A new insight to explore the regulation between S‐nitrosylation and N‐glycosylation

Abstract Nitric oxide (NO) is a signal molecule in plants and animals. Arabidopsis GSNO reductase1 (AtGSNOR1) catalyzes metabolism of S‐nitrosoglutathione (GSNO) which is a major biologically active NO species. The GSNOR1 loss‐of‐function mutant gsnor1‐3 overaccumulates GSNO with inherent high S‐nitrosylation level and resistance to the oxidative stress inducer paraquat (1,1′‐dimethyl‐4,4′‐bipyridinium dichloride). Here, we report the characterization of dgl1‐3 as a genetic suppressor of gsnor1‐3. DGL1 encodes a subunit of the oligosaccharyltransferse (OST) complex which catalyzes the formation of N‐glycosidic bonds in N‐glycosylation. The fact that dgl1‐3 repressed the paraquat resistance of gsnor1‐3 meanwhile gsnor1‐3 rescued the embryo‐lethal and post‐embryonic development defect of dgl1‐3 reminded us the possibility that S‐nitrosylation and N‐glycosylation crosstalk with each other through co‐substrates. By enriching glycoproteins in gsnor1‐3 and mass spectrometry analysis, TGG2 (thioglucoside glucohydrolase2) was identified as one of co‐substrates with high degradation rate and elevated N‐glycosylation level in gsnor1‐3 ost3/6. The S‐nitrosylation and N‐glycosylation profiles were also modified in dgl1‐3 and gsnor1‐3. Thereby, we propose a linkage between S‐nitrosylation and N‐glycosylation through co‐substrates.


