Signatures of rapid change in PRDM9 binding targets are evident in archaic humans

• Main Authors: Tumian, Afidalina, Davies, Robert W., Myers, Simon
• Format: Conference or Workshop Item
• Language: English
• Published: 2015
• Subjects: Q Science (General)
• Online Access: http://irep.iium.edu.my/47189/1/47189.pdf; http://irep.iium.edu.my/47189/; http://www.hugo-international.org/Resources/Documents/HGM2015_Final%20Programme%20and%20Abstract%20Book.pdf

Objectives: Multiple lines of evidence suggest that the rapidly evolving zinc-finger (ZF) protein, PRDM9, is responsible for initiating much or all of recombination in human. PRDM9 shows extreme variation in both the number and sequence of its ZFs, between species and amongst individuals, across mammals. The rapid evolution of the PRDM9 ZF array may be a response to escape a self-destructive drive called biased gene conversion (BGC), which can cause preferential transmission of hotspot disrupting alleles, and leading to erosion of vital recombination sites in the genome through time. Using the recent available Neanderthal and Denisovan high quality genomes, we developed statistical methods that identify the locations where meiotic recombination could have occurred in the past. This is achieved by looking for short words that have undergone rapid losses or gains in each lineage. Methods: In particular, the statistical framework involves an enumerative approach that scans genomewide archaic human genomes which have been mapped to a 6-way primate alignment, exhaustively catalogues all short exact motifs, and identifies statistically highly evolved words. Results: Across all species, remarkably, we observed definitive evidence that on each lineage, specific genomic "words" have been rapidly removed from the genome, at rates highly inconsistent with neutral mutation changes. It seemed possible that they might have been past binding targets of PRDM9, with rapid evolution of the PRDM9 zinc finger array explaining their lineage-specificity, and differences in PRDM9 binding explaining the stronger signals in specific genomic regions. As predicted by BGC, we saw an overwhelming support for bias towards acceleration of motif loss rather than motif gain. In human lineage, it is shown that the method almost successfully identifies the current human hotspot motif, CCNCCNTNNCCNC. Conclusion: It is suggestive that we are able to see positions of ancient hotspots, and could even make maps of ancient hotspot positions, by looking at the distribution of motif losses along the genomes. The findings could further shed light on the dynamic turnover of PRDM9 binding sites, and help understand the details of how recombination has shaped over genomes.