Structure of the Sec23p/24p and Sec13p/31p complexes of COPII.

COPII-coated vesicles carry proteins from the endoplasmic reticulum to the Golgi complex. This vesicular transport can be reconstituted by using three cytosolic components containing five proteins: the small GTPase Sar1p, the Sec23p/24p complex, and the Sec13p/Sec31p complex. We have used a combination of biochemistry and electron microscopy to investigate the molecular organization and structure of Sec23p/24p and Sec13p/31p complexes. The three-dimensional reconstruction of Sec23p/24p reveals that it has a bone-shaped structure, (17 nm in length), composed of two similar globular domains, one corresponding to Sec23p and the other to Sec24p. Sec13p/31p is a heterotetramer composed of two copies of Sec13p and two copies of Sec31p. It has an elongated shape, is 28-30 nm in length, and contains five consecutive globular domains linked by relatively flexible joints. Putting together the architecture of these Sec complexes with the interactions between their subunits and the appearance of the coat in COPII-coated vesicles, we present a model for COPII coat organization.

A novel quality control compartment derived from the endoplasmic reticulum.

Degradation of proteins that, because of improper or suboptimal processing, are retained in the endoplasmic reticulum (ER) involves retrotranslocation to reach the cytosolic ubiquitin-proteasome machinery. We found that substrates of this pathway, the precursor of human asialoglycoprotein receptor H2a and free heavy chains of murine class I major histocompatibility complex (MHC), accumulate in a novel preGolgi compartment that is adjacent to but not overlapping with the centrosome, the Golgi complex, and the ER-to-Golgi intermediate compartment (ERGIC). On its way to degradation, H2a associated increasingly after synthesis with the ER translocon Sec61. Nevertheless, it remained in the secretory pathway upon proteasomal inhibition, suggesting that its retrotranslocation must be tightly coupled to the degradation process. In the presence of proteasomal inhibitors, the ER chaperones calreticulin and calnexin, but not BiP, PDI, or glycoprotein glucosyltransferase, concentrate in the subcellular region of the novel compartment. The "quality control" compartment is possibly a subcompartment of the ER. It depends on microtubules but is insensitive to brefeldin A. We discuss the possibility that it is also the site for concentration and retrotranslocation of proteins that, like the mutant cystic fibrosis transmembrane conductance regulator, are transported to the cytosol, where they form large aggregates, the "aggresomes."

Masking of an endoplasmic reticulum retention signal by its presence in the two subunits of the asialoglycoprotein receptor.

Human asialoglycoprotein receptor H1 and H2b subunits assemble into a hetero-oligomer that travels to the cell surface. The H2a variant on the other hand is a precursor of a cleaved soluble form that is secreted. Uncleaved H2a precursor molecules cannot exit the endoplasmic reticulum (ER), a lumenal juxtamembrane pentapeptide being responsible for their retention. Insertion of this pentapeptide into H1 (H1i5) causes its complete ER retention but not fast degradation as happens to H2a. Cotransfection of H2a elicited, by heterodimerization, the Golgi processing of H1i5 and its surface expression. This occurred to a much lesser extent by cotransfection of H2b. Likewise, coexpression of H1i5 and not H1 stabilized H2a and caused its export to the cell surface. Homodimerization of molecules containing the pentapeptide did not cancel the retention. Thus, only when the pentapeptide is present in both subunits is the ER retention efficiently abrogated. The results show the unexpected finding that identical ER retention signals present in two associated chains can mask and cancel each other's effect. This could have important implications as similar abrogation of ER retention of other proteins could eventually be obtained by engineering and coexpressing an associated protein containing the same retention signal.

Folding and self-assembly do not prevent ER retention and proteasomal degradation of asialoglycoprotein receptor H2a.

