Locus-specific editing of histone modifications at endogenous enhancers

Abstract

Mammalian gene regulation is dependent on tissue-specific enhancers that can act across large distances to influence transcriptional activity. Mapping experiments have identified hundreds of thousands of putative enhancers whose functionality is supported by cell type-specific chromatin signatures and striking enrichments for disease-associated sequence variants. However, these studies did not address the in vivo functions of the putative elements or their chromatin states and did not determine which genes, if any, a given enhancer regulates. Here we present a strategy to investigate endogenous regulatory elements by selectively altering their chromatin state using programmable reagents. Transcription activator-like (TAL) effector repeat domains fused to the LSD1 histone demethylase efficiently remove enhancer-associated chromatin modifications from target loci, without affecting control regions. We find that inactivation of enhancer chromatin by these fusion proteins frequently causes downregulation of proximal genes, revealing enhancer target genes. Our study demonstrates the potential of epigenome editing tools to characterize an important class of functional genomic elements.

Figure 1

Programmable TALE-LSD1 fusion modulates chromatin at an endogenous enhancer. (a) Schematic depicts workflow for identification of nucleosome-free target sequence (black stripe) within enhancer (blue peaks of histone modification) and design of corresponding TAL effector fusion. TAL effector arrays comprising ~18 repeats (colored ovals) that each bind a single DNA base are fused to the LSD1 histone H3K4 demethylase. Upon transient transfection, we assayed for binding to the target site, induced chromatin changes and altered gene expression. (b) ChIP-seq signal tracks show H3K4me2, H3K27ac and TALE binding in K562 cells across a targeted enhancer in the SCL locus. Control tracks show anti-FLAG ChIP-seq signals in mCherry transfected cells and input chromatin. The target sequence of the TALE is indicated below. (c) ChIP-qPCR data show fold-change of H3K4me2 and H3K27ac enrichment in cells transfected with constructs encoding TALE-LSD1, the same TALE but lacking LSD1, or a ‘non-target’ TALE-LSD1 whose cognate sequence is not present in the human genome. Data are presented as log2 ratios normalized to mCherry plasmid transfected control (error bars represent ±s.e.m. n=4 biological replicates). (d) ChIP-seq tracks show H3K4me2 and H3K27ac signals across the target SCL locus for K562 cells transfected with TAL effector-LSD1 or control mCherry plasmid.

Figure 2

TALE-LSD1 fusions targeting 40 candidate enhancers in K562 cells. The FLASH assembly method was used to engineer 40 TALE-LSD1 fusions that recognize 17 – 20 base sequences in nucleosome-free regions of candidate enhancers. These reagents were transfected into K562 cells and evaluated by ChIP-qPCR. Bi-directional plot shows fold change of H3K4me2 (green, left) and H3K27ac (blue, right) at the target locus for each of the 40 fusions, which are ordered by strength of effect and labeled by their target genomic site. Most target sites were evaluated using two qPCR primer sets. Data are presented as log2 ratios normalized to mCherry plasmid transfected control (error bars represent ±s.e.m., n=3 biological replicates). The solid red lines define a 2-fold difference (log2 = −1). The dashed red line demarcates constructs that induce a 2-fold reduction in histone modification levels for two or more of the four values shown. Regulated genes for 9 tested fusions are shown at right (see text and Figure 3). The data indicate that TALE-LSD1 reagents provide a general means for modulating chromatin state at endogenous enhancers.

Figure 3

TALE-LSD1 fusions to endogenous enhancers affect proximal gene expression. (a) Nine TALE-LSD1 fusions that robustly alter chromatin state (see Figure 2) were evaluated for their effects on gene expression by RNA-seq (see Methods). For each of the nine fusions, a bar graph shows normalized gene expression values for the closest expressed upstream and downstream genes. The red and pink bars indicate the gene expression value for two biological replicates in cells transfected with the corresponding ‘on-target’ TALE-LSD1 construct, and the black bars indicate the mean expression in cells transfected with control ‘off-target’ TALE constructs (error bars for the “Control” represent s.e.m, n=20 non-target libraries, see Methods, * indicates p <0.05 using an unpaired t-test). (b) ChIP-seq tracks show H3K4me2 and H3K27ac signals across the Zfpm2 locus. TAL effector-LSD1 fusions were designed to target candidate enhancers (black bars) in the first intron. (c) Bar graph shows relative ZFPM2 expression in K562 cells transfected with the indicated combinations of TALE-LSD1 constructs. Error bars indicate ±s.e.m of 4 RT-qPCR measurements). The data suggest that these enhancers act redundantly in K562 cells to maintain ZFPM2 expression.