Arabidopsis Plasma Membrane Proteomics Identifies Components of Transport, Signal Transduction and Membrane Trafficking

Abstract

In order to identify integral proteins and peripheral proteins associated with the plasma membrane, highly purified Arabidopsis plasma membranes from green tissue (leaves and petioles) were analyzed by mass spectrometry. Plasma membranes were isolated by aqueous two-phase partitioning, which yields plasma membrane vesicles with a cytoplasmic-side-in orientation and with a purity of 95%. These vesicles were turned inside-out by treatment with Brij 58 to remove soluble contaminating proteins enclosed in the vesicles and to remove loosely bound contaminating proteins. In total, 238 putative plasma membrane proteins were identified, of which 114 are predicted to have transmembrane domains or to be glycosyl phosphatidylinositol anchored. About two-thirds of the identified integral proteins have not previously been shown to be plasma membrane proteins. Of the 238 identified proteins, 76% could be classified according to function. Major classes are proteins involved in transport (17%), signal transduction (16%), membrane trafficking (9%) and stress responses (9%). Almost a quarter of the proteins identified in the present study are functionally unclassified and more than half of these are predicted to be integral.

(Received September 1, 2004; Accepted September 24, 2004)

Introduction

The plasma membrane constitutes the interface between the cell and its surroundings. This position implies a wide range of important functions concerning, e.g. transport of compounds into and out of the cell, communication with the cell exterior and defence towards invading pathogens—functions that are fulfilled by, e.g. transport proteins, proteins involved in membrane trafficking and receptor kinases.

The Institute for Genomic Research (TIGR) now has more than 27,000 Arabidopsis proteins annotated (Wortman et al. 2003) and 25% of these are predicted to be integral membrane proteins (Schwacke et al. 2003). The plasma membrane is probably the most diverse membrane of the cell, with a protein composition that varies with cell type, developmental stage and environment; and it is likely to harbor thousands of proteins. For instance, receptor-like protein kinases (RLKs) alone, most of which probably reside in the plasma membrane, are represented by more than 600 genes in the Arabidopsis genome (reviewed by Shiu and Bleecker 2001). After the Arabidopsis genome was completed in 2000 (Arabidopsis Genome Initiative 2000), efforts at characterizing the Arabidopsis chloroplast (Peltier et al. 2002, Schubert et al. 2002, Ferro et al. 2003, Froehlich et al. 2003), mitochondrial (Kruft et al. 2001, Millar et al. 2001, Millar and Heazlewood 2003, Heazlewood et al. 2004), peroxisome (Fukao et al. 2002) and tonoplast proteome (Shimaoka et al. 2004), as well as the Arabidopsis cell wall proteome (Chivasa et al. 2002), have been made. Also, the Arabidopsis plasma membrane has been studied using proteomic approaches (Prime et al. 2000, Santoni et al. 2000, Kawamura and Uemura 2003, Marmagne et al. 2004).

In a recent study by Marmagne et al. (2004), about 100 putative plasma membrane proteins were identified in plasma membrane fractions enriched in hydrophobic proteins and obtained from an Arabidopsis cell suspension culture. Fifty of these proteins had predicted transmembrane domains. As in the present study, Marmagne et al. (2004) used one-dimensional gels. To our knowledge, no non-ionic or zwitterionic detergent has been found to quantitatively resolve plasma membrane proteins from eukaryotic cells in the first (IEF) dimension of 2D-PAGE. Thus, so far, it has not been possible to use 2D electrophoresis for integral proteins of eukaryotic plasma membranes. However, integral proteins of bacterial cell membranes (Molloy et al. 2000) and of intracellular membranes of eukaryotic cells are more easily resolved by IEF using non-ionic and zwitterionic detergents, and can therefore be separated by 2D-PAGE. The reason for this might be that many plasma membrane proteins of eukaryotic cells are highly glycosylated. Furthermore, the plasma membrane-attached cortical cytoskeleton might render the eukaryotic plasma membrane recalcitrant to solubilization.

To compensate for the relatively few protein bands resolved by SDS-PAGE, we used nano-flow reversed-phase chromatography coupled ‘on-line’ to an electrospray ionization mass spectrometer (ESI-MS). Hence, the analysis of the complex peptide mixture originating from one protein band, which often consists of several proteins, was improved. By this approach we were able to identify more than 200 putative plasma membrane proteins in Arabidopsis green leaves. Of these 200 proteins about half are predicted to be integral proteins.

Results and Discussion

Protein identification

By examining the mass spectra with the Mascot search algorithm, 326 Arabidopsis proteins were matched in the National Center for Biotechnology Information (NCBI) database. About 25% of the proteins were identified by a single mass spectrogram, while more than half of the proteins were identified by three or more mass spectrograms (Table 1; supplementary Table 2 online). In some cases, the sequences of the identified peptides matched more than one isoform of a protein (supplementary Table 2 online, footnotes d–t). Thus, 22 additional isoforms, with sequence identity to proteins listed in supplementary Table 2, were not regarded as individual matches and are therefore not included in the total number of 304 proteins identified (Fig. 1; supplementary Table 2 online). According to the ARAMEMNON database (Schwacke et al. 2003), which combines the results of several prediction programs, 128 of the identified proteins have transmembrane domains (Fig. 2). In addition, two proteins lacking transmembrane domains are predicted to be glycosyl phosphatidylinositol (GPI) anchored (Borner et al. 2002). Thus, in total 130 proteins are either predicted to be integral membrane proteins or GPI anchored. However, we consider 16 of these proteins, e.g. enolase and alcohol dehydrogenase, to be incorrectly predicted as transmembrane proteins and thus 114 to be either GPI-anchored or true integral membrane proteins (Fig. 1; Table 1).

Table 1 and Fig. 3 include identified proteins involved in transport, signal transduction, stress responses, membrane trafficking, cellular organization, metabolism, as well as ‘cell-wall-associated’ proteins. Contaminants and putative contaminants are grouped in ‘probable contaminants’. A relatively large number of proteins (23%) could not be assigned a functional group and were therefore assembled in ‘unclassified’ (for a complete list of identified proteins including ‘polysomal proteins’, see supplementary Table 2 online).

Plasma membranes prepared by aqueous two-phase partitioning are of high purity (about 95%) and consist of right-side-out (cytoplasmic-side-in) vesicles (Larsson et al. 1994 and references therein). To reduce contamination by soluble proteins enclosed in the vesicles and by proteins only loosely attached to the membrane, the vesicles were turned inside-out by treatment with Brij 58 in the presence of 0.2 M KCl. Yet, a surprisingly large number of soluble enzymes were identified. These might in part be contaminating proteins, since many of them are highly abundant, as reflected by their high number of ESTs, (Table 1; supplementary Table 2 online), and might therefore be difficult to eliminate during preparation. We also identified 61 ribosomal proteins and five translation factors (supplementary Table 2 online), which probably originate from cytoskeletal-bound polysomes. This class of polysomes, which are attached to actin filaments, were first detected in animal cells (Hesketh 1996) and later also in plants (Davies et al. 1991) and may constitute as much as 80% of total polysomes. These polysomes are attached to the plasma membrane via actin filaments (Medalia et al. 2002).

