Biology Letters
You have accessResearch articles

Queen pheromone modulates the expression of epigenetic modifier genes in the brain of honeybee workers

Carlos Antônio Mendes Cardoso-Junior

Carlos Antônio Mendes Cardoso-Junior

Departamento de Biologia Celular e Bioagentes Patogênicos, Faculdade de Medicina de Ribeirão Preto, Universidade de São Paulo, Ribeirão Preto, Brazil

Behaviour, Ecology and Evolution (BEE) laboratory, University of Sydney, Macleay Building A12, Sydney NSW 2006, Australia

[email protected]

Google Scholar

Find this author on PubMed

,
Isobel Ronai

Isobel Ronai

Behaviour, Ecology and Evolution (BEE) laboratory, University of Sydney, Macleay Building A12, Sydney NSW 2006, Australia

Google Scholar

Find this author on PubMed

,
Klaus Hartfelder

Klaus Hartfelder

Departamento de Biologia Celular e Bioagentes Patogênicos, Faculdade de Medicina de Ribeirão Preto, Universidade de São Paulo, Ribeirão Preto, Brazil

Google Scholar

Find this author on PubMed

and
Benjamin P. Oldroyd

Benjamin P. Oldroyd

Behaviour, Ecology and Evolution (BEE) laboratory, University of Sydney, Macleay Building A12, Sydney NSW 2006, Australia

Google Scholar

Find this author on PubMed

Published:https://doi.org/10.1098/rsbl.2020.0440

    Abstract

    Pheromones are used by many insects to mediate social interactions. In the highly eusocial honeybee (Apis mellifera), queen mandibular pheromone (QMP) is involved in the regulation of the reproductive and other behaviour of workers. The molecular mechanisms by which QMP acts are largely unknown. Here, we investigate how genes responsible for epigenetic modifications to DNA, RNA and histones respond to the presence of QMP in the environment. We show that several of these genes are upregulated in the honeybee brain when workers are exposed to artificial QMP. We propose that pheromonal communication systems, such as those used by social insects, evolved to respond to environmental signals by making use of existing epigenomic machineries.

    1. Introduction

    The transition to social living in the eusocial insects required that the reproductive interests of individual workers be subsumed by the collective interests of the colony [13]. In particular, workers are functionally sterile, whereas queens are highly fecund [4,5]. The regulation of worker fertility required the evolution of effective inter-caste communication systems that rapidly respond to the changing needs of the colony. Communication between nest-mates mainly occurs via pheromones [6,7], chemical signals that are produced by one individual, and cause changes in the behaviour or physiology of another [810].

    In the western honeybee (Apis mellifera), a key pheromone is the queen mandibular pheromone (QMP), which is a blend of fatty acids secreted by the glands on the head of the queen [1114]. Although the QMP composition can slightly differ between A. mellifera subspecies [15,16], it major components, including (2E)-9-oxodecenoic acid (9-ODA) [17], affects several important characteristics of workers, including their reproduction [17,18], retinue response to queens [19,20], learning capacity [21], nest-mate recognition [22] and age at onset of foraging [23]. Queen pheromone is spread among workers solely by physical contact [14,18]. Phenotypic variation in these traits is associated with differential gene expression in the brains of workers [24,25]. Nonetheless, little is known about the intermediate steps between QMP production and release by the queen, the regulation of gene expression in workers and changes in worker behaviour [26,27].

    Epigenetic mechanisms are a likely mediator between a worker's social environment and global gene expression responses [28]. Several classes of epigenetic mechanism described in honeybees are potentially associated with environmental cues [2932]. For example, DNA methylation, a reversible chemical modification of cytosines in CpG contexts, is associated with behavioural maturation in the brains of honeybee nurses and foragers [33,34]. DNA methylation is catalysed and maintained by the DNA methyltransferase (DNMT) family of enzymes [35,36]. Interestingly, in honeybees, the expression of genes associated with the maintenance of DNA methylation levels after DNA replication (Dnmt1a and Dnmt1b) is modulated by different social stimuli from Dnmt3, an enzyme that establishes DNA methylation patterns de novo [28,3739]. In addition, the expression of Dnmt2 (also called Trdmt1), a gene whose enzyme product methylates RNA substrates [40], is affected by different social contexts [28,33]. These studies suggest that epigenetic machineries associated with nucleotide modification are affected by several environmental cues.

