Soluble and filamentous proteins in Arabidopsis sieve elements

Authors


S. Dinant. Fax: +33 1 30 83 30 96; e-mail: sylvie.dinant@versailles.inra.fr; dinant@versailles.inra.fr

ABSTRACT

Phloem sieve elements are highly differentiated cells involved in the long-distance transport of photoassimilates. These cells contain both aggregated phloem-proteins (P-proteins) and soluble proteins, which are also translocated by mass flow. We used liquid chromatography–tandem mass spectrometry (LC-MS/MS) to carry out a proteomic survey of the phloem exudate of Arabidopsis thaliana, collected by the ethylenediaminetetraacetic acid (EDTA)-facilitated method. We identified 287 proteins, a large proportion of which were enzymes involved in the metabolic precursor generation and amino acid synthesis, suggesting that sieve tubes display high levels of metabolic activity. RNA-binding proteins, defence proteins and lectins were also found. No putative P-proteins were detected in the EDTA-exudate fraction, indicating a lack of long-distance translocation of such proteins in Arabidopsis. In parallel, we investigated the organization of P-proteins, by high-resolution transmission electron microscopy, and the localization of the phloem lectin PP2, a putative P-protein component, by immunolocalization with antibodies against PP2-A1. Transmission electron microscopy observations of P-proteins revealed bundles of filaments resembling strings of beads. PP2-A1 was found weakly associated with these structures in the sieve elements and bound to plastids. These observations suggest that PP2-A1 is anchored to P-proteins and organelles rather than being a structural component of P-proteins.

Abbreviations
OPP

oxidative pentose phosphate

ROS

reactive oxygen species

TCA

tricarboxylic acid

INTRODUCTION

In higher plants, photoassimilates are transported from source to sink organs via the phloem, which is also required for long-distance signalling (Turgeon & Wolf 2009; Dinant & Suárez-López 2012). Photoassimilates, including sugars, amino acids and organic acids, proteins, mRNAs, small RNAs and hormones, are translocated by mass flow in highly specialized cells, the sieve elements (SEs) (Sjölund 1997). These cells have a striking subcellular organization, with most of the cellular organelles and the nucleus disappearing during their differentiation, resulting in a clear lumen crossed transversely by sieve plates with large pores. Most of the remaining structures – the sieve element reticulum (SER) and the mitochondria – are stacked at the plasma membrane, and nacreous deposits thicken the cell walls lining the SE (Evert 1990; Sjölund 1997; van Bel 2003). This facilitates the trafficking of metabolites and macromolecules, which enter and exit this pathway either through transporters located in the plasma membrane of the SE or companion cells (CCs) or through specialized plasmodesmata. The sieve tubes also mediate long-distance transport, facilitating the translocation of proteins and RNAs. Several components of this trafficking pathway have recently been identified and shown to recruit large ribonucleoprotein complexes (Turgeon & Wolf 2009; Dinant & Lucas 2012).

SEs contain various classes of proteins (Cronshaw & Sabnis 1990; Sabnis & Sabnis 1995). A broad range of soluble proteins, ranging in size from a few kDa to several hundred kDa, has been described in the phloem sap of various species (Balachandran et al. 1997; Hayashi et al. 2000; Dinant & Lemoine 2010). The largest surveys have been carried out in cucurbits because the collection of large amounts of exuding sap is straightforward in these species, whereas, in most species, including Arabidopsis, only minute volumes can be collected by ethylenediaminetetraacetic acid (EDTA)-facilitated exudation (Deeken et al. 2008; Beneteau et al. 2010). More than 1200 proteins have been identified by liquid chromatography–tandem mass spectrometry (LC-MS/MS) in pumpkin (Lin et al. 2009). Some of these proteins, such as FLOWERING LOCUS T (FT) (Corbesier et al. 2007; Lin et al. 2007), have been shown to be transported in the phloem and to act as long-distance signals.

In addition to mobile sieve tube proteins, which can be collected in exudates, intriguing protein bodies, observed on transmission electron microscopy (TEM) and visible as transcellular strands on light microscopy, are observed in the SE. These structures have been named P-proteins, for phloem-proteins (Cronshaw & Sabnis 1990). P-proteins are found in all dicotyledonous and most monocotyledonous plants (Eleftheriou 1990; Evert 1990). P-proteins may be dispersive or non-dispersive (Arsanto 1982; Cronshaw & Sabnis 1990; Sabnis & Sabnis 1995; Knoblauch & Peters 2010). Non-dispersive P-proteins are observed in about 10% of angiosperm families (Behnke 1991). The best studied P-proteins are the forisomes found in the Fabaceae, which are non-dispersive spindle-shape bodies (Knoblauch & Peters 2010). However, the role of the P-proteins remains unclear.

Little is known about the soluble protein content of SE and the P-proteins in Arabidopsis. Non-dispersed P-protein bodies have been observed in differentiating protophloem SE (Busse & Evert 1999) and in the immature SE of the stem (Khan et al. 2007), whereas filaments have been observed in mature SE (Oparka & Turgeon 1999). We provide here a detailed analysis of Arabidopsis phloem sap protein content and a description of the organization of P-proteins in mature SE. We show that PP2-A1 – a member of the Arabidopsis PP2 lectin superfamily thought to correspond to P-protein components in Cucurbita maxima (Bostwick et al. 1992; Dannenhoffer et al. 1997; Dinant et al. 2003) – is associated with dispersive filaments of P-proteins. Moreover, PP2-A1 was found to be anchored to organelles in the sieve elements.

MATERIALS AND METHODS

Plant material and collection of phloem sap exudate

Arabidopsis Columbia-0 (Col0) plants were grown in the greenhouse (16 h photoperiod). Plants were collected at three stages – rosette, early and late flowering [stages 3.9, 6.0 and 6.5, respectively (Boyes et al. 2001)], for ultrastructure and immunofluorescence studies and for sap exudation. Phloem sap was collected by EDTA-facilitated exudation, as previously described (Beneteau et al. 2010), with exudation for 2 h in a phosphate buffer (50 mm potassium phosphate buffer pH 7.6, 10 mm EDTA). For proteome studies, exudates from 24 mature leaves were pooled and concentrated with a Centricon (10 kDa cut-off) concentrator (Millipore, Billerica, MA, USA). The average yield of proteins in the exudate was estimated at 0.7 ng per mg of leaf fresh weight. Phloem sap sugars were analysed as previously described (Beneteau et al. 2010).

