Essential roles of the unfolded protein response in intestinal physiology

Abstract

The intestinal epithelium serves as an essential interface between the host and microbiota, regulating innate and adaptive immunity, absorption of nutrients and systemic metabolism, and mediating bidirectional communication with the nervous system. The intestinal epithelium suffers constant challenges to the proteostasis machinery due to its exposure to the dynamically changing and microbial laden lumenal gut environment and to the high secretory demand placed on multiple epithelial cell types to accommodate gut and systemic physiology—especially goblet, enteroendocrine and Paneth cells. In all cases, intestinal cells require an active unfolded protein response (UPR) to sustain their physiological function, the main pathway that monitors and adjusts secretory function changes in the environment. A specialised endoplasmic reticulum (ER) stress sensor uniquely expressed in epithelial cells lining mucosal surfaces, termed inositol-requiring transmembrane kinase/endoribonuclease β, has specific roles in intestinal epithelial homeostasis, regulating mucus production and communication with microbiota. Chronic ER stress or genetic mutations affecting key UPR mediators contribute to the occurrence of inflammatory bowel disease and ulcerative colitis, in addition to colon cancer. Here, we review recent advances linking the UPR and ER stress with gut physiology and intestinal disease. Therapeutic strategies to alleviate ER stress or enforce UPR function to improve intestinal function in ageing and in bowel diseases are also discussed.

Introduction

The intestine is a large muscular tube that, in addition to being essential for food digestion and absorption of nutrients and water, serves as a selective barrier separating the host from the environment and as an active modulator of microbial populations colonising the gut mucosa. A large number of nerves emanating from the mesentery innervate the wall of the intestine, regulating peristaltic movements, in addition to other specialised functions of the gut and mucosal immune system. Epithelial cells lining the small and large intestine also produce various hormones and metabolites that signal to other parts of the body, controlling the release of digestive enzymes from the pancreas, bile acids from the liver/gall bladder, glucose metabolism systemically and appetite/satiation responses originating from the brain.

The intestinal epithelium is a specialised single cell-thick layer of polarised cells organised in a three-dimensional structure consisting of tips (villi) that alternate with invaginations (crypts) in the small intestine. The small intestine is formed by four layers, including mucosa, submucosa, muscle layer and adventitia. The cell types lining the lumen of the intestinal epithelium include goblet cells, Paneth cells, enterocytes, enteroendocrine cells, and stem cells, among others.¹ Intestinal epithelial cells (IECs) have essential roles in the absorption of nutrients and serve as physical barriers that separate commensal microbiota from the host environment.² Intestinal stem cells in the small and large intestine constantly replicate to generate the cellular components of the intestinal epithelia through a continuous cycle of renewal every 3 to 5 days in human.³

The lumenal contents of the gastrointestinal tract represent a constant challenge to our immune system. Regulatory mechanisms ensure the tolerance of luminal microbiota normally colonising the gut and avoid responding to their products, but still protect the intestinal mucosa from invading pathogens and potentially harmful components of the diet. IECs provide the first line of defence against noxious luminal agents, where epithelial cells prime and signal to professional immune cells to promote effective responses against pathogens.⁴ IECs are generated from continuously dividing intestinal stem cells, which need to be protected from the severe environment inside the intestinal lumen to ensure constant regeneration of the epithelium.⁵ After being formed at the crypt base, the newly generated transitional epithelial cells migrate upwards and out of the crypt into the lumen along the villus, maturing according to their specific differentiation programmes to acquire specialised functions during this transit. The cycle ends when differentiated IECs undergo apoptosis and are shed into the intestinal lumen.

Stem cells generate all of the different cell types that form the intestinal epithelium¹ including (1) enterocytes, the most abundant cells of the intestinal epithelium involved primarily in barrier function, nutrient transport and water and solute intake from the intestinal lumen into the body; (2) Goblet cells, specialised secretory cells responsible primarily for mucus production and transport of lumenal contents to the mucosal immune system; (3) enteroendocrine cells, hormone-producing cells that coordinate intestinal functions with the rest of the body and within the epithelial barrier; (4) Paneth cells, which produce antimicrobial peptides and provide favourable niches for intestinal stem cells; (5) microfold cells, which partially line the lumenal surface of peyers patches and like goblet cells transport lumenal antigens across the epithelial barrier to the mucosal immune system cells and (6) tuft cells, chemosensory cells that produce proinflammatory molecules on detection of pathogens (figure 1). Secretory progenitor differentiates from the stem cell niche and gives rise to Paneth cells, goblet cells and enteroendocrine cells. Proteostasis is constantly challenged at the intestinal epithelium because of many factors including the highly secretory nature of multiple components of the intestinal epithelium, the constant demands of cellular differentiation, the dynamically changing gut lumenal environment affecting cell state, including other challenges like mechanical stress, pH and pathogen invasion. This produces a constant demand on the proteostasis network of these and all other cell types forming the epithelial barrier.⁶ Notably, the high demand on protein production in secretory cell types can naturally result in chaperone/client-protein imbalance within the endoplasmic reticulum (ER), the origin of the secretory pathway, and the accumulation of misfolded proteins—often referred to as ER stress. To accommodate this demand, the IECs can activate the unfolded protein response (UPR), a dynamic signal transduction pathway that adjusts protein production according to need by controlling almost every aspect of the secretory pathway.⁷ In doing so, the UPR regulates protein translation, in addition to enforcing a global transcriptional reprogramming that upregulates many components of the proteostasis network including genes encoding for chaperones, and proteins participating in quality control mechanisms, protein degradation pathways (ER-associated degradation (ERAD), autophagy), protein translocation into the ER, secretion and organelle biogenesis.⁸ The UPR also regulates the redox environment at the ER and upregulates components involved in protein maturation, including disulphide bond formation, glycosylation and other post-translational modifications. This is particularly relevant for mucus production since mucins are highly complex proteins that may misfold during normal biogenesis or on cell stress.⁹ In this review, we discuss the emerging functions of the UPR in intestinal physiology, stem cell renewal and the involvement of ER stress in diseases affecting the gut. We speculate about the possible involvement of the UPR in the communication between the nervous system and the intestine. Possible therapeutic interventions to reduce ER stress in IECs as a strategy to treat inflammatory bowel diseases (IBDs) are also discussed.

