An Integrated Single-Cell Atlas of the Mouse Ileum Links Nutrient Metabolism with Epithelial and Immune Crosstalk

Abstract

Background

The ileum integrates nutrient absorption with mucosal immunity, yet its cell-type–specific functions remain poorly defined. Disruption of epithelial or immune pathways contributes to nutrient deficiency, Crohn’s disease, and impaired barrier integrity. Single-cell RNA sequencing (scRNA-seq) provides the resolution needed to uncover epithelial differentiation and immune crosstalk that bulk approaches cannot resolve.

Objective

This study aimed to map ileal cellular heterogeneity and define epithelial differentiation, nutrient metabolism programs, and epithelial–immune interactions relevant to health and disease.

Methods

scRNA-seq was performed on ileal cells from 8-week-old male C57BL/6J mice. Gene expression and clustering were analyzed using Seurat, with pseudotime trajectory, cell–cell communication, and pathway enrichment analyses applied to characterize intestinal dynamics.

Results

A total of 32,076 ileal cells were identified, including six epithelial types and multiple immune populations. Enterocyte subclusters showed distinct nutrient-related functions: Ent_C1, C4, C7, and C8 were enriched for vitamin A absorption; Ent_C0, C1, C2 and C7 for carotenoid metabolism; and Ent_C1, C4, C7, C8, and C9 for vitamin B12 absorption. Co-expression of β-carotene oxygenase 2 (Bco2) and interleukin 18 (Il18) occurred across enterocytes, stem cells, and goblet cells, whereas non-canonical goblet cells exhibited high Bco2–Il18 expression together with signatures of fatty acid metabolism and stress responses. Plasmacytoid dendritic cells were identified as central regulators of immune–epithelial interactions.

Conclusions

This study provides the first integrated single-cell atlas of the mouse ileum, profiling both epithelial and immune cells and revealing nutrient metabolism programs and epithelial–immune crosstalk relevant to intestinal health and disease.

Introduction

The ileum is the distal segment of the small intestine and plays essential roles in nutrient absorption, microbial ecology, and mucosal immunity. It is the exclusive site of vitamin B12 and bile acid absorption [1,2], and also contributes to the uptake and metabolism of other nutrients, such as vitamin A and carotenoids [3,4]. Unlike the duodenum and jejunum, the ileum harbors a significantly denser and more diverse microbial community, which shapes both nutrient metabolism and immune regulation [5,6]. For example, segmented filamentous bacteria enhance the conversion of retinal to retinoic acid, a metabolite critical for immunity [7,8]. Such interactions highlight the ileum as one of the most intimate sites of diet–microbiome–host crosstalk, where nutrient absorption, microbial metabolism, and immune responses are closely integrated.

Dysregulation of these processes has been linked to human disease. The ileum is a major site of Crohn’s disease pathology, where epithelial barrier dysfunction, altered nutrient metabolism, and immune dysregulation intersect with microbial imbalance [9]. It is also implicated in malabsorption syndromes and other chronic intestinal disorders [10,11]. The ileum is not only as a site of nutrient uptake but also as a hub of metabolic–immune integration with broad implications for health and disease.

Despite its central role, the ileum remains relatively underexplored at the level of cellular heterogeneity. Most previous studies have relied on bulk transcriptomics, histology, or immunostaining, which average signals across populations and obscure the distinct contributions of specialized epithelial and immune cell subsets. Yet the intestinal epithelium is highly diverse, composed of enterocytes, goblet cells, Paneth cells, enteroendocrine cells (EECs), and stem cell populations, each with unique roles in nutrient metabolism, barrier function, and host–microbiome interactions [12,13].

Recent single-cell RNA sequencing (scRNA-seq) studies have advanced our understanding of intestinal cell diversity [14-16], and most have focused on the proximal small intestine or colon and rarely examined the ileum in relation to nutrient metabolism and epithelial–immune crosstalk. Furthermore, most single-cell studies of the intestine have analyzed either ileal epithelial cells or ileal immune cells in isolation, that limits the ability to infer cell–cell communication. Because datasets often originate from different animals or clinical conditions, such variability complicates interpretation and makes it difficult to link nutrient metabolism in epithelial cells with immune signaling in the same microenvironment.

To address this gap, we performed scRNA-seq on both epithelial and immune cells from the same mouse ileal samples. To reduce biological variability, only male mice were used, as estrous cycle–related hormonal fluctuations in females can influence intestinal physiology, immunity, and nutrient metabolism [17,18]. This integrated design enabled the first single-cell atlas of the mouse ileum simultaneously profiling epithelial and immune cells, resolving cell-type–specific pathways of nutrient metabolism and epithelial–immune interactions.

Materials and Methods

Animals

Eight-week-old male C57BL/6J wild-type (WT) mice and β-carotene oxygenase 2 (BCO2) knockout (KO) mice on a C57BL/6J background were used in this study. The KO strain was originally obtained from the Johannes von Lintig Laboratory at Case Western Reserve University [19] and subsequently backcrossed with C57BL/6J mice (Stock #000664, The Jackson Laboratory, Bar Harbor, ME, USA) for more than 10 generations. Heterozygous (BCO2^+/−) breeders were used to generate homozygous KO and WT littermates for experiments. Genotyping was performed by Transnetyx (Cordova, TN, USA). All mouse strains were maintained in an AAALAC-accredited animal facility at Oklahoma State University maintained at standard rodent housing conditions (approximately 20–26 °C, 30–70% relative humidity, 12 h light/12 h dark cycle), consistent with USDA, NIH, and Guide for the Care and Use of Laboratory Animals recommendations. The animals were group-housed (2 mice/cage) and fed AIN-93G rodent diet (formulated by Research Diets Inc. New Brunswick, NJ, USA). Bedding (Bed-o’ Cobs 1/8” and Bed-r’ Nest. LabSupply Texas, Haltom, TX, USA) and diet were changed weekly. All animal experiments and procedures were performed in accordance with the protocol approved by the Oklahoma State University Institutional Animal Care and Use Committee (OSU ACUP #HS-20-75).

Preparation of Single-cell Samples of Ileal Epithelial and Lamina Propria Tissues

All mice (WT and KO) were euthanized in the morning (between 9:00 and10:00) using a ketamine/xylazine cocktail (4.2 /0.42 mg/g BW, respectively) (VetOne, Boise, ID, USA) after 3-hour fasting. Single-cell suspensions of ileal epithelial and lamina propria tissues were prepared as previously described with minor modifications [20-22]. Briefly, the distal segment of the small intestine (e.g., the ileum, the 3rd 1/3 length of the small intestine) was dissected, flushed, and cut into small pieces. Samples were incubated in Hanks’ balanced salt solution (HBSS) containing 5 mM EDTA at room temperature to isolate epithelial cells. The remaining tissues were digested with 0.20 mg/mL collagenase type VIII (Sigma-Aldrich, catalog #C2139, St. Louis, MO, USA) and filtered through a 70 μm strainer, followed by separation using a 40%/80% Percoll gradient centrifugation. Cells were then carefully collected and washed in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with fetal bovine serum (FBS).

