Abstract

   A super-enhancer (SE) is a cluster of enhancers with a relatively high
   density of particular chromatin features. SEs typically regulate key
   genes that can determine cell identity and differentiation. Identifying
   SEs and their effects may be critical in predicting key regulatory
   genes, such as master transcription factor genes or oncogenes. Signal
   inducible SEs are dense stretches of signal terminal transcription
   factor (TF) binding regions, and may modulate the interaction between
   environmental factors (e.g., Vitamin D) and genetic factors (i.e., risk
   variants) in complex diseases such as multiple sclerosis (MS). As a
   complex autoimmune disease, the etiology and progression of MS,
   including the interaction between Vitamin D and MS risk variants, is
   still unclear and can be explored from the aspect of signal SEs.
   Vitamin D [with its active form: 1,25(OH)[2]D[3]], is an environmental
   risk factor for MS. It binds the Vitamin D receptor (VDR) and regulates
   gene expression. This study explores the association between VDR
   super-enhancers (VSEs) and MS risk variants. Firstly, we reanalyse
   public ChIP-seq and RNA-seq data to classify VSEs into three categories
   according to their combinations of persistent and secondary VDR
   binding. Secondly, we indicate the genes with VSE regions that are near
   MS risk variants. Furthermore, we find that MS risk variants are
   enriched in VSE regions, and we indicate some genes with a VSE
   overlapping MS risk variant for further exploration. We also find two
   clusters of genes from the set of genes showing correlation of
   expression patterns with the MS risk gene ZMIZ1 that appear to be
   regulated by VSEs in THP-1 cells. It is the first time that VSEs have
   been analyzed, and we directly connect the genetic risk factors for MS
   risk with Vitamin D based on VSEs.

   Keywords: vitamin D, vitamin D receptor, inducible super-enhancer, risk
   allele, multiple sclerosis

Introduction

   The transcription of genes depends on the interaction of their promoter
   regions with polymerases that synthesize RNA from genomic DNA in
   combination with an array of regulatory factors ([31]1, [32]2).
   Additionally, transcription is aided by cis-acting regulatory elements
   that can be located relatively distant to the promoter region and can
   bind activator proteins. These enhancer elements are able to increase
   the level of gene transcription.

   Super-enhancers (SEs) are dense clusters of enhancers. They differ from
   typical enhancers (TEs) in the density of enhancer elements. Enhancers,
   including both TEs and SEs, can be annotated by histone status, i.e.,
   H3K27ac and H3K4me1. In addition, high chromatin accessibility [e.g.,
   as determined using FAIRE-seq or DNaseI hypersensitive sites (DHS)],
   master transcription factor binding (e.g., PU.1 for monocytes, RORγt
   for Th17 cells), and pervasive factors in the transcription machinery
   (e.g., p300, MED1, BRD4, and RNA polymerase II) are all highly
   correlated with SE regions and can be used to identify SE regions.

   Although defined arbitrarily according to enhancer signal density, SEs
   have proved extremely valuable in predicting key regulatory regions or
   genes for cell identity or cell differentiation ([33]3). For example,
   inappropriate acquisition of SE in an oncogene, such as c-MYC, will
   increase its expression and lead to oncogenesis ([34]4, [35]5). SE
   regions promote the expression of autoimmune disease-associated genes.
   For example, the drug JQ1 [BET (bromodomain and extra-terminal domain)
   inhibitor] inhibits the expression of inflammatory arthritis risk gene
   CXCR4 by affecting its SE region ([36]6), and tofacitinib [JAK (Janus
   kinase) inhibitor] disproportionately inhibits the expression of
   rheumatoid arthritis risk genes with SE regions compared with those
   risk genes without SE regions ([37]7).

   Previous research has focused on classic SEs identified by chromatin
   accessibility, master transcription factors or pervasive factors in the
   transcription machinery, but recently, a new concept of
   signal-inducible SEs has been proposed ([38]8). It was found that
   estrogen could induce the generation of new signal SE regions, which
   were bound by the terminal transcription factor (TF) ERα of the
   estrogen signaling pathway. The advantage of signal SEs for research is
   that they can provide important information on the signal terminal TF
   cistrome before and after signal stimulation.

   Multiple sclerosis (MS) is a complex autoimmune disease with multiple
   risk factors including genetic variants and Vitamin D deficiency
   ([39]9, [40]10). Until now, the functional variants of many genome-wide
   association study (GWAS) loci have not been identified. In addition,
   the mechanism underlying the interaction between genetic factors and
   Vitamin D in MS etiology and progression is still unclear. Some MS risk
   single nucleotide polymorphisms (SNPs) have been found located in
   Vitamin D Receptor (VDR) binding sites in lymphoblastoid cell lines
   (LCLs) ([41]11) and conversely, VDR binding sites are also enriched in
   MS risk regions identified by GWAS (MS risk SNP ± 100 kb) ([42]12).
   Furthermore, MS risk SNPs are enriched in classic SE regions of CD4^+ T
   cells and monocytes ([43]3, [44]7). The risk alleles could potentially
   modulate the regulatory effects of these SEs on key genes in specific
   cell types.

   We hypothesize that VDR super-enhancers (VSEs) may be signal inducible
   SEs relevant to MS development, and that GWAS-identified MS risk loci
   may influence the function of such VSEs. To assess this, we re-analyzed
   next-generation sequencing (NGS) data from cell stimulation experiments
   employing hormones and their nuclear receptors. In particular, we were
   interested in the 1,25(OH)[2]D[3] (the active form of Vitamin D) and
   VDR couple, and its association with MS. Firstly, we analyzed the
   overlap between VSEs and classic SEs on their genomic regions and
   closest genes. We classified all VSEs into three patterns (VSE1: only
   persistent VDR binding; VSE2: both persistent and secondary VDR
   binding; VSE3: only secondary VDR binding) and analyzed their
   characteristics. Furthermore, we analyzed the enrichment of MS risk
   SNPs in VSE regions, and confirmed that VSEs were significantly
   enriched for MS risk SNPs.

   ZMIZ1 and EOMES have been identified as MS risk genes by cohort
   studies, and are differentially expressed in whole blood between MS
   patients and healthy controls ([45]13–[46]15). ZMIZ1 is highly
   expressed in monocytes and EOMES is predominantly expressed in NK
   cells. ZMIZ1 is known to regulate the activity of various transcription
   factors. ZMIZ1 and a set of genes with a correlated expression pattern
   are under-expressed in blood of MS patients ([47]15). We identified two
   gene clusters in the ZMIZ1-correlated gene set, one with high response
   to 1,25(OH)[2]D[3] and the other with high expression levels in THP-1
   cells, that are potentially affected by VSE2 regions and VSE3 regions,
   respectively.

   Our research shows an association between VDR super-enhancer regions
   and MS risk for the first time.

Materials and Methods

Next Generation Sequencing Data Selection

   We downloaded unstimulated and 1,25(OH)[2]D[3]-stimulated VDR ChIP-seq,
   PU.1 ChIP-seq, FAIRE-seq, and RNA-seq data in THP-1 cells, and other
   hormone/nuclear receptor (i.e., estrogen/ERα and dexamethasone/GR) NGS
   data (SRA format), from the NCBI Gene Expression Omnibus (GEO) database
   ([48]Table 1 and [49]Table S1). Then FASTQ files were converted from
   the SRA files with command “fastq-dump.”

Table 1.

   Control and 1,25(OH)[2]D[3] (D3)-stimulated ChIP-seq, FAIRE-seq, and
   RNA-seq data in THP-1 cells (VDR_1).
   Transcription factor Cell type Signal Treatment Organism Vehicle
   Hormone References