Biased Gene Conversion and Its Impact on Genome Evolution

Among the various forces shaping genome evolution, biased gene conversion (BGC) is one of the most recently recognized and yet potentially most widespread. By systematically favoring the transmission of G and C alleles over A and T alleles at sites of heteroduplex DNA during homologous recombination, BGC acts as a directional evolutionary force that can drive allele frequency changes entirely independently of natural selection. Its genomic footprint is increasingly recognized across eukaryotic genomes, and its ability to mimic the signatures of positive selection presents significant challenges for molecular evolutionary analysis.

The Molecular Basis of GC-Bias

During homologous recombination, the two strands of homologous chromosomes form heteroduplex DNA — regions where strands from the two parental chromosomes are base-paired. At positions where the two parental chromosomes differ (heterozygous sites), the heteroduplex contains mismatches: G:T, A:C, or other mispairs. The mismatch repair (MMR) machinery, principally the MSH2-MSH6 complex (for G:T mismatches) and the MSH2-MSH3 complex (for larger insertion-deletion heterologies), recognizes and resolves these mismatches. However, the repair is not symmetric: G:T mismatches in heteroduplex DNA are repaired to G:C pairs significantly more often than to A:T pairs, reflecting the intrinsic bias in MMR and the thermodynamic preference for G-C base pair stability.

This repair bias means that when an individual is heterozygous for an A and a G allele, and a gene conversion event covers that site, the G allele is transmitted at frequency greater than 0.5 to the progeny — providing a transmission advantage for G (and by similar logic, C) alleles that is independent of any fitness effect those alleles may have.

Quantifying BGC Strength

The per-site strength of BGC is typically expressed as a parameter analogous to selection coefficients in population genetics: the BGC coefficient b, where b > 0 indicates GC bias and b = 0 indicates neutral mismatch repair. Empirical estimates from human sperm genotyping and pedigree studies suggest that the per-site BGC coefficient at recombination hotspots is in the range of 0.01–0.1 per meiosis at converted sites — modest per event, but cumulatively significant over evolutionary time scales given that conversion events affect many sites per generation across the genome.

BGC and Isochore Structure

The vertebrate genome is organized into large-scale domains of homogeneous GC content called isochores — GC-rich (H isochores) and GC-poor (L isochores) regions of roughly 300 kb or more. The origin of isochore structure is a longstanding question in genome biology. BGC provides a compelling mechanistic explanation: GC-rich isochores correspond to chromosomal regions with historically high recombination rates. Over evolutionary time, BGC at high-frequency conversion sites in these regions has enriched them for GC base pairs, explaining the correlation between GC content and recombination rate observed across vertebrate chromosomes. Genome-wide reductions in recombination rate — as seen in regions of reduced chiasma formation, near centromeres, and in sex-limited regions — correlate with lower GC content, consistent with reduced BGC.

BGC Mimicking Positive Selection

The most concerning implication of BGC for evolutionary genomics is its ability to mimic the population genetic signatures of positive selection. Under positive selection, a beneficial allele sweeps through a population, reducing genetic diversity in surrounding genomic regions (a "selective sweep") and elevating its dN/dS ratio. Under BGC, a G or C allele at a converted site will also increase in frequency through gene conversion-mediated transmission bias, reducing local heterozygosity and potentially generating an elevated ratio of replacement to synonymous changes if the affected sites fall in coding sequences. This has led to concern that many reported signals of positive selection in genome-wide scans — particularly in regions of high recombination — may reflect BGC rather than genuine adaptive evolution. Developing statistical methods that jointly model BGC and natural selection is an active area of evolutionary genomics research. For an overview of how gene conversion is detected in the first place, revisit our article on detecting gene conversion in genomic studies.

For more information, visit our homepage or our resources section.