The human asialoglycoprotein receptor H2a precursor, a type II membrane protein, is cleaved to a soluble form that is secreted. Uncleaved precursor molecules are completely retained in the endoplasmic reticulum (ER) and degraded by the proteasome. To find out the causes of its fate we studied folding of H2a precursor, which was very similar to that of its alternatively spliced variant H2b which can exit to the Golgi. Proteasomal inhibition led to accumulation of folded rather than unfolded molecules. Accumulation of ER-retained H2a did not cause an unfolded protein response. Although the receptor is a heterooligomer of the H1 and H2 subunits, single expression led to some self-assembly. Whereas these homooligomers accumulated for H2b they were degraded for H2a. Translocation of H2a into the ER occurred efficiently. Therefore, the retention and proteasomal degradation of uncleaved membrane-bound H2a precursor from the ER do not involve aberrant translocation or misfolding and are not prevented by self-assembly.

Differential role of mannose and glucose trimming in the ER degradation of asialoglycoprotein receptor subunits.

To gain insight into how sugar chain processing events modulate endoplasmic reticulum (ER)/proteasomal degradation we looked at human asialoglycoprotein receptor polypeptides H2a and H2b, variants which differ only by an extra pentapeptide (EGHRG) present in H2a. Membrane-bound H2a is a precursor of a soluble secreted form while H2b reaches the plasma membrane. Uncleaved precursor H2a molecules are completely retained in the ER and degraded as well as a portion of H2b. Inhibition of N-linked sugar chain mannose trimming stabilized both variants. In contrast, inhibition of glucose trimming with castanospermine greatly enhanced the degradation rate of H2a but not that of H2b. We studied a possible involvement of the ER chaperone calnexin, as inhibitors of glucose trimming are known to prevent calnexin binding. Incubation of cells with low concentrations of castanospermine (30 microg/ml) did not interfere with calnexin binding to H2a while causing the same accelerated degradation as high concentrations (>100 microg/ml) which did inhibit the association. Castanospermine treatment after calnexin binding blocked the dissociation of the chaperone but still caused accelerated degradation. The increased degradation could be blocked by a specific proteasome inhibitor, ZL(3)VS. Our results suggest that extensive mannose trimming or retention of glucose residues due to lack of glucose trimming are signals for ER/proteasomal degradation independent of interaction with calnexin.

Endoplasmic reticulum quality control of asialoglycoprotein receptor H2a involves a determinant for retention and not retrieval.

The human asialoglycoprotein receptor H2a subunit contains a charged pentapeptide, EGHRG, in its ectodomain that is the only sequence absent from the H2b alternatively spliced variant. H2b exits the endoplasmic reticulum (ER) even when singly expressed, whereas H2a gives rise to a cleaved soluble secreted ectodomain fragment; uncleaved membrane-bound H2a molecules are completely retained and degraded in the ER. We have inserted the H2a pentapeptide into the sequence of the H1 subunit (H1i5), which caused complete ER retention but, unexpectedly, no degradation. This suggests that the pentapeptide is a determinant for ER retention not colocalizing in H2a with the determinant for degradation. The state of sugar chain processing and the ER localization of H1i5, which was unchanged at 15 degrees C or after treatment with nocodazole, indicate ER retention and not retrieval from the cis-Golgi or the intermediate compartment. H1i5 folded similarly to H1, and both associated to calnexin. However, whereas H1 dissociated with a half time of 45 min, H1i5 remained bound to the chaperone for prolonged periods. The correct global folding of H2a and H1i5 and of other normal precursors and unassembled proteins and the true ER retention, and not exit and retrieval, suggest a difference in their quality control mechanism compared with that of misfolded proteins, which does involve retrieval. However, both pathways may involve calnexin.

Membrane-bound versus secreted forms of human asialoglycoprotein receptor subunits. Role of a juxtamembrane pentapeptide.

