|Home | About | Journals | Submit | Contact Us | Français|
Glucosidase II, one of the early N-glycan processing enzymes and a major player in the glycoprotein folding quality control, has been described as a soluble heterodimer composed of α and β subunits. Here we present the first characterization of a plant glucosidase II α subunit at the molecular level. Expression of the Arabidopsis α subunit restored N-glycan maturation capacity in Schizosaccharomyces pombe α— or αβ—deficient mutants, but with a lower efficiency in the last case. Inactivation of the α subunit in a temperature sensitive Arabidopsis mutant blocked N-glycan processing after a first trimming by glucosidase I and strongly affected seedling development.
Protein N-glycosylation in eukaryotic cells is initiated in the endoplasmic reticulum (ER) by the en bloc transfer of a glycan precursor, Glc3Man9GlcNAc2, from a lipid carrier to specific Asn residues in the nascent polypeptide [1, 2]. After transfer, the N-glycan undergoes extensive remodelling that begins with the removal of the outermost (α1-2)-linked glucose residue by the type II membrane protein glucosidase I. Glucosidase II (GCSII) then removes the middle (α1-3)-linked glucose residue from the glycan thus producing monoglucosylated species that may be recognized by the lectin chaperones calnexin (CNX) and/or calreticulin (CRT). Lectin-glycoprotein interaction enhances folding efficiency and prevents ER exit of folding intermediates and irreparably misfolded glycoproteins. GCSII removes then the innermost (α1-3)-linked glucose residue, which prevents further association with CNX/CRT and allows folded proteins to proceed through the secretory pathway. In contrast, incorrectly folded glycoproteins and folding intermediates are reglucosylated by the ER-localized, glycoprotein conformation sensor UDP-glucose: glycoprotein glucosyltransferase and then engaged in a new folding CNX/CRT cycle. If correctly folded, however, glycoproteins are further deglucosylated by GCSII, but not reglucosylated by the glucosyltransferase. The whole process, based on folding cycles via deglucosylation/reglucosylation reactions, shows GCSII to be a major player in glycoprotein folding quality control [3–5].
All GCSII characterized so far are soluble heterodimers composed of two non-covalently linked, but strongly associated subunits, the so called alpha (GCSIIα) and beta (GCSIIβ) subunits. GCSIIα has been described as a soluble protein having a molecular mass of about 110 kDa and displaying sequence homology with glycoside hydrolase family 31 [6–10]. This subunit contains the catalytic domain and carries a conserved WXDMNE motif . It is generally believed that the major function of GCSIIβ is to mediate ER localization of the GCSIIα/GCSIIβ complex . Indeed, GCSIIβ presents a C-terminal ER retention signal (HDEL or VDEL) and a long negatively charged sequence similar to those found in CRT and other luminal ER resident proteins in most species (mammals and S. cerevisiae, but not in S. pombe)[6, 13–16]. Additional data also indicate other functions for GCSIIβ. For instance it is probably involved in the folding of recombinant mammalian GCSIIα expressed also in mammalian cells, as this last protein is insoluble and forms inactive aggregates when expressed alone [16, 17]. However, once GCSIIα is fully folded, GCSIIβ is no longer necessary for activity . The only exception to the presence of an ER retention signal known so far occurs in Saccharomyces cerevisiae GCSIIβ, which is not necessary for retention of GCSIIα in the ER, but is required for cleavage of the innermost (α1-3)-glucose residue by the last subunit .
While GCSII is well characterized in yeasts and many animal species and the biochemical activity of GCSII was described and characterized in mung bean seedlings and soybean suspension-cultured cells more than 15 years ago , surprisingly no molecular information on plant GCSII was available when we started this study.
We report here the cloning of an A. thaliana GCSIIα encoding gene (At5g63840) and the characterization of the protein through complementation of Schizosaccharomyces pombe CGSIIα deficient mutants. We also carried out N-glycan biochemical and structural analysis of A. thaliana mutants defective in GCSII activity and showed the enzyme to play a key role in N-glycan maturation and seedling development.
