Characterization of a putative 3-deoxy-D-manno-2-octulosonic acid (Kdo) transferase gene from Arabidopsis thaliana

Glycobiology vol. 20 no. 5 pp. 617–628, 2010 doi:10.1093/glycob/cwq011 Advance Access publication on February 1, 2010

Characterization of a putative 3-deoxy-D-manno-2octulosonic acid (Kdo) transferase gene from Arabidopsis thaliana

2,4

2,5

genomes, suggest that AtKDTA encodes a putative KDTA involved in the synthesis of a mitochondrial not yet identified lipid A-like molecule rather than in the synthesis of the cell wall RG-II.

2 Laboratoire Glyco-MEV, UPRES-EA 4358, IFRMP 23, Université de Rouen, 76821 Mont-Saint-Aignan, France; and 3Unité Mixte de Recherche 619 Biologie du Fruit, INRA, Université de Bordeaux, Centre de Recherche INRABordeaux-Aquitaine, 33883 Villenave d'Ornon cedex, France

Keywords: Arabidopsis /Kdo /lipid A /plant / rhamnogalacturonan II

Received on September 8, 2009; revised on January 13, 2010; accepted on January 13, 2010

Introduction

The structures of the pectic polysaccharide rhamnogalacturonan II (RG-II) pectin constituent are remarkably evolutionary conserved in all plant species. At least 12 different glycosyl residues are present in RG-II. Among them is the seldom eight-carbon sugar 3-deoxy-D-manno-octulosonic acid (Kdo) whose biosynthetic pathway has been shown to be conserved between plants and Gram-negative bacteria. Kdo is formed in the cytosol by the condensation of phosphoenol pyruvate with D-arabinose-5-P and then activated by coupling to cytidine monophosphate (CMP) prior to its incorporation in the Golgi apparatus by a Kdo transferase (KDTA) into the nascent polysaccharide RG-II. To gain new insight into RG-II biosynthesis and function, we isolated and characterized null mutants for the unique putative KDTA (AtKDTA) encoded in the Arabidopsis genome. We provide evidence that, in contrast to mutants affecting the RG-II biosynthesis, the extinction of the AtKDTA gene expression does not result in any developmental phenotype in the AtkdtA plants. Furthermore, the structure of RG-II from the null mutants was not altered and contained wild-type amount of Rha-α(1-5)Kdo side chain. The cellular localization of AtKDTA was investigated by using laser scanning confocal imaging of the protein fused to green fluorescent protein. In agreement with its cellular prediction, the fusion protein was demonstrated to be targeted to the mitochondria. These data, together with data deduced from sequence analyses of higher plant

1 To whom correspondence should be addressed: Tel: +33-2-35-146394; Fax: +33-2-35-146615; e-mail: [email protected] 4 Present address: CNRS UMR 5203, Plate-forme de Protéomique Fonctionnelle, Institut de Génomique Fonctionnelle, IFR3, 141 rue de la Cardonille, 34094 Montpellier cedex 05, France 5 Present address: UMR 5004 BPMP, Campus INRA, SupAgro - IBIP, Bâtiment 7, 2 place Pierre Viala, 34060 Montpellier cedex 2, France

It was long believed that 3-deoxy-D-manno-octulosonic acid (Kdo) was synthesized only by Gram-negative bacteria as a component of the lipopolysaccharide (LPS) present in the outer membrane (Raetz 1990). In LPS, Kdo ensures the connection between the lipid A, the hydrophobic glucosamine-based phospholipid anchor of LPS, and the outer core and O-antigen. Kdo is crucial for bacteria since it was demonstrated that the minimal LPS required for the growth of Escherichia coli consists of the lipid A and two Kdo residues (Raetz and Whitfield 2002). Kdo was then discovered as a component of the primary cell walls of higher plants and of cell wall polysaccharides of some green algae (York et al. 1985; Becker et al. 1995). The synthesis pathway of Kdo involves the intermediate formation of D-arabinose, the enantiomer of L-arabinose found in the arabinan and arabinogalactan chains of the plant cell wall polysaccharides. The main steps of this synthesis are (i) the isomerization of D-ribulose-5-P into D-arabinose-5-P catalyzed by D-arabinose-5-P isomerase, (ii) the condensation of phosphoenol pyruvate with D-arabinose-5-P to yield Kdo-8-P and Pi catalyzed by Kdo-8-P synthase and (iii) the dephosphorylation of Kdo-8-P by Kdo-8-P phosphatase. The resulting Kdo is then activated by coupling to CMP (cytidine monophosphate; from cytidine triphosphate and free Kdo) in a reaction catalyzed by CMP-Kdo synthetase prior to its incorporation into the LPS of Gram-negative bacteria mediated by Kdo transferase (KDTA) (Rick 1987). In plants, the biosynthetic pathway leading to Kdo is almost fully conserved (Figure 1). Taking advantage of the complete Arabidopsis thaliana genome sequence, most of the genes have been identified (Wu et al. 2004) and some were functionally characterized (Figure 1). A single gene (At3g54690) encodes a putative homolog of D-arabinose-5-P isomerase (E. coli KpsF; Meredith and Woodward 2006). In Arabidopsis, two genes code for a functional Kdo-8P synthase (AtKDSA1, At1g79500 and AtKDSA2, At1g16340; Matsuura et al. 2003; Delmas et al. 2008), while a single gene occurs in pea (Brabetz et al. 2000) and tomato (Delmas et al. 2003). A single gene (AtKDSB, At1g53000) encodes the homo-

© The Author 2010. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: [email protected]

617

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

Martial Séveno , Emilie Séveno-Carpentier , 2 2 Aline Voxeur , Laurence Menu-Bouaouiche , 2 3 Christophe Rihouey , Frédéric Delmas , 3 2 Christian Chevalier , Azeddine Driouich , 1,2 and Patrice Lerouge

M Séveno et al.

Fig. 1. Biosynthesis and activation of Kdo in plant cells. Arabidopsis locus numbers and corresponding gene names (when functionally characterized) are indicated in the diagram.

log of CMP-Kdo synthetase that was previously characterized in maize (Royo et al. 2000). The only exception is the Kdo-8-P phosphatase for which homologous genes are not clearly predicted in Arabidopsis. Interestingly, all these sequences share strong similarities with their bacterial homologues, thus suggesting that they may have appeared in plants following symbiosis with bacteria. In addition to genes encoding enzymes involved in the Kdo synthesis or activation, Wu et al. (2004) mentioned the occurrence of a unique putative KDTA in Arabidopsis (AtKDTA encoded by At5g03770). However, the in planta function of this putative AtKDTA has not been elucidated so far. Kdo is one of the remarkable monomers of rhamnogalacturonan II (RG-II). This pectic polysaccharide of cell walls is composed of an α-1,4-linked homogalacturonan backbone that is substituted with four structurally different oligosaccharide side chains, namely A to D side chains (O'Neill et al. 2004). At least 12 different glycosyl residues are present in RG-II, including, in addition to Kdo, the rare aceric acid, apiose and 3deoxy-D-lyxo-heptulosonic acid (Dha) (O'Neill et al. 1996; Pérez et al. 2000). As a result of its complex structure, the biosynthesis of RG-II is likely to require at least 22 glycosyltransferases which is estimated to represent half of the glycosyltransferases needed for the synthesis of pectin (O'Neill et al. 2004). Despite its highly complex structure, RG-II is evolutionarily conserved in the plant kingdom as it is present in the primary cell wall of all higher plants predominantly in the form of a dimer that is cross-linked by a borate di-ester between two apiosyl residues (O'Neill et al. 1996; Kobayashi et al. 1996). This suggests that proteins involved in its synthesis appeared early in land plant evolution and that RG-II has fundamental functions in the primary wall organization. RG-II is believed to be involved in the regulation of cell wall properties and plant growth due to the boron-mediated cross-linking of RG-II that generates a covalently cross-linked pectic network (Fleisher et al. 1999; Ishii 618