Trypan blue was used to test cell viability. Samples with cell viability greater than 80% were fixed using the Chromium Next GEM Single Cell Fixed RNA Sample Preparation Kit (10X Genomics, catalog # 1000414, Pleasanton, CA, USA) and then stored at −80 °C. Four mice, each randomly selected from separate cages (one mouse per cage), were used for scRNA-seq to ensure representative sampling across the cohort (n=4).

Library Construction, Sequencing, and Reads Processing

The 10X scRNA-seq libraries were constructed using a Chromium Fixed RNA Profiling for Multiplexed Samples library with the Chromium Fixed RNA Kit (10X Genomics, catalog # PN-1000496). Libraries were sequenced on a NovaSeq 6000 (Illumina) platform to obtain approximately 250 million reads per sample. Reads processing was performed using the Cellranger Software Suite (10× Genomics) following the guidelines. The raw data was merged and converted to FASTQ files, followed by alignment to the mouse genome GRCm38.

Cell Clustering, Marker Gene Identification, and Cell Annotation

scRNA-seq data were analyzed using Seurat (v5.1.0) in R [23-25], a standard and widely used framework in the field. Specifically, cells were first filtered to include cells with a minimum of 200 unique features and a maximum of 25 percent of mitochondrial genes. Genes detected in less than three cells were removed. Doublets were identified and removed using the DoubletFinder algorithm. To reduce the technical noise in the data, the gene counts were normalized by scaling sample-specific size factors, then log-transformed with the NormalizeData function implemented in Seurat. Next, top 2000 highly variable genes were selected by the Seurat function FindVariableGenes by default. Sequencing depth effect was regressed out using the ScaleData function. The RunPCA function was used to identify the principal components from the previously retained 2000 variable genes. Principal component analysis (PCA) was applied to reduce the number of dimensions representing each cell. The top 15 principal components, determined using an elbow plot, were retained as they captured the majority of the data variation. We also used Harmony, the batch effects correction method, to integrate datasets from each sample. Following this, the 15 principal components were used in uniform manifold approximation and projection (UMAP) to further reduce the dimensions. Then, cells were grouped into different clusters by Seurat functions FindNeighbors and FindClusters. Marker genes for each cluster at the optimal resolution were found using either the FindAllMarkers or FindMarkers functions in the Seurat package. Cells were annotated based on the expression of marker genes. Module scores were calculated using the AddModuleScore_UCell function of Seurat, with the gene list provided in Supplemental Table S1. Genes in certain functions and metabolic processes were pulled from MSigDB Curated Gene Sets (https://www.gsea-msigdb.org/gsea/msigdb). The Molecular Signatures Database (MSigDB) is a widely used, comprehensive repository of annotated gene sets for Gene Set Enrichment Analysis (GSEA), co-maintained by the Broad Institute and the University of California, San Diego. Subclustering of specific cell type(s) was accomplished by first subsetting the dataset to the cell type(s) of interest and subsequently following the pipeline described above.

Functional Enrichment Analysis

Marker genes identified from each major cell type or from each subtype of specific cell type(s) were subjected to functional enrichment analysis. We used the clusterProfiler package (v 4.12.6) to determine the enriched cellular pathways in Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Significantly enriched GO and KEGG terms were visualized with dot plots or bar charts.

CytoTRACE Analysis

CytoTRACE was developed for predicting cell absolute developmental potential from scRNA-seq data [26]. CytoTRACE 2 (v 1.0.0) was applied to calculate the potency scores of certain cell types. The results were plotted on the UMAP colored by potency scores as well as bar charts ranging from clusters with the highest potency to the lowest.

Single-cell Pseudotime Trajectory Analysis

Monocle2 (v 2.32.0) and Monocle3 (v 1.3.7) were employed for pseudotime trajectory analysis [27,28]. For Monocle2 analysis, genes with an average expression greater than 0.1 were included in this analysis. The single-cell trajectory was constructed by Discriminative Dimensionality Reduction via learning a Tree (DDRTree) algorithm. Cells were ordered along the trajectory, and pseudotime was calculated. The cellular potency scores from CytoTRACE analysis were used to determine the “root” of the constructed trajectory. The results were visualized using 2-dimensional figures that were colored by pseudotime or cell subtype annotations. A branch heatmap illustrating bifurcation expression patterns was generated using the plot_genes_branched_heatmap function based on differentially expressed genes (DEGs) with an FDR adjusted P value <1e−50 (Supplemental Table S2). For trajectory inference using Monocle3, the input dataset was preprocessed to retain 100 principal components. Batch correction and clustering were performed, followed by UMAP embedding in the corrected space. Cell trajectories were inferred using the learn_graph function. After defining root nodes, pseudotime along the intended trajectory was then calculated using the order_cells function. Finally, trajectories were visualized using the plot_cells function, displaying the trajectory graph in UMAP space with cells colored by pseudotime.

Correlation Analysis

Correlation between BCO2 and interleukin 18 (IL18) expression was assessed using the correlatePairs function, based on Spearman’s rank correlation [29]. Within each IEC type, cells co-expressing both Bco2 and Il18 were subsetted, and the significance of their expression correlation was tested by P (<0.05) and ρ (>0.2) values. The results were visualized using the plotExpression function.

Cell-Cell Communication Analysis

Cell-cell communication analysis was performed using CellChat (v2.1.2) [30]. All identified cell types were analyzed to infer intercellular signaling networks based on the CellChatDB.mouse ligand–receptor interaction database (https://rdrr.io/github/sqjin/CellChat/man/CellChatDB.mouse.html). Interaction counts and strengths were compared across cell types. Ligand–receptor pairs with P values < 0.05 were considered statistically significant, and their corresponding interaction probabilities were deemed reliable.

Data Visualization and Statistical Analysis

R package ggplot2 (v 3.5.1) was utilized to visualize data via pie charts and bar charts. Feature plots, heatmaps, violin plots, and dot plots were generated using Seurat FeaturePlot, DoHeatmap, VlnPlot, and DotPlot functions, respectively.

For marker gene identification, genes with an average log2-transformed fold changes (|log₂FC| ≥1) and an adjusted P-value < 0.05 were classified as DEGs. Differences in gene expression in goblet cell populations between WT and KO mice were evaluated using Student’s t-test, with P < 0.05 considered statistically significant. In enrichment analysis, GO and KEGG terms with adjusted P values less than 0.05 were considered significant.

