OPEN ACCESS ARTICLE

This Article Free via Open Access: OA OA Abstract Full Text (PDF) Supplemental Data OA All Versions of this Article: 19/7/2278    most recent tpc.107.052985v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (6) Citing Articles via Google Scholar Google Scholar Articles by Strasser, R. Articles by Mach, L. Search for Related Content PubMed PubMed Citation Articles by Strasser, R. Articles by Mach, L. Agricola Articles by Strasser, R. Articles by Mach, L.

A Unique ß1,3-Galactosyltransferase Is Indispensable for the Biosynthesis of N-Glycans Containing Lewis a Structures in Arabidopsis thaliana^[W],[OA]

^a Institute of Applied Genetics and Cell Biology, BOKU, University of Natural Resources and Applied Life Sciences, A-1190 Vienna, Austria
^b Department of Chemistry, BOKU, University of Natural Resources and Applied Life Sciences, A-1190 Vienna, Austria

¹ Address correspondence to richard.strasser{at}boku.ac.at.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
In plants, the only known outer-chain elongation of complex N-glycans is the formation of Lewis a [Fuc1-4(Galß1-3)GlcNAc-R] structures. This process involves the sequential attachment of ß1,3-galactose and 1,4-fucose residues by ß1,3-galactosyltransferase and 1,4-fucosyltransferase. However, the exact mechanism underlying the formation of Lewis a epitopes in plants is poorly understood, largely because one of the involved enzymes, ß1,3-galactosyltransferase, has not yet been identified and characterized. Here, we report the identification of an Arabidopsis thaliana ß1,3-galactosyltransferase involved in the biosynthesis of the Lewis a epitope using an expression cloning strategy. Overexpression of various candidates led to the identification of a single gene (named GALACTOSYLTRANSFERASE1 [GALT1]) that increased the originally very low Lewis a epitope levels in planta. Recombinant GALT1 protein produced in insect cells was capable of transferring ß1,3-linked galactose residues to various N-glycan acceptor substrates, and subsequent treatment of the reaction products with 1,4-fucosyltransferase resulted in the generation of Lewis a structures. Furthermore, transgenic Arabidopsis plants lacking a functional GALT1 mRNA did not show any detectable amounts of Lewis a epitopes on endogenous glycoproteins. Taken together, our results demonstrate that GALT1 is both sufficient and essential for the addition of ß1,3-linked galactose residues to N-glycans and thus is required for the biosynthesis of Lewis a structures in Arabidopsis. Moreover, cell biological characterization of a transiently expressed GALT1-fluorescent protein fusion using confocal laser scanning microscopy revealed the exclusive location of GALT1 within the Golgi apparatus, which is in good agreement with the proposed physiological action of the enzyme.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
N-glycosylation is a major posttranslational modification of proteins in all eukaryotic cells. While the early processing steps of protein-bound N-linked oligosaccharides (N-glycans) that take place in the endoplasmic reticulum (ER) and Golgi apparatus are conserved between animals and plants, the later steps of N-glycan maturation differ significantly. Plant glycoproteins contain two major types of oligosaccharides: paucimannosidic and complex N-glycans (Figure 1 ) (Lerouge et al., 1998; Wilson et al., 2001b). Most of these plant N-glycans carry ß1,2-xylose and core 1,3-fucose as well as up to two terminal N-acetylglucosamine (GlcNAc) residues attached to the pentasaccharide (Man₃GlcNAc₂) core structure. ß1,2-xylose and core 1,3-fucose residues are absent in mammalian glycoproteins. Furthermore, plant glycoproteins differ from mammalian ones as they are devoid of ß1,4-linked galactose (Bakker et al., 2001b), sialic acid (Séveno et al., 2004; Zeleny et al., 2006), and further branching, which results in considerably less N-glycan heterogeneity. In plants, the only known outer-chain elongation of N-glycans is the attachment of ß1,3-galactose (Gal) and 1,4-fucose (Fuc) residues, which form a trisaccharide [Fuc1-4(Galß1-3)GlcNAc-R] known as the Lewis a (Le^a) epitope (Fitchette-Lainé et al., 1997; Melo et al., 1997) (Figure 1). The Le^a epitope was detected on glycoproteins from many different plant species (Fitchette et al., 1999; Wilson et al., 2001b). Le^a structures were found primarily on soluble and membrane-bound extracellular proteins and were reported as absent from vacuolar proteins (Fitchette-Lainé et al., 1997; Melo et al., 1997; Fitchette et al., 1999). Although Le^a structures are otherwise widespread in the plant kingdom, most studies have failed to detect Le^a containing N-glycans in Arabidopsis thaliana and other Brassicaceae (e.g., Fitchette-Lainé et al., 1997; Wilson et al., 2001b), with two exceptions where Le^a epitopes were detected by immunological means in Arabidopsis root tips and seedlings, respectively (Jin et al., 2001; Léonard et al., 2002).

View larger version (10K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 1. Plant N-Glycan Structures. Characteristic plant N-glycan structures are shown. The paucimannosidic structure MMXF (for nomenclature, see www.proglycan.com and Supplemental Figure 5 online), which lacks terminal GlcNAc residues at the nonreducing ends, and the complex N-glycan GnGnXF are the major N-glycan species identified in plants. The Lea structure (FA)(FA)XF represents the most sophisticated complex plant N-glycan structure so far.