At5g63840 cDNA, was cloned in two steps by RT-PCR (Access Quick RT-PCR Kit from Promega) from a total RNA extract (RNA extract Kit from Promega) of A. thaliana plants using two pairs of primers (At5g63840 F1/R1 and At5g63840 F2/R2) (Table 1) designed according to positions 1–33, 1391–1415, 1250–1274, 2741–2766 of the A. thaliana cDNA sequence from Genbank: accession number NM_125779. Two blunt-ended fragments obtained from RT-PCR were cloned separately in pTOPO vector (Invitrogen) and the full length cDNA was reconstituted thanks to the Bam HI restriction site in the common part (1250–1415) of these two cDNA fragments and confirmed by sequencing. In order to clone this cDNA in the yeast expression vector pREP3X, the cDNA cloned into pTOPO was digested by Sac I and Xho I before ligation with the pREP3X.
In order to clone the cDNA from At5g63840 into pSGP72 vector, the 3′ end was amplified using primers bearing additional restriction sites. The forward primer At5g63840 F3 matched exactly the cDNA sequence in position 2281–2299, while the reverse primer At5g63840 R3 matched the cDNA sequence in position 2746–2763 and carried two additional restriction sites at 5′ end for Not I and Sac I (Table 1). The 3′ end synthesized was exchanged with the original one from the cDNA already cloned in pTOPO using Sac I and Age I digestions. The construct obtained was transferred to pSGP72 using Not I digestion of both the pTOPO construct and the pSGP72 vector. The final construct carried the full length cDNA from At5g63840 fused at its 3′ end, with the sequence encoding an hemagglutinin-tag (HA-Tag).
Escherichia coli Top10 was used for the first steps of cDNAs cloning. E. coli JA226 strain was used for cloning in S. pombe expression vectors. Bacteria were grown on LB medium supplemented with Kanamycin or Ampicillin, respectively, for constructs in pTOPO or S. pombe expression vectors .
S. pombe cells were grown in rich medium containing 0.5% yeast extract (Difco), 3% glucose, and 75 mg.L-1 adenine. The minimal culture medium was as described in Alfa et al. , supplemented with adenine (75 mg.L-1), uracil (75 mg.L-1), or leucine (250 mg.L-1). Malt extract medium was used for conjugations . The wild type S. pombe strains ADp (h+, ura4-D18, leu1—32, ade6-M216) and Sp61 (h-, ura4-D18, leu1—32, ade1, ade6-M210). SpADIIβ (h-, ura4-D18, leu1—32, ade6-M210, gls2β:: ura4+) were described in D’Alessio et al. .
The wild type strains used as positive controls were h90, ura4-D18, ade6-M216, leu1—32.
Sp61IIα and Sp22IIα mutant constructions were performed as described for the Sp95IIα strain (which is h90), but with strains Sp61 and ADp . The mutants were characterized by Southern blotting analysis as described previously in Fernandez et al. . The mutant genotypes were h-, ura4-D18, leu1—32, ade1, ade6-M210, gls2α::ura4+ and h+, ura4-D18, leu1—32, ade6-M216, gls2α::ura4+, respectively. Sp61IIα was used for transformation with the putative A. thaliana gls2α+ gene. Sp22IIα was used for the gls2αβ construction.
