Previous findings indicated that the regulated activities of the β-hexosaminidases and GnTII contribute to the diversity of N-glycan structures, facilitating the formation of the major paucimannose N-glycan, (Xyl)Man₃(Fuc)GlcNAc₂, and the minor complex N-glycan, GlcNAc₂(Xyl)Man₃(Fuc)GlcNAc₂, in the Golgi apparatus of plant cells (Yoo et al., 2015). This prompted us to investigate in more detail whether the structural diversity of N-glycans is necessary and whether the minor complex N-glycan plays an important role in plant cells. Since complex N-glycans played important roles in the stress response and tolerance of Arabidopsis (Kang et al., 2008), we wondered whether GnTII expression is associated with plant growth and development under stress conditions. It seems likely that GnTII expression can be increased under stress to have a positive effect on stress tolerance and stress response in Arabidopsis. We therefore treated Col-0 plants with either MS medium (mock treatment) or MS medium supplemented with NaCl or TM and analyzed GnTII transcription by qPCR to investigate how it might be altered under stress conditions. While GnTII transcription did not increase in response to short-term stress, we found that it did increase in response to prolonged stress (Figure 1A). This finding suggests that GnTII can play an important role in mediating chronic stress responses.

Since GnTII expression was induced in response to chronic stress, we wondered whether the minor complex N-glycan, GlcNAc₂Man₃(Fuc)GlcNAc₂, which is produced in a GnTII-dependent manner, plays an important role in the plant’s response to prolonged stress. We therefore used the Col-0, gnt2-1 mutants, transgenic Arabidopsis plants overexpressing GnTII (three independent lines OE-1, OE-2, and OE-3), and hex1,2&3 plants lacking β-N-hexosaminidase activity to assess the physiological importance of the minor complex N-glycan under chronic stress (Supplementary Figure S1A). When the plants were grown on MS medium, gnt2-1, OE-1, OE-2, OE-3, and hex1,2&3 seedlings did not significantly differ from Col-0 seedlings in the phenotypic characteristics we assessed, which included fresh weight and root length (Figures 1B,D,F,G,I). However, when cultured on MS medium containing 60mm NaCl or 20μgl⁻¹ TM, gnt2-1 seedlings, but not OE-1, OE-2, OE-3 or hex1,2&3 seedlings, showed significant decreases in fresh weight and root length compared to Col-0 seedlings (Figures 1C,E,F,H,I). This finding suggests that the GnTII-based minor complex N-glycan plays a role in plant growth and development under prolonged stress.

To confirm whether the stress-sensitive phenotype of gnt2-1 under chronic stress is caused by altered structure or diversity of N-glycans, immunoblot and lectin blot analyses were performed on proteins extracted from Col-0, gnt2-1, OE-1, and hex1,2&3 plants grown in the presence or absence of TM to (Supplementary Figure S2). Anti-HRP, a polyclonal antibody previously shown to bind peptides with N-glycans containing α1,3-fucose and/or β1,2-xylose residues (Fanata et al., 2013; Harmoko et al., 2016), anti-fucose, and anti-xylose antibodies interacted with proteins extracted from gnt2-1 in a different pattern compared with proteins extracted from Col-0. This result is consistent with previous findings that α1,3-fucosyltransferase, β1,2-xylosyltransferase and GnTII interact competitively with the GlcNacMan₃Glc₂ N-glycan in plant medial Golgi (Yoo et al., 2015). However, these antibodies interacted less with proteins extracted from hex1,2&3. The interactions of these antibodies with proteins extracted from OE-1 plants were comparable to those with proteins extracted from Col-0 plants. No differences in the interactions of proteins extracted from Col-0, gnt2-1, OE-1, and hex1,2&3 plants with ConA and GSII were found. There were no differences in antibody interactions between proteins extracted from plants grown in the presence or absence of TM in our experiments (Supplementary Figure S2).

