Systematic dissection of genomic features determining transcription factor binding and enhancer function

Abstract

Enhancers regulate gene expression through the binding of sequence-specific transcription factors (TFs) to cognate motifs. Various features influence TF binding and enhancer function-including the chromatin state of the genomic locus, the affinities of the binding site, the activity of the bound TFs, and interactions among TFs. However, the precise nature and relative contributions of these features remain unclear. Here, we used massively parallel reporter assays (MPRAs) involving 32,115 natural and synthetic enhancers, together with high-throughput in vivo binding assays, to systematically dissect the contribution of each of these features to the binding and activity of genomic regulatory elements that contain motifs for PPARγ, a TF that serves as a key regulator of adipogenesis. We show that distinct sets of features govern PPARγ binding vs. enhancer activity. PPARγ binding is largely governed by the affinity of the specific motif site and higher-order features of the larger genomic locus, such as chromatin accessibility. In contrast, the enhancer activity of PPARγ binding sites depends on varying contributions from dozens of TFs in the immediate vicinity, including interactions between combinations of these TFs. Different pairs of motifs follow different interaction rules, including subadditive, additive, and superadditive interactions among specific classes of TFs, with both spatially constrained and flexible grammars. Our results provide a paradigm for the systematic characterization of the genomic features underlying regulatory elements, applicable to the design of synthetic regulatory elements or the interpretation of human genetic variation.

Keywords: gene regulation; systems biology; transcription factor binding.

Fig. 1.

In vitro PPARγ binding is determined by core motif affinity. (A and B) Overview of pooled reporter system. (A) Candidate sequences were cloned into plasmids upstream of a minimal promoter and barcoded luc2 ORF. (B) Plasmid pools were transfected into adipocytes and assayed for PPARγ binding by ChIP-seq (Lower) and for enhancer activity by RNA-seq (Upper). (C) Schematic of pool 1, containing 750 bound genomic PPARγ motif sites, 750 unbound genomic PPARγ motif sites, and these 1,500 sites with the core PPARγ motif disrupted. (D) Log₂-ratio of ChIP enrichment for each genomic sequence with an intact and disrupted central PPARγ motif. (E) PPARγ ChIP enrichments for bound and unbound genomic sequences with intact and disrupted core PPARγ motifs. (F) Schematic of pool 2. The core PPARγ motif from 25 bound genomic sites was swapped into each of the other 24 flanking sequences, yielding a matrix of 625 enhancer constructs. (G) ChIP enrichment for each core motif (columns) and flanking sequence (rows) in pool 2. Core motifs were arranged by affinity measured by MITOMI (Materials and Methods). (H) Fraction of genomic PPARγ motif sites bound by PPARγ, conditional on the H3K27ac ChIP enrichment score (47).

Fig. 2.

Elements in flanking sequence govern enhancer activity. (A, Left) Ratio of expression (log₂[RNA/DNA]) for each genomic sequence with an intact vs. disrupted central PPARγ motif. (Right) Expression corresponding to bound and unbound genomic sites with intact and disrupted core PPARγ motifs. (B) Expression driven by sequence constructs in pool 2, comprising 25 core PPARγ motifs (columns) swapped into 25 flanking sequences (rows). (C) Schematic of identification of TF motifs correlated with enhancer activity. For each TF motif, we calculated the correlation between motif counts and expression in pool 1. (D) Counts of 38 motifs were significantly correlated with expression (FDR < 0.01; red). These motifs are enriched or depleted around all 6,835 bound motif sites in the genome (blue). Arrows indicate motifs depicted in E. (E) Expression of candidate enhancers in pool 1, conditional on the number of occurrences of each motif.

Fig. 3.

Disrupting TF motifs affects enhancer activity. (A) Schematic of motif deletion pools. Pool 4 (block-mutated enhancers): for 25 bound genomic sites, we disrupted 10-bp blocks tiled every 5 bp across the sequence (Top) and swapped 20 bp blocks tiled every 5 bp across the sequence between bound (Middle) and unbound (Bottom) genomic sites, matched by the sequence of the central PPARγ motif. In each case, the central PPARγ motif was left intact. Pool 5 (motif-mutated enhancers): each occurrence of the 38 significantly correlated motifs were disrupted across 375 bound genomic sites. Pool 6 (substituted enhancers): motif sites for each of the 38 correlated motifs were substituted into 90 existing motif sites in bound genomic sites. Pool 7 (synthetic enhancers): motif sites for 15 of the positively correlated motifs were added individually (Left) and in pairs (Right) to three neutral templates in various configurations (see SI Appendix, Supplemental Methods). (B) Example of changes in expression caused by tiled mutations in a bound sequences (red, chr8:90491327–90491472) and unbound sequence (gray, chr14:57223369–57223514). Bars represent the log₂ ratio of the mutant and wild-type expression for the block centered at that position. (C, Right) Median change in expression due to mutations in each motif across 375 bound genomic sites (Fig. 3C). (Left) Correlation between change in counts and change in expression for each motif in the substituted enhancers. (D) Expression of synthetic enhancers containing multiple copies of one motif.

Fig. 4.

Interactions between motifs contribute substantially to enhancer activity. (A and B) Performance of linear model (A) and Lasso model (B) predicting expression levels based on motif counts in independent test dataset (pool 3). (C) Boxplots represent expression of sequences in pool 1 (synergistic plot, Left) or pool 5 (additive and inhibitory plots, Left and Center), conditioned on counts of the two motifs. (D) Modes of interaction between TFs. Pioneer factors (First Row) are required to open chromatin at enhancers in the genome, but do not contribute strongly to transcriptional activation. Some pairs of TF enhance each other’s activity, resulting in superadditive transcriptional output (Second Row). Other pairs of TFs function independently of each other, contributing additively to the transcriptional output (Third Row). Finally, some TFs mutually inhibit each other’s activity, resulting in subadditive transcriptional output (Fourth Row). Add, additive; Inhib, inhibitory; and Syn, synergistic.