A Map of Toll-like Receptor Expression in the Intestinal Epithelium Reveals Distinct Spatial, Cell Type-Specific, and Temporal Patterns

Abstract

Signaling by Toll-like receptors (TLRs) on intestinal epithelial cells (IECs) is critical for intestinal homeostasis. To visualize epithelial expression of individual TLRs in vivo, we generated five strains of reporter mice. These mice revealed that TLR expression varied dramatically along the length of the intestine. Indeed, small intestine (SI) IECs expressed low levels of multiple TLRs that were highly expressed by colonic IECs. TLR5 expression was restricted to Paneth cells in the SI epithelium. Intestinal organoid experiments revealed that TLR signaling in Paneth cells or colonic IECs induced a core set of host defense genes, but this set did not include antimicrobial peptides, which instead were induced indirectly by inflammatory cytokines. This comprehensive blueprint of TLR expression and function in IECs reveals unexpected diversity in the responsiveness of IECs to microbial stimuli, and together with the associated reporter strains, provides a resource for further study of innate immunity.

Keywords: Paneth cells; TLR; innate immunity; intestinal epithelial cells; intestinal homeostasis.

Figure 1:. Epithelial TLR expression differs dramatically between the SI and colon.

(A)Design of TLR reporter mice. A reporter cassette including an epitope tag (HA or FLAG) and an internal ribosome entry site followed by a fluorescent protein (eGFP, eYFP, or tdTomato) was inserted at the 3’ end of the endogenous TLR gene for the TLRs indicated. Note: for simplicity, the exon/intron structure for each TLR is not shown. (B) Expression of TLR reporter genes was confirmed by measuring fluorescence (GFP/YFP or tomato) by flow cytometry from live, CD11c⁺ cells from the spleens of indicated mice. Flow cytometry plots are representative of 2 independent experiments, each including at least 3 mice of each genotype. (C) SI and colon sections of reporter mice stained with Epcam (epithelial cells) and antibodies to either GFP/YFP or tomato (TLR reporters). All SI images were from the ileum. Images were taken at 20× magnification and exposures were set using staining of WT mouse tissues. Images are representative of at least 3 independent experiments, each including 2–5 mice of each genotype. Scale bars are 50 microns. (D) TLR expression in IECs was assessed by RNA-sequencing performed on IECs isolated from WT mice from 2 litters. Values are displayed as a number of transcripts per million total reads. Error bars are mean +/− SEM. (E) Proximal colon sections from radiation chimeras in which WT bone marrow was reconstituted into WT or TLR reporter mice stained with Epcam (epithelial cells) and antibodies to the HA epitope tag (TLR protein). HA staining is separated in bottom panels for easier visualization. Images were taken at 20× magnification and exposures were set using staining of WT mouse tissues. White arrows indicate apical staining, yellow arrows indicate basolateral staining, and stars indicate intracellular staining. Images are representative of 2 independent experiments, each including 1–4 mice of each genotype. Scale bars are 50 microns.

Figure 2:. TLR5 expression is restricted to Paneth cells in the SI epithelium.

(A) SI sections from reporter mice stained for lysozyme (Paneth cells) and antibodies to either GFP/YFP or tomato (TLR reporters). Images were taken at 63× magnification and exposures were set using staining of WT mouse SI. Images are representative of at least 2 independent experiments, each including 2–5 mice of each genotype. Scale bars are 50 microns. (B) SI sections from TLR5^TOM x LGR5-GFP mice were stained for tomato (TLR5), GFP (LGR5), and lysozyme (Paneth cells). Images were taken at 63× magnification and exposures were set using staining of WT mouse SI. Images are representative of 2 independent experiments, each including 2–5 mice. Scale bars are 10 microns. (C) Tomato fluorescence from total IECs (live, CD45^-Epcam⁺) or Paneth cells (live, CD45^-Epcam⁺CD24⁺SSC^hi) isolated from crypt preparations from the SI of WT or TLR5^TOM mice. (D) Flow cytometry plot of total IECs from crypt preparations from TLR5^TOM mice pregated on tomato expression. Tomato+ cells are shown in red and tomato-cells are shown in black. For C-D, flow cytometry plots are representative of 3 independent experiments, each including 2–5 mice. (E) Tomato⁺ and tomato^- IECs were sorted from SI crypt preparations from TLR5^TOM mice and gene expression was assessed by quantitative RT-PCR. Due to low cell numbers, cells from three separate sorting experiments, including IECs from over twenty mice, were pooled to generate results. Results are shown as expression relative to beta actin. (F) Paneth cells (CD24⁺SSC^hi) were gated from SI crypt preparations from indicated mice and reporter fluorescence was assessed using flow cytometry. Plots are representative of at least 3 independent experiments, each including 2–5 mice of each genotype.