By studying the overlap between proteins identified in the present study and those identified in proteomic studies of other cell compartments, the purity of the plasma membrane fraction can be assessed (for a complete listing of overlapping proteins see supplementary Table 2 online). There is an overlap of 11 proteins with the 416 proteins identified in Arabidopsis mitochondria (Millar et al. 2001, Millar and Heazlewood 2003, Heazlewood et al. 2004). Four of these, actin 2/7, H⁺-PP_iase (AVP3), a prohibitin and the expressed protein At1g08480, have predicted transmembrane domains. AVP3 was also identified in the chloroplast proteome (Froehlich et al. 2003) and remarkably the prohibitin isoform lacks ESTs. Of more than 70 identified proteins in the thylakoid proteome of Arabidopsis (Peltier et al. 2002) only the large subunit of Rubisco and the water-soluble photosystem I subunit ‘PSI-E-like protein’ overlap, both of which are abundant chloroplast proteins. Indeed, Rubisco is a common contaminant of almost any subcellular fraction obtained from green leaves. When compared to the 392 and 112 proteins identified in the chloroplast envelope by Froehlich et al. (2003) and Ferro et al. (2003), respectively, 13 proteins overlap, including six integral proteins, PIP1;1, TIP2;1, ERD4, PDR8, AVP3 and the ‘putative photosystem I subunit III precursor’. None of these six proteins is a well-established chloroplast envelope protein. Indeed, PIP1;1 is a canonical plasma membrane aquaporin and PDR8 was recently identified in a study on plasma membrane phosphopeptides (Nühse et al. 2003). Twenty-one proteins identified in the present study overlap with the 163 proteins recently identified by Shimaoka et al. (2004) in tonoplasts obtained from Arabidopsis suspension cells. Of these 21 proteins, six are V-ATPase subunits, four polysomal proteins and three proteins with unknown function. Of the remaining eight proteins, the prohibitin isoform and the mitochondrial H⁺-ATPase β-chain were also identified in mitochondria (Heazlewood et al. 2004). Furthermore, the aquaporin PIP2;7 and the sucrose transporter SUC1, which also overlap, are generally regarded as plasma membrane proteins, and enolase has been found in the cell wall in addition to the cytosol (see below). Shimaoka et al. (2004) also identified the α-soluble NSF attachment protein and the SSR-α signal sequence receptor, both of which are involved in membrane trafficking and thus may reside in several membranes, together with the putative leucine aminopeptidase, which as yet has no clear function in the cell. In a proteomic study of the cell wall, 81 proteins were identified (Chivasa et al. 2002). Four of these, pectinesterase, glyceraldehyde-3-phosphate dehydrogenase, enolase and calreticulin 3, were also found in our study. While pectinesterase is a distinct cell wall protein, the identification of the three others both in the cell wall and in the plasma membrane indicates that all four proteins are secreted. In fact, both glyceraldehyde-3-phosphate dehydrogenase and enolase have been immunolocalized outside the cell (Gozalbo et al. 1998) and to the cell wall (Edwards et al. 1999) in fungi. Furthermore, a calreticulin in Nicotiana plumbaginifolia has been immunolocalized to the plasma membrane (Borisjuk et al. 1998). Thus, the overlap between proteins identified by us as plasma membrane proteins and those identified in studies of other cellular compartments is small. We regard this as an indication of a low level of contamination by other membranes.

About one-third of the identified integral proteins have previously been shown or suggested to be plasma membrane proteins, based on various criteria (complementation of yeast mutants, activity, in vitro immunolocalization, sequence homology, etc.). In comparison with a study on phosphopeptides derived from Arabidopsis plasma membranes, 18 of the identified proteins overlap. Among these are four RLKs, two H⁺-ATPases (AHA1 and 2), the ABC transporter PDR8 and the sugar transporter STP13 (Nühse et al. 2003). One-third of the proteins identified by Marmagne et al. (2004) as plasma membrane proteins were also identified in the present study. Some proteins found in the study by Marmagne et al. (2004), e.g. 12 GTP-binding proteins, as well as several stress-regulated proteins, metabolic enzymes and unknown proteins (12, 12 and 18 proteins, respectively), were not found by us. In contrast, few receptor-like kinases were identified by Marmagne et al. (2004). Since the plasma membranes were isolated by aqueous two-phase partitioning in both studies, the observed differences probably reflect differences between Arabidopsis plasma membranes of cell suspension cultures and green leaves. Twelve of the integral proteins identified by us have been studied using immunolocalization or tagged fusion proteins (Stadler et al. 1995, Cutler et al. 2000, Sohlenkamp et al. 2002, Varet et al. 2003, Uemura et al. 2004). With the exception of the aquaporin TIP2;1, which was located to the vacuolar membrane (Cutler et al. 2000) and the syntaxin SYP71 (Uemura et al. 2004), which was located to the ER, the other 10 were also located to the plasma membrane in these studies. Interestingly, the wheat homolog of TIP2;1 is able to complement a yeast mutant deficient in ammonium transporters, which suggests that the TIP2;1 functions as a transporter of ammonia (NH₃) and that, at least in yeast, it is targeted to the plasma membrane (Jahn et al. 2004). TIP2;1, and the other tonoplast intrinsic protein found by us, TIP1;2, were also identified by Marmagne et al. (2004) as putative plasma membrane proteins. Strikingly, no TIP isoforms were identified in the proteomic study of the tonoplast by Shimaoka et al. (2004).

Transport

As could be expected for the plasma membrane, proteins involved in transport, such as aquaporins, H⁺-ATPases and sugar transporters, comprise a large group of identified proteins (Table 1, Fig. 3). Eight plasma membrane aquaporins (PIPs) of the 13 PIPs present in Arabidopsis were identified. The presence of these eight isoforms correlates well with microarray expression data in green leaves (L. Fraysse, unpublished results). Recently, Santoni et al. (2003) identified five PIP isoforms in Arabidopsis root plasma membranes by MS. Even though few PIPs show strict organ-specific expression, our results and the results by Santoni et al. (2003) taken together point to differences in the protein expression pattern. Four PIP isoforms (PIP1;2, 1;3, 1;4 and 2;6) identified in the present study were not identified in roots, whereas PIP1;5 was only identified in roots (Santoni et al. 2003). These five isoforms are thus potential organ-specific aquaporins of leaves and roots, respectively. At the subcellular level, PIP1;2, PIP2;1 and PIP2;7 have previously been located to the plasma membrane by GFP fusion proteins (Cutler et al. 2000). Four H⁺-ATPase isoforms, AHA1, 2, 4 and 11, were found. Only AHA3, not found by us, was earlier firmly demonstrated to be present in leaf tissue (reviewed in Arango et al. 2003) with a specific localization to the plasma membrane of phloem companion cells (DeWitt and Sussman 1995). Of the two disaccharide transporters, SUC1 and 2, and four monosaccharide transporters, STP1, 4, 3 and 13, identified, only STP13 has not yet been functionally classified or localized. For instance, SUC2 has previously been immunolocalized to the plasma membrane of phloem companion cells in all organs (Stadler et al. 1995). Two additional transport proteins, the β-subunit of the K⁺-channel and the ammonium transporter AMT1;1, have been localized to the plasma membrane by immunogold labeling and GFP tagging, respectively (Tang et al. 1996, Sohlenkamp et al. 2002).