In this study, we reinvestigated the occurrence of Le^a structures in Arabidopsis tissues and provide compelling evidence that at least certain organs contain glycoproteins with Le^a structures. Using an expression cloning strategy, we were able to identify a single gene (GALACTOSYLTRANSFERASE1 [GALT1]), out of six candidate genes, that codes for the Arabidopsis ß1,3-galactosyltransferase involved in Le^a biosynthesis. Overexpression of GALT1 in Arabidopsis increased the Le^a epitope levels in planta, whereas a knockout of the GALT1 gene or its selective downregulation by RNA interference abolished the synthesis of Le^a structures. Recombinant GALT1 produced in insect cells was used to determine the enzymatic properties of the enzyme, and a transiently expressed GALT1-green fluorescent protein (GFP) fusion was used to investigate its subcellular localization.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Expression of Le^a Epitopes in Arabidopsis
Although Le^a structures have been detected on glycoproteins from various plant species, this oligosaccharide structure was initially reported to be absent from Arabidopsis and other members of the Brassicaceae (Fitchette-Lainé et al., 1997; Rayon et al., 1999; Wilson et al., 2001b). However, in a later study, the epitope was detected in Arabidopsis seedlings by immunoblots using Le^a-specific antibodies (Léonard et al., 2002). To address these conflicting results, we performed a systematic analysis of N-glycans present in different organs of Arabidopsis. Total protein extracts were analyzed by immunoblots using JIM84 (Horsley et al., 1993), a monoclonal antibody specifically binding to Le^a structures on plant N-glycans (Fitchette et al., 1999). We found that the Le^a epitope is not detectable in leaves but is clearly present in pedicels, stems, and nodes and to a lesser extent in siliques, shoot apex, flowers, and roots (Figure 2A ). Proteins extracted from cgl1 plants, which do not produce complex N-glycans (von Schaewen et al., 1993; Strasser et al., 2005), did not show any signal, indicating that the epitope is exclusively present on N-glycans (see Supplemental Figure 2 online). To further investigate the N-glycan composition of individual Arabidopsis tissues, matrix-assisted laser-desorption ionization time of flight mass spectrometry (MALDI-TOF MS) analyses were performed using N-glycans present in total protein extracts from various organs. The mass spectra of leaf N-glycans contained three major peaks. These three peaks were assigned to the typical paucimannosidic structure MMXF and the complex-type N-glycans GnMXF/MGnXF and GnGnXF (Figure 4B; Strasser et al., 2004a). These peaks were also the predominant structures in the mass spectra of stem-derived N-glycans. Notably, stem samples contained a small but distinct peak with a mass of 2234.8 D corresponding to the Le^a-containing N-glycan structure (FA)(FA)XF and a second peak with one galactose and one outer-chain fucose residue [(FA)GnXF, mass-to-charge ratio (m/z) = 1926.7; Figure 2B]. These results are in good agreement with the immunoblot data and clearly indicate the presence of Le^a epitopes in Arabidopsis stems.

View larger version (46K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 2. Presence of Lea Epitopes in Arabidopsis. Protein gel blot analysis and mass spectrometry of N-glycans were performed to detect the Lea carbohydrate epitope in individual Arabidopsis tissues. (A) Proteins were extracted from different organs, and 12 µg of total protein were subjected to SDS-PAGE under reducing conditions. Immunoblotting was performed using JIM84 antibodies. fl, flowers; pe, pedicels; si, siliques; st, stem; n, node; sa, shoot apex; cl, cauline leaves; jl, juvenile leaves; al, adult leaves, p, petiole; sl, senescent leaves; r, roots. (B) N-glycans isolated from wild-type stems were analyzed by MALDI-TOF MS. Two Lea peaks [(FA)GnXF/Gn(FA)XF and (FA)(FA)XF] were clearly detected.

View larger version (28K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 4. Overexpression of GALT1 in Leaves. (A) Extracted leaf proteins from different transgenic lines (1 to 12) expressing 35S:GALT1 (10 µg) were subjected to SDS-PAGE under reducing conditions. Immunoblotting was performed using JIM84 antibody. Wild-type plants and 35S:Mm-B3GALT1 (Mm) plants were used as controls. (B) MALDI-TOF MS spectra of N-glycans extracted from wild-type and 35S:GALT1 (line 12) leaves. The peaks containing Lea structures are marked, and their masses are shown in the table.

View larger version (23K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 3. Phylogenetic and Expression Analysis of Putative Arabidopsis GALTs. (A) Phylogenetic analysis of putative Arabidopsis GALT proteins was conducted using MEGA version 3.1. Amino acid sequences of the 20 pfam 01762-containing Arabidopsis proteins from CAZy glycosyltransferase family GT31 and three mammalian B3GALTs (Mm-B3GALT1, UniProt accession number O54904; Hs-B3GALT2, O43825; and Hs-B3GALT5, Q9BYG0) were aligned using ClustalW. The phylogenetic tree was constructed with the neighbor-joining method, and the bootstrap values were determined from 1000 trials. The percentages of bootstrap values for the respected branches are shown. The members of subfamily 1 are highlighted in gray. Bar = 20% divergence. (B) Expression analysis of GALT candidate genes by RT-PCR. RT-PCR was performed with specific primers for the respective genes from subfamily 1 or ubiquitin (UBQ5) and core 1,3-fucosyltransferase (FucTB) as controls (35 PCR cycles for all reactions). Siliques (si), stems (st), cauline leaves (cl), and juvenile leaves (jl) were analyzed. All RT-PCR reactions were performed twice with similar results.

Based on the data from the Le^a immunoblots, we reasoned that the expression of the responsible GALT gene(s) might be upregulated in tissues showing a strong Le^a signal. To investigate this, the transcript levels of the six candidate genes were monitored in siliques, stems, and two types of leaves using RT-PCR. All six putative GALT mRNAs were detected in the four organs (Figure 3B), although at various levels. Two of them, At1g26810 and At4g21060, displayed higher transcript levels in stems than in leaves. The difference in transcript levels between stems and leaves was most pronounced for At1g26810. Hence, the open reading frame of this GALT candidate gene was expressed in Arabidopsis under the control of the cauliflower mosaic virus 35S promoter, and leaves of these transgenic plants were screened with JIM84 for the presence of the Le^a epitope. As a control for this expression cloning strategy, additional transgenic Arabidopsis plants carrying either mouse ß1,3-galactosyltransferase 1 (Mm-B3GALT1) or Arabidopsis FUT13 cDNAs were generated. For all transgenic plants, integration of the heterologous construct was confirmed by PCR from genomic DNA, and its transcription verified by RT-PCR. To investigate the presence or absence of Le^a structures in the transgenic lines, total protein extracts from leaves were immunoblotted with JIM84 antibodies. While the 35S:FUT13 plants did not show any increase in the intensity of the Le^a signal, plants expressing the 35S:Mm-B3GALT1 construct clearly did (see Supplemental Figure 3 online). This indicates that acceptor and donor substrates for the formation of Le^a epitopes are available in leaves and demonstrates that the endogenous activity of FUT13 is not limiting. Interestingly, At1g26810 overexpression resulted in a strong signal (Figure 4A ) comparable to the one found in 35S:Mm-B3GALT1 plants. By contrast, expression of At1g27120, At1g74800, At3g06440, At4g21060, and At5g62620 cDNAs did not lead to any increase of JIM84 staining, indicating that these proteins are not involved in the formation of Le^a epitopes (see Supplemental Figure 3 online). In contrast with the N-glycan spectrum of leaves from untransformed plants, MALDI-TOF MS analysis of total N-glycans isolated from leaves of the transgenic 35S:At1g26810 plants showed the presence of peaks representing N-glycans with Le^a structures (Figure 4B). Hence, At1g26810 encodes a ß1,3-galactosyltransferase (GALT1) capable of mediating the formation of Le^a epitopes in Arabidopsis. We therefore decided to focus on At1g26810 and named this gene GALT1.

