REVIEWS

Eusocial as emerging models for behavioural epigenetics

Hua Yan1*, Daniel F. Simola2,3*, Roberto Bonasio2,3, Jürgen Liebig4, Shelley L. Berger2,3,5,6 and Danny Reinberg1,7 Abstract | Understanding the molecular basis of how behavioural states are established, maintained and altered by environmental cues is an area of considerable and growing interest. Epigenetic processes, including methylation of DNA and post-translational modification of histones, dynamically modulate activity-dependent gene expression in neurons and can therefore have important regulatory roles in shaping behavioural responses to environmental cues. Several eusocial — with their unique displays of behavioural plasticity due to age, morphology and social context — have emerged as models to investigate the genetic and epigenetic underpinnings of social behaviour. This Review summarizes recent studies in the epigenetics of social behaviour and offers perspectives on emerging trends and prospects for establishing genetic tools in eusocial insects.

1Department of Biochemistry Animal behaviour is best understood as an emergent phenotypic states through transcriptional regulation of and Molecular phenomenon predicated on the dynamic interpretation genome-wide gene expression4,7–9. The first strategy is Pharmacology, New York University School of Medicine. of sensory stimuli by the brain. The brain’s response to the use of specific transcription factor (TF) complexes 2Department of Cell and sensory stimuli can be modulated by metabolic pro- and regulatory networks that involve feedback loops Developmental Biology, cesses and the endocrine system1; however, one of the (that is, trans epigenetics), and the second strategy Perelman School of Medicine, most remarkable characteristics of the brain is the abil- involves alteration of chromatin structure through DNA University of Pennsylvania. ity to condition consistent behavioural responses on the methylation, histone post-translational modifications 3 Epigenetics Program, University of Pennsylvania. basis of prior sensory experience. In other words, (PTMs), non-coding RNAs (ncRNAs) and associated 4School of Life Sciences, the architecture of an animal brain encodes a persistence chromatin regulatory proteins (that is, cis epigenetics). Arizona State University, or memory (which is broadly defined here to include all These two processes are able to provide an explanation Tempe, Arizona short- and long-term forms of memory2) of transient for behavioural persistence and plasticity (that is, alter- 85287–4501, USA. sensory stimuli that is used to tune the brain’s sensitiv- ing a behavioural response on the basis of new stimuli) 5Department of Biology, University of Pennsylvania. ity to similar stimuli in the future in order to elicit a by their ability to establish and maintain quantitative 6Department of Genetics, fitting behavioural response. Decades of research into control of transcription for long periods of time, either Perelman School of Medicine, the molecular processes that underlie the persistence of across cell divisions or within terminally differentiated University of Pennsylvania, behavioural responses through memory mechanisms cells such as neurons. Philadelphia, Pennsylvania 19104, USA. have shown a key role for the regulation of gene expres- Several animal species have been developed as mod- 7Howard Hughes Medical sion in neurons, which are one of the major cell types els to investigate the various aspects of animal behav- Institute, New York University of the brain2–6. However, the specific regulatory mecha- iour and brain function, including the fruitfly Drosophila School of Medicine, New York, nisms responsible for establishing and maintaining pat- melanogaster for aggression and reproductive behav- New York 10016, USA. terns of neuronal gene expression in response to sensory iour10, mice and rats for learning and memory11, and the *These authors contributed 12 equally to this work. stimuli remain poorly understood, mainly because zebra finch for neurodegeneration and regeneration . Correspondence to J.L., behavioural responses arise from the integration of However, these systems have notable disadvantages for S.L.B. and D.R. stimuli not only from gene-level networks within a neu- genetic and epigenetic studies of behaviour: fruitflies e‑mails: [email protected]; ron but also from networks that integrate the numerous lack key DNA methylation enzymes13; rodents often [email protected]; neurons (and non-neuronal glial cells) within a brain2,6,7. show strong behavioural variation even in standardized [email protected] 14 doi:10.1038/nrg3787 Over the past 30 years, two general strategies have environments ; and birds can be difficult to maintain 15 Published online 9 September emerged from the historically independent field of epi- in a laboratory . Moreover, none of these organisms are 2014 genetics to explain the induction and maintenance of highly social, and they therefore cannot provide suitable

