Gene Co-Expression Network Analysis Identifies Vitamin D-Associated Gene Modules in Adult Normal Rectal Epithelium Following Supplementation

Abstract

Colorectal cancer (CRC) is a common, multifactorial disease. While observational studies have identified an association between lower vitamin D and higher CRC risk, supplementation trials have been inconclusive and the mechanisms by which vitamin D may modulate CRC risk are not well understood. We sought to perform a weighted gene co-expression network analysis (WGCNA) to identify modules present after vitamin D supplementation (when plasma vitamin D level was sufficient) which were absent before supplementation, and then to identify influential genes in those modules. The transcriptome from normal rectal mucosa biopsies of 49 individuals free from CRC were assessed before and after 12 weeks of 3200IU/day vitamin D (Fultium-D3) supplementation using paired-end total RNAseq. While the effects on expression patterns following vitamin D supplementation were subtle, WGCNA identified highly correlated genes forming gene modules. Four of the 17 modules identified in the post-vitamin D network were not preserved in the pre-vitamin D network, shedding new light on the biochemical impact of supplementation. These modules were enriched for GO terms related to the immune system, hormone metabolism, cell growth and RNA metabolism. Across the four treatment-associated modules, 51 hub genes were identified, with enrichment of 40 different transcription factor motifs in promoter regions of those genes, including VDR:RXR. Six of the hub genes were nominally differentially expressed in studies of vitamin D effects on adult normal mucosa organoids: LCN2, HLA-C, AIF1L, PTPRU, PDE4B and IFI6. By taking a gene-correlation network approach, we have described vitamin D induced changes to gene modules in normal human rectal epithelium in vivo, the target tissue from which CRC develops.

Transcriptional dynamics of colorectal cancer risk associated variation at 11q23.1 correlate with tuft cell abundance and marker expression in silico

“…In addition, WGCNA also does not require a hard thresholding of correlations, a major advantage over other correlation-based methods that rely on arbitrary cut offs. We have recently shown that WGCNA can be efficacious at identifying genes driving transcriptional dynamic changes in the colorectum of patients treated with Vitamin D 21 , even in the absence of statistically significant changes in differential expression analysis. Moreover, as CRC risk-associated eQTL targets are identified by bulk expression methods, findings are inherently limited in their potential to detect cell-specific changes, particularly in rare cell-types; expression changes within which may be masked due to low relative abundance.…”

Transcriptional dynamics of colorectal cancer risk associated variation at 11q23.1 are correlated with tuft cell abundance and marker expression in silico

“…In addition, WGCNA also does not require a hard thresholding of correlations, a major advantage over other correlation-based methods that rely on arbitrary cut offs. We have recently shown that WGCNA can be efficacious at identifying genes driving transcriptional dynamic changes in the colorectum of patients treated with Vitamin D (Blackmur et al, 2022), even in the absence of statistically significant changes in differential expression analysis. Moreover, as CRC risk-associated eQTL targets are identified by bulk expression methods, findings are inherently limited in their potential to detect cell-specific changes, particularly in rare cell-types; expression changes within which may be masked due to low relative abundance.…”

“…In addition, WGCNA also does not require a hard thresholding of correlations, a major advantage over other correlation-based methods that rely on arbitrary cut offs. We have recently shown that WGCNA can be efficacious at identifying genes driving transcriptional dynamic changes in the colorectum of patients treated with Vitamin D (Blackmur et al, 2022), even in the absence of statistically significant changes in differential expression analysis.…”