In vivo binding of PRDM9 reveals interactions with noncanonical genomic sites

PRDM9 binds to two distinct classes of sites in the genome. (A) Schematic representation of the PRDM9^Dom2 and PRDM9^Cst proteins in the B6 and RJ2 mouse strains, respectively. The asterisk indicates the only amino acid substitution outside the zinc finger array due to a nonsynonymous single-nucleotide polymorphism. (B) Venn diagram showing the overlap of PRDM9 ChIP-sequencing peaks in B6, RJ2, and B6 Prdm9^KO mice. (C) Number of strain specific and common peaks retrieved for PRDM9, DMC1, and H3K4me3 ChIP-seq experiments. (D) Classification of PRDM9 peaks in subclasses with (+) or without (−) enrichment for DMC1 and/or H3K4me3 in B6 and RJ2 mice. Question marks indicate situations where a PRDM9-dependent signal (if present) could not be detected because of a common signal also found in the other strain. Class 1 contains PRDM9 peaks that are enriched in DMC1 and/or strain-specific (PRDM9-dependent) H3K4me3. Class 2 sites are negative for DMC1 but can overlap peaks that are enriched for H3K4me3 in both mouse strains. (E) Read distribution from PRDM9, H3K4me3, and DMC1 ChIP-seq experiments at representative B6 class 1, 2A, and 2B sites in B6 and RJ2 mice. Read distribution was calculated from pooled replicates, in 1-bp windows, and normalized by library size and input. (F) Average read enrichment (reads per kilobase pair per million mapped reads [RPKM]) of PRDM9, H3K4me3, and DMC1 in B6 and RJ2 mice centered on class 1, 2A, and 2B sites in each strain.