Diversity and Complexity in DNA Recognition by Transcription Factors

Abstract

Sequence preferences of DNA-binding proteins are a primary mechanism by which cells interpret the genome. Despite these proteins' central importance in physiology, development, and evolution, comprehensive DNA-binding specificities have been determined experimentally for few proteins. Here, we used microarrays containing all 10-base-pair sequences to examine the binding specificities of 104 distinct mouse DNA-binding proteins representing 22 structural classes. Our results reveal a complex landscape of binding, with virtually every protein analyzed possessing unique preferences. Roughly half of the proteins each recognized multiple distinctly different sequence motifs, challenging our molecular understanding of how proteins interact with their DNA binding sites. This complexity in DNA recognition may be important in gene regulation and in evolution of transcriptional regulatory networks.The interactions between transcription factors (TFs) and their DNA binding sites are an integral part of the gene regulatory networks that control development, core cellular processes, and responses to environmental perturbations. However, only a handful of sequence-specific TFs have been characterized well enough to identify all the sequences that they can and, just as importantly, can not bind. Computational analysis of microarray readout of chromatin immunoprecipitation experiments (ChIP-chip) suggests extensive use of low affinity binding sites in yeast (1), and computational models of gene expression during fly embryonic development suggest that low affinity binding sites contribute as much as high affinity sites (2).The availability of TF binding data spanning the full affinity range would improve our understanding of the biophysical phenomena underlying protein-DNA recognition, and would improve accuracy in analyzing cis regulatory elements. Here we report the comprehensive determination of the DNA binding specificities of 104 known and predicted mouse TFs using the universal protein binding microarray (PBM) technology (3). These TFs represent 22 different DNA binding domain (DBD) structural classes that are the major DBD classes found in metazoan TFs.We created (4) N-terminal GST fusion constructs of the DBDs of 104 known and predicted mouse TFs (Fig. S1 and Table S1). Five of these proteins -Max, Bhlhb2, Gata3, Rfx3, and Sox7 -were also represented as full-length fusions to N-terminal GST, yielding a total set of 109 non-redundant proteins represented by 115 samples (5). Each protein was used in two PBM experiments (6,7) (Figs. S2, S3, S4 and Table S2). DNA binding site motifs initially were derived using the Seed-and-Wobble algorithm (3,8); Seed-and-Wobble first identifies the single 8-mer (ungapped or gapped) with the greatest PBM enrichment score (E-score) (3), and then systematically tests the relative preference of each nucleotide variant at each position both within and outside the seed (5). Later analyses incorporated additional motif finding algorithms, including RankMotif++ (9) and Kafal (5).Beyond simpl...

Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning

“…Example motif detectors learned by DeepBind models, along with known motifs from CISBP-RNA 22 (for RBPs) and JASPAR 30 (for transcription factors). A protein's motifs can collectively suggest putative RNA-and DNA-binding properties, as outlined 51 , such as variable-width gaps (HNRNPA1, Tp53), position interdependence (CTCF, NR4A2), and secondary motifs (PTBP1, Sox10, Pou2f2). Motifs learned from in vivo data (e.g., ChIP) can suggest potential co-factors (PRDM1/EBF1) as in Teytelman et al 12 .…”

Distinguishing Direct versus Indirect Transcription Factor-DNA Interactions

“…The approach described in this study is not restricted to yeast or to ChIP-chip data, but could be applied to the analysis of ChIPseq (Johnson et al 2007) or ChIP-PET (Wei et al 2006) data sets for TFs in other organisms, including metazoans. With the generation of diverse PBM data sets for hundreds of metazoan TFs (Berger et al 2008;Badis et al 2009;Grove et al 2009), this approach may not only distinguish direct versus indirect genomic TF binding events in vivo, but also suggest the identities of the interacting TFs.…”

Quantifying the effect of sequence variation on regulatory interactions

“…There are still many transcription factors with unknown motifs that are beyond our model, and that will give rise to false negatives. Recent experimental advances and high-throughput data, such a protein-binding arrays [Badis et al, 2009;Berger et al, 2006], are likely to alleviate this limitation in the near future and permit an even more comprehensive assessment of the effect of sequence variations. Large-scale binding screens also raise the hope that we will soon be able to model small differences in the binding preferences of structurally similar proteins, as well as binding differences of the same transcription factor in variable cellular conditions.…”