A domain-resolution map ofin vivoDNA binding reveals the regulatory consequences of somatic mutations in zinc finger transcription factors

Abstract

Multi-zinc finger proteins are the largest class of human transcription factors with integral roles in genome regulation and function. Often using a subset of their zinc finger domains, these proteins bind to diverse DNA sequences that make up the majority of the human regulatory lexicon. However, the molecular code that underlies the interaction of the zinc finger with DNA is incompletely understood, and for most multi-zinc finger proteins the zinc finger subset that is responsible for in vivo DNA binding is unknown. Here, we present a computational model, derived from a compendium of in vivo and in vitro binding preferences, that captures context-specific binding of zinc fingers to DNA. This model can predict the binding specificity of each zinc finger in its natural context more accurately than existing models, and together with molecular dynamics analyses reveals new structural aspects of DNA recognition by zinc finger proteins, including novel amino acid residues that contribute to sequence specificity. Furthermore, by combining this computational model with in vivo binding data, we identify the sequence preference of each protein and the zinc finger subset that is responsible for in vivo DNA binding, providing DNA binding maps and the associated domains for ~30% of all human multi-zinc finger proteins. We show that compared to non-DNA-binding zinc finger domains, DNA-binding zinc fingers are under stronger selective pressure across species and are depleted of genetic variants in the human population. Finally, we propose that a combination of context as well as intrinsic zinc finger features determines the ability to bind to DNA in vivo.

Neural network modeling of differential binding between wild-type and mutant CTCF reveals putative binding preferences for zinc fingers 1-2

“…We applied this approach to a dataset with ChIP-seq from wild-type and mutated CTCF and, in addition to identifying the known motifs of CTCF ZFs 3-11, we discovered a putative interaction between CTCF ZF 1 and a novel downstream GAGCCA motif that as well as a putative weak interaction between CTCF ZF 2 and an ATT motif connecting the core and discovered downstream motifs ( Figure 4 ). Our discovered downstream motif is supported by in vitro studies of CTCF from HT-SELEX [31] and EMSA [36] and in vitro ZF studies from B1H assays [33, 35], the weak putative motif for ZF 2 was supported by CTCF HT-SELEX data, and the discovered downstream motif occurs more frequently in CTCF ChIP-seq peaks that are not bound by CTCF-s than by those that are. While, to fully demonstrate the validity of our findings, we would need to experimentally test if ZFs 1-2 can interact with these sequences using an assay like EMSA, our combination of existing and novel approaches for using additional datasets to support our findings provide a foundation for computationally following up on potential motif discoveries.…”

“…We obtained additional evidence that our discovered downstream motif interacts with CTCF. The most non-degenerate part of this motif (GAG) is almost identical to the computationally predicted motif for ZF 1 according to multiple models that were trained on in vitro ZF binding data from B1H assays ( Figure 3a, Supplemental Figure 6 ), suggesting that ZF 1 interacts with this downstream motif [33–35]. In addition, a recent study showed that the upstream four nucleotides of this downstream motif are found at CTCF sites in the mouse IgH locus; this study did EMSA on multiple variants of the CTCF motif including two variants containing these upstream four nucleotides and found that CTCF was able to bind both variants [36].…”

“…These models were trained on in vitro data measuring the binding specificities of individual ZFs; they take ZF amino acid sequences as input and output a predicted motif. First, we used RCADE2’s RC.sh to predict the motif for CTCF [35]. To explore alternative methods, we also put the sequences of CTCF’s ZFs into the “Predict PWMs” function of the “Interactive PWM Predictor” [33, 34].…”

Rewiring the specificity of extracytoplasmic function sigma factors

“…To do so requires both the ability to predict the sequence specificity of a novel regulator and the ability to determine significant interactions. Because of the difficulty of meeting both of these challenges, previous work has focused primarily on engineering modular regulatory proteins such as zinc-finger (ZFs) and transcription activator-like (TALs) rather than on de novo prediction of biological targets, a problem that remains largely unsolved ( 46 , 47 ). In this work, we solve the DNA-specificity code of ECF σs and build a computational pipeline that enables us to use these rules to determine statistically significant putative regulons for ∼67% of bacterial ECF σs.…”

Rewiring the specificity of extra-cytoplasmic function sigma factors

“…Because of the difficulty of meeting both of these challenges, previous work has focused primarily on engineering modular regulatory proteins such as zinc-finger (ZFs) and transcription activator-like (TALs) rather than on de novo prediction of biological targets, a problem that remains largely unsolved. For example, despite decades of study on ZFs structure and specificity, the regulatory role of a majority of the 700+ ZFs in the human genome remain unknown (Dogan et al 2020;Zuo et al 2019). In this manuscript, we both solve the DNAspecificity code of ECF σ s, a major class of bacterial regulators, and build a computational pipeline that enables us to use these rules to determine statistically significant regulons for ~67% of bacterial ECF σ s. The resulting dataset is the first of its kind: a comprehensive look at the role of a broadly distributed regulatory family.…”