Figure 1. Composition and specialised functions of intestinal epithelial cells (IECs). Crypt-base-columnar cells are continuously dividing intestinal stem cells that generate IECs, which are composed by distinct specialised cell types that play different functions: Enterocytes (in the small intestine) known as colonocytes in the colon, are the most numerous and function primarily for terminal digestion of nutrient substrates, and nutrient, ion and water absorption. Enterocytes express many catabolic enzymes on their exterior luminal surface to break down molecules to size appropriate for uptaking into the cell and solute transporters to mediate transepithelial absorption. Examples of molecules taken up by enterocytes are: ions, water, simple sugars, vitamins, lipids, peptides and amino acids. Single-cell studies discovered several types of enterocytes with distinct functions at specific positions along the crypt-villus axis. Goblet cells secrete mucin as a primary function to create a protective mucus layer which protects the epithelium from the luminal contents. Goblet cells are also involved in immunoregulation, since they can internalise and deliver luminal antigens to dendritic cells to induce tolerance. Goblet cells are the most numerous among the secretory cell types lining the mucosal surface. Enteroendocrine cells secrete hormonal products, they regulate intestinal motility, satiety, insulin secretion, immune responses, or release of digestive enzymes. For example, enteroendocrine cells secrete the gastrointestinal hormones secretin, pancreozymin and entero-glucagon among others. Subsets of sensory IECs synapse with nerves and are known as neuropod cells. Paneth cells produce antimicrobial peptides and proteins and other components that are important in host defence and immunity such as human alpha-defensin. Paneth cells contain secretory granules that are filled with antimicrobial agents (lysozyme, alpha-defensins, and phospholipase A2) and secreted at basal levels. Proinflammatory stimuli (ie, IFN gamma drastically increase their secretion. Paneth cells also contribute at the base of small intestinal crypts to the primary niche for intestinal stem cells. Microfold cells or M cells sample antigens from the lumen and deliver them to the lymphoid tissue associated with the mucosa. In the small intestine, M cells are associated with Peyer’s patches, secondary myeloid organs involved in immune surveillance. Tuft cells are low-abundant IEC components that play parts in the immune response, and are involved in chemosensation. They are closely related to taste receptor cells and use their chemosensory ability to initiate type II immune responses in the intestinal epithelium on detection of parasites. Stem cells are crypt-base-columnar cells that are continuously dividing to generate all other epithelial cell types. They reside exclusively at the bottom of crypts wedged between Paneth cells. Intestinal stem cells divide every day and produce two equipotent daughter cells.

A brief overview of the UPR

Different perturbations can alter the function of the ER, resulting in chaperone/client-protein imbalance and the accumulation of misfolded or immature protein cargoes of the secretory pathway, which on average represents 30% of the total proteome. ER stress is detected by five distinct transmembrane signal transducers, inositol-requiring transmembrane kinase/endoribonuclease (IRE1) alpha and beta, activating transcription factor-6 (ATF6) alpha and beta, and PKR-like ER kinase (PERK) (figure 2).¹⁰ The ER also functions as the site for biosynthesis of all membrane lipids, and dysfunctions in these pathways can be detected by one or more of the ER stress sensors. Depending on the duration and intensity of the ER stress stimuli, the activation of these UPR sensors orchestrates downstream signalling events and transcriptional responses to restore proteostasis or induce apoptosis, thus determining cell fate under stress. In this section, we summarise the main signalling mechanisms that mediate the UPR to then discuss their implications in intestinal physiology and human disease.

Figure 2. Overview of unfolded protein response (UPR) signalling. The UPR is triggered by three key signal transducers: inositol-requiring enzyme 1 (IRE1) alpha and beta, protein kinase RNA-like ER kinase (PERK) and activating transcription factor-6 (ATF6) alpha and beta. The most evolutionarily conserved branch of the UPR is initiated by the stress sensor IRE1α (termed IRE1 here). This protein, located in the endoplasmic reticulum (ER), functions as both a kinase and an endoribonuclease. On activation, IRE1α splices the mRNA of the transcription factor X-Box-binding protein 1 (XBP1), removing a 26-nucleotide intron. This splicing shifts the reading frame, resulting in the production of XBP1s, a potent transcriptional activator that upregulates various proteostasis effectors. Additionally, IRE1 mediates the direct degradation of certain RNAs via Regulated IRE1-Dependent Decay of RNA (RIDD). PERK acts as a kinase within the integrated stress response, phosphorylating eIF2α to reduce general protein synthesis. However, the phosphorylation of eIF2α selectively increases the translation of ATF4, a transcription factor that regulates genes involved in protein folding, metabolism, and apoptosis. Phosphatases CReP and GADD34 dephosphorylate eIF2α, allowing the resumption of protein translation. Under ER stress, ATF6α, a transmembrane transcription factor, is processed in the Golgi apparatus. This processing releases the cytosolic domain (ATF6f), which then functions as a transcription factor to promote adaptive cellular responses. ERAD, ER-associated degradation.

IRE1α is a ubiquitously expressed protein, whereas IRE1β expression is restricted to the intestinal and lung epithelia (see next sections). IRE1α is a type-I transmembrane protein that contains a kinase and RNase domain in the cytosolic region. On activation, IRE1 oligomerises and auto-phosphorylates activate its endoribonuclease activity through a conformational change. IRE1α cleaves and releases a 26-nucleotide intron from the mRNA encoding X-Box Protein-1 (XBP1), followed by ligation-mediated by the RNA 2',3'-cyclic phosphate and 5'-OH ligase (RtcB). This processing event shifts the coding reading frame with the resultant expression of an active and stable transcription factor termed XBP1s.⁷ XBP1s translocates to the nucleus and engages adaptive transcriptional programmes, upregulating many components of the proteostasis network (ie, ER chaperones and foldases, proteins involved in ER and Golgi biogenesis, ER-to-Golgi vesicle trafficking, quality control mechanisms and ERAD (figure 2). The RNase domain of IRE1 is also involved in RNA degradation (known as regulated IRE1-dependent decay or Regulated IRE1-Dependent Decay of RNA (RIDD)), which controls the stability of various RNAs in a tissue-dependent manner. IRE1α may directly detect the presence of misfolded proteins in the ER¹¹ and is negatively regulated by different ER foldases (including BiP,¹² protein disulphide isomerases¹³ and the collagen carrier Hsp47).¹⁴ IRE1 may be capable of sensing lipid imbalances at the ER membrane.¹⁵

