Cutting Edge: Intestinal Mucus Limits the Clonal Deletion of Developing T Cells Specific for an Oral Antigen Free

The mammalian gastrointestinal tract is home to a complex milieu of microbial and dietary Ags that require appropriate containment. Without this control, the host is at high risk of developing enteric infections that may lead to inflammatory bowel diseases (IBD) and severe intestinal inflammation (1). This segregation of luminal Ags is partially mediated by the structural barriers formed by the tightly bound epithelial cells that line the intestine and through the actions of specialized secretory goblet cells (2). Mucins are the key products produced by secretory goblet cells that expand upon release to form the intestinal mucus layer and protect underlying epithelial cells by physically shielding them from the luminal bolus (3). Notably, impaired mucin production and glycosylation have both been repeatedly described in intestinal biopsies from pediatric and adult IBD patients (4). However, the impact of mucosal defects on IBD pathogenesis or its extraintestinal manifestations are unclear.

Mucin-2 (MUC2) is the major secretory mucin (∼80% glycan by weight) of the intestinal mucus layer in both humans and mice (5–7) and functions as a scaffold for the various components of the mucus layer that overlies the intestinal epithelium (3). The importance of MUC2 to intestinal physiology is underscored by the phenotype of aged MUC2-deficient mice that includes the eventual development of spontaneous colitis and/or colorectal cancer, depending on genetic background (8, 9). At present, the precise mechanisms that lead to intestinal disorders in Muc2^−/− mice remain unclear. Notably, MUC2 deficiency not only causes the loss of the mucus layer but also impacts the localization of gut microbiota, disrupting homeostasis of the intestinal epithelium and eventually resulting in immune cell activation (9–11). Despite these consequences, 6–11-wk-old Muc2^−/− mice on the C57BL/6 background do not exhibit any overt differences in intestinal permeability relative to wild-type (WT) mice based on FITC–dextran assays (12). In addition, MUC2 has been suggested to mediate oral tolerance by polarizing dendritic cells (DCs) to an anti-inflammatory phenotype (13). Consequently, MUC2 may function as a protector of the intestinal barrier and an immune regulator.

To investigate the potential role of MUC2 in regulating intestinal immunity, WT and Muc2^−/− mice were orally gavaged with the model Ag OVA and monitored for CD8 T cell responses. In this study, we report that despite having no overt signs of colitis or intestinal epithelial disruption, OVA rapidly escaped the gut and spread through the circulation of Muc2^−/− mice, leading to the acute activation of peripheral Ag-specific CD8 T cells. Remarkably, oral OVA also precipitated the spontaneous loss of immature thymocytes in Muc2^−/−, but not WT, mice in an Ag-specific manner. Collectively, our studies suggest that defects in MUC2 expression could cause clonal deletion of developing thymocytes specific for dietary and intestinal microbial Ags.

Muc2^−/−, Vβ5 TCR, and Nur77^GFP transgenic (Tg) mice were all bred on a C57BL/6 background and described previously (9, 14, 15). Vβ5 TCR or Nur77^GFP Tg control mice were age- and sex-matched and bred in-house. Five- to ten-week-old WT and Muc2^−/− mice with or without Vβ5 TCR and Nur77^GFP transgenes were gavaged with 100 μl of PBS or 1 mg of grade III chicken OVA (Sigma-Aldrich) dissolved in PBS. All experiments followed protocols approved by the University of British Columbia Animal Care Committee and the Canadian Council on Animal Care.

Single-cell suspensions were incubated in FACS buffer (2% FBS/PBS) using specific Abs against TCRβ (H570597), CD4 (GK1.5), CD8 (53-6.7), CD11b (M1/70), CD11c (N418), CD16/32 (93), CD69 (H1.2F3), SIINFEKL-H-2K^b (eBio25-D1.16), F4/80 (BM8), and MHC class II (M5/144.15.2), and dead cells were excluded using fixable viability dye eFluor 780 (all eBioscience). Two to four million cells pretreated with Fc block (5 μg/ml anti-CD16/32 Ab) were incubated with PE- and BV421-labeled OVA_257–264-H-2K^b tetramers (National Institutes of Health Tetramer Facility). Data acquisition was performed with FACSDiva software using BD LSR II cytometer (BD Biosciences) and analyzed with FlowJo (Tree Star).

