C57BL/6 (B6) SPF mice were maintained in our SPF animal facility. B6 mice were originally purchased from the Jackson Laboratory. GF mice and AF mice were bred in sterile isolators in our GF mouse facility at POSTECH. GF B6 breeding pairs were raised with an ultra-filtered low-molecular-weight, chemically defined elemental diet [designated as antigen-free diet (AFD)], and offspring were designated as AF mice (11, 13, 14). Hence, AF mice are GF mice fed with AFD (13). Three-week-old or 10- to 11-week-old age-matched SPF, GF, and AF mice were used for each experiment. In some cases, GF mice were fed with a commercially available amino acid diet (AAD). Detailed composition of AFD and AAD has been described previously (13, 16). Also, 6- to 10-week-old IL-15 knockout (KO) mice, IL-12Rβ2 KO mice, or CD4-dominant negative (DN) TGFβRII transgenic (Tg) mice were used with age-matched B6 mice. All animal experiments were approved and performed in accordance with ethical guidelines by the Institutional Animal Care and Use Committee of POSTECH (IACUC #POSTECH-2014-0021, #POSTECH-2016-0050, and #POSTECH-2025-0028).

Listeria monocytogenes (LM) strain 10403s carrying a recombinant internalin A with a mutation (InlA^M) (naturally streptomycin-resistant), kindly provided by Brian S. Sheridan, was used for infection experiments. The mice were restricted from food and water for approximately 5 h prior to infection. A total of 5 × 10⁸ colony-forming units (CFU) of LM was orally infected per mouse.

The small intestine and large intestine were open longitudinally to expose the luminal side after Peyer’s patches from the small intestine were removed and then cut into 5-mm pieces. These fragments were incubated with DPBS (WelGene, Korea) containing 3% fetal calf serum (FCS) and 1 mM EDTA (WelGene) at 37°C while being stirred for 30 min. To eliminate epithelial cells, the supernatant was filtered with a 40-μm cell strainer (SPL, Korea), and suspended cells with 75% Percoll (GE Healthcare, USA) were overlaid by 40% Percoll and then centrifuged for 20 min at room temperature without brake. Afterwards, IELs were collected from the interface at a 40%–75% Percoll gradient.

Cell suspensions were prepared and pre-blocked with anti-CD16/CD32 (93). For cell surface staining, the following fluorescent monoclonal antibodies (mAbs) were used: anti-CD45.2 (104), anti-TCRβ (H57-597), anti-TCRγδ (GL3), anti-CD4 (RM4-5), anti-CD8α (53-6.7), anti-CD8β (YTS 156.7.7), Thy1.1 (HIS51), and CD11c (N418). For intracellular staining of interferon-γ (IFN-γ) (XMG1.2), cells were stimulated by PMA and ionomycin for 4 h and then fixed and permeabilized with kits from BD Biosciences. For intracellular staining of granzyme B (GB11), T-bet (4B10), and Blimp1 (3H2-E8), cells were fixed and permeabilized with kits from BD Biosciences or eBioscience (USA). Dead cells were excluded by labeling with propidium iodide or Ghost viability dye (Tonbo, USA). Cells were stained with mAbs for 20 min at 4°C. Samples were analyzed with LSR II or FACSCanto II (BD Biosciences, USA) and analyzed by FlowJo software. To calculate the absolute number of the cell population of interest, we multiplied its frequency, determined after applying multiple gating strategies in flow cytometry, by the total leukocyte count measured using a live cell counter (Vi-Cell XR, Beckman Coulter, USA).

Naïve Thy1.1⁺ OT-I CD8⁺ T cells were obtained from Thy1.1 OT-I. B6 mice or Thy1. OT-I Rag1 KO mice and adoptively transferred into non-irradiated SPF, GF, or AF B6 (Thy1.2) mice. One day after transfer, the host B6 mice were fed with 5 mg/mL ovalbumin (OVA, Sigma-Aldrich, MO, USA, grade V) in drinking water for 1 week.

Mice were intraperitoneally (i.p.) injected with 1 μg of recombinant IL-6 (eBioscience, carrier-free), 2 μg of IL-12p70 (eBioscience, carrier-free), or 2 μg of IL-27 (eBioscience). Adult GF B6 mice were injected every 3 days for 3 weeks. Host mice adoptively transferred with OT-I cells were injected every other day for 7 days.

