Microbiota modulate transcription in the intestinal epithelium without remodeling the accessible chromatin landscape

Abstract

Microbiota regulate intestinal physiology by modifying host gene expression along the length of the intestine, but the underlying regulatory mechanisms remain unresolved. Transcriptional specificity occurs through interactions between transcription factors (TFs) and cis-regulatory regions (CRRs) characterized by nucleosome-depleted accessible chromatin. We profiled transcriptome and accessible chromatin landscapes in intestinal epithelial cells (IECs) from mice reared in the presence or absence of microbiota. We show that regional differences in gene transcription along the intestinal tract were accompanied by major alterations in chromatin accessibility. Surprisingly, we discovered that microbiota modify host gene transcription in IECs without significantly impacting the accessible chromatin landscape. Instead, microbiota regulation of host gene transcription might be achieved by differential expression of specific TFs and enrichment of their binding sites in nucleosome-depleted CRRs near target genes. Our results suggest that the chromatin landscape in IECs is preprogrammed by the host in a region-specific manner to permit responses to microbiota through binding of open CRRs by specific TFs.

Figure 1.

RNA-seq reveals transcriptome alterations in the presence and absence of microbiota in the mouse ileal and colonic epithelium. (A) Overview of experiments described in this study. Schematic of the mouse gastrointestinal tract showing the stomach (dark gray), duodenum (teal), jejunum (dark gray), ileum (blue), cecum (dark gray), and colon (red). Adapted from Stevens (1977). © 1933 by H.H. Dukes; © 1977 by Cornell University. Used by permission of the publisher, Cornell University Press. Approximately 6-cm sections of the duodenum, ileum, or colon were used for intestinal epithelial cell (IEC) isolation (see Methods). DNase-seq and RNA-seq were performed on intestinal epithelial cells (IECs, ∼90% purity) isolated from the ileum and colon of germ-free (GF), conventionally raised (CR), and ex-GF conventionalized (CV) mice. DNase-seq was also performed on IECs isolated from the duodenum of CR mice. (B) Fluorescence-activated cell sorting of pooled duodenal and ileal IECs labeled with antibodies marking either epithelial cells (EpCAM) or endothelial cells/leukocytes/platelets (CD31) reveal that ∼90% of cells were epithelial (EpCAM positive and CD31 negative). Similar results were obtained from colonic IEC preparations (data not shown). (C) Dendrogram of Jensen-Shannon divergence shows that RNA-seq replicates from GF, CR, or CV ileal or colonic IECs cluster. Note that anatomical location and environmental condition, rather than sibling relationship, drives the clustering. (D) Principal component analysis (PCA) confirms tissue type (PC1) and colonization state (PC2 and PC3) explains much of the variance observed in the RNA-seq data. Arrow tips denote sample position in PCA coordinates. (E) Volcano plot showing pairwise comparisons of RNA expression between GF versus CR and GF versus CV conditions for each tissue. Green dots represent genes that are significantly different (FDR < 0.05). (F) Hierarchical clustering of FPKM values for all genes that exhibited differential expression in the pairwise comparisons in D. Gene clusters were submitted to DAVID to determine Gene Ontology functional enrichment. Shown are top enrichments for each gene cluster. See also Supplemental Tables S1, S3, and S4.

Figure 2.

Endogenous DNase activity distinguishes open chromatin in mouse intestinal epithelial cells. (A) Pulse-field gel image of nuclei digested for 15 min at 37°C with increasing concentrations of exogenous DNase I. Note that high-molecular-weight (HMW) DNA is stable at 0°C; however, there is significant DNA digestion even with no addition of exogenous DNase when nuclei are incubated for 15 min at 37°C. (m) Yeast chromosome marker. (B) Endogenous DNase activity is detected within 30 sec after moving nuclei to 37°C, and by 8 min, most HMW DNA is digested. Patterns were consistent for duodenum, ileum, and colon (see Supplemental Fig. S4). The observed digestion pattern is similar to reported digestion patterns using exogenous DNase I (Song and Crawford 2010). For DNase-seq library preparation, nuclei digested for 2, 4, and 8 min were pooled to capture a range of DNase hypersensitivities. Libraries were prepared for duodenal, ileal, and colonic IECs. (C) Average phastCons scores plotted for the top 100,000 DHSs from duodenal, ileal, and colonic IECs centered at the peak maximum. Nongenic DNA flanking ileal DNase hypersensitive sites (DHSs) was used to assess background conservation (control). (D) Feature distribution of the top 100,000 and 25,000 DHSs from each tissue. Note the increased representation of promoter-associated sites (<2 kb from annotated transcription start sites) in the 25,000 DHSs with the highest signal intensity. (E) DNase-seq signal tracks from conventionally raised (CR) duodenal, ileal, and colonic IECs at the villin 1 (Vil1) locus. Note strong peaks at the transcription start site (DHS 1) and within the first intron (DHS 2). A 12.4-kb region including both DHS 1 and DHS 2 drives IEC-specific crypt and villous expression in the duodenum, ileum, and colon (Madison 2002); however, DHS 2 is required for crypt expression. For comparison, DNase-seq signal from the liver is also shown. (F) Venn diagram enumerating differential DHSs along the length of the GI tract. (G) Hierarchical clustering of differential DHSs across replicates of CR duodenal, ileal, and colonic IECs reveals open chromatin sites specific to each tissue. (cs) Colon-specific; (i:c) ileum and colon; (ds) duodenum specific; (d:c) duodenum and colon; (is) ileum specific; (d:i) duodenum and ileum. (H) Feature distribution showing that the majority of segment-specific DHSs are located in intergenic (>2 kb away from a gene body) or intronic regions of the genome. See also Supplemental Figures S2–S4 and Supplemental Tables S4 and S6.