PERK is a type-I transmembrane serine/threonine kinase with a cytosolic protein kinase domain. On activation, PERK phosphorylates the translation initiation factor eIF2α on S51, blocking protein translation to reduce the load of proteins entering the ER.⁷ This is accompanied by an enhanced translation of activating transcription factor 4 (ATF4). Under ER stress, eIF2α phosphorylation allows the bypass of inhibitory upstream open reading frames to engage the downstream canonical AUG of the ATF4 coding region.¹⁶ ATF4 controls the expression of genes involved in amino acid biosynthesis and transport, ER chaperones, autophagy and antioxidant responses, in addition to the proapoptotic C/EBP homologous protein (CHOP) transcription factor. Furthermore, CHOP induces the expression of growth arrest and DNA damage 34 (GADD34), a regulatory subunit of protein phosphatase 1c, thus dephosphorylating S51 of eIF2α to allow protein synthesis to resume (figure 2).¹⁶ PERK may be activated by the binding of misfolded proteins or the dissociation of BiP on ER stress.

ATF6 is a 90 kDa type II ER resident transmembrane protein with a basic leucine zipper domain (bZIP) and a transcriptional activation domain present in its cytosolic region. There are two homologs of this sensor, ATF6α and ATF6β, both of which are ubiquitously expressed and have redundant functions. Under normal conditions, disulphide-bonded ATF6 (both α and β) remains in a monomeric and dimeric form occupied by the abundant HSP70 chaperone BiP. Under ER stress, BiP dissociates from the C-terminal luminal region, allowing modification by protein disulphide isomerases (PDIs) and the translocation of full-length ATF6 to the Golgi apparatus, where it is cleaved by site-1 (S1P) and site-2 (S2P) endoproteases. Such cleavage releases the N-terminal 50 kDa bZIP fragment ‘ATF6p50’ (also termed ATF6f) to the cytosol. ATF6f translocates to the nucleus to induce the expression of various ER chaperones and ERAD components among other proteostatic regulators (figure 2). ATF6 may also sense alterations in the lipid content of the ER membrane.

The use of genetically modified mice for essential UPR components (stress sensors and downstream transcription factors) has revealed the fundamental roles of the pathway in the acquisition and maintenance of specialised secretory phenotypes (reviewed in Cornejo et al).¹⁷ XBP1 deficiency ablates immunoglobulin secretion of plasma B cells, operating downstream of differentiation programmes controlled by the B cell receptor.¹⁸ XBP1 also controls the maturation and function of gastric zymogenic cells, promoting the expansion of the ER content.¹⁹ XBP1s regulate the secretory function of the endocrine and exocrine pancreas, in addition to salivary glands.^{7 20 21} The phenotypes of IRE1 null tissues are mild compared with XBP1s deficiency and may be explained by the overactivation of IRE1 in the absence of XBP1 that triggers abnormal levels of RIDD.^{22 23} It is proposed that the high demand for protein production in specialised secretory cells engages the UPR to adjust protein production and expand the secretory pathway. Experimental disruption of the UPR results in proteotoxicity due to the accumulation of abnormally folded proteins that have a high tendency to aggregate (ie, proteins with transmembrane segments, high hydrophobicity, complex maturation process, etc). Studies in medaka fish have demonstrated that the abnormal phenotypes observed in UPR-deficient animals are due to the abnormal deposition of collagen, a protein that undergoes a complex biosynthetic process.^{24 25} ATF4 deficient animals have reduced glutamine update by IECs, decreased expression of antimicrobial peptides, homeostasis and altered eye lenses (reviewed in Cornejo et al).¹⁷ ATF6α knockout mice do not develop spontaneous phenotypes, but they are hypersusceptible to conditions that trigger ER stress.¹⁷ However, ATF6α and ATF6β double deficiency is lethal, suggesting redundant and complementary roles.²⁶ The UPR, and specifically XBP1, is key for the secretion of cytokines in immune cells, including macrophages, dendritic cells, in addition to regulating the activity of reactive astrocytes.¹⁸ Thus, the UPR represents a central homeostatic system that sustains the function of secretory cells and organs. In the next sections, we discuss the significance of the UPR to intestinal physiology, and the role of ER stress in numerous gastrointestinal disorders, including IBD (Crohn’s disease and ulcerative colitis) and colorectal cancer.

Role of the UPR in IECs and gastrointestinal diseases

The intestine could be viewed as a massive secretory organ that is constantly challenged by a fluctuating environment and physiological demands. The constant secretion of digestive enzymes, mucins, antimicrobial peptides and hormones, in addition to the rapid cellular turnover along the crypt-villus axis, the exposure to an anaerobic microenvironment and a multitude of microbes and their toxins and metabolites imposes high pressure on the proteostasis network.⁹ Genetic evidence in humans, in addition to the manipulation of UPR components in mouse models (summarised in table 1), has revealed that impaired handling of protein misfolding has adverse effects on IEC function, where specialised secretory cells, such as goblet and Paneth cells, are highly vulnerable to ER stress (figure 3). Signs of chronic ER stress have been extensively reported in human intestinal tissue derived from patients affected with Crohn’s disease and ulcerative colitis, in addition to other pathological conditions involving the gut.⁶ In this section, we review functional evidence assessing the contribution of the UPR to IEC homeostasis and gastrointestinal diseases. In addition, the use of small molecules to improve ER proteostasis.

Table 1. Functional involvement of ER proteostasis in intestinal physiology and disease.