Western blotting

Protein concentration was determined by the Bradford method with bovine serum albumin (BSA) as a standard. We heated 8 µg of protein per sample at 95 °C for 5 min, separated the proteins by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) (12% polyacrylamide gel) and then transferred them to a nitrocellulose membrane (Bio-Rad, Hercules, CA, USA) by semi-dry electroblotting. Actin was immunodetected with a mouse monoclonal anti-actin antibody (A0480, Sigma Aldrich, St Louis, MO, USA) (dilution 1:1000). α-tubulin was immunodetected with a mouse monoclonal anti-tubulin antibody (T5168, Sigma Aldrich) (dilution 1:2000). Secondary horseradish peroxidase-coupled anti-mouse and anti-rabbit antibodies were purchased from Invitrogen (61–6520; Carlsbad, CA, USA) (dilution: 1:2000) and Sigma Aldrich (A0545) (dilution 1:10 000), respectively. The membrane was blocked, washed and incubated with antibodies in phosphate-buffered saline (PBS), pH 7.4 supplemented with 5% non-fat milk and 0.05% Tween-20. Proteins were detected by chemiluminescence (Lumigen TMA6, Nugen, GE Healthcare, France), with a luminescence analysis system (Fujifilm LAS4000, Tokyo, Japan).

Preparation of samples for LC MS/MS

Phloem sap proteins were separated by SDS–PAGE in 12% polyacrylamide gels. The gel was stained with Sypro Ruby (Bio-Rad), and 15 bands distributed along the entire length of the lane were excised (2 mm in width) and analysed at the PAPPSO platform (http://pappso.inra.fr/). The proteins were subjected to manual in-gel digestion as follows: gel pieces were washed in successive baths of (1) 50% acetonitrile/50% 25 mm NH4CO3; (2) 100% 25 mm NH4CO3; and (3) 100% acetonitrile. Each bath had a volume of 40 µL. The gel pieces were then incubated in 20 µL of 10 mm dithiothreitol in 25 mm NH4CO3, for 45 min at 37 °C and in 30 µL of 50 mm iodoacetamide in 25 mm NH4CO3, for 45 min at 37 °C, for Cys reduction and alkylation, respectively. Proteins were digested overnight at 37 °C with 10 µL of a solution of 125 ng of modified trypsin (Promega, Madison, WI, USA) dissolved in 25 mm NH4CO3. The peptides were extracted successively with (1) 20 µL of 0.5% trifluoroacetic acid and 50% acetonitrile; and (2) 20 mL of 100% acetonitrile. The extracts were dried in a vacuum centrifuge and suspended in 25 µL of 0.08% trifluoroacetic acid and 2% acetonitrile for LC-MS/MS analysis.

Analysis of proteins by LC-MS/MS

LC-MS/MS was performed on an Ultimate 3000 LC system (Dionex, Sunnyvale, CA, USA) connected to an LTQ Orbitrap Discovery mass spectrometer (Thermo Fisher, Waltham, MA, USA) with a nanoelectrospray ion source. Tryptic peptide mixtures were loaded onto a Pepmap C18 precolumn (0.3 × 5 mm, 100 Å pores, 5 µm; Dionex) at a flow rate of 20 µL min−1. After 4 min, the precolumn was connected to the Pepmap C18 separating nanocolumn (0.075, 15 cm, 100 Å pores, 3 µm) and a linear gradient was established, from 2 to 36% buffer B (0.1% formic acid and 80% acetonitrile) in buffer A (0.1% formic acid and 2% acetonitrile), at a flow rate of 300 nL min−1 over 50 min. Ionization was performed at the liquid junction, with a spray voltage of 1.3 kV applied to an uncoated capillary probe (PicoTip EMITER 10 µm tip ID; New Objective). Peptide ions were automatically analysed by the data-dependent method, as follows: full MS scan (m/z 300–1600) on an Orbitrap analyser and MS/MS on the four most abundant precursors, with the LTQ linear ion trap. In this study, only peptides with +2 and +3 charges were subjected to MS/MS, with an exclusion window of 1.5 min, using classical peptide fragmentation parameters: Qz = 0.22, activation time = 50 ms, collision energy = 35%. The raw data generated by the LTQ-Orbitrap mass spectrometer were converted into an mzXML file with ReADW (http://sourceforge.net/projects/sashimi).

Protein identification from peptides

Proteins were identified with X!Tandem 1 software (X!Tandem tornado 2008.02.01.3; http://www.thegpm.org) against an Arabidopsis protein database associated with a proteomic contaminant database. The X!Tandem search parameters were as follows: trypsin specificity with a tolerance of one missed cleavage, fixed alkylation of Cys and variable oxidation of Met. The mass tolerance was fixed to 10 ppm for precursor ions and 0.5 Da for fragment ions. The final results were filtered by applying a multiple threshold filter at the protein level, consisting of a protein E-value of < 10−4 identified with a minimum of two different peptides, in at least one analysis, with a peptide E-value of < 0.05.

Protein annotations

Proteins were assigned to functional classes according to the GO and MapMan categories on TAIR (http://www.arabidopsis.org/) and BAR (http://bar.utoronto.ca/ntools/cgi-bin/ntools_classification_superviewer.cgi). For proteins involved in metabolic pathways, the annotations were refined with the Plant Metabolic Network (PMN) (http://pmn.plantcyc.org/ARA/server.html?). We compared the proteins identified with those found in the phloem sap of other species, using data reported for pumpkin (Lin et al. 2009) and oilseed rape (Giavalisco et al. 2006), and the accession numbers of the genes encoding the closest orthologue in Arabidopsis (Supporting Information Table S2). Comparisons were carried out with the Venn Selector program from BAR (http://bar.utoronto.ca). The frequencies of proteins belonging to various MapMan functional categories were determined with the Classification SuperViewer program (Provart & Zhu 2003). Normed frequencies were calculated as follows: (Nb_in_Classinput_set/Nb_Classifiedinput_set)/(Nb_in_Classreference_set/Nb_Classifiedreference_set). MapMan annotations [http://mapman.gabipd.org/web/guest/mapmanstore (file Ath_AGI_TAIR9_Jan2010.txt)] were automatically assigned to each gene on the SuperViewer browser. Cell wall proteins were identified by comparison with a list of 386 proteins compiled from published Arabidopsis cell wall proteomes (Chivasa et al. 2002; Borderies et al. 2003; Boudart et al. 2005; Charmont et al. 2005; Kwon, Yokoyama & Nishitani 2005; Bayer et al. 2006; Jamet et al. 2006; Irshad et al. 2008), from which cytosolic contaminants had been eliminated (Ito et al. 2011) (see Supporting Information Appendix S1).