Strains SpADIIβ and Sp22IIα were conjugated at 28°C and diploids were selected in minimal medium supplemented with leucine. Diploids spontaneously sporulated after 7 days in this medium. Tetrads were treated with β-glucuronidase (Sigma) for 5 min at room temperature and vigorously shaken in the presence of acid-washed glass beads. Resulting spores were diluted in water and spread in minimal medium supplemented with adenine and leucine. Haploid colonies and their mating type were identified by colony PCR with primers MT1, MP and MM as described in D’Alessio et al. . Genomic DNA was prepared from 10 mL of culture and the genotypes of spores were determined by PCR with primers specific for each disrupted gene. For gls2α+, the primers used were GIISb and GIIAb as already described in Fernandez et al. . They yielded 542- and 2306-bp bands for the wild type gene and the gls2α::ura4+ insertion, respectively. For gene gls2β+, the primers used were B1s as described in D’Alessio et al.  and B2a (Table 1). They yielded 672- and 2436-bp bands for the wild type gene and the gls2β::ura4+ insertion, respectively. Germinated spores were grown in rich medium without adenine for determination of the adenine genotype. The resultant ADIIαβ double mutant genotype was h-, ura4-D18, leu1—32, ade6-M216, gls2α::ura4+, gls2β::ura4+. This mutant was used for transformation with the putative A. thaliana GCSIIα encoding gene.
S. pombe mutants gls2α and gls2αβ were, respectively, lacking the α and both α and β GCSII subunits, respectively. Transformation was performed by electroporation. Both gls2α and gls2αβ mutants were transformed by constructions in pREP3X and in pSGP72 (see The Forsburg Lab S. pombe pages, http://www.pombe.net). In pREP3X the full length cDNA from At5g63840 was cloned unmodified. In pSGP72 the cDNA from At5g63840 was cloned without the stop codon and fused at the 3′terminus with the HA sequence tag. In both cases the signal peptide from the plant was unchanged. Transformants were isolated in selective medium (minimal medium + adenine) and transformation by Atg63840 was checked by PCR.
Cells from the exponential phase (approximately 1 g in 250 mL of appropriate medium) were harvested (3,000 g, 5 min), resuspended in water, centrifuged (3,000×g, 5 min), and resuspended in 5 mL of Solution A (0.25 M sucrose, 20 mM imidazole, pH 7.5, 1 mM EDTA) containing protease inhibitors (1 mM tosylphenylalanyl chloromethyl ketone, 1 mM phenylmethylsulfonyl fluoride, 1 μM E-64, 1 μM pepstatin, 10 μM leupeptin). Cells were then broken with 10 pulses of 1 min each of vortexing with glass beads and the suspension centrifuged at 5,000×g for 7 min. The supernatant was saved and the pellet resuspended in solution A plus protease inhibitors and treated with the beads as above. The supernatants were pooled and centrifuged at 45 000×g rpm for 60 min. The pellet was then resuspended in 40 mM sodium phosphate buffer pH 7.0, 1 mM EDTA plus the same protease inhibitors as mentioned above.
Short term (30 min) in vivo labelling of S. pombe cells with [14C]glucose and purification of labelled endo-β-N-acetyl-glucosaminidase H (Endo H)-sensitive oligosaccharides were performed as described previously in Fernandez et al.  for S. cerevisiae cells, but with the addition of 50 μL of 50 mM Kifunensine and no 1-deoxynojirimycin. Whatman 1 papers were used for chromatographies. Solvent employed was 1-propanol//water (5:2:4).
Microsomes (100 μg protein) were incubated for 30 min with 1,400 cpm of [glucose-14C]Glc1Man9GlcNAc, 40 mM sodium phosphate buffer pH 7.0 and 0.5% Lubrol in a total volume of 100 μl. The reaction was stopped by adding 1 volume of methanol and heating at 60°C for 5 min. The reaction mix was centrifuged and the supernatant loaded on a Whatman 1 paper. Chromatograhy was run using 2-propanol/acetic acid/water (29:4:9) as solvent. This method was described before .