The sequence of the GALT1 open reading frame has a size of 1932 bp and codes for a protein of 643 amino acids, with a calculated molecular mass of 72.3 kD and a theoretical isoelectric point of 5.0. Apart from a silent base pair change (at position 1206: "G" instead of "A"), the isolated cDNA is identical to the seven exons of the annotated genomic sequence. The 5'-rapid amplification of cDNA ends (5'-RACE) revealed a 5'-untranslated region of 248 bp without any additional translation initiation start site. A hydrophobic sequence is located at the N terminus (Figure 5 ), which suggests that GALT1 has a type II membrane protein topology with a single transmembrane domain and a short cytoplasmic tail as typical for Golgi-located glycosyltransferases. An alignment of the galactosyltransferase domain sequences of GALT1 and mammalian B3GALTs reveals a number of highly conserved residues putatively involved in substrate binding and catalysis (see Supplemental Figure 4 online).

View larger version (86K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 5. GALT1 Amino Acid Sequence. The putative transmembrane domain (amino acids 6 to 23) predicted by HmmTOP_V2 (http://www.enzim.hu/hmmtop/index.html) is highlighted in gray, and putative N-glycosylation sites are underlined. The galactoside binding lectin domain (pfam 00337) is shown in italics, and the galactosyltransferase domain (pfam 01762) is shown in bold. The conserved DXD motif is shaded black, and the asterisk marks the beginning of the fragment that was expressed in insect cells as a soluble secreted form.

Heterologous Expression of GALT1 in Insect Cells
A soluble secreted form of GALT1 was produced by baculovirus-mediated infection of Spodoptera frugiperda Sf21 insect cells, and recombinant GALT1 protein was purified from culture supernatants by means of nickel-chelate affinity chromatography. Immunoblotting with antibodies to the enterokinase cleavage site present in the recombinant fusion protein revealed a major immunoreactive band of 80 kD (Figure 6A ), which is in good agreement with its theoretical molecular mass (73.2 kD), considering that the protein contains seven potential N-glycosylation sites. PNGase F digestion led to a 73-kD band, indicating that recombinant GALT1 is indeed N-glycosylated.

View larger version (18K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 6. Recombinant GALT1 Modifies Glycopeptide and N-Glycan Substrates with ß1,3-Linked Galactose Residues. (A) Immunoblot of purified recombinant GALT1, with (+) and without (–) PNGase F digestion. (B) Dabsylated GnGn glycopeptide (m/z = 2061) was incubated with purified recombinant GALT1 in the presence of the donor substrate UDP-galactose and analyzed by MALDI-TOF MS. In the bottom panel, the acceptor substrate (S) is shown as a control. Asterisks mark substrate fragmentation peaks. (C) GnGn-PA was incubated either with recombinant GALT1 or recombinant Mm-B3GALT1 and UDP-galactose. The reaction products of GALT1 coeluted with the Mm-B3GALT1 products. (D) Reaction products from the GALT1 assay shown in (C) were incubated with different ß-galactosidases prior to rechromatography. X.m. galactosidase, ß1,3-galactosidase from X. manihotis; A.o. galactosidase, A. oryzae ß1,4-/ß1,6-specific galactosidase.

Characterization of Galactosyltransferase Activity
To make a detailed analysis of the linkage that is created by GALT1, activity assays were performed with the oligosaccharide substrate GnGn-PA. The samples were subsequently subjected to HPLC analysis using a novel graphitized fractionation matrix to obtain clean separation of substrate and product peaks. Three product peaks eluting clearly after the substrate GnGn-PA were detected (Figure 6C). The peaks from the GALT1 sample coeluted with the ones from a control reaction (A³Gn-, GnA³-, and A³A³-PA) using recombinant Mm-B3GALT1. In addition, digestion of the GALT1 products with Xanthomonas manihotis ß1,3-galactosidase led to a shift of the corresponding peaks to the elution position of the GnGn-PA substrate (Figure 6D). By contrast, incubation with Aspergillus oryzae ß-galactosidase, which is known to remove ß1,4-linked and ß1,6-linked galactose (Zeleny et al., 1997), did not result in any detectable shift of peaks. These data demonstrate that GALT1 is a ß1,3-galactosyltransferase acting on N-glycan substrates. The specific activity with GnGn-PA as substrate was 3.1 ± 0.3 pmol/min/mg of GALT1 protein. No galactosyltransferase activity was detected in the absence of manganese ions or in the presence of EDTA (see Supplemental Figure 6 online), which indicates that GALT1 is, like the mammalian B3GALTs, a strictly metal ion–dependent enzyme, in agreement with the presence of a DXD motif (Figure 5) presumably involved in binding of the metal ion and the donor substrate.