Production of polyclonal antibodies against PP2-A1

Two rabbit polyclonal antisera were raised against PP2-A1, either against the recombinant PP2-A1-His6 protein produced in Escherichia coli (Beneteau et al. 2010) or against a specific synthetic peptide. An antiserum against the recombinant protein was produced at INRA (Antibody Facility, BIA, Angers-Nantes). An antiserum against a synthetic peptide mapping to a variable region (CNGKEKPQEKK) was purchased from BioGenes (Berlin, Germany). Both antisera were affinity purified with the recombinant protein. The PP2-A1-His6 protein was coupled to CNBr-activated Sepharose 4F, according to the manufacturer's protocol (Amersham Biosciences, Munich, Germany). The sera were incubated with the antigen-Sepharose matrix and the unbound fractions were removed from the resin by thorough washing. Antibodies were eluted with 0.2 m glycine (pH 2), 0.15 m NaCl and neutralized with 1.5 m Tris-HCl, pH 8.8. The specificity of the antibodies was checked by immunoblotting against PP2-A1-His6.

Indirect immunofluorescence analysis

Immunolocalization was performed on leaf and stem sections, with primary antisera against PP2-A1, as previously described (Masclaux-Daubresse et al. 2006), but with the following modifications: (1) after treatment with primary antibody, slides were incubated in Evans blue solution (0.001% in PBS; 10 min); (2) goat anti-rabbit IgG labelled with Alexa Fluor 488 (Molecular Probes, Carlsbad, CA, USA) was used as the secondary antibody. The two antisera raised against PP2-A1 gave identical results. Immunofluorescence was observed with a Leica TCS-SP2-AOBS confocal laser scanning microscope (Wetzlar, Germany). Additional views were obtained with a Leica DMRB-DIC fluorescence microscope. Callose was visualized by incubating sections with 5% aniline blue for 5 min in PBS buffer, after immunohistochemical treatment. Nuclei were stained by incubating sections in DAPI (1 ng µL−1) for 5 min. Fluorescence under UV light was visualized with a Zeiss microscope (Hertfordshire, UK) equipped with a broadband filter. Images were processed with Adobe Photoshop software and ImageJ (National Institutes of Health, Bethesda, MD, USA) for overlays.

TEM ultrastructure

The main vein of mature leaves and petioles was prepared with a modified version of a protocol described elsewhere (Roustaee et al. 2000). Samples were fixed by incubation in 2.5% glutaraldehyde in 100 mm phosphate buffer (pH 7.2) supplemented with 25 mm saccharose, for 3 h at 4 °C. They were then washed and post-fixed by incubation in 1% osmium tetroxide in 100 mm phosphate buffer (pH 7.2) for 2 h at 4 °C. The washed samples were soaked in 1% aqueous tannic acid solution for 30 min, then dehydrated in a graded series of ethanol concentrations. Tissues were then progressively infiltrated with Epon resin, which was polymerized by heating at 60 °C for 24 h. Ultrathin sections (60 nm) were cut on a Leica Ultracut-E microtome, collected on copper grids and stained with 7% uranyl acetate in ethanol for 15 min, then with lead citrate for 5 min. Sections were examined in a CM10 TEM operating at 80 kV (FEI) and equipped with a digital X-60 AMT camera. Two independent experiments were carried out with leaves from independent plants and three independent blocks were observed.

Immunoelectron microscopy

Leaf main veins were processed for immunogold labelling (Newman & Hobot 1999). The samples were fixed by incubation in 4% paraformaldehyde and 0.25% glutaraldehyde in 100 mm phosphate buffer (pH 7.2) for 1 h at room temperature. They were then washed three times, dehydrated in a graded series of ethanol solutions and embedded in 0.1% Benzyl-LR White acrylic resin (Agar). The resin was polymerized by UV illumination at −15 °C for 48 h. Ultra-thin sections (60 nm) were cut and collected on Parlodion-coated nickel grids. Sections were treated for 10 min with aqueous 0.1% glycine, then placed in a 1:20 dilution of fetal calf serum in 0.1% BSA, 0.02 m Tris buffer for 10 min. They were then incubated with primary antibody against PP2-A1-His6 for 45 min, washed in 0.1% BSA in 0.02 m Tris buffer and incubated with a goat anti-rabbit 10 nm colloidal gold-conjugated antibody for 45 min (diluted 1:20). Four individual blocks were observed. In control experiments, the primary antibody was replaced with 0.1% BSA in 0.02 m Tris buffer. The samples were observed under a FEI TEM, at 80 kV. Digital images were obtained with a lateral X-60 AMT camera.

Image analysis

The fine structure of the subunits constituting the ‘string of beads’ was analysed with the NIH image analysis program ImageJ v1.43u (NIH, http://rsb.info.nih.gov/ij/). Greyscale TEM images were first enhanced with brightness/contrast tools, and filament details were then analysed with CLAHE (enhance local contrast), fast Fourier transform) (FFT) and Lipshitz filters.

Accession numbers

The sequence data cited in this work can be found in the EMBL/GenBank data libraries under accession numbers: At4g19840 for PP2-A1; At3g43190 for SUS4; At5g20830 for SUS1; At1g66200 for GLN1;2; At3g56240 for CCH.

RESULTS

Enrichment of the EDTA-facilitated exudate in phloem sap

EDTA-facilitated exudate was obtained from mature rosette leaves sampled from plants after flower initiation. As expected, the protein profiles for whole leaf tissues and for the EDTA-facilitated exudate derived from the same leaves were different (Fig. 1a). We tested for an enrichment of the exudate in phloem sap, by determining its sugar content. A lower proportion of hexoses than of sucrose was found (Fig. 1b). The presence of hexoses may indicate, however, a minor contamination by apoplasmic fluids. To ascertain phloem sap enrichment, we performed Western blotting on the samples with antibodies against protein markers that we expected to be either present or absent in phloem. These markers included actin, which is found in the sieve tube exudates of a range of plant species, and tubulin, which has been shown to be absent from such exudates (Schobert et al. 1998). As expected, actin was present in both leaf tissues and EDTA-facilitated exudate. By contrast, the signal for α-tubulin was barely detectable in the exudate fraction, whereas it was present in leaf tissues (Fig. 1c). There was therefore limited contamination of the exudate with other cell types, although contamination with the loosely bound apoplasm proteins cannot be excluded in such approaches.

Figure 1.