    Another epigenetic mechanism, histone post-translational modifications (HPTMs), change chromatin structure by altering the physico-chemical affinity between DNA and histones and thereby affect gene expression [41]. HPTMs are catalysed by histone modifier proteins [36], which can be divided into three functional classes: writers, erasers and readers. ‘Writer' enzymes add chemical radicals to histone tails by covalent modification. For example, lysine acetyltransferases (KATs) promote the acetylation of lysine residues [36], which reduce the affinity between DNA and nucleosomes. Histone acetylation induces chromatin relaxation and is often associated with increased gene expression [42]. By contrast, ‘eraser’ enzymes remove such chemical radicals from histone tails. Canonical eraser enzymes are the histone deacetylases (HDACs) and sirtuins, which remove acetyl groups from lysine residues, resulting in chromatin compaction and, consequently, the inhibition of gene expression [36,43]. Finally, ‘reader' enzymes recognize epigenetic modifications and induce chromatin remodelling through the recruitment of protein complexes [36]. A honeybee proteome study [30] has shown that histone tails are extensively modified by epigenetic marks, indicating that writer, eraser and reader enzymes are present in the honeybee. Furthermore, the differential accumulation of HPTMs has been associated with caste differentiation and behaviour in bees and ants [30,4447].

    Given that QMP affects behaviours in honeybee workers [19,20,23,48], we hypothesized that the expression of genes associated with epigenetic modification to nucleotides and histones would respond to QMP exposure in the brain of honeybee workers. These epigenetic mechanisms can thus serve as proxies to understand the regulation of global changes in gene expression in a complex social environment.

    2. Material and methods

    (a) Biological material

    To obtain age-matched adult workers, we collected brood frames from four unrelated queenright A. m. ligustica source colonies and kept them in an incubator overnight at 34.5°C. From each source colony (n = 4), workers were randomly allocated to each of two cages (n = 150 bees per cage), generating a total of eight cages for the entire experiment. One cage from each colony pair (QMP+) was furnished with a 0.5 queen equivalent per day QMP strip (Phero Tech Inc. Canada). The strip contains five of the major components of natural QMP in their biologically active ratio [17] that are a highly effective A. m. ligustica queen mimic in cage experiments with young workers [49,50]. The other cage from each colony pair (QMP) contained no QMP strip. Pollen, honey and water were provided ad libitum. Food was replenished when necessary, and the number of dead workers was recorded each day, which was similar in the QMP+ and QMP cages (electronic supplementary material, table S1). Cages were kept in a Thermoline incubator (model l985–250 L) at 34.5°C for 4 days. The metal cages with a glass front (14 × 10 × 7 cm) were spaced at least 5 cm away from one another in the incubator [51]. Workers were collected on dry ice at day 0 (directly from the brood comb), day 1 and day 4. Day 1 was chosen to identify genes with a quick response to the QMP treatment, and day 4 was chosen to identify the genes that are still influenced by QMP exposure after prolonged exposure. At each time point, we collected 8 bees from each cage. We then dissected the brains of the workers on dry ice [52]. Each sample consisted of a single gland-free brain, generating a total of 32 samples per treatment by time combination (electronic supplementary material, figure S1). All experiments were performed at the University of Sydney, Sydney, Australia in January 2019. Detailed methods for gene expression analyses and the in silico identification of DNA methyltransferases and histone modifiers in the honeybee genome are provided in the electronic supplementary material. Raw data are accessible on Dryad [53].