Some less well-characterized transport proteins were also found. COPT1 is one of five Arabidopsis proteins that can functionally complement Saccharomyces cerevisiae high-affinity copper transport mutants (Kampfenkel et al. 1995). The hypothetical protein, At2g03620, is identical to the magnesium transporter MRS2-5. The LAX/AUX1-like permease identified has a strong similarity (77% identity) to the auxin transporter AUX1. In addition, a peptide sequence that matches two isoforms of lysine-histidine transporters was identified. The two nodulin-like proteins, predicted to have 10 and 11 transmembrane domains, respectively, both contain domains with high homology to prokaryotic oxalate/formate antiporters and may thus be antiporters.

Considering transport proteins non-canonical to the plasma membrane, the H⁺-PP_iase (AVP3), two ABC-transporters (PDR8 and MDR11), seven V-ATPase subunits and two TIP isoforms (discussed above) were identified. H⁺-PP_iases are usually regarded as vacuolar membrane proteins, although AVP3 was also identified recently in studies of the chloroplast envelope and the mitochondrial proteomes (Froehlich et al. 2003, Heazlewood et al. 2004). Furthermore, GFP-tagged H⁺-PP_iase (AVP2) was shown to be present in the Golgi apparatus (Mitsuda et al. 2001) and an H⁺-PP_iase has been immunogold localized to the plasma membrane of cauliflower (Ratajczak et al. 1999). So far, most ABC transporters characterized in plants have been localized to the vacuolar membrane (Sanchez-Fernandez et al. 2001). Yet, PDR8 was identified recently in a proteomic study of the chloroplast envelope (Froehlich et al. 2003). No previous localization of the other ABC transporter identified by us, MDR11, has been reported. However, by immunolocalization it has been shown that a PDR homolog in N. plumbaginifolia, NpABC1, as well as MDR11 in Arabidopsis, are present in the plasma membrane (Jasinski et al. 2001, Sidler et al. 1998), indicating that members of both the PDR and MDR subfamilies of ABC transporters may be located in the plasma membrane. There are several reports of the presence of animal and plant V-ATPases in membranes other than the vacuolar membrane and among these the plasma membrane (e.g. Rouquie et al. 1998, Toyomura et al. 2003). Only one of the seven subunits identified by us is part of the membrane-spanning V₀-domain of the V-ATPase and thus has transmembrane domains. Several soluble V₁-subunits were also identified in previous proteomic studies on the plasma membrane (Prime et al. 2000, Santoni et al. 2000) and the chloroplast envelope (Froehlich et al. 2003).

Signal transduction

About half of the identified proteins involved in signal transduction are receptor-like protein kinases (RLKs), which reflects the diversity of these proteins in the plasma membrane. In Arabidopsis, this gene family comprises more than 600 genes. Yet, less than 10 of these have assigned functions (Shiu and Bleecker 2001) and none of the RLKs identified in this report has been further studied. Three calcium-dependent protein kinases (CDPKs), CPK3, 9 and 21 were also identified. Several CDPKs are known to be associated with the plasma membrane, and recently CPK9 and 21 were located to the plasma membrane by GFP fusions (Dammann et al. 2003). We believe that the CDPKs found by us are good candidates for regulating plasma membrane proteins. Recently, Nühse et al. (2003) identified 299 putative plasma membrane phosphopeptides, which illustrates the significance of kinases in the regulation of plasma membrane processes. For instance, water transport by aquaporins and H⁺ pumping by the H⁺-ATPases have been demonstrated to be regulated by phosphorylation by plasma membrane-associated protein kinases (Johansson et al. 1996, Svennelid et al. 1999). Other Ca²⁺-binding signaling proteins identified were calmodulin and two calreticulin isoforms. The two identified phospholipases, a phosphoinositide-specific phospholipase C and a phospholipase D, have both been shown to be present in the plasma membrane and to be expressed in leaves (Otterhag et al. 2001, Wang and Wang 2001).

Membrane trafficking

Several proteins known to be involved in membrane trafficking between the Golgi apparatus and the plasma membrane were identified. Syntaxins are involved in the formation of SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) complexes. Three syntaxins of the SYP1 subfamily (SYP121, 122 and 132), which is the group most closely related to the plasma membrane-associated syntaxins in yeast and mammals (Sanderfoot et al. 2000), were identified. Recently, Uemura et al. (2004) GFP tagged these three syntaxins and located them to the plasma membrane in Arabidopsis cell suspension cultures, whereas the fourth syntaxin found by us, SYP71, was located to the ER in the same study. α-SNAP2 is believed to disassemble the SNARE complex after membrane fusion has occurred. ADP-ribosylation factor 1 (ARF1), a small GTP-binding protein, has also been suggested to be involved in transporting proteins to the plasma membrane, since mutants deficient in arf1 do not target the plasma membrane H⁺-ATPase correctly (Lee et al. 2002). The nodulin-like protein (At5g25250) is homologous to flotillin in Drosophila (FLO^Dm), which is involved in membrane trafficking and targeted to caveolae-enriched plasma membrane fractions (Galbiati et al. 1998).

Clathrin heavy and light chains are involved in endocytosis, and the two putative heavy clathrin chains identified in the present study show homology to their mammalian counterparts. In mammalian cells, the final scission of the clathrin-coated vesicle from the plasma membrane relies on dynamins through the action of their peckastrin homology (PH) domain. Interestingly, the dynamin identified in this study, ADL3, is one of two Arabidopsis dynamin homologs with a PH domain. While the function of ADL3 is still unknown, the other plant PH-domain-containing isoform, ADL6, is suggested to be involved in vesicle formation for vacuolar trafficking at the trans-Golgi (Jin et al. 2001, Mikami et al. 2000). The secretory carrier membrane protein, SC3, might also have a role in endocytosis, since mammalian homologs are suggested to be involved in clathrin-mediated vesicle budding from the plasma membrane and trans-Golgi complex (Fernandez-Chacon and Sudhof 2000). Prohibitins were previously thought to be exclusively located in mitochondria. However, evidence for localization outside mitochondria has been reported for mammalian cells, where prohibitins are suggested to associate with proteins involved in membrane trafficking (Bacher et al. 2002 and references therein).

