Mechanisms of stable lipid loss in a social insect

Abstract

Worker honey bees undergo a socially regulated, highly stable lipid loss as part of their behavioral maturation. We used large-scale transcriptomic and proteomic experiments, physiological experiments and RNA interference to explore the mechanistic basis for this lipid loss. Lipid loss was associated with thousands of gene expression changes in abdominal fat bodies. Many of these genes were also regulated in young bees by nutrition during an initial period of lipid gain. Surprisingly, in older bees, which is when maximum lipid loss occurs, diet played less of a role in regulating fat body gene expression for components of evolutionarily conserved nutrition-related endocrine systems involving insulin and juvenile hormone signaling. By contrast, fat body gene expression in older bees was regulated more strongly by evolutionarily novel regulatory factors, queen mandibular pheromone (a honey bee-specific social signal) and vitellogenin (a conserved yolk protein that has evolved novel, maturation-related functions in the bee), independent of nutrition. These results demonstrate that conserved molecular pathways can be manipulated to achieve stable lipid loss through evolutionarily novel regulatory processes.

Fig. 1.

Maturational changes in fatty acid and carbohydrate metabolism gene expression. Transcripts differentially expressed between nurse and forager fat bodies (false discovery rate, FDR<0.05) were mapped to metabolic pathways based on the union of direct annotations from the Kyoto Encyclopedia of Genes and Genomes (KEGG) and the KEGG annotations for Drosophila orthologs of honey bee genes. Pathway diagrams are modified from portions of KEGG maps for glycolysis/gluconeogenesis (00010), starch and sucrose metabolism (00500), and fatty acid metabolism (00071).

Fig. 2.

Correlations of transcriptional responses in fat bodies to maturation and diet with protein differences in the fat bodies and hemolymph. log₂ expression differences are shown for mRNA (microarrays) and protein (liquid chromatography/mass spectrometry, LC/MS) for all genes quantified in both platforms. Labels are shown for selected genes (lsp2, larval serum protein 2; vg, vitellogenin; obp13, odorant binding protein 13) and categories [energy metabolism: gene ontology (GO) biological processes tricarboxylic acid cycle, ATP synthesis, or oxidative phosphorylation]; carbohydrate metabolism (GO: carbohydrate metabolic process).

Fig. 3.

Age-related changes in responsiveness of metabolic and hormonal signaling pathways in fat bodies to nutritional stimuli. Fat body gene expression for nurses and foragers collected directly from the hive and for pre-nurses, nurses and foragers that were caged and fed either rich or poor diet. Asterisks indicate significance in paired contrasts following mixed-model ANOVA (*P<0.05; **P<0.01, ***P<0.001, n.s. P>0.05). Expression (mean ± s.e.m.) is shown relative to nurses in the hive. Genes: vitellogenin (vg), JH esterase (jhe), JH epoxide hydrolase (jheh), insulin-like peptide 2 (ilp2), thiolase, carnitine O-palmitoyl transferase 1 (cpt1), lipid storage droplet 2 (lsd2), insulin-related receptor 1 (inR1), adipokinetic hormone receptor (akhR) and small neuropeptide F receptor (snpfr). N=19–30 bees/group.

Fig. 4.

Co-expression patterns of genes related to metabolic and hormonal signaling pathways. Pearson correlation matrix for co-expression between genes shown in Fig. 3 (qPCR, N≈250) with average-linkage hierarchical clustering based on Euclidean distance.

Fig. 5.

Effects of queen mandibular pheromone (QMP) on abdominal lipid stores and food consumption. (A) Abdominal lipid stores of hive-reared 1 day old bees (1d), 5 day old bees (5d), nurses (N) and foragers (F), and of cage-reared 5 day old bees fed either rich or poor diet in combination with exposure to QMP or a solvent control. Means + s.e.m. N=30 bees. ANOVA for diet × QMP factorial: P_diet=1.0e–8, P_QMP=0.006, P_diet×QMP=0.20. (B) Effects of QMP on food consumption (total consumption over 4 days for cages containing 35 bees) by bees fed a rich diet of both pollen paste and sugar syrup and exposed to either QMP or a solvent control. N=6 cages. (C) Effects of QMP on food consumption by bees fed a poor diet of sugar syrup only. N=8 cages. In B and C, bars indicate least square means and their standard errors based on ANOVA for QMP exposure and trial. *P_QMP<0.05.

Fig. 6.

Genes with concordant responses to maturation, diet, vg RNAi and QMP. Genes are shown that responded significantly (FDR<0.2) in all four experiments in concordant directions relative to the effects of each factor on a bee's lipid stores. Heatmap shows the log₂ transformed difference estimate in each experiment. Dots indicate annotation of a gene to the GO biological processes listed or manually annotated to glycogen breakdown. Gene names are listed according to the A. mellifera Official Gene Set 2 (Honeybee Genome Sequencing Consortium, 2006) and their orthology to Drosophila melanogaster genes based on Reciprocal Squared Distance or reciprocal BLAST. Oligos corresponding to unannotated ESTs are listed according to the EST name.

Fig. 7.

Verbal model for the regulation of stable lipid loss and its coordination with behavioral maturation. We propose that stable lipid loss occurs through a shift from the utilization of nutrients for lipid biosynthesis to their utilization via energy metabolism pathways. These metabolic pathways are regulated by juvenile hormone (JH) and insulin/insulin-like growth factor (IIS), much as in other species, and by novel regulatory inputs to JH and IIS whose efficacy depends on a bee's maturational state. Lipid gain early in life is likely controlled primarily by diet, but an unknown maturational signal decreases the responsiveness of JH and IIS to diet once bees become nurses. Later, maturational signals related to the transition from hive work to foraging repress the inhibition of JH/IIS by QMP, as part of the declining sensitivity of older bees to QMP (Grozinger and Robinson, 2007). In addition, there is a well-established mutually repressive relationship between JH and Vg that is thought to contribute to the timing of maturation (Amdam and Page, 2010; Guidugli et al., 2005; Rutz and Luscher, 1974). Communication between the brain and fat is implicit in this model because of the localization of IIS and JH synthesis to the brain and the adjacent retrocerebral complex, respectively, whereas metabolic changes and stable lipid loss occur in the fat bodies. Mutually repressive relationships in general are thought to act as bistable switches, so changes in any of these JH-related repressors during maturation could trigger stable lipid loss. Activating connections are shown by lines ending in arrows; inhibitory connections are shown by lines ending in ovals.