Results Arabidopsis contains a single-copy gene for a putative KDTA Chromosome V of A. thaliana harbors a genomic sequence (At5g03770) that could encode a KDTA protein referred to as AtKDTA, as revealed by basic local alignment search tool (BLAST) searches using the E. coli KDTA (Clementz and Raetz 1991) as a query sequence. Full-length sequences for plant KDTA proteins were also identified in Oryza sativa (EEE55697.1), Medicago truncatula (ABE80128.2), Vitis vinifera (CAO15996.1), Populus trichocarpa (EEF04661.1), Ricinus communis (EEF33745.1) and Zea mays (ACN31913.1) as well as in the moss Physcomitrella patens (PHYPADRAFT 109233). The predicted translation products of these plant KDTA genes display significant amino acid identity percentage with various KDTA sequences of prokaryotic organisms (CAZy GT30, http://www.cazy.org; Cantarel et al. 2009) which all belong to the Gram-negative bacterial family. Despite these sequence similarities with the bacterial enzymes, the plant proteins represent a distinct lineage, which could be clearly separated from the Gram-negative bacterial proteins as seen in the phylogenetic tree (Figure 2). The amino acid sequence alignment of plant putative KDTA (Figure 3) indicates that AtKDTA shares from 48 to 67% of identity with other plant KDTA sequences. Sequence identity with the Gram-negative bacterial KDTA varies from 15 to 35%. The best rates were observed for rhizobial species with 31–35% of identity and 50–55% of similarity between plant and bacterial KDTA sequences supporting a direct homology between plant and rhizobial sequences (Figure 2). The A. thaliana putative KDTA is composed of 447 amino acids (Mr = 49871, pI = 9.96) and displays the two identified Pfam domains, Glyco_transf_N (PF04413) from residue 36 to 221 and Glyco_transf_1 (PF00534) from residue 242 to 415, of KDTA protein (http://pfam.sanger.ac.uk/; Finn et al. 2008). The prediction of the 3D structure of AtKDTA (Q8VZA5 entry of ModBase, http://modbase.compbio.ucsf.edu; Pieper et al. 2009) indicated that the protein is composed of two Rossman fold protein domains as reported for GT-B fold of glycosyltransferases (Lairson et al. 2008). Sequence alignments revealed the presence of highly conserved protein motifs within both plant and bacterial KDTA (Figure 3). Although no data are

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

et al. 1999; Ishii et al. 2001). So far, all mutations that affect the RG-II structure deeply alter the plant development (O'Neill et al. 2001; Ahn et al. 2006). Furthermore, the inactivation of the Kdo synthesis resulted in nonviable null mutants. Pollen tubes from mutants of the cytosolic D-arabinose-5-P isomerase or Kdo-8-P synthase were demonstrated to be unable to elongate properly and perform fertilization (Johnson et al. 2004; Delmas et al. 2008). Thus, although the precise function of RG-II remains to be established, RG-II is believed to play a key function in primary cell wall formation and its in muro functionality requires the presence of Kdo. In the present study, we report on the characterization of null mutants for a putative KDTA in A. thaliana to elucidate its potential involvement in the transfer of Kdo into the cell wall RG-II. Our data suggest that AtKDTA encodes a putative KDTA involved in the synthesis of a mitochondrial not yet identified lipid A-like molecule rather than in the biosynthesis of the cell wall pectic polysaccharide RG-II.

Characterization of Arabidopsis Kdo transferase

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016 Fig. 2. Phylogenetic tree of KDTA based on the maximum likelihood method. The scale bar (0.5) represents the number of amino acid residue substitutions per site. Plant and rhizobial KDTA are in bold and underlined, respectively. Accession numbers for the different protein sequences are indicated in the Materials and methods section.

available on amino acids involved in substrate recognition and CMP-Kdo recognition in bacterial KDTA, these motifs may be crucial for transferase activities in both prokaryotic and eukaryotic proteins. Among these, the two conserved sequences W58– G66 and G345–E350 could be involved in the binding of CMP-Kdo onto the KDTAs since these motifs were shown to be implicated in the CMP-Kdo binding site in CMP-Kdo synthases (Jelakovic and Schulz 2002). Furthermore, the asparagine residue (R279 in the Arabidopis sequence) of the conserved PR sequence was demonstrated to be required for the transfer of Kdo onto lipid A in bacteria (Figure 3) (Ekpo and Nano 2006).

Gene expression analysis of the Kdo biosynthetic genes in Arabidopsis plants We analyzed the expression of the genes involved in the Kdo biosynthetic pathway in Arabidopsis: AtKDSA1 and AtKDSA2 encoding the two isoforms of Kdo-8-P synthase (Matsuura et al. 2003), AtKBSB encoding the CMP-Kdo synthetase and AtKDTA encoding the putative KDTA. The expression analysis was performed using total RNAs extracted from mature flowers, stems, immature and mature leaves from wild-type Col0 plants. Results of transcript copy number estimations with real time quantitative reverse transcriptase-polymerase chain reac619

M Séveno et al.

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

Fig. 3. Protein sequence multiple alignments of KDTA from plants and from Gram-negative bacteria. The gray line above the alignments indicates the two conserved sequences likely involved in the binding of CMP-Kdo. The arrow pointed out to the conserved asparagine residue required for the transfer of Kdo onto lipid A in bacteria. Identical amino acids are shaded in black, and similar amino acids are in bold. Abbreviations and accession numbers are indicated in Materials and methods section.

tion (RT-PCR) are presented in Figure 4. All four genes were preferentially expressed in flowers and in immature leaves, i.e. in plant organs displaying cell division activities. This expres620

sion profile is in agreement with our previous work showing that the AtKDSA2 promoter is strongly activated in shoot apical meristem, first rosette leaves and flowers (Delmas et al. 2008),

Characterization of Arabidopsis Kdo transferase

and that the tomato kdsA gene expression and relevant Kdo-8-P synthase activity are preferentially associated with dividing cells in a cell cycle-dependent manner (Delmas et al. 2003). To a lesser extent, the expression of AtKDSA1, AtKDSA2, AtKBSB and AtKDTA was also detected in stems and mature leaves. Isolation of AtkdtA null mutants We obtained two AtKDTA T-DNA insertion lines from the Arabidopsis Biological Resource Center (SALK_035981 and

Biochemical characterization of RG-II isolated from the putative AtkdtA null mutants Studying the biological function of the Arabidopsis putative KDTA is hardly achievable through an enzymatic assay since CMP-Kdo required for the bioassay is an unstable compound (Belunis et al. 1995). As a consequence, to determine whether this putative transferase is involved in the incorporation of Kdo into RG-II, we investigated the RG-II structure from the homozygous AtkdtA1 null mutant. Pectins were extracted from leaf material by treatment of an alcohol-insoluble fraction with endopolygalacturonase (EPG). RG-II was separated from RG-I and oligogalacturonides by size exclusion chromatography (SEC), and its sugar composition was determined by gas liquid chromatography. As shown in Table I, the sugar composition of the AtkdtA1 RG-II fraction was similar to that of RG-II from wild-type Arabidopsis plants, including the Kdo content which

Fig. 5. Characterization of the AtkdtA null mutants. (A) Schematic representation of the T-DNA insertions in the AtkdtA gene. The coding region of the gene consists of 11 exons, represented as black boxes. White boxes represent the untranslated 5′ and 3′ regions of the corresponding mRNA. The location of primers used in RT-PCR experiments to characterize the AtkdtA null mutant is given in the diagram. (B) Molecular characterization of the AtkdtA null mutants by RT-PCR analysis of Arabidopsis AtKDTA mRNA levels in AtkdtA1 and AtkdtA2 mutants.

