ER Processing, Folding, and Quality Control

The endoplasmic reticulum is where roughly one-third of the human proteome — all secreted proteins, most membrane proteins, and ER/Golgi/lysosomal residents — begins its post-translational life. As a nascent chain threads through the Sec61 translocon (established under cotranslational translocation), it faces a folding problem of enormous difficulty: it must find its native three-dimensional structure inside a crowded, oxidizing compartment, often before synthesis is even complete. N-linked glycosylation is the ER's first move: a large branched oligosaccharide, Glc₃Man₉GlcNAc₂, is transferred en bloc from a lipid carrier (dolichol phosphate) onto the side-chain amide nitrogen of asparagine residues that sit within the consensus sequence Asn-X-Ser/Thr (X being any residue except proline). This happens co-translationally, catalyzed by the multi-subunit enzyme oligosaccharyltransferase (OST), which sits adjacent to the translocon. The glycan is not decoration — it is a functional handle. Two ER glucosidases trim the outer glucoses in a strict order, converting the tri-glucosylated glycan into a mono-glucosylated form. This single remaining glucose is the ligand recognized by two homologous lectin chaperones, calnexin (membrane-bound) and calreticulin (soluble, ER-lumenal). Binding recruits the thiol oxidoreductase ERp57, which catalyzes and isomerizes disulfide bonds within the client protein. The cycle of glucose removal, chaperone release, and — if folding is incomplete — glucose re-addition by the enzyme UGGT (UDP-glucose:glycoprotein glucosyltransferase) constitutes a folding-sensing cycle: UGGT is the only enzyme in the pathway that can distinguish folded from unfolded conformations, and it re-glycosylates only proteins that still expose hydrophobic surface, giving them another round of chaperone assistance. This mechanism is one instance of the broader principle of ER quality control (ERQC): only correctly folded and assembled proteins are permitted to exit the ER for the Golgi; anything that persistently fails is retained, and eventually degraded by ER-associated degradation (ERAD), in which slow mannose trimming acts as a "molecular timer" flagging chronic failures for retrotranslocation to the cytosol and proteasomal destruction. The system therefore performs three linked jobs — modify, assist, and judge — using one glycan as the shared currency of communication between them.

Step 1 — Co-translational N-glycosylation. As the ribosome-translocon complex extrudes the nascent chain into the ER lumen, the OST complex — anchored in the ER membrane near the Sec61 channel — scans for N-X-S/T sequons. It catalyzes transfer of the pre-assembled 14-residue oligosaccharide (three glucose, nine mannose, two N-acetylglucosamine residues — Glc₃Man₉GlcNAc₂) from its dolichol-pyrophosphate lipid carrier onto the asparagine's side-chain nitrogen, forming an N-glycosidic bond. Only sequons that are still lumenally exposed and not already buried by folding are modified, which is why glycosylation must occur essentially as the chain emerges, not after folding is complete. Not every sequon is used — local structure and proximity to the membrane affect efficiency — but most secretory and membrane proteins carry multiple such glycans.

Step 2 — Glucose trimming generates the calnexin/calreticulin ligand. Two ER glucosidases act processively and in strict sequence: Glucosidase I removes the outermost (α1,2-linked) glucose almost immediately. Glucosidase II then removes the second glucose, producing the mono-glucosylated species Glc₁Man₉GlcNAc₂ — the specific ligand for calnexin and calreticulin. Glucosidase II can also remove the final glucose, generating the unglucosylated glycan that no longer binds these chaperones; this is the "exit" state from the cycle.

Step 3 — Lectin chaperone binding and oxidative folding. Calnexin (a type I membrane protein) and calreticulin (its soluble paralog) bind the mono-glucosylated glycan through a globular lectin domain, while an extended arm domain recruits ERp57, a protein disulfide isomerase family member specialized for glycoprotein substrates. ERp57 catalyzes disulfide bond formation and rearrangement in the bound client, and calnexin/calreticulin binding itself retards aggregation by shielding exposed hydrophobic surfaces — the chaperone acts partly as a glycan-targeted holdase.