To see whether the product of GALT1 action is indeed a substrate for FUT13, dabsylated GnGn-peptide (m/z = 2061) was sequentially incubated with GALT1 and FUT13 in the presence of the respective nucleotide sugar donor substrates, UDP-galactose and GDP-fucose. For this purpose, FUT13 was also expressed as a secreted soluble form in insect cells and purified using nickel-chelate affinity chromatography. MALDI-TOF MS analysis of the reaction products revealed the presence of peaks with masses indicative of the addition of either one [(FA)Gn/Gn(FA)] or two [(FA)(FA)] galactose and fucose units (Figure 7A ). To show that GALT1 can act on glycoprotein substrates, a glycovariant of human transferrin lacking sialic acid and galactose residues (GnGn-Tf) was sequentially incubated with GALT1 and FUT13 and then subjected to immunoblotting with JIM84. GnGn-Tf thus treated reacted with anti-Le^a antibodies (Figure 7B), while control samples did not show a transferrin signal. Interestingly, apart from the transferrin band (70 kD), another major band of 80 kD was visible in samples incubated with FUT13. This band was identified as GALT1 itself. Since endogenous and heterologous glycoproteins synthesized by Sf21 cells usually lack N-glycans with terminal galactose residues (e.g., Kubelka et al., 1994; Bencúr et al., 2005), we concluded that GALT1 is capable of autogalactosylation during its production in these cells. In Figure 7B, the autogalactosylation of GALT1 was only detectable upon further incubation with recombinant FUT13 in the presence of the donor substrate GDP-fucose. The corresponding band in lanes 2 and 3 did not give a signal because GDP-fucose (lane 2, left panel) and GDP-fucose/FUT13 (lane 3, left panel) were missing in the assay.

View larger version (27K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 7. Generation of the Lea Epitope on Glycopeptide and Glycoprotein Substrates. (A) Dabsylated GnGn-glycopeptide (m/z = 2061) was incubated with recombinant GALT1 in the presence of UDP-galactose and analyzed by MALDI-TOF MS (bottom panel). Subsequently, the product was incubated with recombinant FUT13 and GDP-fucose and analyzed by MALDI-TOF MS (top panel), which resulted in the detection of the (FA)(FA) peak (mass: 2677 D). Asterisks mark substrate and product fragmentation peaks. (B) GnGn-transferrin (GnGn-Tf) was incubated first with recombinant GALT1 and UDP-galactose and then with recombinant FUT13 and GDP-fucose. As controls the acceptor substrate (GnGn-Tf), the donor substrates (UDP-galactose and GDP-fucose), or the glycosyltransferases (GALT1 and FUT13) were omitted in different combinations. The samples were subjected to SDS-PAGE under reducing conditions, and after blotting, Lea epitopes were detected using JIM84 antibodies. Tf, transferrin; Tf*, ß1,3-galactosylated contaminant present in the transferrin sample used in this experiment. Note that GALT1 itself serves also as a substrate for FUT13.

View larger version (29K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 8. GALT1 Is Capable of in Vivo Autogalactosylation. (A) Recombinant GALT1 (5 µg) was incubated (16 h, 37°C) in 30 µL of 50 mM sodium citrate buffer, pH 4.5, either in the absence (lane 1) or presence of 42 milliunits (mU) of ß1,3-galactosidase (lane 2) or 150 mU of ß1,4-/ß1,6-galactosidase (lane 3). Aliquots (2.5 µg GALT1) were then treated (16 h, 30°C) with recombinant FUT13 (5 µg) in 50 µL of 100 mM MES buffer, pH 7.0, containing 0.5 mM GDP-fucose. The samples were then subjected to immunoblotting analysis with antibodies to the Lea epitope (JIM84) or to the enterokinase cleavage site present in recombinant GALT1 (anti-EK). All samples contain a GALT1 fragment (GALT1*) that fails to react with anti-EK antibodies. (B) Recombinant GALT1 (5 µg) was excised from a Coomassie blue–stained SDS-PAGE gel and subjected to tryptic digestion and subsequent MS analysis to identify glycopeptides with terminal galactose residues. To distinguish between the isobaric structures Man4GnF and ß1,3-galactosylated AMF/MAF, a control sample (right panel) was digested with X. manihotis ß1,3-galactosidase (X.m.). This confirmed that the peak with the mass of 2672.1 D consists mainly of the ß1,3-galactosylated structure AMF/MAF, with the ß1,3-galactosidase–insensitive fraction probably representing Man4GnF. The relevant part of the mass spectrum for glycopeptide T279SSTSLQSNTSR290 is shown. The complete mass spectra of this and other GALT1 glycopeptides are included in Supplemental Figure 7 online.

View larger version (23K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 9. GALT1 Uses GnGnXF as a Substrate. (A) The typical complex plant N-glycan substrate GnGnXF-PA was incubated with GALT1 and UDP-galactose to show that this ß1,2-xylose and core 1,3-fucose containing N-glycan can act as an acceptor structure for GALT1. Three peaks were detected (GnA3XF-PA, A3GnXF-PA, and A3A3XF-PA), which coeluted with standards obtained by incubation of the substrate with Mm-B3GALT1. (B) The reaction products of the GALT1 assay were further modified by incubation with recombinant FUT13 and GDP-fucose. As shown by LC-MS analysis, this led to the formation of N-glycans carrying both ß1,3-linked galactose and 1,4-linked fucose [i.e., (AF)GnXF/Gn(AF)XF and (AF)(AF)XF].

View larger version (23K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 10. Analysis of RNAi and Knockout Plants. (A) The structure of the GALT1 gene and the position of the T-DNA insertion are shown. (B) PCR screening of the T-DNA knockout line (SAIL_170_A08) resulted in the identification of plants homozygous for the T-DNA insertion (galt1-1). Sequencing of the PCR products obtained with one gene-specific primer (At1g26810-1F) and one T-DNA–specific primer (LBsail2, T-DNA 1) or At1g26810-3R and LBsail2 (T-DNA 2) confirmed the insertion and revealed a deletion of 36 bp in intron 4 of GALT1. RT-PCR was performed on RNA extracted from galt1-1 plants with primers At1g26810-1F and At1g26810-2R. UBQ5 served as a control. (C) Schematic presentation of the GALT1 RNAi construct. (D) RT-PCR of a galt1RNAi line using RNA extracted from stems with GALT1-specific primers (AT1g26810-1F/-2R). FucTB served as a control. (E) Protein gel blot in which proteins were extracted from stems, and 10 µg were subjected to SDS-PAGE under reducing conditions. Detection was performed either with JIM84 or with anti-HRP antibodies, which recognize ß1,2-xylose and core 1,3-fucose residues on N-glycans, as a loading control. While the Lea epitope is present in wild-type plants (lane 1) and in the 35S:GALT1 line (lane 3), it is not detectable in transgenic galt1-1 (lane 2) and galt1RNAi (lane 4) plants.