Characterization of EDTA-facilitated exudate from Arabidopsis leaf. (a) Comparison of total proteins extracted from EDTA-facilitated exudate (sap) and leaf (tot) (1 µg per lane), visualized by SYPRO-ruby staining. (b) Sugar content of EDTA-facilitated exudate from Arabidopsis leaf. (c) Presence of actin and absence of tubulin in the exudate collected from Arabidopsis leaf (sap). For comparisons, total proteins were extracted from whole leaf (tot) and probed with appropriate antibodies. EDTA-facilitated exudate and total proteins (8 µg per lane) were probed with anti-actin antibody and anti-tubulin antibody. The molecular mass of actin and tubulin were deduced from pre-stained standards (41 and 50 kDa, respectively). EDTA, ethylenediaminetetraacetic acid.

The Arabidopsis EDTA-facilitated exudate contains a large set of proteins

SDS–PAGE analysis of the protein profile revealed the presence of peptides of 10 kDa to over 100 kDa in size (Fig. 1a; Supporting Information Fig. S1). Protein fractions were excised from the gel and analysed by LC-MS/MS, leading to the unambiguous identification of 377 proteins. The proteins were annotated according to the MapMan classification. They differed from the subsets of proteins predicted from the translatome of CCs (Mustroph et al. 2009) to be the most specific to or abundant in CC (Supporting Information Fig. S2). A large proportion of the proteins identified were enzymes involved in various metabolic pathways (Fig. 2a). The main other functional categories identified were oxidative stress and defence (Fig. 2b). A large set of proteins were annotated as cell wall proteins (Fig. 2b). We therefore compared our set of proteins with the available cell wall proteomes for Arabidopsis (Supporting Information Appendix S1). This analysis confirmed that a subset of 79 proteins, including α-galactosidases, glucosidases, curculins and other legume lectins, corresponded to cell wall proteins, thus demonstrating contamination of the exudate with loosely bound cell wall proteins. Eleven additional proteins annotated as cell wall proteins were also considered to be possible contaminants. Ultimately, 287 proteins were retained for the analysis of Arabidopsis EDTA-facilitated exudate proteins (Supporting Information Table S1).

Figure 2.

Distribution into functional categories of the 377 proteins found in the EDTA-facilitated exudate. (a) Distribution of the 127 proteins classified as belonging to metabolic pathways. (b) Distribution of other categories for the remaining 250 proteins. EDTA, ethylenediaminetetraacetic acid; OPP, oxidative pentose phosphate; TCA, tricarboxylic acid.

Proteins involved in the generation of metabolic precursors

Almost half the 287 proteins (125) were assigned to metabolic pathways (Supporting Information Table S1). The main categories identified were the generation of metabolic precursors, amino acid metabolism and carbohydrate degradation (Fig. 2a). The normed frequencies of these categories, calculated with the BAR Classification Superviewer, were significantly higher than those of representative sets of Arabidopsis genes (P value < 10−7) (Fig. 3). The category ‘generation of metabolic precursors’ included 43 enzymes involved in glycolysis, the Calvin cycle, the tricarboxylic acid cycle (TCA) cycle and the oxidative/non-reductive pentose phosphate (OPP) cycle (Supporting Information Table S1), and these pathways had high normed frequencies (P values of 2 × 10−7, 3 × 10−25, 1 × 10−7 and 1 × 10−16 for glycolysis, photosynthesis/photorespiration, the OPP pathway and the TCA cycle, respectively; Fig. 3). The enzymes involved in glycolysis and the Calvin–Benson cycle identified included fructose biphosphate aldolases (FBA1, FBA2), triose-phosphate isomerases (TIM, TPI), glyceraldehyde phosphate dehydrogenases (GAPA-1, GAPA-2, NP-GAPDH, GAPCP-2, GAPB, GAPC2, NP-GAPDH), enolase (ENO2), phosphoglycerate kinase (PGK), phosphoglucose isomerase (PGI1), UDP glucose pyrophosphorylases (UGP1, UGP2), phosphoribulokinase (PRK) and sedoheptulose-bisphosphatase (SBPASE). The enzymes of the TCA cycle were well represented and included two succinyl-CoA ligases, several malate dehydrogenases (MDH), isocitrate dehydrogenase (CICDH), two lipoamide dehydrogenases (mtLPD1, mtLPD2), aconitate hydratase (ACO1), one pyruvate dehydrogenase (MAB1), isopropyl malate isomerase (IIL1) and carbonic anhydrase CA1. The peroxisomal NAD-malate dehydrogenase (PMDH2), a key enzyme of gluconeogenesis, was also found. Several proteins involved in photorespiration and the OPP pathway were identified, including alanine-glyoxylate transaminase (AGT1), phosphoglycolate phosphatase (PGLP1), glycine decarboxylase P-protein (GLDP1), glycerate dehydrogenase (HPR1), several 6-phosphogluconate dehydrogenases, ribose 5-phosphate isomerase, transketolase and transaldolase.

Figure 3.

Frequencies of functional categories for the proteome of the EDTA-facilitated exudate. Normed frequencies (±bootstrap StdDev) for the proteome of the EDTA-facilitated exudate sampled from Arabidopsis leaves. Frequencies significantly above or below the values for the Arabidopsis reference set are indicated by * and °, respectively (P value < 0.05). EDTA, ethylenediaminetetraacetic acid; n.d., not detected; OPP, oxidative pentose phosphate; TCA, tricarboxylic acid.

Two enzymes involved in the degradation of starch were found: the α-amylase AMY1 and the β-amylase BAM5, previously reported to be present in Arabidopsis phloem sap (Wang, Monroe & Sjölund 1995). We did not find the sucrose synthases SUS1 and SUS4, despite previous suggestions that they play a key role in the metabolism of sucrose in phloem cells (Martin et al. 1993; Koch 2004).