Duodenum from SPF, GF, and AF mice was cut longitudinally and embedded in OCT compound (Sakura, USA). The tissues were cut into 5-μm sections and frozen, and the sections were fixed with methanol at −20°C. The slide sections were pre-blocked with anti-CD16/CD32 (93). For immunofluorescence staining, the slides were stained with primary anti-CD4 (GK1.5, biotin) and CD8a (53-60.7, Alexa Fluor 488) mAbs overnight at 4°C and secondary streptavidin (Alexa Fluor 594) for 2 h at room temperature. After then, the stained slides were mounted using ProLong Gold™ antifade reagent with DAPI. Confocal microscopes (Olympus and Zeiss) were used for detection of immunofluorescence.

RNA was extracted from small intestine tissues and sorted IEL populations using TRIzol reagent (Invitrogen, USA), and cDNA was synthesized using the GoScript™ Reverse Transcriptase kit (Qiagen, USA). The Prdm1 primers were obtained from Life Technologies. Real-time polymerase chain reaction (PCR) using TaqMan was performed on ViiA™7 (Applied Biosystems). Relative gene expression levels were normalized by the amount of the gene encoding 18s rRNA.

RNA was extracted from FACS-sorted CD103⁺ CD11b⁺ DCs from the small intestine of SPF, GF, and AF mice (n = 6 per group) using TRIzol reagent (Invitrogen, USA) by following the manufacturer’s procedure. CD103⁺ CD11b⁺ DCs from three mice per group were combined to obtain sufficient amount of RNA for RNA sequencing (RNA-seq). mRNA was isolated from 1 μg of RNA by using oligodT. After removal of rRNA, mRNAs were reverse-transcribed to generate single-stranded cDNA using random hexamer and reverse transcriptase, followed by double-stranded cDNA synthesis. Double-stranded cDNA was fragmented to the appropriate size and used in a standard Illumina library preparation involving end-repair, A-tailing and adapter ligation, and PCR amplification. After quantification of library using the KAPAlibrary quantification kit, RNA-seq library was sequenced on NovaSeq (Illumina, USA) followed by cluster generation. Purified total RNA and RNA-seq were performed by Theragen Etex (Korea).

Two days after oral LM infection, tissues were harvested to measure infection burden. The spleen, mLNs, the entire small intestine with Peyer’s patches, and liver were ground using a 100-μm cell strainer and incubated with 0.05% Triton X-100 for 1 h at 4°C. To exclude luminal LM, entire small intestinal tissues were extensively flushed with sterile phosphate-buffered saline (PBS). The samples were plated in duplicate on brain heart infusion (BHI, MB Cell) Broth with Bacto™ agar plates added with 50 μg/mL streptomycin. Colonies were counted after 2 days at 37°C.

Data are presented as mean ± SEM. Unpaired two-tailed Student’s t-test or one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test was used. p < 0.05 was significant. *p < 0.05, **p < 0.01, ***p < 0.001.

To investigate the role of commensal microbiota and/or dietary macromolecules in IEL development, we analyzed IEL populations in the small intestinal epithelium of adult SPF, GF, and AF mice. Using an appropriate gating strategy, we distinguished all the conventional and unconventional IEL subsets (Figure 1A). Consistent with previous reports (17, 18), GF mice exhibited no significant reduction in the number of TCRαβ⁺ IELs compared to SPF mice (Figures 1A–C). The frequency and total number of conventional TCRαβ⁺ IEL subsets, including CD8αβ⁺, CD4⁺, and CD4⁺ CD8αα⁺, remained largely intact in GF mice (Figures 1D–F). Similarly, unconventional TCRαβ⁺ CD8αα⁺ and TCRγδ⁺ IEL subsets in GF mice were present in similar numbers to those in SPF mice, indicating a minor role of microbiota for the development of unconventional IELs in the small intestine (Figures 1G, H).

However, as observed in adult AF mice, deprivation of dietary macromolecules in GF mice led to a severe depletion of all conventional TCRαβ⁺ IEL subsets (Figures 1A–F). Immunohistological examination revealed a marked reduction of CD4⁺ and CD8⁺ IELs within the small intestinal villi of AF mice (Figure 1B). The total number of conventional TCRαβ⁺ IELs, including CD8αβ⁺, CD4⁺, and CD4⁺ CD8α⁺ subsets, was significantly lower in AF mice compared to SPF and GF mice (Figures 1C–F). In contrast, unconventional TCRαβ⁺ CD8αα⁺ IEL numbers were only partially reduced (~50%) while TCRγδ⁺ IELs remained relatively unaffected, even in AF mice (Figures 1G, H). These results suggest that dietary antigens are crucial for the development of conventional TCRαβ⁺ IEL subsets in the small intestine but have only a moderate effect on unconventional TCRαβ⁺ CD8αα⁺ IELs and no apparent impact on TCRγδ⁺ IELs.