NATURE REVIEWS | GENETICS VOLUME 15 | OCTOBER 2014 | 677

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

Social behaviours experimental paradigms for the large variety of complex this end, key aspects of the brain and the central nerv- The interactions among social behaviours seen in other species. By contrast, the ous system, including gross morphology and neuronal individuals of the same species, species-rich group of eusocial insects (including all connectivity, have been extensively characterized in for example, collaboration and , as well as some and ) show exemplary eusocial insects, such as the honeybee and within a well-defined group such as a colony of eusocial many of the most fascinating, complex and enigmatic some species, and have been shown to vary by mor- 17–21 insects. displays of animal behaviour. phology, age and social context . Furthermore, many Although individual eusocial insects exhibit behav- eusocial insects, especially ants, can be maintained eas- Polyethism iours that reflect ‘simple’ antagonistic or adaptive ily and in large numbers in a laboratory while largely Variation in the allocation of responses (for example, fight or flight)16, they also display preserving their natural social context, which greatly nest-related tasks among (BOX 2) individuals in a colony. much more sophisticated behaviours elicited by coopera- facilitates controlled behavioural analyses . Polyethism typically refers tive interactions among individuals within a social group Polyethism is primarily influenced by two factors: to the special case of or a colony setting, such as nursing, foraging, nest main- caste morphology and age. Caste morphology is speci- age-dependent changes in an tenance, defence and policing. The regulation of such fied during juvenile (that is, larval or nymphal) develop- individual’s behaviour (that is, polyethism queen worker 22,23 age-dependent or temporal cooperative colonial behaviours (that is, (also ment for both and polymorphic castes . polyethism); however, it more known as division of labour)) is a central feature of more Within each caste, behaviour also changes in an age- generally denotes differences advanced forms of (BOX 1) and often involves dependent manner, as seen in the classic transition from in behaviour associated with the strict allocation of behaviours to qualitatively distinct nursing to foraging in honeybees (Apis mellifera)24,25. caste (that is, temporal as well groups of individuals (denoted as castes) that may vary Timing of these behavioural transitions is not necessar- as morphological or physical polyethism). by age or morphology. The existence of behaviourally ily ‘hardwired’ and is often sensitive to dynamic changes specialized castes in particular underscores the relevance in social context, nutrition and physical environment26 of eusocial insects for behavioural epigenetics research. To (FIG. 1a). This intrinsic behavioural plasticity of eusocial insects provides numerous opportunities for experimen- tal manipulation. For example, in honeybees, selectively royal jelly Box 1 | An overview of insect sociality feeding larvae a protein-rich diet that includes induces their development into queens22,23, whereas Part of our fascination with social insects stems from their capacity to learn and execute removing existing nurses from a colony induces forag- 136 complex decisions not only as individuals but also as integrated social groups — a ers to revert to nursing behaviour27 (FIG. 1b). Workers of rare trait in the animal kingdom. Indeed, the best-studied and most sophisticated forms the ant saltator can mate, obtain reproduc- of social behaviour are highly restricted taxonomically and are mainly, although not exclusively, found in the eusocial insect orders (which includes all ants tive status (that is, become ) and function as and some bees and wasps) and Blattodea (which comprises all termites)22,44. queens when an existing queen dies or when workers are 28,29 is also found in select species outside insects, including snapping shrimps and naked artificially isolated individually or in groups (FIG. 1b). mole rats. Gamergates can also be reverted to workers both behav- Eusociality is considered to be the most complex form of sociality and is defined as iourally and physiologically by temporary separation the cooperative care of offspring, which are born from reproductive individuals but are from a colony (C. Penick, unpublished observations). reared by non-reproductive individuals with overlapping generations within a colony. Thus, insights into the molecular mechanisms that Eusociality and other types of sociality (such as communal, quasisocial and semisocial underlie polyethism can be obtained by analysing both sociality) are likely to have derived from a simpler form of subsociality, which is natural and artificially induced variation in behaviour 137 generally characterized as the display of transient parental brood care . Eusociality among individuals in a colony. In addition, the ecologi- may be found in simple societies with a few monomorphic members and basic division cal abundance and diversity of eusocial insect species — of labour (for example, Lasioglossum albipes and Polistes metricus), as well as in complex 30 societies that contain thousands to millions of workers that show highly specialized and which comprise more than 30,000 extant species (see polymorphic division of labour (for example, Apis mellifera, Camponotus floridanus AntWeb) — provide excellent opportunities for compar- and Reticulitermes flavipes) (BOX 2). ative studies on the evolution and function of the genetic In the insect order Hymenoptera, the state of eusociality has evolved more than 10 and epigenetic processes underlying behaviour. times since its first advent in the Cretaceous geological period ~150 million years ago. The revolution in next-generation sequencing tech- By contrast, eusociality has evolved only once in the equally old (if not older) nologies has greatly accelerated behavioural research in lineage (Blattodea: Isoptera)44,138. It was previously assumed that sociality in ants was eusocial insects. In 2006, the honeybee A. mellifera was derived from their -like ancestors, whereas bees constituted an independently the only eusocial insect to feature a draft genome assem- evolved eusocial lineage. However, recent phylogenomic analyses indicate that ants, bly31. Subsequently, high-quality draft genomes of two 139,140 bees and social wasps may share a common ancestor . Notably, sociality in (REF. 32) 137 ant species were published in 2010 , which were hymenopteran lineages has evolved to qualitatively different degrees and has 33 involved lineage-specific adaptations at multiple scales of genomic organization39. soon followed by the genomes of five other ant species In Hymenoptera, a typical colony contains haploid males and diploid females, and most recently by the genomes of the clonal raider 34 which are grouped into reproductive (queen) and non-reproductive (worker) castes. ant Cerapachys biroi , the socially polymorphic sweat 35 The worker caste may be divided further into morphological subcastes, such as minor Lasioglossum albipes and the dampwood termite workers and major workers (also known as soldiers). Additional worker subcastes (for nevadensis36. These genomes have facili- example, media and super-soldiers) exist in few ant species, including leaf-cutters tated generation of the first epigenomic maps of DNA (genera Atta and Acromyrmex) and some Pheidole species. Caste and subcaste methylation and histone PTMs in eusocial insects37–42, differentiation normally occurs during larval development, whereas further and have enabled the field of insect to begin behavioural differentiation occurs in adults. In an unusual example of reproductive to address key molecular questions. Is there a conserved plasticity, workers in some ponerine ant species, such as , can genetic basis for sociality among eusocial insect spe- supersede morphological queens by becoming gamergates (that is, mated egg-laying 39,43 workers recognized as primary reproductives by remaining workers). cies ? What epigenetic processes drive age-, caste- and context-dependent behavioural plasticity27,40?

678 | OCTOBER 2014 | VOLUME 15 www..com/reviews/genetics

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

Box 2 | Development of genetic eusocial insect models

Reproductive division of labour, which is instrinsic to eusocial insects, controlled crosses cannot be overstated (FIG. 3) because it allows presents a unique challenge for the development of genetic lines because, in sophisticated genetic manipulations to be carried out, as is routinely done in the vast majority of species, only a few individuals in a colony are capable of and mice (for example, gene knockout and knock‑in, reproduction. However, in a few species — such as the parthenogenetic as well as temporally controlled and tissue-specific overexpression using honeybee subspecies Apis mellifera capensis132, the Halictid bee (also known UAS–Gal4, FLP–FRT or Cre–loxP systems142,143). In this regard, H. saltator and as the sweat bee) Lasioglossum albipes35, the ponerine ant Harpegnathos M. pharaonis readily breed with males from the same colony28,144, whereas saltator28 and the doryline ant Cerapachys biroi34 — all individuals are capable C. biroi does not mate and all offspring in this species are female maternal of laying female eggs. In addition, a large percentage (~25%) of clones as a result of thelytokous parthenogenesis34,107. Artificial insemination queen-destined brood can be experimentally produced in the pharaoh ant offers an alternative to controlled breeding that enables genetic crosses, and species Monomorium pharaonis141. Moreover, in honeybees, queens can be it has been carried out successfully in honeybees, bumblebees and leaf-cutter reared by placing embryos in honeycomb cells that are specially prepared for ants145–147. The following table lists characteristics of several eusocial insect queens128. In selecting a species for genetic manipulation, the importance of species that are suitable for experimental analyses of social behaviour.

Species Laboratory Available Availability of Colony Individual Worker Poly­ Poly­ Develop­ controlled clones or resources size size caste poly­ gyny andry mental breeding inbred morphism time (egg lines Genome RNAi to adult) sequence Apis mellifera Yes*‡§ No Yes Yes 104 Large* No No Yes 16–24 days (European honeybee) Bombus terrestris Yes*§ No Yes No 102 * Large* No, but No No ~25 days (bumblebee) continuous size range Lasioglossum albipes Yes*|| No Yes No 101 * Medium* No No No ~1 month (sweat bee) Camponotus floridanus No§ No Yes Yes 104 Medium* Dimorphic: No No ~3 months (Florida ) majors and minors Cerapachys biroi No|| Yes* Yes No 102 * Small No Yes NA ~1.5 months (clonal raider ant) Harpegnathos saltator Yes*|| Yes* Yes No 102 * Large* No No¶ No ~2.5 months (Jerdon’s jumping ant) Linepithema humile No No Yes No 106 Small No Yes No ~3 months (invasive Argentine ant) Monomorium Yes* Yes* No No 104 Small No Yes No ~1.5 months pharaonis (pharaoh ant) Pheidole morrisi No No No No 102 * Small Dimorphic: Yes Yes ~1.5 months (big-headed ant) majors and minors Pogonomyrmex No§ No Yes No 103 Large* Dimorphic: No** Yes ~1.5 months species complex majors and (harvester ant) minors# Solenopsis invicta No No Yes No 104 Small No, but Yes §§ Yes ~2 months (red imported fire ant) continuous size range Polistes metricus Yes*‡ No No‡‡ Yes 102 * Large* No No No ~1.5 months (paper wasp) Reticulitermes flavipes Yes*|| Yes* No Yes 106 Small Dimorphic: Yes |||| Yes |||| NA (eastern subterranean soldiers and termite) workers Zootermopsis Yes*|| Yes* Yes No 103 Large* Dimorphic: Yes |||| Yes |||| NA nevadensis soldiers and (dampwood termite) workers

NA, not available; RNAi, RNA interference. *Features that are desirable for genetic manipulation. ‡Rearing conditions cannot be controlled for large colonies or groups of colonies as easily or as precisely as for colonies that can be reared completely in environmentally controlled chambers; may be subject to seasonal environmental effects. §Workers can lay viable haploid male eggs (the exception is parthenogenetic honeybee subspecies A. mellifera capensis). ||Workers can lay viable haploid male and diploid female eggs (all C. biroi workers are reproductive and only lay female eggs). ¶True polygyny involving queens is rare in nature; however, colonies can routinely have multiple gamergates. #Applies to Pogonomyrmex badius only. **Some species, for example, Pogonomyrmex californicus. ‡‡The genome of Polistes dominulus has been sequenced. §§Monogynous and polygynous forms exist. ||||Colonies are founded by a single pair of queen and king, which is later replaced (after their death) by multiple queens and kings.