Results

Single-cell profiling of intestinal epithelial and immune cells in the mouse ileum

scRNA-seq profiling was performed on ileum tissue samples from WT mice (Figure 1A). After the processes of initial quality control, doublet removal, and batch effect correction, 32,076 cells were retained for downstream analysis (Supplemental Figure S1A-C). We identified nine primary clusters based on differential expression of hallmark genes that have been validated in the literature (Figure 1B, C and Supplemental Table S3): enterocytes, stem cells, goblet cells, tuft cells, EECs, Paneth cells, T cells, B cells, and myeloid cells.

Figure 1. The single-cell RNA sequencing (scRNA-seq) analysis on mouse intestinal epithelial and immune cells.

(A) Scheme of study design. (B) Uniform-manifold-approximation-and-projection (UMAP) plot of the nine main cell types identified in the mouse ileums. (C) Feature plots present the average scores of signature genes for cluster annotation of enterocyte, stem cell, goblet cell, tuft cell, enteroendocrine (EEC), Paneth cell, T cell, B cell, and myeloid cell (genes indicated above each plot).

Abbreviations: DEG, Differentially expressed gene; EEC, enteroendocrine cell; UMAP, Uniform Manifold Approximation and Projection.

Gene Abbreviations: Alpi, Alkaline phosphatase intestinal; Apoa1, Apolipoprotein A 1; Cd3d, CD3d molecule; Cd3e, CD3e molecule; Cd68, CD68 molecule; Cd79a, CD79a molecule; Chga, Chromogranin A; Chgb, Chromogranin B; Dclk1, Doublecortin-like kinase 1; Itgax, Integrin subunit alpha X; Lyz1, Lysozyme 1; Mki67, Marker of proliferation Ki-67; Mmp7, Matrix metallopeptidase 7; Ms4a1, Membrane spanning 4-domains A1; Muc2, Mucin 2; Top2a, DNA topoisomerase II alpha; Trpm5, Transient receptor potential cation channel subfamily M member 5; Zg16, Zymogen granule protein 16

Enterocytes were the predominant cell type, constituting approximately 51% of the total cells, followed by stem cells (17%) and goblet cells (13%) (Supplemental Figure S1D). Among immune cell populations, T cells represented the most abundant group, accounting for approximately 10% of total cells (or around 78% of total immune cells), followed by B cells and myeloid cells. GO biological process (BP) analysis revealed predicted functions of the cell clusters that aligned with their known roles in the ileum, further confirming the accuracy of the clustering (Supplemental Figure S1E)

Metabolic specification of ileal epithelial cells

To gain deeper insight into intestinal epithelial differentiation and function in the ileum, we re-clustered all identified IEC populations (Figure 2A). Gene signatures curated from the Mouse Molecular Signatures Database were then used to assess module scores related to nutrient metabolism. Among epithelial subsets, enterocytes exhibited the highest activity associated with both macro- and micro-nutrient metabolism (Figure 2B). Specifically, Bco2 was relatively enriched in enterocytes, whereas β-carotene oxygenase 1 (Bco1) was more prominent in enteroendocrine cells (EECs). Aldehyde dehydrogenase 1 family, member A1 (Aldh1a1) showed higher expression in enterocytes, while retinol dehydrogenase 10 (Rdh10) was detected in other IEC populations. Intestine specific homeobox (Isx) expression was up-regulated more in enterocytes, stem, and goblet cells and other cell types. For vitamin B12 metabolism, cubilin (Cubn) and amnionless (Amn), key mediators of vitamin B12 absorption, were largely expressed in enterocytes, whereas transcobalamin 2 (Tcn2) was more broadly distributed across IEC subsets (Figure 2C).

Figure 2. Analysis of IECs in the ileum reveals a correlation between Bco2 and Il-18 expression.

(A) UMAP plot of all IEC types identified in this study. (B) Module scores of nutrient-metabolism gene sets across enterocyte IECs. (C) Violin plots showing the expression of genes involved in vitamin A/carotenoid and vitamin B12 metabolism across IEC types. (D) Feature plots present the average scores of expression of Bco2 and Il18 in all IECs. (E) Feature plot showing the distribution of IECs co-expressing Bco2 and Il18. (F) Violin plots showing the subclusters co-expressing Bco2 and Il18 in enterocytes, stem cells, goblet cells, tuft cells, EEC, and Paneth cells. (G) Correlation analysis of Bco2 and Il18 expression in enterocytes, stem cells, and goblet cells.

Abbreviations: EEC, Enteroendocrine cell; IEC, Intestinal epithelial cell; UMAP, Uniform Manifold Approximation and Projection

Gene Abbreviations: Aldh1a1, aldehyde dehydrogenase 1 family, member A1; Amn, amnionless; Bco1, β-carotene oxygenase 1; Bco2, β-carotene oxygenase 2; Cubn, cubilin; Il18, interleukin 18; Isx, intestine-specific homeobox; Rara, retinoic acid receptor alpha; Rdh10, retinol dehydrogenase 10; Rxra, retinoid X receptor alpha; Tcn2, transcobalamin 2

Expression of Il18 in IECs is correlated with Bco2 expression

The results revealed that cells expressing Bco2 appeared to co-localize with those cells expressing Il18 on the UMAP visualization (Figure 2D). Therefore, we labeled cells co-expressing both Bco2 and Il18 as “double-positive” and highlighted them in red in Figure 2E. These double-positive cells were predominantly observed among enterocytes, goblet cells, and stem cells. To refine the analysis, we subsequently re-clustered IEC populations at a higher resolution (a resolution of 1.4), resulting in 31 IEC subclusters. We calculated the proportion of double-positive cells within each subcluster, and designated clusters with more than 10% double-positive cells as “double-positive enriched” (Figure 2F). Given that human Bco2 and IL18 are located on chromosome 11 at the 11q22.2-22.3 region of about 100 kb in humans, with Bco2 located upstream of IL18 [31], and mouse Bco2 and Il18 on chromosome 9 at 27.74 cM and 27.75 cM, respectively (https://www.informatics.jax.org/marker/), we hypothesized a potential co-expression relationship in mouse IECs. To test this, we performed Spearman’s rank correlation. The results revealed moderate positive correlations in enterocytes and stem cells, and a weak positive correlation in goblet cells, respectively (Figure 2G-L).

