Introduction Intestinal goblet cells produce the viscous mucus layer covering the intestinal epithelium. In the large intestine mucus completely fills the crypts and forms a thick coating over the mucosal surface. MUC2 mucin is the major macromolecular constituent of intestinal mucus, and is responsible for its high viscosity and forming a protease-resistant matrix that retains molecules vital to host defence. MUC2 contains a large central O-glycosylated domain and N- and C-terminal cysteine-rich domains that homo-oligomerise [1], forming a complex molecular lattice [2,3]. MUC2 is N-glycosylated in the ER, where initial homo-oligomerisation occurs [4], and the complex then moves into the Golgi apparatus where O-glycosylation, the final stages of oligomerisation, and packaging into granules occur [5]. Mucins stored intracytoplasmically in granules form the characteristic goblet cell theca from where they are secreted constitutively and in response to stimuli. Inflammatory bowel diseases (IBD), characterized by chronic or recurrent relapsing gastrointestinal inflammation, are broadly grouped by clinical and pathological features into two groups—Crohn's disease and ulcerative colitis (UC) [6–8]. Evidence from animal models indicates that failure to suppress immunity to the abundant intestinal foreign antigen load can cause inflammation. Maintaining the normal balance between competence to respond to intestinal pathogens while not generating an inflammatory response to commensals appears to depend on the integrity of the mucosal and epithelial barriers, proinflammatory signalling pathways (especially via NF-κB), and regulation of innate and adaptive immune responses in the intestine and draining lymphoid organs. Disruption of any of these components has been shown to result in intestinal inflammation in animal models [9,10]. Defects in these components have been implicated in human IBD, although fundamental knowledge of underlying pathogenesis remains very poorly understood, with even the most well-established defect, CARD15 mutations, elucidating only a subset of white patients with ileal Crohn's disease [9,10], and with a penetrance less than 4%. Furthermore, the pathogenic pathways that distinguish UC from Crohn's disease remain obscure. Goblet cell and secreted mucus phenotypes represent important differences between the two diseases: in Crohn's disease, there is typically an increase in goblet cells and a thicker mucus layer [11,12], whereas in UC there is a reduction in goblet cells, decreases in MUC2 production [13,14] and sulfation [14–16], accumulation of MUC2 precursor [16], and a reduction in secreted mucus. Although it remains unclear whether changes in mucus are causative or secondary to inflammation [17], the lack of these changes in Crohn's disease indicate they are not a universal consequence of intestinal inflammation. Furthermore, a reduction in a specific biochemical fraction of colonic mucin has been described in UC patients and in their unaffected monozygotic twins [18,19], and engineered changes in intestinal mucin glycosylation result in enhanced susceptibility to toxin-induced colitis [20], raising the possibility that genetic defects in mucin physiology or biochemistry predispose to UC. On the other hand, in Muc2 −/− mice the colitis phenotype appears to arise only on a permissive genetic background [21,22], and transgenic depletion of goblet cells not only does not trigger spontaneous inflammation, it unexpectedly reduces dextran sodium sulphate (DSS)-induced colonic injury [23]. Thus, a causal relationship between mucin abundance and colitis remains to be defined. In this study we used random mutagenesis to produce murine models of inflammatory bowel disease. Two models were produced and we sought to characterize the basis for the pathology and identify similarities between these models and human IBD, particularly UC. Materials and Methods Generation of Mice by N-Ethyl-N-Nitrosourea Mutagenesis All animal experimentation was approved by either the Australian National University or University of Queensland Animal Experimentation Ethics Committees. Mice were generated and housed