Stress

Little is known about many of the stress-induced proteins identified. The KED-like protein is classified by its similarity to the KED protein in Nicotiana tabacum, named so because it is rich in lysine (K), glutamate (E) and aspartate (D). The transcript of this gene was shown to be up-regulated as an early response to wounding (Hara et al. 2000). The harpin-induced protein, NHL3, is encoded by a member of a large family of genes in Arabidopsis homologous to the NDR and HIN pathogen up-regulated genes in tobacco. NHL3 was recently immunogold localized to the plasma membrane (Varet et al. 2003). Two erd (early response to dehydration) gene products were found. ERD4 is an integral protein with 10 predicted transmembrane domains, but has not been further characterized. It was, however, identified by Froehlich et al. (2003) as part of the chloroplast envelope proteome. Evidence for the presence of callose synthases was found. However, due to high sequence homology between different isoforms, the sequences of the three peptides identified could only confirm that at least two different isoforms were present.

Cellular organization, and GPI-anchored and cell-wall-associated proteins

Several tubulins and actins were identified. In three cases, however, it was not possible to distinguish between isoforms, since the sequences of the peptides identified were identical (supplementary Table 2 online).

Although no precise function of a specific arabinogalactan protein (AGP) has been established in plants, they are implicated in cell-to-cell signaling during diverse developmental events. Many are also predicted to be GPI anchored (Borner et al. 2002), as predicted for the three AGPs (FLA1, 8 and 9) identified here. The root hair defective protein 3 (RHD3), which has three predicted transmembrane domains, is believed to play an essential role in cell wall biosynthesis and actin organization (Hu et al. 2003). Even though the role of spermine synthase has not been fully established in plants, it seems to have a role in cell wall expansion, since the inactivation of one spermine synthase isoform, ACL5, results in defects in the elongation of Arabidopsis stem internodes by reducing cell expansion (Hanzawa et al. 2000).

Metabolism

Surprisingly, 24 enzymes involved in primary or secondary metabolism were identified and only a few of these have transmembrane domains. Even if the soluble enzymes might be contaminants, some have previously been reported to be associated with the plasma membrane (e.g. enolase and glyceraldehyde-3-phosphate dehydrogenase, see above). However, as discussed in the Conclusions, some of the enzymes identified are highly abundant, soluble proteins and might therefore be contaminants of the plasma membrane preparation.

Probable contaminants

This group comprises contaminants and probable contaminants, such as two putative DNA-binding proteins, two putative transcription factors, four chloroplast proteins, one of which is a putative integral membrane protein and five mitochondrial proteins, of which none is integral. Two of the chloroplast proteins, the large subunit of Rubisco and the water-soluble photosystem I subunit ‘PSI-E-like protein’, were identified previously in a study of the thylakoid proteome (Peltier et al. 2002) and the ‘putative photosystem I subunit III precursor’ was identified in a study of the chloroplast envelope proteome (Froehlich et al. 2003). Thus, there is contamination by a few mitochondrial and chloroplast proteins. Assessing the level of contamination by soluble cytosolic proteins is more difficult.

Unclassified

More than half of the 56 unclassified proteins have predicted transmembrane regions and 25 show no homology to other proteins or are homologous to other unclassified proteins (supplementary Table 2 online). Thirty-two proteins, previously annotated in NCBI as unknown, putative or hypothetical, are now shown to be expressed with a likely localization to the plasma membrane. Two of the soluble proteins, the hypothetical protein At2g39730 and the expressed protein At5g08690, have also been identified in a proteomic study of the chloroplast envelope (Froehlich et al. 2003).

Conclusions

Identification of integral plasma membrane proteins by MS in combination with 2-D gel separation has proved to be difficult. We have, however, by the exclusion of the isoelectric focusing step and the use of reversed-phase HPLC coupled directly to the mass spectrometer (nano-LC/MS/MS), been able to identify a large number of integral proteins present in the plasma membrane. In total, over 200 proteins putatively associated with the plasma membrane were identified, and more than 100 of these are predicted to be integral.

Some of the proteins have already been established to be present in the plasma membrane by other techniques. The majority, however, have not previously been localized. Only a few of the identified proteins are typically associated with cell compartments other than the plasma membrane.

Whether these are contaminants of the plasma membrane preparation or not, cannot be definitely determined before tagged fusion proteins have been made or immunogold localizations have been done. Only a few obvious contaminants, such as highly abundant soluble proteins like the large subunit of Rubisco, were found. However, the use of plasma membrane vesicles treated with Brij should have reduced the number of contaminating soluble proteins. Importantly, the aim was not only to identify integral proteins of the plasma membrane, but also to identify true peripheral proteins. Thus, several interesting groups of soluble proteins, such as CDPKs, candidates to be regulators of plasma membrane proteins, were found.

Concerning integral proteins, there are altogether eight proteins out of over 100 reported that could be contaminants, although five of these have been reported to be true plasma membrane proteins in different species (i.e. isoforms of H⁺-PP_iases, V-ATPases, prohibitins and ABC transporters, see above). In addition, TIP1;2 and TIP2;1 were also identified by Marmagne et al. (2004) as plasma membrane proteins. These seven proteins are thus listed under ‘Transport’ and ‘Membrane trafficking’, and this leaves only the putative photosystem I subunit III precursor as a probable contaminating integral protein. Taken together this suggests a contamination by other membranes of less than 1%.

A group of proteins that mistakenly could be interpreted as contaminants, but are not, are the ribosomal proteins we detect. It is well established that polysomes are linked to actin filaments, which in turn are linked to the plasma membrane (Hesketh 1996, Davies et al. 1991, Medalia et al. 2002). Since we wanted to include peripheral plasma membrane proteins in our study, in addition to integral plasma membrane proteins, the problem of distinguishing between true peripheral proteins and contaminating soluble proteins arises. Notably, the putative contaminating proteins are abundant soluble proteins and not integral membrane proteins of organellar membranes. To wash away all contaminating soluble proteins one would probably have to wash away many true peripheral proteins. However, of a total of 124 soluble proteins identified, only 15 are likely to be contaminants. This is a low level of contamination by soluble proteins. In summary, only one integral membrane protein and 15 soluble proteins are probable contaminants (Table 1, ‘Probable contaminants’).

Evidence of only one ion channel, the β-subunit of the K⁺ channel, was found. To identify more ion channels and other integral proteins with low copy numbers, plasma membrane fractions enriched in integral proteins are probably needed. Furthermore, many plasma membrane proteins are probably only expressed in certain cell types, at discrete developmental stages or as a response to a particular stress. This means that the large majority of plasma membrane proteins remains to be identified. In addition, the functions of many of these proteins are unknown. Almost a quarter of the proteins identified in the present study are functionally unclassified and more than half of these are predicted to be integral.

