Effects of sequence variation on differential allelic transcription factor occupancy and gene expression

· by Gao Wang · Read in about 3 min · (593 words) ·

By Reddy T. et al (2012) Genome Research.

First, for audience not very familiar with the context:

  • Transcription factors: Bind specific DNA motif & controls gene expression
  • Chromatin Immunoprecipitation Sequencing: Cartoon: ChIP-Seq with chips
  • Differential allelic :
    • Alleles are different versions of the same gene.
    • To focus on allelic imbalance means to focus on heterozygotes
  • Differential allelic TF occupancy (DATO):
    • preference to bind one allele over another, within the same cell type
  • Allele specific expression (ASE):
    • Explained in part as functional consequence of differential allelic TF binding
  • EP300: histone acetyltransferase p300, makes DNA more accessible to TF

This paper was published in 2012: when exome sequencing was just becoming widely adopted.


Variant discovery & annotation

  • variant –> disease (GWAS / exome sequencing)
  • variant –> expression –> disease (eQTL/tWAS)
  • variant –> TF occupancy imbalance –> expression & disease (this paper)

This paper studies impact of genetic variation on gene regulation. Specifically it aims to identify DATO sites and understand their properties wrt gene expression and disease


  • Variation in core TF binding motifs: small in fraction, large in effects wrt DATO
  • DATO sites significantly enriched disease variants particularly autoimmune disease


Measure allelic differences in expression in the same cells. RNA-Seq + Chip-Seq

  • why: RNA-seq only limits to heterozygous exonic variants


  • Genome-wide RNA-Seq
  • TF occupancy of RNA Pol2 of EP300 and of 24 sequences specific TFs
  • 1000 Genomes female LCL
  • Aligned sequence reads to both the maternal and paternal versions of the genome


TF hubs

  • TF cluster on the same alleles in regions of open chromatin
  • DATO evenly distributed across autosomes
  • 30% are in hubs (more than 5 TF according to Figure 1D?), compared to 15% non-DATO as hubs
    • Sites are cooperative
    • hubs are sensitive to variation
  • DATO in hubs often reside on the same allele (Figure 1B), a pattern non-exist in non-DATO
    • Existence of coordinated differential allelic gene regulation
    • Except for co-occupancy in a few cases such as EP300, these (novel) hubs are generally not explained by known PPI
  • Variants in hubs are more conserved (Figure 1D)

DATO variants

  • Strong enrichment of DATO variants within 50bp of maximal ChIP signal
  • Heterozygous variants in motif are 3 times more likely to occur in DATO sites
  • Variants in DNA binding motif (of strong effects) explains only 12% DATO

DATO and allele specific expression (ASE)

  • Using RNA-Seq alone, 9% genes with exonic heterozygous variants (4194 transcripts) have ASE
    • Effect sizes are small for autosomal genes
    • 4194 transcripts represents only 39% of transcripts in sample
  • Using RNA Pol2 ChIP-Seq, 6.3% (out of 10,353) with variants in RNA Pol2 sites have ASE
    • Cross clone variations: strong concordance in ASE on autosomes
    • These variants significantly but imperfectly explains variations in expression
  • Combined results: 910 genes in sample have ASE

Heritibility: DATO vs ASE

  • The stronger allele in DATO sites has greater ChIP-Seq signal in corresponding parent in 81% of the cases
    • Compared to 60% for ASE
  • DATO is more strongly determined by local sequences

ASE genes are lower expressed and more tissue specific

Highly expressed genes are more conserved: a closer link between gene expression and evolutionary conservation than previously been shown.

TF occupancy explains variation in gene expression

  • DATO are significantly closer to genes with ASE than without
    • 6.8% sites within 100kb of ASE genes has DATO
    • 3.9% sites within non-ASE genes has DATO
  • non-DATO no closer to genes with ASE
  • Strong positive DATO and ASE correlation observed within 100bp of TSS
    • i.e, highly predictive

Disease significance of DATO

  • 155 autosomal variants in TF binding sites associated with disease
  • 21 of them are DATO variants