621

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

Fig. 4. Expression analysis of Kdo biosynthetic genes in Arabidopsis plants. Total RNAs were prepared from flowers, stems, immature and mature leaves of wild-type Col0 plants. The copy number of Kdo biosynthetic pathway gene transcripts estimated from quantitative RT-PCR experiments in real time is represented by bar graphs. Copy number estimates were conducted using cDNA synthesized from 8ng of total RNAs. Error bars represent experimental variation observed in four independent experiments using different wild-type plants.

GABI_140B12) designated as AtkdtA1 and AtkdtA2, respectively. The T-DNA insertion site of AtkdtA1 is illustrated in Figure 5A; it disrupts the AtKDTA gene within intron 7. This was confirmed by PCR analyses using genomic DNA as compared to control wild-type Col0 plants (not shown). RT-PCR analyses of T4 homozygous AtkdtA1 plants confirmed that AtkdtA1 is a null allele (Figure 5B). Identical results were obtained with AtkdtA2 mutant which is disrupted in exon 10 (Figure 5B). The general development and growth phenotype of AtkdtA null mutants appear to be similar to those of the wildtype Col0 plants (not shown). Thus under standard growth conditions and in contrast to mutants altered in the cytosolic biosynthesis of Kdo (Johnson et al. 2004; Delmas et al. 2008), the knock-out mutation of AtKDTA does not induce any apparent modifications of the plant phenotype.

M Séveno et al.

Table I. Glycosyl residue composition of RG-II isolated from the wild-type (WT) and AtkdtA1 mutant. The structure of monomers was confirmed by electron impact mass spectrometry. Due to the overlapping of GLC peaks, the quantification of Api and Dha was not achieved Monosaccharides

Mol%

Ara Rha Fuc Gal GalA GlcA Api 2-OMeFuc 2-OMeXyl Kdo AceA Dha

WT 14 15 5 13 36 3 Detected 3.5 4 3 3.5 Detected

AtkdtA1 16 18 4 15 31.5 2.5 Detected 3 4 3 3 Detected

AtKDTA is involved in a mitochondrial biosynthetic pathway Sequence analysis of Arabidopsis AtKDTA indicates that this protein is not predicted to be a type II protein as reported for Golgi glycosyltransferases. Prediction of its cellular localization using TargetP suggests that this putative transferase is targeted to the mitochondria. Similar predictions were obtained for other putative plant KDTA (O. sativa, M. truncatula, V. vinifera, P. trichocarpa, R. communis and Z. mays). To investigate the

Fig. 6. Characterization of the side chain C of RG-II isolated from the cell wall of AtkdtA1 null mutant. (A) HPLC trace of a RG-II from the AtkdtA1 null mutant submitted to a mild acid hydrolysis and a coupling to 1,2-diamino-4,5-methylene dioxybenzene (DMB). (B) MALDI-TOF mass spectrum of peak 2. The observed pseudomolecular ions were consistent with the presence of a DMB derivative of the Rha-α1-5-Kdo disaccharide.

622

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

represented 3%. Dha was also clearly detected in the RG-II fraction, and its presence was confirmed by gas liquid chromatography coupled to electron impact mass spectrometry. This suggests that the putative transferase AtKDTA is not involved in the transfer of this Kdo structurally related residue. To further confirm the presence of Kdo in RG-II from AtkdtA1 null mutants, this pectic polymer was submitted to a mild acid hydrolysis, taking advantage that Kdo can be specifically cleaved from the oligogalacturonide backbone in mild acidic conditions (Séveno et al. 2009). The resulting Kdo-containing side chain was then selectively coupled to 1,2-diamino-4,5-methylene dioxybenzene (DMB) into fluorescent derivatives and analyzed by reverse-phase high-performance liquid chroma-

tography (RP-HPLC) (Séveno et al. 2009). As illustrated in Figure 6A, a main peak was detected, collected and analyzed by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry. The observed pseudomolecular ions were consistent with the presence of a DMB derivative of Rha-α(1-5)Kdo (Figure 6B). We also investigated the structure of the side chains A and B of RG-II in AtkdtA1 mutant. Due to the presence of an acidic-labile apiose residue at their reducing end, the side chains A and B can also be specifically released by mild acidic hydrolysis of RG-II (Séveno et al. 2009). B and then A side chains were successively released from AtkdtA1 RG-II by treatment with 0.1 M trifluoroacetic acid (TFA) at 40°C (Figure 7A) and 80°C (Figure 7B) and structurally characterized by electrospray ionization mass spectrometry (ESI-MS). Molecular ions detected in the spectra were in agreement with the structure of A and B side chains and side chain fragments according to literature data (Pérez et al. 2003). Together, these findings demonstrate that the inactivation of AtKDTA does not result neither in the prevention of Kdo transfer into RG-II nor in any structural alteration of this pectic polymer. Therefore, the AtKDTA gene does not likely encode a KDTA involved in RG-II biosynthesis.

Characterization of Arabidopsis Kdo transferase

Discussion

Fig. 7. Characterization of the side chains A and B of RG-II isolated from the AtkdtA1 null mutant. Structure of A side chain (A) and B side chain (B) of Arabidopsis RG-II. ESI-MS of oligosaccharide fragments released by the mild acid hydrolysis in TFA 0.1M at 40°C (C) and 80°C (D) of RG-II isolated from the cell wall of AtkdtA1 mutant. *: potassium adduct. •: A side chain methyl esters. Api, D-apiose; AceA, L-aceric acid; 2-OMeFuc, 2-O-methyl L-fucose; 2-OMeXyl, 2-O-methyl D-xylose. B - Rha and A - 2OMeXyl refer to B side chain lacking L-rhamnose residue and A side chain lacking 2-Omethyl D-xylose, respectively. *: potassium adduct.

cellular localization of AtKDTA, this putative transferase was transiently expressed in tobacco leaves as a fusion protein to green fluorescent protein (GFP) and visualized using confocal microscopy. As shown in Figure 8A (green channel), 48 h after