Goat anti-OVA polyclonal Ab (MP Biomedicals) was conjugated to 5-μm carboxylate-modified latex beads (Life Technologies) using the carbodiimide method. Small aliquots (∼5 μl) of blood were collected in heparinized Microhematocrit Capillary Tubes (Thermo Fisher Scientific) following tail poke before the addition of 10 μl of 50 mM EDTA in PBS containing plus 20 μl of FACS buffer. Plasma fractions were diluted with 50 μl of FACS buffer and 10 μl of 0.5% Tween 20 prior to the addition of 10 μl of carboxylate-modified latex beads (∼50,000) and being rocked overnight at 4°C. Next, beads were incubated with rabbit anti-OVA Ab (Biodesign International) and detected with goat anti-rabbit IgG F(ab′)₂-PE (Jackson ImmunoResearch Laboratories). PBMC pellets were treated with 10 μl 0.5% Tween 20 in PBS and brought up to 100 μl with FACS buffer prior to OVA detection described above.

Thymic tissue was digested in 500 μl of protein-free media containing 0.2 mg/ml collagenase IV (Worthington) for 1 h at 37°C. Next, samples were pressed through mesh, washed, and thymic DC purified using biotinylated CD11c Ab and streptavidin microbeads (Miltenyi Biotec). Thymic DCs were surface labeled with Abs against CD11c, CD11b, and SIINFEKL-H-2K^b prior to intracellular staining with rabbit anti-OVA Ab and goat anti-rabbit IgG F(ab′)₂-PE using BD Cytofix/Cytoperm.

All data are shown as the mean along with error bars representing the SEM. Data from multiple similar experiments were combined for presentation when possible. Otherwise, representative data are shown with the number of replicate experiments indicated. Groups were compared using two-tailed Student t tests to identify significance using GraphPad Prism 7 software.

WT and Muc2^−/− mice were bred onto the Vβ5 TCRβ-only transgene derived from the MHC class I–restricted OT-1 TCR to investigate how MUC2 deficiency affects CD8 T cells upon oral challenge with the model Ag OVA. CD8 T cells from Vβ5 TCR mice possess a semidiverse TCR repertoire that is skewed toward the recognition of OVA_257–264-H-2K^b MHC class I molecules with a range of affinities (16, 17). Six- to ten-week-old WT and Muc2^−/− Vβ5 TCR mice were orally gavaged with PBS or whole OVA protein, and the status of their CD8 T cells assessed at 16 h postchallenge using the acute activation marker CD69. The identification of OVA-specific CD8 T cells was performed using the same tetramer bound to two different dyes to increase the specificity as described by Nelson, et al. (18). Significantly, the mesenteric lymph nodes (MLNs) and spleens of Muc2^−/− Vβ5 TCR mice gavaged with OVA-exhibited reductions in the frequency and numbers of OVA-specific CD8 T cells relative to PBS treatment (Fig. 1A, 1B). The decreased representation of Ag-specific CD8 T cells in OVA-treated Muc2^−/− Vβ5 TCR mice could be due to activation-induced cell death or diminished detection related to T cell activation and concomitant downregulation of TCR and CD8 expression. In addition, Ag-specific CD8 T cells from both MLNs and spleens of Muc2^−/− Vβ5 TCR mice showed dramatic increases in CD69 levels upon gavage with OVA compared with WT mice (Fig. 1C, 1D; MLN: 10.1 ± 2.0 versus 1.8 ± 0.2-fold; spleen: 13.3 ± 3.2 versus 2.2 ± 0.8-fold). Thus, orally delivered OVA robustly activates cognate CD8 T cells at both local and distal sites of Muc2^−/− Vβ5 TCR mice.

One caveat with CD69 as a marker of TCR-induced activation is that its expression can also be triggered by inflammatory mediators like type I IFNs (15). Thus, the above experiments were repeated with Nur77 reporter strains in which TCR signaling is marked by GFP expression (15). MLN and splenic Ag-specific CD8 T cells from Muc2^−/− Nur77^GFP Vβ5 TCR mice strongly induced GFP expression upon OVA gavage relative to those from WT mice (Fig. 1C, 1D; MLN: 10.1 ± 1.8 versus 2.8 ± 0.4-fold; spleen: 12.3 ± 3.2 versus 2.0 ± 0.3-fold). Together, these studies indicate that MUC2 limits oral OVA from eliciting TCR signaling in cognate CD8 T cells.

