Postoperative Ileum Transcriptomics Implicate Sex-Biased Mechanisms in Crohn's Disease Recurrence

Abstract

Background & aims: Despite widespread biologic use, more than 70% of patients with Crohn's disease require resectional surgery, most commonly of the terminal ileum. Gene expression and genetics of the neoterminal ileum at postoperative, surveillance colonoscopies highlight pathways of disease recurrence. Postoperative colonoscopy transcriptomes were interrogated to evaluate the hypothesis that specific molecular mechanisms contribute to recurrent Crohn's pathophysiology.

Methods: Ribo-depleted, paired-end sequencing was run on 339 neoterminal ileal pinch biopsies from 267 patients with (Rutgeerts i2b+) and without (Rutgeerts i2a or lower) recurrent disease. Differential gene and transcript usage were assessed. Expression quantitative trait loci link genetic variation with gene expression. Serial sampling was performed on 70 patients.

Results: At colonoscopy, 4171 genes increased and 3579 genes decreased in recurring vs nonrecurring patients. Although gene expression was highly correlated (r = 0.71), we observed and replicated higher dynamic ranges of gene expression in male compared with female patients. Activation of both pro- (tumor necrosis factor, interferon gamma) and anti-inflammatory (transforming growth factor beta) pathways was observed; importantly, multiple nuclear hormone receptor pathways demonstrated activation, including both estrogen and dihydrotestosterone pathways, whereas progesterone was inhibited. We observed sex-specific expression quantitative trait loci driven by recurrent samples and differential transcript usage related to lipid metabolism, membrane trafficking, and the extracellular matrix.

Conclusions: In recurrent disease of the postoperative neoterminal ileum, markedly greater dynamic ranges of gene expression occur in male compared with female patients. Pathway analyses implicate numerous nuclear hormone pathways, highlighting new mechanisms for therapeutic targeting beyond pro-inflammatory cytokine blockade. This study identifies key covariates and pathways of disease recurrence, many of which are distinct from drivers of initial disease susceptibility.

Keywords: Crohn’s Disease; Genomics; Inflammatory Bowel Disease; RNA-Seq; Single Cell.

Figure 1.

Differential expression using DESeq2. (A) Comparison between all nonrecurring and recurring samples (negative values represent genes with higher expression in nonrecurring samples in A to C). (B) Comparison between nonrecurring and recurring male samples after down sampling to match female numbers. (C) Comparison between nonrecurring and recurring female samples. (D) Comparison of recurrence vs nonrecurrence log2FC values for female patients and male patients show very high correlation (correlation coefficient r = 0.706), but a consistently higher fold change in male patients. (E) PC1 calculated using the top 20 recurrence vs nonrecurrence genes in male patients predicts recurrence status in female patients. (F) Comparison between anti-TNF use and no anti-TNF use (negative values represent genes with higher expression without anti-TNF).

Figure 2.

Expression differences among top pathways and sex-related expression changes in an Affymetrix replication cohort. (A and B) Expression of genes regulated by dihydrotestosterone in single-cell data collected from the terminal ileum of patients with CD. (C–F) Fold change comparisons for genes downstream of top IPA regulators. Male fold change values are on the x-axis and female values are on the y-axis. Affymetrix array differential expression results when comparing (G) female and (H) male samples (40 nonrecurrent and 28 inflamed recurring samples of each sex). (I) Comparison of beta coefficients for each gene by sex.

Figure 3.

Sex differential genetic regulation of gene expression is driven by recurrence status. (A) eQTL effect estimates (n = 81) in male patients vs female patients for differentially expressed genes in recurrent cases. The effect sizes are shown for the minor allele. (B) Response eQTL effect in recurrence (R1) vs nonrecurrence (R0) groups in male vs female patients for MMP3 gene (matrix metalloproteinase) and (C) C6 (complement C6) gene. The slope represents the response eQTL effect estimate. (D) Proportion of eQTLs having same vs opposite direction of eQTL effect in male vs female patients, vs differential expression direction in the recurrence groups. (E) Proportion of largest effect eQTLs (ß > 0.3 or < −0.3) having the same vs opposite direction of eQTL effect in male patients vs female patients, vs differential expression direction in the recurrence groups.

Figure 4.

Differential transcript usage and expression indicate that recurrence status is associated with altered splicing biased toward the rectum. (A) Tissue-specific DTUs are associated with recurrence status. (B) Violin plots of PC2 scores of samples with nonrecurring (R0) and recurring (R1) disease show median and interquartile range as boxes, and full kernel density distribution of scores as surrounding shapes. The significant difference is from a Wilcoxon Rank Sum Test. (C) Tissue-specific DEGs are associated with recurrence status. (D) Violin plot of PC1 scores in nonrecurring and recurring disease show recurrence status is associated with tissue-biased expression patterns (Wilcoxon rank sum test). (E) Pathways enriched in the top 100 genes in PC1. Shading represents Bonferroni adjusted P value. (F) Pathways enriched in the top 100 genes in PC4. Shading represents Bonferroni adjusted P value.