Pathway	Intervention	Effects on intestinal physiology	Ref
IRE1/XBP1	XBP1cKO IECs	Colitis and bacterial infection susceptibility, intestinal inflammation, basal ER stress	27
IRE1/XBP1	XBP1cKO IECs	Enhanced Th17 production and cell responses, increased ROS	33
IRE1/XBP1	IRE1cKO ILC3s	Colitis and bacterial infection susceptibility, decreased IL-17 and IL-22 production	35
IRE1/XBP1	XBP1cKO IECs	Stem cell expansion, tumour formation and colitis susceptibility	38 39
PERK/eIF2α	eIF2αDN IECs	Colitis and bacterial infection susceptibility, Paneth cell dysfunction	44
PERK/eIF2α	CHOP-/- mice	Colitis protection, reduced apoptosis, inflammation and immune cells infiltration	46
PERK/eIF2α	CHOPTg/Tg IECs	Susceptibility to colitis an intestinal inflammation, intestinal mucosa injury	47
ATF6	ATF6-/- mice	Severe colitis, apoptosis and inflammation induction, increased ER stress	56
ATF6	nATF6Tg/Tg IECs	Spontaneous colorectal tumorigenesis, intestinal dysbiosis	59
ER proteostasis	4-PBA/TUDCA	Ameliorated acute and chronic colitis, reduction of ER stress	56 63
ER proteostasis	recombinant BiP	Decreased susceptibility to colitis and immune infiltration, enhanced barrier integrity	60
ER proteostasis	recombinant CT-KDEL	Increased cell migration and wound healing, activation of IRE1α/XBP1s pathway	66 67

cKOconditional knockout miceCT-KDELcoleta toxin subunit-B-KDEL motifDNdominant negativeIECsintestinal epithelial cellsTgtransgenic mice

Figure 3. Integration of unfolded protein response (UPR) to gastrointestinal disease and endoplasmic reticulum (ER) proteostasis targeting. Intestinal epithelial cells (IECs) are exposed to a challenging microenvironment that imposes high pressure on the proteostasis network. Through physiological UPR activation, IECs restore their proteostasis to sustain their function and maintain intestinal homeostasis. However, chronic ER stress triggers non-adaptive UPR signalling that impairs proteostasis recovery. Genetic manipulation of UPR elements has pinpointed their relevance to IEC function and ER stress sensitivity. Overall, UPR abrogation and proteostasis impairment induce cellular and tissue dysfunction, favouring bacterial infection and a proinflammatory state that increases susceptibility to colitis and tumorigenesis. Interventions with small molecules to improve ER proteostasis under specific pathological models are also highlighted. ATF6, activating transcription factor-6; IRE1, inositol-requiring enzyme 1; 4-PBA, 4-phenylbutyrate; PERK, protein kinase RNA-like ER kinase; TUDCA, tauroursodeoxycholate; XBP1, X-Box-binding protein 1.

IRE1α and XBP1 regulation in disease and relevant mechanisms

The activity of IRE1α and XBP1 has been implicated in the basal function of different IECs. At resting conditions, low levels of XBP1 mRNA splicing are detected in the small and large intestine, where Paneth cells have a higher constitutive UPR.⁶ Conditionally knocking out XBP1 in the gut epithelium of animals led to spontaneous ER stress, resulting in massive loss of Paneth cells.²⁷ At the ultrastructural level, surviving Paneth cells depict lower content of secretory granules, and altered ER morphology and content. The dysfunction of Paneth cells in XBP1 deficient animals translated into poor antimicrobial responses, reflected in high susceptibility to Listeria monocytogenes.²⁷ As expected, goblet cells were also affected on targeting XBP1 in the intestine, associated with a near 30% reduction in their abundance. The basal alterations observed in XBP1-deficient IECs resulted in the spontaneous development of chronic intestinal inflammation resembling IBD,²⁷ associated with a proinflammatory status, ulcerations and infiltration of immune cells. In agreement with this, XBP1-deficient animals were more susceptible to experimental colitis induced by dextran sodium sulphate (DSS) administration.²⁷ Importantly, genetic studies in patients affected with Crohn’s disease and ulcerative colitis uncovered rare polymorphisms in XBP1 that are linked to an increased risk of developing the disease. These variants reduced the activity of XBP1, as reported by in vitro studies.²⁷ Since the deletion of just one allele of the XBP1 gene results in spontaneous enteritis, the authors suggested that IECs are highly vulnerable to minor alterations in the function of the IRE1/XBP1 pathway. Nevertheless, results obtained in XBP1-deficient animals need to be assessed with caution because this strategy triggers massive IRE1 overactivation as a compensatory feedback reaction.²⁷ In fact, studies in the liver and pancreas of XBP1-deficient animals demonstrated hyper-phosphorylated IRE1 and high levels of RIDD, which may result in artificial phenotypes that do not strictly depend on the direct role of XBP1.^{22 23} Gain-of-function studies are required using XBP1s transgenesis to complement the available data.

Alterations to other proteostasis components, such as the autophagy regulator 16-like 1 (ATG16L1), are genetic risk factors in Crohn’s disease. The secretory activity of Paneth cells is affected when the expression of ATG16L1 is ablated in IECs in mice.^{28 29} Interestingly, deficiency in ATG16L1 compensates through activation of the UPR, whereas XBP1 deficiency engages autophagy, demonstrating a tight homeostatic balance between these two proteostasis nodes.²⁸ This compensatory association between XBP1 deficiency and autophagy upregulation was also reported in the context of neurodegeneration.^{30 31} XBP1 and ATG16L1 double knockout animals develop severe spontaneous Crohn’s disease-like transmural ileitis. Further studies indicated that IEC-specific deletion of Atg16L1 in mice results in spontaneous ileitis due to IRE1α overactivation in Paneth cells,³² a phenotype exacerbated by XBP1 deficiency. In this context, Paneth cell dysfunction generates severe dysbiosis that does not trigger ileitis but enhances the susceptibility to DSS-induced colitis.