A. thaliana seeds from wild type Colombia and rsw3 mutant were grown for 2 days at the permissive temperature (21°C) and then at 30°C (non-permissive temperature) for 5 days. Seedlings were then collected and homogenized in an Eppendorf tube containing hot denaturing buffer (20 mM Tris—HCl pH 6.8, 0.3% β-mercaptoethanol, 5% (v/v) glycerol and 1% (w/v) SDS). The homogenate was then boiled for 5 min and centrifuged for 10 min at 12,000×g. Proteins were separated by SDS-PAGE in 15% polyacrylamide gels according to Laemmli . Analytical gels were silver stained according to Blum et al. . After transfer from the gel onto a nitrocellulose membrane, glycoproteins were immunodetected using purified rabbit antibodies specific for (β1-2)-xylose or (α1-3)-fucose residues constitutive of plant complex N-glycans  and the ECL (Amersham) amplification procedure. Affinodetection of glycoproteins on blot was carried out as previously described by Faye and Chrispeels  and using ECL (Amersham) for detection. Immunodetection of the HA-tagged proteins was performed as described in Abe et al.  with high affinity monoclonal antibodies (3F10, Roche). Immunoreactive bands were visualized by staining with horseradish-conjugated goat anti rat IgG (Sigma) and chemiluminiscence (Supersignal West Pico Chemiluminiscence substrate, Pierce).
A crude protein extract was obtained from 7-days-old A. thaliana seedlings after homogenization of 60 mg of freeze dried seedlings in 10 ml of 50 mM HEPES buffer pH 7.5, 2 mM sodium metabisulfite and 0.1% SDS. Insoluble material was eliminated by centrifugation at 4,400×g for 15 min at 4°C. Proteins were precipitated by the addition of trichloroacetic acid (12.5% v/v). After centrifugation (10,000×g, 15 min), the pellet was washed twice in 80% acetone and air-dried. The protein pellet was then solubilized by heating for 4 min in 1 mL of 50 mM sodium acetate buffer pH 5.5, 0.1% SDS. The volume was completed to 5 mL with 50 mM sodium acetate buffer pH 5.5 before adding Endo H (0.1 U). The solution was incubated for 18 h at 37°C. Proteins were precipitated by the addition of 4 volumes of ethanol at -20°C. After centrifugation (10,000×g, 15 min), the supernatant was lyophilized. Released N-glycans were purified by chromatography on 200-mg SepPack C18 column (Varian) followed by a Carbograph column (LudgerClean E10 Cartridge). Bound N-glycans were eluted in 50% acetonitrile, 0.1% trifluoracetic acid (TFA). The oligosaccharides were lyophilized before mass spectrometry analysis.
Two hundred milliunits of Jack bean α-mannosidase (Sigma) were desalted by ultrafiltration (Microcon YM-10, Millipore) and incubated overnight with the glycan mixture. The digest was then directly analyzed using MALDI-TOF MS [31, 32].
MALDI-TOF mass spectra of purified N-glycans were obtained on a Voyager DE-Pro MALDI-TOF instrument (Applied Biosystems, USA) equipped with a 337-nm nitrogen laser. Mass spectra were performed in the reflector, delayed extraction mode using 2, 5-dihydroxybenzoic acid (Sigma-Aldrich) as matrix. The matrix, freshly dissolved at 5 mg.mL-1 in a 70:30% acetonitrile/0.1% TFA was mixed with the oligosaccharides, solubilized in water with 0.1% TFA, in a ratio 1:1 (v/v). The spectra were recorded in a positive mode, using an acceleration voltage of 20,000 V with a delay time of 100 ns. They were smoothed once and externally calibrated using commercially available mixtures of peptides and proteins (Applied Biosystems). In this study, the spectra have been calibrated using des-Arg1-Bradykinin (904.4681 Da), Angiotensin I (1296.6853), Glu1-Fibrinopeptide B (1570.6774 Da), ACTH clip 18–39 (2465.1989) and bovine insulin (5730.6087). Laser shots were accumulated for each spectrum in order to obtain an acceptable signal to noise ratio.