The H2a alternatively spliced variant of the human asialoglycoprotein receptor H2 subunit differs from the H2b variant by an extra pentapeptide, EGHRG, present in the ectodomain next to the membrane-span. This difference causes retention and degradation in the endoplasmic reticulum (ER) of H2a when expressed without the H1 subunit in 3T3 cells. In contrast, a significant portion of singly expressed H2b is Golgi-processed and reaches the cell surface. Using a new specific anti-H2a antibody, we found that in HepG2 cells, H2a is rapidly cleaved to a 35-kDa fragment, comprising the entire ectodomain, most of which is secreted into the medium. The cleavage site for the secreted fragment was located at the lumenal end of the membrane span. No membrane-bound H2a exits the ER, indicating that the pentapeptide is a signal for ER retention and degradation of the membrane form but does not hinder secretion of the cleaved soluble form. H2a does not form a membrane receptor complex with H1 as H2b does. H2a is therefore not a subunit of the receptor but a precursor for a secreted form of the protein; signal peptidase is probably responsible for the cleavage to the soluble fragment. Therefore, the juxtamembrane sequence regulates the function of the transmembrane domain of a type II membrane protein as either a signal-anchor sequence (H2b) or as a cleaved signal sequence, which generates a secreted product (H2a).

Cab45, a novel (Ca2+)-binding protein localized to the Golgi lumen.

We have identified and characterized Cab45, a novel 45-kD protein from mouse 3T3-L1 adipocytes. Cab45 is ubiquitously expressed, contains an NH2-terminal signal sequence but no membrane-anchor sequences, and binds Ca2+ due to the presence of six EF-hand motifs. Within the superfamily of calcium-binding proteins, it belongs to a recently identified group of proteins consisting of Reticulocalbin (Ozawa, M., and T. Muramatsu. 1993. J. Biol. Chem. 268:699-705) and ERC 55 (Weis, K., G. Griffiths, and A.I. Lamond. 1994. J. Biol Chem. 269:19142-19150), both of which share significant sequence homology with Cab45 outside the EF-hand motifs. In contrast to reticulocalbin and ERC-55 which are soluble components of the endoplasmic reticulum, Cab45 is a soluble protein localized to the Golgi. Cab45 is the first calcium-binding protein localized to the lumenal portion of a post-ER compartment; Cab45 is also the first known soluble protein resident in the Golgi lumen. Cab45 can serve as a model protein to determine the mechanism of retention of soluble proteins in the Golgi compartment.

Blocking intracellular degradation of the erythropoietin and asialoglycoprotein receptors by calpain inhibitors does not result in the same increase in the levels of their membrane and secreted forms.

The erythropoietin receptor (EPO-R), a type 1 membrane glycoprotein, is degraded mainly in the lysosomes or endosomes, whereas the asialoglycoprotein receptor (ASGP-R) H2a subunit, a type 2 membrane glycoprotein, is degraded exclusively in the endoplasmic reticulum. The present study describes compounds that inhibit the intracellular degradation of these receptors in an efficient manner. However, the levels of cell-surface expression and secretion of their soluble exoplasmic domains were not enhanced to the same extent. The calpain inhibitors N-acetyl-leucyl-leucyl-norleucinal (ALLN) and N-acetyl-leucyl-leucyl-methional (ALLM) inhibited EPO-R degradation profoundly. After 3 h of chase using Ba/F3 cells and NIH 3T3 fibroblasts expressing the EPO-R, virtually all of the receptor molecules were degraded, whereas 80% of the pulse-labelled receptor remained intact in the presence of the inhibitor. EPO-R cell-surface expression was elevated 1.5-fold after 1 h of incubation with ALLN. In the absence of protein synthesis, ALLN caused the accumulation of non-degraded EPO-R molecules in endosomes and lysosomes, as determined by double immunofluorescence labelling of NIH 3T3 cells expressing EPO-Rs. In Ba/F3 cells expressing a soluble EPO-R, ALLN treatment increased secretion of the soluble exoplasmic domain of the EPO-R 2-5-fold. Similarly, in NIH 3T3 cells singly transfected with the ASGP-R H2a subunit cDNA, ALLN inhibited degradation of the ASGP-R H2a subunit precursor, as well as the degradation of the 35 kDa proteolytic fragment corresponding to the receptor ectodomain, by 3-6-fold. However, accumulation of secreted proteolytic fragment in the medium was augmented in the presence of ALLN by only 1.75-fold. In cells expressing the G78R mutant of the ASGP-R H2a subunit, which is not cleaved to the 35 kDa fragment [Yuk and Lodish (1993) J. Cell Biol. 123, 1735-1749], degradation of the precursor was inhibited. Overall, our data suggest the involvement of cysteine proteinases located in the endoplasmic reticulum, as well as in post-Golgi compartments, in degradation of the EPO-R and the ASGP-R H2a subunit. The much lower effect of the inhibitory compounds on cell-surface and secreted forms of the EPO-R and ASGP-R H2a subunit illustrates the complexity and the tight regulation of the cellular localization and stability of membrane proteins.