Proteins involved in the synthesis and interconversion of amino-acids

The amino acid metabolism category was represented by 28 enzymes (Supporting Information Table S1). The main class identified was ‘central amino-acid metabolism’, with four aspartate aminotransferases, one alanine aminotransferase and a glutamate-glyoxylate aminotransferase (ASP1, ASP2, ASP3, ASP5, AOAT1, AOAT4, IMD3). The pathway of S-adenosylmethionine synthesis from aspartate and recycling via L-homocysteine was represented by two S-adenosylhomocysteinases, two methionine adenosyltransferases, methionine synthase and thiol methyltransferase (SAHH1, SAHH2, MAT3, MAT4, METS, HOL1). The enzymes of the glutamate-arginine pathway identified included a 2-oxoglutarate 5-aminotransferase, an ornithine carbamoyltransferase and N-acetyl-gamma-glutamyl-phosphate reductase (AOTA, OTC). The synthesis and degradation pathway for the glycine-serine-cysteine group was represented by several cysteine synthases, alanine-glyoxylate transaminase, phosphoserine aminotransferase and amino methyltransferase (OASB1, OASB, CYSC1, CYSD1, AGT1, PSA). By contrast, the only enzymes acting on nitrogen assimilation identified were one GDH enzyme and one GLU1 enzyme. No glutamine synthetases, such as GLN1;2 – key regulators of nitrogen assimilation in the companion cells (Lothier et al. 2011) – were found. ALDH10, an enzyme involved in the synthesis of the osmolyte glycine betaine, was also found. Several enzymes potentially involved in the biosynthesis of vitamins, including riboflavin, thiamin, folate and pyridoxine, were also found.

A few enzymes involved in the synthesis of precursors of glucosinolate or their degradation by myrosinases were found (UGT74B1, MAM-IL, SUR1, TGG1, TGG2, NIT2, GGP1, BCAT4, IPM1). The degradation pathways for xenobiotics and the methylglyoxate pathway were represented by glyoxalases and dienelactones. Several enzymes potentially involved in the biosynthesis of auxin from tryptophan were identified, including Nit2, UGT74B1, AMI1, MES2, MES3 and SUR1.

Proteins involved in oxidative stress and stress responses

The remaining 162 proteins belonged to various functional categories (Fig. 3). ‘Redox control’ was identified at a high frequency (P values of 10−13), and corresponded to a large set of proteins involved in reactive oxygen species (ROS) detoxification or scavenging (peroxiredoxin, ascorbate peroxidase, monodehydroascorbate reductase, glutathione disulfide reductase, catalase, superoxide dismutase, thioredoxin, glutathione S-transferase). Defence proteins were abundant (PR-proteins), as were putative storage proteins, including germins, cupins and a vegetative storage protein, and lectins, with curculins, jacalins, and legume lectins, some of which were annotated as cell wall proteins. Lectins are thought to be involved in defences against pests or symbiosis (De Hoff, Brill & Hirsch 2009). Three members of the major latex protein (MLP) family were identified.

Proteins involved in signalling

One of the proteins identified in this analysis was FT, a protein trafficking long distance through the phloem to induce flowering at the shoot apex (Giakountis & Coupland 2008). The detection of this protein is consistent with the developmental stage at which the plants were collected, after the induction of flowering. Other signalling proteins were also identified: nine G-box-binding factors, annexins, calreticulins, leucine-rich repeat (LRR) proteins and cyclophilins. Enzymes involved in protein turnover (including components of the 20S proteasome, proteases, protease inhibitors, aminopeptidases and the DJ-1 putative peptidase), folding or transport constituted a major category of the proteins identified, although the frequency of these proteins did not differ significantly from that in the normed Arabidopsis protein set (P value = 0.03). Several RNA-binding proteins were also identified, including the glycine-rich RNA-binding proteins GRP7 and GRP8 and the chloroplast ribonucleoprotein RBP31.

Proteins poorly represented in the exudate

The cytoskeleton was poorly represented, although profilin, actin and actin-depolymerizing factor were found in the EDTA-facilitated exudate, as previously reported in other species. We detected no ribosomal proteins, indicating that the level of contamination with the cytosol of the surrounding cells was low. Overall, categories associated with ‘protein’, ‘RNA’ and ‘DNA’ were poorly represented. No P-protein components were found (PP1, PP2 or SEO-like). Thus, the structural components of P-proteins, if present in the SE, were not detected in the EDTA-facilitated exudate.

Common trends in the sieve tube exudate proteomes of various dicotyledonous species

Comparisons with the proteomes of phloem sap from pumpkin and oilseed rape showed some overlap: 62 proteins were found to be common to pumpkin and 31 to oilseed rape, with 13 proteins found in all three species (Supporting Information Table S2). These proteins included malate dehydrogenase, the ascorbate peroxidase APX1, the FT protein, the dehydroascorbate reductase DHAR2, the RNA-binding protein GRP7, the enolase ENO2, the translationally controlled tumour protein TCTP, the nucleoside diphosphate kinase NDPK1, the adenosylhomocysteinase SAHH1, the thioglucosidase BGLU38, the elongation factor EF-1-alpha and the annexin ANNAT2. Several other functional categories were similarly overrepresented in all three species (Supporting Information Table S3). These overrepresented categories included, in particular those associated with primary metabolism: OPP, the TCA cycle, the Calvin cycle, glycolysis and amino acid metabolism. Nucleotide metabolism, redox/oxidative stress response and metal handling-related proteins were also highly represented. Proteins associated with the stress response, although abundant, were no more frequent than in the Arabidopsis reference set. By contrast, and as expected in enucleated cells, the ‘RNA’ (transcription, processing) and ‘DNA’ categories were present at only very low frequencies.

Sieve tube protein bodies in Arabidopsis

EDTA-exudate corresponds to the soluble fraction from the sieve tube. These cells usually also contain sieve tube proteins in inclusion bodies. We investigated the organization of P-proteins in situ, by preparing vein sections for TEM. Intact cells were preserved by chemical fixation. In the main vein of expanding leaves, typical phloem cell types were identified, including SE, CC and phloem parenchyma cells (Fig. 4). In SE, typical filaments were observed at various locations and in various orientations. The abundance of these filaments close to sieve plates was frequently asymmetric, often with haystack-like P-proteins accumulating on one side. This accumulation of P-proteins on one side of the cell plate was probably due to a surge towards the point of pressure release, as frequently reported. Dispersed filaments were also observed in the lumen of SE. In sieve plates, filaments frequently spanned the pores and their density seemed to be irregular, depending on position within the SE.

Figure 4.

Organization of phloem cells and P-proteins. (a) Observation of phloem anatomy in the main vein of a leaf. (b) Details of a differentiated SE, showing the accumulation of dispersive P-proteins. (c) High magnification of P-proteins showing filaments with globular subunits. CC, companion cell; CW, cell wall; mi, mitochondrion; P-P, P-protein; PPC, phloem parenchyma cell; SE, sieve element; SER, sieve element reticulum.