In order to identify A. thaliana GCSIIα subunit involved in the early trimming of N-glycans, mammalian GCSIIα sequences available were blasted against the A. thaliana genome database at NCBI and a gene encoding a putative GCSIIα subunit (At5g63840) was identified. This gene shows a high degree of similarity (>60%) with alpha subunits cloned from yeasts and mammals, [6, 16–18, 33–35]. At5g63840 encodes a protein containing a 16- or 20 amino acid long, cleavable N-terminal signal peptide according to Psort (http://www.psort.nibb.ac.jp) or to SignalP (http://www.cbs.dtu.dk/services/SignalP/), respectively. Two potential N-glycosylation sites in position N689 (NVT) and N804 (NSS) according to Proscan (http://npsapbil.ibcp.fr/) were also identified and one of these two sites (N804) could be actually glycosylated according to NetN-Glyc 1.0 (http://www.cbs.dtu.dk/services/NetNGlyc/). Furthermore, as observed for yeast and mammalian GCSIIα, none of the ER retention or retrieval signals identified so far is present in the amino acid sequence derived from At5g63840. In contrast, the protein sequence carries a conserved WXDMNE motif, which is the typical consensus sequence for the active site of glycosidases in CAZy family-31 .
The corresponding cDNA was cloned in two PCR-steps and subcloned in pTOPO and S. pombe expression vectors as described under Material and methods.
Microsomes prepared from S. pombe mutants defective in GCSIIα (gls2α) or in both GCSIIα and GCSIIβ (gls2αβ), but transformed with At5g63840 cDNA in the pREP3X vector displayed GCSII activity as they were able to liberate labelled glucose from [glucose-14C]Glc1Man9GlcNAc (Fig. 1). This result strongly suggested that At5g63840 encodes a protein bearing the A. thaliana GCSII catalytic activity. To further confirm this result we studied N-glycan maturation in vivo in S. pombe cells. Wild type and gls2α (GCSIIα minus) cells transformed either with an empty vector or with the pREP3X vector containing At5g63840 cDNA were pulse-labelled for 30 min with [14C]glucose in the presence of both Kifunensine and dithiothreitol. The former was added to minimize N-glycan ER mannosidase degradation and the latter to prevent ER exit of glycoproteins and thus further N-glycan enlargement at the Golgi. Dithiothreitol impedes disulfide bond formation and thus proper folding of most glycoproteins. Proteins from incubated cells were proteolytically degraded with Protease Type XIV (Pronase from Sigma) and N-glycans released from glycopeptides by Endo H treatment.
Consistent with the expected effect of Kifunensine, under described conditions wild type S. pombe mainly accumulated the product of GCSI and GCSII (Man9- GlcNAc2) and traces of Man8GlcNAc2 (Fig. 2a). In contrast, the gls2α mutant lacking GCSII activity accumulated the enzyme substrate Glc2Man9GlcNAc2 (Fig. 2b).
The oligosaccharide pattern obtained with the gls2α mutant transformed with At5g63840 cDNA was similar to that obtained from wild type S. pombe except for trace amounts of Glc2Man9GlcNAc2 (Fig. 2c). This result clearly illustrates that GCSII activity was restored by the At5g63840 cDNA in the S. pombe gls2α mutant at an almost wild type level. Altogether these results obtained both in vitro and in vivo clearly indicate that At5g63840 encodes the first plant GCSIIα subunit cloned so far.
Kaushal et al.  showed a 95 KDa protein to be the GCSIIα subunit in soybean suspension-cultured cells while GCSIIα is a 110 kDa protein in mung bean. The molecular mass of AtGCSIIα deduced from its cDNA sequences was 104 kDa, while western blot analysis of protein extracts obtained from S. pombe gls2α mutant expressing HA-tagged AtGCSIIα showed a polypeptide of approximately 119 kDa (Fig. 3a). Moreover, the molecular mass of yeast-made AtGCSIIα was reduced by about 2 kDa after deglycosylation with Endo H (Fig. 3b). This result shows that when expressed in S. pombe AtGCSIIα is glycosylated with probably a single N-glycan, a data in agreement with previous biochemical characterization of a plant GCSIIα subunit  and in silico prediction from the At5g63840 sequence (see above).
