Cause and consequence of bias in human recombination

Reciprocal crossing-over between homologues during meiosis is a prerequisite for their correct segregation into gametes. The process is initiated by programmed double-strand breaks (DSBs), though these outweigh the number of crossovers (COs) by at least 10-fold. The majority are instead repaired by non-reciprocal exchange resulting in non-crossovers (NCOs), thought to aid homologue pairing and to shape the distribution of COs. Secondarily, both these forms of recombination increase genetic diversity by generating new haplotypes and can directly contribute to genome evolution either by acting as a form of meiotic drive or by increasing GC content.

Induction of a DSB is followed by resection and invasion of one of the overhangs into a non-sister chromatid. Subsequent synthesis with repair of DNA mismatches leads to patches of gene conversion. Ordinarily, there is no net conversion bias as DSBs are equally likely to be induced on each homologue. However, if initiation arises more frequently on one and subsequently formed mismatches are preferentially corrected with information from the other intact homologue, then biased conversion will ensue and transmission distortion (TD) will be observed. We have previously shown through high-resolution human sperm DNA analyses that single SNP variants can act as initiation-suppressing alleles and lead to meiotic drive [1]. Similarly, we have recently shown the first direct evidence of GC-biased gene conversion (gBGC) in mammals as a result of biased repair of heteroduplex exclusively in NCOs [2]. The latter contrasts with yeast where gBGC occurs in CO only, indicating significant differences in repair mechanisms. At the population level, such biased processes may mimic the signature of selection.

The targeted sperm DNA approaches developed at Leicester have provided unprecedented insight into human hotspots, 1-2 kb intervals within which the majority of recombination occurs. Complementary approaches can provide a genome-wide and importantly sex balanced perspective. This project will adopt both targeted and genome-wide approaches to further our understanding of human recombination and its biases:

1. Extent & nature of NCO throughout the human genome. Targeted analysis by screening bulk sperm DNA indicates that the mean NCO conversion tract is likely to be ~300 bp [3], yet current attempts to study NCO at the genome level have used SNP array data with an average of 1 SNP/3 kb, suggesting that only a subset of events have been sampled and an enrichment for long tract lengths [4]. To circumvent this bias and to better establish NCO rates per bp/generation, conversion tract lengths and the genome-wide extent of gBGC, this project will use publically available whole genome sequences. We will make use of the 3-generation CEPH pedigree 1463 sequenced by both Complete Genomics and Illumina using different platforms. Use of a 3-generation pedigree will further validate de novo NCOs detected in the second generation if they are passed onto any of the 17 children in the third. The same bioinformatics approach may be applied to a partial 3-generation pedigree from the 1000 Genomes Project.

2. Direct test of meiotic drive opposing natural selection. In theory mechanistic biases are capable of opposing natural selection, i.e. maintaining deleterious alleles and increasing disease burden. This will be directly tested using sperm recombination assays for the human beta-defensin-126 locus. DEFB126 is a component of the glycocalyx, shielding sperm from the maternal immune system. A 2-bp deletion null allele is associated with reduced fertility, yet is common amongst world populations (20-80%) [5]. LD suggests the deletion lies at the centre of hotspot; if it acts in cis to suppress DSBs then it would be over-transmitted to recombinants. The strength of meiotic drive will be governed by the degree of TD and recombination frequency, both of which can be directly evaluated.

[1] Sarbajna et al. 2012 Hum Mol Genet 21(9):2029
[2] Odenthal-Hesse et al. 2014 PLoS Genet 10(2):e1004106
[3] Jeffreys & May 2004 Nat Genet 36:151
[4] Williams et al. 2015 eLife 4:e04637.
[5] Tollner et al. 2011 Science Trans Med, 3(92):92ra65