe 3ϩ -phytosiderophore:H ϩ symporter involved in Fe 3ϩ uptake (72) but also in the uptake of various other metal cations complexed with either phytosiderophores or nicotianamine (73). The two OPT family members that we have identified (AtYSL4 and AtYSL6) in the tonoplast are classed as Yellow Stripe-like (YSL) transporters (72) and present 69 and 70% similarity with YS1. We also confirmed the presence of other metal transporters such as Nramp3 (36) and AtMTP1 (AtZAT1) (32,33). We did not identify the Na ϩ /H ϩ antiporter (AtNHX1) (74) but found an isoform, AtNHX4, which is also a member of the monovalent cation: proton antiporter-1 (CPA1) family. A putative iron (Fe 2ϩ ) transporter (AtIRT3), a member of the zinc-iron permease family, was also identified (75). The vacuolar calmodulin-regulated Ca 2ϩ -ATPase 4 (ACA4) (30) was also found in the tonoplast. Three K ϩ uptake permeases, members of one of the five potassium transporter families, were identified in the present work (AtKuP7, AtKuP5, and AtKuP4) (75). The latter, AtKuP4 (or AtTRH1), has recently been shown to be required for auxin transport in Arabidopsis roots (76).
The tonoplast contains at least seven amino acid transporters from the 50 distinct amino acid transporter genes encoded by the Arabidopsis genome. We identified four of the nine cationic amino acid transporters (AtCAT2, AtCAT4, At-CAT8, and AtCAT9) present in Arabidopsis. Interestingly a recent molecular and functional characterization of this family shows that AtCAT2 is localized to the tonoplast, whereas AtCAT5 (not found in this proteomics work) is present in the plasma membrane (77). AtCAT2 and AtCAT4 were identified in the tonoplast in the study by Carter et al. (42). We also found five amino acid/auxin permease (AAAP) family members in the tonoplast (Table IV).
Moreover a few other unexpected proteins such as "expressed protein similar to TolB protein precursor" (At1g21680) or "Niemann-Pick C1 similar protein" (At4g38350) were found in the tonoplast membrane. TolB, a periplasmic protein found in most Gram-negative proteomes, is one of the Tol proteins of Escherichia coli and is involved in the translocation of group A colicins. TolB also forms a complex with Pal, an outer membrane peptidoglycan-associated lipoprotein anchored to the outer membrane by its N-terminal lipid moiety (57). The exact role of the Tol system remains to be determined. Its presence in the tonoplast is probably linked to membrane biogenesis as supposed for the Gram-negative organisms. Niemann-Pick C1 protein is a large multitransmembrane glycoprotein that was shown to reside primarily in mammalian late endosomes. Its cytoplasmic tail contains a dileucine endosome-targeting motif, and it transiently associates with lysosomes and the trans-Golgi network. The function of the NPC1 protein is unclear. However, a number of observations suggest that NPC1 may be related to a family of prokaryotic efflux pumps, and thus it may also act as a molecular pump (78) or cholesterol transporter (79) or play a role in docking/ fusion events (80).
