The Diverging Routes of BORIS and CTCF: An Interactomic and Phylogenomic Analysis

Abstract

The CCCTC-binding factor (CTCF) is multi-functional, ubiquitously expressed, and highly conserved from Drosophila to human. It has important roles in transcriptional insulation and the formation of a high-dimensional chromatin structure. CTCF has a paralog called "Brother of Regulator of Imprinted Sites" (BORIS) or "CTCF-like" (CTCFL). It binds DNA at sites similar to those of CTCF. However, the expression profiles of the two proteins are quite different. We investigated the evolutionary trajectories of the two proteins after the duplication event using a phylogenomic and interactomic approach. We find that CTCF has 52 direct interaction partners while CTCFL only has 19. Almost all interactors already existed before the emergence of CTCF and CTCFL. The unique secondary loss of CTCF from several nematodes is paralleled by a loss of two of its interactors, the polycomb repressive complex subunit SuZ12 and the multifunctional transcription factor TYY1. In contrast to earlier studies reporting the absence of BORIS from birds, we present evidence for a multigene synteny block containing CTCFL that is conserved in mammals, reptiles, and several species of birds, indicating that not the entire lineage of birds experienced a loss of CTCFL. Within this synteny block, BORIS and its genomic neighbors seem to be partitioned into two nested chromatin loops. The high expression of SPO11, RAE1, RBM38, and PMEPA1 in male tissues suggests a possible link between CTCFL, meiotic recombination, and fertility-associated phenotypes. Using the 65,700 exomes and the 1000 genomes data, we observed a higher number of intergenic, non-synonymous, and loss-of-function mutations in CTCFL than in CTCF, suggesting a reduced strength of purifying selection, perhaps due to less functional constraint.

Keywords: Amniotes; Bilateria; CTCF; chromatin loops; gene duplication; natural selection; polymorphism.

Figure 1

Structural comparison of CTCF and CTCFL. Numbered boxes indicate C2H2 zinc-finger domains, vertical arrows demarcate the position of introns, and red bars represent low complexity regions. Drawn to scale.

Figure 2

Protein interaction network of CTCF (A) and CTCFL (B): Colour (from orange to red) and node size denote increasing degree.

Figure 2

Protein interaction network of CTCF (A) and CTCFL (B): Colour (from orange to red) and node size denote increasing degree.

Figure 3

Lineage phylostratigraphy of CTCF/CTCFL interactors. Note that Apoikozoa is the common ancestor of animals and choanoflagellates and Holozoa is the common ancestor of Apoikozoa and Ichthyosporea (Mesomycetozoea). Multiple lineage assignments indicate that a particular ancestor cannot be inferred with certainty. Assignment to Opisthokonta indicates that the protein emerged in the ancestor of Opisthokonta or earlier. Green refers to common interactors, arrows on the margins indicate increasing evolutionary age relative to CTCF(L).

Figure 4

Maximum likelihood phylogeny of the CTCF gene family. The underlying alignment matrix consists of 40 protein sequences with 772 characters and 15.56% gaps or undetermined characters. The main aim of this tree is to show that CTCFs and CTCFLs from birds form different clusters. Fish CTCFs have atypical phylogenetic positions in the tree, as reflected by their low bootstrap values. Bootstrap values below 50 were omitted for clarity.

Figure 5

Genomic synteny blocks of CTCF/CTCFL: Boxes (transcript orientation) with identical colors correspond to orthologous genes. Encircled “A” in the grey frame corresponds to the inferred ancestral state (see Genomicus web site). Neo-synteny/Paleo-synteny refers to evolutionary younger/older orthologous clusters.

Figure 6

Hi-C map of mouse sperm cells at a 25 kb resolution showing the chromatin (sub) loop harboring SPO11, RBM38, and CTCFL, and a larger loop containing PCK1, ZBP1, and PMEA1. The region of interest (Chr2:172,970,000-173,224,000) is marked with a yellow frame (A) and zoomed in at a 5 kb resolution (B). Green circles mark off-diagonal interactions.

Figure 7

Histogram showing enrichment for mutations in CTCF(L) based on the 1000 genomes phase 3 data. The y-axis displays the ratio of the normalized number of mutations in CTCFL and CTCF (r = N_CTCFL/N_CTCF).