Step 4 — The fork: UGGT as folding sensor. When glucosidase II removes the final glucose, the client is released from calnexin/calreticulin. At this branch point, UGGT inspects the protein.

• If the protein is natively folded, UGGT does not act, and the glycoprotein proceeds toward ER exit sites.
• If the protein still exposes unfolded/misfolded surface, UGGT adds a single glucose back (using UDP-glucose as donor), regenerating the calnexin/calreticulin ligand and re-engaging the folding cycle.

This iterative glucosylation/deglucosylation cycle can repeat many times, giving the protein repeated opportunities to fold correctly — a proofreading loop built entirely around the chemistry of one sugar residue.

Step 5 — Mannose trimming as a folding-attempt clock. While the glucose cycle runs, ER mannosidase I (and mannosidase-like proteins such as EDEM1–3 in mammals) slowly and irreversibly trims mannose residues from the Man₉ branch, particularly the mannose on the C-branch (the so-called B-branch mannose in some nomenclatures — trimming of the terminal α1,2-mannose on branch B generates Man₈GlcNAc₂ isomer B). Because this trimming is slow relative to glucose cycling, it accumulates over time: a protein that folds quickly exits before much mannose trimming occurs, whereas a chronically misfolded protein accrues progressively trimmed glycans. This converts glycan structure into a rough timer of "how long has this protein been trying to fold."

Step 6 — ERAD: recognition, retrotranslocation, and degradation. Terminally misfolded proteins bearing the trimmed-mannose signal are recognized by lectins OS-9 and XTP3-B, which deliver them to membrane-embedded E3 ubiquitin ligase complexes (in mammals, HRD1 and gp78 are central hubs). The client is threaded backward through the ER membrane — retrotranslocation (or "dislocation") — via channels formed in part by the ligase complex itself (HRD1 can form part of the conduit), a process requiring the cytosolic AAA+ ATPase p97/VCP, which pulls the polyubiquitinated substrate out using ATP hydrolysis. Once in the cytosol, N-glycans are removed by peptide:N-glycanase (PNGase), and the deglycosylated, ubiquitinated protein is degraded by the 26S proteasome.

Step 7 — Systemic feedback: the unfolded protein response. If unfolded protein accumulates faster than the folding/ERAD system can handle — "ER stress" — three ER-transmembrane sensors (IRE1, PERK, ATF6) trigger the unfolded protein response, transcriptionally upregulating chaperones, ERAD components, and, if stress is unresolved, apoptotic pathways. This links the moment-to-moment glycan cycle described above to whole-cell decisions about growth, secretion, and survival — a preview of what becomes central in The Secretory Pathway Overview.

Tunicamycin blocks N-glycosylation and reveals its role. The antibiotic tunicamycin inhibits GlcNAc phosphotransferase, the first enzyme in the dolichol-linked oligosaccharide assembly pathway, preventing formation of the lipid-linked glycan donor and thus blocking all N-glycosylation. In classic experiments through the 1970s–80s (building on work by Stuart Kornfeld, Armando Parodi, and others mapping the dolichol pathway), treating cells with tunicamycin was shown to produce unglycosylated forms of secretory and membrane glycoproteins that frequently misfolded, aggregated in the ER, and failed to be secreted efficiently — even though the polypeptide sequence itself was unchanged. This demonstrated that the glycan is not merely a surface marker added late, but is functionally required, at the time of synthesis, for correct folding and export of many glycoproteins.

Discovery and characterization of glucosidases I and II defined the trimming order. Biochemical pulse-chase studies in the late 1970s and 1980s (Robert Trimble, Armando Parodi, Stuart Kornfeld and colleagues) tracked the fate of the Glc₃Man₉GlcNAc₂ glycan after transfer, using cells and cell-free microsomal systems, combined with specific glucosidase inhibitors such as castanospermine and deoxynojirimycin. These inhibitors blocked trimming at defined steps and caused glycoproteins bearing untrimmed (tri-glucosylated) glycans to accumulate, establishing the strict order of trimming (Glucosidase I removes the outermost glucose first, then Glucosidase II removes the inner two) and showing that this trimming, not glycosylation itself, generates the calnexin/calreticulin ligand.