To obtain more information on the detailed N-glycan structures in Col-0, gnt2-1, OE-1, and hex1,2&3 plants, we analyzed the N-glycans released by peptide:N-glycosidase A treatment of proteins from the four genotypes of plants using an Orbitrap-based mass spectrometer. The N-glycan profiles revealed by the mass spectrometry (MS) data indicated that Col-0 plants produce both the major (Xyl)Man₃(Fuc)GlcNAc₂ N-glycan form and the minor GlcNAc₂(Xyl)Man₃(Fuc)GlcNAc₂ N-glycan, while the gnt2-1 plants produce the major (Xyl)Man₃(Fuc)GlcNAc₂ paucimannose N-glycan but not the minor GlcNAc₂(Xyl)Man₃(Fuc)GlcNAc₂ complex N-glycan (Supplementary Figures S3–S6). The relative amount of the GlcNAc₂(Xyl)Man₃(Fuc)GlcNAc₂ complex N-glycan was higher in proteins extracted from OE-1 plants than in Col-0 plants. In the hex1,2&3 plants, the GlcNAc₂(Xyl)Man₃(Fuc)GlcNAc₂ complex N-glycan was the most abundant, while the paucimannose (Xyl)Man₃(Fuc)GlcNAc₂ was almost undetectable (Supplementary Figures S3–S6). Anti-HRP, anti-fucose, and anti-xylose antibodies interacted weakly with hex1.2&3 plant proteins, while MS analysis showed that the most abundant N-glycan is GlcNAc₂(Xyl)Man₃(Fuc)GlcNAc₂. According to these findings, anti-HRP, anti-fucose, and anti-xylose antibodies should have a lower affinity for antigens containing the GlcNAc₂(Xyl)Man₃(Fuc)GlcNAc₂ complex N-glycan than for antigens containing the (Xyl)Man₃(Fuc)GlcNAc₂ paucimannose N-glycan (Supplementary Figures S2–S6). This finding indicates that the gnt2-1 mutant lacks the minor complex N-glycan which could be linked to the phenotype of abnormal growth in Arabidopsis under prolonged stress.

Callose (β-1,3-glucan) is a key regulator of plasmodesmata permeability as well as a key player in plant growth, development, and stress responses (Jacobs et al., 2003; Cui and Lee, 2016). Since the gnt2-1 mutant showed severe developmental defects, including decreases in fresh weight and root length, we wondered whether the loss-of-function mutation in GnTII is associated with aberrant callose deposition during prolonged stress. To answer this question, we compared the pattern of callose deposition in the plant root apices of Col-0, gnt2-1, OE-1, and hex1,2&3 plants grown on standard MS medium vs. MS medium containing 130mm NaCl or 50μgl⁻¹ TM and (Figure 2). When the plants were cultured on MS medium, none of the four plants—Col-0, gnt2-1, OE-1, and hex1,2&3—showed significant differences in callose deposition in the root apices (Figures 2M–P). However, when grown on MS medium supplemented with 130mm NaCl or 50μgl⁻¹ TM, gnt2-1 plants, but not OE-1 or hex1,2&3 plants, exhibited a distinct punctate pattern of callose deposition in the root apices (Figures 2R,V) compared to Col-0 plants (Figures 2Q,S–U,W,X). This result indicates that the minor complex N-glycan plays a role in root growth and stress responses by regulating callose deposition in root apices during prolonged stress.

Because expression of GnTII was induced by salt stress and ER stress, and gnt2-1 was more responsive to these two forms of stress than Col-0, we hypothesized that the GnTII loss-of-function mutation would result in an enhanced UPR under prolonged stress. Thus, we examined expression of UPR-related genes, such as bZIP17, bZIP28, bZIP60u, bZIP60s, BIP3, CRT, CNX, and PIDL, in Col-0, gnt2-1, OE-1, and hex1,2&3 plants grown on standard MS medium or MS medium containing TM (Figure 3). When the plants were grown on MS medium, there were no significant differences in the expression of the UPR genes between Col-0, gnt2-1, OE-1, and hex1,2&3 plants (Figure 3). However, when the plants were grown in the presence of 20μgl⁻¹ TM, UPR gene expression was significantly higher in gnt2-1 plants than in Col-0, but not in OE-1 or hex1,2&3 plants (Figure 3). Thus, the stress-sensitive phenotype of gnt2-1 is associated with UPR activation by prolonged stress (Figures 1B–I). It is possible that the lack of the minor complex N-glycan results in an increase in unfolded proteins, which in turn causes gnt2-1 to have a higher UPR under prolonged stress.