Altering RG-II composition and structure is of an indisputable interest to understand its function in cell wall and its relationship with growth. Hence the isolation of plant mutants is needed to offer such a means of investigation. In this study, we aimed at performing a functional analysis of a candidate gene putatively encoding a KDTA referred to as AtKDTA. AtKDTA was identified in the Arabidopsis genome, based on amino acid sequence alignments with Gram-negative bacterial KDTAs (Figures 2 and 3). The best sequence homologies were observed for Rhizobial species with 31–35% of identity and 50–55% of similarity between plant and bacterial KDTA sequences. Sequence alignments revealed the presence of highly conserved protein motifs within both plant and bacterial KDTA which could be implicated in catalytic activity. Two biochemical lines of evidence strongly suggest that AtKDTA is not involved in RG-II biosynthesis in A. thaliana. First, Kdo was detected in the glycosyl residue composition of AtKDTA1 null mutant in a ratio that is consistent with a presence of one Kdo residue for one RG-II monomer. Second, the Rhaα1-5-Kdo side chain of RG-II was structurally characterized in this mutant after release by mild acidic hydrolysis and coupling to DMB (Figure 7). Furthermore, this putative transferase is not predicted to be a membrane protein as required for its localization in the Golgi apparatus, and the AtKDTA-GFP fusion protein does not accumulate in Golgi structures where RG-II is synthesized. As a consequence, we postulate that AtKDTA is not responsible for the incorporation of Kdo in RG-II. Other proteins such as sialic acid transferase-like enzymes (At3g48820 and At1g08660) have to be considered. Sialic acids are not synthesized in plants and therefore these putative transferases are not involved in the transfer of this residue onto glycopolymers (Séveno et al. 2004; Daskalova et al. 2009). Arabidopsis sialic acid transferase-like proteins were demonstrated to be located 623

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

infiltration, the AtKDTA-GFP fusion protein was mainly located in mobile 1-μm dots. To identify the nature of these dots, we coexpressed AtKDTA-GFP with a Golgi marker, the ERD2-cyan fluorescent protein (CFP) fusion protein (Saint-Jore et al. 2002) (Figure 8A). While the ERD2-CFP fusion protein accumulated in both endoplasmic reticulum and Golgi apparatus, the AtKDTA-GFP fusion protein appeared in dots that are clearly distinct from Golgi bodies. To determine whether the AtKDTA-GFP-labeled structures are related to mitochondria, we used the MitoTracker® orange to specifically stain these organelles. MitoTracker® orange is a mitochondrion-selective dye that enters the cells and is sequestered in the mitochondria where it reacts with thiols on proteins to form a fluorescent conjugate. Figure 8B shows a cell expressing the AtKDTA-GFP construct whose mitochondria are labeled with MitoTracker®. Mitochondria appear as dot structures of 1-μm diameter that are highly mobile. AtKDTA-GFP accumulated in the same dots as MitoTracker ® and showed a ring-like labeling surrounding the mitochondria (arrow in Figure 8B). Additional localization experiments were carried out on tobacco mesophyll cells that contain chloroplasts. No colocalization of AtKDTA-GFP with chloroplasts was observed (not shown). Together these data demonstrate that, under our labeling conditions, the fusion protein AtKDTA-GFP is localized in the mitochondria.

M Séveno et al.

in the Golgi apparatus by a proteomic approach (Dunkley et al. 2006) and by the analysis of the localization of the corresponding GFP fusion proteins (Daskalova et al. 2009) but their biological functions in this compartment have not been elucidated so far. Since Kdo and sialic acid transferases share common features such as the use of CMP-activated sugars as cosubstrates (Münster et al. 1998), these Golgi transferases may be considered as putative candidates for the transfer of Kdo and/or Dha into RG-II. A relationship between AtKDTA and RG-II biosynthesis is also highly questioned due to the absence of phenotypical alteration in the AtkdtA mutant lines when compared to wild-type plants. Indeed all mutations so far identified that target the RG-II composition and thereby structure deeply alter plant development (O'Neill et al. 2001; Ahn et al. 2006). Furthermore, the inactivation of the D-arabinose-5-P isomerase or Kdo-8-P synthase that is involved in the Kdo synthesis in the cytosol resulted in nonviable null mutants (Johnson et al. 2004; Delmas et al. 2008). It would be expected that preventing Kdo incorporation into the RG-II would also result in the inability of pollen from the null mutant to perform fertilization. In addition, the Kdo structurally related monosaccharide Dha was also clearly identified in the RG-II fraction, thus suggesting that AtKDTA is not involved in the transfer of this monomer in the Ara-β(1-5) Dha side chain of RG-II. 624

The cellular localization of AtKDTA was investigated through confocal imaging of the protein fused to GFP. The fusion protein was observed in 1-μm size mobile dots. Coexpression with Golgi marker clearly showed that these dots are not Golgi stacks as expected for glycosyltransferases involved in the RG-II biosynthesis. In contrast, AtKDTA-GFP accumulated in the same dots as MitoTracker®, a mitochondrial marker, and showed a ring-like labeling surrounding the mitochondria. As a consequence, in agreement with its predicted cellular localization, we conclude that AtKDTA protein is targeted to the mitochondria. Recently, BLAST search for candidate genes in the Arabidopsis genome revealed the presence of a set of genes potentially encoding homologues of LpxA, LpxB, LpxC, LpxD and LpxK involved in the biosynthesis of lipid A, the hydrophobic glucosamine-based anchor of LPS in bacteria (Wu et al. 2004). We searched in other sequenced plant genomes and found that Lpx genes are conserved in the plant kingdom. Furthermore, most of these genes encode proteins that are predicted to be targeted to mitochondria. Although the presence of lipid A-like molecules is still highly speculative, this observation suggests that the putative Arabidopsis AtKDTA may be involved in the biosynthesis of a not yet identified Kdo-lipid A mitochondrial membrane component in plants and not in the transfer of Kdo into RG-II. The presence of lipid A-like molecules in plants is

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

Fig. 8. Subcellular localization of AtKDTA. (A) Tobacco leaf epidermis expressing both ERD2-CFP (red) and AtKDTA-GFP (green) constructs at 48 h after infiltration. The two images were merged to show that AtKDTA-GFP fusion protein accumulated in mobile dots was distinct from the Golgi stacks (arrows). Bars = 5 μm. (B) Tobacco leaf epidermis expressing the AtKDTA-GFP construct (green) (48 h) was infiltrated with the MitoTracker® dye (red). The fluorescence was observed 4 h after MitoTracker® orange treatment. Note that AtKDTA-GFP dots and MitoTracker®-stained mitochondria colocalize (arrows). Bars = 5 μm.

Characterization of Arabidopsis Kdo transferase

PpKDTA, PHYPADRAFT_109233; Populus trichocarpa, PtKDTA, EEF04661.1; Rhodopirellula baltica, RbKDTA, RB11688; Ricinus communis, RcKDTA, EEF33745.1; Rhizobium leguminosarum, RlKDTA, RL0902; Rickettsia prowazekii, RprKDTA, RP089; Rhodopseudomonas palustris, RpaKDTA, RPA1158; Ralstonia solanacearum, RsoKDTA, RSc0693; Rhodobacter sphaeroides, RspKDTA, RSP_2394; Sinorhizobium meliloti, SmKDTA, SMc00894; Vibrio cholerae, VcKDTA, VC0233; Vitis vinifera, VvKDTA, CAO15996.1; Xanthomonas campestris, XcKDTA, XCC3337; Yersinia pestis, YpKDTA, YPO0055; Zea mays, ZmKDTA, ACN31913.1. An extract of the multiple alignments presented in Figure 3 was rendered by the ESPript/ENDscript program (http://espript.ibcp.fr; Gouet et al. 2003). Putative cellular localization of the proteins was analyzed using TargetP program (http://www.cbs.dtu.dk/services/TargetP; Emanuelsson et al. 2007).

Materials and methods Plant materials Wild-type and AtkdtA mutant plants of A. thaliana are in the Columbia (Col0) ecotype background. For germination, seeds were surface sterilized for 15 min in 12.5% (v/v) sodium hypochlorite and 0.02% (v/v) Triton X-100, rinsed at least five times and plated in Petri plates containing Murashige and Skoog's growth medium (Murashige and Skoog 1962). After cold treatment at 4°C for 2 days in the dark, the plates were incubated in a growth chamber at 22°C in a cycle of 16h light/8-h darkness. After 10 days of growth the plantlets were transferred in soil in a growth chamber in the same conditions.