A sensitive, cytometric bead assay was developed, detecting as little as 1 × 10⁻¹¹ g/ml OVA in PBS, to measure Ag trafficking in WT and Muc2^−/− mice after oral gavage (Fig. 2A, 2B). At 6 h postgavage, plasma of Muc2^−/− mice contained large amounts of OVA regardless whether they carried the Vβ5 TCR transgene (data not shown) or not (Fig. 2C). A kinetic analysis revealed that OVA was tangible in Muc2^−/− plasma by as early as 1 h, peaked between 3 and 6 h, and low levels remained detectable until 72 h postgavage (Fig. 2D). Further, OVA found its way into the blood of Muc2^−/− mice when supplied in drinking water, indicating that our findings are not an artifact of oral gavage (Supplemental Fig. 1A). In addition, oral gavage of a second protein bovine β-lactoglobulin also resulted in its trafficking into the circulation of Muc2^−/−, but not WT, mice (Supplemental Fig. 1B). Collectively, these findings suggest that MUC2 is critical for maintaining barrier function to some specific luminal Ags.

Systemic Ag could impact thymocyte selection through direct presentation within the thymus or indirectly via a cytokine storm produced by activated peripheral CD8 T cells. To visualize Ag presentation, CD11c⁺ APCs isolated from thymuses of WT and Muc2^−/− mice were stained for intracellular OVA and surface OVA_257-264/H-2K^b class I molecules and live gated as shown (Supplemental Fig. 1C). Thymic DCs from OVA-treated Muc2^−/−, but not WT, mice were enriched with cells labeling with OVA_257–264/H-2K^b and whole OVA Abs (Fig. 2E, 2F). Overall, these findings indicate that MUC2 helps restrict the capture and presentation of oral OVA in the thymus.

An i.v. injection of influenza virus hemagglutinin peptide (HA_126–138) induces massive deletion of HA_126–138-specific αβ TCR Tg double-positive (DP) thymocytes (19). To investigate whether the dissemination of oral OVA in Muc2^−/− Vβ5 TCR mice has insidious consequences for T cell development, cohorts of age- and sex-matched WT and Muc2^−/− Vβ5 TCR mice were orally gavaged with PBS or OVA, and thymic subsets quantitated 1 d later. Strikingly, OVA gavage of Muc2^−/− Vβ5 TCR mice resulted in marked 22 and 45% reductions in the proportion and numbers of DP thymocytes, whereas thymuses from WT Vβ5 TCR mice were unaffected (Fig. 3A, 3B). By contrast, T cell development in normal (non-Tg) WT or Muc2^−/− mice was not changed by oral gavage with OVA (Fig. 3A; data not shown). Altogether, these observations demonstrate that oral OVA travels systemically in Muc2^−/− mice in sufficient quantities to impact thymocyte selection.

We next investigated whether oral OVA causes thymocyte activation and TCR signaling in WT and Muc2^−/− Vβ5 TCR mice without or with the Nur77^GFP reporter (Supplemental Fig. 2A–C). Markedly, Muc2^−/− DP and CD8 single-positive (SP) thymocytes showed sizable rises in both CD69 and GFP levels upon OVA versus PBS gavage, whereas no changes were observed in WT mice. Next, examination of TCRβ-chain expression revealed that the frequencies and numbers of both TCR^lo DP and immature CD8 SP (TCR^lo CD4⁻ CD8⁺) thymocytes were decreased in Muc2^−/− Vβ5 TCR mice orally gavaged with OVA (Fig. 3C, 3D). By contrast, oral OVA gavage did not alter these thymocyte subsets in WT Vβ5 TCR mice. Thus, the preferential loss of TCR^lo DP and immature CD8 SP cells in Muc2^−/− mice suggests that these immature thymocytes are especially susceptible to OVA stimulation.