in a PC2 specific pathogen-free facility and fed autoclaved food and water. For mutagenesis, we treated 8- to 15-wk-old male C57BL/6 mice three times at weekly intervals with 85 mg/kg N-ethyl-N-nitrosourea (ENU, Sigma-Aldrich) in 10% ethanol in citrate buffer (pH 5.0). After identifying heritable phenotypes, affected B6 mice were out-crossed with NOD H-2k congenic mice, and the F1 mice were intercrossed. DNA from affected F2 mice was scanned using simple sequence length polymorphism markers and agarose gel electrophoresis. For experiments mice were housed either under SPF or clean conventional conditions and showed the same colitis phenotype in both housing conditions. Assessment of Inflammation and Intestinal Permeability Scoring of aberrant crypt architecture (score range 0–5), increased crypt length (0–3), goblet cell depletion (0–3), general leukocyte infiltration (0–3), lamina propria neutrophil counts (0–3), crypt abscesses (0–3), and epithelial damage and ulceration (0–3) was performed on the proximal and distal colon at 6, 12, and 18 wk of age (full details in Table S1). The mesenteric lymph nodes (MLNs) were dissected free of fat, disaggregated between two microscope slides, and the number of recovered leukocytes determined. 2 × 106 MLN leukocytes were cultured in 1 ml of RPMI1640 containing 10% fetal calf serum and stimulated with 50 ng/ml PMA and 750 ng/ml ionomycin. The distal one-third of the colon was cleared of luminal material, weighed, washed four times in antibiotics, diced into 1 mm square pieces, and then cultured as explants for 24 h in 1 ml of RPMI1640 containing 10% fetal calf serum. MLN and explant culture supernatants were frozen at −70 °C until assayed for the inflammatory cytokines interleukin (IL)-1β, tumour necrosis factor (TNF)-α, interferon (IFN)-γ (BD Biosciences), and IL-13 (R&D Systems) according to the manufacturer's instructions. To assess intestinal permeability, mice were orally gavaged with 4 kDa FITC-dextran (Sigma, 400 mg/kg body weight in PBS), blood samples obtained at 2 and 5 h, and plasma FITC-dextran concentrations determined by measuring fluorescence at 520 nm emitted in a 96-well plate excited with a 474 nm laser using a FLA5100 scanner (Fuji) versus a FITC-dextran standard curve. Immunoglobulin coating of the faecal bacterial flora was determined using an adaptation of a flow cytometry technique used in human IBD [24]. Induction of Colitis with Dextran Sodium Sulfate Mice were treated with 0.5%, 2%, or 3% dextran sodium sulphate (DSS) in drinking water administered continuously for 3–63 d. Body weight, stool scores, and faecal occult blood were assessed daily. In 2% and 3% DSS experiments on day 7 blood was obtained for full blood counts and serum biochemistry, mice were humanely killed and dissected and the length of the large intestine was measured, and intestinal tissue was fixed for histological analysis. For assessment of DSS-induced inflammation, increased crypt length and goblet cell depletion were excluded from the colitis scores, because these parameters are fundamental components of the Winnie and Eeyore phenotypes. In the 0.5% DSS experiment, mice were weighed weekly, monitored daily, and humanely killed when too ill to continue based on a scoring system involving loss of body weight, diarrhoea, rectal prolapses and bleeding, behaviour, and appearance. Antibodies, Immunohistochemistry, Immunofluorescence, and Western Blotting Muc2 peptides (mM2.2, CPEDRPIYDEDLKK; mM2.3, NGLKPVRVPDADNC) were synthesized (Auspep), conjugated to BSA using glutaraldehyde, emulsified in Freund's adjuvant (Invitrogen), and used to immunize rabbits. Peptide-specific antibodies were purified from rabbit serum by affinity chromatography. The 4F1 antibody reactive with the MUC2 nonglycosylated VNTR peptide repeat [25] was purified from hybridoma supernatant. We purchased antibodies against FLAG (Sigma, clone M2), GRP78 (Santa Cruz, polyclonal N-20), and β-actin (Novus Biologicals, clone AC-15). Standard immunohistochemical procedures with HRP-polymer detection were used to detect MUC2 and GRP78. Standard immunofluorescence staining for Muc2 (detected with anti-rabbit