Unlike T cells in other tissues, such as small intestinal LPs, IELs in the small intestine are poised to exert immediate effector functions upon TCR stimulation (19–21). Conventional TCRαβ⁺ IEL subsets readily produce both IFN-γ and granzyme B, whereas unconventional IELs express granzyme B and perforin following TCR stimulation (5, 22–24). Previous studies have shown that the cytolytic activity of IELs is moderately reduced in GF mice, suggesting a partial role for microbiota (8).

To precisely determine the influence of gut microbiota- and diet-derived antigens on IEL effector function, we first examined cytolytic activity in conventional IEL subsets from SPF, GF, and AF mice. Granzyme B production was clearly detected in all IEL subsets except for TCRαβ⁺ CD4⁺ IELs (Figure 2A). Among granzyme B-producing IELs, levels of intracellular granzyme B were not significantly reduced in GF mice compared to SPF mice (Figure 2B). However, in AF mice, granzyme B production was severely impaired across both conventional (TCRαβ⁺ CD8αβ⁺ and TCRαβ⁺ CD4⁺ CD8αα⁺) and unconventional (TCRαβ⁺ CD8αα⁺ and TCRγδ⁺) IELs, indicating that dietary antigens, even if unprocessed by the gut microbiota, are sufficient to induce granzyme B production in IELs (Figures 2A, B).

In terms of IFN-γ production, the requirement of gut microbiota and dietary antigens in IFN-γ production varied across conventional IEL subsets. IFN-γ production in conventional TCRαβ⁺ CD8αβ⁺ IELs was partially dependent on gut microbiota, while further deprivation of dietary antigens led to a severe reduction in IFN-γ production in this subset (Figures 2C, D). In contrast, IFN-γ production by TCRαβ⁺ CD4⁺ IELs was unaffected in GF mice but was completely abrogated in AF mice (Figures 2E, F).

T-bet (T-box expressed in T cells) is a transcription factor previously identified in certain IEL subsets in normal SPF mice (25, 26). While other transcription factors such as Eomesodermin can promote granzyme B expression, T-bet acts as a potent enhancer of granzyme B expression (27, 28). Consistent with the diminished granzyme B expression, T-bet was significantly reduced in conventional TCRαβ⁺ CD8αβ⁺ IELs in AF mice compared to SPF and GF mice (Supplementary Figures S1A, B). Likewise, T-bet expression was also downregulated in unconventional TCRαβ⁺ CD8αα⁺ and TCRγδ⁺ IELs from AF mice relative to those in SPF and GF mice (Supplementary Figures S1C, D). Flow cytometry and qPCR analyses further revealed that Blimp1 expression was significantly reduced in TCRαβ⁺ CD8αβ⁺ IELs from AF mice compared to GF mice (Supplementary Figures S1E, F), in line with their tissue localization patterns (29). Together, these findings highlight that dietary antigens are critical for optimal IEL effector function, at least in part, through the regulation of transcription factors such as T-bet and Blimp-1.

Exposure to solid food after weaning and subsequent microbial changes are considered as a critical factor for the development or maturation of the mucosal immune system (30). To investigate the ontogeny of IELs and IEL effector function under physiological conditions, we examined IEL populations in SPF mice before and after weaning onto a normal chow diet (NCD). Conventional TCRαβ⁺ CD8αβ⁺ IELs were scarce in 3-week-old mice but gradually increased following weaning onto NCD (Figures 3A, B), consistent with previous findings (1). The proportion of unconventional TCRαβ⁺ CD8αα⁺ IELs among TCRαβ⁺ CD8α⁺ cells was relatively higher in pre-weaned SPF mice compared to adults, reflecting the age-dependent expansion of conventional TCRαβ⁺ CD8αβ⁺ IELs (Supplementary Figure S2A). TCRγδ⁺ IELs also increased after weaning onto NCD (Supplementary Figure S2B). Furthermore, upon weaning onto NCD, granzyme B expression was substantially upregulated in both conventional TCRαβ⁺ CD8αβ⁺ IELs and unconventional TCRαβ⁺ CD8αα⁺ and TCRγδ⁺ IELs (Figures 3C, D, and Supplementary Figures S2A, B).