Ari Helenius and colleagues established calnexin as a glycan-dependent chaperone. In a series of experiments in the early 1990s, Helenius's group (and independently David Williams's laboratory) showed that calnexin associates transiently with newly synthesized glycoproteins — including influenza hemagglutinin and the T-cell receptor — in a manner that depended on the presence of the mono-glucosylated glycan. Using glucosidase inhibitors (castanospermine, which prevents trimming to the monoglucosylated form) they showed that blocking trimming prevented calnexin association entirely, while forcing glycoproteins to remain glucosylated (using inhibitors of glucosidase II) trapped them on calnexin indefinitely. This provided direct causal evidence that calnexin binding is glycan-structure-specific, not simply protein-specific, and that release requires deglucosylation.

UGGT was identified as the folding sensor using in vitro reglucosylation assays. Armando Parodi's laboratory, working initially in trypanosomes and then extending to mammalian systems, showed that a purified glucosyltransferase activity (UGGT) would selectively reglucosylate denatured glycoproteins but not their natively folded counterparts, using UDP-glucose as sugar donor in cell-free assays. This was a decisive experiment: it showed the cell possesses an enzyme that can directly discriminate folded from unfolded conformations of the same polypeptide backbone, biochemically confirming the "iterative proofreading cycle" model rather than a simple time-delay model.

ERAD and retrotranslocation were established through yeast genetics and in vitro reconstitution. Genetic screens in Saccharomyces cerevisiae in the 1990s (work from the Sommer, Wolf, and Rapoport laboratories) identified HRD (HMG-CoA reductase degradation) and DER (degradation in the ER) genes required for turnover of misfolded ER proteins, including the mutant secretory protein CPY* (a misfolded carboxypeptidase Y variant used as a standard ERAD substrate) and mutant forms of the yeast mating pheromone receptor. Cell-free reconstitution experiments — following radiolabeled substrate from the ER lumen, through membrane extraction, to cytosolic proteasomal degradation — combined with ATP-depletion and proteasome-inhibitor (e.g., lactacystin, MG132) controls, demonstrated that misfolded proteins are physically retrotranslocated out of the ER and degraded by the ubiquitin-proteasome system, not by an ER-lumenal protease, resolving a long debate about where ER quality-control degradation actually occurs.

Example 1 — Predicting the fate of a glycosylation-site mutant. A membrane glycoprotein has three N-linked sequons: Asn53 (Asn-Leu-Thr), Asn110 (Asn-Pro-Ser), and Asn240 (Asn-Val-Ser). A researcher mutates Asn110 to Gln (N110Q) and expresses the protein in cultured cells, then pulses with radiolabeled methionine and chases, analyzing glycoform by SDS-PAGE mobility shift and pulse-chase turnover. Reasoning: first check the sequon rule. Asn-Pro-Ser is not a valid consensus sequon because proline in the X position disrupts the turn geometry required for OST recognition — so Asn110 was likely never glycosylated in the wild-type protein anyway, meaning N110Q should have no effect on glycan number or electrophoretic mobility compared to wild-type. If, however, the experimentalist mistakenly assumed all three Asn residues were modified and instead mutated Asn53 (a genuine N-X-S/T sequon, X=Leu), the consequence would differ: the mutant would run with faster mobility (one fewer ~2–3 kDa glycan), and — critically — if Asn53's glycan happened to be required for calnexin engagement to fold a nearby disulfide-bonded domain, the mutant might show increased ER retention, delayed maturation kinetics in the pulse-chase, and eventual ERAD-mediated turnover instead of Golgi export. The key lesson: not all consensus sequons are functional (X = Pro is excluded, and even valid sequons are sometimes skipped due to local structure), so predicting a phenotype requires checking real sequence context, not just counting N-X-S/T motifs.

