Evolution of transcription factor binding in metazoans - mechanisms and functional implications

Abstract

Differences in transcription factor binding can contribute to organismal evolution by altering downstream gene expression programmes. Genome-wide studies in Drosophila melanogaster and mammals have revealed common quantitative and combinatorial properties of in vivo DNA binding, as well as marked differences in the rate and mechanisms of evolution of transcription factor binding in metazoans. Here, we review the recently discovered rapid 're-wiring' of in vivo transcription factor binding between related metazoan species and summarize general principles underlying the observed patterns of evolution. We then consider what might explain the differences in genome evolution between metazoan phyla and outline the conceptual and technological challenges facing this research field.

Figure 1. Insects and mammals have dramatically different densities of conserved elements

The central panel (white background) compares sequence constraint (Conserved elements tracks) in a 100 kb window around the tgo gene in Drosophila melanogaster (top) and the homologous ARNT gene in Homo sapiens (bottom); the difference in constraint density in this window is representative of whole-genome differences. A higher fraction of the fruit fly genome is conserved (across fifteen insect genomes) compared to the human genome (across 33 placental mammals): at the whole-genome level, 37-53% of the D. melanogaster genome lies in conserved elements, compared to 3-8% of the human genome. Grey background panels show gene annotations in these regions (Ensembl tracks). For the gene-dense region in Drosophila, forward strand genes are on the top track and reverse strand genes on the bottom. Figure adapted from Ensembl Genome Browser, with conserved elements from phastCons/UCSC. See main text for further discussion.

Figure 2. Genome-wide TF binding profiling in Drosophila and Mammals

Percentage binding overlaps are shown for the developmental TF Twist in whole embryos from divergent Drosophila species (A) and for the tissue-specific TF CEBPA in livers from mammalian species (B). In B, the main graph shows overlaps for 6 vertebrate species, while the inset data is for four mouse species and rat. In all cases, species are ordered by their evolutionary relationships as shown in the phylogenetic trees below each graph. Species in bold were used as the reference genome for comparison of the corresponding ChIP-Seq data. Where name abbreviations are used in the main graph, full species names are shown in the phylogenetic trees below.

Figure 3. Sources of metazoan TF binding divergence

A. Random genetic drift. Point mutations, indels and genomic rearrangements can lead to binding events from non-bound sequences in the last common ancestor. This mechanism is most efficient for transcription factors with short binding sequences, such as CEBPA (binding motif logo shown on the left). From top to bottom, the examples in the diagram exemplify the birth of CEBPA binding events from the ancestor sequence by a point mutation, an insertion, or a genomic rearrangement with a different chromosome. B. Repetitive element expansions. Expansion of repetitive sequences carrying binding motifs by transposable elements can give rise to numerous binding events across mammalian genomes. This mechanism is especially relevant for transcriptional regulators such as CTCF, whose long binding sequence cannot easily arise by genetic drift. The diagram depicts birth of multiple CTCF binding sites through expansion of SINE transposable elements. The central inset contains a partial B2 element sequence harbouring a high-affinity CTCF binding event. C. Capture of ancient repeat events. In contrast to B, some repetitive elements contain low affinity binding motifs that differ in a few key mutations from high-affinity binding sequences. Once expanded throughout the genome by transposable elements, these binding sequences can easily mutate to high-affinity binding events by genetic drift. This mechanism is exemplified in the diagram for the transcriptional repressor NRSF. The hERV family of transposons contains low-affinity, non-binding motifs for NRSF that can be exapted as high-affinity binding sites upon a few key mutations.