β1,3-galactosyltransferase on chromosome 6 is essential for the formation of Lewis a structure on N-glycan in Oryza sativa

β1,3-galactose is the component of outer-chain elongation of complex N-glycans that, together with α1,4-fucose, forms Lewis a structures in plants. Previous studies have revealed that N-glycan maturation is mediated by sequential attachment of β1,3-galactose and α1,4-fucose by individual β1,3-galactosyltransferase (GalT) and α1,4-fucosyltransferase (1,4-FucT), respectively. Although GalT from several species has been studied, little information about GalT from rice is available. I therefore characterized three GalT candidate genes on different chromosomes in Oryza sativa. Seeds of rice lines that had T-DNA insertions in regions corresponding to individual putative GalT genes were obtained from a Rice Functional Genomic Express Database and plants grown until maturity. Homozygotes were selected from the next generation by genotyping PCR, and used for callus induction. Callus extracts of two independent T-DNA mutant rice which have T-DNA insertions at the same gene on chromosome 6 but in different exons showed highly reduced band intensity on a western blots using an anti-Lewis a antibody. Cell extracts and cultured media from suspension culture of the one of these mutant rice were further analysed by N-glycan profiling using matrix-associated laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF). Identified N-glycan species containing β1,3-galactose from both cell extracts and cultured media of knock-out mutant were less than 0.5% of total N-glycans while that of WT cells were 9.8% and 49.1%, respectively. This suggests that GalT located on rice chromosome 6 plays a major role in N-glycan galactosylation, and mutations within it lead to blockage of Lewis a epitope formation.


Introduction
Glycoproteins produced in eukaryotic cells undergo a major post-translational modification called N-glycosylation.The initial steps of N-glycosylation in the endoplasmic reticulum (ER) and Golgi apparatus are somewhat conserved between plants and animals.The later steps of N-glycosylation in plants, however, are quite different from those of animals.In particular, the attachment of α1,3-fucose, β1,2-xylose, β1,3-galactose, and α1,4-fucose to N-glycans occurs in plants but not in mammals.The plant N-glycosylation pathway has been extensively studied and is well understood (Bosch et al. 2013;Lerouge et al. 1998;Nagashima et al. 2018).Briefly, α1,3-fucose and β1,2xylose independently bind to a pentasaccharide core structure (Man3GlcNAc2) that contains one or two terminal N-acetylglucosamines (GlcNAc).N-glycan outer-chain elongation is initiated by β1,3-galactose attachment by β1,3-galactosyltransferase (GalT) and is completed by the later addition of α1,4-fucose to the terminal GlcNAc to form a trisaccharide known as a Lewis a (Le a ) structure (Strasser et al. 2007).In this step, α1,4-fucosyltransferase (1,4-FucT) specifically transfers fucose from GDP-fucose to GlcNAc-bound β1,3-galactose (Wilson, 2001;Leonard et al. 2002Leonard et al. , 2005a)).In rice, N-glycans containing Le a structures were detected in callus extract, suspension culture media and secreted heterologous recombinant proteins (Jung 2022;Jung et al. 2016Jung et al. , 2017Jung et al. , 2019;;Shin et al. 2010;Shin et al. 2011).However, the specific enzyme responsible for β1,3-galactose attachment to GlcNAc in rice, and the gene coding this enzyme in the rice genome, were not determined.
In a previous study, a specific GalT mediating the biosynthesis of N-glycans containing Le a structures in Arabidopsis thaliana was identified (Strasser et al. 2007).Formation of Le a structures was blocked in GalT-lacking A. thaliana plant and restored by overexpression of GalT.Characterization of A. thaliana GalT revealed presence of galactosyltransferase sequence domain (pfam 01762), galactoside-binding lectin domain (pfam 00337) and transmembrane domain as human GalT has in its DNA and amino acid sequence.
In this study, several rice mutants with T-DNA insertions at putative GalT positions were used to determine GalT, which plays a major role for the formation of Le a structures in rice.A mutant rice cell line showed highly decreased level of Le a on a western blot against Le a structures, and its N-glycan profiling.