P-proteins are highly organized structures

We examined the organization of these filaments at higher magnification (Fig. 5). They were found to consist of globular subunits arranged in straight or slightly curved strands, resembling strings of beads. They had an even thickness, of ∼15 nm, and repeat distances of ∼10–12 nm. They were frequently more than 100 nm long. P-proteins are thus highly organized structures. The filaments were occasionally found associated in pairs, consistent with interfilament junctions, or as branched dichotomous filaments (Fig. 5.1a,b). When observed at a higher magnification, the globular subunits appeared to be heterogeneous, with uneven stacks of smaller units presenting successions of regular ellipsoidal stacks and more heterogeneous subunits (Fig. 5.2b). Using Lipshitz filters, we observed that the subunits constituting the filaments were themselves assemblies of smaller structural subunits of 3 to 5 nm in size (Fig. 5.1b, 5.2b1 and 5.3b1).

Figure 5.

Image analysis for filamentous P-proteins. 1, 2, 3: Initial TEM image. 1a, 1b, 2a, 2a1, 2b, 2b1, 3a and 3b: Higher magnification, before and after treatment of the image with ImageJ filters, for details of the filaments as indicated in 1, 2 and 3.

PP2-A1 proteins form small bodies present in the sieve tubes

In Arabidopsis, the closest orthologue of the phloem lectin PP2, PP2-A1, is encoded by a gene specifically expressed in the phloem (Dinant et al. 2003). PP2-A1 transcripts are abundant and specific to the CC translatome (Mustroph et al. 2009), consistent with the long-standing hypothesis that PP2 proteins are P-protein components (Read & Northcote 1983a). We investigated the location of PP2-A1 by indirect immunofluorescence, using primary antibodies against recombinant PP2-A1-His6 or against a specific synthetic peptide, purified against the recombinant protein. On confocal laser scanning microscopy (CLSM) and epifluorescence microscopy, a signal was obtained exclusively for vascular tissues. On transverse sections, the signal was found in the phloem transport areas, associated with CC/SE complexes (Fig. 6a–f). No signal was found in phloem parenchyma cells, cambium or other vascular cells. There was also no signal in CC/SE close to the cambium zone, suggesting a non-uniform location. The two antibodies gave similar results, recognizing specific structures in SE (Figs 6 & 7, Supporting Information Fig. S3). No signal was observed when the primary antibody was omitted.

Figure 6.

Immunolocalization of PP2-A1 in phloem cells. Transverse and longitudinal sections of the floral stem after immunohistolocalization, with specific, purified anti-PP2-A1 antibodies. (a–f) Transverse sections, (g–i) longitudinal sections. (a–c) Bright-field, Alexa 488 and overlay; (d–f) enlarged views of a, b and c; (g–i) Nomarski optics, Alexa 488, overlay. In (g), dense material, reminiscent of P-protein bodies, is visible within the SE (white arrows). Ca, cambium; CC, companion cell; Ph, phloem; En, endoderm; SE, sieve element; SP, sieve plate; Xyl: xylem. (a–c) bars = 50 µm; (d–i) bars = 10 µm.

Figure 7.

Immunolocalization of PP2-A1 in the sieve elements and counterstaining with aniline blue. Transverse (a–h) and longitudinal (i–l) sections of the floral stem after immunohistolocalization with specific PP2-A1 antibodies and counter-staining with DAPI and aniline blue. (a–d) Bright-field, DAPI, Alexa 488, overlay; (e–h) Bright-field, aniline blue, Alexa 488 and overlay; (i–l) Bright-field, aniline blue, Alexa 488 and overlay. CC, companion cell; SP, sieve plate; SE, sieve element; N, nucleus; with bright-field images, dense material reminiscent of P-protein bodies is visible within the SE (black arrow). Bars = 10 µm.

At higher magnification, fluorescence was found to be restricted to the SE. It appeared as discrete bodies present either across sieve plates or at the periphery of the cells, close to the plasma membrane (Fig. 6g–i). Double labelling with DAPI, which stains nuclei, confirmed the absence of signal from adjacent CC (Fig. 7a–d; Supporting Information Fig. S3b–d). Double labelling with aniline blue, which stains callose, identified the sieve plates and confirmed that the signal was restricted to the SE (Fig. 7e–h). The signal was often associated with transcellular strands present within SE and observed under bright light or with aniline blue staining (Fig. 7i–l; Supporting Information Fig. S3e–h). It appeared to emanate from a material attached to the sieve plates and passing from one side to the other in the SE, as expected for P-protein bodies. The signal was intense in some cells, but was not evenly distributed within the SE, often appearing to be stronger on one side of the SE.

Association of PP2-A1 with the P-proteins and plastids

The polyclonal antibody against recombinant PP2-A1-His6 was used for subsequent subcellular TEM studies. High-resolution immunolocalization with the anti-PP2-A1-His6 antibody detected an antigen on P-protein filaments in mature SE (Fig. 8a). The gold signals were uneven on the filaments and no labelling of the P-protein aggregates in the SE was observed. This antigen was most frequently detected on dense P-protein material located within the sieve pores (Fig. 8b). In addition, a strong immunogold labelling signal was found in the matrix associated with starch grain-associated plastids (Fig. 8c,d). We observed very limited labelling in the companion cells (Fig. 8e), confirming that PP2-A1 accumulated mostly in the sieve elements. No background signal was obtained in the absence of the primary anti-PP2-A1 serum (Fig. 8f), confirming the specificity of the antibodies.

Figure 8.

Subcellular localization of PP2-A1, as determined by transmission electron microscopy. Immunogold-labelled PP2-A1 was observed in sieve elements, associated with P-proteins and amyloplasts, on transmission electron microscopy (a–d, f). Localization in sieve element; (e) localization in companion cell. (a) Association with bundles of P-protein filaments assembled in dense structures at the sieve pores. (b) Association with P-protein filaments within the SE. (c,d) Association with amyloplasts within the SE. (e) Localization of PP2-A1 in the companion cells. (f) TEM image of the control treatment (no PP2-A1 serum), showing minimal background. Arrows indicate the gold label. SE, sieve element; CC, companion cell; CW, cell wall; P-P, dispersive P-proteins; SP, sieve plate; Po, sieve pore; pl, amyloplast; mi, mitochondrion; ER, endoplasmic reticulum; nuc: nucleus; vac: vacuole. Bars = 500 nm.

DISCUSSION

The phloem specializes in the long-distance allocation of a range of products, and this process requires the orchestration of local supply and demand in source and sink organs. The content of the phloem sap has been examined in detail in a few species (Dinant et al. 2010), mostly cucurbits, because it is easy to collect large volumes of sieve tube exudate from the plants of this group. However, phloem sap composition depends on the plant species, in addition to developmental stage and growth conditions. A recent study of pumpkin sieve tube exudates showed that the contents of the fascicular and extrafascicular sieve tube exudates differ considerably (Zhang et al. 2010a). Many studies have focused so far on extrafascicular sieve tube exudates. These data, indicating that the functions of the fascicular and extrafascicular phloem may be specialized and different, call into question the reliability and generalizability of the results obtained for cucurbits (van Bel et al. 2011a), which emphasizes the need to obtain descriptions of phloem sap composition for other plant species.

