Ablation of Intestinal Epithelial Sialylation Predisposes to Acute and Chronic Intestinal Inflammation in Mice

Abstract

Background & Aims

Addition of sialic acids (sialylation) to glycoconjugates is a common capping step of glycosylation. Our study aims to determine the roles of the overall sialylation in intestinal mucosal homeostasis.

Methods

Mice with constitutive deletion of intestinal epithelial sialylation (IEC Slc35a1^–/– mice) and mice with inducible deletion of sialylation in intestinal epithelium (TM-IEC Slc35a1^–/– mice) were generated, which were used to determine the roles of overall sialylation in intestinal mucosal homeostasis by ex vivo and mutiomics studies.

Results

IEC Slc35a1^–/– mice developed mild spontaneous microbiota-dependent colitis. Additionally, 30% of IEC Slc35a1^–/– mice had spontaneous tumors in the rectum greater than the age of 12 months. TM-IEC Slc35a1^–/– mice were highly susceptible to acute inflammation induced by 1% dextran sulfate sodium versus control animals. Loss of total sialylation was associated with reduced mucus thickness on fecal sections and within colon tissues. TM-IEC Slc35a1^–/– mice showed altered microbiota with an increase in Clostridium disporicum, which is associated a global reduction in the abundance of at least 10 unique taxa; however, metabolomic analysis did not show any significant differences in short-chain fatty acid levels. Treatment with 5-fluorouracil led to more severe small intestine mucositis in the IEC Slc35a1^–/– mice versus wild-type littermates, which was associated with reduced Lgr5⁺ cell representation in small intestinal crypts in IEC Slc35a1^-/-;Lgr5-GFP mice.

Conclusions

Loss of overall sialylation impairs mucus stability and the stem cell niche leading to microbiota-dependent spontaneous colitis and tumorigenesis.

Graphical abstract

Summary.

The study investigated the role of overall sialylation in intestinal function. Our data demonstrate that loss of overall sialylation impairs mucus stability and the stem cell niche leading to microbiota-dependent spontaneous colitis and tumorigenesis in mice.

The intestinal tract is constitutively exposed to potentially noxious stimuli including a dense and diverse microbial consortium, and host- and microbe-derived proteases and glycosidases, bile acids, and microbial metabolites.1, 2, 3 Having coevolved with this dynamic environment, the intestine has adapted several mechanisms to maintain homeostasis and restore its function on injury. Mucus production by intestinal goblet cells is an essential mechanism to maintain homeostasis by forming an encapsulating barrier around the fecal material in mice and likely humans, protecting the mucosa from invasion by the microbiota, and injury from mechanical and enzymatic forces, as the feces moves down the colon.⁴ The underlying epithelial barrier is under a constant state of renewal because of the activity of stem cells near the base of intestinal crypts to replace dying, damaged, or infected epithelial cells at the mucosal surface.⁵ These stem cells are governed by apical and basal signals emanating from their surrounding niche and are further activated after injury to promote epithelial restitution on injury.

Mucus is made up of the gel-forming mucin-family molecule, MUC2.⁶ MUC2 is comprised mainly of glycans, >80% by weight, which is made up of >100 structures containing combinations of at least 5 monosaccharides: galactose (Gal), N-acetylgalactosamine (GalNAc), N-acetylglucosamine (GlcNAc), sialic acid (Sia), and Fucose (Fuc).⁷ The main type of glycosylation on MUC2 is “O-linked,” because of the modification of –OH (hydroxyl) groups of Serine/Threonine (Ser/Thr) abundant in the MUC2 mucin domain.⁸ There are approximately 1804 predicted O-glycosylation sites in human MUC2 and 334 in murine Muc2, compared with only 21 and 20 predicted N-glycosylation sites, respectively. Complex O-glycans are essential for mucus barrier integrity by shielding the MUC2 core from microbial proteolytic degradation.⁹ Furthermore, like dietary glycans (eg, starch), host mucin O-glycans ex vivo have recently been shown to be a source of microbiota-derived anti-inflammatory metabolites including short-chain fatty acids (SCFA), such as butyrate, acetate, propionate,¹⁰ which are known to induce regulatory T cells that dampen inflammatory tones.^11,12 Collectively, these studies show the impact of host-glycosylation on host-microbiota symbiosis; however, the contribution of individual monosaccharides to homeostasis is unclear.

Sia is a major capping monosaccharide found on 50% of complex O-glycans on mucus.^8,13 This monosaccharide has unique biochemical properties versus other mucin sugars, namely a 9-carbon backbone (vs 6 on Gal, GlcNAc, GalNAc, and Fuc) that is acidic in nature because of both N-acetyl and carbonyl groups along the backbone that give it strong anionic charge.¹⁴ A large family of sialyltransferases (more than 20) mediate sialylation of glycans, which can occur in several conformations including α2,3 linkages to underlying Gal, or α2,6 to Gal or GalNAc, or α2,8-9 to itself, thus introducing diversity to glycans.¹⁴ Sialylation of glycoconjugates plays diverse roles in physiology, including neural development and immune functions.^15,16 Sialylation status can also drive or serve as a biomarkers for various pathophysiological processes including progression of various cancers including colon carcinoma.^17,18 One key function of Sia is to inhibit proteolytic attack on peptide core of the glycoconjugate, which has been observed on diverse glycoproteins including podoplanin,¹⁹ growth-factors,^20,21 and von Willebrand factor multimers,²² and recently suggested for mucins, such as MUC2.^23,24

Studies on specific sialyltransferases have suggested the importance of intestinal epithelial sialylation in preserving mucus structure and integrity and mucosal homeostasis. ST6GALNAC1, which modifies the initial GalNAc of mucin-type O-glycans with Sia via an α2,6 linkage, is implicated in governing human MUC2 sialylation and mucus protection from glycoproteases.²³ In mice, ST6GALNAC1 loss was associated indirectly with defects in stem cell function via its impact on microbial composition and function, characterized by increased mucin degrader representation and overproduction of butyrate, which impairs colonic stem cell activity.²³ In another study, ST6GALNAC6-mediated sialylation has been shown important for mucus cross-linking via its impact on the net charge of the glycosylated mucus.²⁴ Although these studies implicate specific forms of sialylations in host-microbial and intestinal epithelial homeostasis, there is still substantial sialylation occurring on glycoconjugates including mucins, therefore the role of overall sialylation remains to be studied.

Sialylation occurs in the trans-Golgi by transferring Sia from the donor substrate cytidine-5’-monophosphate-Sia (CMP-Sia) to acceptor N- and/or O-glycans. CMP-Sia transporter Slc35a1, which exchanges CMP-Sia for CMP across the Golgi membrane using an antiport mechanism, is essential for sialylation. To address the role of overall sialyation in the intestine, we generated mice with either constitutive or inducible deletion of the CMP-Sia transporter, Slc35a1, specifically in intestinal epithelial cell (IEC) using the Cre-LoxP technology. Our results show loss of intestinal epithelial Slc35a1 leads to spontaneous colitis and tumorigenesis, which was linked to intestinal mucus defects and altered mucus ecology. In addition, mice lacking intestinal epithelial Slc35a1 were also susceptible to acute injury and had defects in epithelial stem cell function. These studies indicate a critical role for sialylation of epithelium in maintaining microbe and immune homeostasis.

Results

Conditional Deletion of Slc35a1 Ablates Intestinal Epithelial Sialylation and Alters Muc2 O-glycosylation

Recent studies have addressed how specific linkages of Sia to mucins (eg, α2,6-sialylated glycans) contribute to mucus function.23, 24, 25 However, these approaches are limited in that sialylation is still present on mucins in the form of other linkages (α2,3 and possibly α2,8). To address this, we first looked at the relationship α2,6 and α2,3 linked Sia via lectin staining with Sambucus nigra (SNA), which recognizes α2,6-sialylated glycans, and Maackia amurensis lectin II (MALII), which recognizes both α2,3 and 3-O-sulfated Gal.²⁶ We found that the proximal colon was primarily α2,6-derived, giving rise to mostly an α2,6-linked sialylated b1 barrier layer (Figure 1A). However, the distal colon mucus binds tightly to SNA and MALII (Figure 1A). Although we have previously shown the MALII ligand is likely a sulfated epitope in mouse colon goblet cells,⁴ if α2,3 Sia is present, it is not likely found in the proximal region and would be found in the distal colon. In our previous study, we demonstrated the heterogeneity of Sia linkages along the mouse colon,⁴ indicating that the strategy of deleting a single sialyltransferase is not feasible for studying the function of sialylation in the colon.