Sequence analysis
DNA and amino acid sequences of Arabidopsis GalT were obtained from the study of Strasser et al.These sequences were analysed by using BLAST search in Uniprot (http:// www.unipr ot.org/ blast/) and NCBI (https:// blast.ncbi.nlm.nih.gov/ Blast.cgi).These candidates were analysed again to define whether they contain both conserved galactosyltransferase sequence domain (pfam 01762) and galactosidebinding lectin domain (pfam 00337), by searching obtained DNA sequences of these rice GalT candidates through CAZY database (http:// afmb.cnrs-mrs.fr/ CAZY/).
RNA-Seq FPKM Expression Values and TRAP-Seq FPKM Expression Values of each genes were searched and compared through Rice Genome Annotation Project (http:// rice.uga.edu/) locus search.

Obtaining of mutant rice lines and the generation of plants
Mutant rice seeds that corresponded to individual rice GalT candidates were screened in the rice functional genomic express database (http:// signal.salk.edu/ cgi-bin/ RiceGE).A single line for chromosome 1 (3A-11407), two lines for chromosome 2 (3B-00747, NF8522) and two lines for chromosome 6 (3D-02972, 1C-03248) were identified as containing insertions in the putative GalT genes.
Individual mutant rice seeds were washed with 70% ethanol and 1% NaOCl (v/v), then placed on Murashige and Skoog (MSO) medium containing 2.4% plant agar (w/v) (Duchefa).After 2-3 weeks incubation at 28 °C under 16 h light and 8 h dark conditions, seedlings were moved to greenhouses and grown until bare offspring.

Genotyping PCR for identification of homozygous T-DNA insertion
Rice leaves were harvested and homogenized using liquid nitrogen.Genomic DNA was extracted by using a genomic DNA extraction kit (ZymoResearch).To determine homozygous T-DNA insertions into both alleles of the putative GalT region, PCR using genomic DNA was conducted using the primer sets shown in Supplementary Table 3. (Supplementary Fig. 1).Briefly, genomic DNA and three primers were mixed with 2X GoTaq® Green Master Mix (M712, Promega), and PCR was conducted under the following PCR conditions: 1 cycle at 94 °C for 5 min; 30 cycles at 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min, followed by 1 cycle at 72 °C for 5 min.Mutant plants showing homozygous T-DNA insertions were selected and its seeds were harvested for further analysis.
Propagated calli were inoculated into N6CI (without phytagel) media and incubated at 28 °C in the dark using a rotary shaker with a rotation speed of 110 rpm to setup the cell suspension culture.10 g of each cell line was inoculated into a 500 ml sized flask with 100 ml of media, and sub-cultured every 5 days.Spent media and cultured cells were harvested and kept at − 80 °C until the next experiments.

SDS-PAGE and Western blot analysis against to plant N-glycans
To estimate N-glycan patterns in T-DNA mutants, total proteins were extracted from calli.For 100 mg of mass, 200 µL of 1X phosphate-buffered saline was added and homogenized by using a pestle tip.Mutant rice had T-DNA insertion on α-mannosidase I, which is an enzyme mediating early stage of N-glycosylation in plants and showing down-regulated levels of plant-specific N-glycans, was used as positive control (Jung 2022).Protein extracts (30 µL/ sample) were separated via 12% SDS-polyacrylamide gel electrophoresis and stained using Coomassie brilliant blue (SDS-PAGE).For western blot analysis, proteins were trans-blotted and incubated with antibody against plant-specific N-glycans, α1,3-fucose (Agrisera), β1,2-xylose (Agrisera) or Le a epitope (JIM84), respectively, as primary antibody, and then detected by anti-rabbit or rat antibody conjugated with alkaline phosphatase.Colour development was conducted using BCIP/NBT (Milipore).