Interleukin-17-producing T helper (Th17) cells are a subtype of CD4 T cells present in the intestine that aggravate inflammatory conditions, and their development depends on adherent microbes in the gut. ER stress in IECs was shown to regulate microbially induced Th17 development.³³ XBP1 deficiency in IECs enhanced Th17 cell production even when microbiota were eliminated.³³ Purine metabolites were accumulated by the absence of XBP1 in IECs, which contributed to this immunomodulatory phenotype in the gut.³³ IEC-associated ER stress may also elicit barrier-protective immune responses. Immunoglobulin A (IgA) is the major secretory immunoglobulin isotype present in the mucosal surface, regulating microbiota composition and excluding luminal factors from contacting IECs. ER stress and XBP1 deficiency were shown to regulate IgA responses, influencing the extent of intestinal inflammation.³⁴ Innate lymphoid cells (ILCs) are a group of immune cells that exhibit lymphoid characteristics yet lack antigen-specific receptors found on T and B cells, where group 3 ILCs (ILC3s) are key players in intestinal homeostasis. IRE1α/XBP1 is activated in ILC3s from mice exposed to experimental colitis and in human IBD samples with high inflammatory profiles.³⁵ Mice with IRE1α deletion in ILC3s were highly vulnerable to bacterial infections and colitis,³⁵ suggesting that the UPR can influence intestinal physiology from immune cells.

The UPR has also been linked with the homeostatic control of intestinal stem cells. ER stress was shown to induce epithelial cell differentiation, resulting in an apparent reduction of epithelial stem cells.^{36 37} Genetic deletion of Xbp1 gene in IECs increases intestinal stem cell numbers.³⁸ Again, the overactivation of IRE1α in the context of XBP1 deficiency drives the expansion of the intestinal stem cell repertoire due to abnormal levels of ER stress signalling.³⁸

Another disease impacted by ER stress signalling is colorectal cancer. Colorectal cancer is one of the leading causes of death in the western society ranked as the third most lethal neoplasia. Hypomorphic XBP1 function (or IRE1α overactivation) in mice was associated with an increased propensity to develop tumours of the intestinal epithelium in the context of experimental colitis.³⁹ Induction of colorectal cancer with azoxymethane followed by three cycles of DSS caused a profound increase in tumorigenesis in IECs deficient for XBP1, which may be the result of a combination of inflammatory reactions, and tumour-promoting signals emanating from myeloid infiltrates.³⁹ In contrast to the known role of ER stress in cancer,^{40 41} this study revealed an unexpected role for XBP1 in suppressing tumour formation. Analysis of gene expression databases in patients with cancer indicated that reduced XBP1 expression in colorectal cancer samples was associated with poor long-term patient survival and reduced p53 pathway activity.³⁹ Simultaneous deficiency of epithelial XBP1 and Rnaseh2b (a DNA repair factor associated with intestinal carcinoma) induced the generation of aggressive metastatic intestinal adenocarcinoma in mice. XBP1 was shown to restrain intestinal stem cell proliferation and consecutive carcinogenesis via p53.³⁹

Control of PERK and eIF2/ATF4 in gastrointestinal diseases

ATF4 and CHOP participate in a feedback loop to dephosphorylate eIF2α and restore protein synthesis through upregulation of the protein phosphatase 1 (PP1) regulatory subunit GADD34.⁴² GADD34 forms a complex with PP1 to dephosphorylate eIF2α. At basal levels, the constitutive repressor of eIF2α phosphorylation (CReP) associates with PP1 to dephosphorylate eIF2α.⁴³ Conditional expression of non-phosphorylatable Ser51Ala mutant of eIF2α in IECs increased the susceptibility of animals to oral Salmonella infection and DSS-induced colitis.⁴⁴ Paneth cells of eIF2α Ser51Ala mutant mice exhibited a reduction in the number of secretory granules, in addition to developing a fragmented ER and altered mitochondrial morphology under normal conditions.

Salubrinal is a small molecule that targets CReP-PP1 activity, sustaining protein translation under ER stress.⁴³ Salubrinal administration was reported to protect against DSS-induced colitis, suppressing the expression of proinflammatory cytokines.⁴⁵ In contrast with these findings, Chop deficient animals are also protected against experimental colitis in two independent models, reducing apoptosis in the intestine, in addition to producing immunomodulatory effects.⁴⁶ Finally, overexpression of CHOP in IECs using transgenic mice increased the susceptibility to develop colitis and intestinal inflammation, in addition to increasing the levels of mucosal tissue injury.⁴⁷ However, at basal levels, these animals did not develop spontaneous intestinal inflammation. Mucosal tissue regeneration was also impaired in CHOP transgenic mice after mechanical injury.⁴⁷ These results suggest that PERK signalling may regulate intestine physiology through different signalling outputs (ie, translation, control of apoptosis, etc). Finally, in the context of intestinal stem cell homeostasis, inhibition of PERK-eIF2α signalling resulted in stem cell accumulation in the organoid culture of the primary intestinal epithelium, reinforcing the idea that the UPR plays an important role in the regulation of stem cell differentiation into IECs.³⁷

Tryptophan (Trp) and its metabolites, particularly kynurenic acid, are involved in the progression of bowel disease.⁴⁸ Clinical studies have shown that reduced tryptophan serum levels and an elevated kynurenic acid/tryptophan ratio correlate with disease activity in both Crohn’s disease and ulcerative colitis.^{49 50} In animal models, tryptophan supplementation has been found to reduce the expression of UPR genes.⁵¹ Moreover, tryptophan deficiency is linked to the activation of the ATF4-CHOP pathway, although its specific role in UPR regulation within the context of intestinal bowel disease remains to be further explored.⁵²

ATF6 regulation in gastrointestinal diseases

Several studies have defined the involvement of ATF6 signalling in intestinal biology. IECs derived from Crohn’s disease patients or cultured organoid presented patterns of ATF6 activation.⁵³ Genetic targeting of the SP1 protease enhances the susceptibility of animals to DSS-induced colitis, correlating with reduced levels of the ATF6 target genes BiP and Grp94.⁵⁴ Importantly, S1P cleaves a family of transcription factors related to ATF6 not strictly involved in ER stress.⁵⁵ ATF6α-deficient animals develop more severe colitis after administration of DSS, associated with exacerbated ER stress, EIC apoptosis and inflammation in the gut.⁵⁶ In line with these results, pharmacological activation of ATF6α with the small molecule AA147 improved intestinal barrier integrity, disease score and morphological alterations in two models of experimental colitis.⁵⁷