Alexa 633, Invitrogen), the lectin Dolichos biflorus agglutinin (DBA, Sigma, detected with streptavidin Alexa 488, Invitrogen), and DAPI (Invitrogen) was analysed by multitracking on a LSM510 META confocal microscope (Zeiss). Standard PAGE (NuPAGE gels, Invitrogen) and Western blotting were performed with detection by chemiluminescence or dual-label infrared fluorescence on an Odyssey instrument (Li-Cor). Biochemical Characterization of Muc2 Faecal matter was gently removed from the small and large intestines before the mucosa was scraped from the submucosa using a coverslip and extracted in ten volumes of 6 M guanidine HCl extraction buffer, reduced, and alkylated as described [26], and subjected to agarose gel electrophoresis and Western blotting as described [27,28]. Cloning and Expression of Recombinant Truncated Muc2 N-Terminal Proteins The coding sequence for the murine Muc2 N-terminal D3 domain (AJ511872 mRNA nucleotides 2452–3696, hereafter called rMuc2-D3) was reverse transcribed and amplified from murine intestinal RNA extracted from both wild-type C57BL/6 and Winnie mice using Platinum Pfx Taq polymerase (Invitrogen) and the following primers: 5′-CCCAAGCTTCCTTGCATCCACAACAAAGA-3′, 5′-CTAGTCTAGAACCATCCTGGGTCATGTTAAG-3′. The amplicon was cloned into the HindIII and XbaI restriction sites of the pSecTag3xFLAG expression vector constructed by cloning three consecutive FLAG tags into the pSecTag vector (Invitrogen), establishing an N-terminal signal peptide followed by 3 FLAG tags, and a C-terminal myc tag. Correct cloning was confirmed by sequencing. MKN45 cells were transfected with wild-type and Winnie rMuc2-D3 with Lipofectamine 2000 (Invitrogen). After 48 h culture medium was replaced with serum-free medium for 24 h before collecting and concentrating it through a centrifugal membrane filter (Millipore). Cells were lysed in RIPA buffer and lysates subjected to Western blotting. Morphological Studies Intestinal tissue was fixed in 10% buffered formalin or frozen in OCT (Tissue Tek) and sections were stained with hematoxylin and eosin (H&E), or with Alcian blue/Periodic Acid Schiff (PAS). For electron microscopy (EM) tissue was fixed in 4% glutaraldehyde, then post-fixed in osmium tetroxide, embedded in resin, and semi-thin sections stained with toluidine blue and thin sections with uranyl acetate. Gene-Expression Analysis RNA was extracted separately from the distal and proximal large intestine of three C57BL/6, three Winnie, and three Eeyore 8-wk-old mice using Trizol (Invitrogen) and cleaned on RNeasy columns (Qiagen). RNA integrity (RNA integrity number >8) was verified using a Bio-analyser (Agilent) and equal quantities of RNA from the proximal and distal large intestine of each mouse pooled. Samples were labelled for GeneChip analysis using the One-Cycle Target Labelling and Control Reagents (Affymetrix). All steps of target labelling, hybridisation, and scanning were performed according to the manufacturer's protocol. The entire microarray dataset can be accessed at NCBI GEO Accession No. GSE9913 (http://www.ncbi.nlm.nih.gov/geo/) and a heat map of genes with altered expression is provided in Figure S1. For quantitative RT-PCR RNA was reverse-transcribed using Superscript III (Invitrogen) and amplified in a Rotor Gene RG-3000 (Corbett Life Science) using Platinum SYBR Green qPCR Supermix (Invitrogen) and the following cycling conditions and primers: Hspa5 (95 °C 10 s; 56 °C 20 s; and 72 °C 30 s: 5′-TGCAGCAGGACATCAAGTTC-3′ and 5′-GTTTGCCCACCTCCAATATC-3′), the unspliced isoform of XBP-1 (95 °C 10 s; 60 °C 45 s: 5′- CAGCACTCAGACTATGTGCACCTC-3′ and 5′- AAAGGATATCAGACTCAGAATCTGAAGA-3′), the spliced isoform of Xbp-1 (95 °C 10 s; 60 °C 45 s: 5′-GAGTCCGCAGCAGGTGC-3′ and 5′-CAAAAGGATATCAGACTCAGAATCTGAA-3′) and β-actin (94 °C 10 s; 53 °C 30 s; 72 °C 30 s: 5′-GAAATCGTGCGTGACATCAAA-3′ and 5′-CACAGGATTCCATACCCAAGA-3′). Assessment of Intestinal Hypertrophy and Apoptosis 20 crypts cut in longitudinal section at each region of the intestine were measured using a micrometer. To determine the percentage of cells undergoing apoptosis, all of the epithelial cell nuclei within ten consecutive longitudinal crypt