N-glycan analysis using mass spectrophotometry
Separation and identification of N-glycans from rice total glycoprotein were conducted as described in previous studies with a slight modification (Jung 2022;Shin et al. 2011).Rice calli and cultured media were collected every 5 days after sub-culturing.Calli were homogenized in liquid nitrogen, and then re-suspended using 2 µL/mg of PBS for protein extraction.Both cell extracts and cultured media were centrifuged at 15,000 × g for 30 min at 4 °C.The supernatant was lyophilised, re-suspended, and dialysed using dialysis tubing with distilled water.Collected samples were clarified at 15,000×g for 30 min at 4 °C and applied to a Con A-Sepharose 4B column (GE Healthcare, Uppsala, Sweden), and glycoproteins were eluted with 0.5 M α-D-methylglucoside.After lyophilisation, each sample (2 mg) was digested with 50 µg trypsin (T0303; Sigma-Aldrich) and chymotrypsin (C4129; Sigma-Aldrich) at 37 °C for 16 h; the enzymes were then deactivated by boiling for 10 min.N-glycans were released by incubation with 1.0 mU PNGase A (Roche Diagnostics, Mannheim, Germany) in 60 µL 0.5 M citrate/phosphate buffer at pH 4.0 for 16 h and purified using a Carbograph Ultra Clean column (Alltech, Deerfield, IL, USA) as described previously (Packer et al. 1998).Pyridylamination of N-glycans was performed using a pyridylamination manual kit (Takara Bio Inc., Otsu, Japan) according to the manufacturer's instructions.For further purification of the 2-aminopyridine (PA)-N-glycans, NP-high performance liquid chromatography (HPLC) was performed using an TSK gel Amide-80 column (4.6 × 250 mm; Tosoh Co., Tokyo, Japan) and a Waters 2690 Alliance HPLC separation module (Waters Co., Milford, MA, USA) equipped with the 474 Fluorescence Detector (Waters Co.), as described previously (Guile et al. 1996).Fluorometric detection was carried out at excitation and emission wavelengths of 310 and 380 nm, respectively.The molecular masses of PA-N-glycans were determined using the 4800 Plus matrix-assisted laser desorption ionization time-of-flight/time-of-flight (MALDI-TOF/TOF) analyzer (Applied Biosystems, Framingham, MA, USA).The PA-glycans purified from the pooled NP-HPLC fractions (75-130 min) were subjected to MALDI-TOF/MS analysis as described previously (Shin et al. 2010).The mass spectra were collected in the reflector, positive ion mode using 2,5-dihydroxybenzoic acid (10 mg/mL in 50% acetonitrile/0.1% trifluoroacetic acid and 10 mM NaCl) as a matrix.Purified PA-N-glycans were mixed with an equal volume of matrix solution, and 2 µL of the mixture were placed on the target plate and air-dried (Pfenninger et al. 1999).An m/z range of 1000-2400 was measured and analysed using DATA Explore software (Applied Biosystems, Framingham, MA, USA).Peaks were assigned according to the calculated molecular masses [M + Na + ] of PA-N-glycans, and mannose, galactose, fucose, xylose, and GlcNAc residues were identified.N-glycan structures were assigned by comparison with the plant N-glycans identified to date using SimGlycan version 2.8 (PRE-MIER Biosoft Int., Palo Alto, CA, USA).Relative abundances were calculated based on the areas of the corresponding peaks in the mass spectra, as described previously (Gil et al. 2008).All experiments were repeated at least in triplicate, with similar results.
Q69JV6, A0A0P0VKS6 and Q0E054 showed high similarity overall, with small differences in N-terminus regions; these three were assumed to be the same gene, so Q69JV6 was chosen for further analysis, and A0A0P0VKS6 and Q0E054 were not analysed further.All these isotypes have galactosyltransferase (pfam 01762), galactoside-binding (pfam 00337) and transmembrane domains, similar to Arabidopsis GalT (Supplementary Table 1).Subcellular localisation of these three proteins was Golgi apparatus, which is the place for N-glycosylation process.In the rice genome, genes coding for Q5ZDR9, Q69JV6 and Q67X53 were located on chromosomes 1, 2 and 6, respectively.Hence, these loci were chosen for further experiments.
RNA-seq data was searched through Rice Genome Annotation Project.Notably, Q67X53 showed relatively high fragments per kilobase of transcript per million (FPKM) in the shoot than other part, while that of Q5ZDR9 was medium level compared to other part, and Q69JV6 showed zero level.Overall, TRAP-Seq FPKM Expression Values of Q67X53, Q69JV6, and Q5ZDR9 in the seedling, callus and panicles were ranged from 1.08 to 22.06 FPKM, and panicles showed the highest FKPM from all three different gene (Supplementary Table 2).