Figure 3.

Differential open chromatin between ileal and colonic IECs correlates with differential gene expression. (A) Volcano plot showing pairwise comparison of RNA expression between conventionally raised (CR) ileal and colonic epithelium. Blue and orange dots represent genes more highly expressed in the ileum or colon, respectively (FDR < 0.05). (B) The fold difference in DNase signal intensity from CR ileal versus colonic IECs plotted against the average DNase signal observed in 250-bp windows. Significantly differential windows are highlighted in red and blue (FDR < 0.0001). (C) Representative signal track view highlighting two genes, diacylglycerol O-acyltransferase 1 (Dgat1) and aquaporin 8 (Aqp8), that exhibit differential open chromatin and transcript abundance in the ileum or colon. (D) Two-sided Kolmogorov-Smirnov goodness-of-fit test shows a positive relationship between the presence of a nearby tissue-specific DHS (within 2 kb of and including the gene body) and increased transcript abundance in that tissue. The y-axis shows the cumulative fraction of genes linked to a nearby tissue-specific DHS. Deviation from the null expectation that linked genes display a normal distribution centered on a fold change of 1 (black line) suggests that segment-specific DHSs are enriched near genes of higher expression in that tissue. See also Supplemental Figures S5 and S6.

Figure 4.

Life in the presence or absence of microbiota does not affect the intestinal epithelial accessible chromatin landscape. (A) Hierarchical clustering of differential DHSs across all replicates of conventionally raised (CR) versus germ-free (GF) ileal and colonic IECs. Note the similarity between GF and CR conditions for each tissue. (B) Density scatter plot showing the correlation of DNase-seq signal intensity for the top 100,000 DHSs for CR colon and CR ileum (top), GF ileum and CR ileum (bottom left), and GF colon and CR colon (bottom right). (C) The fold difference in DNase signal intensity plotted against the average DNase signal observed in 250-bp windows. Significantly differential windows are highlighted in red and blue (FDR < 0.0001). Comparing across tissues (CR colon vs. CR ileum) discovered thousands of differential DNase hypersensitive sites (see Fig. 3B). Comparing tissues in the presence or absence of microbiota reveals undetectable change in the open chromatin landscape in response to microbiota. (D) Representative signal track highlighting multiple genes in the ileum or colon that show differences in transcript abundance in the presence of microbiota but no change in the open chromatin landscape. (Angptl4) Angiopoietin-like 4; (Fgf15) fibroblast growth factor 15; (Cyp4b1) cytochrome P450, family 4, subfamily b, polypeptide 1; (Ang4) angiogenin, ribonuclease A family, member 4. See also Supplemental Tables S3 and S7.

Figure 5.

