Epigenetic and transcriptional determinants of the human breast

Abstract

While significant effort has been dedicated to the characterization of epigenetic changes associated with prenatal differentiation, relatively little is known about the epigenetic changes that accompany post-natal differentiation where fully functional differentiated cell types with limited lifespans arise. Here we sought to address this gap by generating epigenomic and transcriptional profiles from primary human breast cell types isolated from disease-free human subjects. From these data we define a comprehensive human breast transcriptional network, including a set of myoepithelial- and luminal epithelial-specific intronic retention events. Intersection of epigenetic states with RNA expression from distinct breast epithelium lineages demonstrates that mCpG provides a stable record of exonic and intronic usage, whereas H3K36me3 is dynamic. We find a striking asymmetry in epigenomic reprogramming between luminal and myoepithelial cell types, with the genomes of luminal cells harbouring more than twice the number of hypomethylated enhancer elements compared with myoepithelial cells.

Figure 1. Differential expression and isoform analysis.

(a) Experimental overview (b) DE genes in myoepithelial and luminal (lum) epithelial cell types across three donors, luminal upregulated (red), myoepithelial (myo) upregulated (green). (c) Exon–intron junction mCpGs provide an inherited signature of exon expression. Average number of CpGs (black, bottom panel) and average mCpG levels (whole-genome bisulphite shotgun, 20 bp bins) at exon junctions +/− 200 bp in luminal (solid line with round dots) and myoepithelial (dashed line with triangles). Exons are divided into four groups namely: (1) exons expressed in both the cell types (exon reads per kilobase of transcript per million reads mapped (RPKM) >0.1 in luminal and myoepithelial RM084, purple); (2) luminal-specific exons (isoform exons expressed in luminal but not in myoepithelial, red); (3) myoepithelial-specific exons (isoform exons expressed in myoepithelial but not in luminal, green) and (4) exons not expressed in either cell types (all other exons, blue). A statistical significant difference was observed between the not expressed exons and all other groups, t-test P value <10⁻¹⁸. All other comparisons show weak or no statistically significant difference. (d) H3K36me3 density in exon bodies provides a transient record of exon expression. Average H3K36me3 signal levels for exons in expressed genes (gene RPKM >0.1) in luminal RM080 (red) and myoepithelial RM080 (green). Exons are broken down into four groups namely: (1) exons expressed in both the cell types (exon RPKM >0.1 in luminal and myoepithelial); (2) luminal-specific exons (isoform exons expressed in luminal but not in myoepithelial); (3) myoepithelial-specific exons (isoform exons expressed in myoepithelial but not in luminal); and (4) exons not expressed in either cell types (all other exons). Fold enrichment of average H3K36me3 signal levels within exons revealed an increase H3K36me3 signals in cell type-specific exons in corresponding cell populations. (e) Model of inherited and transient epigenetic exon marking. Exon boundary DNA methylation (black dot) and exon body H3K36me3 (red flag) marking of exons in luminal and myoepithelial cell populations.

Figure 2. Intron retention and mammary cell type-specific miRNAs and long noncoding RNAs.

(a) Upper panel—enumeration of expressed protein coding (pc) and noncoding (nc) genes, expressed introns and intron/exon ratio from hESCs (left) to luminal (lum) epithelial cells (right). Expression is defined by an RPKM >1. Lower panel—mRNA expression of U1 and U2 spliceosome subunits and SWI/SNF complex components in hESC derived CD56+ ectoderm cultured cells compared with RM084 luminal cells. (b) DNA methylation profile at intron boundaries. Average DNA methylation level profile at intron 5′ and 3′ +/− 200 bp with 20-bp bins in luminal RM066 (top panel) and myoepithelial RM045 (third panel) WGBS libraries. Average number of CpGs (second and fourth panel). Introns are divided into retained introns (red in luminal and green in myoepithelial (myo)) and not retained introns (blue) according to intron retention analysis in RM084. Differences in DNA methylation level across exon–intron boundaries are calculated by subtracting minimum DNA methylation level within introns (intron 5′+200 bp or 3′−200 bp, valley) from maximum DNA methylation level within exons (intron 5′−200 bp or 3′+200 bp, peak) and P values of t-test between retained introns and not retained introns are shown in DNA methylation panels. P values for CpG density panels are calculated by t-test on average No. of CpGs across the 400-bp boundaries between retained introns and not retained introns. (c) Visualization at UCSC genome browser of NOXA1 retained intron event (hg19 chr9:140327716–140327904—highlighted in blue) in luminal epithelial (red tracks) and myoepithelial cells (green tracks) across different assays: RNA-seq (RPKM values), WGBS-seq (CpG fractional methylation) and H3K36me3 ChIP-seq signal track. The `CG' track shows the location of CpG dinucleotides. (d) Boxplots show NOXA1 expression level (log10(RPKM)) in luminal epithelial (left) and myoepithelial (right) cell types. Validation of the loss of detectable NOXA1 protein in both breast cell types across RM071, RM066 and RM084 individuals with IHC compared to kidney used as a positive control. Scale bar: 50 μm. (e) Entropy heatmap of cell type-specific miRNAs across mammary cell types. (f) Entropy heatmap of cell type-specific lincRNAs (Supplementary Data 7) across mammary cell types. Fibr, fibroblast.

Figure 3. Regulatory asymmetry between myoepithelial and luminal cell types.

(a) Plot of the relative abundance of TF binding sites overlapping UMRs in luminal (lum) and myoepithelial (myo) cells reveals regulatory asymmetry between the breast cell types. Red line represents total UMR abundance. (b) Fraction of DE genes (upregulated in luminal: red, upregulated in myoepithelial: green) associated with luminal (bottom panel) and myoepithelial (top panel) UMRs. Left panel shows proximal UMRs (UMRs within transcriptional start site (TSS) +/− 2 kb), and right panel shows distal UMRs (TSS +/− 20 kb). (c) Fraction of DE genes (upregulated in luminal: red, upregulated in myoepithelial: green) associated with both luminal and myoepithelial UMRs. Left panel shows proximal UMRs (UMRs within TSS +/− 2 kb) and right panel shows distal UMRs (TSS +/− 20 kb). (d) Differential expression (upregulated in luminal: red, upregulated in myoepithelial: green) of proximal UMRs overlapping with binding sites of luminal enriched (FoxA1, Gata3 and Znf217), and myoepithelial enriched TFs (Egr1). The size and shade of the circles represents fold change between luminal and myoepithelial gene expression on the log2 scale. (e) RNA yield (microgram per million cells) from luminal (red) and myoepithelial (green) cells extracted from three individuals.