Subcellular Localization of GALT1
The Golgi apparatus is not only a central sorting point within the secretory pathway. This compartment also plays a central biosynthetic role in processing of complex carbohydrate structures. The Golgi is the site of maturation of glycoprotein-bound N-glycans, and the subcellular location of many N-glycan processing enzymes has been determined to be in the Golgi (Dirnberger et al., 2002; Saint-Jore-Dupas et al., 2006; Strasser et al., 2006). An indication that GALT1 is also located in this subcellular compartment was provided by the observation that Le^a structures have been detected in the Golgi by immunostaining using JIM84 (Horsley et al., 1993; Fitchette et al., 1999). To investigate the subcellular location of GALT1, a GFP fusion construct was generated. This fusion protein encompasses the cytoplasmic tail, the predicted transmembrane domain, and the putative stem region (CTS) of GALT1. For many Golgi N-glycosylation enzymes, it has been demonstrated that the CTS region is sufficient to retain a reporter molecule in the plant Golgi (Essl et al., 1999; Dirnberger et al., 2002; Saint-Jore-Dupas et al., 2006; Strasser et al., 2006). GALT1-GFP was transiently expressed in Nicotiana benthamiana leaf epidermal cells, and confocal laser scanning microscopy revealed fluorescence in small mobile vesicles similar as previously found for other Golgi-located enzymes (Figures 11A and 11D) (e.g., Strasser et al., 2006). To further investigate whether this labeling corresponds to Golgi structures, GALT1-GFP was coexpressed with a Golgi marker construct consisting of the CTS region of rat 2,6-sialyltransferase fused to red fluorescent protein (ST-mRFP; Runions et al., 2006). Both GALT1-GFP and ST-mRFP accumulated in the Golgi and exhibited significant colocalization, since 98% of the GALT1-GFP–containing structures were also positive for ST-mRFP (Figures 11A to 11C). Since it has been suggested that some N-glycosylation enzymes are located in the Golgi as well as in the ER (Saint-Jore-Dupas et al., 2006), GALT1-GFP was coexpressed with the ER marker YFP-HDEL (Figure 11E; Irons et al., 2003). No significant colocalization of these two fluorescent proteins was observed (Figure 11F). This indicates that GALT1 is exclusively located in the Golgi apparatus, which is consistent with the proposed physiological action of the enzyme.

View larger version (41K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 11. GALT1-GFP Accumulates in the Golgi. N. benthamiana leaf epidermal cells transiently expressing GALT1-GFP ([A] and [D]), ST-mRFP (B), and YFP-HDEL (E) and the corresponding overlays of the images ([C] and [F]). The punctate fluorescence of GALT1-GFP and its colocalization with ST-mRFP strongly indicates that GALT1-GFP accumulates in the Golgi, which is in good agreement with its proposed physiological action. Bars = 10 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

GALT1: a Novel ß1,3-Galactosyltransferase with a Unique Architecture
GALT1 is classified as a member of family GT31 in the CAZy database, which contains 20 Arabidopsis sequences with significant similarity to mammalian B3GALTs. Phylogenetic analysis indicates that GALT1 belongs to a subfamily (subfamily 1) of six Arabidopsis proteins. However, no information is as yet available on the physiological functions of any other member of subfamily 1. The same applies to the other 14 B3GALT-like Arabidopsis proteins present in CAZy family GT31. None of them seems to be involved in N-glycan maturation. However, it can be envisaged that some of these B3GALT candidates participate in processes such as the biosynthesis of arabinogalactan proteins (Tan et al., 2004) or other cell wall components (Reid, 2000).

In addition to the galactosyltransferase domain, all members of subfamily 1 contain a putative galactoside binding lectin domain with similarity to human galectins (Ackerman et al., 2002). Interestingly, such a lectin domain is not present in mammalian B3GALTs (Hennet, 2002). However, a C-terminal lectin domain is present in various mammalian polypeptide GalNAc-transferases (Hassan et al., 2000). For the latter enzymes, it has been suggested that binding of the lectin domain to GalNAc-modified substrates serves to modulate their catalytic properties. However, the domain organization of subfamily 1 members with an N-terminal lectin domain and a C-terminal catalytic domain differs from that of polypeptide GalNAc-transferases. Hence, the function of the putative lectin domain of GALT1 remains to be determined.

Le^a: A Tissue-Specific Structural Element of Complex Arabidopsis N-Glycans
It has been previously reported that glycoproteins from Arabidopsis and other members of the Brassicaceae lack Le^a structures (Fitchette et al., 1999; Rayon et al., 1999; Wilson et al., 2001b). In this study, we have clearly shown that Arabidopsis does contain glycoproteins with Le^a moieties, which is in good agreement with results obtained by Jin et al. (2001) and Léonard et al. (2002), reporting the presence of Le^a epitopes in Arabidopsis root tips and seedlings, respectively. Furthermore, our analyses revealed that Le^a structures are organ-specifically expressed in Arabidopsis, with undetectable amounts in mature rosette leaves and relatively high levels in stems. The tissue specificity of this modification, an unprecedented observation for plant N-glycans, accounts for previous failures to detect the presence of Le^a-containing N-glycans in Arabidopsis, since most of the earlier studies were performed on leaves (Fitchette et al., 1999; Rayon et al., 1999).

Two different enzymes are involved in the biosynthesis of Le^a structures, GALT1 and FUT13. It has been speculated earlier that the scarcity of Le^a structures in Arabidopsis might be due to the absence of a functional FUT13 protein (Wilson et al., 2001c). However, this assumption has been discarded with the identification of an enzymatically active Arabidopsis FUT13 (Léonard et al., 2002). This is in good agreement with our finding that overexpression of FUT13 does not lead to elevated Le^a levels in Arabidopsis leaves, indicating that this enzyme is not the limiting factor in Le^a biosynthesis in this organ. Our data indicate that GALT1 rather than FUT13 is the rate-limiting enzyme. This is supported by the following lines of evidence: (1) GALT1 is expressed in a tissue-specific manner, and its expression levels correlate with the Le^a levels of the respective organs; (2) overexpression of GALT1 increases the occurrence of Le^a epitopes in tissues with otherwise low Le^a contents (e.g., leaves); (3) the catalytic proficiency of GALT1, at least in vitro, is rather low compared with other plant glycosyltransferases (Bencúr et al., 2005; Strasser et al., 2005); (4) while the activities of most other glycosyltransferases involved in N-glycan processing can be readily measured in plant extracts (e.g., Staudacher et al., 1995; Palma et al., 2001; Léonard et al., 2002; Strasser et al., 2004b), direct detection of endogenous GALT1 activity has not yet been achieved.

