Uncovering Secretory Secrets

Secretory proteins are first inserted cotranslationally into the endoplasmic reticulum (ER) and then transported through the secretory pathway to the cell surface or to inner hydrolytic compartments in a traffic process that is mediated mostly by the Golgi complex. Addition of the oligosaccharide glycan Glc₃Man₉GlcNAc₂ to Asn residues (N-glycosylation) is a common cotranslation modification of secretory proteins (reviewed in Helenius and Aebi, 2001). Although individual glycoproteins exist that are not affected markedly when synthesized without glycans, no eukaryotes have been identified that can survive without N-glycosylation, indicating that some proteins need N-linked glycans to function properly and/or to avoid rapid degradation.

Depending on the conformation of the glycoprotein and the cell type, the oligosaccharide can be modified after transfer to the polypeptide, mainly by glycosidases and glycosyltransferases present in the Golgi complex (reviewed in Lerouge et al., 1998). Golgi-mediated modifications are critical in vertebrates, in which they determine the life of many serum proteins and specify cell identity, but they are dispensable for plant life (von Schaewen et al., 1993) and are very much simplified in yeast. The first modifications of N-linked glycans, however, occur before glycoproteins leave the ER. In a process that seems to be common to all eukaryotes, the oligosaccharide is modified within minutes after the completion of protein synthesis by the removal of the three terminal Glc residues and the transient readdition of one Glc residue through the action of ER-located enzymes. This appeared for years a puzzling process, until it was discovered that two lectins that reside in the ER, calnexin and calreticulin, exclusively bind glycans with the structure Glc₁Man₉GlcNAc₂ and that the readdition of one Glc residue occurs only on glycoproteins that have not yet reached the correct final folding or assembly (reviewed in Helenius and Aebi, 2001). Thus, the structure recognized by the lectins is present only in folding and assembly intermediates or in structurally defective (misfolded) proteins. The cycle of removal and addition of Glc, therefore, is part of the quality control mechanism that operates in the ER to ensure that only structurally correct proteins proceed along the secretory pathway: newly synthesized glycoproteins are retained in the ER by interactions with the lectins and, during the cycles of binding and release, have the opportunity to fold properly. Defective glycoproteins eventually are targeted for degradation (Helenius and Aebi, 2001).

Two glucosidases operate in the ER: glucosidase I removes the outermost, α1,2-linked Glc residue; subsequently, glucosidase II removes the two remaining α1,3-linked residues and the single residue that is readded by the ER-located glucosyltransferase. Two recent articles report the phenotypes of plants in which the synthesis of these ER-located glucosidases has been inhibited genetically (Taylor et al., 2000; Boisson et al., 2001).

An Arabidopsis mutant impaired in the synthesis of glucosidase I was isolated during the screening of a T-DNA insertion collection in a search for mutants affected in seed development (Boisson et al., 2001). The mutation is lethal in the homozygous state: the seeds do not germinate, and the embryo is blocked at the heart stage. The mutant seeds also lack typical storage vacuoles, have abnormally enlarged cells, and have altered cell walls. Glycan modifications occurring in the Golgi complex begin with the removal of part of the Man residues by Golgi mannosidase I, followed by the addition of one terminal GlcNAc residue by Golgi GlcNAc-transferase I. The structure that is formed, GlcNAc₁Man₅GlcNAc₂, is then subjected to further Golgi modifications by other glucosidases and by glucosyltransferases that mediate the addition of GlcNAc, Fuc, Xyl, and Gal residues to produce the so-called complex glycans (Lerouge et al., 1998). The actions of mannosidase I and GlcNAc-transferase I, which require the previous removal of the three Glc residues, are committed steps in the production of complex glycans. Consistently, the glycoproteins of the Arabidopsis mutant seeds lack the typical complex structures and contain exclusively Glc₃Man_7-8GlcNAc₂ glycans (Boisson et al., 2001).

However, it is unlikely that the absence of complex glycans on glycoproteins is the causal agent of the lethal phenotype, because it has been shown that an Arabidopsis mutant lacking GlcNAc-transferase I does not have growth or reproduction defects (von Schaewen et al., 1993). This strongly suggests that the severe phenotype of the glucosidase I mutant is attributable to the fact that ligands for calnexin and calreticulin are not generated, impairing ER quality control (Boisson et al., 2001).