An alternatively spliced miniexon alters the subcellular fate of the human asialoglycoprotein receptor H2 subunit. Endoplasmic reticulum retention and degradation or cell surface expression.

Two types of cDNAs encoding the H2 subunit of the human asialoglycoprotein receptor had been cloned, differing only by the presence (H2a) or absence (H2b) of a segment of 15 base pairs (bp), encoding five amino acids (Glu-Gly-His-Arg-Gly) immediately carboxylterminal (exoplasmic) to the single membrane-spanning segment. We have cloned and sequenced this region of the H2 gene and showed that the two H2 forms are alternatively spliced variants differing in the presence of a 15-bp miniexon. Both H2 messenger RNAs were found in HepG2 cells, H2b accounting for about 92% of the H2 mRNAs. When expressed in NIH 3T3 cells without the H1 receptor subunit, the two-variant polypeptides exhibit different subcellular fates. H2a is completely retained in and degraded in the endoplasmic reticulum or a related pre-Golgi compartment. In contrast a substantial amount of H2b is processed by Golgi enzymes and reaches the cell surface. Thus, the sole difference determining the subcellular localization of the two forms if the five-amino acid insert in H2a. When a virion-packaged retroviral vector containing H2a cDNA infected 3T3 cells, 70% of the resulting clones expressed H2b and 30% H2a. Thus the 15-bp H2a miniexon can be spliced out, at least during the retrovirus life cycle.

Intracellular degradation of unassembled asialoglycoprotein receptor subunits: a pre-Golgi, nonlysosomal endoproteolytic cleavage.

The human asialoglycoprotein receptor is a heterooligomer of the two homologous subunits H1 and H2. As occurs for other oligomeric receptors, not all of the newly made subunits are assembled in the RER into oligomers and some of each chain is degraded. We studied the degradation of the unassembled H2 subunit in fibroblasts that only express H2 (45,000 mol wt) and degrade all of it. After a 30 min lag, H2 is degraded with a half-life of 30 min. We identified a 35-kD intermediate in H2 degradation; it is the COOH-terminal, exoplasmic domain of H2. After a 90-min chase, all remaining intact H2 and the 35-kD fragment were endoglycosidase H sensitive, suggesting that the cleavage generating the 35-kD intermediate occurs without translocation to the medial Golgi compartment. Treatment of cells with leupeptin, chloroquine, or NH4Cl did not affect H2 degradation. Monensin slowed but did not block degradation. Incubation at 18-20 degrees C slowed the degradation dramatically and caused an increase in intracellular H2, suggesting that a membrane trafficking event occurs before H2 is degraded. Immunofluorescence microscopy of cells with or without an 18 degrees C preincubation showed a colocalization of H2 with the ER and not with the Golgi complex. We conclude that H2 is not degraded in lysosomes and never reaches the medial Golgi compartment in an intact form, but rather degradation is initiated in a pre-Golgi compartment, possibly part of the ER. The 35-kD fragment of H2 may define an initial proteolytic cleavage in the ER.

Kainyl-bovine serum albumin: a novel ligand of the kainate sub-type of glutamate receptor with a very high binding affinity.