Figure 1.

Generation of a mouse model with a global loss of intestinal epithelial sialylation. (A) Epifluorescent images of mucus/bacteria staining with different lectins in proximal and distal colon and a fecal pellet. Arrows, secreted mucus. b1, b2, b1 and b2 mucus barrier layers. Scale bars, 50 μm. (B) Left: model showing role of Slc35a1 in controlling access of CMP-Sialic acid to the Golgi compartment. Right: breeding strategy of IEC Slc35a1^–/– mice. (C) Slc35a1 gene expression in colon epithelium (normalized to hprt gene). (D) Left: Epifluorescent images of SNA lectin (binds α2,6 linked Sia; green) and Muc2 (red) in proximal colon. Dual positive cells are yellow. Insets are split images of white-boxed regions. Arrowhead, SNA expressing crypt. Scale bars, 20 μm. (E) Proportion of SNA⁺ goblet cells in crypts. (F) Graph showing signal intensity of SNA in goblet cell (GC) and lamina propria (LP). (G) Epifluorescent lectin staining of CFPE sections. Scale bars, 50 μm. (H) HID-AB staining of distal and proximal colon sections prepatrial and postpartial desulfation (“Desulf.”, boxed region). Black: HID⁺, recognizing acidic sulfated glycoprotein; blue, HID-negative, nonsulfated acidic glycans. Scale bars, 50 μm. Results are representative of 3 independent experiments each with N = 4 or 5 mice per experiment. Data are means ± standard deviation. ∗P < .05, ∗∗P < .01.

To address these caveats, we bred mice with floxed Slc35a1 (Slc35a1^f/f), which encodes CMP-Sia transporter (Figure 1B),²⁷ with VillinCre mice that express Cre recombinase specifically in IEC. The resultant new mouse line lacks overall sialylation constitutively in gut epithelial cells (Slc35a1^f/f;VillinCre or IEC Slc35a1^–/– mice). Real-time quantitative polymerase chain reaction (RT-qPCR) analysis confirmed the elimination of Slc35a1 transcripts in colon epithelial cells in IEC Slc35a1^–/– mice (Figure 1C). To verify that Slc35a1 deletion led to a loss of sialylation in epithelial cells including goblet cells, we stained colon sections with SNA, showing that sialylation was specifically deleted in epithelial cells, most obvious in goblet cells, but not in the lamina propria in IEC Slc35a1^–/– mice (Figure 1D–F). The staining of MALII lectin was comparable between wild-type (WT) and IEC Slc35a1^–/– distal colon (Figure 1G), which correlates with increased high-iron diamine (HID) staining in the distal colon. (Figure 1H). We also observed increased HID staining in proximal colon GCs (Figure 1H). These results indicate that colonic mucin sulfation was intact and increased in the absence of α2,3/6-sialylation.

In addition, we generated a mouse line with tamoxifen (TM)-inducible deletion of intestinal sialylation in adult mice (Slc35a1^f/f;VillinCreER^T2, TM-IEC Slc35a1^–/– mice) (Figure 2A and B). To definitively define the glycan structures in the mutant mice, we performed glycomic analysis on colon mucus extracted from WT and TM-IEC Slc35a1^–/– mice 10 days after the last induction. Mucins were extracted from fecal contents, and glycans were released by reductive β-elimination and subjected to liquid chromatography and mass spectrometry (LC-MS).²⁸ We first analyzed the glycome by comparing overall acidic status (determined by overall levels of sialylation and sulfation), finding acid status contributed by sialylation was virtually gone in TM-IEC Slc35a1^–/– mice (Figure 2C). Comparison of relative abundances of various subgroupings of glycosylation (eg, fucosylated, sialylated, sulfated) revealed all combinations of sialylated glycans were missing (Figure 2D), confirmed also by analysis of individual glycans (Figure 2E). Supplementary Table 1 shows all the O-glycomics analyses. Further analysis indicated that loss of sialylation was associated with an increased abundance of sulfated and fucosylated epitopes (Figure 2F), which is consistent with the increased HID, AB (pH 1.0), and Ulex europaeus agglutinin I (UEA1) staining in the proximal colon (Figure 2G–I). These results demonstrate the first model of complete sialylation ablation in the colon, enabling the determination of the overall role of sialylation in intestinal homeostasis.

Figure 2.

Impact of induced epithelial sialylation loss on Muc2 O-glycosylation. (A) Breeding strategy of TM-IEC Slc35a1^–/– mice. (B) Representative whole colon image of CFPE colon coil sections dually stained with Muc2/SNA from WT and TM-IEC Slc35a1^-/- mice. Scale bars, 4 mm. (C, D) Pie chart of acidic (C) and various subsets (D) of mucin-type O-glycans on purified colonic mucus. (E) Heatmap of individual sialylated mucin-type O-glycans extracted from mouse colonic Muc2. Color scale shows glycan intensity. Gray represents glycans not found. (F) Graph of nonsialylated mucin-type O-glycans extracted from mouse colonic mucus. Black dots indicate sulfated epitopes increased in KO. (G) Top: Tiled-images of HID-AB staining of whole colon in WT and TM-IEC Slc35a1^-/- mice. Bottom: high-mag images of different regions. Scale bars, 4 mm (top) and 50 μm (bottom). (H) Alcian blue (pH 1.0)-stained images of proximal colon in WT and TM-IEC Slc35a1^-/- mice. Scale bars, 50 μm. (I) Z-stacked confocal images of UEA1 lectin staining of proximal colon in WT and TM-IEC Slc35a1^-/- mice. Results are representatives of three independent experiments with N = 4 or 5 mice per experiment. Scale bars, 50 μm.

Loss of Total Intestinal Epithelial Sialylation Renders Mice Highly Susceptible to Colitis

Using hematoxylin and eosin staining sections of colon coils, we observed that IEC Slc35a1^–/– mice showed spontaneous inflammation characterized by inflammatory cell infiltration, goblet cell loss, and epithelial hyperplasia (Figure 3A). F4/80 staining revealed many of the inflammatory cells were macrophages, which were significantly elevated in IEC Slc35a1^–/– versus WT littermates (Figure 3B). Western blotting to detect the active forms (p10 band) of caspase-1, which is also involved in chronic colitis in complex O-glycan-deficient mice, indicated a 3- to 7-fold increase in IEC Slc35a1^-/- mice compared with WT mice as assessed by densitometry of the Casp1 p10 band (Figure 3C and D). Notably, although sialylation was lost in the small intestine in IEC Slc35a1^–/– mice (Figure 3E), we did not notice any significant histologic abnormalities in the duodenum, jejunum, or ileum (Figure 3F). These studies suggest ablation of intestinal epithelial sialylation has a location-specific impact on baseline mucosal homeostasis.

Figure 3.

Mice with intestinal epithelial sialylation deficiency are susceptible to spontaneous colitis. (A) Top: gross picture of colon and tiled images of hematoxylin and eosin (H&E)-stained whole colon coil sections from 3-month-old male mice. Bottom: representative images of H&E-stained distal colon. Graph: histologic score of the distal colon. Scale bars, 4 mm (top) and 50 μm (bottom). (B) Immunofluorescence staining for macrophage marker F4/80 on distal colonic tissues. Graph: number of F4/80⁺ epithelial cells per high-power field (HPF). Scale bars, 50 μm. (C) Representative Western blot of distal colonic lysates derived from 10-month-old mice. (D) Densitometry of the caspase 1 p12 subunit visualized via Western blot of colon lysates. (E) Z-stacked confocal images of SNA lectin staining of mouse small intestine. Scale bars, 50 μm. (F) H&E-stained images of small intestine in WT and IEC Slc35a1^-/- mice. Scale bars, 50 μm. A, B, E, and F: Results are representative of 3 independent experiments each with n = 4 or 5 mice per experiment. Data are means ± standard deviation.∗P < .05. C and D: Results are representative of two independent experiments each with N = 2 mice per experiment. Data are means ± standard deviation. ∗P < .05.