Materials and Methods

Plant material

Arabidopsis ecotype Columbia-0 was grown on soil at 22°C with a short-day light regime (9 h light/15 h dark, 170 µE) and 70% relative humidity. Green leaves (including petioles) were harvested after 6–8 weeks and used for plasma membrane isolation.

Plasma membrane isolation

About 100 g of plant material was homogenized, using a knife blender, in 150 ml of 330 mM sucrose, 50 mM MOPS-KOH, pH 7.5, 5 mM EDTA, 0.2% casein hydrolysate, 0.6% polyvinylpolypyrrolidone (PVPP), 5 mM ascorbate, 5 mM DTT. PVPP, ascorbate and DTT were added immediately before use. Immediately after homogenization, PMSF was added to a final concentration of 0.5 mM together with 1.5 ml of a protease inhibitor cocktail for plant cell and tissue extracts (P9599; Sigma, St. Louis, MO, U.S.A.) in DMSO. The homogenate was filtered through a 200 µm nylon mesh and centrifuged at 10,000×g for 15 min; the supernatant was saved and centrifuged at 30,000×g for 50 min. The resulting microsomal pellet was resuspended in 6 ml of resuspension medium: 330 mM sucrose, 5 mM K-phosphate, pH 7.8, 0.1 mM EDTA, 1 mM DTT (freshly added) and 50 µl protease inhibitor cocktail. The resuspended membranes (6.0 ml) were added to an 18.0 g phase mixture to produce a 24.0 g aqueous polymer two-phase system with a final composition of 6.1% (w/w) Dextran 500, 6.1% (w/w) PEG 3350, 5 mM K-phosphate, pH 7.8, and 3 mM KCl. Plasma membranes were then purified by aqueous polymer two-phase partitioning as described previously (see Larsson et al. 1994, for the most recent update). The final upper phases were diluted at least two-fold with 330 mM sucrose, 5 mM K-phosphate, pH 7.8, 0.1 mM EDTA, and plasma membranes were pelleted by centrifugation at 100,000×g for 1 h. The whole preparation procedure was performed at 4°C. The plasma membrane pellet, containing 4–6 mg of protein, was resuspended in 0.5 ml of resuspension medium and 5 µl protease inhibitor cocktail, and stored in liquid nitrogen until used. Protein concentration was determined according to Bearden (1978) with BSA as standard.

The plasma membrane vesicles, which are largely cytoplasmic-side-in, were turned inside-out by treatment with the detergent Brij 58 (Johansson et al. 1995) to remove soluble proteins enclosed in the vesicles as well as loosely bound, contaminating proteins. This was done by mixing, at room temperature, stock solutions of 2 M KCl and of 2% (w/v) Brij 58 in 330 mM sucrose, 5 mM potassium phosphate, pH 7.8, with plasma membranes to give a detergent to protein ratio of 10 : 1 (w/w) and a KCl concentration of 0.2 M. Since this treatment also results in smaller vesicles (Johansson et al. 1995), the final centrifugation was run for 2 h at 100,000×g to increase recovery, which was 50–60% on a protein basis. The final plasma membrane pellet was resupended in half the original volume (usually 250 µl) of resuspension medium and 2.5 µl protease inhibitor cocktail, and used for SDS-PAGE.

Gradient SDS-PAGE

SDS-PAGE was performed essentially according to Laemmli (1970) using 12 cm gradient (12–20% acrylamide) separation gels. Gels were run at 4°C for 1 h at 5 mA followed by 18 h at 8 mA. Protein was stained with Coomassie Brilliant Blue R 250.

Trypsin digestion and nano-LC/MS/MS

The 30 protein bands above 15 kDa that could be detected were cut out and digested with sequencing grade modified trypsin (Promega, Madison, WI, U.S.A.) essentially according to Wilm et al. (1996). Briefly, the gel pieces were washed in 100 mM bicarbonate buffer and then destained in acetonitrile. The gel pieces were then rehydrated in bicarbonate buffer and thereafter reduced and alkylated with DTT and iodoacetamide. After further treatment with acetonitrile the pieces were dried in a speed vac before being rehydrated on ice for 40 min in trypsination buffer (12.5 ng trypsin µl^–1, 100 mM NH₄HCO₃, 5 mM CaCl₂). The proteins were digested overnight at 37°C. Peptides were extracted by incubating the gel pieces at 37°C with shaking, twice in acetonitrile and once with 5% formic acid (v/v) in between. All extractions were saved and pooled, and then lyophilized and redissolved in 5% formic acid. Using an Ultimate nano-LC system (LC-Packings, Amsterdam, The Netherlands) peptides were first bound to a pre-column (300 µm i.d. × 5 mm, 5 µm 100 Å^–1, C18 PepMap™), desalted by washing with 2.5% (v/v) acetonitrile, 1% (v/v) formic acid and finally loaded onto a Nano™ column (75 µm i.d. × 15 cm, 3 µm 100 Å^–1, C18) placed on-line to the MS and eluted with a gradient of 2.5–98% acetonitrile for 1 h and 20 min at a flow rate of 200 nl min^–1. MS/MS spectra were obtained using nano-electrospray ionization on an ion trap mass spectrometer (LCQ; Finnigan, San Jose, CA, U.S.A.) operating in a data-dependent acquisition mode.

Database search and peptide matching

Using the Sequest program (Thermoquest, San Jose, CA, U.S.A.), the MS/MS spectra were transformed into DTA files, which were used to search the Arabidopsis taxonomy of the National Center for Biotechnology Information (NCBI) database with the MASCOT MS/MS ion search option available at http://www.matrixscience.com. The peptide mass tolerance was set to 2 Da and the fragment mass tolerance to 0.8 Da with carbamidomethyl cysteine as a fixed modification and oxidation of methionine as a variable modification. One missed trypsin cleavage was allowed. The scoring algorithm of Mascot, the MOWSE score, is probability based and combines experimentally determined mass data with available amino acid sequence data (Perkins et al. 1999). Peptides were considered as matches either if they were classified as ‘significant’ (i.e. P > 0.05, which with our search parameters equals a MOWSE score of 40 or more), or if they showed ‘homology’ (a MOWSE score between 15 and 39, depending on the sequence of the peptide) and at the same time represented a protein with a theoretical molecular weight corresponding to the apparent molecular weight after SDS-PAGE. In a few cases single MS/MS spectra matched more than one peptide, most often due to highly homologous peptide sequences. However, only MS/MS spectra matching one peptide were considered. To further eliminate peptides identified incorrectly, MS/MS spectra corresponding to peptides with high homology to other sequences in the NCBI and The Arabidopsis Information Resource (TAIR) databases were evaluated by manual examination of the individual spectra. The number of unique peptides found and the number of acquired mass spectrograms for each protein are listed in supplementary Table 2 online. Predictions of transmembrane domains were performed by ARAMEMNON (Schwacke et al. 2003) available at http://aramemnon.botanik.uni-koeln.de. The Basic Local Alignment Search Tool (BLAST) was used at the NCBI and at TAIR to establish sequence homologies.