Previous studies indicated that the β1,2-GlcNAc residues at the nonreducing termini of N-glycans are potentially involved fruit ripening and/or leaf senescence in plants (Ghosh et al., 2011; Fanata et al., 2013). Additionally, when the Col-0, gnt2-1, OE-1, and hex1,2&3 plants were cultivated in soil, the gnt2-1 mutant, but not the OE-1 or hex1,2&3 plants, displayed an early flowering phenotype with reduced rosette leaf number and flowering time compared with Col-0 plants under long-day (LD) condition (Supplementary Figure S1B). Thus, we wondered if the absence of the 6-arm β1,2-GlcNAc residue of N-glycan is related to the onset of dark-induced senescence in gnt2-1 plants. To address this question, we used a dark-induced senescence assay in which we exposed detached rosette leaves from 4-week-old Col-0, gnt2-1, OE-1, and hex1,2&3 plants to darkness for 3day. After 3day in the dark, the younger leaves of the gnt2-1 plants lost more chlorophyll than those of Col-0 plants. However, chlorophyll loss in younger leaves of OE-1 or hex1,2&3 plants was comparable to that of Col-0 plants (Figures 4A,B). Ethylene-mediated chlorophyll degradation is related to the onset of senescence in plant leaves (Hortensteiner, 2006; Song et al., 2014; Qiu et al., 2015). To further verify the initiation of accelerated dark-induced senescence in gnt2-1, we used qPCR to examine the expression of several senescence-associated genes, such as EIN3, ORE1, NYE1, and ACS2, before and after dark treatment. The expression of senescence-associated genes was higher in gnt2-1 plants than in Col-0 plants, but not in OE-1 or hex1,2&3 plants (Figure 4C). These results are consistent with previous findings of increased shelf life of tomato (Solanum lycopersicum) fruits with RNAi-suppressed β-D-N-acetylhexosaminidase genes and early onset of dark-induced senescence in the rice gnt1 mutant (Meli et al., 2010; Fanata et al., 2013). As a result, structural diversity of N-glycans is important for normal plant growth and longevity.

Cytokinins have been shown to inhibit chlorophyll breakdown in tissues, which delays leaf senescence (Gan and Amasino, 1995; Kim et al., 2006). We used qPCR to examine the expression of cytokinin-related genes, such as AHK2, AHP1, AHP2, ARR2, ARR10, and ARR12, in the detached rosette leaves of 4-week-old Col-0, gnt2-1, OE-1, and hex1,2&3 plants before and after a 3-day dark treatment to see if they were involved in the early onset of dark-induced senescence in gnt2-1 plants (Figure 5). In the senescent leaves of plants of all four genotypes after dark treatment, the expression levels of AHP1, AHP2, ARR2, ARR10, and ARR12 were increased (Figures 5B–F), while the expression of AHK2 was decreased (Figure 5A). Cytokinin-related gene expression was significantly lower in gnt2-1 plants than in Col-0 plants after dark treatment (Figure 5). However, except for ARR2, the expression levels of the genes in the OE-1 and hex1,2&3 plants were comparable to those in the Col-0 plants.

