An all-to-all approach to the identification of sequence-specific readers for epigenetic DNA modifications on cytosine

Epigenetic modifications of DNA play important roles in many biological processes. Identifying readers of these epigenetic marks is a critical step towards understanding the underlying mechanisms. Here, we present an all-to-all approach, dubbed Digital Affinity Profiling via Proximity Ligation (DAPPL), to simultaneously profile human TF-DNA interactions using mixtures of random DNA libraries carrying different epigenetic modifications (i.e., 5-methylcytosine, 5-hydroxymethylcytosine, 5-formylcytosine, and 5-carboxylcytosine) on CpG dinucleotides. Many proteins that recognize consensus sequences carrying these modifications in symmetric and/or hemi-modified forms were identified. We further demonstrated that the modifications in different sequence contexts could either enhance or suppress TF binding activity. Moreover, many modifications can affect TF binding specificity. Furthermore, symmetric modifications showed a stronger effect in either enhancing or suppressing TF-DNA interactions than hemi-modifications. Finally, in vivo evidence suggested that USF1 and USF2 might regulate transcription via hydroxymethylcytosine-binding activity in weak enhancers in human embryonic stem cells. Overall design: An all-to-all approach, dubbed Digital Affinity Profiling via Proximity Ligation (DAPPL) was developed to simultaneously profile human TF-DNA interactions using mixtures of barcoded TF proteins and random DNA libraries. First, in order to optimize the DAPPL approach, the 16-mer DNA library was applied to profiling ETS proteins binding consensus sequences. Second, Many TF readers of epigenetic modifications were identified using 1,239 barcoded TF/co-factors and a mixture of four DNA libraries carrying different epigenetic modifications (i.e., 5-methylcytosine, 5-hydroxymethylcytosine, 5-formylcytosine, and 5-carboxylcytosine) and one DNA library without any modification on CpG dinucleotides. Third, in vivo evidence suggested that USF1 and USF2 might regulate transcription via hydroxymethylcytosine-binding activity in weak enhancers in human embryonic stem cells. Various groups: This study included 1,543 human TF clones, which were randomly divided into eight groups (e.g., Group_1, Group_2, etc). Each group of TFs was subjected to the screening assay using symmetric or hemi-epigenetic modified DNA libraries respectively. Cyc1 and Cyc2: two DAPPL assays using one round selection and two rounds selection separately. Here, only the pilot experiments (two files named with DAPPL_ETS) were optimized under two conditions, one round selection and two round selections. All other DAPPL assay (16 files named with DAPPL_Group) were performed using one round selection only. Hemi_modification: only one strand of the dsDNA library used in DAPPL assay contains epigenetic modified cytosine. Symmetric_modification: both strands of the dsDNA library used in DAPPL assay contain epigenetic modified cytosine. All of the computer programs and scripts used are publicly available at https://github.com/HitTracy/DAPPL