NATURE REVIEWS | GENETICS VOLUME 15 | OCTOBER 2014 | 679

© 2014 Macmillan Publishers Limited. All rights reserved threshold level,the animalundergoes an epigenetic behavioural state age (red dashed line). When the foraging stimulus (apple) exceeds the phenotypes. Intheplot,threshold forforaging behaviourdecreases with threshold value factors withaninternalizedsignal value( regulating Juvenilehormone(JH)titre signal transduction pathways(forexample,byneuroendocrine pathways transduced and‘quantified’intoasignal or stimulus through intracellular are firstperceived (forexample,byantennalolfactoryreceptors), andthen determines an individual’s sensitivity to novel environmental stimuli, which epigenetic transfer function to define a response threshold the P insects, allF factors (E). Square brackets around paternalfactorsindicatethat,ineusocial current physiologicalcondition;andbioticabioticenvironmental mRNA); developmentalhistory(D)thatresults inits maternal example, factors (I), which include genotype, epigenotype and maternal factors (for environmental stimulusdependsonseveral factors:parentally inherited plasticity. Figure 1| REVIEWS 680 a c b Stimulus: Stimulus: Modulation ofresponse thresholds byinheritedfactors(I) Example: Example: Epigenetic response threshold modelofbehavioural plasticity Modulation ofresponse thresholds byenvironmental factors(E) |

0 OCTOBER 2014 Effect: Effect: generation. Thethree primaryfactors(I,EandD)are integrated byan ×

Epigenetic response threshold modelofbehavioural Larva

1 Threshold Threshold malesandsomeF a Precocious foraging Behavioural heterochrony Genetic polymorphism Queen induction Larval castedifferentiation Nutrition | The phenotypic response of an individual to a transienta to individualresponsephenotypic an The of | f (I, E, D) can induce stable behavioural to changesepigenetic (Juvenile)

Parental factors(I;inherited) | sequence Genome

Parental reproductives (P VOLUME 15 1 females do not inherit paternal factors from femalesdonotinheritpaternalfactorsfrom Age (Adult) Allele A Age

structure Chromatin ( BOX 3 s ) thatexceedstheresponse 0 ) (Adult) ) ). Onlytheenvironmental Queen or season Temperature © 2014 Macmillan Publishers Limited. Allrights reserved factors Maternal Allele B Worker Environmental factors(E;notinherited) f (I, E, D ) that that Nutrition dimorphic species or supersoldiers in stimulus (forexample, JH),asinthecaseofmajorcaste qualitative manner, for example, byprohibiting sensitivityto a given panel). Geneticdivergence mayalsoalterresponse thresholds ina the threshold for a given cue to increase environmental sensitivity (left colony) candirectly influenceresponse thresholds, forexample,bylowering occurring geneticvariation(thatis, allelicvariationamongindividualsina interactions betweeninheritedfactorsandenvironmental cues,naturally status (rightpanel). , whereas an induced loss of status leads to a reversion to worker individuals in reproductive polyethism.For example,lossofexistingreproductive according to changes in reproductive social context, which results in needed toprevent pupation. Aresponse threshold changes in adults the JHtitre decreases through larvalstages untilitislessthanthelevel internalized signalvalue( periods ofJHsensitivity)(leftpanel).Inmanycases,thequantification the example, topermitregulation ofcastefateinlarvae(thatis,transient changes according toenvironmental factors(includingnutrition),for transition from nursing(orange) to foraging (blue). Stimulus: Stimulus: Example: Example: Effect: Effect:

Threshold Threshold Gamergate transition andreversion Adult castedifferentiation Loss ofreproductive individuals Worker castepolymorphism Developmental pathwayactivation Genetic divergence context Social Harpegnathos c |Asphysiologicalconditionisdeterminedbythe JH Worker s

) Threshold

changes rather thanthethreshold itself;notably, Nursing

saltator Age (Juvenile) Age (Adult) Gamergate Developmental factors(D;age) inducesatransition from workerto Pheidole

www.nature.com/reviews/genetics s >ƒ(I,E,D) Progeny (F sp. ants Worker Nature Reviews b Foraging | A response threshold Epigenetic response Response threshold ( Transient stimulus(s) 130 1

) (right panel). determination in determination in Major Minor |

Genetics f ) REVIEWS

In this Review, we focus on the latter question through is sufficient to alter cell type identity4,59. This suggests a discussion of how epigenetic processes may regulate that epigenetic processes that regulate gene expression Castes behavioural plasticity in eusocial insects by providing the could provide the major mechanism for the regulation of Specialized behavioural groups molecular machinery to translate transient environmen- neuronal memory and therefore emergent properties within a eusocial colony that often correspond to tal cues into stable transcriptional patterns that can be of the brain, such as animal behaviour. One important morphological features and maintained throughout the marked developmental tran- nuance to this model pertains to genetic polymorphisms that are generally considered sitions of insects (that is, across cell divisions during met- (see below). Any genetic variant that alters an epigenetic to be a stable, if not amorphosis and in terminally differentiated cells such as response to an environmental cue may consequently permanent, characteristic of an individual. For example, postmitotic neurons). We also discuss the prospects for affect the likelihood of a particular behavioural response members of the queen caste studying mechanisms of transgenerational inheritance (for example, single-nucleotide polymorphisms (SNPs) hatch as adults with wings using eusocial insects and for establishing genetic and that contribute to pollen hoarding in honeybees60). In this and can reproduce, whereas genomic tools in these species. way, naturally occurring genetic variation in a colony may members of the worker caste bias the behavioural repertoires of individuals (FIG. 1a). (or castes) are wingless and do not normally reproduce. Epigenetic stabilization of transcription Eusocial insects primarily use three types of envi- TFs and histone acetylation coordinate caste fate in Behavioural epigenetics ronmental factors to control individual phenotype: ants. Recent comparative genomic analyses have identi- An emerging multidisciplinary nutrition, temperature and chemical compounds (for fied several TFs that are associated with caste fate and field of research that aims to 22,26,44–46 39 understand how epigenetic example, ) . These factors are applied behavioural plasticity in eusocial insects . A genome- processes transform transient to various ends and over different timescales to regulate wide study of nearly 30 solitary and eusocial insect environmental cues into phenotypic traits such as caste fate (for example, workers genomes analysed the number of TF binding sites that persistent molecular patterns feeding royal jelly to queen-destined larvae in A. mel- are found near promoters of orthologous genes, and of gene expression in order to lifera)22,23,47, adult reproduction (for example, social revealed that evolution of TF binding sites is more modulate animal behaviour. dominance in H. saltator)45,48–50 and adult behaviour (for divergent among eusocial insects than between solitary Queen example, age-dependent nestmate interactions in many and eusocial insects. Genes with the most significant A morphologically and/or species)6,26,51–53. On the basis of studies in a few eusocial evolutionary changes are enriched for neuroendocrine behaviourally distinct insect species, most notably in A. mellifera, it is under- function, show differential expression between castes in reproductive caste in eusocial insects that often shows stood that neuroendocrine signalling pathways — for both H. saltator and Camponotus floridanus ants, and specialization in both example, Insulin/Insulin-like growth factor (IGF) signal- are associated with neuronal-related TFs, such as Cyclic reproduction and dispersal ling (IIS) and Juvenile (JH) signalling — are AMP response element-binding protein (CREB), Empty abilities. Depending on the primarily responsible for transducing these environmen- spiracles (EMS) and Grainyhead (GRH). These findings species, a colony may contain tal cues (BOX 3). However, it remains unclear how these suggest that neuronal gene networks have been targeted one queen (monogyny) or multiple queens (polygyny). typically transient cues, as well as subsequent intracel- for regulatory rewiring in two independent eusocial Queens, together with males, lular and intercellular signalling cascades, induce stable lineages (ants and bees). constitute the ‘’ of a behavioural states in the brains of eusocial insects54. CREB was independently identified in the honeybee, eusocial insect colony. We propose that combinations of cis epigenetic pro- in which mRNA levels of the target genes of CREB were

