MUC2 is the first human secretory mucin to be identified and characterized (21). The major building block of colonic mucus is MUC2, a glycoprotein consisting of ~5,200 amino acids and ~80% glycans, mostly O-glycans (22). The MUC2 protein has three von Willebrand D domains (VWD), a VWD' domain (lack of an E module compared with VWD), a central high frequency of hydroxy-amino acids (Ser and Thr) together with a Pro (PTS) sequence interrupted by two CysD domains and a C-terminal with VWCN (a globular structure with a VWC fold), one VWD and three von Willebrand C domains (VWC) with a VWC' domain (half of a C domain compared with VWC) followed by a cystine-knot (CK) domain (Fig. 1A) (23). Membrane-tethered and secreted non-gel-forming mucins do not contain these important polymerizing domains, such as VWD and PTS, and thus do not form gels (24).

Cysteine residues located at the N- and C-termini of MUC2 are highly glycosylated, which increases the hydrophilicity of MUC2 and allows lubrication of the intestinal mucosa (25). MUC2 contains 215 cysteine amino acids that make up >10% of the total amino acids outside the PTS mucin domains (26). As all cysteines need to be interlinked (oxidized) to exit the ER, the correct assembly of MUC2 is a formidable challenge (26). A MUC2 monomer contains up to 1,600 O-glycans and 30 N-glycans, resulting in more than 3,300 terminal sugar residues (27).

MUC2 monomers are N-glycosylated within the ER, a process that enables correct MUC2 folding, stable dimer formation and maturation (22). The CK domain in the C-terminus of MUC2 forms a dimer via disulfide bonds between three cysteine amino acids via the ER machinery (Fig. 1B) (28). The correct folding of MUC2 is assisted by various molecular chaperones that prevent the misfolding and aggregation of proteins in the ER, such as binding immunoglobulin protein (BiP, GRP78), anterior gradient 2 (AGR2), calnexin and calreticulin (29).

Next, MUC2 dimers are transported to the Golgi, and the hydroxy amino acids (Ser and Thr) of the PTS domains become O-glycosylated, in which >80% of all Ser and Thr residues carry O-linked glycans (30). The typically O-glycosylated compounds are largely concentrated in the PTS domains, which resemble bottle brushes with protein cores at their centers and oligosaccharides as their bristles (31). In the trans-Golgi network, the glycosylated dimers are then trimerized via disulfide bonding in VWD3, and a net-like structure is generated (Fig. 1B) (32). The glycosylated MUC2 monomer has a mass of ~2.5 MDa and the polymer may have a mass of >100 MDa (33). An important function of O-glycans is to cover the protein backbone and thus protect the mucin from proteolytic enzymes while simultaneously binding water to generate a gel (33). Glycans also act as attachment sites for bacteria, as they are important food sources (34). An increase in the number of small glycans is found in most patients with active UC, and a decrease in the number of more complex compounds has been detected (31).

The distribution of glycosyltransferases (GalNAc-Ts) throughout the Golgi apparatus is not homogenous, and each compartment has a distinct composition of GalNAc-Ts (35). To achieve controlled, sequential extension, the GalNAc-Ts are spatially arranged according to the order of monosaccharide addition (27). First, O-glycosylation is a covalent linkage initiated by 1 of the 20 GalNAc-Ts that add N-acetylgalactosamine (GalNAc) to the Ser or Thr amino acid residues in the MUC2 PTS sequence (36). The Pro amino acids in the PTS ensure a non-folded structure allowing the GalNAc-Ts of the Golgi apparatus to access the protein core (37). The structures formed during this initial stage of modification are known as Tn antigens. Next, the glucan chains are further extended to form four core structures known as Core 1–4 (Fig. 1C) (38). The addition of galactose (Gal) by Core 1 β1-3-galactosyltransferase (C1GalT1) is known as core 1 (39). The addition of N-acetylglucosamine (GlcNAc) by Core 3 β1-3-N-acetylglucosaminyltransferase (β3GnT6) to peptide-bound GalNAc is known as Core 3 (40). Core 2 is formed by the addition of β1-6GlcNAc to the GalNAcs of Core 1 through Core 2 β1,6-N-acetylglucosaminyltransferase 1/3 (C2GnT1/3) (41). Core 4 is formed by the addition of β1-6GlcNAc to the GalNAc of Core 3 through Core 2 β1,6-N-acetylglucosaminyltransferase-2 (C2GnT2) (38). Groups that can be attached to core structures include galactose, sialic acid, sulfate and fucose (42). The order in which the glycans interact with different GalNAc-Ts plays a critical role in O-glycosylation since specific monosaccharides, such as fucose or N-acetylneuraminic acid, can limit further elongation (27). There are large differences in MUC2 characteristics between animal species, such as mouse colonic MUC2, which largely contains Core 2 and minor amounts of Core 3 and Core 4, whereas human MUC2 predominantly contains Core 3 structures, Core 4 to a lesser extent, and essentially lacks Core 1 (42).