Other Proteins-From our data, we can pinpoint another important uncharacterized class of proteins: the band 7 protein family. Twenty members of this family (also known as stomatin prohibitin flotillin Hbc (SPFH) domain proteins) are predicted from the Arabidopsis genome; 10 of them were identified in the vacuole fraction (Table V). Co-fractionation with the tonoplastic membrane in the presence of sodium carbonate at pH 11.5 followed by salt wash strongly suggests that the band 7 proteins are true integral membrane proteins. Interestingly this observation is in agreement with the fact that the band 7 proteins are putatively myristoylated or have one TMD. In most cases, this modification is essential for protein function to mediate membrane association or protein-protein interaction. The SPFH protein domain is characteristic of the prohibitins and the stomatins, which are putatively involved in cell cycle and ion channel control. Multiple stomatin orthologues from bacteria, plants, and animals have been identified. In Caenorhabditis elegans, multifunctional extracellular protein-2 (MEC-2), a stomatin-like protein, is involved in mechanosensation (81). Another example is the protein UNC-1, a close C. elegans homologue of the mammalian protein stomatin, which may be involved in anesthetic sensitivity and could represent a molecular target for volatile anesthetics (82). Cumulative evidence suggests that band 7 stomatins may modulate membrane functions especially those of detergent-resistant microdomains or lipid rafts (for a reviewed, see Ref. 83). The presence of this family in the vacuolar membrane may play a crucial role in the regulation of The table gives the protein acronym (Protein), gene designation (Gene); protein identification number (Accession no.), deduced molecular mass (kDa), theoretical pI, the number of transmembrane domains (TMD) or post-translational modification (PTM). Myristoylation sites were predicted using the Plant-Specific Myristoylation Predictor (93) and Psort II (54). Score gel represents the Mascot score of in-gel digestion protocol. Cov, protein sequence coverage. No. peptides, number of peptides assigned using the in-gel digestion protocol. Score Carter  either the biogenesis of the tonoplast membrane or the regulation of transporters and metabolite flux across the tonoplast. Consequently we confirmed the subcellular localization of one member of this family using a GFP fusion protein approach (see below). In eukaryotes, peptides derived from proteasomal degradation of intracellular proteins have been shown to translocate from the cytosol into the ER via TAP (84). TAP is a macromolecular peptide-loading machinery composed of TAP1, TAP2, tapasin, and several auxiliary factors (e.g. calreticulin and ERp57). According to the proteins characterized in our tonoplastic fraction, similar machinery for protein degradation and peptide import into the vacuole may be present in Arabidopsis. Indeed we have identified a TAP2 homologue, a calreticulin, ERp57, and other members of the TAP machinery, and this could explain the number of proteasome subunits observed in our sample (Supplemental Table II).
Confirmation of the Vacuolar Localization of Selected Proteins-The quality of the purified vacuolar fraction, as assessed by enzymatic and immunological assays, was the first criterion for establishing the likelihood of vacuolar localization of the proteins identified in this study. Indeed mass spectrometry analyses did not reveal the presence of major proteins from other organelles. Moreover several proteins identified (H ϩ -ATPase, TIPs, ABC transporters MRP1 and MRP2, etc.) were already known as vacuolar proteins. To further characterize the vacuolar system studied here, the subcellular localization of several proteins was investigated by transient expression in transfected tobacco plants and in Arabidopsis protoplasts of GFP fusion proteins. We chose the following proteins from Tables I and V and Supplemental Table II: the cell elongation protein Dwarf1/DIMINUTO (Dwarf1, At3g19820), a band 7 family protein (At1g69840), an integral membrane protein (At1g19450), a putative lipocalin (At5g58070), and the 9-cis-epoxycarotenoid dioxygenase (CCD1, At3g63520). Confocal microscopy imaging was carried out on transfected tobacco plants or on protoplasts isolated from Arabidopsis cell suspensions (Fig. 4). As a positive control in accordance with previously published data (36), we expressed GFP-Nramp3 (At2g23150) and confirmed its vacuolar cellular location in tobacco cell leaves (Fig. 4A). The GFP-TIP2.1 (␦-TIP, At3g16240) construct was also used as vacuolar protein control and is presented in Fig. 4B. Inspection of the GFP-Dwarf1-transformed plant using fluorescence microscopy confirms its presence in the tonoplast compartment (Fig. 4D). Analysis of the At3g19820 Dwarf1 amino acid sequence predicts a transmembrane domain at the N terminus of the protein. It has been proposed that Dwarf1 is a peripheral or an integral membrane protein (85). The Arabidopsis DWARF1 gene encodes a protein involved in steroid synthesis, and a vacuolar subcellular localization is in good agreement with the proposed role for the Dwarf1 protein as a biosynthetic enzyme; because sterols as well as steroid hormones are relatively hydrophobic moieties, one would expect that synthesis may occur in a membrane environment. The At1g69840 gene product is a band 7 family protein presenting similarity to hypersensitive-induced response protein with relatively high homology to regions of stomatin and prohibitin (see above). Its presence in the vacuolar membrane was also confirmed (Fig. 4E).