As mentioned above, depending on the organism, different roles for GCSIIβ subunits have been proposed. In most cells GCSIIβ is apparently required for ER retention and/or proper folding of GCSIIα whereas in S. cerevisiae it is not required for those roles but to allow cleavage of the innermost glucose unit. Although we have not yet identified A. thaliana GCSIIβ encoding gene (a Blast search showed the At55g56360 gene to be the best candidate; a comparison of this gene with those encoding mouse, S. pombe and S. cerevisiae GCIIβ subunits is provided in supplemental Fig. S1) or fully characterized the role of the protein, preliminary results strongly suggest A. thaliana GCSIIβ subunit to display similar roles as those of homologous proteins in most eukaryotic cells: as shown in Fig. 1, AtGSIIα expressed in gls2αβ (both GCSIIα and GCSIIβ minus S. pombe mutant cells) was able to release the glucose unit from [glucose-14C]Glc1Man9GlcNAc. Moreover, in vivo processing of the N-glycan in yeast double mutant cells expressing AtGCSIIα resulted in the formation of almost similar amounts of Glc2Man9GlcNAc2 and Man9GlcNAc2, thus confirming the ability of the plant catalytic subunit to remove both the middle and innermost glucose residues. Lack of complete N-glycan deglucosylation might reflect a shorter ER permanence of the plant catalytic subunit when expressed in the absence of GCSIIβ subunits. Accordingly, cell free enzymatic assays performed with yeast microsomes prepared from cells preincubated for two h with cyclo heximide, a protein synthesis inhibitor, showed that the plant catalytic subunit synthesized in S. pombe cells expressing its endogenous GCSIIβ subunit (gls2α mutant) had an active ER localization that was longer than when synthesized in yeast cells expressing no GCSIIβ proteins (gls2αβ mutant) (Fig. 4). These results put together strongly suggest that the role of AtGCSIIβ is to determine the ER localization of the catalytic subunit rather than to allow removal of the innermost glucose unit or to allow proper folding of AtGCSIIα.
Burn et al. (2002) reported a Ds-insertional mutant in At5g63840 and found that plants did not survive if both copies of the gene carry the mutation. Inheritance of the tagged allele in heterozygotes was further reduced by severe effects on male and, to a lesser extent, female gametogenesis. Examination of two T-DNA insertional mutants (N_503451 and N_624837) again showed that plants mutated in both copies did not survive. The rsw3 mutant of Arabidopsis shows a one nucleotide change in the At5g63840 sequence that results in the substitution of Ser599 by a Phe residue in AtGCSIIα. The phenotype is severe, but strongly temperature-sensitive allowing ready propagation of the mutant at its permissive temperature and analysis of the mutant phenotype in plants grown at or transferred to the restrictive temperature .
In order to get further information on the effect of this gene mutation in the early trimming of plant N-glycans, we characterized N-glycan structures in rsw3 seedlings that had germinated from seeds collected from plants grown at the permissive temperature (21°C). Seeds were germinated either for 2 d at 21°C followed by 5 d at 30°C or for 7 d entirely at 30°C. Roots emerge on about day 4 and mutant roots were very stunted and swollen with slightly longer roots if the first 2 days were spent at 21°C (Fig. 5).
