Evolution of the complementary sex-determination gene of honey bees: Balancing selection and trans-species polymorphisms

Abstract

The mechanism of sex determination varies substantively among evolutionary lineages. One important mode of genetic sex determination is haplodiploidy, which is used by ∼20% of all animal species, including >200,000 species of the entire insect order Hymenoptera. In the honey bee Apis mellifera, a hymenopteran model organism, females are heterozygous at the csd (complementary sex determination) locus, whereas males are hemizygous (from unfertilized eggs). Fertilized homozygotes develop into sterile males that are eaten before maturity. Because homozygotes have zero fitness and because common alleles are more likely than rare ones to form homozygotes, csd should be subject to strong overdominant selection and negative frequency-dependent selection. Under these selective forces, together known as balancing selection, csd is expected to exhibit a high degree of intraspecific polymorphism, with long-lived alleles that may be even older than the species. Here we sequence the csd genes as well as randomly selected neutral genomic regions from individuals of three closely related species, A. mellifera, Apis cerana, and Apis dorsata. The polymorphic level is approximately seven times higher in csd than in the neutral regions. Gene genealogies reveal trans-species polymorphisms at csd but not at any neutral regions. Consistent with the prediction of rare-allele advantage, nonsynonymous mutations are found to be positively selected in csd only in early stages after their appearances. Surprisingly, three different hypervariable repetitive regions in csd are present in the three species, suggesting variable mechanisms underlying allelic specificities. Our results provide a definitive demonstration of balancing selection acting at the honey bee csd gene, offer insights into the molecular determinants of csd allelic specificities, and help avoid homozygosity in bee breeding.

Similar but not the same: insights into the evolutionary history of paralogous sex-determining genes of the dwarf honey bee Apis florea

“…A putative csd ortholog has been reported in A. florea, showing exceptionally high-nucleotide diversity exceeding those described for any other Apis (Liu et al, 2011). However, their results are unexpected and may result from pseudogenic fragments with similarity to csd occurring in the genome, which has already given misleading assignments to putative csd alleles in other Apis species (Cho et al, 2006;Hasselmann et al, 2008b).…”

“…These genes can be altered through randomly occurring mutations, with major consequences for their subsequent evolution (for example, neo-functionalization and pseudogenization; Conant and Wolfe, 2008). Pseudogene sequences and non-coding genomic fragments of csd occurring in the A. mellifera genome, which were non-functional but still had high similarity to csd, were previously identified and interpreted as likely trans-specific alleles (Cho et al, 2006; for discussion see Hasselmann et al, 2008b). In our present study, none of the evolutionary young csd alleles provide indications for being trans-specific.…”

A worldwide survey of genome sequence variation provides insight into the evolutionary history of the honeybee Apis mellifera

“…We used six publicly available sequences (~1 kb each) from noncoding regions of the A. cerana genome to estimate the level of divergence between the A. mellifera reference sequence representing a single copy of six different regions sequenced by ref. 49. We used these sequences because the more diverged 75-bp A. cerana reads produced by our sequencing cannot be unambiguously mapped to locations in the A. mellifera genome.…”

“…We calculated Watterson's estimator (θ w ) of the population mutation rate (θ) per base using the number of SNPs segregating in each population 48 . We estimated N e in each sample from our estimates of θ w and an estimate of the mutation rate derived from divergence with A. cerana 49 . For a haplodiploid system, θ = 3N e µ, where N e is the effective population size and µ is the mutation rate per base.…”

Community annotation: Procedures, protocols, and supporting tools: Table 1.

“…Many of the findings were supported by laboratory experiments and developed into full research publications (Cho et al 2006;Collins et al 2006;Dearden et al 2006;Evans et al 2006;Forêt and The number of genes and gene models is summarized in Table 1. The number of gene models in the OGS increased from 10,157 to 10,314, despite 302 OGS models being dropped due to splits and merges.…”