To further confirm that dietary antigens drive the IEL development, neonatal SPF or GF mice were weaned onto either NCD or AF diet (AFD). Regardless of the presence of gut microbiota, both SPF and GF mice weaned onto an AFD for 3 weeks showed a marked depletion of conventional TCRαβ⁺ CD8αβ⁺ IELs (Figures 3E–H). In contrast, weaning neonatal SPF or GF mice onto NCD for 3 weeks did not reveal a significant effect of AFD feeding on the frequency or total number of unconventional TCRαβ⁺ CD8αα⁺ IELs (Supplementary Figures S2C, E). Consistent with the levels of TCRγδ⁺ IELs observed in AF mice (Figure 1H), the abundance of TCRγδ⁺ IELs was similar between mice weaned onto NCD and those weaned onto AFD (Supplementary Figures S2D, F). Conversely, we examined IEL development in neonatal AF mice weaned onto a sterile NCD. Weaning neonatal AF mice onto NCD for 4 weeks did not significantly alter the total number of unconventional TCRαβ⁺ CD8αα⁺ IELs (Supplementary Figure S3A). The abundance of TCRγδ⁺ IELs remained similar among GF, AF, and AF mice weaned onto NCD (Supplementary Figure S3B). However, pre-weaned AF mice maintained onto NCD for 4 weeks developed TCRαβ⁺ CD8αβ⁺ IELs at levels similar to GF mice (Figures 3I, J), accompanied by an increase in granzyme B expression within this subset (Supplementary Figure S3C). Granzyme B expression in unconventional IEL subsets was restored to the level observed in GF mice (Supplementary Figure S3C).

Based on these findings, we also tested whether deprivation of dietary antigens leads to the impairment of TCRαβ⁺ CD8αβ⁺ IEL generation by using commercially available AAD, which are depleted of proteins, as reported previously (16, 31). Weaning neonatal GF mice onto both AFD and AAD effectively prevented the development of IELs, particularly TCRαβ⁺ CD8αβ⁺ IELs (Supplementary Figures S4A, B). In addition, granzyme B expression on the residual IELs that developed in AAD-fed GF mice was very low (Supplementary Figure S4C). Collectively, these findings indicate that dietary antigens exposed during the weaning period are essential for both the development and functional maturation of IELs.

Next, to investigate whether continued exposure to dietary antigens is required for maintaining conventional IEL subsets, adult GF mice with a full IEL compartment were fed with AFD for 8 weeks. Surprisingly, we found that the proportion and total number of conventional TCRαβ⁺ CD8αβ⁺ IELs remained stable despite AFD feeding (Figures 4A, B). However, their effector function, as measured by IFN-γ production and granzyme B expression, was significantly impaired by AFD feeding (Figures 4C–E). AFD feeding to adult GF mice also led to the impaired effector function in unconventional IELs, such as TCRαβ⁺ CD8αα⁺ and TCRγδ⁺ IEL subsets, although both IEL populations were also long-lived (Figures 4F–K). These findings suggest that while IELs are long-lived, their effector function requires continued stimulation by dietary antigens.

While continuous stimulation with dietary antigens is critical for maintaining the effector function of both conventional and unconventional IELs, intestinal cytokine milieu influenced by dietary macromolecules may also regulate their effector function. Previous studies have reported that intestinal epithelium shows high expression of IL-15 and TGF-β (32, 33) and that IELs exert cytolytic activity upon exposure to IL-12, IL-18, and IL-15 (34). To address this issue, in subsequent analyses, we focused on conventional TCRαβ⁺ CD8αβ⁺ IELs to avoid the complexity and heterogeneity associated with unconventional IEL subsets. Moreover, TCRαβ⁺ CD8αβ⁺ IELs can be effectively studied by using experimental approaches such as adoptive transfer of TCR-transgenic CD8⁺ T cells combined with feeding of cognate antigen as a model dietary antigen.

