PMC

a. Regional surveys. tSNE of 11,665 cells from the duodenum, jejunum and ileum, colored by region (left) or post-hoc annotation (right). n=2 mice. b. Regional enterocyte signatures. Relative expression of genes (rows) across cells (columns), sorted by region. c. Regional differences in ISC differentiation. Diffusion-map embedding of 8,988 cells colored by region (left), cluster (center left), or expression of novel regional markers of ISCs (Gkn3, Bex1) or enterocytes (Fabp1, Fabp6). E: Enterocyte, EP: Enterocyte progenitor.

a. Overview. b. Cell type clusters. tSNE of 7,216 single cells (points), colored by cluster assignment (n=6 mice). E: Enterocyte. c. Cell type signatures. Relative expression level (row-wise Z-score of log₂(TPM+1), color bar) of genes (rows) across cells (columns), sorted by types (color code). d,e. Mptx2 is a novel Paneth cell marker. (d) Combined smFISH of Mptx2 (green) and immunofluorescence assay (IFA) of the Paneth cell marker Lyz1 (red). Dashed line: Crypt, arrow: Paneth cell. Scale bar: 20μm. (e) In situ hybridization (ISH) of Mptx2 (red). Scale bar: 50μm.

a,b. Changes in cell composition. (a) IEC subsets (colored by clusters) in control (n=4), Salmonella-infected (n=2), and helminth-infected mice (3 and 10 days; n=2 each). b. Frequencies (y axis) of each cell type in each mouse (dots) under each condition (* FDR < 10⁻⁵; ** FDR < 10⁻¹⁰, Wald test). Error bars: SEM. c. Anti-microbial lectin induction in Salmonella infection. Distribution of expression (y axis) in enterocytes (blue) and all other cells (grey). d. Shifts in tuft cell proportions in helminth infection. Frequencies (y axis) of each tuft subset in each mouse (dots, n=2 mice). Error bars: SEM. (* FDR < 0.25; ** FDR < 0.05, Wald test).

a. unsupervised clustering. tSNE of 533 EECs colored by sub-cluster. (n=8 mice) b. EEC subtype signatures. Relative expression of subtype-enriched genes (FDR < 0.01, rows) across cells (columns). c. Hormone-based EEC classification. Distribution of expression (x axis) of EEC TFs and hormones (columns) in cells from each subset (rows). Grey bars: traditional nomenclature by hormone expression. d. smFISH of Cck (green), Ghrl (red) and Gcg (white). Scale bar, 50μm. Inset (x5): triple-positive SILA cell e. Regional distribution of EEC subsets. Proportion (y axis) of each subset in the three regions (n=2 mice). P.A: Prog. -A, P.L: Prog. –late, P.E: Prog. –early, P.M: Prog. –middle. Error bars: SEM. * FDR<0.25, ** FDR<0.1, *** FDR<0.01, χ² test (Methods) f. Enterochromaffin heterogeneity. smFISH of Reg4 (green) and Tph1 (red) co-stained with IFA of ChgA (white). Yellow and white arrows: Tph1⁺/Reg4⁺ and Tph1⁺/Reg4⁻ EECs respectively. Scale bar: 20μm.

a,b. RANKL-induced organoid M cells. (a) 384 differentiated M cells (blue) in a tSNE of 5,434 epithelial cells from organoids (n=4 pooled wells per treatment). (b) Proportions (y axis) of IEC types (x axis) in control or RANKL-treated organoids. c–e. FAE M cells in vivo. (c) Pearson correlation coefficient (color bar) for each pair of 4,700 FAE cells (n=5 mice, large clusters down-sampled to 50 cells for visualization). Arrow: 18 M cells. (d). Mean expression (color bar) in each FAE cluster (rows) of genes (columns) for known (grey) or novel (black) markers (left) or TFs (right), enriched (FDR < 0.05, Mann-Whitney U-test) in M cells in vivo. E: enterocyte, EEC: enteroendocrine, EC: enterochromaffin.