Enterocyte subclusters exhibit functional specialization

After characterizing nutrient metabolism across ileal epithelial populations, we next focused on enterocytes, the principal absorptive cell type of the small intestine. Re-clustering identified ten distinct enterocyte subclusters (Figure 3A, B). Subclusters Ent_C2, C3, and C5 were classified as immature enterocytes based on their high expression of the early stages of enterocyte differentiation genes, including solute carrier family 16 member 1 (Slc16a1), eukaryotic translation elongation factor 1 alpha 1 (Eef1a1), protease, serine 32 (Prss32), and keratin 8 (Krt8). In contrast, Ent_C1, C4, C7, C8, and C9 showed elevated expression of apolipoprotein A1 (Apoa1), cadherin related family member 2 (Cdhr2), fatty acid binding protein 2 (Fabp2), and fatty acid binding protein 6 (Fabp6), and were thus classified as mature enterocytes. Ent_C0 and C6 expressed a combination of both immature and mature markers, suggesting they represent an intermediate enterocyte population (Figure 3A, B, D). Immature enterocytes expressed both proliferation markers marker of proliferation ki-67 (Mki67) and fibroblast growth factor binding protein 1(Fgfbp1), whereas intermediate enterocytes expressed only Fgfbp1 (Figure 4C), suggesting that Fgfbp1 may serve as a marker for the transitional state between active proliferation and full maturation. Consistent with these observations, Monocle pseudotime analysis demonstrated a clear transition trajectory from immature to mature enterocytes (Figure 4E).

Figure 3. Enterocyte subclusters share core functions but exhibit distinct functional preferences.

(A) UMAP plot of ten enterocyte clusters. (B) Dot plot showing the expression of top marker genes for each enterocyte subcluster. (C) Violin plot showing the expression of Mki67 and Fgfbp1 across enterocyte clusters. (D) UMAP plot of immature, intermediate, and mature enterocytes. (E) Monocle analysis projected onto UMAP, with colors representing progression from early (blue) to late (yellow) differentiation states. (F-K) Feature plot showing the distribution of enterocytes expressing Rbp2 (F), Bco2 (G), Apoa4 (H), Apob (I), Cubn (J), and Tcn2 (K), the average scores of relative gene expression are shown. (L) Bar plot showing the GO BP enrichment results based on upregulated DEGs across enterocyte subclusters.

Abbreviations: AvgExpr, Average Expression; GO BP, Gene Ontology Biological Process; Ent, Enterocyte; PctExpr, Percentage of Expression; UMAP, Uniform Manifold Approximation and Projection.

Gene Abbreviations: Apoa1, apolipoprotein A1; Apoa4, apolipoprotein A4; Apob, apolipoprotein B; Bco2, β-carotene oxygenase 2; Cdhr2, cadherin-related family member 2; Cubn, cubilin; Eef1a1, eukaryotic translation elongation factor 1 alpha 1; Fabp2, fatty acid binding protein 2; Fabp6, fatty acid binding protein 6; Fgfbp1, fibroblast growth factor binding protein 1; Krt8, keratin 8; Mki67, marker of proliferation Ki-67; Prss32, serine protease 32; Rbp2, retinol binding protein 2; Slc16a1, solute carrier family 16 member 1; Tcn2, transcobalamin 2

Figure 4. Canonical and non-canonical goblet cells follow distinct differentiation trajectories and exhibit diverse functions.

(A) UMAP plot showing six distinct goblet cell clusters. (B) Violin plots showing marker gene expression of goblet cell clusters. (C) Pseudotime trajectory of goblet cells, colored by pseudotime progression from dark blue (least differentiated) to light blue (most differentiated). (D) Pseudotime trajectory of goblet cells with each cluster distributed on the trajectory branches. (E) Violin plots of Bco1, Bco2, and Il18 expression across goblet-cell subclusters in WT and BCO2-KO ileum. (F) GO enrichment analysis of DEGs between canonical and non-canonical goblet cells.

Abbreviation: DEG, Differentially expressed gene; UMAP, Uniform Manifold Approximation and Projection.

Gene Abbreviations: Alpi, Alkaline phosphatase, intestinal; Apob, Apolipoprotein B; Bco1, Beta-carotene oxygenase 1; Bco2, Beta-carotene oxygenase 2; Il18, Interleukin 18; Mcm6, Minichromosome maintenance complex component 6; Mki67, Marker of proliferation Ki-67; Muc2, Mucin 2; Olfm4, Olfactomedin 4; Slc6a4, Solute carrier family 6 member 4 (serotonin transporter); Tff3, Trefoil factor 3; Zg16, Zymogen granule protein 16

We next examined cellular heterogeneity among enterocyte subclusters by analyzing the expression of genes related to vitamin A, carotenoid, and vitamin B12 metabolism (Figure 3F-K). Subclusters Ent_C1, C4, C7, and C8 were enriched for vitamin A absorption pathways, as indicated by the expression of retinol binding protein 2 (Rbp2) and apolipoprotein A4 (Apoa4). In contrast, Ent_C0, C1, C2, and C7 showed high expression of Bco2, whereas Bco1 expression was low across subclusters. For vitamin B12 absorption, Cubn was predominantly expressed in Ent_C1, C4, C7, C8, and C9, whereas Tcn2 was more broadly distributed across enterocyte subclusters. Gene Ontology Biological Process (GO BP) enrichment analysis (Figure 3L) further highlighted that different enterocyte subclusters are specialized for distinct biological processes, supporting the concept of functional heterogeneity. Together, these findings reveal a functional diversity among enterocyte subclusters, with specific subsets specialized for distinct nutrient metabolism pathways.

Goblet cell heterogeneity reveals canonical and non-canonical lineages

Because goblet cells are critical for mucus secretion, barrier integrity, and epithelial–immune regulation, we next investigated their transcriptional heterogeneity in the ileum. A total of 4,331 goblet cells were identified and resolved into six subclusters (Goblet_C0–C5) by UMAP (Figure 4A). Marker genes and proliferation status assigned subclusters to three groups: proliferating (Mki67, olfactomedin 4 (Olfm4), minichromosome maintenance complex component 6 (Mcm6)); canonical (mucin 2 (Muc2), zymogen granule protein 16 (Zg16), trefoil factor 3 (Tff3)); and non-canonical [alkaline phosphatase, intestinal (Alpi), apolipoprotein B (Apob), solute carrier family 6 member 4 (Slc6a4) ) (Figure 4B) [32]. Monocle pseudotime revealed two branches from proliferating goblet cells toward canonical (Goblet_C0, C1, C4) or non-canonical (Goblet_C2, C3) fates (Figure 4C, D). Non-canonical cells showed elevated Il18 and Bco2, but not Bco1 (Figure 4E). Gene expression dynamics along pseudotime revealed distinct waves of transcription during goblet cell differentiation (Supplemental Figure S2). Canonical goblet cells upregulated mucus-related genes (Muc2, Zg16, Tff3), whereas non-canonical goblet cells induced metabolic and immune-associated genes (Bco2, Apob, Il18), supporting two divergent differentiation programs. GO-BP analysis revealed functional specialization: canonical cells were enriched for protein O-linked glycosylation, secretion, and unfolded protein response, whereas non-canonical cells were enriched for fatty-acid metabolism, ATP metabolism, lipid transport, and oxidative stress (Figure 4F).