We next investigated the effects of sialylation deficiency after inducible deletion of Slc35a1 in adult mice (TM-IEC Slc35a1^–/– mice). Unlike the IEC Slc35a1^–/– model, TM-IEC Slc35a1^–/– mice exhibited mild inflammation in the colon, despite total ablation of sialylation (Figure 4A). We therefore challenged the TM-IEC Slc35a1^–/– mice using a low-dose dextran sulfate sodium (DSS).²⁹ Five days after the final TM treatment, mice were exposed to 1% DSS in drinking water for 7 days followed by 4 days of sterilized water (Figure 4B). Our findings show that TM-IEC Slc35a1^–/– mice exhibited more severe colitis, characterized by weight loss, diarrhea, and fecal bleeding relative to control animals (Figure 4C). Gross and histologic examination revealed a shorter colon, mucosal destruction, larger areas of epithelial ulceration, erosions, and inflammatory cell infiltration into the colon mucosa in TM-IEC Slc35a1^–/– mice (Figure 4D and E), although no changes were observed in the small intestine (Figure 4F). Similar results were observed in colons of IEC Slc35a1^–/– mice (Figure 4G–M), indicating that colon disease susceptibility in both models is likely through similar mechanisms.

Figure 4.

Inducible and constitutive loss of intestinal epithelial sialylation renders the colon hypersusceptible to acute injury via DSS. (A) H&E-stained images of distal colon in WT and TM-IEC Slc35a1^–/– mice 5 days post-TM injection. Graph: histologic score of the distal colon. (B) Schematic of low-dose DSS treatment regimen. (C) Body weight change and clinical disease activity index of mice with acute intestinal injury induced by low dose (1%) DSS. (D) Top: representative gross pictures of colon. Bottom: H&E-stained images of distal colon. Scale bars, 100 μm (E) histologic colitis score and colon length. (F) H&E-stained images of small intestine in WT and TM-IEC Slc35a1^–/– mice after 1% DSS-treatment and 7 days recovery. Scale bars, 50 μm. (G) Schematic of 2% DSS-treatment experiments. (H) Body weight change. (I) Clinical disease activity index. (J) Survival rate. (K) Top: representative gross pictures of large intestines. Bottom: H&E-stained images of distal colon. Scale bars, 50 μm. (L) Bar plot of mean colon length. (M) Bar plot showing histologic colitis score. All results are representative of three independent experiments each with N = 4 or 5 mice per experiment. Data are means ± standard deviation.∗P < .05.

Spontaneous Colitis in IEC Slc35a1^–/– Mice Is Microbiota-dependent

In normal conditions, the proximal colon-derived O-glycosylated mucus encapsulates the fecal microbiota to protect the distal colon.⁴ Therefore, we examined how complete ablation of epithelial glycan sialylation impacted colonic mucus function. Confocal analysis of CFPE proximal and distal colon cross-sections dually stained with Muc2/fluorescence in situ hybridization probes showed modest levels of secreted Muc2 in the proximal colon of IEC Slc35a1^–/– mice (Figure 5A), which corresponded to thinning and disruption of the b1 but not b2 mucus barriers, leading to closer proximity of the microbiota with the tissues (Figure 5B). A similar b1/b2 phenotype was observed in TM-IEC Slc35a1^–/– distal colon (Figure 5B). To determine if the microbiota impacted spontaneous disease, we treated IEC Slc35a1^-/- mice with established colitis with a broad-spectrum antibiotic cocktail (neomycin sulfate, vancomycin, metronidazole, ampicillin) (Figure 5C and D).³⁰ We found that the antibiotic treatment reduced the severity of spontaneous disease in IEC Slc35a1^-/- mice (Figure 5E and F), revealing the contribution of the microbiota to the colitis phenotype in the absence of sialylation.

Figure 5.

Loss of the gut epithelial sialome impacts mucus function and associates with microbial-induced colitis. (A) Dual Muc2:fluorescence in situ hybridization stain of CFPE proximal colonic sections. Arrowheads: secreted mucus. Graphs: quantitation of number and volume of secreted Muc2 boli. Scale bars, 20 μm. (B) Z-stack of confocal images of distal colon sections. Graph: thickness of b1 (green) and b2 (blue) mucus barrier layers. Scale bars, 50 μm. (C) Antibiotics treatment regimen. (D) qPCR for the 16S ribosomal rRNA gene from genomic DNA extracted from fecal pellets. (E) Representative images of H&E-stained distal colon. Scale bars, 50 μm. (F) Histologic score of the distal colon. Results are representative of three independent experiments with N = 4 or 5 mice per experiment. Data are means ± standard deviation. ∗P < .05, Student t test.

Inducible Loss of Epithelial Sialylation Impacts Microbial Ecology

Because sialylation of glycoconjugates including mucins directly impacts luminal ecology,^23,31, 32, 33 and protects from spontaneous colitis (Figure 3A), we addressed how the loss of total glycan sialylation impacts the microbiota composition and function. To rule out the impact of inflammation itself on the microbial community, we conducted the study on TM-IEC Slc35a1^–/– mice. We first induced Slc35a1 deletion for 9 days, and then euthanized the mice and harvested colon contents of the proximal and distal regions for microbial community analysis via 16S sequencing of the V4 region, and the mid colon for SCFA production (Figure 6A).³⁴ As expected, we found that loss of glycan sialylation altered microbial communities, as determined through the Bray-Curtis dissimilarity index (a β-diversity metric) in the proximal and distal colons of mice, with most of the proximal and distal colons of TM-IEC Slc35a1^–/– segregating away from the control animals (Vehicle-treated Slc35a1^f/f;VillinCreER^T2+) (Figure 6B, Supplementary Table 2). In-depth taxonomic profiling revealed increased representation of Clostridiaceae in the TM-IEC Slc35a1^–/– proximal and distal colons versus control animals (Figure 6C). We unexpectedly observed that among the significantly changed taxa at the genus level (unadjusted P values < .05) (Supplementary Tables 3 and 4), only 1 taxa, identified as Clostridium disporicum was increased in proximal and distal colons of TM-IEC Slc35a1^–/– versus WT, whereas most others were reduced in relative abundance (observed reads/total reads/sample) (Figure 6D and E), most notably genera Lachnoclostridium, and Agathobaculum, and a taxon assigned to genus Dwanesavagella (aka Candidatus savagella), a segmented filamentous bacteria,³⁵ which was not detected in TM-IEC Slc35a1^-/- littermates (Figure 6F). Sialyation defects have been linked to SCFA overproduction by the microbiota²³; however, the altered microbiota did not translate into any significant differences in SCFA production in luminal content of TM-IEC Slc35a1^–/– mice at the time point analyzed (Figure 6G). Supplementary Table 5 shows SCFA analysis results. Analysis of glycan length between WT and TM-IEC Slc35a1^–/– mice showed no obvious changes to this parameter, suggesting no major change in overall glycan metabolism (Figure 6H). Because the TM-IEC Slc35a1^–/– and control mice are cohoused, it is unlikely that the community itself is pathogenic. Therefore, we conclude that the colitis phenotype is most likely caused by the compromised mucus barrier functions toward the microbiota.

Figure 6.

Inducible loss of the gut epithelial sialome rapidly modulates microbiota community composition. (A) Schematic of gut microbiota study. (B) Principal coordinate analysis of the Bray-Curtis dissimilarity distance matrix for the bacterial microbiota. (C) Bar chart showing the relative abundance of major bacterial taxa in WT (N = 5) versus TM-IEC Slc35a1^–/– (KO, N = 8) mice, with each lane representing an individual mouse. (D) Statistically different taxa (Genus level) between WT and TM-IEC Slc35a1^–/– (KO) mice. ∗Unadjusted P < .05, Student t test; ^#unadjusted P < 0.05, Wilcoxon rank-sum test. (E) Graph showing increased relative abundance for Clostridium disporicum in TM-IEC Slc35a1^–/– (KO) versus WT mice (unadjusted P < .05). (F) Graph showing reduced relative abundance of selected genera in KO than WT (all had unadjusted P < .05). (G) Bar chart showing concentrations of SCFAs in the WT and KO mice. (H) Bar chart illustrating comparison of glycan length between WT and KO mucins. Data are means ± standard deviation.