sections were counted together with the number of apoptotic bodies. Additionally, TUNEL staining was conducted with an In Situ Death Detection Kit according to the manufacturer's instructions (Roche Diagnostics) and photographed with an Olympus BX60 fluorescence microscope. To assess proliferation mice were given 100 μg/g body weight 5-bromodeoxyuridine (BrdU) i.p. 2 h prior to killing for tissue removal. Sections from paraffin-embedded Swiss rolls of the small and large intestine were antigen-retrieved by heating for 20 min in 10 mM citric acid (pH 6) and cooled to room temperature, then placed in 2 N HCl for 60 min followed by boric acid-borate buffer (pH 7.6) for 1 min and then 0.01% trypsin (in 0.5 M Tris-HCl) for 3 min. Sections were then stained with the biotinylated anti-BrdU antibody and detected with streptavidin-HRP (BD Pharmingen). BrdU-positive and -negative nuclei were counted in ten crypts from proximal and distal colon and ten crypt–villus units from the small intestine. Human Tissue Samples For ultrastructural studies, intestinal tissue biopsies were obtained at colonoscopy from five female and three male patients; four of the total number had distal UC and four were unaffected individuals undergoing colorectal cancer screening. Paired biopsies from proximal and distal colon were taken for conventional H&E histology and for EM. Two of the UC patients were on no treatment, one was on maintenance thiopurine and 5-aminosalicylate treatments, and the other was on a tapering dose of prednisolone. For immunohistochemistry, archival colonic biopsy tissue was prepared from five male and five female UC patients aged 24–45 y with left-sided or extensive colitis. Three had colitis in complete endoscopic remission (one with no immunosuppressive or 5-aminosalicylate treatment), four had mild colitis (three no treatment), and three had moderately severe colitis (one no treatment). Immunohistochemical detection of the Muc2 and the ER chaperone protein GRP78 was conducted using paraffin sections from these ten UC patients and normal tissue obtained from resection margins of patients undergoing surgery for colorectal cancer. The collection of tissue and clinical data followed informed consent and was approved by the Mater Health Services Human Research Ethics Committee Approval No. 396A. Statistical Analyses Due to difficulties in verifying normality of distributions when the sample size is small, we have taken a conservative approach and used the nonparametric Mann-Whitney U-test or Kruskal-Wallis test with Dunn's multiple comparison test for multiple comparisons, and data are presented using box plots. For larger datasets (n > 10), data were assessed with probability plots to determine if they were normally distributed prior to parametric analysis by ANOVA followed by the Bonferroni post-hoc test. Survival analysis was conducted using Kaplan-Meier plots and the Mantel log-rank test. All statistical analyses were performed using Prism v4.03 (Graphpad Software) or Systat v10.2 (Systat Software). The statistical test used and the sample sizes for individual analyses are provided within the figure legends. Results Identification of Two Novel Goblet Cell Mutants We identified Winnie mice amongst the G3 progeny of an ENU-treated C57BL/6 founder by their visible phenotype of spontaneous watery diarrhoea and high incidence of rectal bleeding and prolapse. Compared with wild-type littermates, Winnie small and large intestines were characterized by fewer goblet cells with smaller thecae, the presence of PAS-positive Alcian blue-negative material in the cytoplasm, and a reduction in secreted mucus (Figure 1). This phenotype differs from Muc2 −/− mice, which completely lack Alcian blue-positive mucin stored in goblet cells and secreted mucus [21]. In order to map Winnie, we outcrossed an affected G4 to NODk mice, and intercrossed their F1 progeny; 46/172 (27%) F2 mice had the diarrheal phenotype, consistent with a fully penetrant recessive trait (Figure 2A). Genotyping of 19 affected mice for a panel of 115 microsatellite markers mapped the mutation to a 14.5 Mb interval on Chromosome 7 encoding 198 transcripts