In the context of colorectal cancer, ATF6α was proposed as a marker for early dysplastic changes.⁵⁸ The use of transgenic mice that overexpress active ATF6α in IECs demonstrated the occurrence of spontaneous colorectal tumorigenesis associated with intestinal dysbiosis and abnormal innate immune responses.⁵⁹ Interestingly, the presence of intestinal microorganisms was required for tumour formation induced by active ATF6α. Finally, using human cancer data sets, it was proposed that ATF6α activation is associated with an increased risk of postoperative disease relapse.⁵⁹

Targeting ER proteostasis with chemical chaperones or hormesis-inducing agents

Available evidence suggests that the ER protein folding machinery is key to sustaining intestinal homeostasis. For example, the genetic ablation of the ER chaperone P58IPK enhances the susceptibility of mice to experimental colitis.⁵⁶ BiP/GRP78 is secreted to a high level in the gut. Administration of recombinant soluble BiP decreases the susceptibility to experimental colitis, improving intestinal function and reducing infiltration of immune cells and inflammatory markers.⁶⁰ The levels of the ER foldase ERdj5 are significantly higher in the colonic tissues of patients with IBD. ERdj5 is a member of the PDI family, enzymes that catalyse disulphide bond reduction, oxidation and isomerisation.⁶¹ Studies using ERdj5 knockout mice indicated severe inflammation in mouse colitis models and weakened the gut barrier function because of increased inflammation.⁶²

Generic strategies to reduce ER stress have been tested in models of colitis using chemical chaperones. Chemical chaperones are a group of low-molecular mass compounds that stabilise the folding of proteins and buffer abnormal protein aggregation, reducing ER stress, where the most studied chemical chaperones are 4-phenylbutyrate (4-PBA) and tauroursodeoxycholate (TUDCA). Administration of 4-PBA or TUDCA mitigates features of acute and chronic colitis induced by DSS in mice, modulating ER proteostasis.^{56 63} Moreover, TUDCA was able to ameliorate the phenotypes triggered by ATF6α or ERdj4 deficiency in mouse models of colitis.^{56 62 64}

Hormesis refers to a biological favourable response triggered by subjecting an organism/cell to low exposure to toxins and other stressors. Conditions that stimulate hormesis can engage adaptive stress signalling events rendering cells resistant to a high dose of the same stimuli. Strategies to induce ER hormesis (ie, the use of non-toxic doses of pharmacological ER stress agents) protect animals against neurodegeneration and ageing.⁶⁵ A strategy to induce ER hormesis in the gut was generated using CTB-KDEL, a non-toxic recombinant cholera toxin B-subunit modified with a KDEL motif to induce ER retention after intake. CTB-KDEL exhibited wound-healing properties in the colon through the activation of the UPR in the epithelium.^{66 67} In vitro studies demonstrated that the protective effects of CTB-KDEL were mediated by the expression of adaptive UPR responses through activation of the IRE1α/XBP1s axis.⁶⁶

IRE1β: specialised UPR sensor in the gut

IRE1 represents the most evolutionary conserved UPR pathway in metazoans, and the only one present in yeast. As mentioned, two IRE1 paralogues are present in mammals, where IRE1β expression is restricted to the epithelial cells lining the mucosal surface of the intestinal and respiratory tracks. In the intestine, functional studies suggested that IRE1α and IRE1β have distinct functions possibly related to the high specialisation level of the mucosal epithelia (figure 4).⁶⁸ IRE1β are predominantly expressed in goblet cells and less so in absorptive cells, reaching 50-fold more expression than IRE1α. IRE1β-deficient animals are viable but show dysmorphic goblet cells with reduced expression of mucin 2.³² IRE1β was recently found to be required for microbiota-induced goblet cell maturation and mucus barrier assembly in the colon—and to shape the colonising microflora itself (70). IRE1β deficiency results in the accumulation of misfolded mucin 2 in immature goblet cells, associated with basal ER stress levels and ER abnormalities.⁶⁹ There is controversy, whereas IRE1β can catalyse XBP1 mRNA splicing because it has lower cleavage activity in vitro, but it can perform RIDD.70,72 In fact, IRE1β can regulate the levels of mucin 2 mRNA through RIDD in vitro.⁶⁹ The upregulation of mucin 2 mRNA observed in IRE1β knockout mice is not observed when IRE1α is targeted in IECs.³² In Paneth cells, IRE1α and IRE1β may have redundant complementary roles, because single knockout animals for each gene do not develop spontaneous phenotypes whereas double knockout mice show a dramatic collapse in Paneth cell function with almost null lysozyme staining after histological analysis.³² These observations suggest that IRE1β may regulate XBP1 mRNA splicing in other cell types beyond goblet cells. Microsomal triglyceride transfer protein is essential for lipid metabolism in absorptive enterocytes. IRE1β is proposed to regulate MPT levels through RIDD,⁷³ suggesting an additional function of this IRE1 paralogue in IECs. In the context of gastrointestinal disease, loss of IRE1β results in an earlier onset of DSS-mediated colitis, in addition to impaired recovery, and increased animal mortality.^{69 74}

Figure 4. Cell type-specific inositol-requiring enzyme 1 beta (IRE1β) functions in the intestine. The IRE1β paralogue is expressed in intestinal epithelial cells (IECs), accomplishing different roles. In goblet cells, IRE1β tightly regulates mucin 2 (MUC2) expression through Regulated IRE1-Dependent Decay of RNA (RIDD). X-Box-binding protein 1 (XBP1s) can further regulate endoplasmic reticulum (ER) expansion and cell maturation during ER stress to sustain proper mucus production. IRE1β deficiency leads to unfolded MUC2 accumulation and cell dysfunction, impairing mucus production and sensitising cells for bacterial infection. In Paneth cells, IRE1β by itself does not trigger any phenotype; XBP1 splicing or RIDD have not been addressed. Finally, IRE1β in enterocytes can attenuate IRE1α hyperactivation after XBP1 deficiency. In addition, IRE1β regulates microsomal triglyceride transfer protein (MTP) through RIDD; however, its contribution to chylomicron synthesis or function has not been studied. IRE1β deficiency in enterocytes correlates with increased ER stress in the mouse small intestine and colon.