| INTRODUCTION
S-nitrosylation is a crucial mechanism for the exertion of nitric oxide (NO) biological functions in animal and plants. During S-nitrosylation modification, NO molecules are covalently added to consensus cysteine residues flanked by acidic and basic amino acid residues of peptides forming endogenous S-nitrosothiols and modifying protein structures and functions (Greco et al., 2006;Seth & Stamler, 2011;Stamler, Toone, Lipton, & Sucher, 1997). In animals, three major isoforms of NO synthases (NOSs: iNOS, eNOS, nNOS) have been identified since 1989 (Alderton, Cooper, & Knowles, 2001) however not any NOS similar to that in animal models has been identified in plants, although the S-nitrosoglutathione reductase (GSNOR) has been characterized. GSNOR irreversibly catalyzes the metabolism of GSNO which is a major biological active form of NO to GSSG and NH 3 as main products thus indirectly controls the S-nitrosylation level (Malik, Hussain, Yun, Spoel, & Loake, 2011). In Arabidopsis, mutations in the singled-copied GSNOR1 gene cause defects in development (Feechan et al., 2005;Kwon et al., 2012). The gsnor1-3 is a loss-of-function mutant in GSNOR1 gene with increased SNO level, accompanied by severe developmental defects such as semidwarf, bushy, and reduced fertility (Feechan et al., 2005;Kwon et al., 2012;Lee, Wie, Fernandez, Feelisch, & Vierling, 2008). In our previous study (Chen et al., 2009), we performed a genetic screen for the paraquat resistant mutant and characterized the par2-1 mutant gsnor1-3 plays critical roles in studying NO function and GSNOR1 regulated S-nitrosylation (Feng et al., 2013;Hu et al., 2015Hu et al., , 2017Tada et al., 2008;Yang et al., 2015;Zhan et al., 2018). In order to further analyze the role of GSNOR1 in Arabidopsis development, we performed another genetic screen for the suppressor of gsnor1-3 based on its paraquat resistance. Map-based clone revealed that the suppressor was a mutation in one subunit of the oligosaccharyltransferse (OST) complex which catalyzes the transfer of oligosaccharide onto a nascent protein in N-glycosylation.
N-glycosylation is a ubiquitous protein modification in eukaryotes, almost 70% of eukaryotic proteins are glycosylated (Mononen & Karjalainen, 1984). In secretory pathway, glycosylation is essential for proteins to be secreted or integrated in membranes. Nascent polypeptides in the endoplasmic reticulum (ER) lumen which are covalently attached with oligosaccharide to asparagine (Asn) sidechains will be folded into the native structures and delivered to the Golgi apparatus for further folding and conformational maturation, while underglycosylation will cause proteins misfolding and trigger ER stress, unfolded proteins response (UPR), and ER-associated degradation (ERAD) (Aebi, 2013;Lannoo & Van Damme, 2015;Moremen, Tiemeyer, & Nairn, 2012). The ERAD system translocates misfolded proteins across the ER membrane into the cytosol where ubiquitin-conjugated enzymes target these misfolded proteins for degradation.
Here, we isolated a suppressor of gsnor1-3 from an ethylmethane sulfonate (EMS)-mutagenized library based on the hypersensitive phenotype to paraquat. Positional cloning showed that the suppressor mutant nominated as dgl1-3 was due to a point mutation in the exon of DGL1 which encodes a subunit of the OST complex. We hybridized gsnor1-3 with dgl1-3 and other ost mutants such as stt3b and ost3/6. The double mutants exhibited some intermediate phenotypes, for example gsnor1-3 dgl1-3 and gsnor1-3 ost3/6 were sensitive to paraquat, whereas gsnor1-3 ost3/6 was also resistant to glycosylation inhibitor tunicamycin; the dgl1-3 mutant was partial embryo-lethal and post-embryonic development defective while the gsor1-3 −/− dgl1-3 −/+ self-pollinated F2 population contained 1/4 of gsnor1-3 −/− dgl1-3 −/− with fertility. Immunoblot analysis showed that the profiles of S-nitrosylation and N-glycosylation were altered in dgl1-3 and gsnor1-3 compared to that in WT, respectively. By enriching glycoproteins in gsnor1-3 and using mass spectrometry analysis, TGG2 (thioglucoside glucohydrolase2) was identified as one of the 26 co-substrates of Snitrosylation and N-glycosylation which was degraded faster with elevated N-glycosylation level in gsnor1-3 ost3/6. On the basis of our research and reported information, we speculated that disulfide may play a role as a link between S-nitrosylation and N-glycosylation.
To screen gsnor1-3 dgl1-3 mutant, EMS-mutagenized M2 seeds based on gsnor1-3 in the Col-0 background were germinated and grown on 1/2 MS agar plates for 7 days then transferred to 1/2 MS with or without 0.1 μM paraquat for additional 5 days. In the first 7 days, the seedlings were grown on the vertically placed plates for elongation of roots, after transferring the seedlings were placed in the inverted orientation for bending roots. By comparing the relative bending roots lengths with and without paraquat treatment, paraquat hypersensitive mutant was identified. Genetic analyses were performed by pair-wise crossing of individual mutants followed by assessing segregation patterns in F1 and F2 generations.

| Genetic complementation
A DGL1 (At5g66680) genomic DNA fragment of 4226-bp containing 3′-URT was obtained by PCR from wild-type (Col-0) plants, verified by sequencing and then cloned into the XhoI and SpeI sites of a binary vector pER8 to yield pER8-DGL1. The Flag or GFP sequence was linked to the C or N terminal of DGL1 in a head-to-tail configuration.
The constructed vector was transformed into both dgl1-3 −/+ and gsnor1-3 −/− dgl1-3 −/+ plants by floral dipping and the plants with the homogenous dgl1-3 mutation background in T3 generation were isolated by sequencing the PCR products amplified by primers anchored between the mutated site and 5′-UTR.

| Analysis of trypan blue and DAB staining
For trypan blue staining, the seedlings were stained with lactophenol-trypan blue (10 ml of lactic acid, 10 ml of glycerol, 10 g of phenol, 10 mg of trypan blue, dissolved in 10 ml of distilled water) (Keogh, Deverall, & McLeod, 1980). The whole seedlings were boiled for approximately 1 min in the stain solution and then decolorized in chloral hydrate (2.5 g of chloral hydrate dissolved in 1 ml of distilled water) for at least 30 min. The seedlings were viewed and photographed under a stereoscope. DAB(Beyotime, Cat#: ST003) was dissolved into a 50 mM Tris-HCl, pH 3.8 solution with a concentration of 1 mg/ml. The seedlings were submerged into the DAB solution in dark for 5-6 hr then dehydrated in 95% ethanol (Yokawa, Kagenishi, Kawano, Mancuso, & Baluška, 2011).

