Signatures of selection among sex-determining alleles of the honey bee

Abstract

Patterns of DNA polymorphisms are a primary tool for dissecting signatures of selection; however, the underlying selective forces are poorly understood for most genes. A classical example of diversifying selection is the complementary sex-determining locus that is found in the very large insect order Hymenoptera (bees, wasps, ants, and sawflies). The gene responsible for sex determination, the complementary sex determiner (csd), has been most recently identified in the honey bee. Females are heterozygous at this locus. Males result when there is only one functional allele present, as a result of either homozygosity (fertilized eggs) or, more commonly, hemizygosity (unfertilized eggs). The homozygotes, diploid males, do not reproduce and have zero fitness, which implies positive selection in favor of rare alleles. Large differences in csd cDNA sequences within and between four populations were found that fall into two major groups, types I and II. Type I consists of several allelic lineages that were maintained over an extended period, an indication of balancing selection. Diversifying selection has operated on several confined parts of the protein, as shown by an excess of nonsynonymous differences. Elevated sequence differences indicate another selected part near a repeat region. These findings have general implications about the understanding of both the function of the multiallelic mechanism and the adaptive processes on the level of nucleotide sequences. Moreover, the first csd sequence data are a notable basis for the avoidance of diploid males in bee selection programs by allele-assisted breeding.

Fig. 2.

Amino acid alignment of the hypervariable region with a variable number of (N)_1–4/Y repeats that were excluded in the evolutionary sequence analysis. Several gaps must be introduced to maximize homology. Flanking amino acids shown in bold are conserved. Amino acid sequence among replicate variants (as shown in Fig. 1) is the same, except for an amino acid difference for replicate D2-38 (position 390, Y → H). The numbers above the alignment represent the position in the overall sequence alignment, including gaps.

Fig. 3.

Evolutionary tree of csd type I/II alleles. The numbers of nonsynonymous differences (b_N) and synonymous differences (b_S) per site and for each branch in the genealogy are presented above each branch as b_N × 1,000/b_S × 1,000 with a transition/transversion ratio of R = 2. The statistical significance was tested with a one-tailed Z test (P < 0.05) and is indicated by asterisks. The tree is unrooted, and its topology differs from that in Fig. 1 because of the omission of replicate variants.

Fig. 1.

csd genealogy based on neighbor-joining analysis of genetic differences obtained from the deduced amino acid sequence, excluding ambiguous sites in which gaps have to be introduced in the alignment. The sequences fall into two major branches, types I and II. Fifteen allelic lineages were found; other sequences represent replicate variants of the same lineage, as indicated by bushes at the end of some tips (see text for further details). A genealogy of nearly identical topology was obtained when a hypervariable repeat region (rich in NY) was included. Numbers to the left of allele and replicate numbers indicate bootstrap percentages (of 1,000 resamplings) in support of each note. Bootstrap values <90% were not included. The scale bar indicates amino acid differences per site. Geographical origin is indicated by letter: D, Davis; B, Berlin; S, Stellenbosch; and A, Ribeirão Preto.

Fig. 4.

Mean pairwise diversity plots of K_s and K_a values, which were constructed for windows of 50 bases sliding along the csd sequence alignment. Only sequences from separate allelic lineages were used in the analysis, which did not include the replicate variants. (A) Average of these statistics for all pairwise comparisons of type I lineages. The arrow indicates the position of the hypervariable repeat region (see Fig. 2) not included in this analysis. (B) The comparison between types I and II. Filled boxes indicate base positions at the middle of the window (step size of 10 bases) in which nonsynonymous differences significantly exceed synonymous differences (one-tailed Z test, P < 0.05). Striped boxes indicate base positions in the sequence in which the reverse significant condition is found (K_s > K_a, P < 0.05).