including the Muc2 and Muc6 mucin genes (Figure 2B). We sequenced exons for the two mucin genes and found a single missense mutation (G9492A, GenBank accession no. AJ511872, http://www.ncbi.nlm.nih.gov) resulting in substitution of cysteine with tyrosine in the D3 domain at the N terminus of Muc2 (Figure 2C). Figure 1 Histological Phenotype of Mice with Muc2 Mutations PAS/Alcian blue stained intestinal sections from Winnie and wild-type C57BL/6 mice. Note the reduced size of Alcian blue staining thecae (stored mucin) and the presence of PAS-positive/Alcian blue negative accumulations (arrows) in Winnie goblet cells. L, lumen. Figure 2 Generation and Characterization of Mice with Muc2 Mutations (A) Dendrogram showing genesis of the Winnie mutation; filled symbols indicate the diarrhoea phenotype. (B) Microsatellite genotype on Chromosome 7 of 17 affected Winnie F2 mice. (C) Domain organization of the Muc2 protein showing the N- and C-terminal vWF D-domains (D1–D4), the C-terminal B, C, and CK domains, and the two central glycosylated tandem repeat domains (VNTR). The sites of the Winnie and Eeyore mutations are shown, together with the region cloned to express the rMuc2-D3 recombinant protein. (D) Results of complementation cross of Win/Win and Eey/+. PAS-stained sections of representative distal large intestinal histological phenotypes demonstrate noncomplementation (preservation of the eey/eey histological phenotype in eey/win). The possibility existed that another, unidentified mutation might be present to explain Winnie. We identified a second strain, Eeyore, from a different G0 founder, with a similar phenotype also inherited as a fully penetrant recessive trait (Figure 2D). In a Win/Win × Eey/+ complementation test, eight of 22 offspring exhibited the clinical and histopathological phenotype (Figure 2D). Thus, the noncomplementation of Winnie with Eeyore provided strong evidence that both strains harbour a mutation in the same gene and that these mutations are solely responsible for the phenotype. Sequencing of Muc2Eey identified a unique missense mutation (T2996C, GenBank accession no. AJ511873) resulting in a serine to proline substitution in the C-terminal D4 domain (Figure 2C). Winnie and Eeyore remain the only characterized diarrheal/colitis phenotypes in the Australian Phenomics Facility and Phenomix Australia murine ENU mutagenesis programs. Muc2 Mutant Mice Develop Spontaneous Colitis The progressive incidence of rectal prolapse and colitis-associated mortality in Winnie and Eeyore mice is shown in Figure 3A. At 1 y approximately 40% of Eeyore and 25% of Winnie mice had died or were humanely killed due to the development of rectal prolapse or debility (based on a multifactorial scoring system, see Materials and Methods), whereas no wild-type littermates developed rectal prolapses. Proximal and distal colon were thickened and colon weight was greater in Winnie and Eeyore than in wild-type mice, and the thickening and weight increased progressively from 6 to 18 wk of age (Figure 3B). Despite its thickening, the colon was not shortened in mutant mice (Figure S2). Colitis was assessed histologically in Winnie mice at 6, 12, and 18 wk of age, revealing mild inflammation in the large intestine (Figure 3C). The inflammatory infiltrate was usually mild and did not become more severe with age. Classical signs of murine colitis, including crypt elongation, neutrophilic infiltrates, goblet cell loss, crypt abscesses, and focal epithelial erosions were present, particularly in the distal large intestine (examples shown in Figure 3F–3K). Figure 3 Spontaneous Colitis in Mice with Muc2 Mutations (A) Incidence of premature death or colitis-associated pathology requiring humane killing (chiefly rectal prolapse) in Winnie (n = 309) and Eeyore (n = 355) mice. Mice entering experiments or killed for other reasons were treated as censored observations (designated by upward ticks), and the number of uncensored mice remaining at 50 d intervals is included underneath the graph. Incidence rates in the two strains were compared using the Mantel Log-rank test. (B) Weight of the colon after removal of luminal faecal