    (b) Statistical analysis

    Expression data were first assessed using the D'Agostino & Pearson normality test in GraphPad Prism software v.7. For genes that passed the normality test, we applied a generalized linear mixed model (GLMM). GLMM was used to compare the expression of the QMP+ and QMP treatments at day 1 and day 4 with ‘colony' as a random effect and ‘treatment' and ‘age' as fixed effects. To model the gene expression, we used link = identity, family = Gaussian. For those genes that did not fit the Gaussian distribution (Dnmt1a, Dnmt1b, Dnmt2, Kat2a, Kat3b, Kat6b, Kat7, Kat8, Sirt6 and Polybromo-1), we applied a log10 transformation (electronic supplementary material, table S4). For genes that deviated significantly from a Gaussian distribution even after a log10 transformation (Dnmt1a, Dnmt1b, Dnmt2, Kat6b, Kat8, Sirt6 and Polybromo-1), we used an R package called Aligned Rank Transformation (ARTool) to model gene expression using ‘colony' as a random effect and ‘treatment' and ‘age' as fixed effects. We used day-0 data as a baseline for gene expression. For Hdac1, for which a significant interaction between ‘treatment' and ‘age' was found, we report all pairwise comparisons across treatment and age (electronic supplementary material, table S4). GLMM and ARTool analyses were performed in R [54] loading the packages lme4, car, wrs2, artool and emmeans. An adjusted p-value for multiple pairwise comparisons (Tukey correction for each gene) lower than 0.05 was considered significant for all statistical tests.

    3. Results

    Using the protein sequences of the 15 genes studied, we first acquired in-silico evidence (e.g. subcellular location, predicted domains and homology with other species) that each gene was a bona fide epigenetic modifier of DNA, RNA or histones (electronic supplementary material, table S2). In 1-day-old workers, the expression of ten genes associated with epigenetic processes (Dnmt1b, Dnmt2, Dnmt3, Kat2a, Kat3b, Kat8, Hdac1, Hdac3, Sirt1 and Sirt7) was affected by exposure to QMP (adjusted p < 0.05, figures 1 and 2, electronic supplementary material, table S4). At the age of 4 days, six genes (Dnmt1b, Dnmt2, Kat3b, Kat8, Sirt7 and Polybromo-1) were differentially expressed in the brains of workers (adjusted p < 0.05, figures 1 and 2, electronic supplementary material, table S4). For all differentially expressed genes, at least two colonies responded to QMP treatment by an increased expression (electronic supplementary material, figure S2). Age was a statistically significant effect for 13 of the 15 genes (p < 0.05, electronic supplementary material, table S4), the exceptions being Kat7 and Dnmt3. A significant interaction between treatment and age was found for Hdac1, while treatment by day interactions were non-significant for all other genes (electronic supplementary material, table S4). This interaction arose because the direction of expression level reversed between QMP treatment groups between days 1 and 4 (figure 2b and electronic supplementary material, table S4).

    Figure 1.

    Figure 1. Relative expression of four nucleotide modifier (DNA methyltransferase) genes in the brains of 0-, 1- and 4-day-old honeybee workers, exposed to queen mandibular pheromone (QMP+) or not (QMP). Each box shows the interquartile range (25th–75th percentiles) and the median (line), while whiskers represent the minimum and maximum points. Red dots inside boxes represent individual brains. Relative expression was calculated for each gene at all three ages. Day 0 was used as the baseline for gene expression. Statistical information: GLMM test or ARTool of differences between means with Tukey correction for multiple pairwise comparisons, * p < 0.05, ** p < 0.01, *** p < 0.001, 8 bees per cage, 4 cages (colonies) per treatment/day combination (total 32 bees per treatment/day combination).

    Figure 2.

    Figure 2. Relative expression of 11 histone modifier genes in the brains of 0-, 1- and 4-day-old honeybee workers, exposed to queen mandibular pheromone (QMP+) or not (QMP). (a) Relative expression of histone acetyltransferases genes (writer enzymes). (b) Relative expression of histone deacetylases and sirtuin genes (eraser enzymes). (c) Relative expression of the Polybromo-1 gene (reader enzyme). Each box shows the interquartile range (25th–75th percentiles) and the median (line), while whiskers represent the minimum and maximum points. Red dots inside boxes represent individual samples. Relative expression was calculated for each gene at all three ages. Day 0 was used as the baseline for gene expression. Statistical information is as in figure 1.

    4. Discussion

    Our study shows that the presence of QMP is associated with an increase in the expression of 11 of 15 genes that are related to epigenetic processes in the brain of honeybee workers. As predicted, our data indicate that epigenetic mechanisms are likely mediators between queen pheromone signalling and the regulation of worker gene expression. Given that QMP alters worker behaviour [19,20,23,48], we suggest that queen–worker communication via QMP makes use of existing epigenetic mechanisms that regulate transcriptomic changes. In this way, epigenetic mechanisms can propagate pheromonal information.