Worker cesses (which involve DNA methylation, histone PTMs found to vary with age-dependent behavioural states A non-reproductive caste in and ncRNAs) and trans epigenetic processes (which (that is, foraging, maturation and aggression) using a eusocial insects. Workers involve TFs and ncRNAs) underlie behavioural plastic- compendium of transcriptome profiles from individ- cooperatively care for the ity by their ability to maintain transcriptional patterns ual brains61. CREB and other TFs — including those brood of the colony, forage for over time. Notably, this mechanism broadly applies to involved in Bone morphogenetic protein (BMP) signal- food, clean up the nest and defend it against invaders. two key aspects of adult behavioural identity: the main- ling (such as MAD, Medea and Schnurri) and chromatin Workers constitute the body tenance of chronological memory of neurotransmission regulation (such as GAGA factor) — are also associated or ‘soma’ of a eusocial events in postmitotic neurons and the maintenance of with caste-specific recruitment of the transcriptional co- insect colony. inherited memory of cellular identity, which is speci- activator CREB-binding protein (CBP) in C. floridanus40.

Royal jelly fied during larval development and transmitted through This suggests some degree of coordination between TF 3,7 A nutrient-rich secretion metamorphosis . Such an epigenetic model of behaviour (trans epigenetics) and chromatin (cis epigenetics) regu- produced by mandibular and complements our understanding that behavioural states latory processes, as CBP is the major acetyltransferase hypopharyngeal glands in correspond to patterns of synaptic connectivity encoded for a key histone residue, histone H3 lysine 27 (H3K27), honeybee nurses, which feed in neural networks53, as neuronal connectivity, synaptic in insects62 (FIG. 2a). Indeed, changes in H3K27 acetyla- it to larvae to induce their development into plasticity and firing sensitivity are themselves environ- tion (H3K27ac) correlate with changes in both CBP and (that is, virgin queens). mentally sensitive cell-specific phenotypes that depend mRNA expression between major worker and minor worker on dynamic non-genetic mechanisms to maintain state. castes in C. floridanus. Furthermore, developmental and Gamergates Consistent with this model, regulation of gene tran- neuronal genes show evolutionary increases in the num- A unique reproductive caste comprised of mated, fertile scription by established epigenetic mechanisms has key ber of CBP-binding sites in C. floridanus compared with 40 workers. In some ponerine roles in learning, memory and neuronal development their orthologues in D. melanogaster , which mirrors species (for example, in insects and mammals55,56. Although transcriptional the observation of increased evolutionary variability in Harpegnathos saltator), regulation is by no means the only key layer of regula- CREB-binding sites (see above). gamergates emerge from the tion of neuronal gene expression — subcellular mRNA The proposed role for CBP in morphological and existing cohort of workers 57,58 when a queen dies or is localization and local translation also have important behavioural plasticity in ants is consistent with observa- artificially removed from a and established roles — it is notable that transplantation tions in other organisms. For example, CBP is required for colony. of mRNA complements between neuronal cell types long-term memory formation in fruitflies and mice63,64,