Figure 3:. SI TLR5 expression gradually decreases during the neonatal period.

(A) SI sections from d1 WT and TLR reporter mice were stained with antibodies to Epcam (epithelial cells) and GFP/YFP or tomato (TLR reporters). Images were taken at 20× and exposures were set using staining of WT sections. Images are representative of 2 independent experiments, each including 2–5 mice of each genotype. Scale bars are 50 microns. (B) Timecourse of tomato reporter expression in SI sections from TLR5^TOM mice stained with DAPI (nuclei), lysozyme (Paneth cells) and tomato (TLR5) and images were separated for easier comparison. Arrows denote tomato⁺ Paneth cells while asterisks denote tomato⁺ nonepithelial cells in the lamina propria. Images were taken at 20× and exposures were set using staining of WT sections. Images are representative of 2 independent experiments, each including 2–3 mice. Scale bars are 50 microns. (C) Tomato fluorescence from total IECs isolated from WT or TLR5^TOM SI at the indicated time points assessed using flow cytometry. Data are representative of 3 independent experiments, each including 2–5 mice of each genotype. (D) The percentage of tomato⁺ SI IECs from indicated mice beginning one week after birth quantified by flow cytometry using gating strategy from (C). Results are pooled from 2–3 independent experiments and each time point includes mice from at least 2 separate litters. Error bars are mean +/− SEM. (E) IECs were isolated from the SI of WT mouse from our colony (SPF) and WT mice housed in germ-free isolators (GF) and quantitative RT-PCR was performed to measure TLR expression over time. Values are normalized to expression of beta actin and then expressed as fold over the expression observed in IECs collected at day 7. For SPF mice, data represent 4 mice per time point and the experiment was repeated 2 times. For GF mice, results from 2 separate experiments were pooled to generate results. Error bars are mean +/− SD. Stars denote significant differences between expression values at a time point compared to the value at day 7.Significance was calculated using one-way ANOVA. All points without stars are not significant. (F) Expression of TLR5 over time in SI IECs isolated from SPF and GF mice. Expression is relative to expression of beta actin. Significance was analyzed using students t-test. For E-F * p ≤ 0.5, ** p ≤ .01, *** p ≤ 0.001.

Figure 4:. The distinct patterns of TLR expression in adult IECs are maintained in intestinal organoids.

(A) Images from SI organoids, SI organoids skewed to enrich for Paneth cells, and colon organoids. Paneth cell skewed organoids are smaller, rounder, and contain dark granules. Images are representative of at least 5 independent experiments. (B) RNA was isolated from SI, Paneth cell skewed, and colon organoids at day 6 after passage and the gene expression in these organoids was assessed using Nanostring. Each point represents an independent experiment. Values shown are the normalized number of transcripts detected. Error bars are mean +/− SEM. * p ≤ 0.5, ** p ≤ .01, *** p ≤ 0.001. (C)Reporter fluorescence (YFP/GFP or tomato) in live, CD45^-Epcam⁺ total IECs isolated from indicated organoids on day 6 after passage compared to baseline WT fluorescence in these channels. Flow cytometry plots are representative of 3 independent experiments, each including 2–5 mice of each genotype. (D) TLR expression in SI and colon organoids at day 6 after passage assessed using RNA-sequencing. Values are transcripts per million. Error bars are mean +/− SEM. (E) Tomato fluorescence in total IECs (live, CD45^- Epcam⁺) or Paneth cells (live, CD45^- Epcam⁺CD24⁺SSC^hi) from SI organoids from WT and TLR5^TOM mice. Flow cytometry plots are representative of 3 independent experiments. (F) Tomato fluorescence in total IECs (live, CD45^- Epcam⁺) or Paneth cells (live, CD45^- Epcam⁺CD24⁺SSC^hi) from organoids generated from d7 TLR5^TOM mice. WT mouse organoids are shown as a control. Flow cytometry plots are representative of 2 independent experiments.