As discussed, IRE1α hyperactivation in XBP1 deficient IECs mediates in part the adverse effects observed in the intestine of these animals. IRE1β expression was shown to protect against IRE1α hyperactivation in the gut,³² possibly due to its inhibitory effects over IRE1α activity under ER stress.⁷⁰ In this line, increased levels of ER stress are observed in the small intestine and colon of mice lacking IRE1β.^{32 69} However, this phenotype could be triggered by the accumulation of misfolded mucin 2. As introduced just above, IRE1β is dynamically regulated by the microenvironment in the intestine, mediating signalling crosstalk with microbiota. IRE1β is required for mucus barrier assembly in the colon and goblet cell maturation induced by microbiota.⁷⁵ Overall, it is proposed that IRE1β evolved at mucosal surfaces to mediate the dynamic interaction between gut microbes and the colonic epithelium, in addition to regulating proteostasis of IECs to sustain gut function and host defence.

IRE1β and AGR2: stress sensing and goblet cell function

The intestinal mucosa constitutes a barrier with various components, including a thick layer of secreted mucins. The glycocalyx, composed of glycoproteins, glycolipids, and sulphated polysaccharides, resides beneath this mucus layer and above epithelial cells. The barrier further incorporates transmembrane mucins expression in the surface of IECs, in addition to collagen-rich basement membrane. Mucin 2 is a large protein of more than 5100 amino acids, highly O-glycosylated that is produced abundantly in the intestine. Mucin 2 synthesis requires complex posttranslational modifications at the ER, including glycosylation, stabilisation by disulphide bond formation and oligomerisation. Then, mucin 2 traffics through the secretory pathway and is deposited into secretory granules for storage and release to the lumen. As mentioned, IRE1β regulates goblet cell differentiation and thus mucin biosynthesis, folding, and secretion—in part by also affecting mucin 2 mRNA transcript levels and secretory function. Mutations in the gene encoding for mucin 2 are associated with bowel disease and the accumulation of immature unglycosylated mucin 2 is observed in human ulcerative colitis.^{76 77} Disulphide isomerases such as ERdj5 and AGR2 are required for proper mucin 2 production.^{62 78} However, one study was not able to observe defects in mucin 2 production in AGR2-deficient animals.⁷⁹

The mechanisms underlying ER stress sensing by IRE1β are starting to be elucidated. Early in vitro studies suggested that IRE1β directly recognises misfolded proteins.⁸⁰ However, very recently two studies uncovered a novel mechanism that may couple the UPR to mucin production through AGR2 (figure 5). In analogy to BiP binding to IRE1α, AGR2 binding to IRE1β disrupts its oligomerisation, blocking its endonuclease activity.⁸¹ Ablation of AGR2 expression in goblet cells induces spontaneous IRE1β activation, suggesting that alterations in AGR2 availability in the ER determine the threshold for IRE1β activation. AGR2 represses IRE1β activity in vitro and the introduction of mucin 2 to the system reversed AGR2-mediated repression of the IRE1β⁸² (figure 5). Thus, in goblet cells, AGR2 may participate in sensing the efficacy of mucus production by assisting mucin 2 production, relaying this information into IRE1β through direct interactions with its ER luminal domain.

Figure 5. Integration of inositol-requiring enzyme 1 beta (IRE1β) and anterior gradient protein 2 homolog (AGR2) sensing mechanism and adjustment of goblet cell function. AGR2 interacts with IRE1β to disrupt its oligomerisation, thereby blocking its endonuclease activity. On endoplasmic reticulum (ER) stress, AGR2 dissociates to counter mucin 2 (MUC2) unfolding, enabling IRE1β activation and signalling. XBP1s expression allows goblet cell adaptation through ER expansion and possibly by regulating AGR2 gene transcription. Ablation of AGR2 expression in goblet cells induces spontaneous IRE1β hyperactivation, increasing X-Box-binding protein 1 (XBP1) splicing. However, the subsequent cellular effects of XBP1s expression and Regulated IRE1-Dependent Decay of RNA (RIDD) activation in the absence of ER stress have not been addressed. UPR, unfolded protein response.

Interestingly, AGR2 functions as both an ER stress and inflammation sensor through a homodimerisation mechanism regulated by ER proteostasis.⁸³ AGR2 is highly expressed in intestinal biopsies from patients with IBD and colorectal cancer.^{83 84} AGR2 is a genuine XBP1s target and is also found to be secreted by stressed epithelial cells and affects the cellular microenvironment by chemoattracting immune cells.^{83 85} In HT29 cells, the induction of ER stress promotes UPR activation, accompanied by the upregulation and secretion of AGR2.⁸⁶ Extracellularly, AGR2 primarily facilitates the acquisition of invasive, protumorigenic phenotypes in epithelial cells and contributes to fibrosis in Crohn’s disease models.^{86 87} Additionally, AGR2 overexpression may function as a feed-forward loop to inhibit IRE1β activation.

Brain-gut axis, controlling ER proteostasis at a distance?

In the last years, there has been an explosive increase in studies describing a close bidirectional communication between the gut and the brain in neuropsychiatric disorders such as anxiety, depression and autism, among others.⁸⁸ Importantly, gastrointestinal physiology is influenced by signals generated both locally in the intestine and from the brain. Neurotransmitters, immune signalling, hormones and neuropeptides produced in the intestine can affect brain physiology.⁸⁹

Multiple studies in Caenorhabditis elegans indicate that the function of neuronal XBP1 has important consequences in extending lifespan.790,94 Interestingly, these protective effects of neuronal XBP1s were mapped to activating the IRE1/XBP1 pathway in the intestine of the worm as a mirror response through a cell-non-autonomous mechanism involving the release of small clear vesicles.⁹¹ Neuronal XBP1 expression triggered massive changes in gene expression in the periphery including the upregulation of genes involved in proteostasis,⁹³ lipid metabolism⁹⁴ and lipophagy.⁹² The functional outputs of XBP1 expression in neurons may depend on the specific neuronal type affected, including the modulation of antipathogen mechanisms in the intestine.^{95 96} Overall, these reports suggest that the worm’s nervous system acts as a sensing unit to propagate signals to the intestine to orchestrate global adaptation processes to ER stress. Whether neuronal UPR regulates intestinal proteostasis in mammals remains to be determined. Furthermore, how UPR regulates immune cells in the intestine and the communication to the brain is unknown.