To validate Bco2-Il18 co-expression, we analyzed scRNA-seq data from age- and sex-matched Bco2 knockout mice. Il18 expression was reduced in canonical and non-canonical goblet cells but remained unchanged in proliferating cells in KO mice (Figure 4E).

Immune cell heterogeneity and epithelial–immune interactions in the mouse ileum

To characterize the immune landscape of the mouse ileum, a total of 4,277 immune cells were identified and re-clustered into 14 distinct populations based on the marker genes (Figure 5A, B, Supplemental Table S3). T cells were defined by the expression of Cd3d and Cd3e and further subdivided into Cd4 T cells (expressing Cd4), Cd8 T cells (expressing Cd8a), and natural killer T cells (expressing killer cell lectin-like receptor subfamily k member 1 (Klrk1) and killer cell lectin-like receptor subfamily b member 1c (Klrb1c)). Within the Cd4 T cell population, two subtypes were distinguished, naïve Cd4 T cells (expressing C-C chemokine receptor type 7 (Ccr7) and selectin l (Sell)) and regulatory T cells (Tregs, expressing forkhead box p3 (Foxp3) and interleukin 2 receptor alpha (Il2ra)).

Figure 5. Characterization of immune cell clustering and cell-cell communication network in the ileum.

(A) UMAP plot of immune cell clusters. (B) Dot plot showing the expression of marker genes for each immune cell cluster. (C) Connectome web illustrating ligand–receptor interaction strength among all cell populations. Node size represents the cell number of cells in each cell type, while the thickness and color of the connecting lines indicate the interactions strength between two nodes. (D) Visualization of functional clustering of all detected signaling pathways. Each dot represents the communication network of a signaling pathway. Dot size represents communication probability. Colors represent distinct functional groups of signaling pathways.

Abbreviations: AvgExpr, Average expression; Cd4, CD4⁺ T cell; cDC, Conventional Dendritic Cell; Commu.Prob., communication probability; Dim, Dimension; EEC, enteroendocrine cell; ILC, Innate Lymphoid Cell; Mac, Macrophage; nIEL, Natural Intraepithelial Lymphocyte; NKT, Natural Killer T cell; PctExpr, Percentage of expression; pDC, Plasmacytoid Dendritic Cell; Treg, Regulatory T cell

Gene Abbreviations for Figure 5B: Bst2, Bone marrow stromal antigen 2; C1qa, Complement C1q A chain; C1qc, Complement C1q C chain; Cd3d, CD3d molecule; Cd4, CD4 molecule; Cd8a, CD8a molecule; Cd8b1, CD8b molecule 1; Cd79a, CD79a molecule; Ccr7, C-C motif chemokine receptor 7; Foxp3, Forkhead box P3; Gata3, GATA binding protein 3; Gzmb, Granzyme B; Iglc1, Immunoglobulin lambda constant 1; Il2ra, Interleukin 2 receptor subunit alpha; Itgae, Integrin subunit alpha E; Itgax, Integrin subunit alpha X; Klrb1c, Killer cell lectin-like receptor subfamily B member 1C; Klrk1, Killer cell lectin-like receptor subfamily K member 1 (also known as NKG2D); Mcpt1, Mast cell protease 1; Mcpt2, Mast cell protease 2; Mki67, Marker of proliferation Ki-67; Ms4a1, Membrane spanning 4-domains A1 (also known as CD20); Mzb1, Marginal zone B and B1 cell-specific protein; Prf1, Perforin 1; Rorc, RAR-related orphan receptor C; S100a8, S100 calcium binding protein A8; S100a9, S100 calcium binding protein A9; Sell, Selectin L (also known as CD62L); Siglech, Sialic acid binding Ig-like lectin H; Wdfy4, WD repeat and FYVE domain containing 4.

Gene Abbreviations for Figure 5D: Group 1: CADM, Cell Adhesion Molecule; CD39, Ectonucleoside triphosphate diphosphohydrolase 1 (ENTPD1); CD40, CD40 molecule (TNF receptor superfamily member 5); CD45, Protein tyrosine phosphatase receptor type C (PTPRC); CD52, CD52 molecule; CD6, CD6 molecule; CD80, CD80 molecule (B7-1); CD86, CD86 molecule (B7-2); CNTN, Contactin; CXCL, C-X-C motif chemokine ligand; FLT3, Fms-related tyrosine kinase 3; FN1, Fibronectin 1; GP1BA, Glycoprotein Ib platelet subunit alpha; ICOS, Inducible T-cell costimulator; IFN-II, Interferon type II (typically refers to IFN-γ); IL16, Interleukin 16; IL2, Interleukin 2; IL4, Interleukin 4; IL6, Interleukin 6; LCK, LCK proto-oncogene, Src family tyrosine kinase; LAIR1, Leukocyte-associated immunoglobulin-like receptor 1; LIGHT, TNFSF14 (Tumor necrosis factor ligand superfamily member 14); MHC-II, Major Histocompatibility Complex Class II; PLAUR (PLAU), Plasminogen activator, urokinase receptor (PLAU is the ligand, PLAUR is the receptor); RANKL, Receptor activator of nuclear factor κB ligand (TNFSF11); SELPGL (SELP-LG), Selectin P ligand (SELPLG); TWEAK, TNFSF12 (TNF-related weak inducer of apoptosis); TNF, Tumor Necrosis Factor.

Group 2: 2-AG, 2-Arachidonoylglycerol; ADGRE, Adhesion G protein-coupled receptor E; ADGRL, Adhesion G protein-coupled receptor L; APP, Amyloid beta precursor protein; CD48, CD48 molecule; CDH1, Cadherin 1 (E-cadherin); CCL, C-C motif chemokine ligand; CEACAM, Carcinoembryonic antigen-related cell adhesion molecule; CLEC, C-type lectin domain family; GALECTIN, Galectin family (galactoside-binding lectins); GDF, Growth Differentiation Factor; ICAM, Intercellular Adhesion Molecule; JAM, Junctional Adhesion Molecule; KIT, KIT proto-oncogene receptor tyrosine kinase; KLR, Killer cell lectin-like receptor; L1CAM, L1 Cell Adhesion Molecule; LAMININ, Laminin (extracellular matrix glycoprotein); MHC-I, Major Histocompatibility Complex Class I; NECTIN, Nectin cell adhesion molecule; NOTCH, Notch receptor; PECAM1, Platelet and Endothelial Cell Adhesion Molecule 1; PECAM2, Platelet and Endothelial Cell Adhesion Molecule 2; Prostaglandin, Prostaglandins (lipid-derived signaling molecules); SEMA4, Semaphorin 4; TGFb, Transforming Growth Factor Beta; THBS, Thrombospondin; THY1, Thy-1 Cell Surface Antigen.