Example 2 — Interpreting a clinical glycoprotein folding disorder. A patient presents with liver damage and low serum levels of a plasma glycoprotein, alpha-1 antitrypsin (A1AT), despite normal-to-elevated mRNA levels in hepatocytes (measured by RT-qPCR) and normal translation rates (measured by polysome profiling). Liver biopsy shows PAS-positive, diastase-resistant globular inclusions inside hepatocytes — consistent with intracellular aggregates. Genetic testing reveals the patient is homozygous for the Z allele (Glu342Lys), the classic mutation underlying alpha-1 antitrypsin deficiency. Reasoning: normal transcription and translation rule out a synthesis defect; the PAS-positive inclusions localized to the ER (confirmable by co-staining with an ER marker such as calnexin or BiP) indicate a folding/trafficking defect rather than a lysosomal storage disease. The Glu342Lys substitution destabilizes a loop-sheet region of A1AT, slowing its native fold and causing polymerization of partially folded monomers via domain-swapping — the mutant protein spends abnormally long engaging the calnexin/calreticulin cycle, accumulates trimmed-mannose glycans, and a fraction is degraded by ERAD (explaining reduced serum secretion), while another fraction escapes degradation and aggregates within the ER lumen (explaining the PAS+ inclusions and hepatocyte toxicity). This dual outcome — reduced circulating (functional) protein plus toxic ER accumulation — is why A1AT deficiency causes both emphysema (loss-of-function: too little circulating antitrypsin fails to inhibit neutrophil elastase in the lung) and liver cirrhosis (gain-of-toxic-function: aggregates damage hepatocytes), a combination that only makes sense once you know the disease is fundamentally a protein-folding/ER-quality-control disorder, not a simple loss-of-function mutation.

Ari Helenius built much of the conceptual and experimental foundation for calnexin/calreticulin-mediated ER quality control. Trained originally in virology, Helenius became interested in how enveloped virus glycoproteins — obligate ER clients — fold and traffic, using influenza hemagglutinin and Semliki Forest virus glycoprotein as tractable, abundant, well-characterized model substrates. His laboratory's pulse-chase and glucosidase-inhibitor experiments in the late 1980s and early 1990s converted calnexin from an obscure ER membrane protein of unknown function into the central lectin chaperone of a defined biochemical cycle, and helped establish the general principle that glycan trimming states carry folding-status information — a concept subsequently extended by Armando Parodi's biochemical work identifying UGGT as the discriminating sensor enzyme.

Cystic fibrosis therapeutics. The ΔF508 CFTR mutation causes the chloride channel to fold too slowly and be retained/degraded by ER quality control rather than losing intrinsic channel activity. This insight directly motivated "corrector" drugs (e.g., lumacaftor, elexacaftor, part of the combination therapy Trikafta) that stabilize the mutant fold enough to pass ER quality control and reach the plasma membrane, where separate "potentiator" drugs (e.g., ivacaftor) then restore channel gating.

Alpha-1 antitrypsin deficiency and serpinopathies. As worked through above, understanding that Z-allele A1AT is retained via calnexin cycling and aggregates in the ER has guided strategies to enhance ERAD clearance of toxic polymers or to chemically chaperone the mutant fold, rather than simply trying to replace enzyme activity.

Cancer therapeutics exploiting ER stress. Multiple myeloma cells produce enormous quantities of immunoglobulin and are exquisitely dependent on ER folding capacity and ERAD/proteasome flux; proteasome inhibitors (bortezomib, carfilzomib) exploit this dependency by overwhelming ERAD-linked degradation, triggering terminal UPR-driven apoptosis selectively in these secretory-overloaded cells.

Vaccine and biologics manufacturing. Recombinant glycoprotein production (monoclonal antibodies, subunit vaccines) in mammalian or yeast expression systems must account for calnexin/calreticulin cycling kinetics and glycosylation site occupancy, since incomplete glycosylation or chaperone bottlenecks reduce yield, alter glycan-dependent immunogenicity, and can trigger unwanted ER-stress responses in production cell lines.

Diabetes and Wolfram syndrome. Pancreatic beta cells synthesizing large amounts of proinsulin are highly sensitive to ER folding load; chronic ER stress and UPR activation are implicated in beta-cell dysfunction and death in some forms of diabetes, and mutations in ER stress-response genes (e.g., WFS1) cause Wolfram syndrome, directly linking ER quality control machinery to human monogenic disease.

Ask the tutor: How does the calnexin/calreticulin cycle differ mechanistically from the general Hsp70 (BiP)-based chaperone system also operating in the ER?