a. Tuft cell subsets. tSNE of 166 tuft cells colored by sub-cluster (n=6 mice). b. Tuft-1 and Tuft-2 gene signatures. Relative expression (in droplet-based data) of the top 25 genes (rows) for Tuft-1 and Tuft-2 cells (columns) (FDR < 0.01 in both datasets). c. Tuft-2 cells express TSLP. Distribution of expression of Il25 and TSLP in Enterocytes (E), Tuft-1 and Tuft-2 subsets (* FDR<0.1, *** FDR <0.0001, Mann-Whitney U-test). d–g. Tuft-2 cells express Ptprc (CD45). (d) Distribution of expression of Ptprc and known markers in indicated subsets (full-length scRNA-seq). (e) Left: smFISH of Ptprc (CD45, green) co-stained with DCLK1 antibody (red). Scale bar: 50μm. Right: IFA co-staining of DCLK1 (red), Gfi1b-GFP (green) and CD45 (white) in the same tuft cell. Merge in bottom panel. Scale bar: 15μm. (f) FACS histogram of CD45 protein levels in Gfi1b-GFP⁺ cells (green), background (light grey) and monocytes (dark grey). (g) Proportion (y axis) of tuft subsets in 3′ droplet scRNA-seq (n=3 pooled mice) of EpCAM⁺ (left) or EpCAM⁺/CD45⁺ cells (* p<0.05, *** p<0.0005, hypergeometric test).

a–d. Diffusion-map embedding of 5,282 cells (points) progressing through stages of enterocyte differentiation (Methods). a,b Cells are colored by their cluster assignment (). Diffusion component 1 and 3 (DC-1 and DC-3) are associated with the transition from stem cells to progenitors (a), while DC-2 distinguishes between proximal and distal enterocyte fate commitment (b). c,d Cells are colored by the expression (log₂(TPM+1), color bar) of known and novel TFs associated with stages of differentiation (c), or with proximal or distal enterocyte differentiation (d). e. TF genes differentially expressed between proximal and distal cell fate. Heatmap shows the mean expression level (color bar) of 44 TFs differentially expressed between the proximal and distal (color legend) enterocyte clusters of (FDR < 0.05, Mann-Whitney U-test). f. Novel regional stem cell markers () identify distinct populations in diffusion map space. Close-up of stem-cell region of diffusion space (b, inset square) colored by expression level (log₂(TPM+1), color bars) pan-ISC marker Lgr5 (left), proximal ISC marker Gkn3 (center) and distal ISC marker (Bex1). Dashed line helps visualize separation of ISCs.

a. EEC subset discovery and regional location. tSNE of the 533 enteroendocrine cells (EECs) identified from the droplet-based datasets for whole SI and regional samples (color legend, n=8 mice, Methods). b. Agreement in hormone detection rates between 3′ droplet and full-length scRNA-seq. Scatter plot shows the detection rate (fraction of cells with non-zero expression of a given transcript) for a set of known EEC hormones, TFs and marker genes (color legend) in EECs from the full-length dataset (x axis), and from the 3′ droplet-based dataset (y axis). Linear fit (dashed line) and 95% confidence interval (shaded) are shown. c. Expression of key genes across subset clusters. tSNE plot shows cells colored by their assignment to the 12 clusters (top left plot; identical to ) or by the expression (log₂(TPM+1), color bar) of markers of immature EECs (Neurog3), genes encoding gut hormones (Sct, Sst, Cck, Gcg, Ghrl, GIP, Nts, PYY) or markers of enterochromaffin cells (Tac1, Reg4). d. Co-expression of GI hormones by individual cells. Left: Heatmap shows the expression (color bar) of canonical gut hormone genes (rows) in each of 533 individual EECs (columns), colored based on their assignment to the clusters in (color bar, top). Right: Heatmap shows for each cluster (columns) the percentage of cells (color bar, inset text) in which the transcript for each hormone (rows) is detected.