Isolation of AtkdtA null mutants To isolate the homozygous AtkdtA1 null mutant, Arabidopsis genomic DNA was PCR amplified with the primers KdoTransf5′ (5′-TGTTGTCTGCTTTTTGTTCCA-3′) and KdoTransf3′ (5′-TTCGTCTTCCTTGGTGTAAGC-3′). PCR reactions were performed using the LBa1 primer (5′TGGTTCACGTAGTGGGCCATCG-3′), located within the LB sequence of the T-DNA, and KdoTransf5′ or KdoTransf3′ primer in order to amplify the T-DNA junction. The use of primers KdoTransf5′ and KdoTransf3′ generates a 606 bp fragment of the AtKDTA genomic sequence. PCR reactions were performed with genomic DNA extracted as described by Edwards et al. (1991). Fifty nanograms of genomic DNA was used for each PCR with primer combinations KdoTransf5′–KdoTransf3′, KdoTransf5′–LBa1 or LBa1– KdoTransf3′. Same strategy was used for the isolation of the AtkdtA2 null mutant using the T-DNA primer GABI-T-DNA (5′-CCATTTGGACGTGAATGTAGACAC-3′), AtKDOTF2 (5′-GCCTCGTAGACTATCGTGTGG-3′) and AtKDOTR (5′-TTTGACTTGTCGAGCCATTTC-3′).

Amino acid sequence analysis The phylogenetic analysis and drawing of the KDTA multiple alignments were performed on the phylogeny.fr server (http:// www.phylogeny.fr; Dereeper et al. 2008) with the PhyML program (Guindon and Gascuel 2003; Anisimova and Gascuel 2006) after blocks selection using Gblocks (Castresana 2000). The phylogenetic tree was drawn with TreeDyn (Chevenet et al. 2006). The multiple KDTA sequences alignment used for phylogenetic analysis was generated by the multAlin program (http://bioinfo.genotoul.fr/multalin; Corpet 1988). Accession numbers and abbreviations for the different protein sequences are as follows: Aquifex aeolicus, AaKDTA, aq_326; Acidobacteria bacterium, AbKDTA, Acid345_4718; Anaeromyxobacter dehalogenans, AdKDTA, Adeh_2615; Arabidopsis thaliana, AtKDTA, Q8VZA5; Agrobacterium tumefaciens, AtuKDTA, Atu0695; Brucella melitensis, BmKDTA, BMEII1029; Bacteroides thetaiotaomicron, BtKDTA, BT_2747; Caulobacter crescentus, CcKDTA, CC_0302; Chlorobaculum tepidum, CteKDTA, CT0593; Chlamydia trachomatis, CtrKDTA, CT208; Escherichia coli, EcKDTA, b3633; Fusobacterium nucleatum, FnKDTA, FN1606; Geobacter sulfurreducens, GsKDTA, GSU2259; Haemophilus influenzae, HiKDTA, HI0652; Helicobacter pylori, HpKDTA, HP0957; Leptospira interrogans, LiKDTA, LA1477; Legionella pneumophila Paris, LpKDTA, lpp2288; Mesorhizobium loti, MlKDTA, mlr8269; Medicago truncatula, MtKDTA, ABE80128.2; Neisseria meningitidis, NmKDTA, NMB0014; Oryza sativa, OsKDTA, EEE55697.1; Opitutus terrae, OtKDTA, Oter_2533; Pseudomonas aeruginosa, PaKDTA, PA4988; Physcomitrella patens,

RNA extraction, cDNA synthesis and RT-PCR analysis Wild-type or AtkdtA seedlings grown under light conditions for 10–12 days were transferred to soil for growth and used for RNA analysis. Total RNA were isolated using the TRIzol® Reagent (Invitrogen) and were treated with DNase RQI (Promega, Charbonnières-les-bains, France) according to the manufacturer's protocol. Two micrograms of total RNA from plantlets was reverse transcribed into cDNA using oligo(dT)16 as a primer, SuperScript™ II RNase H− reverse transcriptase (Invitrogen) in a total volume of 20 μL. The cDNA was then diluted 10 times, and 1 μL of the diluted cDNA was used as a template for PCR analysis. PCR reactions were performed using AtKDOTF1 (5′-TCGCTTATCTGGTTTCACGC-3′) and AtkdtAi3 (5′-ATCTGTTGGCCATGATGTGG-3′) as primers for detecting KDTA transcripts in the AtkdtA1 mutant, and using AtkdtAi5 (5′-GCATCTTCATTACATAGAGG-3′) and AtKDOTR as primers in the AtKdtA2 mutant. After an initial denaturation step of 5 min at 95°C, the reaction program was as follows: 30 s at 95°C as a denaturation step, 30 s at 60°C for primer annealing, 1 min 20 s at 72°C as an elongation step, for 30 cycles, and a final step of 5 min at 72°C. 625

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

debatable. Recent reports on this hypothetic bacteria-like glycolipid suggested its presence onto the chloroplast membrane (Raetz and Whitfield 2002; Armstrong et al. 2006). Lipid A may have appeared in plants following symbiosis with bacteria which would explain the high homology of all proteins of its biosynthetic pathway with bacterial sequences. This biochemical process led to the accumulation in the plant cytosol of Kdo that might have then been channeled to the synthesis of the plant cell wall RG-II during the early land plant evolution. Efforts are currently made in our laboratory to provide biochemical evidence for the presence of a lipid A-like on plant mitochondria and to unravel its putative function in higher plants. Since numerous genes encoding homologues of the Toll-like receptors are also found in the Arabidopsis genome, this plant lipid A may be involved in some yet unknown signaling mechanisms.

M Séveno et al.

Isolation of Arabidopsis RG-II One gram of frozen leaves was suspended in aqueous 80% (v/v) ethanol, heated at 70°C for 15 min and centrifuged at 2500 rpm for 5 min. The insoluble residue was treated for 4 h at 4°C with 0.1 N NaOH to saponify the methyl and acetyl esters. The suspensions were adjusted to pH 5 with 10% (v/v) acetic acid and then treated for 16 h at 30°C with an EPG from Aspergillus niger (30 units, Megazyme, Wicklow, Ireland) and then with α-amylase (500 units) for 16 h at 25°C. The suspensions were centrifuged, and the insoluble residues were washed with water. The EPG-soluble fractions were dialyzed (1 kDa cutoff dialysis tubing) against deionized water and freezedried. RG-I and RG-II polysaccharides were purified from the EPG-solubilized material by elution from SEC on a Sephadex G-75 (2.5 × 90 cm) (Amersham Pharmacia Biotech Inc., Uppsala, Sweden; 1.6 × 38 cm) column in 50 mM ammonium formate as buffer. 626