Loss of Epithelial Sialylation Alters Stem Cell Homeostasis In Vivo and Ex Vivo to Promote Spontaneous Tumorigenesis

Previous work has shown that loss of complex O-glycans in gut epithelium decimated mucus function and led to microbiota-induced colitis-associated cancer.^9,36 We therefore tested whether prolonged loss of total sialyation can lead to tumorigenesis. We used the IEC Slc35a1^–/– mice because this develops more severe spontaneous disease at birth. We let the IEC Slc35a1^–/– mice age to 18 months, and found spontaneous tumors in distal colons in aged IEC Slc35a1^–/– mice (6–12 months old), which increased both in size and incidence (30%) in mice 12–18 months of age (n = 11 for all aged mice) (Figure 7A and B). These appeared to be tubular adenomas, with mixed hyperplastic and neoplastic epithelium with evidence of invasive carcinoma (Figure 7A and C). To gain insight into how the loss of intestinal epithelial glycan sialylation impacts tumor susceptibility, we first investigated the proliferation status of IEC Slc35a1^–/– via Ki67 staining on distal colon cross sections, finding increased proliferative zones in the crypts (Figure 7D), as expected based on the hyperplastic appearance of the mucosa. This was concomitant with heightened DNA damage via double-stranded breaks as shown by γH2A.X staining (Figure 7E) as expected in chronic inflammatory environments.³⁶

Figure 7.

Loss of the intestinal epithelial sialome impacts colonic epithelial stem cell function and predisposes to spontaneous tumorigenesis. (A) Left: representative gross morphology of murine colon of 15-month-old mice. Right: representative H&E staining of colonic tissue (15 months) in low and high magnifications. Black arrow: tumor. Black arrowhead: invasive glands. Yellow arrowhead: inflammatory cell infiltration. Scale bars, 200 mm. (B) Top: mean tumor volume. Bottom: tumor incidence (percentage of mice with tumors). (C) Quantification of average numbers of invasive glands/section. (D) Representative IHC staining for Ki67 in distal colon. Graphs: enumeration of proliferating nuclei in E. Scale bars, 50 mm. (E) Immunostaining for γH2A.X in distal colon sections. Graph: quantification of γH2A.X. nuclei in distal colons. Scale bars, 50 mm. (F) Representative AB staining of a normal colon crypt showing strong/diffuse staining at the base of crypts where stem cells reside. Arrows: diffuse AB staining. Scale bars, 50 mm. (G) Breeding strategy of Lgr5-GFP mice used in the experiment. (H) Immunostaining for Lgr5/Ki67 in distal colon sections. Graphs: number of Lgr5⁺ stem cells per HPF. HPF, high-power field. Scale bars, 50 mm. (I) Representative first passage of distal colon–derived organoids at successive days after culture. Scale bars, 200 mm. (J) Micrograph of a representative cultured spheroid on Day 4. Graph: spheroid diameter quantification. Scale bars, 200 mm. (K) IF staining Ki67 on FFPE colonoids. Graph: enumeration of Ki67⁺ nuclei of colon organoids. Scale bars, 50 mm. Results are representative of 3 independent experiments each with N = 4 or 5 mice per experiment. Data are means ± standard deviation. ∗P < .05.

Recent studies have suggested intestinal sialylation defects lead to the overproduction of microbial-derived butyrate and consequently butyrate-dependent inhibition of colonic stem cell function that delays regeneration.²³ However, we did not observe enhanced butyrate in our Sia ablation model (Figure 6G), suggesting an alternative mechanism. We noted that diffuse Alcian blue (AB)-stained crypt bases of WT mice were missing in IEC Slc35a1^–/– mice (Figure 7F), suggesting an altered stem cell niche. To determine if this impacted stem cell numbers, we crossed IEC Slc35a1^–/– with Lrg5 ^GFP-CreER transgenic mice³⁷ to examine this directly (Figure 7G). Unexpectedly, we found that the numbers of Lgr5⁺ cells in the distal colon of IEC Slc35a1^–/– were reduced relative to WT mice, with the remaining Lgr5⁺ cells in IEC Slc35a1^–/– mice still showing proliferation signals (Figure 7H). To determine if this led to any intrinsic impact on epithelial stem cell function, we harvested distal colon crypts from WT and IEC Slc35a1^–/– mice for spheroid culture to monitor stem cell activity ex vivo. We found that IEC Slc35a1^–/– organoids grew larger (Figure 7I), even after passage, where they maintained a higher proliferative state (Figure 7J and K). However, this proliferative response does not seem to be effective after acute injury caused by DSS (Figure 4). Collectively, these results suggest that epithelial sialylation has a dual impact on mucosal homeostasis that impacts tumor susceptibility: a defective mucus system that leads to microbial-induced inflammation; and an altered stem cell niche that enhances stem cell activity in response to spontaneous inflammation but not acute injury.

Loss of Epithelial Sialylation Induced Delayed Gut Regeneration After 5-FU-Induced Injury

Although sialylation deficiency is also present in small IEC (Figure 3E), the mutant mice did not exhibit any noticeable differences in small intestinal phenotype compared with WT mice (Figure 3F). However, we observed a significant reduction in small intestinal Lgr5⁺ stem cells when sialylation was deleted in the epithelial cells (Figure 8A). Therefore, we used 5-fluorouracil (5-FU), a chemotherapy drug used for colorectal cancer treatment, to induce mucositis in the small intestine and investigate how loss of epithelial sialylation affects injury and regeneration in the small intestine. To do this, we administered 5-FU (50 mg/kg) intraperitoneally to IEC Slc35a1^{–/–; Lgr5-GFP} and WT^Lgr5-GFP mice for 5 consecutive days and allowed them to recover under normal conditions (Figure 8B). We monitored body weight daily during the treatment period and expressed it as a relative change (%) from Day 0. Both IEC Slc35a1^{–/–; Lgr5-GFP} and WT ^Lgr5-GFP mice showed significant weight loss during 5-FU treatment. However, IEC Slc35a1^{–/–; Lgr5-GFP} mice showed slower weight recovery than WT mice (Figure 8C). Hematoxylin and eosin staining revealed significant villi shortening with fattened and vacuolated cells, inflammatory cell infiltration in the lamina propria, loss of normal crypt architecture, vacuolization, and neutrophil infiltration in the ileal mucosa in both IEC Slc35a1^{–/–; Lgr5-GFP} and WT^Lgr5-GFP mice after 5-FU treatment (Figure 8D). Interestingly, even after 5 days of recovery, IEC Slc35a1^{–/–; Lgr5-GFP} mice still had ileal mucositis, whereas the ileum of WT mice had recovered from the injury (Figure 8D). After 5-FU treatment, Lgr5⁺ stem cells decreased and recovered over time in WT^Lgr5-GFP mice, but not in IEC Slc35a1^{–/–; Lgr5-GFP} mice (Figure 8E). We also found reduced proliferative activity (Figure 8E) and differentiation (Figure 8F) in IEC Slc35a1^{–/–; Lgr5-GFP} small intestinal epithelium. Additionally, severe mucositis was observed in the distal colon in IEC Slc35a1^{–/–; Lgr5-GFP} compared with WT^Lgr5-GFP mice (Figure 8G). Thus, loss of glycan sialylation is associated with delayed small and large intestinal regeneration after 5-FU-induced injury through changes in stem cells.

Figure 8.