NATURE REVIEWS | GENETICS VOLUME 15 | OCTOBER 2014 | 681

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

and defects in CBP, notably in the histone acetyltrans- environment through histone acetylation and associated ferase domain, cause intellectual disabilities in humans Trithorax group protein complexes that contain H3K4 (Rubinstein–Taybi syndrome)65. Histone acetylation, methyltransferase activity, and by antagonizing the activ- including H3K27ac, is a key regulator of transcription ity of Polycomb repressive complexes (PRCs), which and enhancer activity in neurons in fruitflies66 and catalyse trimethylation of H3K27 (H3K27me3)65,68. mice3,67. Therefore, CBP may confer epigenetic memory of gene transcription in neurons and other cell types by DNA methylation is linked to morphological and behav- maintaining a transcriptionally accessible chromatin ioural plasticity in eusocial insects. DNA methylation is a classic epigenetic process that is most widely known to maintain stably repressive chromatin in mammalian Box 3 | The insect neuroendocrine system, IIS signal transduction and JH X chromosome inactivation and genomic imprinting65,69, The neuroendocrine system is a mechanism involving secreted signals that allows distal as well as in the developmental silencing of promoters 65,70 cell–cell communication and that is found in all species with a nervous system. The throughout the genome . Specific CpG dinucleotides secreted signals, , are typically produced in endocrine glands and regulate a are targeted for methylation either through the recruit- range of cell and tissue types, including neurons, through which they can have a ment of the de novo DNA methyltransferase DNMT3 by substantial influence on behaviour148. TFs or histone modifiers (for example, G9a methyltrans- Perhaps the best-studied insect neuroendocrine signal transduction pathway ferase)70, or through propagation of pre-existing DNA involves Insulin/Insulin-like growth factor (IGF) signalling (IIS), which is a conserved methylation by the maintenance DNA methyltransferase 1 target of nutrition-mediated signals . Nutrient uptake is recognized by DNMT1 (REFS 65,70). Unlike in , in which insulin-producing cells in the brain or fat body, which respond by synthesizing and DNA methylation occurs globally and promoter meth- secreting ligands called Insulin-like peptides (ILPs) into the haemolymph. ILPs are 71 recognized by cognate ILP transmembrane receptors (InRs) that are expressed on all ylation represses gene transcription , in most insects 36 cells and in turn activate the intracellular IIS pathway, leading to altered gene — with a possible exception in termites — DNA meth- regulation. Overall, IIS has been implicated in the regulation of morphology, behaviour ylation occurs predominantly over transcribed genes, and ageing in eusocial insects101. However, the effect of IIS on caste fate is perhaps specifically exons37,38 (FIG. 2b). In contrast to some insects best understood in the honeybee, in which nurse workers synthesize, secrete and feed that lack one (for example, silkworms and red flour royal jelly to queen-destined larvae, hence establishing long-term differences in lack DNMT3 (REF. 72)) or both (for example, physiology, behaviour and reproduction in the adult queen. In addition to containing D. melanogaster13,72) key DNA methyltransferases, nutritional signals that activate IIS, one of the key functional components of royal jelly DNMT1 and DNMT3 are present and functional in is Royalactin (also known as Major royal jelly protein 1) — a 57 kDa protein that 73 37,42,74 38 47 eusocial insects , including bees , ants , social activates the Epidermal growth factor receptor (EGFR) , which in turn seems 41,75 36,76 149 wasps and termites , as well as in some solitary to activate a common target of IIS, the InR substrate (IRS) . 72 77 Recent studies have shown a functional link between IIS and the insect insects such as parasitoid wasps . These findings have neuroendocrine system through Juvenile hormone (JH)150. JH is synthesized in the spurred interest in studying potential functional roles of corpora allata, which is a pair of tissues in the insect brain that function as endocrine DNA methylation in caste fate and behavioural plasticity. glands. JH has a well-characterized role in regulating worker behaviour in honeybees, Indeed, a few recent studies have shown that DNA and this seems to be conserved in ants29. For example, treatment of larvae from several methylation can regulate both reproductive caste fate and species of the dimorphic myrmicine Pheidole with methoprene (which is a behaviour in eusocial insects. In bumblebees, chemical chemical analogue of JH) is sufficient both to induce major-destined workers and to inhibition of DNA methyltransferase activity promotes activate a conserved but latent developmental trajectory that expresses a super-soldier worker reproduction in queenless colonies42. Consistent 130 caste (FIG. 1c). Similarly, application of methoprene to larvae induces development of with this finding, queen larvae in honeybees have lower majors in the formicine ant Camponotus floridanus (D.F.S., unpublished observations) genome-wide methylation levels than worker larvae, and and development of queens in many ant and bee species, including Apis mellifera and Harpegnathos saltator22,151. knockdown of DNMT3 using RNA interference (RNAi) The methoprene-tolerant (Met) gene encodes a well-characterized transcription in worker larvae induces the development of queen-like factor (TF) and JH receptor152. A recent transcriptome analysis revealed that many IIS ovaries in adults, which mimics the effect of royal jelly and JH targets are differentially expressed in the brain6,101, and further functional treatment78. This indicates that DNA methylation might analyses suggested that the TF Ultraspiracle (USP), which dimerizes with the Ecdysone be required to inhibit queen development in honeybees, receptor, may mediate the effects of JH in the brain153. Among the known targets of JH is in which case royal jelly may function partly by remov- the well-studied egg yolk precursor protein Vitellogenin (Vg), the expression of which ing this inhibition. DNA methylation may also be medi- 152,154,155 has been closely linked to reproduction and fertility in various insects . In ated by the IIS pathway, both because royal jelly activates 156 eusocial insects, Vg has a key role in both queen reproduction and worker behaviour . IIS (BOX 3) and because IIS may regulate DNA methyla- Interestingly, this dual role of Vg involves an inversion of the regulatory relationship tion and histone modification in mammals through the between JH and Vg. At the late pupal stage in honeybee development, JH abundance correlates positively with Vg to promote reproduction, whereas in adults JH mitogen-activated protein kinase (MAPK) or phos- functionally inhibits Vg as the level of JH increases with age from nurses to foragers154. phoinositide 3‑kinase (PI3K)–AKT–glycogen synthase 79–82 Indeed, knockdown of Vg accelerates the transition to foraging in honeybees123. There kinase‑3 (GSK‑3) signalling pathways . Consistent is also evidence that both IIS and JH may be inhibited by Vg, which suggests that with this hypothesis, royal jelly also contains a func- behavioural regulation by this pathway may involve negative (that is, stabilizing) tional fatty acid histone deacetylase inhibitor (HDACi)83 feedback101,155. Unlike honeybees and ponerine ants (for example, H. saltator), the known as 10HDA, which should facilitate gene tran- genomes of which have one vg gene locus, multiple vg paralogues have been identified scription by increasing chromatin accessibility65. In in myrmicine ants (for example, Pogonomyrmex barbatus) and dolichoderine ants (for addition, among the Pogonomyrmex sp. seed harvester example, Linepithema humile). Expression analyses of these paralogues suggest that ants (BOX 2), obligately sterile interspecific hybrids show functional divergence of Vg is linked to behavioural specializations specifically in more lower levels of global DNA methylation than faculta- advanced ant species157. tively sterile conspecific progeny. This suggests that the

682 | OCTOBER 2014 | VOLUME 15 www.nature.com/reviews/genetics

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

Parthenogenetic regulation of worker caste fate involves an interaction Analyses of DNA methylation in the brains of hon- 84 Pertaining to , between genetic factors and DNA methylation . eybees that are engaged in nursing and foraging tasks which is a form of asexual Given these striking results, it is perhaps surprising identified several dozen genes with expression levels that reproduction that produces that genome-wide analyses have found relatively few changed inversely with DNA methylation status between viable embryos from eggs 27 without fertilization by sperm differentially methylated regions in the brains of age- behavioural groups . Moreover, foraging bees that were 27,37 — notably, haploid male matched adult queens and workers . Similarly, com- experimentally reverted to nursing recapitulated many production in Hymenoptera. In parison of whole-body DNA methylation patterns in two of the nursing-specific DNA methylation patterns, some parthenogenetic insects ant species revealed very few changes between worker which affect genes involved in learning, axon migration, such as Cerapachys biroi, castes in C. floridanus, and between workers and gamer- transcription and translation. Given the involvement of female reproductives can lay 38 86 diploid eggs by , gates in H. saltator ; however, substantial changes were DNA methylation in learning in mammals , methyla- thereby producing clonal found between males and females, between workers and tion patterns associated with foraging in eusocial insects female offspring. queens, and even between virgin and mature queens38. may facilitate learning or memory72. Interestingly, genes These results suggest that, unlike its role as a fairly stable involved in the regulation of chromatin accessibility27 and Thelytokous 85 The parthenogenetic epigenetic mark in mammals, DNA methylation may be JH signalling also showed differential DNA methyla- production of female offspring used in a more dynamic manner in eusocial insects, in tion, which suggests that DNA methylation may target from unfertilized eggs. which it changes with developmental stage, age and social genes with pleiotropic regulatory effects. In addition to context. In support of this view, a comparative analysis of regulating gene expression, DNA methylation also seems honeybee heads between worker and queen castes at lar- to regulate alternative splicing. Significant correlations val and adult stages identified nearly fivefold more differ- between DNA methylation and alternative splicing were entially methylated genes (DMGs) in larvae than in adults, reported in honeybees, ants and termites27,36–38 (FIG. 2b), and a larger proportion of DMGs were upregulated in the and RNAi knockdown of DNMT3 in honeybees led to worker larvae than in the queen larvae85. some changes in gene splicing patterns87. By contrast, no

a Neuroendocrine Neuron b signalling pathways (e.g. IIS) TF Off Off

Ex1 Ex2 Ex3

Repressive PTMs m7G AAAA m7G AAAA Histone modifier On On

TF TF Ex1 Ex2 Ex3

Chromatin regulator Active PTMs ncRNA m7G AAAA m7G AAAA m7G AAAA m7G AAAA m7G AAAA m7G AAAA m7G AAAA m7G AAAA Chromatin m7G AAAA m7G AAAA m7G AAAA m7G AAAA regulator m7G AAAA m7G AAAA