| Glycoprotein enrichment
Plant material was homogenized in liquid nitrogen and extracted in extraction buffer (20 mM Tris-HCl, pH 7.0 and 20 mM β-mercaptoethanol). Homogenate was filtered through Miracloth and centrifuged at 10,000 g for 10 min. Proteins in cleared extracts were precipitated by 80% saturation of ammonium sulfate and centrifugation at 10,000 g for 10 min at 4°C. Protein precipitates were resuspended in column buffer (20 mM Tris-HCl, pH 7.0, and 500 mM NaCl), centrifuged, and loaded onto Con A-Sepharose (Nanocs, Cat#: AR-CAN-1, 1-ml bead volume) equilibrated with column buffer. After washing with column buffer, bound proteins were eluted with 1%

| Mass spectrometry analysis
The enriched glycoproteins in gels were digested by trypsin (0.01 mg/ml; Promega, Cat #: V5111) . The digested peptides were sent to National Center for Protein Sciences (Beijing) (http://www.phoenix-center.cn) for analyzing by mass spectrometer (Thermo Q Exactive, USA).
Raw data were used for a search against Arabidopsis protein database (www.ncbi.nlm.nih.gov/guide/proteins/; Version, Aug 8, 2018) with the taxonomy restriction to "Arabidopsis thaliana". The BioWorks TurboSequest software was used for the database searching using the following parameters: the mass tolerances for peptides and fragment ions were set to 0.5 Da; (FDR) < 0.05; PSM FDR, Protein FDR, and Site FDR were set under 0.01; minima peptide length was 6; top MS/MS peaks per 100 Da (TOF) was 10.

| Biotin-switch and Western blot assay
S-nitrosylated proteins were analyzed by the biotin-switch assay as described previously (Feng et al., 2013;Yang et al., 2015).

| Genetic screen for the suppressor of gsnor1-3
Paraquat (PQ) is a kind of nonselective herbicide which has an efficient inducer of cell death in animal and plant cells (Suntres, 2002).
In our previous research, we got a par2-1 mutant with strong paraquat resistance which is allelic to gsnor1-3/hot5 (Chen et al., 2009).
The gsnor1-3/hot5/par2-1 mutant showed anti-cell death, reduced fertility and heat acclimation phenotypes. We hypothesized that mutations render gsnor1-3 sensitive to paraquat may represent important genetic loci that are involved in regulation of cell death or NO metabolism. Therefore, we carried out a genetic screen for paraquat sensitive mutants by surveying ethylmethane sulfonate (EMS)generated library based on gsnor1-3. As the mutants after paraquat treatment should be feeble and require to recover, a low concentration of paraquat and delicate method were used for screening the mutants ( Figure S1, Supporting Information). About 8,000 lines of seeds in the library were germinated on 1/2 MS medium vertically for 7 days then transferred to 1/2 MS mediums with or without 0.1 μM paraquat and invertly placed and cultured for additional 5 days. The bent roots lengths of seedlings grown on mediums were measured and the relative root lengths were calculated with roots grown on 1/2 MS medium without PQ as controls. After screening, we got a double mutant named gsnor1-3 dgl1-3 based on the gene cloned in the mutant (Figure 1a,b). Indeed, gsnor1-3 dgl1-3 was so tender that high humidity or pests in greenhouse caused plant death easily. Therefore we crossed gsnor1-3 −/− dgl1-3 −/+ background plants