a,b. Paneth cell subsets. (a) tSNE of 10,396 single cells (points) obtained using a large cell-enriched protocol (Methods), colored by clusters annotated post-hoc. n=2 mice. b,c. Paneth cell subset markers. (b) Expression (row-wise Z-score, color bar) of genes specific (FDR<0.05, Mann-Whitney U-test, log₂ fold-change > 0.5) to each of the two Paneth cell subsets (average of 724.5 cells per subtype, down-sampled to 500 for visualization) shown in (a). c. Two Paneth subsets reflect regional diversity. Expression of the same genes (rows) as in (b) in Paneth cells from each of three small intestinal regions (176.3 cells obtained per each of the regions on average, columns; ); 11 of 11 Paneth-1 markers are enriched in the ileal Paneth cells, while 7/10 Paneth-2 markers are enriched in duodenal or jejunal Paneth cells (FDR <0.05, Mann-Whitney U-test). d. Validation of regional enterocyte markers. smFISH of Lct (red) and Fabp6 (white) in the duodenum (proximal, left) and ileum (distal, right). Dotted line: boundary between crypt and villi, green and yellow arrows: proximal and distal enterocytes, respectively. Scale bar, 50μm. e. Regional variation of intestinal stem cells. Expression (row-wise Z-score) of genes specific to stem cells from each intestinal region (FDR<0.05, Mann-Whitney U-test, log₂ fold-change > 0.5). There are 1,226.3 obtained cells per each of the three regions on average, down-sampled to 500 for visualization (columns).

a. Genes significantly induced in response to infection in a non-cell-type specific manner. tSNE visualization of 9,842 single IECs (dots) from control wild-type mice (left), mice infected with H. polygyrus for three or 10 days (middle) and mice infected with Salmonella (right). Cells are colored by the expression (log₂(TPM+1), color bar) of the indicated genes. Genes were selected as significantly differentially expressed in response to infection in a non-cell-type specific manner (FDR < 0.001 in both the 3′ scRNA-seq and full-length scRNA-seq datasets). b,c. Expression of the Tuft-1 and Tuft-2 signatures in the dataset of control, Salmonella and H. polygyrus infected cells. (b) Violin plots of the distribution of the respective signature scores (left and middle) and the expression of Dclk1 (right, log₂ (TPM+1, y axis) in cells (dots) in each of the tuft subsets (x axis). (c) tSNE mapping of the 409 tuft progenitor, Tuft-1 and Tuft-2 cells, colored by the scores for each signature (color bar, left and middle) and their assignment to subtype clusters via kNN-graph clustering (right). d. Induction of anti-parasitic genes by goblet cells in helminth infection. Distribution of expression (log₂ (TPM+1), y axis) of three anti-parasitic immunity genes up-regulated by goblet cells in response to H.polygyrus infection (FDR < 0.05, Mann-Whitney U-test), in control and infected mice. e. Anti-parasitic protein secretion by goblet cells during H. polygyrus infection. Immunofluorescence assay (IFA) of FFPE sections of RELMβ (top-left, red), E-cadherin (Bottom left, green) and their merged view (right) after 10 days of helminth infection. White arrow: sections of H. polygyrus. Scale bar, 200μm.

a–d. M cells in RANKL treated organoids. a–c tSNE of 5,434 single cells (dots) from control (left) or RANKL-treated (middle, right) intestinal organoids; or coloring each cell (b–c) by the expression (log₂(TPM+1), color bar) of the canonical M cell markers TNF-alpha induced protein 2 (Tnfaip2, M-sec, b) and glycoprotein 2 (Gp2, c). n=4 pooled wells per treatment condition. d. Expression of M cell marker genes^– in each of the organoid cell clusters. Violin plots show the distribution of expression levels (log₂(TPM+1)) for each of 10 previously reported M cell marker genes (columns), in the cells (dots) in each of 13 clusters, including mature M cells (red), identified by k-NN clustering of the 5,434 scRNA-seq profiles from organoids. e,f. M cell gene signature in vitro. Heatmaps show for each cell type cluster of organoid-derived intestinal epithelial cells (columns) the mean expression (color bar) of genes (rows) for known (grey bars) or novel (black bars) M cell markers (e) or transcription factors (f), identified as specific (FDR<0.05, Mann-Whitney U-test) to M cells both in vitro and in vivo (Methods). g. Congruence of in vitro and in vivo-derived M cell gene signatures. Violin plot shows the distribution of the mean expression of the in vitro-derived signature genes (y-axis) across the in vivo M cells (blue) and all other cells derived from the FAE (grey).

