Biased Allele Expression and Aggression in Hybrid Honeybees may be Influenced by Inappropriate Nuclear-Cytoplasmic Signaling

Abstract

Hybrid effects are often exhibited asymmetrically between reciprocal families. One way this could happen is if silencing of one parent's allele occurs in one lineage but not the other, which could affect the phenotypes of the hybrids asymmetrically by silencing that allele in only one of the hybrid families. We have previously tested for allele-specific expression biases in hybrids of European and Africanized honeybees and we found that there was an asymmetric overabundance of genes showing a maternal bias in the family with a European mother. Here, we further analyze allelic bias in these hybrids to ascertain whether they may underlie previously described asymmetries in metabolism and aggression in similar hybrid families and we speculate on what mechanisms may produce this biased allele usage. We find that there are over 500 genes that have some form of biased allele usage and over 200 of these are biased toward the maternal allele but only in the family with European maternity, mirroring the pattern observed for aggression and metabolic rate. This asymmetrically biased set is enriched for genes in loci associated with aggressive behavior and also for mitochondrial-localizing proteins. It contains many genes that play important roles in metabolic regulation. Moreover we find genes relating to the piwi-interacting RNA (piRNA) pathway, which is involved in chromatin modifications and epigenetic regulation and may help explain the mechanism underlying this asymmetric allele use. Based on these findings and previous work investigating aggression and metabolism in bees, we propose a novel hypothesis; that the asymmetric pattern of biased allele usage in these hybrids is a result of inappropriate use of piRNA-mediated nuclear-cytoplasmic signaling that is normally used to modulate aggression in honeybees. This is the first report of widespread asymmetric effects on allelic expression in hybrids and may represent a novel mechanism for gene regulation.

Keywords: Africanized; Apis mellifera; PIWI; PIWI-interacting small RNAs; aggression; cytoplasmic incompatibility; hybrid incompatibility; parental effects.

FIGURE 1

Gene counts in bias categories. Average maternal/paternal bias of all genes in each bias category in each hybrid family (EA, European maternity; AE, Africanized maternity). Gray columns = European allele, Black = Africanized allele. The total number of genes in each category and the number falling in each QTL type are given in the columns on the right. Only genes falling into a single bias category across samples are included in counts for QTL types. ^∗Significantly more genes in this category are present within these QTL than expected by chance (Bonferroni corrected p-value = 0.005).

FIGURE 2

Sting response time of individuals. Individual bees (573 total) were given an electrical shock from a constant current stimulator and the time in seconds for them to sting a suede patch was recorded. Genotypes of reciprocal hybrids are given on the X-axis: Africanized maternity hybrids AC (Apis mellifera carnica father) and AL (A.m. ligustica father), and Africanized paternity hybrids CA (A.m. carnica mother) and LA (A.m. ligustica mother). Data presented is untransformed, letters designate significant differences.

FIGURE 3

Heat map of maternally biased gene clusters. (A) Gene cluster on chromosome 3, overlapping a QTL associated with stinging behavior. (B) Gene cluster on chromosome 12, overlapping a QTL associated with alarm pheromone production (isopentyl acetate). Position along the chromosome (in Mbp), relative allele usage of each reciprocal family within each sample, and OGS 3.2 gene ID is given for every tested gene within these clusters. Relative allele usage calculated as maternal read count/total read count. ^∗Statistically significant allelic bias. NT, not tested. N/A in gene ID column are transcripts that had no clear match to a protein coding gene.

FIGURE 4

Proposed model of epigenetic regulation of aggression through piRNAs. (A) Divergent selective pressure for aggressive reproductive offspring on males and females creates genomic conflict, but this conflict is balanced by the need for appropriate colony-level aggression. Fathers attempt to increase aggression (through a shift toward aerobic glycolysis, AG) by silencing genes using sperm-loaded piRNAs. Mothers negate this silencing through genomic licensing using oocyte-loaded piRNAs. (B) AHB and EHB differ in aggression due to both genetic effects and a greater epigenetic potential for aggression in AHB, selected for in either their native or introduced range.