Bovine serum albumin has been conjugated with kainylaminooxyacetylglycine to afford a multivalent kainylated protein called kainyl-bovine serum albumin (KA-BSA). This derivative, radiolabelled with 125I to more than 5000 Ci/mmol, was found to interact in the chick, goldfish and rat brain to specific membranous sites displaying the pharmacological properties attributed to the kainate sub-type of glutamate receptor. Measurements of the kinetics of association and dissociation of KA-BSA showed a quasi-irreversible binding with dissociation constants in the subpicomolar and nanomolar range. The chemical properties and the binding characteristics of KA-BSA suggest that it interacts mainly with kainate binding sites present in clusters in the membrane. Localization of the KA-BSA binding sites, by autoradiography in the chick cerebellum and by immunoperoxidase staining in the goldfish cerebellum, revealed an exclusive association with the molecular layer.

3-O-methylation of mannose residues. A novel reaction in the processing of N-linked oligosaccharides occurring in Mucor rouxii.

Yeast- and mycelial-form cells of the dimorphic fungus Mucor rouxii incubated with [U-14C]glucose were found to synthesize Man-P-dolichol, Glc-P-dolichol, and Glc3Man9GlcNAc2-P-P-dolichol. The structure of the oligosaccharide moiety of the latter was similar to that of the same compound isolated from other eucaryotic cells. Oligosaccharides that migrated on paper chromatography as Man6-30GlcNAc standards were obtained upon treatment of delipidated proteins with a protease and endo-beta-N-acetylglucosaminidase H. The oligosaccharides that migrated apparently as single substances on paper chromatography could be separated into three different populations by paper electrophoresis in sodium borate buffer. The fastest migrating substances contained only mannose and N-acetylglucosamine residues, whereas the other two contained, in addition, different proportions of 3-O-methylmannose units. The oligosaccharides with the highest content of 3-O-methylmannose residues appeared to be completely resistant to alpha-mannosidase degradation; they were, however, cleaved by endo-beta-N-acetylglucosaminidase H. Mycelial cells synthesized a much higher proportion of 3-O-methylmannose-containing oligosaccharides than yeast cells. Cells incubated with [methyl-14C]methionine were found to label only the N-linked oligosaccharides containing 3-O-methylmannose residues. It is concluded that transfer of Glc3Man9GlcNAc2 to protein is followed by excision of glucose and probably one or two mannose residues, followed by further mannosylation and in some cases also methylation of oligosaccharides. This represents a novel reaction in the processing of N-linked oligosaccharides.

Evidence that transient glucosylation of protein-linked Man9GlcNAc2, Man8GlcNAc2, and Man7GlcNAc2 occurs in rat liver and Phaseolus vulgaris cells.

Formation of protein-linked Glc1Man9GlcNAc2 , Glc1Man8GlcNAc2 , and Glc1Man7GlcNAc2 was detected in rat liver slices and Phaseolus vulgaris seeds incubated with [U-14C]glucose. Similar compounds were not synthesized in Saccharomyces cerevisiae cells incubated under similar conditions. Rat liver microsomes were incubated with [glucose-U-14C] Glc3Man9GlcNAc2-P-P-dolichol or UDP-[U-14C]Glc as glycosyl donors. Only in the latter condition protein-linked Glc1Man8GlcNAc2 and Glc1Man7GlcNAc2 were formed. Addition of mannooligosaccharides that strongly inhibited alpha 1-2-mannosidases to incubation mixtures containing rat liver microsomes and UDP-[U-14C]Glc did not prevent formation of protein-bound Glc1Man8GlcNAc2 and Glc1Man7GlcNAc2 . Furthermore, the presence of amphomycin in reaction mixtures containing liver membranes and UDP-[U-14C]Glc completely abolished synthesis of glucosylated derivatives of dolichol without affecting formation of protein-linked Glc1Man9GlcNAc2 , Glc1Man8GlcNAc2 , and Glc1Man7GlcNAc2 . The results reported above indicated that under the experimental conditions employed protein-bound Glc1Man9GlcNAc2 , Glc1Man8GlcNAc2 , and Glc1Man7GlcNAc2 were formed by glucosylation of unglucosylated oligosaccharides. Results obtained in pulse-chase experiments performed in vitro also supported this conclusion. UDP-Glc appeared to be the donor of the glucosyl residues. The rough endoplasmic reticulum was found to be the main subcellular site of protein glucosylation. It is tentatively suggested that this process could prevent extensive degradation of oligosaccharides by mannosidases during transit of glycoproteins through the endoplasmic reticulum.