Group 3: 12oxoLTB4, 12-oxo-leukotriene B4; 5-HT, 5-Hydroxytryptamine (Serotonin); Adenosine, Adenosine (purine nucleoside signaling molecule); ANGPT, Angiopoietin; ApoB, Apolipoprotein B; BTLA, B and T Lymphocyte Associated; CD160, CD160 molecule; CD200, CD200 molecule; CD23, Fc epsilon receptor II (FCER2); CD276, CD276 antigen (B7-H3); CD96, CD96 molecule; CHAD, Chondroadherin; CysLTs, Cysteinyl leukotrienes; DHEA, Dehydroepiandrosterone; DHEAS, Dehydroepiandrosterone sulfate; Dopamine, Dopamine (catecholamine neurotransmitter); FASLG, Fas ligand (TNF superfamily member 6); Glutamate, Glutamic acid (neurotransmitter); IL1, Interleukin 1; LIFR, Leukemia inhibitory factor receptor; MIF, Macrophage Migration Inhibitory Factor; NCAM, Neural Cell Adhesion Molecule; NKG2D, Natural Killer Group 2 Member D; NPY, Neuropeptide Y; PARs, Protease-Activated Receptors; PDGF, Platelet-Derived Growth Factor; Pregnenolone, Pregnenolone (precursor steroid hormone); PROCR, Protein C Receptor (endothelial); PVR, Poliovirus Receptor (CD155); SELL, L-selectin (CD62L); VEGI, Vascular Endothelial Growth Inhibitor; VISFATIN, Visceral fat–derived adipokine, aka Nicotinamide phosphoribosyltransferase (NAMPT).

Group 4: 27HC, 27-Hydroxycholesterol; ADGRG, Adhesion G protein-coupled receptor G; AGRN, Agrin; ANGPTL, Angiopoietin-like protein; Androsterone, Androsterone (a steroid hormone); BMP, Bone Morphogenetic Protein; CDH, Cadherin; Cholesterol, Cholesterol (lipid molecule); CLDN, Claudin; CypA, Cyclophilin A (also known as PPIA); Desmostero, Desmosterol (cholesterol precursor); DESMOSOME, Desmosome (cell-cell junction structure); Dopamine, Dopamine (catecholamine neurotransmitter); EGF, Epidermal Growth Factor; EPHA, Ephrin type-A receptor; EPHB, Ephrin type-B receptor; FLRT, Fibronectin Leucine Rich Transmembrane Protein; GAS, Growth Arrest-Specific protein; GRN, Progranulin (Granulin precursor); GUCA, Guanylate Cyclase Activator; HH, Hedgehog signaling molecule; HSPG, Heparan Sulfate Proteoglycan; MPZ, Myelin Protein Zero; Netrin, Netrin (axon guidance molecule); OCLN, Occludin; PROS, Protein S (encoded by PROS1); PTPR, Protein Tyrosine Phosphatase Receptor; SEMA3, Semaphorin 3; SEMA5, Semaphorin 5; SEMA6, Semaphorin 6; SEMA7, Semaphorin 7; Serotonin, Serotonin (5-HT, neurotransmitter); TRAIL, TNF-Related Apoptosis-Inducing Ligand; UNC5, UNC-5 netrin receptor; WNT, Wingless/Integrated signaling molecule.

All Cd8 T cells in this dataset expressed Cd8a but not Cd8b1, suggesting they are natural intraepithelial lymphocytes (nIELs) rather than lamina propria T cells. These nIELs were the most abundant (41.45%) immune cells and were further categorized into normal nIELs, including cytotoxic nIELs (expressing granzyme b (Gzmb) and perforin 1 (Prf1)), and proliferating nIELs (expressing Mki67) (Figure 5A, B). B cells were identified by the expression of Cd79a, with a subpopulation of plasma B cells marked by high expression of immunoglobulin lambda constant 1 (Iglc1) and marginal zone b and b1 cell-specific protein (Mzb1).

Additionally, five distinct myeloid populations were identified, including plasmacytoid dendritic cells (pDCs), expressing sialic acid binding Ig-like lectin h (Siglech) and bone marrow stromal antigen 2 (Bst2)), conventional dendritic cells (cDCs) expressing wdfy family member 4 (Wdfy4) and Itgax), macrophages expressing complement component 1q subcomponent subunit a (C1qa) and complement component 1q subcomponent subunit c (C1qc), mast cells expressing mast cell protease 1 (Mcpt1) and mast cell protease 2 (Mcpt2), and neutrophils expressing s100 calcium-binding protein a8 (S100a8) and s100 calcium-binding protein a9 (S100a9)), respectively (Figure 5A, B). Each immune population displayed transcriptional signatures consistent with its established roles in intestinal immunity and barrier regulation.

Finally, we applied CellChat to our scRNA-seq dataset to map the cell-cell communication network among major ileal cell types. This analysis revealed a complex network of potential interactions involved in immune-to-epithelial, epithelial-to-immune, immune-to-immune, and epithelial-to-epithelial signaling (Figure 5C, D). Significantly, pDCs emerged as a central communication hub (Figure 5C). The strongest interactions were observed within pDCs, between pDCs and cDCs, and between pDCs and enterocytes (Figure 5C). pDCs were highly responsive to signals from nearly all major epithelial and immune populations, receiving diverse ligands associated with immune modulation (CD28, CD40, MHC-I/II), inflammation (transforming growth factor beta (Tgfβ), interleukin 4 (Il4), interleukin 6 (Il6), interferon gamma (Ifnγ), tumor necrosis factor (Tnf)), and developmental pathways (Notch) (Supplemental Figure S3). Additionally, pDCs actively communicated with immune cells (particularly cDCs, T cells, and macrophages) via molecules such as CD45, MHC-I/II, Il6, and immune checkpoint ligands (Cd200 and B and T lymphocyte associated (Btla)). They also engaged in epithelial compartments through signals involving Tgfβ, platelet and endothelial cell adhesion molecule 1 (Pecam1), Sell, and lipid metabolism pathways (cholesterol, desmosterol, ectonucleoside triphosphate diphosphohydrolase 1 (Entpd1), amyloid beta precursor protein (App), and cyclophilin A (CypA)) (Supplemental Figure S4). These findings emphasized the crucial role of pDCs in coordinating cellular communication and thus maintaining intestinal homeostasis.