Even in 35S:GALT1 organs with relatively high Le^a levels (e.g., stems), only a small fraction of N-glycans carries this modification. This could be simply due to the rather low specific activity of GALT1. However, overexpression of its far more active mouse counterpart Mm-B3GALT1 (Hennet et al., 1998; Zhou et al., 1999) led to similar low Le^a levels as found in 35S:GALT1 plants. This suggests that GALT1 is not the sole factor limiting the occurrence of Le^a epitopes on plant glycoproteins. Possible other contributing factors could be the in situ supply with the GALT1 donor substrate UDP-galactose (Seifert, 2004) or the limited availability of appropriate acceptor substrates (i.e., N-glycans with terminal GlcNAc residues). Given the complex architecture of the GALT1 protein, it is also possible that structural constraints permit the enzyme to act only on a small subset of endogenous glycoproteins. This is substantiated by the fact that until now only few purified plant glycoproteins have been found to carry Le^a epitopes (Fitchette-Lainé et al., 1997; Melo et al., 1997; Kimura et al., 2005).

It is of interest that N-glycans with free terminal Galß1-3GlcNAc moieties occur only in very small quantities in plants (Wilson et al., 2001b), including our 35S:GALT1 lines. This could be due to one of the following scenarios: (1) in situ modification of nascent Galß1-3GlcNAc structures by FUT13 is quick and quantitative; or (2) unmodified Galß1-3GlcNAc moieties are rapidly degraded due to the action of ß-galactosidases (transient galactose), in analogy to the transient modification of plant N-glycans with terminal GlcNAc residues. In the latter case, GlcNAc residues added in the Golgi are quickly removed upon arrival of the respective glycoprotein in vacuolar compartments (Vitale and Chrispeels, 1984). A similar scenario could be envisaged for the removal of galactose residues from N-glycans carrying terminal Galß1-3GlcNAc moieties.

Subcellular Localization of GALT1
JIM84 binds specifically to the Le^a epitope and has been used for many years as a Golgi marker (Horsley et al., 1993; Fitchette et al., 1999). This epitope has been detected in the Golgi apparatus of Arabidopsis cells by immunofluorescence and immunogold staining (Jin et al., 2001; Léonard et al., 2002), suggesting that the enzymes responsible for its formation are also located in this subcellular compartment. We have now obtained good evidence that GALT1 resides within the Golgi apparatus. A fusion protein of the 60 N-terminal amino acids of GALT1 and GFP displayed significant colocalization with a Golgi marker. This indicates that the N-terminal part of GALT1 consisting of its cytoplasmic tail (C), the single transmembrane domain (T), and a putative stem region (S) is sufficient for the proper subcellular localization of GALT1. The pivotal role of the CTS region for Golgi targeting and retention has been previously demonstrated for other plant N-glycosylation enzymes with a similar type II transmembrane protein topology (Essl et al., 1999; Dirnberger et al., 2002; Saint-Jore-Dupas et al., 2006; Strasser et al., 2006). Thus, our findings correspond well with previous studies and with the proposed physiological action of the enzyme.

Properties of GALT1-Deficient Plants
Surprisingly little is still known about the function of N-glycans in plants. In mammals, complex N-glycans regulate a diversity of biological processes by modulating receptor–ligand interactions, for example, during embryonic development (Haltiwanger and Lowe, 2004). In Arabidopsis, deficiencies in ER-resident N-glycosylation enzymes cause strong developmental phenotypes (Boisson et al., 2001; Burn et al., 2002; Gillmor et al., 2002), which are probably related to the quality control function of N-glycans during protein folding in this compartment. By contrast, Arabidopsis mutants that completely lack complex N-glycans and thus also Le^a structures are viable and display no obvious phenotype under standard growth conditions (von Schaewen et al., 1993). Hence, it comes at no surprise that Arabidopsis plants with an inactivated GALT1 gene are normal with respect to morphology and growth rate. Similar results have been previously reported for other Golgi-resident enzymes acting at late stages during N-glycan maturation (Strasser et al., 2004a, 2006). However, it can be envisaged that plants with altered complex N-glycosylation might display conditional phenotypes, for example, under certain biotic or abiotic stress conditions. Such conditional phenotypes have been recently described for Caenorhabditis elegans mutants with modified complex N-glycans (Shi et al., 2006). The GALT1-deficient Arabidopsis lines are valuable tools to establish the impact of Le^a-containing N-glycans on the response of plants to various stress conditions and thus will help to elucidate the still unknown physiological function(s) of these oligosaccharides.

	METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Plant Materials and Growth Conditions
Wild-type plants of Arabidopsis thaliana (ecotype Columbia [Col-0]) and transgenic plants were grown as described previously (Strasser et al., 2004a). Nicotiana benthamiana plants were grown in a growth chamber at 22°C with a 16-h-light/8-h-dark photoperiod. Arabidopsis cgl1 seeds were obtained from the Nottingham Arabidopsis Stock Centre (ID N6192; http://arabidopsis.info/).

Immunoblots from Arabidopsis Organs
Plant material was ground in liquid nitrogen, resuspended in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 2 mM KH₂PO₄, pH 7.4) supplemented with 1% (v/v) Triton X-100, and cleared by centrifugation (two times for 3 min at 16,000g). The protein content was determined in the supernatant using a BCA protein assay kit (Pierce) and BSA as a standard. Ten to twelve micrograms of protein were subjected to 12% SDS-PAGE under reducing conditions and blotted onto Hybond Enhanced Chemiluminescence nitrocellulose membranes (GE Healthcare). The blot was blocked in PBS containing 0.1% (v/v) Tween 20 and 3% (w/v) BSA for 1 h and incubated with a 1:40 dilution of JIM84 antibody (kindly provided by Paul Knox, University of Leeds, UK). The detection was performed after incubation in a 1:20,000 dilution of a horseradish peroxidase–conjugated goat anti-rat-IgM (µ-chain) antibody (Jackson) with Supersignal West Pico Chemiluminescent Substrate (Pierce).