Mild acid hydrolysis of RG-II One milligram of purified RG-II was hydrolyzed in 200 μL of 0.1 M TFA for 16 h at 40°C. For the visualization of the A chain, the sample was then submitted to an additional hydrolysis at 80°C for 1 h. The solutions were then freeze-dried before analysis by ESI-MS-MS. Gas liquid chromatography analysis The samples were submitted to a 16-h methanolysis at 80°C with 500 μL of dry 2 M methanolic-HCl. After evaporation of the methanol, the methyl glycosides were then converted into their trimethylsilyl derivatives and separated by gas liquid chromatography. The gas chromatograph was equipped with a flame ionization detector, a wall-coated open tubular fused silica capillary column (length 25 m, i.d. 0.25 mm) with CPSil 5 CP as stationary phase and helium as gas vector. The oven temperature program was 2 min at 120°C, 10°C/min to 160° C, 1.5°C/min to 220°C and then 20°C/min to 280°C. The quantification of sugar was done by integration of peaks and determination of the corresponding molar values using response factors established with standard monosaccharides. Gas liquid chromatography coupled to electron impact mass spectrometry was performed using an Agilent (Wilmington, DE) GC 6890 chromatograph coupled to an Agilent 5973 mass spectrometer. Data collection was performed on HP Chemstation A.03.03. The separation of monosaccharides was performed on an Agilent HP5 column (30 m × 0.25 mm × 0.25 mm). Oven temperature was programmed from 60°C for 1 min to 140°C at 20°C/min, then a second ramp from 140 to 160°C at 5°C/min. Finally, a third ramp was used from 160 to 235°C at 2°C/min. This final temperature was maintained for 3 min. Injector and detector temperatures were 250 and 230°C, respectively. The injector was in splitless mode. Helium was used as carrier gas. Mass spectra (MS) were recorded in the electronic impact (70 eV) mode. The transfer line temperature was 250°C with source temperature at 250°C. Analysis of DMB-labeled side chain C A 14 mM solution of DMB (Sigma-Aldrich, Saint-Quentin Fallavier, France) was prepared by dissolving DMB in an aqueous solution of 80 mM β-mercaptoethanol, 40 mM sodium hydrosulphite and 2.8 M acetic acid. One milligram of the RG-II fraction was solubilized in 90 μL of deionized water and 90 μL of the DMB solution. The mixture was heated at 50°C for 2.5 h in the dark. Twenty microliters of the resulting solution was injected in a liquid chromatograph (Kontron, Milan, Italy) equipped with a reverse-phase C18 column (300 × 4.5 mm, C18 monomeric, Vydac, cat. # 238TP54) and SFM25 fluorescence spectrophotometer (Kontron). Elution of the DM B d er iv at iv es wa s pe rf or m e d at a fl o w r a t e o f 0.9 mL min−1 at room temperature using solvent A (acetonitrile:methanol:water, 4:6:90 [v/v]) and solvent B (acetonitrile: methanol:water, 11:7:82 [v/v]), with an A to B linear gradient from 50:50 to 0:100 (v/v) over 40 min. The DMB derivatives were detected by fluorescence using excitation and emission wavelengths of 373 and 448 nm, respectively. Mass spectrometry MALDI-TOF mass spectrum of the disaccharide side chains released by mild acid hydrolysis and coupled to DMB was

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

Real time quantitative RT-PCR Real time quantitative PCR was performed as to analyze the gene expression for AtKDSA1, AtKDSA2, AtKDSB and AtKDTA using the following primers, respectively: AtkdsA1i-5′ (5′-GAGGTTGGTGACCCAAACGG-3′) and AtkdsA1i-3′ (5′-CATGGTGCTTCTTTAAAACGTTCA-3′); AtkdsA2i-5′ (5′-CTGCATTTTGCAGGTCCGTAC-3′) and AtkdsA2i-3′ (5′-TATCATATCAGGCAACATCAGCTACTTTGA-3′); AtCMPKdoi-5′ (5′-GTTGACACACCCGATGATGTC-3′) and AtCMPKdoi-3′ (5′-CACACACAATGCCACAACTC-3′); AtkdtAi-5′ (5′-GCATCTTCATTACATAGAGG-3′) and AtkdtAi-3′ (5′-ACTTGTTGGCCATGATGTGG-3′). Reaction mixtures (25 μL) consisted of 12.5 μL of Platinum® quantitative PCR Supermix-UDG (Invitrogen) with 1 μL of SYBR Green I dye (Molecular Probes, Invitrogen, Cergy Pontoise, France) solution (1/3000 dilution in water of the manufacturer's 10,000× stock solution), 0.2 μM of each primers and cDNA corresponding to 8 ng of total RNA. Cycling conditions consisted of an initial uracyl DNA glycosylase treatment for 2 min and denaturation step of 95°C for 5 min as a “hot start”, 40 cycles of 95°C for 30 s and 60° C for 30 s. The PCR reactions and quantifications were carried out in the real time PCR detection system iCycler iQ (BioRad, Marne-la-Coquette, France). During PCR amplification, fluorescence emission was monitored. The increase in fluorescence emission of the dye used is proportional to the amount of PCR product accumulated that is in turn proportional to the starting amount of DNA template. A standard curve was constructed using templates of known copy number for the target sequence. The copy number of the samples was estimated by plotting the threshold cycle (the cycle at which fluorescence is considered to be significant above the background level and within linear range) against the log of the starting copy number (iCycler iQ real time detection system from Bio-Rad manual). To construct the standard curve, serial dilutions of cloned AtKDSA1, AtKDSA2, AtKDSB and AtKDTA fragments inserted into the pGEM-T Easy vector (Promega) were used. The number of copies in each dilution was calculated with the following formula: (number of moles) × (6.02 × 10 23 ) = number of copies. All standard samples were assayed in duplicate wells, and experimental samples were assayed in quadruplate wells.

Characterization of Arabidopsis Kdo transferase

acquired on a Voyager DE-Pro MALDI-TOF instrument (Applied Biosystems, Courtabeuf, France) equipped with a 337 nm nitrogen laser. MS were performed in the reflector delayed extraction mode using 2,5-dihydroxybenzoic acid (SigmaAldrich, Saint-Quentin Fallavier, France) as matrix. These spectra were recorded in a positive mode using an acceleration voltage of 20,000 V with a delay time of 100 ns. ESI-MS was performed on a Q-TRAP mass spectrometer (Applied Biosystems). Samples in 95% water/5% acetonitrile and 0.02% formic acid (v/v) were infused through Proxeon nanospray capillaries (Proxeon Biosystems, Odense, Denmark). The ion source conditions were adjusted for optimal sample ionization. According to this, infusion needle potential values were adapted between 0.9 and 1.5 kV. Spectra were acquired in a positive mode with a scan rate of 1000 amu s−1 and accumulated until a satisfactory signal to noise ratio had been obtained. Other voltages were as recommended by the manufacturer.

laser of 488 nm for GFP and 543 nm were alternatively used. GFP fluorescence was detected using a 545 nm dichroic beam splitter and a 505–530 nm bandpass filter, and MitoTracker® orange fluorescence was detected with a 545 nm dichroic beam splitter and a 560–615 nm bandpass filter. Image analysis and deconvolutions were performed with the Huygens Remote Manager (Ponti et al. 2007).

