Mechanisms of Binding Specificity among bHLH Transcription Factors

Abstract

The transcriptome of every cell is orchestrated by the complex network of interaction between transcription factors (TFs) and their binding sites on DNA. Disruption of this network can result in many forms of organism malfunction but also can be the substrate of positive natural selection. However, understanding the specific determinants of each of these individual TF-DNA interactions is a challenging task as it requires integrating the multiple possible mechanisms by which a given TF ends up interacting with a specific genomic region. These mechanisms include DNA motif preferences, which can be determined by nucleotide sequence but also by DNA's shape; post-translational modifications of the TF, such as phosphorylation; and dimerization partners and co-factors, which can mediate multiple forms of direct or indirect cooperative binding. Binding can also be affected by epigenetic modifications of putative target regions, including DNA methylation and nucleosome occupancy. In this review, we describe how all these mechanisms have a role and crosstalk in one specific family of TFs, the basic helix-loop-helix (bHLH), with a very conserved DNA binding domain and a similar DNA preferred motif, the E-box. Here, we compile and discuss a rich catalog of strategies used by bHLH to acquire TF-specific genome-wide landscapes of binding sites.

Keywords: ChIP-seq; E-box; bHLH; co-factors; dimerization; pioneer factors; transcription factor binding sites.

Figure 1

Structure of Tcf3-Neurod1 heterodimer bound to a CATCTG E-box. The bHLH domains of Neurod1 and Tcf3 are shown in red and blue, respectively. Neurod1 binds the CAT half-site of the E-box in the forward strand (pink) and Tcf3 binds the CAG half-site in the reverse strand (green). This representation has been produced with VMD v1.9.4 from the published X-ray crystal PDB:2ql2 from Longo et al. [40].

Figure 2

Heatmap representing motif similarity values among bHLH TFs calculated in Lambert et al. [1] from data derived from high-throughput in vitro assays of bHLH homodimers. Hierarchical clustering identifies three groups with a preference for E-boxes containing CAC (cluster 1, red), CAT (cluster 2, blue) and CAG half-sites (cluster 3, green). Homodimers of clusters 1 and 2 bind symmetrical CAT-CAT or CAC-CAC E-boxes, while members of cluster 3 require the presence of CAG in at least one of the half-sites (this difference is indicated by *). A TF can appear in multiple clusters if it is represented by multiple annotated motifs, but when all of them belong to the same cluster, the TF is only shown once. bHLH classes determined by Atchley and Fitch [15] and Ledent et al. [16] are shown on the right column.

Figure 3

Representation of the phylogenetic relationships, alignment of the basic domain, and different classification systems of bHLH factors. The phylogenetic tree and the alignment were downloaded from the online database provided by Lambert et al. (http://humantfs.ccbr.utoronto.ca/dbdsTable.php?dbd=bHLH, accessed on 15 April 2021). The tree was inferred from the alignment of the whole bHLH domain, but here we only represent the basic domain as it contains the most relevant positions with respect to binding. Importantly, the tree does not imply true ancestral phylogenetic relationships among bHLH classes. The amino acids in the five positions that better separate the phylogenetic classes are colored, taking as a reference amino acids described by Atchley and Zhao [66], although we find some minor differences in those diagnostic amino acids, because they used bHLH sequences from multiple species, while we focused in human bHLH factors. In the right, different classification systems are displayed: the subfamily as annotated by Simionato et al. [17], the phylogenetic classes by Atchley et al. and Ledent et al. [15,16], the Murre classes based on both structural and functional criteria [13], the phylogenetic classes by Skinner et al. [63] inferred from the sequence of the whole protein, and finally, our clusters derived from in vitro binding affinity experiments. The boxes are colored in gray when no information about the classification was available for the corresponding gene in the corresponding original study.

Figure 4

Network representation of protein-protein interactions among bHLH TF catalogued in the STRING database. Only experimental/biochemical score was taken into account, from experiments of human and mouse proteins. When experimental data was available for both species, the highest score was considered. STRING score is reflected in the width of edges in the network, while the size of each node represents its degree centrality. The shape of each node indicates the bHLH classification by Atchley and Fitch [15] and Ledent et al. [16]. Nodes are colored by the motif similarity cluster derived from Figure 2. (The difference of CAG is indicated by *).

Figure 5

Summary measures of ChIP-seq experiments conducted on bHLH TF using data collected from GTRD (https://gtrd.biouml.org, accessed on 17 May 2021) (A) Number of GSE records for each available bHLH transcription factor (32 had no available data in GTRD, see Table S1. Boxplot shows the number of peaks detected by MACS2 for each transcription factor averaging over the sum of experiments with the same GSE record ID, variability of numbers could reflect technical or biological differences among conditions and replicates. (B) The number of studies conducted in each class of bHLH in humans and rodents. (C) Barplot showing the distribution of tissues evaluated by ChIP-seq experiments stratified by species and representing the contribution of each bHLH class.