Next, we sought to determine whether oral OVA administration skews the frequency of OVA-specific cells among residual DP and CD8 thymocytes in Muc2^−/− Vβ5 TCR mice. However, MHC tetramers are somewhat constrained in detecting Ag-specific DP thymocytes because their CD8 molecules bind promiscuously to MHC class I tetramers independent of peptide, and they have low TCR expression, whereas CD8 SP thymocytes exhibit strict Ag-dependent binding like peripheral CD8 T cells (20, 21). Nevertheless, the proportion and numbers of Muc2^−/− Vβ5 TCR DP thymocytes that stain brightly with OVA-tetramer were reduced 3-fold and 5-fold, respectively, 1 d post-OVA gavage versus PBS (Fig. 3E, 3F). In addition, the frequency and numbers of OVA-specific Muc2^−/− Vβ5 TCR CD8 SP thymocytes decreased 2-fold and 4-fold upon OVA gavage. By contrast, no significant changes in frequency or cell numbers were observed among WT Vβ5 TCR thymocytes. Further, analysis of mature CD8 SP thymocytes from unmanipulated non-TCR Tg Muc2^−/− mice revealed that TCR Vβ repertoire varies from their WT counterparts (Supplemental Fig. 2D). Collectively, these results suggest that MUC2 expression shapes the TCR repertoire of developing T cells.

Central tolerance plays critical roles in preventing self-destructive immune responses through the removal of autoreactive thymocytes (negative selection) or instruction of a regulatory T cell fate. By contrast, nonresponsiveness to intestinal luminal Ags has been predominately attributed to peripheral tolerance because such Ags are generally not thought to reach the thymus (22). In this study, we present what we believe to be the first demonstration that oral Ag profoundly affects thymocyte selection when intestinal mucus is limiting. Specifically, we have found that MUC2 deficiency enables OVA to migrate rapidly from the intestinal lumen to the circulation, leading to the loss of cognate Ag-specific thymocytes. The increased intestinal permeability to OVA caused by MUC2 loss may be primary because of impact on barrier function or secondary, resulting from alterations in intestinal epithelial homeostasis, proinflammatory cytokine production, or gut microbiome (8, 9, 11, 23). Regardless, these findings may have major implications for the pathogenesis of IBD and other maladies associated with a leaky gut.

The remodeling of the TCR repertoire of Muc2^−/− Vβ5 TCR thymocytes by oral OVA likely results from Ag recognition either directly through thymocyte–APC interactions or indirectly via factors released by activated peripheral T cells. The size of the DP thymocyte loss (22% in proportion and 45% in cell number) in Muc2^−/− Vβ5 mice upon OVA gavage may be larger than predicted if solely based on their recognition of Ag because Zehn and Bevan (17) showed that ∼6% of peripheral Vβ5 TCR CD8 T cells are OVA-reactive—a number calculated from the proportion of T cell blasts that secrete IFN-γ upon peptide stimulation in vitro. However, the frequency of OVA-reactive cells in Vβ5 TCR mice could be much greater testing thymocytes and using an assay that detects lower affinity cells. Moreover, the Ag dose required for deletion of preselection DP thymocytes is considerably lower than what is necessary for peripheral T cell activation (24, 25). Accordingly, the loss of DP thymocytes in Vβ5 TCR Muc2^−/− mice upon OVA gavage may largely be attributed to direct Ag contact given the impressionability of preselection thymocytes.

The importance of intestinal barrier function in health and disease is becoming better recognized as studies show associations between increased intestinal permeability and IBD as well as other diseases including type 1 and 2 diabetes, obesity, rheumatoid arthritis, allergy, and some types of cancer (26). Interestingly, increased intestinal permeability often predates disease onset although the cause–effect relationship has yet to be established. Besides microbial infection, our findings indicate that a defective intestinal barrier poses the added risk of intestinal luminal Ags imprinting the TCR repertoire by affecting the fate of developing thymocytes. Moreover, we speculate that the negative selection of thymocytes expressing TCRs reactive against intestinal luminal Ags may lead to losses in the diversity of the TCR repertoire and possibly immunodeficiency. Collectively, our results provide new insights into how defects in intestinal mucus may contribute to the pathogenesis of immune deficiency, inflammation, and autoimmune diseases.

We are grateful to National Institutes of Health Tetramer Facility and Drs. Pam Fink (University of Washington) and Kris Hogquist (University of Minnesota) for providing Vβ5 TCR and Nur77^GFP breeder mice. The visual abstract was created from elements generated by Biorender.com and those drawn using Intaglio software (Purgatory Design).

The authors have no financial conflicts of interest.