Transient glucosylation of protein-bound Man9GlcNAc2, Man8GlcNAc2, and Man7GlcNAc2 in calf thyroid cells. A possible recognition signal in the processing of glycoproteins.

Calf thyroid slices incubated with [U-14C]glucose synthesized protein-bound Glc3Man9GlcNAc2, Glc2-Man9GlcNAc2, Glc1Man9GlcNAc2, Glc1Man8GlcNAc2, and Glc1Man7GlcNAc2. Although label in the glucose residues of the last three compounds could be detected within 5 min of incubation, appearance of radioactivity in the mannose residues of the alpha-mannosidase-resistant cores of Glc1Man8GlcNAc2 and Glc1Man7GlcNAc2 took more than 30 and 60 min, respectively, to appear after label was detected in the same mannose residues of Glc1Man9GlcNAc2. The glucose residues were removed upon chasing the slices with unlabeled glucose. The last compound to disappear was Glc1Man9GlcNAc2. Calf thyroid microsomes incubated with UDP-[U-14C]Glc synthesized the five protein-bound oligosaccharides mentioned above. Although addition to GDP-Man to the incubation mixtures greatly diminished the formation of Glc3Man9GlcNAc2 bound either to dolichol-P-P or to protein, labeling of Glc1Man9GlcNAc2, Glc1Man8GlcNAc2, and Glc1Man7GlcNAc2 was not affected. Addition of kojibiose prevented deglucosylation of protein-bound Glc3Man9GlcNAc2 without affecting the formation of Glc1Man8GlcNAc2 and Glc1Man7GlcNAc2 and only partially diminishing that of Glc1Man9GlcNAc2. These results indicate that Glc1Man8GlcNAc2 and Glc1Man7GlcNAc2 were formed by glucosylation of the unglucosylated species and not be demannosylation of Glc1Man9GlcNAc2 and that probably part of the latter compound was formed in the same way.

Separation of dolichol monophosphate mannose and dolichol monophosphate glucose by thin-layer chromatography.

Dolichol monophosphate mannose and dolichol monophosphate glucose from mammalian or mold cells were separated by thin-layer chromatography. This separation was not due to a difference in the size of the lipid moieties. It was found that the lipid moieties of dolichol monophosphate mannose and dolichol monophosphate glucose are the same size when synthesized by the same cell. The thin-layer chromatographic separation appeared to be due, therefore, to the saccharide part of the molecules.

Protein glycosylation in Trypanosoma cruzi. The mechanism of glycosylation and structure of protein-bound oligosaccharides.

As reported previously (Parodi, A.J., and Cazzulo, J.J. (1982) J. Biol. Chem. 257, 7641-7645), label was incorporated first to the glucose residues of protein-bound Glc1Man9GlcNAc2, Glc1Man8GlcNAc2, and Glc1Man7GlcNAc2 when Trypanosoma cruzi cells, the causative agent of Chagas disease, were incubated with [U-14C]glucose. It is now reported that the glucose residues are removed from the oligosaccharides after a chase period. The relative proportion of Man9GlcNAc2, Man8GlcNAc2, Man7GlcNAc2, and Man6GlcNAc2 appeared to be the same after 120 and 180 min of chase, thus indicating that these compounds were the fully processed protein-bound oligosaccharides. No complex type protein-bound oligosaccharides were detected. Evidence is presented indicating that Glc1Man7GlcNAc2 was formed mainly by glucosylation of Man7GlcNAc2 and not by demannosylation of Glc1Man9GlcNAc2. Man9GlcNAc2 was the first oligosaccharide to be labeled when cells were incubated with [2-3H]mannose. Based on these and previous results, the overall mechanism of protein N-glycosylation appeared to be: (formula; see text) The structure of the oligosaccharides appeared to be similar to some of those present in human glycoproteins. T. cruzi cells isolated from distant locations in South America were found to share a common mechanism of protein glycosylation.