• Expression of neuropeptides Unmethylated cytosine Methylated cytosine • Synthesis of hormones • Activity of motor neurons Behavioural regulation Exon mRNA • Neuronal signal propagation m7G AAAA Ex1 Ex2 Ex3 m7G AAAA

Figure 2 | Epigenetic mechanisms of gene regulation in the insect histone H3 lysine 27 trimethylation (H3K27me3; red circles)) to active (for Nature Reviews | Genetics brain. a | The neuroendocrine system regulates gene expression in example, H3K27 acetylation; green circles). Here, chromatin is depicted the brain by targeting transcription factors (TFs), non-coding RNAs using the ‘beads on a string’ model, with DNA (black line) wrapped around (ncRNAs) and chromatin regulatory proteins, which involves histone histone octamers. b | Correlation of gene body DNA methylation with post-translational modifications (PTMs) and chromatin remodelling. TFs gene expression and alternative splicing is shown. At the genome-wide (such as Cyclic AMP response element-binding protein (CREB)) and level, hypermethylation predominantly occurs at genes with medium to ncRNAs operate partly through recruitment of histone modifiers (a high levels of transcription. In addition, inclusion of alternatively spliced specific class of chromatin regulatory proteins), such as CREB-binding exons normally correlates with hypomethylated regions. However, in protein (CBP), which is a transcriptional co-activator that has histone some cases, the opposite is also true, which suggests the recruitment of acetyltransferase activity. In this way, signalling pathways may alter the different factors to different gene loci38. IIS, Insulin/Insulin-like growth chromatin landscape around target genes from repressive (for example, factor signalling; m7G, 7-methyl-guanosine cap.

NATURE REVIEWS | GENETICS VOLUME 15 | OCTOBER 2014 | 683

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

association between DNA methylation levels and splicing fate determination. Furthermore, increased genetic patterns has been found in the solitary hymenopteran variation within a colony, which ostensibly decreases wasp Nasonia vitripennis77, which indicates the possi- kin relatedness, is paradoxically associated with more bility that certain taxonomically restricted functions of complex social structures. This has led to the conclu- DNA methylation may be involved in regulating plastic sion that genetic variation must be more evolutionarily behaviour in eusocial insects. favourable than previously thought92,98–101. Major worker Many studies on the effects of genetic variation have (Also known as a soldier). Gene regulation by non-coding RNAs. Various types of focused on reproductive caste ratio (that is, queen pro- A large worker produced in 98 some ant species, for example, ncRNAs, including microRNAs (miRNAs),­ enhancer duction) . However, these studies involved colonies Camponotus floridanus and RNAs (eRNAs) and long ncRNAs, participate in cell with unusual reproductive strategies for eusocial insects, Pheidole morrisi. Major fate determination, neuronal plasticity, embryogenesis including polymorphic queen castes and interspecific workers are typically and disease progression through transcriptional and hybridization102,103. Other studies have also examined aggressive and specialize in post-transcriptional gene regulation7,88. Conventionally, genetic effects on worker behaviour, which is a more nest defence and carrying heavy or large food items. ncRNAs are divided into short (<200 nucleotides) quantitative and variable phenotype than caste fate. and long (>200 nucleotides) classes. Both conserved and Perhaps the earliest of such reports linked age-dependent Minor worker lineage-specific short ncRNAs, including mi­RNAs, have division of labour in workers to allozyme isomorphs A small worker produced in been identified in ants and honeybees, and many of in the polyandrous honeybee A. mellifera104. Different some ant species, for example, 39,89 Camponotus floridanus and these short ncRNAs show caste-specific expression . colonies of the ant Leptothorax nevadensis rudis were Pheidole morrisi. Minor Interestingly, in C. floridanus and H. saltator, expression also shown to exhibit differences in the proportion of workers carry out most tasks in of CpG island-derived ncRNAs positively correlates with foraging workers of similar age105. More recently, genetic the nest, including foraging. the expression of adjacent protein-coding genes, many variation due to polyandry was linked to worker caste of which have roles in neuronal development39. Long polymorphism in Acromyrmex echinatior leaf-cutter DNA methyltransferase 99 106 An enzyme that catalyses ncRNAs share many characteristics with mRNAs, such as ants and P. badius harvester ants . Such variation DNA methylation. There a multiexonic structure, polyadenylation and transcrip- may not simply be regarded as a functional side effect, as are two functional DNA tion by RNA polymerase II (Pol II); many long ncRNAs honeybee colonies show a positive relationship between methyltransferase (DNMT) also have tissue-specific expression patterns90. As several colonial genetic variation and ergonomic efficiency classes in metazoa. DNMT1 carries out long ncRNAs can interact with both genomic DNA and (which is a proxy for fitness that is sensitive to evolu- 60 maintenance DNA methylation epigenetic regulators, they may recruit or stabilize epi- tionary selection) . Therefore, worker behaviour is also using a pre-methylated DNA genetic modifications at specific genomic loci in a simi- subject to genetic effects in both ant and bee lineages. sequence as a template, lar manner to TFs (although probably in cis)7 (FIG. 2a). Notably, these examples of genetic effects on caste whereas DNMT3 is responsible Although there has not yet been a comprehensive iden- and behaviour have been attributed to polyandry, which for de novo methylation of DNA. DNMT2 was tification of long ncRNAs in any eusocial insect, two long suggests that genetic variation per se is not a require- originally defined as a DNA ncRNAs were recently characterized in the honeybee ment for the expression of temporal or morphological methyltransferase but was in association with worker ovary size91. These prelimi- polyethism. Instead, it argues that certain polymorphic later correctly recognized as a nary findings suggest that the classes of short and long alleles can predispose or bias the probability that a given tRNA methyltransferase. ncRNAs might constitute yet another layer of epigenetic larva will develop into a particular caste or exhibit a par- CpG island regulation of caste identity in eusocial insects. ticular behaviour, and that the representation of these An intergenic genomic region alleles in a colony can be advantageous107. Although such that contains a greater Genetic effects on behavioural plasticity consideration of genetic effects may complicate straight- density of unmethylated CG Despite the clear role of the environment in regulating forward analyses of the role of environmental and epige- dinucleotides than expected compared with genome-wide caste fate and behaviour, the phenotype of an organism netic effects on behaviour, it remains an important factor density. CpG islands were emerges from the dynamic interplay between environ- that should be examined. To this end, it would be par- originally defined as regulatory mental and genetic factors21,53,92,93 (FIG. 1a): behavioural ticularly revealing to use a monogamous but dimorphic regions in mammals. As insects responses depend on both nervous system ontogeny and species, as may be found in the Camponotus or Pheidole largely lack intergenic DNA (BOX 2) methylation, it remains unclear immediate sensory perception; interpretation of sensory genera , to assess whether genetic variation that is whether CpG island-like cues depends on the expression of specific receptors, solely due to maternal heterozygosity is also associated sequences are functional in such as those involved in olfaction, gustation and vision; with caste fate or worker behaviour92,108. Nonetheless, insects. and long-term memory in neurons depends on proper the often-invoked metaphor that dichotomizes nature recruitment of chromatin regulators such as CBP94. In this and nurture should be replaced by an integrated view Allozyme A variant form of an enzyme context, genetic variation in sequences that regulate gene of nature plus nurture, in which both genetic and that is encoded by a different expression (for example, TF binding sites and ncRNAs) environmental factors have complementary modula- allele of the same genetic can be used to modulate the thresholds that define an tory roles in directing the ontogenetic trajectory of an locus. Allozyme analysis was individual’s developmental or behavioural response individual98 (FIG. 1a). used to infer genetic variation 95 (FIG. 1c) before direct DNA sequencing to environmental factors . Indeed, caste fate became widely used. in some species, including Pogonomyrmex badius, Transgenerational epigenetic inheritance cursor and Wasmannia auropunctata, is not and inclusive fitness theories provide an Kin selection sensitive to environmental variation. Such observations evolutionary explanation for the success of eusocial- A form of natural selection of environmentally invariant caste polymorphism led ity109,110. Individuals of a colony may help to increase that favours the reproductive 96,97 success of relatives even at a to early speculations and several recent demonstra- their colony’s fitness indirectly by refraining from 92,98 cost to an individual’s own tions that genetic variation has an important (albeit reproduction in favour of rearing successive genera- survival and reproduction. fairly uncommon) and complementary role in caste tions of their siblings or relatives. However, as these