Subcellular localization of AtKDTA-GFP protein The cDNA of AtKDTA containing the full-length ORF was PCR amplified from the U17171 clone (from RIKEN cDNA clone RAFL06-11-M15 GenBank accession number AY065115) with primers 5′-GGGGACAAGTTTGTACAAAAAAGCCTTCATGAAGCTCGAGTGTTTG-3′ and 5′-GGGGACCACTTTGTACAAGAAAGCTGCGTCTTTGCAAATGTGATTC-3′. The AtKDTA sequence was then inserted in pGEM-T and sequenced. The cDNA was first inserted into pDONRTM 207 (Invitrogen). Then, the AtKDTA cDNA was subcloned into pMDC83, which contains the GFP6 cDNA, the dual P35S and the Nos T (ABRC, Ohio State University, Ohio; Curtis and Grossniklaus 2003) by using the Gateway® technology (Invitrogen). The ERD2-CFP construct (Saint-Jore et al. 2002), used as a Golgi marker, was kindly given by N. Paris (BPMP, Montpellier, France). The AtKDTA-GFP expression vector was introduced in the Agrobacterium strain GV3101 pMP90. Leaves of 5- to 6-week-old tobacco plants growing at 22°C were infiltrated with transformed agrobacteria as described by Brandizzi et al. (2002). For confocal imaging, the fluorescence emitted by the reporter proteins was observed 48–72 h after transient expression. The ERD2-CFP construct used as a Golgi marker was previously described (Chatre et al. 2005). For mitochondria labeling, leaves expressing the AtKDTA-GFP construct were infiltrated with 100 nM MitoTracker® Orange CMTMRos (Molecular Probes™) in Murashige and Skoog's buffer and plants were then left in the dark for 4 h before observation. For colocalization with chloroplasts in tobacco mesophyll cells, epidermis was removed prior to observation. The confocal imaging was carried out on a confocal microscope from Leica (TCS SP2 AOBS, Wetzlar, Germany) with a ×40 oil immersion objective. For imaging the coexpression of ERD2CFP and AtKDTA-GFP constructs, excitation lines of an argon ion laser of 458 nm for CFP and 488 nm for GFP were alternatively used with line switching mode on the multitrack setting of the microscope. CFP fluorescence was detected between 460 and 485 nm and GFP fluorescence between 509 and 602 nm. For imaging both the AtKDTA-GFP fluorescence and MitoTracker® labeling, we used a Zeiss LSM510 Meta laser scanning microscope (Jena, Germany) with a ×63 oil immersion objective. To do so, excitation lines of an argon ion

None declared.

Funding This research was supported by the Centre National de la Recherche Scientifique, the Région Haute-Normandie, the Region Aquitaine and the University of Rouen. F.D. was supported by grant no. 00-512858 from the Ministère de la Recherche et de la Technologie (France). Conflict of interest statement

BLAST, basic local alignment search tool; CFP, cyan fluorescent protein; CMP, cytidine monophosphate; Dha, 3-deoxyD-lyxo-heptulosonic acid; DMB, 1,2-diamino-4,5-methylene dioxybenzene; EPG, endopolygalacturonase; ESI-MS, electrospray ionization mass spectrometry; GFP, green fluorescent protein; Kdo, 3-deoxy- D -manno-octulosonic acid; KDTA, Kdo transferase; LPS, lipopolysaccharide; MALDI-TOF, matrix-assisted laser desorption ionization-time of flight; RG-II, rhamnogalacturonan II; RP-HPLC, reverse-phase high-performance liquid chromatography; RT-PCR, reverse transcriptasepolymerase chain reaction; TFA, trifluoroacetic acid. Acknowledgements We thank the cell imaging facilities of Universities de Rouen and Montpellier. References Ahn JW, Verma R, Kim M, Lee JY, Kim YK, Bang JW, Reiter WD, Pai HS. 2006. Depletion of UDP-D-apiose/UDP-D-xylose synthases results in rhamnogalacturonan-II deficiency, cell wall thickening, and cell death in higher plants. J Biol Chem. 281:13708–13716. Armstrong MT, Theg SM, Braun N, Wainwright N, Pardy RL, Armstrong PB. 2006. Histochemical evidence for lipid A (endotoxin) in eukaryote chloroplasts. FASEB J. 20:2145–2146. Anisimova M, Gascuel O. 2006. Approximate likelihood ratio test for branchs: A fast, accurate and powerful alternative. Syst Biol. 55:539–552. Becker B, Lommerse JPM, Melkonian M, Kamerling JP, Vliegenthart JFG. 1995. The structure of an acidic trisaccharide component from a cell wall polysaccharide preparation of the green alga Tetraselmis striata Butcher. Carbohydr Res. 267:313–321. Belunis CJ, Clementz T, Carty SM, Raetz CRH. 1995. Inhibition of lipopolysaccharide biosynthesis and cell growth following inactivation of the KdtA gene in Escherichia coli. J Biol Chem. 270:27646–27652. Brabetz W, Wolter FP, Brade H. 2000. A cDNA encoding 3-deoxy-D-mannooct-2-ulosate-8-phosphate synthase of Pisum sativum L. (pea) functionally complements a kdsA mutant of the Gram-negative bacterium Salmonella enterica. Planta. 212:136–143. Brandizzi F, Frangne N, Marc-Martin S, Hawes C, Neuhaus JM, Paris N. 2002. The destination for single-pass membrane proteins is influenced markedly by the length of the hydrophobic domain. Plant Cell. 14:1077–1092.

627

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

Abbreviations

M Séveno et al.

628

Johnson MA, von Besser K, Zhou Q, Smith E, Aux G, Patton D, Levin JZ, Preuss D. 2004. Arabidopsis hapless mutations define essential gametophytic functions. Genetics. 168:971–982. Kobayashi M, Matoh H, Azuma J. 1996. Two chains of rhamnogalacturonan II are cross-linked by borate-diol ester bonds in higher plant cell walls. Plant Physiol. 110:1017–1020. Lairson LL, Henrissat B, Davies GJ, Withers SG. 2008. Glycosyltransferases: Structures, functions, and mechanisms. Annu Rev Biochem. 77:521–555. Matsuura K, Miyagawa I, Kobayashi M, Ohta D, Matoh T. 2003. Arabidopsis 3-deoxy-D-manno-oct-2-ulosonate-8-phosphate synthase: cDNA cloning and expression analyses. J Exp Bot. 54:1785–1787. Meredith TC, Woodward RW. 2006. Characterization of Escherichia coli Darabinose 5-phosphate isomerase encoded by KpsF: Implication for group 2 capsule biosynthesis. Biochem J. 395:427–432. Münster AK, Eckhardt M, Potvin B, Mühlenhoff M, Stanley P, GerardySchahn R. 1998. Mammalian cytidine 5'-monophosphate N-acetylneuraminic acid synthetase: A nuclear protein with evolutionarily conserved structural motifs. Proc Natl Acad Sci USA. 95:9140–9145. Murashige T, Skoog F. 1962. A revised medium for rapid growth and bio-assays with tobacco tissue culture. Physiol Plant. 15:473–497. O'Neill MA, Warrenfeltz D, Kates K, Pellerin P, Doco T, Darvill AG, Albersheim P. 1996. Rhamnogalacturonan-II, a pectic polysaccharide in the walls of growing plant cell, forms a dimer that is covalently cross-linked by a borate ester. J Biol Chem. 271:22923–22930. O'Neill MA, Eberhard S, Albersheim P, Darvill AG. 2001. Requirement borate cross-linking of the cell wall rhamnogalacturonan II for Arabidopsis growth. Science. 294:846–849. O'Neill MA, Ishii T, Albersheim P, Darvill AG. 2004. Rhamnogalacturonan II: Structure and function of a borate cross-linked cell wall pectic polysaccharide. Annu Rev Plant Biol. 55:109–139. Pérez S, Mazeau K, Hervé du Penhoat C. 2000. The three-dimensional structures of the pectic polysaccharides. Plant Physiol Biochem. 38:37–55. Pérez S, Rodriguez-Carvajal MA, Doco T. 2003. A complex plant cell wall polysaccharide: Rhamnogalacturonan II. A structure in quest of function. Biochimie. 85:109–121. Pieper U, Eswar N, Webb BM, Eramian D, Kelly L, Barkan DT, Carter H, Mankoo P, Karchin R, Marti-Renom MA, et al. 2009. MODBASE, a database of annotated comparative protein structure models and associated resources. Nucleic Acids Res. 37:D347–D354. Ponti A, Gulati A, Bäcker V, Scwarb P. 2007. Huygens remote manager: A web interface for high-volume batch deconvolution. Imaging Microsc. 9:57–58. Raetz CRH. 1990. Biochemistry of endotoxins. Annu Rev Biochem. 59:129–170. Raetz RH, Whitfield C. 2002. Lipolysaccharide endotoxins. Annu Rev Biochem. 71:635–700. Rick P. 1987. Lipopolysaccharide biosynthesis. In: Neidhardt FC, editor. Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. Washington: ASM. Royo J, Gomez E, Hueros G. 2000. A maize homologue of the bacterial CMPKdo synthetases: Similar pathways operate in plants and bacteria for the activation of Kdo prior to its incorporation into cellular envelopes. J Biol Chem. 275:24993–24999. Saint-Jore CM, Evins J, Brandizzi F, Batoko H, Moore I, Hawes C. 2002. Redistribution of membrane proteins between the Golgi apparatus and endoplasmic reticulum in plants is reversible and not dependent on cytoskeletal networks. Plant J. 29:661–678. Séveno M, Bardor M, Paccalet T, Gomord V, Lerouge P, Faye L. 2004. Glycoprotein sialylation in plants? Nat Biotechnol. 22:5–6. Séveno M, Voxeur A, Rihouey C, Wu A-M, Ishii T, Chevalier C, Ralet M-C, Driouich A, Marchant A, Lerouge P. 2009. Structural characterisation of the pectic polysaccharide rhamnogalacturonan II using an acidic fingerprinting methodology. Planta. 230:947–957. Wu J, Patel MA, Sundaram AK, Woodard RW. 2004. Functional and biochemical characterization of a recombinant Arabidopsis thaliana 3-deoxy-Dmanno-octulosonate 8-phosphate synthase. Biochem J. 381:185–193. York WS, Darvill AG, McNeil M, Albersheim P. 1985. 3-desoxy-D-manno-2octulosonic acid (KDO) is a component of rhamnogalacturonan II, a pectic polysaccharide in the primary cell walls of plants. Carbohydr Res. 138:109–126.