MUC2 forms large net-like structures that are densely packed in the secretory granules of goblet cells (43). Fully synthesized MUC2 is released from goblet cells through baseline secretion or via Ca²⁺⁻dependent stimulated secretion (44). The pH along the secretory pathway shifts gradually from 7.2 in the ER to 6.0 in the trans-Golgi network, and to 5.2 in the secretory granule (45). Moreover, the intragranular Ca²⁺ concentration increases, suggesting that the packing of MUC2 may be pH- and Ca²⁺-dependent (45). The N-terminal trimers spontaneously form concatenated rings with long-extended mucin domains standing perpendicular to this structure and joining to other MUC2 molecules end-to-end through their C-terminus (46).

After secretion, the expression of Na⁺/H⁺ exchanger 3 (NHE3), the dominant isoform of the apically expressed Na⁺/H⁺ exchangers at the surface of epithelial cells contributes to an acidic mucosal milieu (47). NHE3 transports luminal sodium in exchange for cellular hydrogen and helps facilitate a pH gradient at the brush border as hydrogen diffuses out of the unstirred layer (10). The mucus is then converted to a more voluminous, loose outer mucus layer with increasing pH and removal of N-terminus-bound single calcium ions (10). Upon secretion, the densely packed mucin expands >1,000-fold into a large net (48). In general, the outer layer is twice as thick as the inner layer (49). The inner mucus layer forms a barrier impervious to bacteria and thus protects the colonic epithelium, which is essentially sterile (30).

In addition, the hydrolytic activities of some proteases also play a role in the process of MUC2 volume expansion (50). Calcium-activated chloride channel regulator 1 is highly abundant in intestinal mucus, with an N-terminal metalloprotease domain, a central von Willebrand A domain and a C-terminal fibronectin type III domain that likely has both metallohydrolase activity and structural roles and can process N-terminal MUC2 via proteolysis (51). Transglutaminase 3, the dominant cross-linking enzyme in the mouse colon, catalyzes cross-linking of the MUC2 protein and increases the resistance of MUC2 to degradation (52). Under the synergistic effects of numerous factors, MUC2 forms a net-like structure and establishes a sturdy mucus barrier.

IRE1 proteins, which include IRE1α and IRE1β, possess both an ER luminal sensor domain and a cytosolic endoribonuclease (RNase) domain and show kinase activity (66). Kinase activity enables trans-autophosphorylation, which activates RNase activity (67). Activated IRE1 splices X-Box Binding Protein 1 (XBP1) mRNA and removes an inhibitory 26-nucleotide intron from the XBP1 transcript (68). The mRNA is subsequently re-ligated by RNA terminal phosphorylase B, generating a functionally active isoform of XBP1 (XBP1s) (69). XBP1s bind to ER stress response element (ERSE), ER stress response element II (ERSE II) and UPR element sequences (70). This process upregulates the expression of genes that encode proteins that promote protein refolding (ER chaperone proteins), aid in the destruction of proteins that are beyond repair (components of the ERAD machinery) and encode proteins that facilitate the expansion of the ER, thus increasing protein folding capacity (71). In addition to splicing XBP1, IRE1 RNase activity can cleave other RNA targets in a process known as regulated IRE1-dependent decay (72).

PERK is a transmembrane protein located in the ER, and its N-terminal regulatory motif is located in the lumen and adjoins the cytosolic eIF2α kinase domain (73). PERK is maintained in an inactive state by the interaction of its ER luminal domain with BiP (74). Once a misfolded protein is produced, BiP dissociates from PERK due to its preferential binding to hydrophobic residues of misfolded proteins (75). Once disassociated, following phosphorylation at Thr980 by autophosphorylation, PERK dimerizes to form an active homodimer (76). The activation of PERK leads to the phosphorylation of eukaryotic initiation factor 2 (eIF2α) at Ser51 (77). Phosphorylated eIF2α perturbs 80S ribosome assembly inhibiting protein translation, thus blocking the production of an additional influx of nascent polypeptides that could worsen the ER stress (78).