Loss of epithelial sialylation in the small intestine leads to delayed regeneration after 5-FU-induced mucositis. (A) Tiled images showed Lgr5 stem cells in the ileum. Scale bars, 100 mm. (B) Schematic of 5-FU treatment experiments. (C) Body weight change monitoring. (D) H&E-stained images of ileum on Day 0, 6, and 10. Graph: histologic colitis score. Scale bars, 50 mm. (E) Confocal Z-stack images of ileum tissue sections at different timepoints. Graphs: number of Ki67⁺ epithelial cells and Lgr5⁺ stem cells per HPF.. Scale bars, 50 mm. (F) Confocal Z-stack images of Muc2 staining in ileum tissue sections at different timepoints. Graphs: number of Muc2⁺ goblet cells per HPF. Scale bars, 50 mm. (G) H&E-stained images of distal colon on Day 0, 6, and 10. Graph: histologic colitis scored. Scale bars, 50 mm. Results are representative of 3 independent experiments each with N = 4 or 5 mice per experiment. Data are means ± standard deviation.∗P < .05. HPF, high-power field; ns, nonsignificant.

Discussion

Glycosylation is an essential posttranslational modification in the mammalian gut that protects from acute and chronic diseases including inflammatory bowel disease and colorectal cancer.^9,36,38,39 Glycoprotein sialylation has emerged as a critical protective posttranslational modification in the intestine.23, 24, 25 Our study is the first to explore the impact of total ablation of glycan sialylation in the intestine by knocking out the CMP-Sia transporter Slc35a1 specifically in gut epithelial cells. Using this model, we found that mice lacking epithelial sialylation presented with microbiota-driven spontaneous colitis, colorectal cancer, and a hypersusceptibility to acute injury by DSS and 5-FU, which were dependent on multiple factors. First, consistent with recent reports on the protective functions of specific types of Sia on mucins,23, 24, 25 we found the total loss of overall sialylation impaired the function of gut mucins by reducing thickness of the proximal-derived b1 barrier layer. Although colitis was linked to an altered microbiota, this microbiota seemed functionally similar to WT as assessed through metabolomic approaches. Second, sialylation has mucus-independent impacts on epithelial stem-cell function, which delayed regenerative responses to both acute injuries by DSS or the chemotherapeutic agent 5-FU, a known inducer of mucositis.⁴⁰ Importantly, the epithelial intrinsic functions of sialylation were dependent on region and mode of injury. Spontaneous colitis was caused by an abnormal response to the microbiota, which enhanced stem cell activities that were linked to susceptibility to tumorigenesis. However, acute injury led to reduced stem cell responses because of reduced overall Lgr5⁺ cell representation in vivo. Overall, these studies reveal that glycan sialylation is essential for maintaining host-microbiota interactions and governing stem cell responses during epithelial injury.

Total loss of intestinal epithelial sialylation impaired mucus barrier functions, mostly of the b1 barrier layer. We have previously demonstrated that loss of epithelial O-glycosylation via deletion of the O-glycan core synthesizing enzymes C1GalT1 and C3GnT (B3GNT6) led to the destruction of the Muc2 barrier layer and ensuing microbiota-dependent chronic colitis and cancer.³⁶ Because this model ablates all complex glycans, it is difficult to discern if specific sugars on the glycans play disproportionately protective roles. Our results are in line with a recent study showing α2,6-linked sialylated GalNAc is a protective moiety on mucin oligosaccharides that can protect against glycoproteolytic degradation in vitro and impairs mucus thickness in vivo.²³ However, it is surprising that we did not see a more dramatic loss of mucus integrity despite the total loss of Sia in our model versus the partial loss in other models.²³ Another possibility for mucus defects is that the loss of sialylation can alter the colon mucin network structure; altered charge distribution on α2,6-linked sialylated GalNAc-deficient glycans has been reported to affect mucus architecture and make it more penetrable to the microbiota.²⁴ One interesting phenotype of the Slc35a1-deficient colonic mucus and goblet cells was the compensatory increase in mucin sulfation. Glycan sulfation on mucins has been implicated in preserving mucin integrity in the presence of bacterial enzymes,⁴¹ even acting as a rate-limiting modification in complex glycan breakdown by glycosidases.⁴² It is interesting to note that only the highly-sulfated b2 layer was not as impacted as the b1 layer in our model. Whether compensatory sulfation in response to Slc35a1 deficiency protects mucus integrity or affects its fecal-adherent network structure remains to be determined. Importantly, although our current study focuses on sialylated O-glycosylated Muc2, which is the primary component of colon mucus, the potential role of sialylation loss on N-glycans warrants further investigation.

The development of colon tumors in the absence of intestinal epithelial sialylation was in some respects surprising. Glycan sialylation is generally a protumorigenic modification, particularly when tumor cells acquire sialylated neoantigens. Increased exposure of sialyl-Tn antigen is associated with neoplasia with a generally poor prognosis.^17,18,43, 44, 45, 46 However, it is well established that chronic inflammation itself is a driver of neoplasia through multiple mechanisms that may indirectly relate to sialylation.47, 48, 49, 50 Indeed, we noticed increased DNA damage in the inflamed tissues and activation of caspase 1 mice lacking intestinal O-glycans.^36,51 These previous findings link impaired O-glycan expression to colonic mucus barrier breaches, which leads to caspase-1-dependent inflammation and tumor development. Our current research further identifies an increase in caspase-1 activation in intestinal sialylation-deficient mice, underscoring the role of inflammasome-mediated inflammation in tumorigenesis within colonic epithelial cells.

Ablation of intestinal epithelial sialylation had a dramatic impact on microbiota profiles. In a previous study with a lack of α2,6 sialylation, there were several increases in mucin degraders, such as Akkermansia muciniphila and others, which had a functional impact on metabolite profiles including increased production of SCFA, such as butyrate.²³ In contrast, we did not observe an increase in these mucin utilizers; however, this could easily be attributed to differences in baseline community and differences in the model. For example, our baseline community surprisingly had little representation of Bacteroides and Prevotella spp in our facility, as well as undetectable A muciniphila. Furthermore, because we wanted to limit the impact of inflammation on the microbiota, we used an inducible model system and monitored change in the luminal microbiota over ∼10 days in WT versus littermate TM-IEC Slc35a1^–/– mice. It was notable how fast loss of sialylation changes the microbiota. One unexpected feature of the altered microbiota in the absence of host epithelial sialylation is the fact that most of the taxa that were significantly changed by Sia loss underwent a reduction in overall abundance, whereas only the Firmicute C disporicum was consistently more represented. This suggests that although few species seem to thrive in the absence of Sia, many others may rely more heavily on this sugar and are less adaptive in its absence. One limiting factor is that the data are relative values; whether the reduced overall abundance impacts absolute numbers is still not clear. It would be important to look at the longer term to see if the relative abundance rebounds over time. The other limiting factor is our smaller sample sizes, which allowed for only limited statistical conclusions after correcting for multiple comparisons. Ultimately, despite the changes we observed in the microbiota, we did not see any changes in SCFA composition, suggesting that altered microbiota function caused by Sia loss was not contributing to the disease per se. However, because ABX treatment ameliorated spontaneous disease, it is clear the microbiota is involved in the disease progression, and this is likely via the mucus defects.

Unlike the colon, epithelial Sia loss did not alter the homeostasis of the small intestine. However, when we challenged the small intestine with 5-FU, we saw worsened mucositis in the small intestine of IEC Slc35a1^–/– mice versus WT littermates. We traced this back to altered stem cell response, because IEC Slc35a1^–/– mice had reduced Lgr5⁺ cells in the small intestinal crypts versus WT mice, which delayed overall regeneration. We saw a similar delay in the regeneration of colon crypts in response to 5-FU and DSS colitis. Paradoxically, the spontaneous disease was associated with the enhanced proliferative capacity of organoids of IEC Slc35a1^–/– mice ex vivo. These results suggest the specific response of epithelial stem cells to sialylation loss is contingent on the nature of the insult, with altered microbiota-host relationships at the gut-lumen interface boosting IEC Slc35a1^–/– stem cell function, whereas acute injury by chemical toxins impairs stem cell function. Ultimately, these acute (chemically induced) injuries enhance morbidity, but chronic (microbe-induced) injury increases the likelihood of spontaneous tumors. Although beyond the scope of the current study, future studies should look more deeply at molecular changes of colon stem cells, such as via spatial transcriptomics in the absence of sialylated epithelial glycans and their niche.