Supplementary material

Supplementary material mentioned in the article is available to online subscribers at the journal website www.pcp.oupjournals.org.

Acknowledgments

We would like to thank Adine Karlsson for preparing plasma membranes and running gels, and Stefanie Wienkopp and Helge Egsgaard (Risø National Laboratory) for their assistance and advice using the nano-ESI-MS. Funding from Formas to P.K. and from the Swedish Foundation for Strategic Research to C.L. are gratefully acknowledged.

Fig. 1 Properties of the identified proteins. The polysomal proteins probably originate from cytoskeletal-bound polysomes, anchored to the plasma membrane via actin filaments (Hesketh 1996, Davies et al. 1991, Medalia et al. 2002). ^a Of which one is a probable contaminant. ^b Of which 15 are probable contaminants.

Fig. 2 Predicted topology of integral proteins. Number of transmembrane as predict by ARAMEMNON (Schwacke et al. 2003).

Fig. 3 Functional classification of the putative plasma membrane proteins.

Table 1

Proteins identified by nano-LC/MS/MS

AGI acc no.	Protein		MW	TD	EST	UP/MS
Transport
At1g59870	PDR8	ABC transporter	165831	13	63	>3/>3
At3g28860	MDR11	ABC transporter	137215	10	20	1/2
At2g18960	AHA1	PM H+-ATPase	104614	10	194	>3/>3
At4g30190	AHA2	PM H+-ATPase	104734	10	110	1/>3
At3g47950	AHA4	PM H+-ATPase	105954	10	3	1/1
At5g62670	AHA11	PM H+-ATPase	105954	10	21	3/3
At4g29900	ACA10	PM Ca2+-ATPase	117469	9 a	8	1/3
At3g61430	PIP1;1	Aquaporin	30897	6	55	2/>3
At2g45960	PIP1;2	Aquaporin	30772	6	227	2/>3
At1g01620	PIP1;3	Aquaporin	30875	6	106	2/>3
At4g00430	PIP1;4	Aquaporin	30901	6	71	2/>3
At3g53420	PIP2;1	Aquaporin	30683	6	210	3/>3
At2g37170	PIP2;2	Aquaporin	30662	6	22	2/>3
At2g39010	PIP2;6	Aquaporin	31258	6	54	2/3
At4g35100	PIP2;7	Aquaporin	29802	6	125	3/>3
At3g26520	TIP1;2	Aquaporin	25889	7 a	284	1/>3
At3g16240	TIP2;1	Aquaporin	25125	6	86	3/>3
At1g71880	SUC1	Sucrose transporter	55222	12	23	1/1
At1g22710	SUC2	Sucrose transporter	54968	12	48	2/>3
At1g11260	STP1	Glucose transporter	57972	12	64	3/>3
At5g61520	STP3	Monosaccharide transporter	55002	12	7	1/1
At3g19930	STP4	Monosaccharide transporter	57457	11	17	>3/>3
At5g26340	STP13	Hexose transporter?	57382	12	23	3/>3
At4g13510	AMT1;1	Ammonium transporter	54055	10	21	1/3
At2g38290	AMT2	Ammonium transporter	50906	11	11	1/1
At1g04690	KAB1	K+ channel β-subunit	36744	–	43	1/>3
At5g59030	COPT1	Copper transporter	18558	2	4	1/3
At2g03620	MRS2-5	Mg2+ transporter?	47347	2	1	1/1
At3g54140		Peptide transporter?	64620	10	14	1/2
At5g01240		LAX1/AUX1 permease?	54833	9	4	1/2
At5g40780	LHT1	Lysine/histidine transporter?	49796	11	18	1/1
At2g39210		Nodulin-like protein antiporter?	66515	10	7	1/2
At5g14120		Nodulin-like protein antiporter?	63564	11	30	1/2
At1g15690	AVP3	H+-PPiase	81452	14	182	2/2
At1g78900	VHA-A	V-ATPase	69111	1	93	>3/>3
At1g76030	VHA-B1	V-ATPase	54188	–	37	>3/>3
At4g38510	VHA-B2	V-ATPase	54385	–	60	3/>3
At4g38920	AVA-P3	V-ATPase	13927	3	93	1/1
At3g58730	VHA-D	V-ATPase	29098	–	30	1/1
At4g11150	VHA-E1	V-ATPase	26272	–	39	>3/>3
At3g42050	VHA-H	V-ATPase	50253	–	27	>3/>3
Signal transduction
At1g48210	RLCK VIII	Serine/threonine kinase	39572	–	10	1/3
At2g43230	RLCK VIII	Kinase?	45338	–	3	2/3
At3g51550	CRPK1L	Receptor kinase?	98829	1	42	1/1
At5g49760	LRR VIII	Receptor kinase?	104624	1	12	2/2
At3g23750	LRR IX	Receptor kinase?	100472	1	5	2/2
At1g53430	S-Dom	Receptor Ser/Thr kinase?	114818	1	5	>3/>3
At3g14840	S-Dom	Receptor Ser/Thr kinase?	115546	1	8	2/2
At4g23180	DUF26	Serine/threonine kinase?	70297	1	9	2/2
At5g38990	CrPK1	Kinase?	98519	1	13	1/3
At2g37710	L-Lectin	Receptor kinase	71900	2	5	2/2
At3g28450	LRR X	Leucine-rich kinase?	67814	2	7	1/1
At4g08850	LRR XI	Receptor kinase?	116135	3	19	>3/>3
At2g31880	LRR VII	Leucine-rich kinase?	71523	2	18	1/1
At3g02880	LRR III	Leucine-rich kinase?	68167	2	15	>3/>3
At2g26730	LRR III	Leucine-rich kinase?	72106	1	8	2/2
At5g41260	RLCK II	Kinase?	70613	–	5	1/2
At3g26700	PPC:1	Kinase?	41124	1	2	1/2
At3g45780	PHOT1	Phototropin	112246	–	20	2/2
At4g18760		Leucine-rich repeat	46441	1	–	2/2
At4g23650	CPK3	Ca2+-dependent kinase	59527	1 b	29	>3/>3
At3g20410	CPK9	Calmodulin-domain kinase	60724	1 b	10	1/1
At4g04720	CPK21	Ca2+-dependent kinase	60199	1 b	8	>3/>3
At3g61050	CaLB	Ca2+-dependent lipid binding protein?	55174	2	11	>3/>3
At1g66410		Calmodulin 4	16844	–	24	>3/>3
At1g09210	CRT2	Calreticulin	48355	–	32	1/1
At1g08450	CRT3	Calreticulin	50098	–	12	3/3
At1g18080	ARCA	G-protein β-subunit?	36124	–	99	>3/>3
At3g08510	PLC2	Phospholipase	66480	–	23	>3/>3
At4g35790	PLD	Phospholipase	99595	–	50	>3/>3
At3g44310		Nitrilase 1	38154	–	36	>3/>3
At4g38580	ATFP6	Copper chaperone?	17241	–	10	2/2
At3g11490		rac GTPase?	49118	–	–	1/1
At1g05150		O-GlcNAc transferase?	90911	–	13	1/1
At1g10400		Putative glucosyl transferase	53433	1	2	1/1
At4g38740	ROC1	Peptidylprolyl isomerase	18589	–	49	>3/>3
At2g16600	ROC3	Cytosolic cyclophilin	18708	–	24	3/>3
At5g58710	ROC7	Cyclophilin	22004	–	11	1/1
Membrane trafficking
At3g09740	SYP71	Syntaxin	30135	1	21	>3/>3
At3g11820	SYP121	Syntaxin	38105	1	15	>3/>3
At3g52400	SYP122	Syntaxin	38042	1	12	3/>3
At5g08080	SYP132	Syntaxin	34318	1	3	1/1
At3g56190	ASNAP	α-Soluble NSF attachment protein	33190	–	19	3/>3
At1g61250	SC3	Secretory carrier membrane protein	32763	4	33	2/2
At2g47170	ARF1	ADP-ribosylation factor 1	20591	–	43	>3/>3
At1g59610	ADL3	Dynamin-like protein	91907	–	26	1/2
At1g35720	AnnAt1	Annexin	36182	–	66	>3/>3
At2g38750	AnnAt4	Annexin	36370	–	7	1/1
At5g22780		α-Adaptin C homolog	113083	–	13	2/2
At3g08530		Clathrin heavy chain?	191202	–	46	>3/>3
At3g11130		Clathrin heavy chain?	194404	–	61	1/2
At5g44140		Prohibitin	30798	1	–	1/1
At5g25250		Nodulin protein?	52391	–	10	>3/>3
At3g12390		Nascent peptide complex, α?	21969	–	17	3/>3
At1g72150		Cytosolic factor?	64121	–	84	>3/>3
At2g21160	SSR-α	Signal sequence receptor?	28264	1	30	1/>3
At5g66680		Oligosaccharyltransferase?	48656	1	35	>3/>3
At2g01720		Ribophorin I?	60199	2	16	1/1
At1g52600		Signal peptidase subunit?	20066	2	10	2/2
Stress
At5g62390		KED-like protein	51706	–	11	>3/>3
At5g06320	NHL3	Harpin-induced protein	25947	1	26	>3/>3
At3g63080		Glutathione peroxidase?	19486	–	4	>3/>3
At3g50360		Centrin	19396	–	1	1/1
At1g30360		ERD4	82282	10	53	>3/>3
At1g20450	ERD10	Dehydrin ERD10 (LTI49)	29530	–	80	2/3
At3g16470	JR1	Lectin	32103	–	43	1/3
At1g65960	GAD2	Glutamate decarboxylase	56504	1 b	65	2/2
At4g36750		Quinone reductase family	28850	–	6	>3/>3