684 | OCTOBER 2014 | VOLUME 15 www.nature.com/reviews/genetics

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

117 Inclusive fitness non-reproductive workers are functionally sterile and humile ants . Initial limited support for ASM was 38 The sum of the reproductive otherwise phenotypically distinct from their parents, found in C. floridanus and H. saltator ants , in which fitness of an individual and reproductive conflicts are expected to arise between gene expression levels correlated with increased DNA the indirect fitness received parental and sibling genomes. In ants, bees and wasps, methylation in an allele-dependent manner. However, by relatives other than the individual’s own offspring that such genomic conflicts are exacerbated owing to it was not determined whether the observed ASM was were produced as a result of haplodiploid sex determination, which creates relatedness caused by parental imprinting. Parental effects may also help from the individual. asymmetries between maternal, paternal and offspring be achieved by mechanisms other than ASM, such as dif- genotypes. One possible resolution of genomic conflicts ferential deposition of maternal RNAs118, inheritance of may be achieved through the parental regulation of off- parental chromatin state119 or both120 (FIG. 1a). Evaluating spring phenotypes, such as fertility potential, through these possible mechanisms for resolving genomic con- mechanisms of transgenerational epigenetic inheritance, flicts may provide a key perspective on the broader use for example, imprinting by allele-specific DNA meth- of transgenerational epigenetic inheritance in . ylation (ASM)69,111–115 (FIG. 2). By carrying out reciprocal Moreover, as many epigenetic mechanisms are sensitive crosses between two genetic backgrounds, recent stud- to environmental conditions, these studies may also pro- ies have shown substantial parental effects on worker vide new insights into the influence of the environment defence behaviour in A. mellifera honeybees116, as well in generating phenotypic plasticity in both current and as on nursing and foraging behaviours in Linepithema future generations121,122. a b All eusocial insects Wasps Bees Ants

C. biroi A. mellifera Bee Ant P. metricus

M. pharaonis Termites

L. albipes Termite Wasp

Z. nevadensis H. saltator Small-molecule pharmaceutical treatment or RNAi treatment Genetic manipulation (e.g. CRISPR–Cas)

Germline mutation or transgenic female and male reproductives

Gal4 driver × UAS transgene

Somatic changes of gene expression and protein activity Temporally controlled or tissue-specific knockdown or overexpression Figure 3 | Genetic approaches in eusocial insects. a | Pharmaceutical temporally controlled or tissue-specific overexpression and knockdown). compounds and RNA interference (RNAi) are currently used to change Blue scale bars represent 1 mm, and blackNature scale Reviews bars represent | Genetics gene expression in somatic cells to alter caste fate and behaviour in 5 mm. CRISPR–Cas, clustered regularly interspaced short palindromic eusocial insects. b | A schematic is shown for generating mutant and repeat–CRISPR-associated. C. biroi image courtesy of D. Kronauer (The transgenic lines by germline genetic manipulation in lineages of wasps, Rockefeller University, New York, USA); C. floridanus and Z. nevadensis termites, bees and ants, such as Polistes metricus, Zootermopsis nevadensis, images courtesy of A. Smith (University of Illinois at Urbana-Champaign, Apis mellifera, Lasioglossum albipes, Cerapachys biroi, USA); H. saltator image courtesy of C. Penick (North Carolina State Monomorium pharaonis and Harpegnathos saltator. With appropriate University, USA); L. albipes image courtesy of S. Kocher (Museum of manipulation, the reproductive females in these species can either mate Comparative Zoology, Massachusetts, USA); M. pharaonis image courtesy or be inseminated artificially to generate progeny from controlled genetic of L. Pontieri (University of Copenhagen, Denmark); P. metricus image crosses, which are required for some genetic manipulations (for example, courtesy of P. Coin (Durham, North Carolina, USA).