The enteric nervous system (ENS), also termed the ‘second brain’, is formed by a diversity of neuronal cell types and complex, integrated circuits that permit the autonomous regulation of many processes in the bowel.⁹⁷ The submucosal plexus is located between circular muscle and bowel mucosa, and controls fluid secretion and absorption, modulates blood flow and responds to stimuli from epithelium and lumen to support bowel function. XBP1s have important roles in the nervous system beyond proteostasis, regulating synaptic activity, connectivity, morphology,^{98 99} contributing to sustained brain function during ageing¹⁰⁰ (reviewed in Martínez et al)¹⁰¹ Whether neuronal UPR can modulate intestinal physiology through the ENS remains to be established.¹⁰²

Discussion

Although ER stress is classically viewed as a pathological factor, understanding the role of the UPR in intestine biology represents a good example of the physiological role of ER stress in tissue homeostasis. Most cells lining the intestinal epithelia secrete abundant proteins ranging from mucins, antibacterial agents, extracellular matrix components, digestive factors and inflammatory mediators. The bidirectional communication between the mucosal surface, commensal microorganisms, IECs and immune cells generates a complex niche that operates as an ecosystem that dynamically regulates the intestinal function. This is why it is not surprising that experimental strategies or disease conditions that disrupt ER homeostasis or UPR signalling in IECs result, not only in proteotoxicity and impaired function of specialised secretory cells (ie, Paneth and goblet cells) but also in altered gut microbiota composition, inflammatory status and barrier function.

The UPR field has experienced transforming advances by the generation of small molecules and gene therapy strategies to target ER proteostasis.^{103 104} Although UPR defects are genetically linked to bowel diseases, as discussed only a few compounds have been tested in the context of intestinal disease (figure 2, table 1). The use of generic compounds to alleviate ER stress (ie, chemical chaperones) has demonstrated its pathological role in IBD. Molecules present in food such as trehalose¹⁰⁵ have chemical chaperone activities, in addition to enhancing autophagy, two outputs that should have positive effects on gut function under pathological conditions. Another interesting example is Zerumbone, a naturally occurring sesquiterpene molecule obtained from dietary bitter ginger, which has protective effects in vivo by reducing ER stress.¹⁰⁶ Several studies have found that nicotinamide adenine dinucleotide (NAD+) reduces tissue deterioration and multiple reports have shown that NAD+ precursors reduce ER stress levels in various settings. Similarly, resveratrol and curcumin, potent antioxidants produced by plants, have shown their ability to reduce ER stress.¹⁰⁷ We predict that the selection of dietary components that are enriched in natural compounds impacting protein stability and ER proteostasis (ie, flavonoids¹⁰⁸ or probiotics containing microorganisms that produce such molecules) will have relevant effects in reducing the symptoms of bowel disease and may improve intestinal function as we age. On the other hand, inhibiting the UPR (or promoting non-adaptative UPR) should have positive effects on improving current treatments for colorectal cancer, as it has been already described in other malignancies in several reports.⁴⁰

ER proteostasis is particularly relevant in goblet and Paneth cells as demonstrated using genetic targeting of central UPR components in IECs. Since IRE1α and IRE1β have distinct structural and enzymatic properties,⁶⁸ it might be feasible to generate allosteric modulators (small molecules) that enhance the activity of IRE1β aiming to improve epithelial cell function for IBD treatment. We predict that therapeutic strategies to boost UPR adaptive signalling should enhance colon crypt goblet cell function (eg, mucin folding and secretion, stronger Paneth cell function), promote survival and proliferation of colon crypt stem cells and transit-amplifying cells and have immunomodulatory effects in the intestinal milieu. Overall, accumulating evidence has expanded the classical view of the UPR as a sole adjustor of protein production in the IECs. It has relevant roles in intestinal barrier function, inflammation and immunosurveillance, microbiota composition and stem cell renewal. As the field of ER proteostasis and intestinal physiology continues to expand, we envision transforming discoveries in the near-future integrating proteostasis pathways in the gut with interorgan communication (cell-nonautonomous control) and healthy ageing.

The intestine is a window to the external world, suffering constant challenges due to variable dietary components, constant intake of adverse microorganisms that may compete with existing microbiota and a reactive immune system. The emergence of IRE1β as an epithelial specialised UPR sensor in evolution reflects the specific requirements and demands of IECs on UPR signalling to sustain not only ER proteostasis and secretion but also a favourable and equilibrated milieu. The recent discovery of a bidirectional communication between goblet cell function and host microbiota through IRE1β signalling reflects a tight connection that orchestrates specialised phenotypic acquisition and maintenance of IEC intestinal function.⁷⁵ Although most of the available data linking UPR with gut function are focused on goblet and Paneth cells, more specific studies are needed to assess the potential contribution of ER proteostasis to other cell types composing the epithelia, in addition to the function of the pathway in immune cells surveilling the gut. Of note, many studies in the context of cancer have shown that ER stress signals can be transmitted between cells, a phenomenon that affects anticancer immune responses and angiogenesis.⁷ It remains to be determined if ER-stressed IECs can propagate UPR signals to neighbour cells, impacting their function. In addition, since the UPR has important roles in controlling synaptic function and connectivity,¹⁰¹ it will be interesting to test the possible involvement of the UPR in the brain–gut axis (both in neurons and IECs) and the mesentery. As reported in C. elegans, the idea of a cell-nonautonomous control of intestinal function by neuronal UPR remains a fascinating open question as to how it might relate to the mesentery. With advancing age, gastrointestinal problems may increase, not only due to the ageing processes but also due to individual exposomes (diet, environmental exposure) and the superimposed effects of comorbidities (autoimmune diseases, allergies, etc), which can have adverse effects on the intestine.¹⁰⁹ Since ER proteostasis is key to sustaining healthy ageing,¹¹⁰ it remains to be determined if the activity of the UPR in IECs is attenuated with ageing and its impact on the functional decline in intestinal function as we age.^{111 112}