It is noteworthy that SLC35A1 mutation causes a rare congenital disorder of glycosylation, namely SLC35A1-CDG,⁵² which is characterized by thrombocytopenia,^27,52,53 developmental delay, intellectual disability, seizures, and vascular system abnormalities.^52,54 The thrombocytopenia is likely caused by decreased production and increased clearance of unsialylated platelets based on mouse models of platelet-specific Slc35a1 loss.²⁷ There is no report on patients with SLC35A1-CDG have intestinal dysfunction, which is an interesting topic to investigate in the future.

Materials and Methods

Access to Data

All authors had access to the all data and have reviewed and approved the final manuscript.

Mice

To create mice with a targeted sialylation deficiency in the IEC, we bred Slc35a1-floxed (Slc35a1^f/f) females with either VillinCre or VillinCreER^T2 males. This produced either permanent (IEC Slc35a1^–/–) or activatable (TM-IEC Slc35a1^–/–) sialylation knockout mice, along with their WT siblings. To activate the deficiency, mutant and WT/control littermates received daily intraperitoneal injections of 1 mg TM (∼40 mg/kg, MP Biomedicals) in a 1:9 ethanol/sunflower seed oil solution for 5 consecutive days. We analyzed the mice for the genetic deletion at least 5 days after the final TM treatment. Lgr5^GFP-CreER mice were acquired from Jackson Laboratories. All experimental and WT mice had a background of either pure C57BL/6J (BL/6) or a BL/6 and 129/SvImJ (129/Sv) blend. No significant genetic background-related effects were observed. Both male and female mice were used in most of the experiments. All animals were fed a standard diet (PicoLab Rodent Diet 20; LabDiet 5053) and kept in specific-pathogen-free conditions. The Institutional Animal Care and Use Committee of the Oklahoma Medical Research Foundation approved all animal research activities.

Tissue Processing

Colons with fecal materials were harvested from mice, and inundated with ice-cold Carnoy solution (60% methanol, 30% chloroform, and 10% acetic acid) or 10% formalin for fixing, and then processed for embedding into paraffin blocks. These Carnoy or formalin-fixed paraffin-embedded tissues were then sectioned longitudinally (5 μm thick) and deparaffinized and rehydrated using standard xylene and alcohol gradients for subsequent staining procedures.

Histology

For routine histology to determine changes in tissue morphology, paraffin sections were stained with hematoxylin and eosin. The brightfield images were obtained with a Nikon Eclipse E600 microscope equipped with a Nikon DS-2Mv camera, using the software NIS -Elements AR (V3.0) (Nikon Instruments). Histologic scoring was performed according to ours and other published methods,^9,55 with scoring systems appropriate for spontaneous colitis (including immune cell infiltration, crypt hyperplasia, and abscesses) or DSS-induced colitis, the latter which incorporated crypt dropout and ulceration status.

Immunostaining

Deparaffinized and rehydrated CFPE were incubated with streptavidin and biotin blocking solution (Vector Laboratories) to block endogenous biotin and avidin, followed by incubating in CAS-BlockTM (Invitrogen) serum-free blocking buffer for 10 minutes at 4°C. Sections were then incubated with rabbit monoclonal anti-Muc2 antibody (1:200, ab272692; Abcam) and/or biotinylated MALII (1:200, B-1265; Vector Laboratories); FITC-SNA (1:200, FL-1301-2; Vector laboratories) for 2 hours at room temperature, or overnight at 4°C, rinsed with phosphate-buffered saline, and then incubated with DyLight 594 or 647-labeled donkey anti-rabbit IgG (8 μg/mL, Jackson Immunoresearch) and DyLight 488-conjugated Streptavidin (5 μg/mL, Jackson Immunoresearch) for 1 hour at room temperature in the dark. Slides were washed 3 times in dH₂O, stained with 20 ng/μL DAPI for 5 minutes, washed again in dH₂O, and mounted with Permafluor mounting medium (Invitrogen). For some staining, fluorochrome-conjugated lectins were used, including DyLight 594-labeled UEA1 (1:200, DL-1067-1; Vector Laboratories) and FITC-SNA (1:200, FL-1301-2; Vector Laboratories). In this case, the lectins were incubated with the secondary antibody when costained with Muc2 as described previously. For lectin staining only, sections were incubated directly with the fluorochrome-conjugated lectins and then processed as described previously. For F4/80, Ki67, lysozyme, GFP, and γH2A.X staining, antigen retrieval was performed by incubating sections at 95°C–100°C for 20 minutes in a working dilution of Antigen Unmasking Buffer (Vector Biolabs, Burlingame, CA), and then cooling to ambient temperature. Formalin-fixed-paraffin-embedded sections were blocked with CAS-BlockTM (Invitrogen) serum-Free Protein Block for 10 minutes and then incubated at 4°C overnight with rat monoclonal antibody against mouse F4/80 (5 mg/mL, clone C1:A3-1; Bio-Rad), rabbit polyclonal anti-Ki67 (1:100, RM-9106-S0; ThermoFisher Scientific), rabbit monoclonal anti-lysozyme (5 mg/mL, ab108508, Abcam), rabbit polyclonal anti-GFP (5 mg/mL, ab290; Abcam) and rabbit monoclonal anti-γH2A.X (1:100, #9718; Cell Signaling Technologies) antibodies. For detection, sections were incubated with secondary antibodies for 1 hour at room temperature in the dark. Slides were washed 3 times in dH₂O, stained with 20 ng/μL DAPI for 5 minutes, washed again in dH₂O, and mounted with Permafluor mounting medium (Invitrogen). For anti-Ki67 IHC staining, sections were incubated with horseradish peroxidase–conjugated donkey anti-rabbit antibody (2 μg/mL; Jackson ImmunoResearch Laboratories) for 40 minutes at RT, detected using the 3,3’-Diaminobenzidine Staining Kit (Vector Laboratories), and counterstained with hematoxylin, dehydrated, and mounted. The slides were used for subsequent confocal imaging (described later).

Fluorescence In Situ Hybridization

Fluorescence in situ hybridization was performed as previously described.⁹ Briefly, deparaffinized sections were incubated at 37°C overnight with a Texas Red–conjugated universal bacterial probe EUB338 (5’-GCTGCCTCCCGTAGGAGT-3’; bp 337–354 within bacteria EU622773; Eurofins MWG Operon) in hybridization buffer (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl, 0.1% sodium dodecyl sulfate). The sections were rinsed with wash buffer (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl) and phosphate-buffered saline twice each and were coimmunostained for Muc2 staining as described previously.

RNA Extraction and RT-qPCR

Total RNA was extracted from snap-frozen colon tissues, or isolated colon crypts, using the RNeasy Kit (Qiagen) and 1.5 μg was reverse-transcribed using the Omniscript RT Kit (Qiagen). The cDNA was then used for expression analysis using an established primer set against the murine genes. For bacterial 16S rRNA analysis, bacterial gDNA was extracted from mouse fecal pellets using the Stool Mini Kit (Qiagen) and 5 ng was used for PCR. Hrpt1 was used as a housekeeping control. All expression analysis was carried out using the Bio-Rad CFX manager 3.0 and BioRad Maestro (using the ^△△Ct method for relative normalized expression and ^△Ct for relative expression). The sequences of primers were summarized in Supplementary Table 6.

Histochemical Staining of Mucins

High-Iron Diamine-Alcian Blue Staining

HID-AB staining for sulfated acidic mucins in situ were performed as previously described.⁵⁶ Briefly, deparaffinized and rehydrated sections were incubated overnight in solution containing 0.4% N, N-dimethyl-para-phenylenediamine dihydrochloride, 0.24% N, N-dimethyl-meta-phenylenediamine dihydrochloride and 1.7% ferric chloride, then rinsed thoroughly in dH₂O, and stained with AB pH 2.5, counterstained with nuclear fast red, and dehydrated and mounted as described previously. Staining intensities were further evaluated as described next (quantitative imaging).