The synthesis of ER glucosidase II was inhibited in potato by antisense downregulation (Taylor et al., 2000). These transgenic plants have no apparent phenotype when grown in the greenhouse, but they have severely reduced growth and tuber production when grown in the field. They also show signs of plasmolysis and have reduced cell wall lignification and pectin content. This is accompanied by an increase in transcripts for BiP, the HSP70 chaperone of the ER. BiP also is part of the ER quality control system and has a more general role than the mechanism based on the terminal Glc residues, because it recognizes the polypeptide backbone of nascent or defective secretory proteins (Vitale and Denecke, 1999). Most likely, BiP transiently binds hydrophobic sequences exposed in folding and assembly intermediates, inhibiting irreversible aggregation in the crowded ER environment. When proteins are correctly folded and assembled, these sequences are buried inside the protein, effectively concealing BiP binding sites.

An increase in BiP synthesis is diagnostic of ER stress and is induced by conditions that negatively affect protein folding in the ER (Vitale and Denecke, 1999). This induction, termed “unfolded protein response,” is common to the other helpers of folding in the ER, such as protein disulfide isomerase and calreticulin (Shorrosh and Dixon, 1991; Denecke et al., 1995). Thus, transgenic potato plants appear to undergo ER stress when grown in an open field. Taylor et al. (2000) hypothesize that the lack of glucosidase II activity impairs the calnexin/calreticulin quality control mechanism, negatively affecting the folding of newly synthesized glycoproteins, and that the increase of BiP synthesis is a partial compensation for this defect.

Why is the phenotype manifested only in the field? This is not easy to explain, but possibly the stress is favored by the more severe environmental conditions experienced in an open field than in the greenhouse (Taylor et al., 2000). It also should be noted that the antisense strategy greatly reduces but does not completely block glucosidase II activity. Residual glucosidase II activity may be sufficient to create substrates for calreticulin and calnexin, provided that growth conditions are optimal.

Pulse chase experiments using ER glucosidase inhibitors in developing cotyledons or plant cell cultures have failed to show adverse in vivo effects on protein intracellular transport (Chrispeels and Vitale, 1985; Lerouge et al., 1996). Plant calnexin (Huang et al., 1993) and calreticulin (Denecke et al., 1995) have been cloned, and interactions have been shown between calnexin and the newly synthesized subunits of vacuolar H⁺-ATPase (Li et al., 1998) and between calreticulin and BiP (Crofts et al., 1998). However, glycan-mediated binding of glycoproteins to calnexin or calreticulin in plants has not been demonstrated directly. The involvement of Glc residue removal in regulating glycoprotein assembly in the plant ER has been shown using an in vitro translation and translocation system combined with ER glucosidase inhibitors (Lupattelli et al., 1997). The studies by Taylor et al. (2000) and Boisson et al. (2001) extend this observation, showing that an impairment in Glc removal has adverse effects on plant growth and therefore that ER quality control probably is fundamental for the synthesis of secretory proteins required for plant viability. Indeed, several plant embryogenesis mutants are affected in other aspects of the secretory pathway, providing support for the notion that its full integrity is critical for plant life (reviewed in Sanderfoot et al., 2000; see Lukowitz et al., 2001). The in vivo pulse chase experiments performed previously may have been too limited in both the length of treatment with glucosidase inhibitors and the plant material used to allow the detection of detrimental effects (Chrispeels and Vitale, 1985; Lerouge et al., 1996). It will be interesting to determine if only a few key proteins are affected negatively by the absence of Glc removal or if this is a more general phenomenon.

The common detrimental effect on cell wall structure observed in the two studies is particularly intriguing. In yeast, ER glucosidase I and II are not essential for growth, but mutated strains have defects in cell wall synthesis and reduced levels of 1,6-β-glucan (Simons et al., 1998). One of the possibilities raised at the time of this discovery involved the glycosidases as processing enzymes of a hypothetical transient primer for the synthesis of 1,6-β-glucan. However, it has been shown that this is not the case; instead, one Golgi protein involved in the synthesis of 1,6-β-glucan is selectively unstable in strains defective in glucosidase I (Abeijon and Chen, 1998). In this respect, it is interesting that potential N-glycosylation sites are present in RSW1, a catalytic component of cellulose synthase (Arioli et al., 1998), and in two plant glycosyltransferases for which a role in the synthesis of matrix polysaccharides of the cell wall has been confirmed (Edwards et al., 1999; Faik et al., 2000). Moreover, the lethal cyt1 mutation of Arabidopsis, which results in N-glycosylation deficiency (because it affects the enzyme that produces GDP-Man, a precursor for the synthesis of N-linked glycans), also causes a marked reduction in cellulose content (Lukowitz et al., 2001). This may be attributable to the synthesis of unglycosylated RSW1 or the unglycosylated form of other unknown subunits of cellulose synthase (Lukowitz et al., 2001). Therefore, plant glycoproteins necessary to synthesize the cell wall may require perfectly functional quality control in the ER, acting through the cycle of Glc removal and addition.