Total N-Glycan Analysis
Total N-glycan analysis was performed from 500 mg of leaves as described previously (Wilson et al., 2001b; Strasser et al., 2004a). For stems and other organs, N-glycans were extracted from 20 mg of plant material by a modified version of the in-gel release method (Kolarich and Altmann, 2000) as described recently (Léonard et al., 2005b).

RT-PCR
Total RNA was purified from stems, siliques, cauline, and rosette leaves of Arabidopsis wild-type plants using a SV total RNA isolation kit (Promega). First-strand cDNA was synthesized from 500 ng of total RNA at 42°C using oligo(dT) primers and AMV reverse transcriptase (Promega). Semiquantitative RT-PCR was performed using UBQ5-D/-U primers for ubiquitin and core 1,3-fucosyltransferase B (FucTB, At1g49710) FTB-9/-10 oligos as a control and the following GALT-specific primers: At1g26810-1F/-2R, At1g27120-1/-2, At1g74800-1/-2, At3g06440-1/-2, At4g21060-1/-2, and At5g62620-1/-2.

Ectopic Expression of cDNAs in Arabidopsis
cDNA for At1g27120 and At1g74800 was obtained from RIKEN (clones pda11283 and pda08691, respectively; http://rarge.gsc.riken.jp/cdna/cdna_keyword.pl). The coding region was amplified with primers At1g27120-7/-8 and At1g74800-4F/-5R using Turbo Pfu polymerase (Stratagene) and inserted into the XbaI site of plant expression vector pPT2 (Strasser et al., 2005). In pPT2, the expression of the DNA sequences is under the control of the cauliflower mosaic virus 35S promoter. The coding region of At1g26810 (GALT1) was amplified with BD Advantage 2 Polymerase (BD Biosciences) from cDNA obtained from siliques using the primers At1g26810-5/-6 and subcloned into pGEM-T vector (Promega). The GALT1 coding region was subsequently cloned into KpnI-BamHI sites of pPT2-gal, which is a derivative of pPT2 containing an additional KpnI cloning site. At3g06440 was amplified from leaf cDNA using primers At3g06440-3/-4 and BD Advantage 2 Polymerase and subcloned into pGEM-T. The coding region was excised with BamHI-KpnI and ligated into pPT2-gal. The coding region of At5g62620 was amplified from cDNA of siliques using Turbo Pfu polymerase with primers At5g62620-5/-6 and cloned into the XbaI site of pPT2. We attempted unsuccessfully to amplify the At4g21060 open reading frame using oligos based on the deposited cDNA sequence (NM_118224). Hence, we performed 5'-RACE using a Smart RACE kit (BD Biosciences) and primer At4g21060-8R to obtain additional information about the At4g21060 cDNA. We found that the 5'-end of the At4g21060 cDNA is different from the one deposited in the database. We designed oligo At4g21060-14F based on the new 5'-end, and in combination with oligo At4g21060-10R, we could amplify the open reading frame (EF428438), which was cloned into the XbaI site of pPT2. The construct for Mm-B3GALT1 expression was obtained by amplification of the Mm-B3GALT1 coding region from mouse genomic DNA using oligos MGAL-F/-R and ligation into the XbaI-BamHI sites of pPT2. For expression of FUT13, the open reading frame was amplified from leaf cDNA with primers AthFTC-7F/-2R and subcloned using a ZERO Blunt TOPO PCR cloning kit. The coding region was removed by XbaI-BglII digestion and inserted into the XbaI-BamHI sites of vector pPT2. All inserts were sequenced using a BIG Dye Termination Cycle sequencing kit (Applied Biosystems) to rule out the presence of mutations in the amplified sequences.

To obtain the 5'- and 3'-untranslated regions of GALT1 mRNA, 5'- and 3'-RACE reactions were performed using a Smart RACE kit and reverse primers At1g26810-18R and At1g26810-3R (5'-RACE) and forward primer At1g26810-19F (3'-RACE).

Generation and Screening of Overexpression Lines
The transgenic expression lines (35S:At1g27120, 35S:At1g74800, 35S:At3g06440, 35S:At4g21060, 35S:At5g62620, 35S:GALT1, 35S:MmB3GALT1, and 35S:FUT13) were generated by floral dipping of Col-0 wild-type plants with agrobacteria containing the pPT2 constructs as described (Strasser et al., 2005). Selection of transgenic lines was performed on basal Murashige and Skoog medium containing kanamycin. Positive seedlings were transferred to soil and screened by PCR for the presence of the expression construct and by RT-PCR using a gene-specific forward primer and the universal reverse primer pPT2-1 binding to the untranslated 3'-region of the RNA. For each line, 10 to 25 individuals were then subjected to protein gel blotting analysis using JIM84.

Expression of GALT1, Mm-B3GALT1, and FUT13 in Insect Cells
The N-terminal deletion construct of GALT1 containing amino acids 31 to 643 was generated by PCR using primers 15F1-Pst/15R1-Kpn. The PCR product was cleaved with PstI and KpnI restriction enzymes and ligated into pVTBacHis1 baculovirus transfer vector, digested with the same enzymes. In this construct, the truncated GALT1 protein is placed downstream of the melittin signal peptide, a 6xHis tag, and an enterokinase cleavage site (Bencúr et al., 2005). Fragments encoding N-terminally truncated versions of Mm-B3GALT1 (amino acids 34 to 326) and FUT13 (amino acids 32 to 401) were generated and subcloned essentially as above using primers MGAL-4F/-5R (Mm-B3GALT1) and primers AthFTC-8F/-11R (FUT13, At1g71990). Expression in Spodoptera frugiperda Sf21 cells, purification and quantification of recombinant proteins, and enzymatic deglycosylation of GALT1 using PNGase F (peptide N-glycosidase F; Roche) were performed using previously published procedures (Bencúr et al., 2005; Strasser et al., 2006) with minor modifications. Recombinant GALT1 was detected on immunoblots with a mouse monoclonal antibody raised to the enterokinase recognition sequence (anti-EK; Invitrogen).