Alcian Blue (pH 1.0) Staining

AB (pH 1.0) staining for sulfated acidic mucins in situ was performed as previously described.⁵⁷ Briefly, CFPE sections were immersed for 2 minutes in 0.1 M HCl. Staining with AB was performed with 1% AB 8GX (Sigma, Lot# MKCH5435) in 0.1M HCl (pH 1.0) for 30 minutes. After staining, slides were first rinsed in 0.1 M HCl containing 10% (v/v) Methanol (MeOH) (pH 1.0) 3 times, then rinsed with 0.5 M NaCl/HCl (pH 1.0) 3 times and counterstained with nuclear fast red, then dehydrated and mounted using Permount.

Imaging

The Nikon C2⁺ confocal microscope was used for regular confocal imaging with the NIS-element AR software (Nikon). Z-stacked confocal images were acquired with a 20x dry objective, or a 60x oil-immersion objective, with optical sections taken at 0.3–0.5 μm intervals. Images were analyzed using Imaris software (versions 7.7.2 and 9.5).

Quantitative Imaging

In situ mucus thickness on CFPE was determined as follows. For lectin-stained sections, 80–300 measurements were measured in Fiji/ImageJ in blinded fashion (ie, measurer did not know the identities of mice). Data was acquired from 4x, 10x, 20x, and 40x images with appropriate scales for each magnification used to calibrate Fiji. Each measurement was taken perpendicular to the upper and lower borders of the mucus barrier layer, which are clearly delineated when this layer is well-preserved. Three to 5 images/mouse (n = 3–8 mice) were analyzed, measurements were pooled among replicates, recorded in Microsoft Excel, and plotted in GraphPad Prism.

Low-Dose Dextran Sulfate Sodium Treatment

Cohoused littermates were treated with 1% DSS (∼40 kDa) dissolved in drinking water and filter sterilized (0.2 μm) or water for 7 days, followed by 4 days on filter-sterilized tap water. Mice were then euthanized for downstream analysis of mucus quality and histology of the colon. Body weight, the presence of occult blood per rectum, and stool consistency were measured daily during the course of treatments to obtain a clinical score as previously described.⁵⁸

Antibiotic Treatments

WT and IEC Slc35a1^–/– mice (cohoused littermates) were given 1 g/L each of metronidazole, neomycin sulfate, ampicillin, and 0.5 g/L of vancomycin (Sigma-Aldrich), and 0.1% sucrose in sterilized tap water. After 4 weeks of treatment, mice were switched to sterile tap water for 2 days, and microbial status was monitored by fluorescence in situ hybridization on fecal sections to confirm depletion of the microbiota. Mice were euthanized and tissue were collected for further studies.

Protein Extraction, Electrophoresis, Western Blotting, and Densitometry

Tissues from the distal colon were lysed by using RIPA lysis buffer. After beadmilling at 48 Hz for 1 minute at room temperature with a mini bead-beater (BioSpec Products), the samples were placed on ice for 30 minutes. The lysates were cleared by centrifugation and collected and quantified using the DC (detergent compatible) Protein Assay kit (Bio-Rad). Lysate proteins were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (15% acrylamide resolving gel) and transferred to a PVDF membrane by semidry transfer at 20 V for 45 minutes. The PVDF membranes were blocked by 5% skim milk in TBST buffer for 1 hour and then incubated with 1:1000 rabbit anti-proCaspase1 antibody (EPR16883, Abcam) or 1:5000 mouse anti-β-actin (MA1-91399, Invitrogen) at 4°C overnight. Membranes were washed with TBST, incubated with horseradish peroxidase–conjugated goat anti-rabbit or anti-mouse IgG for 1 hour at room temperature, and detected with SuperSignal West Pico PLUS detection agent (ThermoScientific). Densitometry of the caspase 1 p10 band was performed by using Fiji/ImageJ.

5-FU Treatment and Scoring

According to a published method with some modifications,⁵⁹ to induce mucositis, the 5-FU group mice were intraperitoneally administered with 5-FU (50 mg/kg) once daily for 5 days. After that, the mice were housed in the normal condition to recover for 5 days. The morphometric analyses of the intestine, and the histologic score, were performed as described by Leocadio et al.⁶⁰ Briefly, microscopic slides were reviewed by a histologist blinded to the experimental groups and damage extent. Data represent median values (and range) of scores from 0 to 3: Score 0, normal histologic findings; Score 1, villus blunting, loss of crypt architecture, sparse inflammatory cell infiltration, vacuolization and edema normal muscle layer; Score 2, villus blunting with fattened and vacuolated cells, crypt necrosis, intense inflammatory cell infiltration, vacuolization and edema, and normal muscle layer; Score 3, villus blunting with fattened and vacuolated cells, crypt necrosis, intense inflammatory cell infiltration, vacuolization and edema, and muscle layer showing edema, vacuolization and neutrophilic infiltration. Villus alterations were not considered for colon scores.

Organoid Culture

Colon crypts were isolated and cultured using an established method.³⁸ Briefly, isolated crypts were embedded in Matrigel on ice and seeded in a 24-well plate. 500 μL culture medium was added, and organoids were cultured at 37°C in cell culture incubator. On Day 4, organoids were passaged at 1:3 dilution and cultured at 37°C in a standard incubator (humidified, 5% CO₂). Images of organoids were taken using EVOS Cell Imaging System (Thermo Fisher Scientific) with a 20x objective at various time points.

Organoid Processing

Organoids were processed by our published method.³⁸ Organoid-containing matrigel was collected by mechanical disruption and aspiration from the well with a pipette together with wash media (DMEM/F12 with 10% fetal bovine serum, and 1x penicillin/streptomycin), transferred to a 1.5-mL microtube, and centrifuged (300 x g). The supernatant was discarded and the pellet was resuspended in 4% PFA and fixed for 10 minutes. Four percent Ultrapure low melting point agarose (Invitrogen) was then mixed with fixed matrigel/organoids 1:1 (v/v), and transferred into the middle of a ∼40°C presolified 4% low melting point agarose layer in a cryomold, and then overlayed with fresh low melting point agarose to form a “sandwich” structure. The agarose block was fixed with 4% PFA overnight at 4°C, and then processed for paraffin-embedding. Sections of 5 μm thickness were collected for histology and immunostaining as described previously.

Fecal Mucus Extraction and Glycan Analysis by Liquid Chromatography-Mass Spectrometry

Mucus Isolation

For mucus isolation from fecal pellets, fresh pellets were placed directly into a 2-mL tube and snap frozen in liquid N₂ and stored at -80°C until processing. At time of processing, frozen pellets were placed in 1 mL guanidium chloride (GuCl) extraction buffer (6 M GuCl, 0.1 M Tris pH 8.0, 1 mM EDTA) containing 2x cOmplete Protease Inhibitors (Millipore-Sigma) with 10x cOmplete Protease Inhibitors. The solution was bead-milled at 25 Hz for 1 minute at room temperature with a mini bead-beater (BioSpec Products). All samples were then extracted overnight at 4°C on a rotator. The sample was pelleted by centrifugation and pellets were resuspended and reextracted 3 more times. Following final reextraction, samples were reduced twice with 100 mM dithiothreitol to solubilize the mucin (once overnight at 37°C, the other for 4 hours at 37°C). Soluble mucins were then alkylated with 250 mM iodoacetamide on a rotator overnight at room temperature in the dark. The reduced and alkylated mucins were filtered through a 125-μm and 10-μm syringe filter in succession. The filtered mucins were then dialyzed into ddH₂O in a SpectraPor 1000 MWCO Dialysis Membrane (Repligen) using a SpectraFlo Dynamic Dialysis System, and the postdialyzed samples were then freeze-dried and stored at -20°C until analyzed.