material in C57BL/6 (WT), Winnie (Win), and Eeyore (Eey) mice at 6 (Eey 6–9 wk), 12, and 18 (WT and Win only) wk of age, n = 4–9; box plots show median, quartiles, and range. (C) Histological colitis scores (see Materials and Methods) in WT and Win mice at 6, 12, and 18 wk of age, n = 4–6; scores from individual mice are shown. (D–K) Histology of normal distal colon from a C57BL/6 mouse (D and E) and examples of inflammation in the rectum (F–I) and distal large intestine (J and K) of untreated Winnie mice showing leukocytic infiltration (G and I), occasional branching crypts (F and H), crypt abscesses (J and K) and focal ulcerations (I); scale bars = 20 μm. Note the layer covering the mucosal surface in (F) is a granulocytic serous exudate. Statistics (B and C): p-values for Kruskal-Wallis nonparametric analysis are shown, Dunn's multiple comparison test versus wild type, ** p 15 kb) to express in vitro. Therefore, to ascertain whether the Winnie mutation affects biosynthesis of Muc2, we cloned partial cDNAs encoding the Winnie and wild-type Muc2 N-terminal D3 oligomerisation domain (rMuc2-D3, Figure 2C) and expressed these as recombinant proteins in MKN45 gastric cancer cells which produce the MUC5AC mucin (but not MUC2) and are therefore likely to express mucin-specific chaperones. Western blotting of cell lysates demonstrated increased oligomerisation of the D3 proteins carrying the Winnie mutation (Figure 7). Wild-type D3 proteins were present intracellularly mainly as monomers and dimers, with a small amount of tetramer (11%). In contrast, 61% of Winnie D3 proteins were in the form of a ladder of higher-order oligomers. Wild-type D3 proteins were secreted from MKN45 cells almost exclusively as dimers with a small tetramer band, whereas, despite production intracellularly, Winnie D3 proteins were not secreted, thus confirming the mutation leads to a defect in biosynthesis or secretion. Figure 7 Altered Muc2 N-Terminal Oligomerisation Caused by the Win Mutation (A) PAGE/Western blotting analysis of oligomerisation of the rMuc2-D3 wild-type proteins and proteins with the Win mutation in MKN45 cells and secretions following transfection (n = 4 separate transfections for each group). The position of the origin (o) and markers are shown. A reduced sample (D3 Red) on each gel shows the migration of the D3 monomer. Recombinant proteins were detected with the M2 anti-FLAG antibody, and no reactivity was seen with untransfected cells. Note the hyperoligomerisation of the Winnie D3 domain intracellularly and its failure to be secreted. (B) Densitometric analysis of oligomerisation of the rMuc2-D3 Winnie and wild-type proteins in MKN-45 cellular lysates following transfection. Relative expression of monomer, dimer, and higher-order oligomers was determined by densitometry and expressed as a percentage of the total densitometric value for each sample (lane). Statistics: individual data points and p-values from Mann-Whitney U-tests shown. Muc2 Mutations Lead to Accumulation in the ER Both Muc2 mutations result in a reduction of Muc2 secretion and altered oligomerisation, but the presence of cytoplasmic accumulations of Muc2 precursor raised the possibility of additional pathogenic effects. Since alterations resulting from point mutations can result in functional and sometimes toxic effects not observed when the protein is absent, such effects could contribute to the phenotype of Winnie and Eeyore and explain differences of these mutants from Muc2 −/− mice. To this end, we next examined both mutant strains for ultrastructural changes. Resin sections revealed that Winnie and Eeyore small and large intestines contained large numbers of cells distended with membranous vacuolar material, corresponding with the areas in which Muc2 precursor was detected by confocal microscopy. Vacuolisation was more frequent in the large intestine, and vacuolated cells usually contained stored mucin granules in thecae, demonstrating that they were goblet cells, although these thecae were smaller than in wild-type goblet cells (Figure 8A, 8E, 8I, 8M, 8P, and 8S). By transmission EM, goblet cells possessed dilated rough ER containing