| The phenotypes of dgl1-3 and gsnor1-3 dgl1-3 mutants
Besides the sensitivity to paraquat, the cotyledons of both dgl1-3 and gsnor1-3 dgl1-3 were dark, suggesting accumulation of DU ET AL. seedlings, the dark cotyledons phenotype was segregated in a 1:3 ratio (dark: green = 34: 125, χ 2 = 1.1), indicating that the mutation is recessive in a single nuclear gene. But the F2 of self-pollinated dgl1-3 −/+ segregated dark cotyledons seedlings less than 25% of total, fluctuated from 8% to 15%. In fact, when we used gsnor1-3 −/− dgl1-3 −/+ (Col-0 background) to cross gsnor1-3 −/+ (Landsberg erecta background) to generate F2 population for mapping, we noticed that the segregation ratio was 1:3 in gsnor1-3 −/− background population but in other background populations the number of dark cotyledons seedlings was less than that in theory. Thus the dgl1-3 mutant was embryo lethal as reported (Lerouxel et al., 2005)   Previous researches have reported that human OST48 is allelic to DGL1 and consists of 456 residues with the first 42 residues in N terminal including a signal sequence meanwhile the nine residues in the C terminal constitutes a cytosolic segment which interacts with DAD1 (Fu, Ren, & Kreibich, 1997;Mohorko et al., 2011). Pig OST48 contains a double lysine motif at the very C terminus which was suggested to confer ER residency (Hardt, Aparicio, & Bause, 2000;Hardt, Aparicio, Breuer, & Bause, 2001;Mohorko et al., 2011). Because of the homolog in function and structure between DGL1 and OST48, we believed some special structure or functions might be possessed by both terminals of DGL1 and the tags in our transgenic lines possibly be sheared before protein maturation.

| The gsnor1-3 mutation endows OST mutants with tunicamycin resistance
DGL1 encodes a critical subunit of the OST complex which catalyzes transfer of oligosaccharide onto nascent peptides in N-glycosylation.
The mutation of DGL1 causes the underglycosylation of proteins.
F I G U R E 3 N-glycosylation pattern and tunicamycin resistance analysis. (a) Immunoblotting with antibody (HRP) raised against β(1,2)-xylose and α(1,3)-fucose N-glycan epitopes. Arrowheads show altered glycoproteins in gsnor1-3 and gsnor1-3 ost double mutants compared to that in Col-0 and relative ost mutants. (b) 7-day-old seedling germinated and vertically grown on 1/2 MS medium were transferred to 1/2 MS medium without (Control) or with 0.5 mg/ml tunicamycin (TM) in inverted orientation for additional 5 days. Bar = 0.5 cm. (c) Relative fresh weight of per plant in panel (b). Asterisks indicate significant differences (***p < 0.001, Student's t test) The OST complex consists of at least eight subunits in Arabidopsis, we crossed stt3b (SALK_134449C) and ost3/6 (SALK_067271C) with gsnor1-3. Unfortunately the seeds of dgl1 SALK lines obtained from donators were inactive or without mutation. All those related mutations and Col-0 were also detected by immunoblot analysis with the HRP antibody. The immunoblot showed that the profiles of glycoproteins in gsnor1-3 were differently expressed to that in Col-0; meanwhile the depression of glycosylation in stt3b and ost3/6 were highly elevated in double mutants (Figure 3a right, arrowheads). Tunicamycin (TM) is a Streptomyces-derived inhibitor of eukaryotic protein N-glycosylation and also an inducer of endoplasmic reticulum (ER) stress. Plants treated with TM will accumulate unglycosylated, misfolded proteins in ER. We tested if gsnor1-3 endowed the stt3b and ost mutants resistance to TM. An experiment similar to screening for the gsnor1-3 suppressor was carried out but TM inhibition was so strong that the bending roots were ceased to elongate, thus we measured the fresh weight of per plant (Figure 3b).
The relative fresh weight of gsnor1-3 after TM treatment was higher than that of Col-0 indicating a strong resistance to TM. Similarly all the gsnor1-3 ost double mutants showed higher resistance than relative ost single mutants (Figure 3c). From these experiments we have reasons to believe that the elevated S-nitrosylation of proteins promoted N-glycosylation.