Usefulness of the EDTA-facilitated exudation method

Little information is available about the phloem sap proteome in Arabidopsis, mostly because this species displays no exudation after the cutting of petioles or stems and because only nanolitre volumes are obtained by stylectomy (Zhu et al. 2005; Hunt et al. 2006). EDTA-facilitated exudation (King & Zeevaart 1974) constitutes a feasible alternative method for many dicotyledonous species (van Bel et al. 2011b), and has been used in Arabidopsis for analysis of the amino acid or mRNA content of phloem sap (Deeken et al. 2008; Zhang et al. 2010b). Its use for the analysis of phloem sap protein content has been limited, mostly because the EDTA added to chelate Ca2+ ions was thought to generate artefacts (van Bel et al. 2011b). This assumption is supported by the different protein profiles obtained for phloem sap harvested by stylectomy and by EDTA-facilitated exudation, in barley (Gaupels, Knauer & van Bel 2008). Thus, thorough studies of the protein profiles obtained by EDTA-facilitated exudation are required, to assess the reliability of the method.

We applied the EDTA-facilitated exudation method to Arabidopsis for the analysis of phloem sap protein content. We identified 377 proteins by LC-MS/MS. These proteins were probably the most abundant, with other proteins present but not detected, because more than 1200 proteins have been identified in pumpkin exudates (Lin et al. 2009). We checked for enrichment in sieve tube exudate proteins by Western blotting with antibodies against actin or tubulin, positive and negative markers of phloem sap content, respectively (Schobert et al. 1998). Tubulin was barely detectable in the EDTA-facilitated exudate fraction, but was abundant in the whole leaf sample, consistent with previous reports on sieve tube exudates (Schobert et al. 1998). This finding confirms that the fraction was enriched in phloem sap. By contrast, actin was abundant in both extracts, as previously reported (Schobert et al. 1998). Consistent with this finding, the proteome of the EDTA-facilitated exudate included actin, actin-polymerizing factor and profilin, but no tubulin.

We identified several proteins reported in other studies as cell wall components, suggesting a leakage from the apoplasm. The method of exudation is based on the use of EDTA to prevent the occlusion of sieve pores. However, EDTA and other calcium-chelating agents have been reported to solubilize calcium-pectin complexes from the cell wall and to release bound enzymes and other loosely bound apoplasmic proteins (Masuda, Komiyama & Sugawara 1988, 1989; Boudart et al. 2005). We found that the exudate contained some cell wall proteins, consistent with apoplasm contamination. We therefore check for additional contamination due to the cytoplasm of the surrounding cells. Our data suggest that if such contamination occurs, it is probably limited. Firstly, proteins reported to be present in the cytosol of the companion cells, such as the glutamine synthetase GLN1;2 or the sucrose synthases SUS1 and SUS4, were not found in the exudate. Secondly, the proteins present in the exudate did not correspond exclusively to abundant proteins encoded by ribosome-associated mRNAs transcribed in companion cells (Mustroph et al. 2009). Thirdly, several classes of abundant proteins, such as ribosomal proteins, were not identified in the proteome. Finally, we identified a number of proteins repeatedly reported to be present in sieve tube exudates from other species. These observations provide evidence that the EDTA-facilitated exudate is enriched in sieve tube proteins. This method, as a first attempt to characterize sieve tube proteins, may therefore serve as a useful starting point for future studies.

Subsets of proteins consistently found in phloem exudates

Several proteins present in the Arabidopsis exudate have also been found in the phloem sap proteomes of castor bean, oilseed rape, rice or pumpkin (Barnes et al. 2004; Giavalisco et al. 2006; Aki et al. 2008; Lin et al. 2009). These proteins include a cytosolic malate dehydrogenase (MDH), the small RNA-binding protein GRP7, a component of the flowering autonomous pathway (Streitner et al. 2008), the dehydroascorbate reductase 2 (DHAR2) and the growth integrator TCTP (Berkowitz et al. 2008; Brioudes et al. 2010), which is expressed in the phloem and forms part of large RNP complexes in pumpkin phloem sap (Ham et al. 2009). A comparison of the pumpkin, oilseed rape and Arabidopsis phloem sap proteomes identified a further eight proteins: the ascorbate peroxidase APX1, the flowering factor FT, also found in rice; the enolase ENO2; the nucleoside diphosphate kinase NDPK1; the adenosylhomocysteinase SAHH1; the thioglucosidase BGLU38; the elongation factor EF-1-alpha; and the annexin ANNAT2. This subset of proteins may be considered a hallmark of the phloem sap proteome.

The sieve tube: a specialized compartment for the synthesis of precursor metabolites

Several enzymes involved in the generation of metabolic precursors were found in the EDTA-facilitated exudate. They included enzymes of glycolysis, photorespiration, the oxidative pentose phosphate (OPP) cycle and gluconeogenesis. These observations are consistent with reports for the phloem sap of Ricinus communis, which contains a full complement of glycolytic intermediates, despite displaying only weak enzymatic activities (Geigenberger et al. 1993). Precursors, such as Glc6P and Fr6P, have also been found in cucurbit sap (Richardson, Baker & Ho 1982; Fiehn 2003) and glycolytic activities have been reported in Robinia pseudoacacia and Cucurbita pepo (Kenneke, Ziegler & De Fekete 1971; Lehmann 1973a,b). These data suggest that metabolic activity occurs in the SE, potentially generating both metabolite precursors and reducing power. However, we did not identify all the enzymes of these pathways, suggesting that the proteome reported here corresponds to only a limited fraction of sieve tube proteins. We would also expect such activities to be tightly regulated. Indeed, it has been shown that respiration is partly inhibited and glycolysis is restricted in castor bean phloem sap, possibly due to the low levels of oxygen present (van Dongen et al. 2003, 2004).

A second consequence of the low oxygen levels in phloem sap is the preferential use of SuSy and UGPase, which consume less ATP (Geigenberger et al. 1993; Geigenberger 2003), for glycolysis. UGPases were identified in the exudate, together with nucleoside diphosphate kinases, which might assist the sucrose synthase pathway by converting UTP to ATP. However, we found no SuSy or phosphofructokinase, which is predicted to be active in the sieve elements (Koch 2004), in the phloem sap analysed here. This again suggests that the proteome reported here may not be exhaustive. It may also indicate that some of these enzymes are anchored to membranes, as suggested for the SuSy present in the sieve elements of castor bean (Wächter et al. 2003). Such anchored proteins would not be mobile in the sieve tube exudate.