We wondered if cytokinin responsiveness was altered in the gnt2-1 mutant since altered cytokinin signaling is linked to the early onset of dark-induced senescence. To address this question, Col-0, gnt2-1, OE-1, OE-2, OE-3, and hex1,2&3 plants were grown for 12day on either standard MS medium or MS medium containing 40nm benzylaminopurine (BAP), a synthetic cytokinin that induces plant growth and development responses (D’Aloia et al., 2011). The extent to which this exogenous cytokinin inhibited primary root growth was significantly less in gnt2-1 plants than in Col-0, OE-1, OE-2, OE-3, or hex1,2&3 plants (Figures 6A–C), suggesting that the reduced cytokinin responsiveness is responsible for altered cytokinin signaling and an earlier onset of dark-induced leaf senescence in gnt2-1. The Arabidopsis histidine protein kinase CKI1 has been shown to localize to the plasma membrane, and treatment of Arabidopsis protoplasts with an N-glycosylation inhibitor, TM, fully abolished CKI1-GFP expression (Hwang and Sheen, 2001). Thus, the changes in the structural diversity of N-glycans may have impact on the conformation, localization, and/or function of the histidine protein kinases, possibly resulting in decreased cytokinin responsiveness in gnt2-1.

To determine if the decreased cytokinin responsiveness in gnt2-1 results in altered cytokinin signaling, we used qPCR to evaluate the expression of cytokinin-regulated genes in Col-0, gnt2-1, OE-1, and hex1,2&3 plants grown on standard MS medium or MS medium containing 40nm BAP for 12day. In comparison with mock-treated plants, AHP1, ARR7, CRF5, and CRF6 expression levels appeared to be higher in the presence of BAP in all four genotypes (Figures 6E,G–I). ARR2 expression was increased only in the presence of BAP in the hex1,2&3 plants (Figure 6F). However, the degree to which AHP1, ARR7, CRF5, and CRF6 gene expression changed in gnt2-1 in the presence or absence of BAP was less than in Col-0, OE-1, or hex1,2&3 plants. Thus, the decreased cytokinin responsiveness results in altered cytokinin signaling in gnt2-1, which may be a consequence of changes in the structural diversity of N-glycans caused by the GnTII loss-of-function mutation.

The decreased cytokinin responsiveness of gnt2-1, which could be explained by changes in the structural diversity of N-glycans, indicated that the mutant plants could also have altered auxin transport. To determine whether gnt2-1 exhibited an altered auxin response, Col-0, gnt2-1, OE-1, and hex1,2&3 plants were grown for 12day on standard MS medium or on MS medium containing 1μm N-1-naphthylphthalamic acid (NPA), a chemical inhibitor that blocks polar auxin transport (PAT) by interfering with the cycling of auxin transporters, like PIN1 (Geldner et al., 2001; Abas et al., 2021). Roots were allowed to expand against gravity along the agar surface by orienting plates vertically (Figure 7). Positive gravitropism was more impaired in gnt2-1 seedlings than in Col-0 seedlings following NPA treatment. However, positive gravitropism of OE-1 or hex1,2&3 seedlings was comparable to that of Col-0 seedlings (Figures 7A,B). Additionally, NPA treatment substantially decreased total fresh weight in gnt2-1 seedlings compared to Col-0 seedlings. The total fresh weights of OE-1, OE-2, OE-3, and hex1,2,&3 seedlings, on the other hand, were similar to those of Col-0 seedlings (Figure 7C). Under salt stress, the expression of PIN2–green fluorescent protein (GFP) signals in the root apical meristem of Col-0 and gnt2-1 seedlings was investigated to see what caused gnt2-1 seedlings to be more sensitive to NPA. The basal membrane in the root apical tissues of Col-0 seedlings exhibited a polarized distribution of PIN2–GFP (Figure 8A). While basal PIN2–GFP localization was still observed in the root apical tissues of gnt2-1 seedlings, PIN2–GFP signals were diffusely distributed across the plasma membrane (Figure 8A). Additionally, the intensity of PIN2–GFP signals was weaker in the root apical meristem of gnt2-1 than in Col-0 (Figures 8A,B). These results indicate that GnTII loss-of-function mutation may affect the structural integrity of auxin transport carriers, such as PINs. Thus, changing the structural diversity of N-glycan in gnt2-1 may affect a number of proteins involved in development or stress response.