Moreover, activating transcription factor 4 (ATF4) escapes translational attenuation by eIF2α phosphorylation because ATF4 has upstream open reading frames (ORFs) in its 5′-untranslated region (79). These upstream ORFs prevent translation of ATF4 under normal conditions and are bypassed only when eIF2α is phosphorylated; therefore, ATF4 translation occurs (80). ATF4 translation activates the expression of the transcription factor CCAAT/enhancer binding protein-homologous protein, a master regulator of ER stress-induced apoptosis and plays pro-apoptotic roles in the stress response (81). Chronic or prolonged PERK signaling can lead to goblet cell apoptosis in IBD (82).

ATF6 proteins are type II transmembrane proteins that contain basic leucine zipper (bZIP) motifs in their cytosolic domains and a C-terminus protrudes that into the ER lumen (83). After disassociation from BiP under ER/oxidative stress conditions, ATF6 translocates from the ER to the Golgi apparatus, where it is cleaved by site-1 protease (S1P) and site-2 protease (S2P) to remove the luminal and transmembrane domains (84). This process results in the generation of a cytosolic fragment with transcription factor activity which is designated as ATF6f, the released ATF6f then translocate to the nucleus and regulates the expression of genes encoding BiP, XBP1s and ERAD components (85).

Unlike IRE1α, which plays a broad role in the UPR by splicing XBP1 mRNA and degrading misfolded proteins, IRE1β has specialized functions in secretory epithelial cells, especially in goblet cells (86). Vertebrates express two IRE1 paralogs, IRE1α, which is universally expressed and IRE1β, which is specifically expressed within mucus-secreting cells in the respiratory and gastrointestinal tracts (87). Single-cell RNA sequencing revealed that the abundance of IRE1β mRNAs is ~50-fold greater than that of IRE1α mRNAs in the goblet cells of the small intestinal epithelium (88). Loss of IRE1β expression results in the accumulation of misfolded MUC2 precursor proteins in the ER of immature goblet cells, and defects in MUC2 maturation and its accumulation in the ER lead to marked ER abnormalities and signs of ER stress (89).

The role of IRE1β in intestinal mucus barrier homeostasis can't be replaced by IRE1α (90). The transplantation of microbiota from conventionally raised mice to germ-free mice restored goblet cell numbers; this restoration was completely abolished in IRE1β-deficient mice, despite normal IRE1α expression, but it failed to compensate for IRE1β function (91). Compared with normal colonic tissues, the tumor tissues exhibited a significant increase in XBP1s mRNA expression levels, no significant difference in IRE1α mRNA expression, and a significant decrease in both IRE1β and MUC2 mRNA and protein levels (92). Moreover, the expression levels of 23 genes associated with the mechanistic target of rapamycin complex 1 signaling pathway were increased in the IRE1β-knockout intestinal epithelial cells (IECs) compared with IRE1α knockout IECs in both an ER stress mouse model treated with tunicamycin, and a colitis mouse model induced by 2.5% DSS (93). These findings indicated that, unlike IRE1α, IRE1β exerts a novel non-canonical splicing target gene unique in IECs.

AGR2 selectively binds to IRE1β, but not IRE1α. AGR2, a member of the protein disulfide isomerase (PDI) family with redox and molecular chaperone functions that process and modify immature mucins into mature mucins, is highly expressed in goblet cells (94). AGR2 forms disulfide bonds with the N- and C-terminal cysteine-rich sections of MUC2 (94). Impaired AGR2 function, such as H117Y mutation and S-glutathiolation modification, directly suppresses MUC2 biosynthesis, inducing ER stress and driving IBD pathogenesis (58,95). Bio-Layer Interferometry revealed a half-maximal binding concentration of 19 µM for binding of the IRE1β luminal domain (LD) to AGR2, whereas an immobilized IRE1α LD resulted in significantly weaker binding signals with AGR2 (96). The interaction between IRE1β and AGR2 does not involve disulfide bonds, and AGR2 binding favors the formation of IRE1β LD monomers (96). Both the C81S and H117Y mutations in AGR2 abrogate its ability to bind and inhibit IRE1β activity (97). Importantly, both the N-terminus (residues 21–1,397) and the C-terminus (residues 4,198-6,178) of MUC2 have a derepression effect on the IRE1β-AGR2 pair, and the derepression effect on IRE1α is relatively weak under the same conditions (96).

These results suggest that the IRE1β-mediated UPR likely works by binding with AGR2 in a manner such as that of BiP and IRE1α, by antagonizing IRE1β LD dimerization (Fig. 2) (98). The IRE1β-AGR2 pathway may be an important UPR pathway in goblet cells.