To further dissect how intestinal cells coordinate their functions, we grouped all detected signaling pathways into four categories based on functional similarity (Figure 5D). Group 1 included pathways associated with immune regulation and inflammatory signaling, such as Cd45, MHC-II, Tnf, Il2, Il4, Il6, Il16, and interferon type II (Ifn-II, or Type II interferon). Group 2 consisted of immune-modulatory and intestinal development pathways, including NOTCH, App, growth differentiation factor (Gdf), and tyrosine-protein kinase KIT (KIT). Group 3 comprised additional immune regulation signals (e.g., Cd276, Cd200, Btla, natural killer group 2, member d (Nkg2d), and Sell) along with lipid transport pathways like apolipoprotein A (ApoA), ApoB, and Visceral fat–derived adipokine (Visfatin). Group 4 encompassed signals involved in intestinal development (e.g., Wnt, epidermal growth factor (Egf), bone morphogenetic protein (Bmp)), epithelial barrier integrity (e.g., DESMOSOME, cadherin (Cdh), claudin (Cldn), occluding (Ocln), myelin protein zero (Mpz), agrin (Agrn), granulin (Grn)), and lipid signaling (e.g., cholesterol, desmosterol, 27-hydroxycholesterol (27HC), androsterone, and CypA) (Figure 5D). Collectively, these results revealed a tightly regulated signaling network essential for maintaining the balance between immune defense, inflammatory control, and epithelial homeostasis in the ileal microenvironment.

Discussion

In this study, we generated a comprehensive single-cell transcriptomic atlas of the mouse ileum, that revealed marked cellular heterogeneity and functional specialization. Enterocyte subclusters demonstrated distinct roles in nutrient metabolism, with specific subclusters enriched for metabolism of vitamin A, carotenoids, and vitamin B12, respectively. Goblet cells differentiated into canonical and non-canonical lineages, with the latter characterized by high co-expression of Bco2 and Il18, linking carotenoid metabolism to immune signaling. Within the immune compartment, pDCs emerged as central mediators of epithelial–immune communication. Together, these findings stress the heterogeneity across ileal cell types and uncover novel pathways integrating nutrient metabolism with mucosal immunity.

Our dataset aligned with the established knowledge of intestinal cellular composition, demonstrating an IEL to IEC ratio of approximately 9%, that closely matched the reported ~10% in the literature [33,34]. We identified all major IEC types. However, the immune cell population accounted for about 14% of total cells, within which we identified three IEL subtypes and 11 lamina propria immune cell subtypes. These specific characteristics of immune cell populations may be influenced by overall immune cell numbers, tissue region (e.g., ileum specifically), mouse strain, age, and sex, as well as by the cell isolation procedure employed in this study.

Cell–cell communication underlies the coordinated functions necessary for biological processes, particularly in complex tissues such as the intestine [35]. scRNA-seq enables the analysis of intercellular communication at high resolution, revealing how diverse cell populations interact to maintain the delicate balance between immune tolerance and activation in the ileum [30, 36]. Dendritic cells (DCs) act as antigen-presenting cells that bridge IECs and the immune system. They are primary categorized into pDCs and cDCs, with cDCs typically regarded as the predominant subtype involved in antigen presentation and T cell priming [37-39]. In our study, ligand–receptor interaction analysis revealed that pDCs, rather than cDCs, occupied a central position in the signaling network among IECs and immune cells in the mouse ileum. While pDCs are recognized primarily for their role in controlling viral infections [40, 41], and recently, Wang et al. reported that pDCs can disrupt intestinal immune environments in lupus, acting as dominant signaling sources contributing to immune dysregulation [42]. Our results may suggest a broader function of pDCs as a critical mediator of immune-epithelial crosstalk, essential for intestinal homeostasis in the ileum.

Vitamin A and carotenoid metabolism are central to intestinal epithelial dynamics and immune regulation [43, 44]. Bco2, a mitochondrial carotenoid-cleaving enzyme, is positioned adjacent to Il18 in both the human [31] and mouse genome, and SNPs within this locus have been associated with altered serum Il18 levels and risk of metabolic disease [31, 45, 46]. Our prior work showed that Bco2 knockout drives mitochondrial dysfunction, ROS accumulation, and diet-related disorders such as obesity and liver disease [47-49]. Here, we extended this link by demonstrating Bco2–Il18 co-expression in enterocytes and goblet cells, suggesting a coordinated role for carotenoid metabolism and epithelial immune signaling in maintaining nutritional and metabolic homeostasis.

The distinguished significance between canonical and non-canonical goblet cell lineages is an emerging concept in intestinal biology. Within the intestinal epithelium, Il18 production is predominantly attributed to IECs, particularly enterocytes, goblet cells, and stem cells, the major epithelial populations represented in our dataset. In goblet cells, Bco2 and Il18 expression was highly increased within the non-canonical goblet lineage, that suggested the transition of goblet cell differentiation from proliferating to non-canonical goblet cells. IL-18 was previously shown to inhibit goblet cell maturation and mucus production in colitis [46, 50]. Our findings in healthy mouse ileal tissue could support this statement, as non-canonical goblet cells exhibited reduced expression of mucus-related genes, including Muc2, Zg16, and Tff3, compared to canonical goblet cells. Interestingly, the downregulation of Il18, but not Bco1, in both canonical and non-canonical goblet cells in Bco2 KO mice suggested that Bco2 may regulate the development of non-canonical goblet cells through modulation of Il18 expression, thereby influencing host–microbiome interactions via goblet cell–mediated mechanisms.

Our single-cell map demonstrates that vitamin A metabolism is partitioned across distinct enterocyte subclusters. Rbp2 and Apoa4 were enriched in subclusters likely supporting retinol uptake/trafficking and chylomicron assembly, respectively [3]. We also observed high Aldh1a1 expression in enterocytes and Rdh10 distribution in other epithelial cells, suggesting local interconversion between retinol and retinal that can feed retinoic acid (RA) production [3, 51]. Together with RA-responsive control of Isx [52, 53], these patterns indicated that vitamin A metabolism is wired into subcluster-specific programs rather than being uniformly expressed. Functionally, such partitioning may buffer dietary fluctuations (retinyl ester vs carotenoid intake) by distributing uptake, binding, esterification, and lipoprotein assembly across enterocyte subsets [3].

Carotenoid metabolism also segregates across subclusters. Enterocytes enriched for Bco2 point to a role for mitochondrial carotenoid catabolism in mitigating oxidative stress while generating apocarotenoid metabolites [54, 55]. By contrast, Bco1 expression was observed mainly outside these subclusters (e.g., in EECs), suggesting complementary pathways for carotenoid processing across lineages [56, 57]. Coupled with Il18 expression in specific epithelial compartments, these data support a model in which carotenoid metabolism through Bco2 intersects with innate signaling and redox pathways in a subcluster-dependent manner, thereby linking dietary carotenoids to epithelial metabolism and immune regulation through regulation of goblet cell function.