A highly conserved feature of phloem sap: redox state control

Low levels of oxygen may also trigger the induction of genes responsible for ROS handling. These genes encode a range of enzymes involved in redox status maintenance and the regeneration of active forms of antioxidants, such as glutathione or ascorbate (Blokhina & Fagerstedt 2010). In the Arabidopsis EDTA-facilitated exudate, the frequency of proteins relating to antioxidant pathways and stress responses was high, as reported in many other phloem sap studies (Walz et al. 2002, 2004; Giavalisco et al. 2006; Kehr 2006; Malter & Wolf 2011).

Amino acid metabolism

Essential and non-essential amino acids are abundant in Arabidopsis phloem sap (Hunt et al. 2006, 2010). Phloem loading with these compounds depends on nitrogen remobilization, which is controlled by CC cytosolic glutamine synthetase, glutamate synthase, glutamate dehydrogenase and asparagine synthase activities (Ngai, Tsai & Coruzzi 1997; Tercé-Laforgue et al. 2004; Masclaux-Daubresse et al. 2006; Bernard & Habash 2009). It is also coupled to the activity of specific transporters (Rentsch, Schmidt & Tegeder 2007). In Arabidopsis, the main amino acids transported over long distance in sieve tubes are glutamate, aspartate, glutamine and asparagine (Hunt et al. 2006, 2010). Several enzymes responsible for the interconversion of these amino acids were found in the EDTA-facilitated exudate, such as alanine and aspartate amino transferases. These enzymes may be also involved in the maintenance of carbon–nitrogen balance in phloem sap, as suggested for alanine aminotransferase (Miyashita et al. 2007). No attempts to measure the activities of these enzymes in phloem sap samples have been reported. Furthermore, several of these metabolic steps usually take place in organelles such as mitochondria, chloroplasts and peroxisomes, whereas the presence of these enzymes in the soluble mobile fraction is more suggestive of cytosolic activity. This raises a number of questions, concerning the location of such enzymatic activities in SE, in particular, as these cells have no chloroplasts and peroxisomes (Raven 1991). Further investigations are required to determine the role of these enzymes in amino acid partitioning in sieve tubes.

Pearl-lace structure of Arabidopsis P-proteins

In addition to the soluble proteins present in the exudate, we studied the organization of P-proteins in the SE of Arabidopsis, by TEM. They consisted of filaments with a cross-hatched appearance, with a periodicity of about 15 nm. These filaments were unevenly distributed in the lumen of SE, although they generally accumulated on one side of the sieve plate. It has been suggested that P-proteins may be anchored to the plasma membrane or SER (Smith, Sabnis & Johnson 1987; Ehlers, Knoblauch & van Bel 2000; Knoblauch et al. 2001), but we found no evidence of P-protein binding to membranes in Arabidopsis SE. No P-protein bodies were observed in CC.

At higher magnification, the P-proteins resembled strings of beads similar to those described in Acer rubrum (Cronshaw & Sabnis 1990) and corresponding to subunits of about 15 nm in diameter. Single filaments were the most frequent organization observed in SE, but we also observed pairs of zipped filaments and branched filaments with an uneven dichotomy. Thus, even if the mechanism underlying P-protein formation is an ordered process, we observed some degree of variability in P-protein organization. At a higher magnification, the structural subunits appeared irregular, suggesting local variations in organization or composition.

PP2-A1 is associated with P-proteins

The first dispersive P-proteins to be described in detail were PP1, a 96 kDa protein found exclusively in cucurbits (Bostwick et al. 1992; Clark et al. 1997; Golecki et al. 1998; Golecki, Schulz & Thompson 1999), and PP2, a 48 kDa dimeric lectin that binds covalently to PP1 (Read & Northcote 1983a,b; Bostwick et al. 1992; Bostwick, Skaggs & Thompson 1994). PP2-like proteins have been identified in many angiosperms and PP2 genes specifically expressed in CC-SE complexes have been found in several species, including celery and Arabidopsis (Dinant et al. 2003), suggesting that PP2 may be a common component of P-proteins.

PP2-A1, the closest orthologue of the pumpkin CbmPP2 gene, is expressed in the phloem and its mRNA accumulates in CC-SE complexes (Dinant et al. 2003). PP2-A1 mRNA is indeed abundant in phloem cells (Zhao et al. 2005; Deeken et al. 2008; Mustroph et al. 2009). We showed by immunofluorescence that PP2-A1 accumulated in SE. Labelling was restricted to discrete bodies present in the lumen of SE, along cytoplasmic strands visible on the examination of fixed tissues by light microscopy, similar to those frequently observed in the SE of many species (Esau 1969). This observation is consistent with the TEM observation of dispersed P-proteins in SE in other species (Cronshaw & Esau 1968) and was further confirmed by TEM immunogold labelling, which showed that PP2-A1 was associated with dispersed P-proteins in the lumen of SE and stacks of P-proteins across sieve pores.

However, the immunogold labelling of these P-proteins was sparse. By contrast, we observed strong labelling associated with amyloplasts. This suggests that PP2-A1 is not exclusively attached to P-proteins. Moreover, the lack of detection of a PP2-A1 peptide in the EDTA-facilitated exudate proteome suggests that PP2-A1 may not be mobile in Arabidopsis SE, instead being mostly anchored either to P-proteins or to organelles. As PP2-A1, a GlcNAc lectin, has been shown to interact with phloem sap proteins, potentially playing a role in the shuttling of glycoproteins between CC and SE (Beneteau et al. 2010), it remains unclear how this role can be compatible with PP2-A1 being a structural component of P-proteins.

ACKNOWLEDGMENTS

We thank Dr Alain Guillot (PAPPSO Platform, Jouy-En-Josas) for LC-MS/MS analyses, Olivier Tranquet for producing PP2-A1 antibodies, Halima Morin for immunocytochemistry, and Laurence Lavenant and Laurent Marché for producing recombinant proteins. We also thank Laurence Bill for technical assistance. We would like to thank Dr Catherine Bellini for critical reading of the manuscript. JB held an INRA-BV/CEPIA fellowship. TL was supported by an INRA-FORMAS fellowship. This work was supported by a grant from the CEPIA department (ACI 2009–2010).

Ancillary