Genome Science Institute’s Seventh Annual Educational Symposium
Boston, MA, November 2015
Chromatin state patterns and sequence motif analysis at sex-biased
transcription factor binding sites provide insights into the hierarchical
regulation of sex-biased gene expression in mammalian liver
Gracia M. Bonilla and David J. Waxman
Department of Biology and Graduate Program in Bioinformatics, Boston University
Widespread sex-biased patterns of gene expression in mammalian liver affect numerous
biological processes, leading to sex differences in metabolism and disease risk. The sexdifferential stimulation of hepatocytes by male versus female plasma patterns of growth
hormone (GH) dictates liver sex differences by regulation at the transcriptional level via the
combinatorial interactions of several GH-responsive liver transcription factors (TFs),
including HNF6 and STAT5, whose sex-biased binding to liver chromatin correlates strongly
with the sex-biased expression of neighboring genes. We hypothesize that the sex-biased
binding by these factors is facilitated by the binding of co-regulators that function in specific
chromatin contexts that are distinguishable by features related to sex-biased chromatin
accessibility and histone marks. These co-regulators may function in a hierarchical manner,
with some factors involved in the opening of condensed chromatin, and other factors
interacting collaboratively at pre-existing accessible chromatin sites. To investigate this
proposal, we clustered sex-biased TF binding sites according to the sex bias of their
chromatin state, previously established for male and female mouse liver using genome-wide
DHS analysis and six histone marks. We identified sex-biased HNF6 and STAT5 binding
sites in proximity to sex-biased genes, located in regions in the same chromatin state, as
well as regions in different chromatin states between male and female liver. These include
(1) female-biased TF sites in enhancer chromatin states in female liver and inactive states
in male liver, and correspondingly for male-biased TF binding sites, indicating that sexdifferential chromatin states can be a determinant of sex-differential binding; and (2)
female-biased TF binding at sites that do not show sex-differences in chromatin accessibility
and are in the same active chromatin state in male and female liver, and correspondingly
for male-biased TF binding sites, indicating that factors other than accessibility to chromatin
and local chromatin state may confer sex bias in TF binding. Sequence motifs differentially
enriched near these TF binding sites identified potential cofactors of HNF6 and STAT5, which
could contribute to the sex bias of TF binding at genomic regions in a sex-independent
chromatin state, and that may do so in a hierarchical manner. This analysis identified
factors that have been associated with chromatin opening, including GATA2 and TAL1 as
potential collaborators of STAT5 in male liver, but not in female liver, and the factors
FOXL1, and PBX1, which are associated with binding at pre-existing accessible chromatin
states. These findings provide insights into the hierarchy of TF binding interactions that
mediate sex-specific gene regulation in mammalian liver. In doing these analyses, we
developed a tool that automates the clustering of genomic regions by the patterns of
differential regulation of their local chromatin environment, and that facilitates the
integration of these regions with genome-wide maps of chromatin accessibility, histone
modifications, transcription factor binding motifs, and gene expression. This tool can be
used generally to integrate multiple genomic features regulated by a common underlying
mechanism and study their function. Supported in part by NIH grant R01-DK33765 (to
DJW).