    Some expression responses are noteworthy, as they have been observed in other species. In particular, there is an increased expression of Dnmt3 in the brains of honeybee workers in response to QMP in the workers' environment. Similar effects have been seen in the ants Lasius flavus and L. niger [38], where (ant) queen pheromones influence the expression of Dnmt3. We also showed that expression of Kat 8 is upregulated in the brains of QMP-treated honeybee workers. This gene is differentially spliced in L. flavus ants exposed to queen pheromone [55]. Although intra- and interspecific differences in the composition and response of queen pheromones are expected among social insects [15,56,57], these results suggest that social insects use similar epigenetic programs (e.g. acetylation/deacetylation of histones and nucleotide methylation) to respond to queen pheromones.

    Several of the histone modifier genes associated with acetylation/deacetylation processes had increased expression in workers exposed to queen pheromone. While it is still not possible to determine whether the differential expression of epigenetic modifier genes led to changes in the honeybee epigenomes, we hypothesized that queen signals influence the modification of histones to promote chromatin reorganization, thereby altering gene expression in worker brains. In line with this hypothesis, it was recently shown that honeybee queens regulate worker fertility through polycomb repressive complex 2 (PRC2) activity and differential histone methylation marks [58]. We propose that queens, via QMP, may influence modifications to histones to regulate behavioural plasticity in the brains of honeybee workers, just as they do in ovaries.

    Pheromonal modulation of gene expression in honeybee workers changes over time (this study, [24,25]). Gene expression in QMP workers is relatively stable from day 0 to day 1 when compared to QMP+ workers, suggesting that QMP actively promotes the expression of several epigenetic modifier genes within 24 h. Only six genes were differentially expressed after 4 days of QMP exposure, indicating that the expression of the majority of epigenetic modifiers is dynamically switched on and off [24]. The expression of Hdac1, the only gene in which we found an interaction between ‘treatment' and ‘age', suggests that in the absence of a QMP signal, older workers require higher levels of Hdac1 expression, whereas younger workers do not.

    Although our demonstration of increased expression of DNA methyltransferase genes and histone modifier genes in response QMP is unequivocal, it is not clear whether these alterations lead to causal alterations in the epigenome itself [59,60], or to changes in the expression of genes that might be so methylated. For example, the methylomes of Ooceraea biroi ants are genotype-specific and unresponsive to a major change in a social context [61]. This finding, along with others [59,60,62], questions the role of methylation in the dynamic regulation of gene expression in social insects. Therefore, the alterations in the mRNA levels of epigenetic genes identified in this study need to be interpreted with caution. Future research should prioritize functional assays (e.g. knockdown of epigenetic modifier genes or inhibition of protein activity) to establish a causal relationship between differential expression of epigenetic genes and their respective epigenetic products in the DNA, RNA or histones.

    Our study provides evidence that many genes associated with epigenetic modification are differentially expressed in the brains of honeybee workers in response to queen pheromone. Changes in the expression of the genes studied here may drive changes in gene expression in the brains of adult workers, providing a plausible mechanism by which a queen can influence both the rate of behavioural maturation and reproductive behaviour of her workers.

    Ethics

    Honeybees are not in the list of animals that require ethical approval.

    Data accessibility

    All data that support this study are available in the supplementary material or at Dryad Digital Repository (https://doi.org/10.5061/dryad.s4mw6m959).

    Authors' contributions

    C.A.M.C.J. designed the study, carried out the cage experiments, dissections, performed molecular laboratory work, analysed data and wrote the manuscript draft. I.R. carried out the protein characterization and wrote the manuscript. B.P.O. and K.H. revised the manuscript and supervised the work, providing substantial contributions to the conception and design of this study. All authors gave final approval for publication and agree to be held accountable for the work performed.

    Competing interests

    We declare we have no competing interests.

    Funding

    This study was funded by Australian Research Council (DP180101696) and Fundação de Amparo à Pesquisa do Estado de São Paulo (2016/15881-0 and 2017/09269-3).

    Footnotes

    Electronic supplementary material is available online at https://doi.org/10.6084/m9.figshare.c.5221443.

    Published by the Royal Society. All rights reserved.

    References