A full-length cDNA encoding GnTII was cloned from the Arabidopsis leaf cDNA library by PCR using a pair of gene-specific primers (Supplementary Table S1). The full-length cDNA of GnTII was cloned into a pCAMBIA vector under the control of the CaMV 35S promoter. The plasmid pCAMBIA::35Sp:GnTII containing the hpt gene as a selection marker was transformed into Agrobacterium tumefaciens strain GV3101. The Agrobacterium strain with the pCAMBIA::35Sp:GnTII was subsequently used to transform A. thaliana ecotype Columbia (Col-0) by the floral dip method (Geldner et al., 2001). A total of 50 independent transgenic lines (T0) were selected by hygromycin-resistance screening. The copy number and homozygosity of each transgenic line were determined by their mortality on the selection plates with hygromycin (25mgl⁻¹). A total of 19 independent homozygous transgenic lines (T3) were identified. The seeds from three representative GnTII-overexpressing lines (OE-1, OE-2, and OE-3) were selected for further functional analyses. Mutant seeds of gnt2-1 (Salk_063549) and gnt2-2 (Flag_394A11) were obtained from the Arabidopsis Biological Resource Center and from the INRA, respectively, and screened by PCR with GnTII and T-DNA-specific primer pairs (Supplementary Table S1). The PIN2:PIN2-GFP transgenic line was obtained from Dr. Inhwan Hwang at Pohang University of Science and Technology, Korea (Abas et al., 2021). The PIN2:PIN2-GFP transgenic line was crossed to the gnt2-1 knockout mutants to obtain gnt2-1 mutants expressing the PIN2:PIN2-GFP. Arabidopsis Col-0, gnt2-1, OE-1, OE-2, OE-3, and hex1,2&3 plants were grown in a growth chamber at 22°C under long-day conditions (16-h/8-h light/dark photoperiod, 100–200μmolm⁻² s⁻¹ photon flux density, and 60–70% relative humidity) on 1×Murashige and Skoog (MS) medium (Duchefa), pH 5.8, supplemented with 3% (w/v) sucrose and 0.25% (w/v) gellan gum (PhytoTechnology Laboratories). All seeds were cold-treated at 4°C for 4day in darkness and then incubated at 22°C.

Arabidopsis seeds were placed in Petri dishes on either standard solid MS medium or solid MS medium supplemented with 20μgl⁻¹ TM, 60mm NaCl, 40nm BAP, or 1μm NPA and were allowed to germinate and grow. To analyze GnTII expression following treatment with NaCl and TM, seedlings were transferred to sterilized 3mm paper and submerged in 5ml standard liquid MS medium or liquid MS medium containing 200mm NaCl or 5mgl⁻¹ TM (Kang et al., 2008; Nagashima et al., 2011). To synchronize sample harvesting, seedlings were treated with NaCl or TM for the indicated periods of time just prior to being harvested on the 17th day after imbibition.

Five-day-old plants were treated for 4day with 130mm NaCl and with 50μgl⁻¹ TM at 22°C, respectively. To detect callose deposition, seedlings were immersed in 100mm K₂HPO₄ and 0.005% decolorized aniline blue for 2h in a Falcon tube wrapped in aluminum foil for light protection (Kang et al., 2008). Callose deposition was documented using a FV1000 confocal laser scanning microscope (Olympus) equipped with a DAPI filter set. The optimal excitation wavelength for aniline blue was 370nm, and the emission maximum was 509nm.