a. Tuft-1 and Tuft-2 cells. tSNE visualization of 102 tuft cells (dots) from the plate-based full-length scRNA-seq dataset (), labeled by their sub-clustering into Tuft-1 (orange) and Tuft-2 (brown) subtypes. n=8 mice. b. Gene signatures for Tuft-1 and Tuft-2 cells. Heatmap shows the relative expression (row-wise Z-scores, color bar) of the consensus Tuft-1 and Tuft-2 marker genes (rows; orange and brown, respectively), across single cells from the plate-based dataset (columns) assigned to Tuft-1 and Tuft-2 cell clusters (orange and brown, respectively). Top 25 genes shown for each subtype (all FDR < 0.01 and log₂ fold change > 0.1 in both plate- and droplet-based datasets). c. Tuft-2 signature genes are enriched in immune functions. Shown are the significantly enriched (Methods, FDR <0.1, -log₁₀(q-value), x axis) GO terms (y axis) in the gene signature for the Tuft-2 subset. d. Expression of neuron- and inflammation-related genes in Tuft-1 and Tuft-2 subsets, respectively. Plot shows for each gene (y axis) its differential expression (x axis) between Tuft-1 and Tuft-2 cells. Bar indicates Bayesian bootstrap estimates of log₂ (fold change), and hinges and whiskers indicate 25% and 95% confidence intervals, respectively. e. IL-33 not detected in tuft cells. Distribution of expression of Il33 in cell subsets (x axis), in full-length scRNA-seq. (* FDR<0.1, *** FDR <0.0001, Mann-Whitney U-test). f–g. Tuft-2 cells enriched for TSLP. f. Combined smFISH and IFA of TSLP (green) with DCLK1 (red), scale bar 10μm. g. Relative quantification (RQ) of mRNA expression by qPCR of Alpi, TSLP and Dclk1 (tuft cell markers) from Tuft-1, Tuft-2 or randomly selected EpCAM⁺ single cells identified from full-length scRNA-seq 96-well plate (16 cells per group). (* p<0.05, ** p<0.005, t-test). h. Validation of CD45 expression in Tuft-2 cells. IFA showing co-expression of the tuft cell marker, DCLK1 and CD45 (left) and CD45 (right, with higher intensity), yellow boxes show three representative tuft cells. Scale bar, 200μm. i. Isolation of Tuft-2 cells based on CD45 expression using FACS. tSNE of 332 EpCAM⁺/CD45⁺ FACS-sorted single cells (points, n=3 pooled mice), colored by unsupervised clustering (top left), the expression of the Tuft cell marker Dclk1 (top right), or the signature scores for Tuft-1 and Tuft-2 cells (bottom left and right, respectively).

a. QC metrics and post-hoc cluster annotation by the expression of known cell type markers. tSNE visualization of 1,522 single cells where individual points correspond to single cells. Top left corner to bottom right corner, in order: Cells are colored by their assignment to clusters, using a k-nearest neighbor (kNN) graph-based algorithm (Methods; Legend shows the cluster post-hoc annotation to cell types); mean expression (log₂(TPM+1), color bar) of several known marker genes for a particular cell type or state (indicated on top; same as in ); the mouse from which they originate (color legend) and its genotype, the FACS gate used to sort them (color legend), the number of reads per cell (color bar) and the number of genes detected per cell (color bar). n=8 mice. b. Cell-type-specific signatures. Heatmap shows the relative expression level (row-wise Z-scores, color bar) of genes (rows) in consensus cell-type-specific signatures (same genes as , with the exception of enterocytes), across the individual post-mitotic IECs (columns) in the full-length scRNA-seq data. Color code marks the cell types and their associated signatures. c. Mptx2, a novel Paneth cell marker. tSNE of the cells from the droplet-based 3′ scRNA-seq (left, as in ) and plate-based full-length scRNA-seq (right, as in a) datasets, colored by expression (log₂(TPM+1), color bar) of the mucosal pentraxin Mptx2. d. Cell-type-enriched GPCRs. Heatmap shows the relative expression (row-wise Z-scores, color bar) of genes encoding GPCRs (rows) that are significantly (FDR < 0.001, Mann-Whitney U-test, Methods) up- or down-regulated in the cells (columns) in a given cell type (top, color coded as in a) compared to all other cells, in the plate-based full-length scRNA-seq data. e. Cell-type-specific Leucine-rich repeat (LRR) proteins. Heatmap depicts the mean relative expression (column-wise Z-score of mean log₂(TPM+1) values, color bar) of genes (columns) encoding LRR proteins that are significantly (FDR < 0.001, Mann-Whitney U-test) up- or down-regulated in a given cell type (rows) compared to all other cells, in the plate-based full length scRNA-seq data. f. Cell type TFs and GPCRs. Average relative expression (Z-score of mean log₂(TPM+1), color bar) of the top TFs (left) and GPCRs (right, columns) enriched in each cell type (rows).