aggregates of variable electron density, which is the ultrastructural appearance of protein misfolding and ER stress (Figures 8F–8H, 8J–8L, 8Q, 8R, 8T, and 8U). The size and extent of these ER accumulations varied; cells with more were substantially distended, usually lacked stored mucin granules and often contained swollen mitochondria with disrupted cristae (Figure 8V). The two distinct lineages of mucus-producing cells in the proximal colon both showed vacuolisation, although vacuoles were more extensive in the longer-lived lineage that resides in the base of the crypts [32]. Some vacuoles were surrounded by membranes lacking ribosomes, possibly representing accumulations in the Golgi apparatus. Vacuole accumulation was restricted to Muc2-producing cells, with most nongoblet epithelial cells showing normal ultrastructure. However, some enterocytes and Paneth cells also contained ER vacuoles, although these vacuoles were fewer and smaller than in goblet cells, consistent with the comparatively low levels of Muc2 production that we observed in these cells by immunohistochemistry (Figure 8X). Figure 8 Ultrastructural Evidence of Endoplasmic Reticulum Stress in the Intestinal Epithelium of Mice with Muc2 Mutations Semi-thin sections from resin embedded large intestine of wild-type (A), Winnie (E), and Eeyore (I) mice and small intestine of wild-type (M), Winnie (P), and Eeyore (S) mice stained with toluidine blue. Transmission electron micrographs from the large intestine (B–D, F–H, J–L, and V) and small intestine (N, O, Q, R, T, U, W, and X) of wild-type (B–D, N, and O), Winnie (F–H, Q, R, and V), and Eeyore (J–L, T, and U) mice. Note the reduced size of goblet cell thecae (indicated by a T) containing stored mucin granules and the presence of vacuoles (indicated by a V) surrounded by rough endoplasmic reticulum (RER) in Winnie and Eeyore. Other abbreviations: G, Paneth cell granule; GA, Golgi apparatus; L, lumen. Scale bars 40 μm (A, E, and I), 20 μm (L, O, and R), 5 μm (B, F, J, and U), 2 μm (C, G, K, M, P, S, and V), 1 μm (W), 50 nm (D, H, L, N, Q, and T). Given the increased susceptibility of heterozygous mice to DSS, to assess whether one mutant allele could initiate ER stress we examined heterozygous mutants. These mice lacked the diarrhoea phenotype, and their intestinal morphology appeared normal by standard light microscopy. However, glutaraldehyde fixation and resin embedding revealed variable, though often extensive, vacuolisation in goblet cells deep in the crypts of the proximal colon (Figure 9), demonstrating subclinical pathology and establishing that these mutations are not simple mendelian recessive traits. In heterozygotes the surface goblet cell lineage in the proximal colon and the morphologically similar goblet cells in the distal colon and small intestine were mostly free of vacuolisation (Figure 9). However, vacuoles were present in some Paneth cells in the small intestine despite their lower level of Muc2 production, which may be due to the extended lifespan of these cells (unpublished data). Figure 9 Evidence of Mucin Misfolding in Heterozygous Mice Intestinal tissue from mice heterozygous for the Win mutation (Win/+) was fixed with glutaraldehyde, embedded in resin, and semi-thin sections stained with toluidine blue. Note the variable degree of vacuolisation of the goblet cells in the base of the crypts in the proximal large intestine (solid white arrows) and the lack of/minimal vacuolisation in surface goblet cells in the proximal large intestine (hollow arrows), distal large intestine and small intestinal villi. Tissue from wild-type C57BL6 mice is shown for comparison. Scale bars = 20 μm. Accumulation of nonglycosylated Muc2 precursor (Figure 6C), altered electrophoretic migration (Figure 6A), ER vacuolisation (Figure 8), and the hyperoligomerisation of the Winnie D3 domain (Figure 7) together demonstrate that a proportion (but not all) of Muc2 produced in Winnie and Eeyore goblet cells is inappropriately assembled, accumulates in the ER, and leads to ER vacuolisation, smaller goblet cell thecae, and reduced secretion of mature Muc2. It appears that at some stage of their