According to ARAMEMNON database prediction, the At1g19450 gene product is an integral membrane protein with 12 transmembrane domains. Among the biological processes to which this protein may contribute are phosphate transport, oligopeptide transport, cell adhesion, carbohydrate transport, phospholipid biosynthesis, and tissue regeneration. The At1g19450 gene product shows 71% similarity to a putative sugar transporter from sugar beet that was cloned and localized to the vacuole in transgenic yeast and tobacco (86). This putative transporter is a member of a subgroup of a large gene family, currently termed the MFS (70). Fig. 4F confirms the vacuolar localization of this transporter in tobacco cells.
The At5g58070 gene product is a putative lipocalin family member or outer membrane lipoprotein-like. Lipocalins are widely distributed among vertebrates, and a few have been isolated from invertebrates and plants. Predominantly lipocalins are small secreted proteins. Some members of the family exhibit high affinity and selectivity for hydrophobic molecules such as cholesterol, pheromones, or fatty acids. Others have been shown to bind to specific cell surface receptors and to form macromolecular complexes. The lipocalins have been classified as transport proteins. The non-solubilization of the protein during our membrane washing procedure indicated that this protein could be an anchor-linked protein, and the fluorescence microscopic analysis presented in Fig. 4G demonstrated the presence of lipocalin protein on the tonoplast membrane. As the localization of these latter two proteins (i.e. lipocalin and the sugar transporter) presented some heterogeneity in the leaf model, we performed a transient expression analysis in Arabidopsis protoplasts (Fig. 4, I-K). It is interesting to note that lipocalin seems not to be exclusively localized on the tonoplast (Fig. 4I), whereas its presence on the vacuolar membrane is confirmed by observing a fluorescent vacuole being released from a cell (Fig. 4J). Inversely the putative sugar transporter is clearly exclusively targeted to the tonoplast (Fig. 4K).
The last protein that we chose to study as a GFP fusion construct was the carotenoid cleavage dioxygenase (CCD1, At3g63520). Apocarotenoids, derived from the oxidative cleavage of carotenoids, are involved in important metabolic and hormonal functions in diverse organisms (87,88). Since the discovery of VP14 in maize (87), the "prototypic 9-cisepoxycarotenoid dioxygenase" (NCED), several carotenoid cleavage enzymes have been characterized and grouped in different classes according to substrate specificities (88). Class 1 is represented by the NCEDs that are involved in abscisic acid biosynthesis (89). CCD1 is a member of the class 2 family, catalyzing a dioxygenase reaction leading to the synthesis of ␤-ionone and C 14 dialdehyde (90). So far, nine carotenoid cleavage dioxygenase genes have been identified in the complete Arabidopsis genome, and among them, only five NCEDs (2, 3, 5, 6, and 9) have been shown to be targeted to plastids (91). However, no information was available con-cerning the localization of CCD1; as shown in Fig. 4H, our analyses clearly identified CCD1 in the vacuole membrane. As expected and taking into account the overall results presented previously, the vacuolar localization of all the GFP- tagged proteins was confirmed.
Conclusion-The analysis of a proteome at the level of subcellular structures represents an analytical strategy that combines traditional biochemical methods of fractionation and tools for protein identification. We have shown the strategy developed in this work to be very efficient to get a broader view of the vacuole membrane proteome. We identified several new membrane proteins such as channels and transporters that mediate the translocation of molecules and ions across tonoplast membranes. We consider that most of the proteins identified in this study are genuine constituents of the tonoplast. This is based on the identification of key tonoplastic components, the verification of vacuolar localization for selected proteins using GFP fusion proteins, and the low contamination of our preparations by other organelles. Through our vacuolar proteomics strategy, it seems clear that the tonoplast membrane is a complex structure that receives membrane and protein contributions from a variety of subcellular sources and pathways. The high degree of protein diversity of the tonoplast membrane is indicative of a highly complex organelle. Further validation of the results presented here by relevant functional studies should provide a better explanation for the biogenesis and maintenance of these unique organelles.