Site-specific programming of the host epithelial transcriptome by the gut microbiota

Abstract

Background: The intestinal epithelium separates us from the microbiota but also interacts with it and thus affects host immune status and physiology. Previous studies investigated microbiota-induced responses in the gut using intact tissues or unfractionated epithelial cells, thereby limiting conclusions about regional differences in the epithelium. Here, we sought to investigate microbiota-induced transcriptional responses in specific fractions of intestinal epithelial cells. To this end, we used microarray analysis of laser capture microdissection (LCM)-harvested ileal and colonic tip and crypt epithelial fractions from germ-free and conventionally raised mice and from mice during the time course of colonization.

Results: We found that about 10% of the host's transcriptome was microbially regulated, mainly including genes annotated with functions in immunity, cell proliferation, and metabolism. The microbial impact on host gene expression was highly site specific, as epithelial responses to the microbiota differed between cell fractions. Specific transcriptional regulators were enriched in each fraction. In general, the gut microbiota induced a more rapid response in the colon than in the ileum.

Conclusions: Our study indicates that the microbiota engage different regulatory networks to alter host gene expression in a particular niche. Understanding host-microbiota interactions on a cellular level may facilitate signaling pathways that contribute to health and disease and thus provide new therapeutic strategies.

Figure 1

The microbiota elicits specific transcriptional responses in different locations of the gut epithelium. (A) Immunostaining using epithelial cell type-specific antibodies and quantification. n = 8 (GF) and 9 (CR); *P < 0.05, ***P < 0.001 (two-way ANOVA). All pictures are in the same magnification. The scale bar indicates 100 μm. Data show mean ± standard error of the mean. (B) Schematic overview of the experimental setup. Tip and crypt epithelial fractions were harvested by LCM from ileum and colon of GF and CR mice and then used for microarray analysis. (C) Hierarchical clustering of the microarray data. (D) Venn diagram of microbiota-dependent genes (P < 0.001 for GF versus CR) in the four different LCM fractions.

Figure 2

Quantitative PCR validation of site-specific microbiota-induced transcriptional responses in the gut epithelium. Tip and crypt epithelial fractions were harvested via LCM from ileum and colon of GF and CR mice. RNA was isolated from each fraction, linear amplified and directly used for quantitative PCR analysis. n = 4 to 5 per group. Data show mean ± standard error of the mean. *P < 0.05, ***P < 0.001, ****P < 0.0001 (two-way ANOVA). ns, not significant; nd, not detectable.

Figure 3

Microbiota-responsive gene functions in different locations of the gut epithelium. Gene ontology analysis of microbiota-dependent genes in the LCM fractions. Only terms with P < 10⁻⁵ in any of the four epithelial fractions and more than five members were considered. Figure depicts log-transformed adjusted P-values for GF versus CR comparison.

Figure 4

Regulatory factors enriched among promoters of microbiota-responsive genes. Transcription factor binding sites were predicted in promoter sequences (1 kb) of microbiota-responsive genes. Lists of microbially responsive genes were also compared with target genes inferred from published ChIP-seq data. Results were merged and only the 50 most significantly enriched regulators considered for each of the four epithelial fractions. TF, transcription factor.

Figure 5

The colonic epithelium responds quicker to microbial exposure than the ileal. (A) Schematic overview of the experimental setup. GF mice were colonized with a normal microbiota. Tip and crypt epithelial fractions were harvested by LCM on days (d) 0, 1, 3, 5 and 7 during the time course of colonization and then used for microarray analysis. (B) Hierarchical clustering of the microarray data.

Figure 6

Clustering of microbially altered genes according to their responses during colonization of GF mice. For each LCM fraction, normalized gene expression values were analyzed for their changes over time during colonization. Those genes with significant changes between at least two time points were clustered into four different clusters solely according to their expression profile using a consensus-clustering approach. Besides their expression pattern, clusters of different tissue fractions are completely unrelated (for example, cluster 1 of ileum crypt or tip). Data show mean ± standard deviation. Numbers indicate how many genes are contained in the respective cluster. The noted function refers to the most abundant GO term associated with the clustered genes.