a. Generalized and pathogen-specific response genes. Volcano plots show for each gene (dot) the differential expression (DE, x axis), and its associated significance (y axis; (-log₁₀(Q value); Likelihood-ratio test) in response to either Salmonella (top) or H. polygyrus (bottom). Genes strongly up-regulated in Salmonella (FDR < 10⁻⁶) or H. polygyrus (FDR < 5×10⁻³) are highlighted in purple or red, respectively. All highlighted genes are significantly differentially expressed (FDR < 0.05) in both the 3′ scRNA-seq and the higher depth full-length scRNA-seq datasets. Left panels: all genes differentially expressed in the noted pathogen infection vs. uninfected controls; middle panels: the subset differentially expressed in both pathogens vs. control; right panels: the subset differentially expressed only in the noted pathogen but not the other (Methods). b. Global induction of enterocyte-specific genes across cells during Salmonella infection. tSNE of 9,842 single IECs from control wild-type mice (left) and mice infected with Salmonella (right). Cells are colored by the expression of the indicated genes, all specific to enterocytes in control mice (–) and strongly up-regulated by infection (FDR < 10⁻¹⁰ in both the 3′ scRNA-seq datasets and in the higher depth full length scRNA-seq dataset). c. IEC programs in Salmonella infection. Enriched (–log₁₀(q), x axis) GO terms in genes induced in Salmonella-treated IECs vs. control. d. Cell-intrinsic changes following Salmonella infection. Relative expression (row-wise Z-scores, color bar) of 104 genes (top) of which 58 (bottom) are specific to Salmonella infection, significantly up-regulated (FDR < 0.05, Mann-Whitney U-test, log₂ fold-change > 0.1) in enterocytes (columns) from Salmonella infection. 10 representative genes are labeled. e. Up-regulation of pro-inflammatory apolipoproteins Serum Amyloid A 1 and 2 (Saa1 and Saa2) in distal enterocytes under Salmonella infection. Violin plot shows log₂(TPM+1) expression level (y axis) of Saa1 (top) and Saa2 (bottom) across all post-mitotic cell-types from control and Salmonella-treated mice (n=4 mice, sample identity shown by color legend) (* FDR < 0.01; ** FDR < 0.0001, Mann-Whitney U-test). f. Up-regulation of anti-microbial peptides (AMPs) by Paneth cells following Salmonella infection. Violin plots show log₂ (TPM+1) expression levels (y axis) of genes encoding AMPs (panels) and the mucosal pentraxin Mptx2 (bottom right) in the cells (dots) from control and Salmonella–infected mice (n=4 mice, sample identity shown by color legend) (* FDR < 0.1; ** FDR < 0.01, ** FDR < 0.0001, Mann-Whitney U-test). g. Paneth cell numbers detected (using graph-clustering, Methods) after Salmonella infection. Frequencies (y-axis) of Paneth cells in each mouse (dots) under each condition (color legend). Error bars: standard error of the mean (SEM). (** FDR < 0.01, Wald test).

