Quantifying the arms race between LINE-1 and KRAB-zinc finger genes through TECookbook

Abstract

To defend against the invasion of transposons, hundreds of KRAB-zinc finger genes (ZNFs) evolved to recognize and silence various repeat families specifically. However, most repeat elements reside in the human genome with high copy numbers, making the ChIP-seq reads of ZNFs targeting these repeats predominantly multi-mapping reads. This complicates downstream data analysis and signal quantification. To better visualize and quantify the arms race between transposons and ZNFs, the R package TECookbook has been developed to lift ChIP-seq data into reference repeat coordinates with proper normalization and extract all putative ZNF binding sites from defined loci of reference repeats for downstream analysis. In conjunction with specificity profiles derived from in vitro Spec-seq data, human ZNF10 has been found to bind to a conserved ORF2 locus of selected LINE-1 subfamilies. This provides insight into how LINE-1 evaded capture at least twice and was subsequently recaptured by ZNF10 during evolutionary history. Through similar analyses, ZNF382 and ZNF248 were shown to be broad-spectrum LINE-1 binders. Overall, this work establishes a general analysis workflow to decipher the arms race between ZNFs and transposons through nucleotide substitutions rather than structural variations, particularly in the protein-coding region of transposons.

Figure 1.

(A) Overall schematics for liftIn and liftOut operations of TECookbook; (B) Conversion of alignment file from RepeatMasker output to liftOver chain block for liftIn operation; (C) Example of liftOut operation to extract all genomic sites at defined repeat locus;

Figure 2.

(A) ZNF10 ChIP-exo signals mapped onto LINE-1 ORF2 repeat coordinates; (B) PFM of L1-ZNF10 locus for each sub-group; (C) PEM derived from ZNF10 Spec-seq data; (D) Predicted energy distribution for all sites lifted out of each sub-group; (E) Arms race model between LINE-1 and ZNF10; (F) Mutagenesis effects at L1-ZNF10 locus (19).

Figure 3.

(A) ZNF382 ChIP-exo signals mapped onto LINE-1 ORF2 repeat coordinates; (B) PFM of L1-ZNF382 locus for each sub-group; (C) PEM derived from ZNF382 Spec-seq data; (D) Predicted energy distribution for all sites lifted out of each sub-group; (E) Arms race model between LINE-1 and ZNF382; (F) Mutagenesis effects at L1-ZNF382 locus (19).

Figure 4.

(A) The L1 ORF2 domain organization (adapted from Adney et al.) and some known ZNFs that target LINE-1; (B) Sub-group specificity of each ZNF based on maximal Fold Change over Control of ChIP-seq/exo data mapped on repeat coordinates.