Identification of homozygous T-DNA insertion in mutant rice
Rice seeds of a single cell line for chromosome 1 (3A-11407), two cell lines for chromosome 2 (3B-00747, NF8522) and two cell lines for chromosome 6 (3D-02972, 1C-03248) were obtained from Rice Functional Genomic Express Database.To confirm the homozygous T-DNA insertion, genomic DNA was extracted from calli of each mutant rice cell line and used for genotyping PCR.Genotyping PCR was conducted using three primers, Left primer and Right primer plus LBP or RBP (Supplementary Fig. 1A, Supplementary Table 3).From genotyping PCR, wild type produces only PCR product from Left primer and Right primer while homozygote shows a band only from Left primer and LBP, or Right primer and RBP (Supplementary Fig. 1A).In the case of heterozygote, both bands were detected (data not shown).Cell lines proven to be homozygous were further propagated on agar plates and used for further studies.

Screening of mutant contains reduced Le a epitopes
To screen rice GalT gene responsible for galactosylation of N-glycans, protein extracts from callus of each cell were analysed by SDS-PAGE and the proteins transferred to membranes for western blot using antibodies against Le a epitope (Fig. 1).3D-02972 and 1C-03248 both had a T-DNA insertion on Os06g0229200 and showed dramatically reduced Le a epitope detection while wild type and other GalT candidates showed strong multiple bands.A positive control, α-mannosidase I mutant have T-DNA insertion in the ER/Golgiα-mannosidase I, which plays a role in very early stage of N-glycosylation, showed bands that are assumed as a result of insufficient knock-out by T-DNA insertion on the intron or minor activity of α-mannosidase I isotype (Jung 2022).From this result, Os06g0229200 seems have important functions in the galactosylation of N-glycans in rice.For further analysis, 3D-02972 was chosen and inoculated into liquid culture media (N6CI).

N-glycan profiling of identification of homozygous T-DNA insertion in mutant rice
Cells and cultured medium from suspension culture of 3D-02972 were isolated at 3-5 days after sub-culturing and analysed separately (Fig. 2).Protein extracts from suspension-cultured cells were analysed by western blot against plant-specific N-glycans, α1,3-fucose, β1,2-xylose, or Le a epitopes (Fig. 2A-D).Protein extracts from suspension-cultured cells or the secreted glycoproteins in culture medium of 3D-02972 showed no clear bands on a western blot against Le a epitopes (Fig. 2B, E, and F).This confirms a knock-out of GalT in a 3D-02972 mutant that led to blockage of galactosylation of both intra-and extracellular glycoproteins.Notably, band intensity on a western blot using antibodies against α1,3-fucose and β1,2xylose was much stronger than that of the wild type with similar amounts of protein (Fig. 2A, C, and D).This is probably because galactosylation of N-glycan was blocked and the substrates of this enzyme (GnMXF3, GnGnXF3) were accumulated.