As illustrated in Fig. 6a, protein patterns obtained after SDS-PAGE and silver staining of the gel were quantitatively and qualitatively similar in both cases. The seedling protein extracts were further analysed for glycosylation patterns by either immunodetection of glycoproteins containing plant complex N-glycans using purified antibodies specific for (β1-2)-xylose- (Fig. 6b) or (α1-3)-fucose-containing (Fig. 6c) glycoepitopes  or by lectin affinity of glycoproteins containing high mannose-type N-glycans using the lectin Concanavalin A (ConA) as probe , (Fig. 6d). Several glycoproteins with complex N-glycans reacting with anti-xylose and anti-fucose antibodies, and at least three glycoproteins with high-mannose N-glycans reacting with ConA were detected in blots from wild type (WT) Arabidopsis seedlings. In contrast, glycoproteins reacting with antibodies specific for complex N-glycans were almost completely absent from rsw3 seedling extracts, while glycoproteins reacting with ConA were strongly increased. Moreover, Lewis A N-glycan-containing glycoproteins immunodetected in wild type seedling extracts using an antibody specific for this glycan structure were not detected in glycoproteins from rsw3 seedlings (data not shown). These results indicate that in the rsw3 mutant seedlings grown for 5 days under restrictive conditions maturation of N-glycans is blocked at an early stage, this block leading to the accumulation of glycoproteins with high-mannose type N-glycans and to a lack of complex type ones. Trace amounts of glycoproteins with complex N-glycans detected in rsw3 seedling extracts were either synthesized during the first 2 days of growth at the permissive temperature, or may reflect residual enzyme activity in the protein product of the rsw3 allele of At5g63840. The latter seems likely since development, albeit highly abnormal, can continue through to flowering even at 30°C (Burn et al. 2002).
To further characterize the structure of the N-glycans reacting with ConA in the rsw3 mutant N-glycans were released from glycoproteins after an Endo H treatment of seedling protein extracts and analysed by MALDI-TOF mass spectrometry. Analyses of WT A. thaliana N-glycans revealed the presence of ions at m/z=1,054, 1,216, 1,378, 1,540 and 1,702, assigned to the sodium adducts of high mannose type N-glycans from Man5GlcNAc to Man9-GlcNAc, previously described in Arabidopsis (Fig. 7). Interestingly, two oligosaccharide structures at m/z=1,864 and 2,026 were exclusively detected in the rsw3 mutant. These molecular ions corresponded to structures containing one GlcNAc and 10 and 11 hexose residues, respectively. Consistent with GCSII inactivation at non-permissive temperature, these structures could be attributed to the presence of one or two terminal glucose residues on the oligosaccharide (Fig. 7A, structures F to K). The presence of N-glycans displaying only one glucose unit might reflect a leaky characteristic of the rsw mutation.
Enzymatic degradation using Jack bean α-mannosidase coupled with MALDI-TOF MS analysis was conducted as described previously,  to further identify the structures of N-glycans specific for rsw3 (Fig. 7). Ions at m/z=1,054, 1,216, 1,378, 1,540 and 1,702 were detected exclusively in the rsw3 mutant after digestion with this glycosidase. Observed products corresponded to oligosaccharide structures containing one GlcNAc and five to eight hexoses, including one to two terminal glucose residues (the mannosidase employed is an exoglycosidase). These structures are presented Fig. 7B, structures d to j).
Using heterologous expression in the yeast S. pombe, we have shown that the protein encoded by At5g63840 is an authentic GCSIIα subunit bearing GCSII catalytic activity. Inactivation of GCSIIα in A. thaliana blocks N-glycan maturation at an early stage, this block resulting in the accumulation of N-glycans with one or two terminal glucose residues and in the absence of complex N-glycans. Consequently, At5g63840 encodes a GCSIIα subunit playing a key role in N-glycan maturation that cannot be balanced by other N-glycan trimming enzymes.