Microbiota do not substantially remodel the intestinal epithelial chromatin landscape upon acute colonization. (A) The fold difference in DNase signal intensity from conventionalized (CV) versus germ-free (GF) ileal or colonic IECs plotted against the average DNase signal observed in 250-bp windows. Significantly differential windows are highlighted in red and blue (FDR < 0.0001). (B) Representative signal track highlighting multiple genes in the ileum or colon that show differences in transcript abundance upon colonization with microbiota but no detectable change in the open chromatin landscape. (Bambi) BMP and activin membrane-bound inhibitor; (Sprr2b ) small proline-rich protein 2B; (Fabp6) fatty acid binding protein 6; (Plec) plectin. See also Supplemental Tables S3 and S7. (C) Two-sided Kolmogorov-Smirnov goodness-of-fit test shows a weak relationship between the presence of a nearby tissue-specific DHS (within 2 kb of the gene body) and increased transcript abundance in the GF versus CV ileum comparison at FDR < 0.05. The y-axis shows the cumulative fraction of genes linked to a nearby tissue-specific DHS. Deviation from the null expectation that linked genes display a normal distribution centered on a fold change of 1 (black line) suggests that CV-specific DHSs are enriched near genes of higher expression in CV ileal IECs. (D) Percent of differentially expressed genes that have a differential DNase hypersensitive site within their regulatory domain at two cutoffs (FDR < 0.0001 and FDR < 0.05). See also Supplemental Figure S7 and Supplemental Table S8.

Figure 6.

Integrating gene expression and open chromatin data identifies candidate transcription factors regulating response to microbiota colonization in the ileum. (A) Integration of our data set with published studies comparing ileum gene expression in the presence and absence of microbiota reveals a set of genes consistently up- or down-regulated by microbiota across at least four studies. Significant functional enrichments are shown for each gene set (see Supplemental Fig. S9; Supplemental Table S11). (B) Heat map of known transcription factors (TFs; including DNA binding transcription factors and transcription cofactors) that consistently display differential RNA expression levels in response to microbiota across multiple experimental studies in the ileum. Relative expression levels are indicated, where white represents no data. TFs are annotated with their predicted DNA binding domain family. Highlighted with blue or red circles are TFs with motif (C) or binding support (E). (C) Transcription factor binding site (TFBS) prediction in DHSs within the regulatory domain of genes consistently differentially regulated by microbiota in the ileum (see Supplemental Tables S12, S13). Fold enrichments were calculated relative to a GC matched background (Guturu et al. 2013). Motifs are colored based on fold enrichment ratios between down and up gene sets. (Teal) Enriched in DHSs near down genes; (brown) enriched in DHSs near up genes. Highlighted with blue or red circles are motifs matching TFs with differential expression (B) or binding support (E). (D) Scatter plot showing P-values for IPA upstream regulator analysis for the ileum up and ileum down gene lists identifies TFs and other factors that have previously been shown to influence expression of genes within these lists. (E) Plot showing the overlap of ChIP-seq peaks from multiple TFs (measured in various tissues) (see Supplemental Table S14) with DHSs within the regulatory domain of genes either consistently up-regulated (y-axis) or down-regulated (x-axis) by microbiota in the ileum. Fold enrichments were calculated relative to a uniformly distributed null model. Highlighted are the TFs where the up/down fold ratio is at least one standard deviation away from the mean of all fold ratios.

Figure 7.

Integrating gene expression and open chromatin data identifies candidate transcription factors regulating response to microbiota colonization in the colon. (A) Integration of our data set with published studies comparing colon gene expression in the presence and absence of microbiota reveals a set of genes consistently up- or down-regulated by microbiota across at least four studies. Significant functional enrichments are shown for each gene set (see Supplemental Fig. S9; Supplemental Table S11). (B) Heat map of known transcription factors (TFs; including DNA-binding transcription factors and transcription cofactors) that consistently display differential RNA expression levels in response to microbiota across multiple experimental studies in the colon. Relative expression levels are indicated, where white represents no data. TFs are annotated with their predicted DNA binding domain family. Highlighted with blue or red circles are TFs with motif (C) or binding support (E). (C) Transcription factor binding site (TFBS) prediction in DHSs within the regulatory domain of genes consistently differentially regulated by microbiota in the colon (see Supplemental Tables S12, S13). Fold enrichments were calculated relative to a GC matched background. Motifs are colored based on fold enrichment ratios between down and up gene sets. (Teal) Enriched in DHSs near down genes; (brown) enriched in DHSs near up genes. Highlighted with blue or red circles are motifs matching TFs with differential expression (B) or binding support (E). (D) Scatter plot showing P-values for IPA upstream regulator analysis for the colon up and colon down gene lists identifies TFs and other factors that have previously been shown to influence expression of genes within these lists. (E) Plot showing the overlap of ChIP-seq peaks from multiple TFs (measured in various tissues) (see Supplemental Table S14) with DHSs within the regulatory domain of genes either consistently up-regulated (y-axis) or down-regulated (x-axis) by microbiota in the colon. Fold enrichments were calculated relative to a uniformly distributed null model. Highlighted are the TFs where the up/down fold ratio is at least one standard deviation away from the mean of all fold ratios.