Discussion
In this study, rice GalT, the gene encoding the enzyme is responsible for galactosylation of N-glycans on an N-glycosylation pathway, has been identified.Two T-DNA insertional mutants of Os06g0229200 on rice chromosome 6 showed highly reduced galactosylation.Although there were 0.3 and 0.4% of N-glycans containing β1,3-galactose, it is clear that Os06g0229200 plays a critical role in galactosylation.
The presence of residual galactose would probably because of the presence of minor type of GalT or a minor peak due to impurities.Other GalT candidates showed the presence of Le a , which are comparable to that of wild type, so their genes that are disrupted by T-DNA insertions may have minor GalT activity, or are probably not related to N-glycosylation in rice.However, even if those genes may have a minor role in N-glycan galactosylation, it seems insignificant level.
Not only 3D-02972 and 1C-03248, but also other mutant GalT candidate rice plants showed no clear visual defects in a glass house (data not shown), and successfully produced seeds that could be germinated.It is still unclear what functions and importance Le a epitopes on glycoproteins are, but at least it does not seem to affect growth and maturation under controlled conditions.Os06g0229200 showed higher α1,3-fucose and β1,2-xylose level than wild type from cell extracts.Although substrate accumulation could increase absolute amounts, retention of these N-glycans in the cells suggests that Le a may have some relationship with the secretion of glycoproteins.In a previous study, glycoproteins with Le a epitopes were purified and identified using JIM 84 antibody that can bind Le a epitopes, and as a result, it was found that Le a epitopes are abundantly present on extracellular glycoproteins (Fitchette et al. 1999).More recent study found that the glycoproteins have Le a on their N-glycans were mainly involved in cell wall biosynthesis in A.Thaliana, and Six glycoproteins identified in rice are associated with diverse functions but these are mainly secreted or located in the plasma membranes (Beihammer et al. 2021).Indeed, because two knock-out mutant rice of N-acetylglucosaminyltransferase I and α-mannosidase I showed secretion of heterologous recombinant proteins into culture media in previous studies (Jung 2022;Jung et al. 2017Jung et al. , 2019)), secretion would probably not entirely depend on its N-glycan pattern.Further studies, such as comparing the level of secretion of a single protein in the GalT over expressing cell line with the Os06g0229200 or observation of Os06g0229200 under certain stress conditions, are expected to answer what is the exact functions of Le a epitopes.
The decrease in α1,4-fucose on the Os06g0229200 line was expected due to the substrate specificity of α1,4-fucosyltransferase (Leonard et al. 2002(Leonard et al. , 2005;;Strasser et al. 2007;Wilson et al. 2001).In other words, the subsequent α1,4-fucosylation of N-glycan, which is mediated by 1,4-FucT, seems to be blocked by the absence of its enzymatic substrate, β1,3galactose of N-glycan (Galb1,3GlcNAc).It means rice α1,4-FucT has the same mechanisms as that of A. thaliana and of Silene alba.
In a recent study, the inactivation of rice α1,3fucosyltransferase and β1,2-xylosyltransferase by multiplex CRISPR/Cas9 was conducted (Jung et al. 2021).Although those two enzymes were completely knocked-out and their products, which are α1,3-fucose and β1,2-xylose on the N-glycans of glycoproteins, were not detected, downstream β1,3galactosylation and α1,4-fucosylation were processed, unlikely to other plant species such as Nicotiana benthamiana and A. thaliana (Strasser et al. 2004(Strasser et al. , 2008)), unexpectedly.Therefore, the finding of this study provides a specific target position of rice GalT for CRISPR/Cas9 and will help with the generation of rice plants that produce glycoproteins with 'Humanized N-glycan' such as GnGn.In summary, GalT, which is essential for the formation of Le a epitopes in the plant N-glycosylation pathway, was specified in rice.Its coding region was located on chromosome 6, and knock-out of this single gene was enough to block Le a formation in rice.This study provides knowledge of subsequent studies on Le a in plants.

Fig. 1
Fig. 1 Screening of cell line showing decreased level of Le a epitopes in callus extracts by Western blot using anti-body against to Le a epitope.A SDS-PAGE using protein extracts.B Detection of Le a epitope by Western blot using JIM84 antibody.Lane M, pre-trained molecular weight standard (PageRuler protein ladder); Lane gntI, samples obtained from rice cell line with mutation in N-acetylglucosamynyltransferase I; Lane WT, samples obtained from wild-type rice cell line.Lane 1, samples obtained from 3A-11407 rice cell line.Lane 2, samples obtained from 3D-00747 rice cell line.Lane3, samples obtained from NF-8522 rice cell line.Lane 4, samples obtained from 3D-02972 rice cell line.Lane 5, samples obtained from 1C-03248 rice cell line

abFig. 2
Fig. 2 Characterization of N-glycans by Western blot using anti-body against to plant specific N-glycans.A SDS-PAGE using protein extracts used for panels (B), (C) and (D).B Detection of Lea epitope by Western blot using JIM84 antibody.C Detection of α1,3 fucose by Western blot using anti-α1,3-fucose anti-body.D Detection of β1,2-xylose by Western blot using β1,2-xylose anti-body.E SDS-PAGE using 7 day cell-cultured media used for panels (F).F Detection of Lea epitope by Western blot using JIM84 antibody.Lane M, pre-trained molecular weight standard (PageRuler protein ladder); Lane gntI, samples obtained from rice cell line with mutation in N-acetylglucosamynyltransferase I; Lane WT, samples obtained from wild-type rice cell line.Lane 4, samples obtained from 3D-02972 rice cell line

Table 1
Comparison of N-glycans from wild-type and GalT mutant rice