| The S-nitrosylaiton pattern was strongly changed in gsnor1-3 dgl1-3
Using biotin-switch analysis, we checked the S-nitrosylation patterns in relative mutants. In line with previous research, gsnor1-3 had much more bands representing nitrosylated proteins, but here we cannot tell the level difference of S-nitrosylation between gsnor1-3 and gsnor1-3 dgl1-3 as some bands in gsnor1-3 dgl1-3 were missing while some bands were aggravated. However, the profiles of S-nitrosylation were apparently different between gsnor1-3 and gsnor1-3 dgl1-3 (Figure 4a).
We checked the paraquat sensitivity of gsnor1-3 ost double mutants by measuring the relative bending root lengths of 7-day-old seedlings transferred to 1/2 MS mediums containing 0.1 μM paraquat. The gsnor1-3 ost3/6 double mutant showed a hypersensitivity to paraquat while gsnor1-3 stt3b still had a strong resistance F I G U R E 4 Biotin-switch assay and paraquat resistance analysis. (a) In vivo biotin-switch assay of 7-day-old seedlings. Arrowheads indicate different bands of proteins between gsnor1-3 and gsnor1-3 dgl1-3. (b) 7-day-old seedling germinated and vertically grown on 1/2 MS medium were transferred to 1/2 MS medium without (Control) or with 0.1 μM paraquat (PQ) in inverted orientation for additional 5 days. Bar = 5 mm. (c) Relative bending root lengths of seedlings in panel (b). Asterisks indicate significant differences (***p < 0.001, Student's t test) (Figure 4b,c). As mentioned above, stt3a stt3b double mutant was embryo lethal while stt3b had no obvious underglycosylation defective phenotype, we suspect that because stt3b is a weak underglycosylation mutant that is not able to lead a paraquat sensitive phenotype.

S-nitrosylation and N-glycosylation
As described above, dgl1-3 suppressed the paraquat resistance and changed the S-nitrosylation pattern of gsnor1-3 while gsnor1-3 rescued the embryo lethal phenotype of dgl1-3; meanwhile the S-nitrosylation and N-glycosylation profiles were changed in dgl1-3 and gsnor1-3. On the basis of these phenomena we suspected that S-nitrosylation and N-glycosylation crosstalks with each other through some co-substrates. To identify the co-substrates, the concanavalin A (Con A)-conjugated agarose beads were used in immunoprecipitation (IP) to enrich distinct glycoproteins in gsnor1-3 compared to that in Col-0 (Figure 5a were also found in the list of S-nitrosylated proteins reported in the Arabidopsis nitrosoproteomic analysis . Interestingly, nine out of 116 proteins related to unfolded protein reaction (UPR) such as CRT1, CRT3, PDILs were also identified (Table S1, Supporting Information). We picked TGG2 protein as a target for forward research because it had been proved to be a glycoprotein and the glycosylated and unglycosylated TGG2 proteins were easy to distinguish on a denaturating polyacrylamide gel by Western blot (Liebminger et al., 2012;Ueda et al., 2006). Using chemically synthesized antigen (sequence: AHALDPSPPEKLT) (Ueda et al., 2006), we prepared the antibody of TGG2.
As glycoproteins in underglycosylated mutants were fast degraded for misfolding through ERAD pathway, TGG2 protein volume in ost3/6 was lower with higher electrophoretic mobility compared to that in Col-0. However, no difference of TGG2 was observed between stt3b and Col-0 suggesting that stt3b was a weak underglycosylation mutant which was coincident with the paraquat resistant phenotype (Figure 4b,c). TGG2 in dgl1-3 was suspected to be degraded to low molecular weight fragments (arrowheads) which were completely degraded in gsnor1-3 dgl1-3. In addition, TGG2 is a potential target of S-nitrosylation which degrades faster in gsnor1-3, it is possible that S-nitrosylation promoted TGG2 degradation. As one N-glycan side chain was removed from a glycoprotein, the protein molecular weight would minus 1 kDa, when comparing the electrophoretic mobility of TGG2 in ost3/6, gsnor1-3 stt3b, and gsnor1-3 ost3/6, TGG2 was obviously glycosylated in gsnor1-3 ost3/6 which had the same molecular weight as that in gsnor1-3 stt3b intimating that S-nitrosylation promotes N-glycosylation of TGG2 and a crosstalk between S-nitrosylation and N-glycosylation involving co-substrates.