GALT1 Activity Assays
For GALT1 enzyme activity assays, a dabsylated GnGn-glycopeptide (Altmann et al., 1995) with N-glycans carrying two nonreducing terminal GlcNAc residues was used as a substrate. The standard reaction mixture consisted of 10 µL (0.55 to 0.98 mg/mL) purified GALT1, 50 µM acceptor substrate, 1 mM UDP-galactose, 10 mM MnCl₂, and 0.1 M MES buffer, pH 6.8, in a total volume of 20 µL. After incubation for 20 h at 30°C, the glycopeptides were purified from the assay mixture using 10 mg of C18 reversed phase matrix previously equilibrated with 5% (v/v) acetonitrile and eluted with 40% (v/v) acetonitrile plus 10% (v/v) isopropanol before analysis by MALDI-TOF MS. For assays with pyridylaminated (PA) substrates, purified GALT1 was incubated with 5 µM GnGn-PA or GnGnXF-PA as acceptor substrate along with 1 mM UDP-galactose and 10 mM MnCl₂ in 0.1 M MES buffer, pH 6.8, for 16 h at 30°C. Samples were analyzed by HPLC using a porous graphitic carbon column (Thermo Hypersil Keystone-HYPERCARB) with the dimensions 150 x 3 mm and a particle size of 5 µm. Samples were eluted with 0.3% (v/v) formic acid (titrated to pH 3.0 with NaOH) as buffer A and 95% (v/v) acetonitrile in water as buffer B at a flow rate of 0.8 mL/min. Gradients (35 min) from 10 to 20% and 15 to 25% buffer B were applied for separation of A³Gn/GnA³ and A³GnXF/GnA³XF isomers, respectively. Control assays with Mm-B3GALT1 were performed as described for GALT1. The identity of the Mm-B3GALT1 product peaks was verified by LC-MS analysis and by sequential exoglycosidase digestion using jack bean (Canavalia ensiformis) ß-hexosaminidase (Sigma-Aldrich) and Xanthomonas manihotis ß1,3-galactosidase (NEB) followed by rechromatography on a C18 reversed phase column as previously described (Kubelka et al., 1994).

Activity Assays Using Transferrin
Human transferrin (Sigma-Aldrich) was converted to the GnGn-glycoform (GnGn-Tf) by neuraminidase (Sigma-Aldrich; from Clostridium perfringens) and Aspergillus oryzae ß-galactosidase (Zeleny et al., 1997) treatment as described (Bencúrová et al., 2004). Two micrograms of GnGn-Tf was first incubated overnight at 30°C in 12.5 mM MnCl₂, 125 mM MES buffer, pH 7.0, and 9.8 µg of purified recombinant GALT1 in the presence or absence of 0.5 mM UDP-galactose in a total volume of 20 µL. Subsequently, samples were incubated with 2.7 µg of FUT13 in 10 mM MnCl₂, 0.1 M MES buffer, pH 7.0, and 0.5 mM GDP-fucose overnight at 30°C in a total volume of 25 µL. Modified Tf proteins (0.1 µg) were then subjected to 12% SDS-PAGE under reducing conditions, blotted onto a nitrocellulose membrane, and incubated with JIM84 antibody as described above.

LC-MS Analysis of Tryptic Glycopeptides
Purified recombinant GALT1 (5 µg) from insect cells was separated by SDS-PAGE (10% polyacrylamide) under reducing conditions, and polypeptides were detected by Coomassie Brilliant Blue staining. The 80-kD band was excised from the gel, destained, carbamidomethylated, in-gel trypsin digested, and analyzed as described (Kolarich and Altmann, 2000; Kolarich et al., 2006).

Digestions with ß-Galactosidases
Degalactosylation experiments were performed with two different ß-galactosidases. One ß-galactosidase was purified from A. oryzae and found to selectively remove ß1,4- and ß1,6-linked galactose residues (Zeleny et al., 1997). The other enzyme used was the commercially available recombinant ß1,3-galactosidase from X. manihotis. All digestions were done in 50 mM sodium citrate, pH 4.5, for 16 h at 37°C, using either 14 to 42 mU of ß1,3-galactosidase or 150 mU of ß1,4-/ß1,6-galactosidase. One milliunit is the amount of enzyme causing the liberation of 1 nmol of 4-nitrophenol per minute using 5 mM pNP-ß-Gal (Sigma-Aldrich) as substrate.

Generation and Screening of T-DNA Insertion and RNAi Lines
A GALT1 T-DNA insertion line (SAIL_170_A08) was obtained from the Nottingham Arabidopsis Stock Centre, and lines homozygous for the insertion were identified by PCR using one gene-specific primer (At1g26810-1F or At1g26810-3R) and a T-DNA left border–specific primer (LBsail2) and verified by sequencing. RNA was extracted from leaves of homozygous T-DNA insertion lines, and the absence of functional GALT1 transcripts was verified by RT-PCR with At1g26810-1F/-3R primers and UBQ5-D/-U as control primers.

To construct the GALT1 RNAi vector, the intron 2 (XTI2) sequence from the Arabidopsis ß1,2-xylosyltransferase (At5g55500) was PCR amplified from genomic DNA with primers ARA-XTI2-fw/-rv, BamHI-KpnI digested, and ligated into BamHI-KpnI cut vector pUC18 to create puc18XTI2. A 422-bp fragment corresponding to exons 4 and 5 of the GALT1 gene was PCR amplified using primers At1g26810-9F/-10R, XbaI-BglII cut, and inserted into the XbaI-BamHI site of puc18XTI2 to create puc18XTI2-GALT1-s. To amplify the 422-bp antisense GALT1 fragment, PCR was performed with primers At1g26810-11F/-12R. The PCR product was KpnI-EcoRI digested and ligated into KpnI-EcoRI cut puc18XTI2-GALT1-s to create puc18XTI2-GALT1-sas. Subsequently, the sense-intron-antisense cassette was removed by XbaI-BamHI digestion and cloned into XbaI-BamHI cut plant expression vector pPT2 to create pPT2-GALT1-RNAi. Transgenic Arabidopsis RNAi lines were generated by floral dipping of Col-0 plants as described above. The insertion of the RNAi construct into the plant genome was monitored by PCR from genomic DNA with the primers ARA-XTI2-fw/pPT2-1. Changes in the amount of GALT1 transcript in stems were analyzed by RT-PCR with GALT1 primers At1g26810-1F/-2R and core 1,3-fucosyltransferase B with primers FTB-9/-10 as a control. Protein gel blot analysis was performed using JIM84 as described above or anti-HRP antibody as described previously (Strasser et al., 2004a).