To determine whether these cytokines contribute to the effector function of dietary antigen-induced IELs, we first analyzed IELs in IL-15-deficient (IL-15KO), CD4-DN TGF-βR2 Tg (CD4-DN TGF-βR2 Tg), and IL-12Rβ2-deficient (IL-12Rβ2 KO) mice. Interestingly, the proportion of conventional TCRαβ⁺ CD8αβ⁺ IELs increased in IL-15KO mice. However, their total numbers and granzyme B expression remained unchanged in IL-15 KO mice (Supplementary Figures S5A–C). In CD4-DN TGF-βR2 Tg mice, the levels of TCRαβ⁺ CD8αβ⁺ IELs were significantly reduced, yet granzyme B expression was similar to WT mice (Supplementary Figures S5E–G). In contrast, IL-12Rβ2 KO mice exhibited a profound depletion of TCRαβ⁺ CD8αβ⁺ IELs, and the few remaining IELs failed to express granzyme B (Figures 5A–D). These results suggest that IL-12 signaling is critical for both the induction of TCRαβ⁺ CD8αβ⁺ IELs and their effector function, whereas either IL-15 or TGF-β signaling is dispensable for their effector function.

Notably, RNA-seq analysis of purified intestinal CD103⁺ CD11b⁺ DCs, the most abundant DCs in the small intestine, revealed that Il12b mRNA expression in CD103⁺ CD11b⁺ DCs was similar between SPF and GF mice but significantly reduced in AF mice (Figure 5E). Of note, levels of CD103⁺ CD11b⁺ DCs were reduced in AF mice compared to SPF and GF mice (13). To further investigate the role of IL-12 in regulating TCRαβ⁺ CD8αβ⁺ IEL effector function, we examined the effects of injecting innate cytokines, such as IL-6, IL-12, and IL-27, into adult GF mice fed with AFD. Since the effector function of IEL subsets declined following prolonged AFD feeding (Figures 4A–D), adult GF mice were given repeated cytokine injections for 3 weeks, starting at 5 weeks after switching the mice onto AFD. Intriguingly, IL-12 injection restored granzyme B expression in TCRαβ⁺ CD8αβ⁺ IELs (Figures 5F, G). IL-12 injection did not affect granzyme B expression on splenic and LP-activated CD8⁺ T cells (Supplementary Figures S6A, B). Collectively, these data suggest that dietary antigens as well as IL-12, presumably produced by intestinal DCs, are critical for the prolonged function of TCRαβ⁺ CD8αβ⁺ IELs.

We hypothesized that dietary antigen-induced IELs might provide protection against foodborne pathogens. To assess this idea, we orally infected adult GF and AF mice with LM expressing a mutated internalin A, which enables the bacterium to recognize murine E-cadherin expressed on intestinal epithelial cells in mice (35). Although LM can breach the intestinal epithelial barrier in an internalin A-independent manner (36), we selected this LM strain because it efficiently transverses the intestinal epithelium, where IELs are abundant. Mice were infected with LM via oral gavage to mimic listeriosis, and bacteria burden was assessed by analyzing CFU in the small intestine, liver, and spleen. In line with our hypothesis, we observed that AF mice exhibited a significantly higher infection burdens in the small intestine and liver, but not in spleen, compared to GF mice (Figure 6A). These findings suggest that dietary antigen-induced IELs contribute to early and local protection against LM intestinal infection.

To further investigate the role of dietary antigen-induced TCRαβ⁺ CD8αβ⁺ IELs in LM protection, we utilized SPF Rag1 KO mice adoptively transferred with OT-I transgenic CD8⁺ T cells, which are specific to OVA; these mice lack all IEL subsets and OVA feeding allows for the repopulation of OT-I cells in the intestinal epithelium (Figure 6B). In addition, we administered IL-12 to examine its role in the induction and effector function of TCRαβ⁺ CD8αβ⁺ IELs in an antigen-specific manner. We found that in Rag1 KO mice adoptively transferred with OT-I CD8⁺ T cells, OVA feeding combined with IL-12 treatment not only enhanced the expansion of OT-I CD8⁺ T cells in the intestinal epithelium (Figures 6C, D) but also increased granzyme B production, compared to IL-6 treatment (Figure 6E).

Next, we orally infected these mice with LM lacking OVA expression to evaluate whether OT-I CD8⁺ IELs could provide early protection in an antigen-independent manner (Figure 6F). Although OVA feeding facilitated the repopulation of OT-I CD8⁺ T cells within the intestinal epithelium, these cells alone failed to provide early protection against LM infection. However, IL-12 treatment significantly reduced LM levels in the small intestine and liver at day 2 post-infection (Figure 6G). Collectively, these findings suggest that dietary antigen-induced IELs play a critical role in providing innate-like protective function, thereby contributing to rapid and local protection against foodborne infectious pathogens.