| DISCUSSION
In this study, by the isolation of paraquat sensitive mutant under gsnor1-3 background, we characterized a new mutation on DGL1 nominated as dgl1-3. Genetic and biochemical experiments showed that dgl1-3 suppressed the bushy phenotype and changed the S-nitrosylation pattern of gsnor1-3. As one important subunit of OST complex, it is believed that DGL1 functions in binding lipid-linked oligosaccharide donor substrates (Pathak et al., 1995). The F I G U R E 5 Affinodetection of specific glycoproteins in gsnor1-3 and N-glycosylation analysis of TGG2. (a) Affinodetection with concanavalin A (Con A) of 7-day-old seedling proteins extracted from Col-0 and gsnor1-3, and separated on a 15% denaturating polyacrylamide gel. Arrowheads show altered glycoprotein profiles in gsnor1-3. CBS for coomassie brilliant blue staining. (b) Immunoblotting of proteins extracts from 7-day-old seedlings with the antibody raised against TGG2. Arrowheads indicate TGG2 fragment in dgl1-3 and glycosylated TGG2 in gsnor1-3 ost3/6 compared to that in ost3/6 and gsnor1-3 stt3b dysfunction of DGL1 in Arabidopsis (dgl1-1and 2) causes severe development defects, such as embryo lethality, reduced cell elongation, and post-embryonic development cease (Lerouxel et al., 2005), while rice Osdgl1 mutant exhibited shorter root length, smaller root meristem, and cell death in the root (Qin et al., 2013). Here, we also observed that dgl1-3 was embryo lethal like dgl1-2 and ceased postembryonic development as dgl1-1. Furthermore, the cell death level in the cotyledons of dgl1-3 was elevated. Both dgl1-1 and 2 are T-DNA insertions in the promoter region while Osdgl1 is a point mutation resulting in premature termination of protein synthesis.
Using total cDNA of Col-0 and dgl1-3 as PCR templates and primers for the full length of DGL1 (1,314 bp), we amplified the same length of bands, and by qRT-PCR the expression of DGL1 in Col-0 and dgl1-3 were at the same level (data not shown). Therefore, the development defects of dgl1-3 must be caused by dysfunction of DGL1 protein and the Gly186 is of great importance.
By segregation ratio assay of F2 population from different backgrounds, we further observed that gsnor1-3 rescued the embryo lethality phenotype of dgl1-3. Furthermore, gsnor1-3 dgl1-3 double mutant was reproductive meaning gsnor1-3 rescued the post-embryo cease phenotype of dgl1-3. DGL1 has two cysteine residues, by assuming that these cysteine residues were S-nitrosylated and transforming cysteine to serine mutated DGL1 to dgl1-3 background, the transformed plant developed normally (data not shown). Therefore, it is impossible for S-nitrosylation rescuing the dysfunction of DGL1, not to mention that DGL1 is just a subunit of OST complex and its S-nitrosylation is hardly to influence the enzyme activity of OST during the N-glycosylation reaction in gsnor1-3 dgl1-3.
Both S-nitrosylation and N-glycosylation are post-translational modifications with enormous amount of substrates. As the S-nitrosylation and N-glycosylation patterns were changed in dgl1-3 and gsnor1-3, the gsnor1-3 dgl1-3 also showed the intermediate phenotypes, we hypothesized that S-nitrosylation and N-glycosylation might crosstalk with each other through some co-substrates, but what is the possible molecular mechanism of the crosstalk? We speculated that disulfide bonds might play a critical role. S-nitrosylation is commonly regarded as stable regulatory modification; however, Wolhuter et al. (2018) showed that S-nitrosothiols rapidly react with thiols to form disulfides. That is, rat aortic smooth muscle cells treated with NO donor, transient S-nitrosothiols were formed and rapidly transited to disulfides. Cherepanova, Shrimal, and Gilmore (2014) reported that in HeLa cells many N-glycosylation sites were either closely bracketed by disulfide or that the N-glycosylation sequon (NXT/S) contained a disulfide-bonded cysteine as the X residue. In the N-glycosylation process a transient mixed disulfide between the OST complex and a free thiol in a glycoprotein substrate was formed that delayed disulfide