Indeed, in some mammalian tissues, prevention of N-glycan trimming by ER glucosidases using either specific inhibitors or mutant cell lines never completely hinders deglucosylation of the oligosaccharide precursor and formation of complex N-glycans. The GCSII-independent trimming of the terminal glucosyl residues is due to the presence of an alternative deglucosylation pathway involving a cis/medial Golgi endomannosidase, that releases a Glc1,2Man di- or tri-saccharide from the Glc1,2Man9GlcNAc2 . Our results clearly demonstrate that complex N-glycans containing (α1-3)-fucose and (β1-2)-xylose residues are neither present in plant cells after treatment with glucosidase inhibitors [39,], nor in GCSI  or GCSIIα-deficient Arabidopsis mutants (the present paper). Together these observations indicate that plants lack an alternative pathway to N-glycan trimming by ER glucosidases. This result agrees with the reported absence of the endomannosidase from plant tissues .
Removal of terminal glucose residues from N-glycans by GCSI and GCSII plays a central role in the ER quality control of newly synthesized N-glycoproteins [42–44]. N-glycan trimming by ER-resident glucosidases is not essential for suspension-cultured plant cell viability [39, and Gomord et al, unpublished results] the same as for yeast and cultured mammalian cells, thus implying that not only N-glycan processing, but also the CNX/CRT cycle is not essential for these cells. This situation strongly contrasts with lethal phenotypes obtained after inactivation or knocking out GCSI and/or GCSII in multicellular organisms including the model plant Arabidopsis. Indeed, it was previously shown in A. thaliana that glucose trimming by GCSI is required for seed development  and cellulose biosynthesis . The present study, together with results previously published by Burn et al.  illustrate that GCSII activity is also essential for normal plant and seed development in A. thaliana. These results suggest that glucose trimming from the oligosaccharide precursor is strictly required for proper folding through CNX/CRT cycles of some glycoproteins essential for plant growth and cell differentiation, but not for single cell viability. Defects in plant growth and cell differentiation in Arabidopsis rsw3 mutants do not seem to be a consequence of the blockage of complex N-glycan biosynthesis as other complex N-glycan deficient A. thaliana mutants affected in glycosyltranferase activities show no obvious phenotypes except for the N-glycan patterns [46, 47].
This work was supported by grant from the Centre National de la Recherche Scientifique (CNRS) and the French Ministère de la Recherche to VG and by an Australian Research Council grant to RW. P. Soussilane was supported by a fellowship from MIIAT-BP. T. Paccalet was supported by a post-doctoral fellowship from CNRS (VG191063200). Work in Argentina was supported by the NIH (USA) (Grant GM044500), by the Howard Hughes Medical Institute and by the National Agency for the Promotion of Science and Technology.
Electronic supplementary material The online version of this article (doi:10.1007/s10719-008-9201-1) contains supplementary material, which is available to authorized users.
Pravina Soussillane, CNRS, UMR 6037, IFRMP 23, Bâtiment Biologie Extension, Faculté des Sciences, Mont-Saint-Aignan, France.
Cecilia D’ Alessio, Instituto Leloir, Av Patricias Argentinas 435, Buenos Aires, Argentina.
Thomas Paccalet, CNRS, UMR 6037, IFRMP 23, Bâtiment Biologie Extension, Faculté des Sciences, Mont-Saint-Aignan, France.
Anne-Catherine Fitchette, CNRS, UMR 6037, IFRMP 23, Bâtiment Biologie Extension, Faculté des Sciences, Mont-Saint-Aignan, France.
Armando J. Parodi, Instituto Leloir, Av Patricias Argentinas 435, Buenos Aires, Argentina.
Richard Williamson, Plant Cell Biology Group, Research School of Biological Sciences, The Australian National University, G. P. O. Box 475, Canberra, ACT 2601, Australia.
Carole Plasson, CNRS, UMR 6037, IFRMP 23, Bâtiment Biologie Extension, Faculté des Sciences, Mont-Saint-Aignan, France.
Loïc Faye, CNRS, UMR 6037, IFRMP 23, Bâtiment Biologie Extension, Faculté des Sciences, Mont-Saint-Aignan, France.
Véronique Gomord, CNRS, UMR 6037, IFRMP 23, Bâtiment Biologie Extension, Faculté des Sciences, Mont-Saint-Aignan, France.