Glycan Extraction

Fecal mucin samples, extracted and purified as described previously, were lyophilized and 1 μg 2’-fucosyllactose (Carbosynth, Compton, Berkshire, UK) was added as an internal standard. Mucin O-glycans were released by Carlson reductive β-elimination method using 100 μL of 50 mM NaOH with 1 M NaBH₄ at 45°C for 16 hours. Samples were cooled to 25°C and neutralized by adding 10-μL aliquots of 2-M acetic acid until bubbling stopped, vortex mixing thoroughly between additions. Solid-phase extraction was performed using 250 mg ENVI-Carb graphitic carbon cartridges (Supelco, Bellefonte, PA). Postquenching, the neutralized reaction mixture was loaded on preconditioned ENVI-Carb solid-phase extraction cartridges. Conditioning of the cartridges was done manually (using positive pressure) using 3 mL 80% acetonitrile (ACN) containing 0.1% TFA, followed by washing with 5 mL water. The sample was passed through the cartridge dropwise using a minimal amount of pressure, after which the solid-phase extraction cartridge was washed with 5 mL water followed by 2 x 2.2 mL of 50% ACN containing 0.1%TFA, which was collected. The 50% ACN/0.1% TFA eluant were dried in vacuo using SpeedVac (ThermoFisher Scientific). After drying down the sample for ∼3 hours to remove TFA, samples were snap frozen in liquid nitrogen and lyophilized. Samples were reconstituted in water and subjected to mass spectrometry analysis.

Liquid Chromatography-Mass Spectrometry

High-performance liquid chromatography was done using Agilent 1290 Infinity system, supplied with 1290 Infinity binary pump, 1290 Infinity autosampler, and a 1290 Infinity temperature-controlled sample compartment (Agilent Technologies, Santa Clara, CA). A Hypercarb (Thermo-Fisher; 100 mm × 2.1 mm; particle size 3 μm) high-performance liquid chromatography column was used for separation of glycans. The mobile phases used in ultra-high-performance liquid chromatography for separation of glycans are as follows: Mobile phase A = 10 mM ammonium bicarbonate; mobile phase B = 10 mM ammonium bicarbonate in 80% (v/v) ACN. The ultra-high-performance liquid chromatography gradient conditions (flow rate = 0.250 mL/min) are as follows: 0 minutes = 0% B; 5 minutes = 5% B; 15 minutes = 20% B; 20 minutes = 40% B; 25 minutes = 80% B, 27 minutes = 80% B, 27.1 minutes = 5% B. MS was performed using an Agilent 6530 Quadrupole Time-of-Flight with jet stream electrospray ionization source. Parameters for Quadrupole Time-of-Flight–MS in negative electron spray ionization mode are as follows: sheath gas flow = 10 L/min; sheath gas temperature = 350°C; drying gas flow = 10 L/min; drying gas temperature = 300°C; nozzle voltage = 1000 V; nebulizer = 35 psi; Fragmentor voltage = 175 V). Data were acquired in high resolution mode at a mass setting of 3200 m/z. To avoid instrumental drift reference ion solution containing 10 μM purine (m/z 119.0360 for [M-H]^-) and 2.0 μM HP-0921 (m/z 966.0007 and 1033.9881 for [M-H]^- and [M+HCOO]^-, respectively) in 95:5 ACN:H₂O was added postcolumn at 800 μL/min by an Agilent 1260 Infinity II isocratic pump (with 1/16th of it getting to the MS and remaining getting recycled). Full-scan spectra were collected at a rate of 2 Hz extended dynamic range with a mass range of 100–3200 m/z and the data were saved in profile format.

LC-MS data acquisition and analysis were performed using MassHunter Workstation software (Agilent Technologies): Data Acquisition Workstation (v B.06.01, SP1) and Qualitative Analysis (v B.07.00, SP2). A database was constructed with formulas corresponding to combinations of 1 or more of the following: N-acetylhexosamine (HexNAc), hexose (Hex), fucose (Fuc), or N-acetylneuraminic acid (Neu5Ac). At a minimum, all combinations contained 1 reduced HexNAc, and included also were formulas for O-sulfated glycans. The database was used with the Find-by-Formula algorithm in Qualitative Analysis to survey the data for formula matches within a ± 10.00 ppm mass accuracy limit and, when applicable, a ± 0.350 minute retention time window of the same peak in other samples. Peak areas and retention times were processed further in Microsoft Excel. To account for differences in sample masses, peak areas for all detected glycans were normalized to the total glycan signals for each sample. Glycans were analyzed and plotted in RStudio (R version 4.3.1) using an in-house R script, to quantify relative abundances of different groups of glycans (eg, neutral, acidic, fucosylated, sialylated) and see how they differed by genotype.

16S Library Preparation and Next-Generation Sequencing

16S library preparation for individual samples at Gut4Health (Vancouver, Canada) was prepared similar to the method described in De Wolfe and Wright.⁶¹ Briefly, the V4 region of the 16S rRNA gene was amplified with barcode primers containing the index sequences using a KAPA HiFi HotStart Real-time PCR Master Mix (Roche). PCR product amplification and concentration was monitored on a Bio-Rad CFT Connect Real-Time PCR system. Amplicon libraries were then purified, normalized, and pooled using the SequalPrep normalization plate (Applied Biosystems). Library concentrations were verified using a Qubit dsDNA high-sensitivity assay kit (Invitrogen) and KAPA Library Quantification Kit (Roche) following manufacturer details. The purified pooled libraries were submitted to the Bioinformatics + Sequencing Consortium at UBC, which verifies the DNA quality and quantity using an Agilent high-sensitivity DNA kit (Agilent) on an Agilent 2100 Bioanalyzer. Sequencing was performed on the Illumina MiSeq v2 platform with 2 x 250 paired end-read chemistry. Sequences were analyzed through Qiime2 for quality control, Divisive Amplicon Denoising Algorithm 2 (DADA2), a robust algorithm to generate high-quality assigned sequence variants to generate core-diversity metrics including alpha diversity (eg, Shannon Diversity Index), and beta-diversity (eg, Bray-Curtis dissimilarity). Established algorithms in Qiime2⁶² to assign taxonomy to assigned sequence variants using the new Greengenes2⁶³ were used. Basic statistical analysis of taxonomic differences between genotype and sex were determined via an in-house R script that will subset dataframes based on specific taxa at family and genus level, determine whether distributions follow a normal or nonnormal for each taxon, then perform t tests or Wilcoxon rank sum tests for normal and nonnormal data, respectively, and plot that data into boxplots. Taxa where unadjusted P values showing below the significance threshold (P < .05) were plotted as stated in the legends, and adjusted P values determined from false discovery rate correction using the Benjamini-Hochberg method are included in Supplementary Tables 3 and 4.

Gas Chromatography Mass Spectrometry

To analyze SCFA we followed protocols be Estaki et al.⁶⁴ Briefly, ∼50 mg of midcolon stool was homogenized with isopropyl alcohol, containing 2-ethylbutyric acid at 0.01% (vol/vol) as an internal standard, at 30 Hz for 13 minutes using metal beads. Homogenates were centrifuged twice, and the cleared supernatant was injected into a Trace 1300 gas chromatograph equipped with a flame-ionization detector with an AI1310 autosampler (Thermo Fisher Scientific) in spitless mode. Data were processed using Chromeleon 7 software. Half of the fecal material was freeze-dried to measure the dry weight, and measurements are expressed as micromoles per gram of dry weight. Data were analyzed and plotted in R.

Statistics

Unless otherwise indicated, all data were plotted and analyzed using GraphPad Prism version 8.0. Error bars represent standard deviation of biologic replicates. Data were put through the D’Agostino and Pearson K2 normality tests to determine the downstream statistical analysis. Data sets passing were subjected to Student t test (2 groups) or 1-way analysis of variance (3 or more groups). Data sets not passing were subjected to the Mann-Whitney test (2 groups), or Kruskall-Wallis test (3 or more groups). For multiple comparisons among 3 or more groups, Dunnett or Tukey multiple comparison posttest (following analysis of variance) or Dunn posttest (following Kruskall-Wallis) was used to calculate significance. A P value < .05 was considered to be significant.

Supplementary Material

Excel Table 1

Glycomics Analysis.

Excel Table 2

Bacterial 16s Sequencing Analysis.

Excel Table 3

T-Test Results of 16s Sequencing Analysis.

Excel Table 4

Wilcoxon Test Results of 16s Sequencing Analysis.

Excel Table 5

SCFA Analysis.