Brain regions and molecular pathways responding to food reward type and value in honey bees

The ability of honey bees to evaluate differences in food type and value is crucial for colony success, but these assessments are made by individuals who bring food to the hive, eating little, if any, of it themselves. We tested the hypothesis that responses to food type (pollen or nectar) and value involve different subsets of brain regions, and genes responsive to food. mRNA in situ hybridization of c‐jun revealed that brain regions responsive to differences in food type were mostly different from regions responsive to differences in food value, except those dorsal and lateral to the mushroom body calyces, which responded to all three. Transcriptomic profiles of the mushroom bodies generated by RNA sequencing gave the following results: (1) responses to differences in food type or value included a subset of molecular pathways involved in the response to food reward; (2) genes responsive to food reward, food type and food value were enriched for (the Gene Ontology categories) mitochondrial and endoplasmic reticulum activity; (3) genes responsive to only food and food type were enriched for regulation of transcription and translation; and (4) genes responsive to only food and food value were enriched for regulation of neuronal signaling. These results reveal how activities necessary for colony survival are channeled through the reward system of individual honey bees. Female forager bees were collected from colonies derived from naturally mated queens maintained according to standard methods at the University of Illinois Bee Research Facility (Urbana, IL). The colonies were typical of North American populations of Apis mellifera, which are hybrids of various European‐derived subspecies, mostly A. m. ligustica. Different individual colonies were used in each of the three behavioral experiments described below. Colonies for Experiments 1–3 were located in a large outdoor, screened enclosure (6 × 20 × 3 m3) to precisely control their access to food. Additional bees were collected from one colony to detect the presence of octopaminergic and dopaminergic neuronal activation. Bees from Experiments 2–3 were used for both RNAseq and mRNA in situ hybridization (ISH); bees from Experiment 1 were only used for RNAseq. One complication associated with this method is that the stimulus varied between dancers and non‐dancers; this absolute difference in reward may also be driving the differences in brain gene expression in addition to differences in relative reward assessment. An alternative would have been to sample bees that danced or did not dance for the same sucrose concentration (Barron et al. 2007b), but this would have led to concerns that inter‐individual differences in response may be related to a variety of factors and not just reward assessment. Of the two, we believe that the method used was preferable for capturing the transcriptomic response to differences in reward value. The low concentration feeder was presented first on each collection day. Bees arriving at the feeder were marked on the abdomen with paint (Testors Corporation, Rockford, IL, U.S.A.) so we could easily identify them upon return to the hive. Dancing bees were those that performed ≥8 ‘round dance’ circuits, which indicates robust dancing (Barron et al. 2007a,b). Non‐dancers were those that did not perform any dance circuits within approximately 20 seconds after returning to the hive (Seeley & Tovey 1994) and then walked beyond the ‘dance floor’, an area near the hive entrance where most dances occur. One pair of sucrose feeders was available from 1000 h to 1130 h daily for collections for RNAseq, and a second pair of feeders was available from 1330 h to 1445 h daily for collections for ISH. Bees from both colonies were used for both types of analysis, as in Experiment 2 (RNAseq, N = 20; ISH, N = 14).