life cycle these goblet cells cease producing mature mucin, whilst still retaining Muc2 precursor in ER vacuoles (see middle right image of large intestine in Figure 6C). Aberrant Mucin Assembly Triggers an ER Stress Response GRP78 is a heat shock protein that chaperones proteins in the ER during folding and is up-regulated in ER stress when it accumulates with misfolded proteins, ensuring they do not exit the ER normally but are removed for degradation [33]. Quantitative PCR showed a 2- to 3-fold increase in Hspa5 (which encodes GRP78) mRNA in Winnie and Eeyore proximal and distal colon (Figure 10A). Protein misfolding engages a series of molecular events that result in reduced protein translation and transcription of genes involved in the unfolded protein response (UPR) [34]. Splicing of the Xbp-1 mRNA by the enzyme IRE-1 is one key component of the UPR and we observed significantly decreased levels of the unspliced form of Xbp-1 in the proximal colon of Winnie and increased levels of the spliced form of Xbp-1 in the proximal colon of Eeyore mice (Figure 10A). The ratio of spliced to unspliced Xbp-1, which is used as a measure of UPR activation [35], increased 2.5- and 2.7-fold in Winnie and Eeyore proximal colon, respectively. Western blotting demonstrated increased GRP78 protein in both Winnie and Eeyore intestinal tissue (Figure 10B). In goblet cells from mutant mice GRP78 was detected immunohistochemically within the membranous accumulations, consistent with an association with misfolded Muc2, but notably not within thecae containing mucin packaged in granules for secretion (Figure 10C). Analysis of the Winnie and Eeyore intestinal transcriptome demonstrated increased transcription of the genes encoding the GRP78 and GRP94 chaperones, the γ-subunit of the SEC61 pore through which misfolded proteins exit the ER, ubiquitin peptidases, and key elements of Ca2+ metabolism and signalling pathways consistent with disrupted Ca2+ sequestration ensuing from ER stress (see Table S2). Figure 10 Evidence of ER Stress and UPR Activation in Mice with Muc2 Mutations (A) mRNA expression of hspa5 (GRP78) and the unspliced and spliced forms of Xbp-1 in the proximal and distal colon of wild-type C57BL/6 (WT), Winnie (Win), and Eeyore (Eey) mice determined by quantitative PCR and expressed relative to β-actin with the mean ratio for the WT group in each tissue corrected to 1. Statistics: n = 3 individual data points shown, Kruskal-Wallis nonparametric analysis (p-values shown) with Dunn's multiple comparison test (*p 2 Winnie:wild type, green for 0.5 and <2.0. The entire microarray dataset can be accessed at NCBI GEO Accession No. GSE9913 (http://www.ncbi.nlm.nih.gov/geo/). (57 KB PDF) Click here for additional data file. Figure S7 Evidence of MUC2 Precursor Accumulation in Ulcerative Colitis Individual confocal images of the composite confocal images shown in Figure 12B. Staining with the MUC2 precursor antibody 4F1, DBA lectin, and DAPI as indicated. Scale bars = 10 μm. (560 KB PDF) Click here for additional data file. Figure S8 Morphological Evidence of ER Stress in Ulcerative Colitis Intestinal biopsies from two healthy individuals, the unaffected proximal colon of three UC patients, and the affected distal colon of one of those UC patients, were examined by light (left) and electron (right) microscopy. Arrows indicate the presence of vacuoles in goblet cells. Abbreviations: NG, nongranular material; RER, rough endoplasmic reticulum; T, theca/mucin granules; V, vacuole. Scale bars are individually annotated. (2.9 MB PDF) Click here for additional data file. Table S1 Histological Scoring of Murine Colitis (67 KB DOC) Click here for additional data file. Table S2 Comparison of the Intestinal Transcriptome of C57BL/6, Winnie and Eeyore Mice—Genes Involved in ER Stress, Antimicrobial Defence, Wound Repair, Epithelial Growth, Cell Cycle, and Apoptosis (61 KB DOC) Click here for additional data file. Table S3 Comparison of the Intestinal Transcriptome of C57BL/6, Winnie and Eeyore Mice—Genes Involved in Inflammation, Metabolism, Detoxification, and the Mucus Barrier (60 KB DOC) Click here for additional data file.