Cubn and Amn, which encode the intrinsic factor–vitamin B12 receptor complex, were concentrated in specific enterocyte subclusters, whereas Tcn2 was more broadly expressed across IECs. This distribution suggests that vitamin B12 absorption is spatially restricted to specialized enterocyte subsets, while export into the circulation is more widely distributed [58-60]. Such functional specialization may help explain selective vitamin B12 malabsorption when ileal architecture is disrupted, as occurs in Crohn’s disease or after ileal resection [9, 61, 62], and could also account for variability in vitamin B12 status under different dietary and bile acid conditions [63].

Several limitations should be acknowledged in this study. The analysis was conducted in a relatively small number of animals (n = 4, from 4 separate cages), all male, and lacked protein-level validation or integration with microbiome data. Small sample sizes are a common challenge in single-cell studies due to cost and technical demands [64,65]. In addition, we used fixed-cell scRNA-seq, which offers advantages by reducing stress-induced transcriptional artifacts, minimizing batch effects, and preserving fragile cell populations, but may also compromise RNA integrity, reduce capture efficiency, and preclude downstream functional assays [66]. Despite these limitations, the depth of our dataset, which includes large numbers of ileal epithelial and immune cells and captures transcriptional heterogeneity at high resolution [15,16], provides a valuable resource for the field. While additional validation and extension to female mice and disease models will be important [17,18], our single-cell atlas enables peers to explore diverse questions related to nutrient metabolism, immune signaling, and intestinal health. In particular, the mapping of vitamin A, carotenoid, and vitamin B12 metabolism at single-cell resolution offers a foundation for advancing nutrition-focused research and the development of targeted dietary interventions.

In conclusion, this study provides an integrated single-cell atlas of the mouse ileum, linking epithelial differentiation, nutrient metabolism, and immune interactions. By resolving how enterocyte and goblet cell subclusters specialize in vitamin A, carotenoid, and vitamin B12 pathways, and how these programs intersect with immune regulation, we highlight the nutritional dimension of ileal cellular heterogeneity. Beyond advancing basic understanding of intestinal biology, our findings offer a foundation for future nutrition-focused research aimed at preventing or treating disorders of malabsorption, inflammation, and metabolic disease through targeted, cell-based nutritional strategies.

Abbreviations

27HC

27-hydroxycholesterol

Ada

adenosine deaminase

Agr2

anterior gradient 2

AGRN

agrin

Aldh1a7

aldehyde dehydrogenase family 1, subfamily A7

Alpi

alkaline phosphatase, intestinal

AMP

antimicrobial peptide

ApoA

apolipoprotein A

ApoB

apolipoprotein B

Apoa1

apolipoprotein A1

App

amyloid beta precursor protein

Bco2

beta-carotene oxygenase 2

BMP

bone morphogenetic protein

Bst2

bone marrow stromal antigen 2

BTLA

B and T lymphocyte associated

C1qa

complement component 1q subcomponent subunit a

C1qc

complement component 1q subcomponent subunit c

Ccr7

C-C chemokine receptor type 7

Ccl2

C-C chemokine receptor type 2

Cd3

cluster of differentiation 3

Cd3d

Cd3 delta chain

Cd3e

Cd3 epsilon chain

Cd79A

Cd79 alpha

CDH

cadherin

Cdhr2

cadherin related family member 2

Ccnd1

cyclin d1

CypA

cyclophilin A

cDC

conventional dendritic cell

Chga

chromogranin a

Chgb

chromogranin b

Clca1

chloride channel accessory 1

Clca4a

chloride channel accessory 4a

CLDN

claudin

Dclk1

doublecortin like kinase 1

DC

dendritic cell

EEC

enteroendocrine cell

Eef1a1

eukaryotic translation elongation factor 1 alpha 1

EGF

epidermal growth factor

Enpp7

ectonucleotide pyrophosphatase/phosphodiesterase 7

Entpd1

ectonucleoside triphosphate diphosphohydrolase 1

Fabp2

fatty acid binding protein 2

Fabp6

fatty acid binding protein 6

FBS

fetal bovine serum

Fgfbp1

fibroblast growth factor binding protein 1

Foxp3

forkhead box p3

GDF

growth differentiation factor

GO

Gene Ontology

GRN

granulin

Gzmb

granzyme b

IEC

intestinal epithelial cell

IEL

intraepithelial lymphocyte

IFN-II

interferon type II

IFNγ

interferon gamma

IGF1

insulin-like growth factor 1

Iglc1

immunoglobulin lambda constant 1

IGSF6

immunoglobulin superfamily member 6

Il18

interleukin 18

IL16

interleukin 16

IL2

interleukin 2

IL2ra

interleukin 2 receptor alpha

IL4

interleukin 4

IL6

interleukin 6

Itgax

integrin subunit alpha x

KEGG

Kyoto Encyclopedia of Genes and Genomes

Klrk1

killer cell lectin-like receptor subfamily k member 1

Klrb1c

killer cell lectin-like receptor subfamily b member 1c

KIT

tyrosine-protein kinase KIT

Krt8

keratin 8

Lrp1

low density lipoprotein receptor-related protein 1

Ly6m

lymphocyte antigen 6 complex locus m

Lyz1

lysozyme 1

Mcm

minichromosome maintenance complex

Mcm6

minichromosome maintenance complex component 6

Mcpt1

mast cell protease 1

Mcpt2

mast cell protease 2

MHC-I

major histocompatibility complex class I

MHC-II

major histocompatibility complex class II

Mki67

marker of proliferation Ki-67

Mmp7

matrix metallopeptidase 7

MPZ

myelin protein zero

Ms4a1

membrane spanning 4-domains a1

Muc2

mucin 2

Myb

myeloblastosis oncogene

Mzb1

marginal zone b and b1 cell-specific protein

NKG2D

natural killer group 2, member d

nIEL

natural intraepithelial lymphocyte

Nlrp6

nod-like receptor family pyrin domain containing 6

Notch

notch receptor

OCLN

occludin

Olfm4

olfactomedin 4

pDC

plasmacytoid dendritic cell

Pecam1

platelet and endothelial cell adhesion molecule 1

Prf1

perforin 1

PRR

pattern recognition receptor

Prss32

protease, serine 32

Reg3a

regenerating islet-derived 3 alpha

S100a8

S100 calcium-binding protein a8

S100a9

S100 calcium-binding protein a9

Sell

selectin l

Siglech

sialic acid binding ig-like lectin h

Slc16a1

solute carrier family 16 member 1

Slc25a48

solute carrier family 25 member 48

Slc2a5

solute carrier family 2 member 5

Slc6a4

solute carrier family 6 member 4

Slc7a15

solute carrier family 7 member 15

Tff3

trefoil factor 3

TGFβ

transforming growth factor beta

TNF

tumor necrosis factor

UMAP

uniform manifold approximation and projection

Wdfy4

WDFY family member 4

Zg16

zymogen granule protein 16

Data Availability

Data described in the manuscript, code book, and analytic code will be made available upon request.