Mechanisms of transcription factor evolution in Metazoa

Abstract

Transcriptions factors (TFs) are pivotal for the regulation of virtually all cellular processes, including growth and development. Expansions of TF families are causally linked to increases in organismal complexity. Here we study the evolutionary dynamics, genetic causes and functional implications of the five largest metazoan TF families. We find that family expansions dominate across the whole metazoan tree; however, some branches experience exceptional family-specific accelerated expansions. Additionally, we find that such expansions are often predated by modular domain rearrangements, which spur the expansion of a new sub-family by separating it from the rest of the TF family in terms of protein-protein interactions. This separation allows for radical shifts in the functional spectrum of a duplicated TF. We also find functional differentiation inside TF sub-families as changes in expression specificity. Furthermore, accelerated family expansions are facilitated by repeats of sequence motifs such as C2H2 zinc fingers. We quantify whole genome duplications and single gene duplications as sources of TF family expansions, implying that some, but not all, TF duplicates are preferentially retained. We conclude that trans-regulatory changes (domain rearrangements) are instrumental for fundamental functional innovations, that cis-regulatory changes (affecting expression) accomplish wide-spread fine tuning and both jointly contribute to the functional diversification of TFs.

Figure 1.

Stacked bar plots depicting TF family sizes in analyzed species. The left-hand side of the graph shows a phylogenetic tree of the analyzed species. WGD events are denoted with a ‘2x’-symbol. Branches with accelerated gene gain rates are highlighted with stars. The stars are colored according to the TF family with accelerated gain rate. Time scale is approximate and largely based on (41). See ‘Materials and Methods’ section for more details.

Figure 2.

Relationship between domains and number of genes per TF family for all analyzed species. Linear regression lines are shown for each TF family. (A) Number of DACs (different domain arrangements) per TF family plotted against number of genes in the respective TF family. Full scale graph shows a log–log plot, inset shows linear axis. Each of the points represents one species. (B) Number of unique domains per TF family plotted against number of genes in the respective TF family. Each point represents one species.

Figure 3.

Wordclouds of the GO terms found to be enriched in the DACs of the bHLH TF family in human. For each DAC a pictogram of the domain arrangement is shown. Each GO term is scaled according to the P-value found in the enrichment test (smaller P-values mean bigger font size).