bond formation until the glycan was added to the Asparagine residue; disulfide in a nascent polypeptide can also be opened by OST to form a mixed disulfide allowing access of OST to an inaccessible sequon (Cherepanova et al., 2014). The work of Wolhuter and Cherepanova reminded us a question: what if the thiols near N-glycosylation sequons of the cosubstrates were occupied by S-nitrosothiols? Especially in gsnor1-3 which accumulated much more GSNO and S-nitrosothiols in the cells, it is possible that the equilibrium between S-nitrosothiols and disulfides could have been broken or both on an elevated level. We suspected that the occupancy of the thiols near N-glycosylation sequons should consume less energy and facilitate the formation of mixed or bracketed disulfides than N-glycosylation. The way to test and verify this hypothesis is to make TGG2 transgenic plants with Snitrosylated sites mutations then to check in vivo if the mutated sites have infected on N-glycosylation and vice versa. This could be our next work to deal with but the exact S-nitrosylation and N-glycosylation sites of TGG2 should be identified first.
Analysis of consensus sequence of S-nitrosylated peptides revealed that the nitrosylated cysteine residues are usually flanked in motifs that contain acidic, but not basic, amino acid residues . In addition, the cysteine thiol micro-environment has been proposed important factors for the specificity of protein S-nitrosylation. Hao, Derakhshan, Shi, Campagne, and Gross (2006) suggested that the key determinants of S-nitrosylation specificity are likely to be undefined 3-D structural features. As N-glycosylation plays an important role in glycoprotein quality-control (GQC) system and controls the primary structure of nascent polypeptides, we have reasons to believe that underglycosylation might lead to a change in the 3-D structures of peptides and effects on S-nitrosylation. On the other hand, the destination of underglycosylated proteins are finally degraded, thus S-nitrosylation of these proteins should be a waste of energy, unless in the cells of gsnor1-3 where accumulated too much GSNO.
Interestingly, we can also conclude from the reports mentioned above that in time sequence N-glycosylation and S-nitrosylation should occur before the formation of disulfide bonds, but what is the sequence between N-glycosylation and S-nitrosylation? It is known that N-glycosylation is actually a co-translational modification: when a nascent glycopeptide is translated by membrane-bound ribosomes and enters the ER lumen, it is glycosylated by ER membrane anchored OST complex. Dose S-nitrosylation happen even before N-glycosylation? It is mysterious to answer this question.
In our previous work, we assumed that PAR2/GSNOR1/HOT5 functions downstream of superoxide to regulate cell death (Chen et al., 2009). Here, because H 2 O 2 as a superoxide in dgl1-3 was depressed in gsnor1-3 dgl1-3, it seems our assumption was reasonable.

| CONCLUSION
By characterizing the suppressor of gsnor1-3, we identified a new point mutation on the DGL1 gene which genetically repressed the paraquat resistance of gsnor1-3. We also presented genetic and biochemical evidences to shed light on the crosstalk between S-nitrosylation and N-glycosylation based on co-substrates. As TGG2 was one of the co-substrates of S-nitrosylation and N-glycosylation, the elevated S-nitrosylation level rescued its underglycosylation in gsnor1-3 ost3/6 double mutant. We hypothesized that the transformation DU ET AL. | 9 between S-nitrosothiols and disulfides bracketing N-glycosylation sites regulated the crosstalk between S-nitrosylation and N-glycosylation, but further investigation are needed.

ACKNOWLEDGMENTS
We would like to thank Drs Sun Mengxiang and Guo Hongwei for providing stt3b and ost3/6 mutant seeds, Dr. Gary Loake for providing the gsnor1-3 seeds. Thank Drs Wu Tingquan, Yao Chunpeng, and Mr Jose Eric Romero Gonzalez for improving the manuscript. We thank all the members of Zuo group for helpful discussion. This study was supported by grants from the National Natural Science