At5g44070	PCS1	Phytochelatin synthase	55409	–	13	1/1
At5g02500	HSP70–1	DnaK-type chaperone	71712		537	1/2
At1g77020		Heat shock protein DnaJ?	40459	–	4	>3/>3
At2g24200		Leucine aminopeptidase?	54760	1	39	1/1
At1g70890		Major latex protein?	17992	–	14	2/3
At4g23670		Major latex protein?	17621	–	90	1/1
At3g28940		AIG2 protein?	19567	–	28	1/1
At1g54410		Dehydrin?	10789	–	66	1/3
At2g15970		Cold acclimation protein?	21762	5	52	1/2
At2g37790		Aldo/keto reductase family	39267	–	6	1/>3
At1g08470		Strictosidine synthase?	44294	–	12	1/1
At4g03550	GSL5	Callose synthase	206897	14	5	2/2
At3g07160	GSL10	Callose synthase	222076	16	8	1/1
Cellular organization
At3g18780		Actin 2	42064	1 b	209	2/2
At4g14960		Tubulin α-6 chain	49520	–	167	>3/>3
At5g44340		Tubulin β-4 chain	50361	–	119	>3/>3
At5g19770		Tubulin α-5 chain-like	50250	–	53	3/>3
At5g12250		Tubulin β-6 chain	51237	–	38	3/>3
At5g58070		Outer membrane lipoprotein?	21421	–	48	1/1
At2g03880		Selenium-binding protein?	72850	1	2	1/1
GPI-anchored and cell wall-associated proteins
At4g12730	FLA2	Fasciclin arabinogalactan-protein	43592	– c	47	2/2
At2g45470	FLA8	Fasciclin arabinogalactan-protein	32768	2 c	44	>3/>3
At1g03870	FLA9	Fasciclin arabinogalactan-protein	26156	– c	33	>3/>3
At1g52400	BG1	Glycosyl hydrolase family 1	60877	1	84	1/2
At1g02310		Glycosyl hydrolase family 5	48698	1	2	1/1
At1g75680		Glycosyl hydrolase family 9	58117	2	41	1/3
At5g58090		Glycosyl hydrolase family 17	52600	1c	15	1/1
At2g45820		Remorin	21069	–	16	>3/>3
At5g20230		Blue copper-binding protein	20212	1 c	15	1/1
At4g12420		Pectinesterase	65767	1 c	13	>3/>3
At3g13870	RHD3	Root hair defective 3	89606	3	21	1/1
At2g21660	GRP7	Glycine-rich protein	16937	1 b	163	2/3
At4g39260	GRP8	Glycine-rich protein	16626	–	54	1/1
At5g07530	GRP17	Glycine-rich protein	53385	2	3	1/1
Metabolism
At2g36530	ACLB2	Enolase	47974	1 b	224	1/2
At5g49460		ATP citrate lyase	66356	–	18	1/3
At5g43940		Alcohol dehydrogenase	41528	1 b	62	1/1
At1g04410		Malate dehydrogenase?	35890	1 b	73	>3/>3
At5g14740		Carbonic anhydrase 2	28668	–	90	>3/>3
At5g61510		Quinone oxidoreductase protein?	34658	–	5	>3/>3
At4g14880	OASA1	Cytosolic O-acetylserinelyase	33898	1 b	81	1/3
At5g28020	ATCYSD2	Cysteine synthase	18542	1 b	9	1/2
At3g63520		9-cis-epoxycarotenoid dioxygenase	61154	–	26	1/1
At5g19530	ACL5	Spermine synthase	38864	–	13	1/1
At1g20620		Catalase 3	57059	–	646	2/>3
At5g27380	GSH2	Glutathione synthetase	54053	–	15	1/1
At4g09320	NDPK1	Nucleoside-diphosphate kinase	18973	1 b	41	2/3
At4g37930		Glycine hydroxymethyltransferase protein?	57535		182	1/1
At5g17920		S-methyltransferase	84646		584	2/2
At3g43190		Sucrose synthase?	93514	–	6	1/1
At1g13440	GAPDH	Glyceraldehyde-3-phosphate dehydrogenase	37004	–	456	>3/>3
At3g17390	SAMS3	S-adenosylmethionine synthetase?	42769	–	92	>3/>3
At1g17290	ALAAT1	Alanine aminotransferase?	47747	–	35	1/1
At4g22690	CYP706A1	Cytochrome P450	30344	–	67	1/2
At1g13110	CYP71B7	Cytochrome P450	57628	4	18	1/2
At3g03470	CYP89A9	Cytochrome P450?	59616	–	21	2/2
At3g12290		5,10-Methylenetetrahydrofolate dehydrogenase	31798	1 b	22	3/>3
Probable contaminants
At1g31330		PS I subunit III precursor?	24329	2	100	1/2
At3g61260		DNA-binding protein?	23244	–	16	>3/>3
At2g32080		DNA-binding protein?	32181	–	34	1/2
At1g17880		Transcription factor?	17935	–	13	1/>3
At1g73230		Transcription factor btf3?	17993	–	20	2/2
AtCg00490		Rubisco, large subunit	52956		–	3/3
At2g39730		Rubisco activase	52347	–	444	2/2
At4g28750		PS I subunit PSI-E?	11703	–	106	2/>3
At2g20360		NADH-ubiquinone oxireductase?	44022	1 b	30	1/1
At5g55070		2-Oxoglutaratedehydrogenase subunit	50079	–	20	1/1
At3g28710	VATD2	ATPase?	41335	–	49	1/>3
AtCg00480		ATP synthase CF1 β-chain	53957	–	–	1/1
At5g08670		Mitochondrial H+-ATPase β?	59805	–	104	3/3
At2g20580	RPN1	26S proteasome subunit S2	98082	3 b	29	2/2
At4g24820	RPN7	26S proteasome subunit?	44425	1 b	24	1/1
At1g68010	HPR	Hydroxypyruvate reductase	42449	–	75	1/1
Unclassified
At1g57990		Unknown protein	44495	10	20	1/1
At1g68100		Expressed protein	51007	7	13	1/2
At3g25290		Expressed protein	42703	5	13	1/1
At4g12980		Putative protein	42430	5	7	1/3
At3g45600		Putative protein	32494	4	12	1/1
At3g18524		Hypothetical protein	50568	4	4	1/3
At5g20650		Putative protein	15945	3	21	1/1
At1g16860		Expressed protein	50950	2	20	1/1
At4g33360		Putative protein	38122	2	14	1/1
At1g09070		Expressed protein	34282	1	34	2/2
At5g11560		Putative protein	106139	1	15	2/2
At3g62360		Putative protein	133888	1	15	3/>3
At3g17440		Expressed protein	30566	1	4	2/2
At2g20990		Expressed protein	61933	1	30	>3/>3
At3g54200		Putative protein	26053	1	9	1/1
At4g23470		Expressed protein	29661	1	44	1/3
At4g15630		Expressed protein	20267	4	5	1/3
At2g20230		Expressed protein	30101	4	26	2/2
At4g29870		Putative protein	18765	3	4	2/2
At5g60620		Putative protein	43562	3	3	1/1
At1g08480		Expressed protein	15803	2	9	1/1
At3g17780		Unknown protein	15235	2	8	1/1
At1g48440		Expressed protein	15253	2	13	1/2
At3g19340		Unknown protein	57151	2	3	1/1
At1g14870		Unknown protein	17643	2	7	2/>3
At5g11420		Putative protein	39900	1	25	1/1
At1g29250		Expressed protein	14564	1	5	1/2
At4g20040		Putative protein	52488	1	2	1/1
At1g17620		Expressed protein	28726	1	26	2/>3
At1g65270		Expressed protein	32628	1	12	1/1
At1g52200		Expressed protein	21678	1	5	1/1
At1g66970		Unknown protein	82269	1 c	23	1/1
At1g64500		Unknown protein	41675	–	14	1/3
At5g62740		Putative protein	31696	–	15	2/2
At3g01290		Expressed protein	31529	–	19	>3/>3
At1g69840		Unknown protein	31671	–	10	>3/>3
At4g13010		Putative protein	34642	–	28	>3/>3
At3g10300		Unknown protein	35451	–	12	1/1
At4g00710		Hypothetical protein	55677	–	6	1/2
At4g35230		Putative protein	58953	–	3	2/3
At4g38690		Putative protein	36894	–	11	>3/>3
At5g07350		Putative protein	19895	–	24	1/1
At3g50690		Putative protein	49707	–	2	1/2
At1g22530		Unknown protein	75962	–	26	>3/>3
At3g28715		Hypothetical protein	41331	–	28	1/3
At3g04830		Expressed protein	33728	–	11	1/3
At4g32740		Putative protein	26128	–	3	1/1
At5g01750		Expressed protein	24450	–	37	1/2
At1g13470		Hypothetical protein	42266	–	1	3/>3
At1g80180		Expressed protein	15211	–	3	1/1
At1g13120		Unknown protein	69939	–	6	1/1
At4g28440		Putative protein	16563	–	7	1/1
At1g10590		Expressed protein	15502	–	13	1/2
At2g30930		Expressed protein	17040	–	15	3/3
At1g14880		Unknown protein	17461	–	4	>3/>3
At2g21280		Expressed protein	38007	–	9	1/1
At1g29690		Expressed protein	62526	–	8	1/1