Tissue was ground in liquid nitrogen, resuspended in phosphate-buffered saline buffer (pH 7.4, 137mm NaCl, 10mm phosphate, and 2.7mm KCl) and cleared by centrifugation (10min at 15000×g). The protein content was determined using a protein assay kit (Bio-Rad) and bovine serum albumin as a standard. Each protein (20μg) was mixed with SDS–polyacrylamide gel electrophoresis (PAGE) loading buffer, denatured at 95°C for 5min, and subjected to 10% SDS–PAGE under reducing conditions. Separated proteins were either stained with Coomassie Brilliant Blue R-250 or transferred to a nitrocellulose membrane (Hybond-ECL, Amersham). Blots were blocked in 5% (w/v) non-fat dry milk in Tris-buffered saline (TBS) buffer (pH 7.6, 20mm Tris–HCl, and 137mm NaCl) for 1h and incubated in a 1:10,000 dilution of rabbit anti-horseradish peroxidase- (Sigma), anti-α1,3-fucose-, anti-β1,2-xylose antibodies (Agrisera), and anti-GFP (Abcam) in TBS supplemented with 0.1% (v/v) Tween 20. Detection was performed after incubation in a 1:3,000 dilution of a horseradish peroxidase-conjugated goat anti-rabbit antibody (Bio-Rad) in TBS-Tween with Western Blotting Detection Reagents (ECL, Amersham). For lectin blot analysis, concanavalin A (ConA, Sigma) was used to detect N-glycans with terminal mannose residues, and Griffonia simplicifolia lectin (GSII, EY laboratories) was used to detect N-glycans with terminal GlcNAc residues.

N-glycan purification from protein was performed as previously described (Yoo et al., 2015). The sodiated samples (treated with 1mm of NaOH in 80% MeOH) were directly infused into a 30-μm fused silica emitter (New Objective). The relative quantitation of N-glycan was performed on the Q Exactive^™ Plus Orbitrap Mass Spectrometer (Thermo Scientific) equipped with a Nanospray Flex Ion Source for direct infusion at a 0.5μlmin⁻¹ flow rate. The full MS spectra were obtained for quantification with 30-s data acquisition time in the 600- to 2,000-Da mass range. N-linked glycan structures were assigned using GlycoMod platform¹ (Cooper et al., 2001).

Total RNA was extracted from seedlings using a NucleoSpin RNA Plant Kit (Macherey-Nagel) following the manufacturer’s instructions. For each sample, 1μg of purified RNA was used for first-strand cDNA synthesis using a ReverTraAce-α Kit (Toyobo) according to the manufacturer’s instructions. The synthesized cDNA samples were diluted (1:50) with sterile diethylpyrocarbonate -treated water. Quantitative PCR (qPCR) was performed with a CFX96 real-time PCR system (Bio-Rad). qPCR was conducted in a 10μl reaction volume including 0.5μl of each primer (10 pm), 4μl of template cDNA, and 5μl of iQ^™ SYBR Green Supermix (Bio-Rad). The thermal profile used was as follows: 1cycle of 50°C for 2min and 95°C for 5min, followed by 40cycles of 95°C for 10s and 60°C for 30s. Finally, a melting-curve analysis (1cycle) from 65°C to 95°C was carried out. Expression data shown are the means±SEM of three independent biological replicates and were normalized to β-TUBULIN expression. Primers used in this study are presented in Supporting Information Supplementary Table S1.

Plants were grown for 2week on solid MS medium, and then transferred to soil and grown for a further 2week. All green leaves were detached and placed on plastic trays wrapped with a double layer of aluminum foil. Petri dishes were kept in continuous darkness for 3day in a growth chamber at 22°C (Oh et al., 1996).

Chlorophyll was extracted by grinding the excised leaf tissue in 80% acetone (1ml 10mg⁻¹ of tissue) with a ground-glass homogenizer. The homogenate was centrifuged at 1,500g for 5min, and the absorbance of the supernatant solution was determined at 663 and 645nm. Total chlorophyll contents were calculated according to the following equation (Arnon, 1949): total chlorophyll (mgl⁻¹)=8.02(A₆₆₃−A₇₁₀)+20.2(A₆₄₅−A₇₁₀).

Expression of PIN2-GFP was analyzed in the cell division and elongation zone of Arabidopsis roots. Roots were dipped in 10ng/ml propidium iodide (PI, Sigma) in H₂O for 5min in the dark and then removed excess staining by rinsing in water. Fluorescence imaging was performed using a model FV1000 confocal laser scanning microscope (Olympus) with excitation at 488 and 543nm and emission at 510–540nm for GFP and 587–625nm for PI.