Figure 5:. TLR5 signaling initiates a common gene program in colon IECs and Paneth cells.

(A) Expression of TLR5 in Paneth cell skewed and colon organoids assessed using RNA-sequencing. Values are transcripts per million. Significance was determined using one-way ANOVA ** p ≤ .01. (B) Volcano plots showing the results of RNA-sequencing on SI organoids, colon organoids, and Paneth cell skewed organoids. Genes are plotted by the average fold change vs. significance of their induction after stimulation with flagellin for 4 hours. Log10(q values) were capped at 20 for this analysis. Selected significantly induced genes are labeled. (C) The top 30 most highly induced genes from flagellin-stimulated Paneth cell skewed organoids and colon organoids, listed from highest to lowest fold change in Paneth cell skewed organoids. Genes were pre-filtered on significance, with q values of <0.05. Genes from the top 30 most induced genes in stimulated Paneth cells that are induced less than 1.5-fold in colon organoids are highlighted in orange, and genes from the top 30 most induced genes in colon organoid that are induced less than 1.5 fold in Paneth cell skewed organoids are highlighted in purple. p-values listed as NA were not calculated because the counts for indicated genes included at least one outlier that did not fall within the normal distribution. (D) Top 10 gene sets enriched in Paneth cell skewed organoids stimulated with flagellin using gene set enrichment analysis. For A-D, RNA-sequencing was performed on RNA from unstimulated and flagellin-stimulated SI organoids isolated from 3 independent experiments, colon organoids isolated from 4 independent experiments, and Paneth cell skewed organoids isolated from 6 independent experiments.

Figure 6:. Transcript analysis reveals distinct patterns of TLR- and cytokineinduced gene expression in SI and colon IECs.

(A) Heat map of gene expression results from Nanostring analysis. Organoids were stimulated with the indicated ligands or cytokines for 4 hours and total RNA was isolated and used for Nanostring analysis. Values shown are fold change over unstimulated organoids, calculated from the average of 3 independent experiments. (B) Heat map of Nanostring results for a subset of genes associated with antimicrobial responses. Stimulation conditions that resulted in reproducible induction of multiple genes are shown. Values were calculated as in (A). (C) Reg3γ expression levels in organoids stimulated with TLR ligands and cytokines determined using Nanostring. Results shown are normalized transcript counts. (D) Fold induction of genes significantly induced by TLR ligands but not by IL-22 or IFNγ. TLR ligands that induced the strongest responses in Paneth cell skewed and colon organoids are shown. For C-D, each point represents Nanostring results from a separate experiment. Error bars are mean +/− SEM. Significance was determined by comparing normalized transcript counts to unstimulated condition using one-way ANOVA. * p ≤ 0.5, ** p ≤ .01, *** p ≤ 0.001.

Figure 7:. In vivo confirmation of a TLR-induced gene program.

(A) Nanostring gene expression analysis of total RNA from colon IECs harvested 1 hour after intraperitoneal injection of 20 μg of flagellin or PBS. Results shown are normalized transcript counts. Each point represents an individual mouse. Error bars are mean +/− SEM. Significance was determined using one-way ANOVA. * p ≤ 0.5, ** p ≤ .01, *** p ≤ 0.001. (B) Comparison of average fold change between flagellin-stimulated ex vivo colon IECs and colon organoids. Selected genes are labeled. Added line is x=y. (C) Ccl20 and Cxcl1 levels in colon homogenates from indicated radiation chimeras 1 hour after intraperitoneal injection with PBS or 20 μg flagellin measured using ELISA. Error bars are mean +/− SEM. Results are representative of two independent experiments, each including 3–6 mice per condition. Significance was determined using one-way ANOVA. * p ≤ 0.5, ** p ≤ .01, *** p ≤ 0.001.