Proteins are grouped according to functional similarity. Accession numbers according to the Arabidopsis Genome Initiative (AGI) are given in column 1. The protein names are given in columns 2 and 3 together with commonly used abbreviations, if available. The theoretical calculated molecular weight (MW) in column 4 is from NCBI. The numbers of predicted transmembrane domains (TD) are given according to the ARAMEMNON database in column 5 and the number of ESTs according to TAIR in column 6. The number of identified peptides unique for the protein (UP) is given first in column 7, and is followed by the number of mass spectrograms recorded for the peptides (MS). Abbreviations: acc no., accession number; ?, similar/putative. For a complete list of identified proteins, including the class ‘polysomal proteins’, see Supplementary Table 2 online.

^a Protein with incorrectly predicted number of TDs by ARAMEMNON (Schwacke et al. 2003).

^b Soluble protein, which is incorrectly predicted to have TDs by ARAMEMNON.

^c Protein predicted to be GPI anchored according to Borner et al. (2002).

Abbreviations

 
• AHA

  plasma membrane H⁺-ATPase

 
• H⁺-PP_iase

  H⁺-inorganic pyrophosphatase

 
• PIP

  plasma membrane intrinsic protein

 
• TIP

  tonoplast intrinsic protein

 
• SYP

  syntaxin.