a–b. Relationships between EEC subsets. (a) Dendrogram shows the relationship between EEC clusters as defined by hierarchical clustering of mean expression profiles of all the cells in a subset (Methods). Estimates for the significance of each split are derived from 100,000 bootstrap iterations using the R package pvclust (■ p<0.1, * p < 0.05; ** p < 0.01, p < 0.001, χ² test). (b) Heatmap shows cell-cell similarities (Pearson’s r, color bar) between the 11 significant PCs scores (p<0.05, Methods) across the 533 EECs (rows, columns). Rows and columns are ordered using cluster labels obtained using unsupervised clustering (Methods). c. Subset specificity of gut hormones and related genes. Scatter plot shows each gene’s specificity to its marked cell subset (y axis; defined as the proportion of cells not in a given subset which do not express a given gene) and its sensitivity in that subset (defined as the fraction of cells of a given type which do express the gene, Methods). Subsets are color coded as in the legend. Genes are assigned to the subset where they are most highly expressed on average. Genes were chosen based on their known annotation as gut hormones (Cck, Gal, Gcg, Ghrl, GIP, Iapp, Nucb2, Nts, Pyy, Sct, Sst), enterochromaffin markers (Tph1, Tac1) and canonical EEC markers (Chga, Chgb). d. GPCRs enriched in different EEC subtypes. Heatmap shows the expression levels (row-wise Z-score, color bar) averaged across the cells in each of the EEC sub-types (columns) of 11 GPCR-encoding genes (rows) that are differentially expressed (FDR <0.25, Mann-Whitney U-test) in one of the EEC subtype clusters. The free fatty acid receptors (Ffar) 1 and 4 show specific expression patterns: Ffar1 highest in SIN cells, and also expressed by the Cck-expressing subsets previously termed I-cells (SIL-P, SILA and SIK-P), while Ffar4 is highest in the GIP-expressing subsets (SIK and SIK-P). These receptors are known to induce the expression of GIP and Gcg to maintain energy homeostasis. Ffar2 was expressed by some progenitors and by EC cells, but notably absent from GIP-expressing cells, while the oleoylethanolamide receptor Gpr119, important for food intake and glucose homeostasis, is expressed highest in SILA cells.

a,b. Quality metrics for scRNA-seq data. Shown are distributions of the number of reads per cell (left), the number of genes detected with non-zero transcript counts per cell (center) and the fraction of reads mapping to the mm10 mouse transcriptome per cell (right) in the droplet-based 3′ scRNA-seq data (a) and the plate-based full-length scRNA-Seq data (b). c–f. Agreement across batches. (c) Contribution of batches to each cluster. Each pie chart shows the batch composition (color coded legend) of each detected cluster (post-hoc annotation and number of cells are marked on top) in the droplet-based 3′ scRNA-seq dataset. All 10 replicates contribute to all clusters, and no major batch effect is observed. (n=6 mice). (d) Cell type proportions across batches. Shown is the proportion of detected cells (y axis) in each major cell type (x axis) in the droplet-based 3′ scRNA-seq dataset in each of 10 batches (dots, n=6 mice). Grey bar: mean; error bars: standard error of the mean (SEM). (e) Agreement in expression profiles across mice. Box and whisker plot shows the Pearson correlation coefficients (x axis) in average expression profiles (average log₂(TPM+1)) for cells in each cluster (y axis), across all pairs of mice. Black bar indicates median value, box edges correspond to the 25th and 75th percentiles, while whiskers indicate a further 1.5*IQR where IQR is the interquartile range. Note that clusters with additional sub-types (e.g., Tuft, enteroendocrine cells) show more variation, as expected. (f) Scatter plots comparing the average log₂(TPM+1) gene expression values between two scRNA-seq experiments from the droplet-based 3′ scRNA-seq dataset (top, x and y axis), two scRNA-seq experiments from the plate-based full length scRNA-seq dataset (center, x and y axis), or between the average of a plate-based full-length scRNA-seq (x axis) and a population control (y axis, bottom). Pearson correlation is marked top left. g. Additional QC metrics and post-hoc cluster annotation by the expression of known cell type markers. tSNE visualization of 7,216 single cells, where individual points correspond to single cells. Top left corner to bottom right corner, in order: Cells are colored by their assignment to clusters (top left, identical to ), mean expression (log₂(TPM+1), color bar) of several known marker genes for a particular cell type or state (indicated on top), the mouse from which they originate (color legend), the number of reads per cell (color bar), the number of genes detected per cell (color bar) and the number of transcripts as measured by unique molecular identifiers (UMIs) per cell.