Downloaded from http://glycob.oxfordjournals.org/ by guest on February 25, 2016

Cantarel BL, Coutinho PM, Rancurel C, Bernard T, Lombard V, Henrissat B. 2009. The Carbohydrate-Active EnZymes database (CAZy): An expert resource for Glycogenomics. Nucleic Acids Res. 37:D233–D238. Castresana J. 2000. Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Mol Biol Evol. 17:540–552. Chatre L, Brandizzi F, Hocquellet A, Hawes C, Moreau P. 2005. Sec22 and memb11 are v-SNAREs of the anterograde endoplasmic reticulum-Golgi pathway in tobacco leaf epidermal cells. Plant Physiol. 139:1244–1254. Chevenet F, Brun C, Banuls AL, Jacq B, Chisten R. 2006. TreeDyn: Towards dynamic graphics and annotations for analyses of trees. BMC Bioinformatics. 10:439. Clementz T, Raetz CR. 1991. A gene coding for 3-deoxy-D-manno-octulosonicacid transferase in Escherichia coli. Identification, mapping, cloning, and sequencing. J Biol Chem. 266:9687–9696. Corpet F. 1988. Multiple sequence alignment with hierarchical clustering. Nucleic Acids Res. 16:10881–10890. Curtis MD, Grossniklaus U. 2003. A gateway cloning vector set for highthroughput functional analysis of genes in planta. Plant Physiol. 133:462–469. Daskalova SM, Pah AR, Baluch DP, Lopez LC. 2009. The Arabidopsis thaliana putative sialyltransferase resides in the Golgi apparatus but lacks the ability to transfer sialic acid. Plant Biol. 11:284–299. Delmas F, Petit J, Joubès J, Séveno M, Paccalet T, Hernould M, Lerouge P, Mouras M, Chevalier C. 2003. The gene expression and enzyme activity of plant 3-deoxy-D-manno-2-octulosonic acid-8-phosphate synthase are preferentially associated with cell division in a cell cycle-dependent manner. Plant Physiol. 133:348–360. Delmas F, Séveno M, Northey JGB, Hernould M, Lerouge P, Mc Court P, Chevalier C. 2008. The synthesis of the rhamnogalacturonan II component 3deoxy-D-manno-2-octulosonic acid (Kdo) is required for pollen tube growth and elongation. J Exp Bot. 59:2639–2647. Dereeper A, Guignon V, Blanc G, Audic S, Buffet S, Chevenet F, Dufayard JF, Guindon S, Lefort V, Lescot M, et al. 2008. Phylogeny.fr: Robust phylogenetic analysis for the non-specialist. Nucleic Acids Res. 36: W465–W469. Dunkley TPJ, Hester S, Shadforth IP, Runions J, Weimar T, Hanton SL, Griffin JL, Bessant C, Brandizzi F, Hawes C, et al. 2006. Mapping the Arabidopsis organelle proteome. Proc Natl Acad Sci USA. 103:6518–6523. Edwards K, Johnstone C, Thompson C. 1991. A simple and rapid method for the preparation of plant genomic DNA for PCR analysis. Nucleic Acids Res. 19:1349. Ekpo P, Nano FE. 2006. Arg276 of GseA, a Chlamydiatrachomatis Kdo transferase, is required for the synthesis of the chlamydial genus-specific epitope in Escherichia coli. FEMS Microbiol Lett. 96:49–53. Emanuelsson O, Brunak S, von Heijne G, Nielsen H. 2007. Locating proteins in the cell using TargetP, SignalP, and related tools. Nat Protoc. 2:953–971. Finn RD, Tate J, Mistry J, Coggill PC, Sammut JS, Hotz HR, Ceric G, Forslund K, Eddy SR, Sonnhammer EL, et al. 2008. The Pfam protein families database. Nucleic Acids Res. 36:D281–D288. Fleisher A, O'Neill MA, Ehwald R. 1999. The pore size of non-graminaceous plant cell walls is rapidly decreased by borate ester cross-linking of the pectic polysaccharide rhamnogalacturnan II. Plant Physiol. 121:829–838. Gouet P, Robert X, Courcelle E. 2003. ESPript/ENDscript: Extracting and rendering sequence and 3D information from atomic structures of proteins. Nucleic Acids Res. 31:3320–3323. Guindon S, Gascuel O. 2003. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol. 52:696–704. Ishii T, Matsunaga T, Pellerin P, O'Neill MA, Darvill AG, Albersheim P. 1999. The plant cell wall polysaccharide rhamnogalacturonan II self-assembles into a covalently cross-linked dimer. J Biol Chem. 274:13098–13104. Ishii T, Matsunaga T, Hayashi N. 2001. Formation of rhamnogalacturonan IIborate dimer in pectin determines cell wall thickness of pumpkin tissue. Plant Physiol. 126:1698–1705. Jelakovic S, Schulz GE. 2002. Catalytic mechanism of CMP:2-keto-3-deoxymanno-octonic acid synthetase as derived from complexes with reaction educt and product. Biochemistry. 41:1174–1181.