NATURE REVIEWS | GENETICS VOLUME 15 | OCTOBER 2014 | 685

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

Haplodiploid Eusocial insects as genetic models have also offered important molecular interpretations A genetic system of sex Long-term success in addressing the epigenetic basis of a response threshold model for colonial division of determination that is mainly of eusociality and in establishing eusocial insects as labour131 (FIG. 1). Of course, these admittedly few pub- found in the insect order bona fide model organisms will depend on the continued lished studies represent only the tip of the iceberg for Hymenoptera, including (BOX 2; FIG. 3) ants, bees and wasps. development of genetic and genomic tools , epigenetics studies in eusocial insects, and the major Hymenopteran females such as the ability to reduce gene activity using RNAi or questions that drive the fields of behavioural epigenet- are diploid (that is, they pharmacological compounds, or to manipulate genomes ics and sociobiology remain unanswered. What are the have two complete sets of and epigenomes through transposon-mediated and gene principle aspects of a genetic and epigenetic architecture chromosomes), whereas males targeting approaches for mutagenesis and transgenesis that direct the production of distinct morphological and are normally haploid and have only one set of chromosomes. (for example, the CRISPR–Cas system). The use of RNAi behavioural castes? What is the extent to which epige- 3 Some species, such as the fire has been successfully demonstrated in eusocial insects, netic modifications, such as histone acetylation , drive ant Solenopsis invicta, also including honeybees47,78,123, ants32,124, social wasps125 and age-, caste- or social context-dependent behavioural plas- produce viable diploid males. termites126. Pharmaceutical compounds may prove to ticity in brains? Does the evolution of sociality require Relatedness asymmetries be especially fruitful to inhibit or activate epigenetic a distinct genetic toolkit or simply the elaboration and Differences in the degree of regulators of pathways that are not amenable to RNAi, tuning of existing conserved pathways (for example, the genetic similarity between as the appreciation that epigenetic regulatory proteins IIS–JH, CREB and CBP pathways)? Do reproductive parents and offspring that arise have crucial roles in a range of human diseases has led to eusocial insects use epigenetic mechanisms to coordinate in eusocial insects owing to the development of many compounds that target various division of labour transgenerationally in offspring? the haplodiploid mode of sex 94,127 determination and that is conserved pathways . A strategic combination of functional characteriza- exacerbated by polyandry The ability to carry out controlled genetic crosses tion, experimental and genetic manipulation, as well as (a form of polygamy in and other straightforward genetic manipulations — comparison of key genes, pathways and genomes at dif- which a female mates with historically, the primary consideration for granting the ferent scales of biological organization (cells, tissues and multiple males). ‘model organism’ status — remains one of the greatest colonies), using several taxonomically diverse eusocial CRISPR–Cas challenges for many eusocial insect species owing to ‘model organisms’ will be crucial to resolve these ques- (Clustered regularly difficulties in laboratory breeding and inherent limita- tions. In practice, bee species such as A. mellifera have interspaced short palindromic tions in generating large numbers of reproductive indi- provided many of the greatest successes in experimen- repeat–CRISPR-associated). viduals. Despite the large sizes of many eusocial insect tal manipulation and breeding, partly as a result of their A technique for generating 6 site-specific mutant or colonies, which can exceed 10 individuals in nature comparatively long history as ecological and molecu- 4 132 transgenic organisms using (and at least 10 individuals in the laboratory), normally lar research models . In addition to bees, wasps are the Cas9 protein–guide RNA only one to several females can reproduce. However, increasingly being used as behavioural models that complex to generate mutations several specific eusocial species — such as the bees are representative of early stages of eusociality, whereas or to direct exogenous DNA to specific genomic regions where A. mellifera and L. albipes, H. saltator, C. biroi the variety of ants fully spans the range of eusociality, it is incorporated. and Monomorium pharaonis, the paper wasp Polistes met- especially at the most derived and socially complex end ricus and the termite Z. nevadensis — show a particularly of the spectrum. In addition, ants can be reared and bred unique potential either for the experimental generation completely in a laboratory, thereby ensuring the greatest of large numbers of reproductive individuals or for con- possible control over genetic and environmental vari- trolled breeding (BOX 2). Indeed, successful generation of ation, in addition to offering substantial advantages in

transgenic F1 males has recently been demonstrated in cost, convenience and efficiency. However, many ants A. mellifera by adapting and optimizing techniques have a particular disadvantage in terms of long genera- that involve piggyBac transposons128. This achievement tion times (BOX 2). Nonetheless, this spectrum of model suggests that the first transgenic animals from other choices should provide an unprecedented opportunity eusocial lineages may be imminent. to understand the roles and interactions of genetic, epi- genetic and environmental factors on caste fate, social Conclusion behaviour and division of labour in eusocial insects. The recent genomic and epigenomic studies discussed Given the burgeoning interest in eusocial insects (at least here have provided novel insights into the genetic, epige- as measured by the growing number of related review netic and molecular underpinnings of social behaviour articles75,129,133–135), we envision a rich future for eusocial and evolution of eusocial insect lineages27,39,40,129,130, and insects as models for behavioural epigenetics.

1. Gerozissis, K. Brain insulin: regulation, mechanisms 6. Zayed, A. & Robinson, G. E. Understanding the 11. Takahashi, J. S., Pinto, L. H. & Vitaterna, M. H. of action and functions. Cell. Mol. Neurobiol. 23, relationship between brain gene expression and social Forward and reverse genetic approaches to behavior 1–25 (2003). behavior: lessons from the . Annu. Rev. in the mouse. Science 264, 1724–1733 (1994). 2. Margulies, C., Tully, T. & Dubnau, J. Deconstructing Genet. 46, 591–615 (2012). 12. Nottebohm, F. The road we travelled: discovery, memory in Drosophila. Curr. Biol. 15, R700–R713 7. Bonasio, R. Emerging topics in epigenetics: ants, choreography, and significance of brain replaceable (2005). brains, and noncoding RNAs. Ann. NY Acad. Sci. neurons. Ann. NY Acad. Sci. 1016, 628–658 (2004). 3. Dulac, C. Brain function and chromatin plasticity. 1260, 14–23 (2012). 13. Raddatz, G. et al. Dnmt2‑dependent methylomes lack Nature 465, 728–735 (2010). 8. Acar, M., Becskei, A. & van Oudenaarden, A. defined DNA methylation patterns. Proc. Natl Acad. 4. Kim, J. & Eberwine, J. RNA: state memory and Enhancement of cellular memory by reducing Sci. USA 110, 8627–8631 (2013). mediator of cellular phenotype. Trends Cell Biol. 20, stochastic transitions. Nature 435, 228–232 14. Crabbe, J. C., Wahlsten, D. & Dudek, B. C. Genetics of 311–318 (2010). (2005). mouse behavior: interactions with laboratory 5. Giurfa, M. & Sandoz, J. C. learning and 9. Bonasio, R., Tu, S. & Reinberg, D. Molecular signals of environment. Science 284, 1670–1672 (1999). memory: fifty years of olfactory conditioning of the epigenetic states. Science 330, 612–616 (2010). 15. Bier, E. & McGinnis, W. in Inborn Errors of proboscis extension response in honeybees. Learn. 10. Jasinska, A. J. & Freimer, N. B. The complex genetic Development (eds Epstein, C. J., Erikson, R. P. & Mem. 19, 54–66 (2012). basis of simple behavior. J. Biol. 8, 71 (2009). Wynshaw-Boris, A.) 25–45 (Oxford Univ. Press, 2003).

686 | OCTOBER 2014 | VOLUME 15 www.nature.com/reviews/genetics

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

16. Hunt, G. J. Flight and fight: a comparative view of 38. Bonasio, R. et al. Genome-wide and caste-specific DNA 61. Chandrasekaran, S. et al. Behavior-specific changes in the neurophysiology and genetics of honey bee methylomes of the ants Camponotus floridanus and transcriptional modules lead to distinct and defensive behavior. J. Insect Physiol. 53, 399–410 Harpegnathos saltator. Curr. Biol. 22, 1755–1764 predictable neurogenomic states. Proc. Natl Acad. Sci. (2007). (2012). USA 108, 18020–18025 (2011). 17. Withers, G. S., Fahrbach, S. E. & Robinson, G. E. This is the first study of DNA methyaltion in ants, 62. Tie, F. et al. CBP-mediated acetylation of histone H3 Effects of experience and juvenile hormone on the which reveals several conserved characteristics in lysine 27 antagonizes Drosophila Polycomb silencing. organization of the mushroom bodies of honey bees. two species, including non-CpG methylation, Development 136, 3131–3141 (2009). J. Neurobiol. 26, 130–144 (1995). enrichment of methylcytosine in exons, ASM and 63. Graff, J. & Tsai, L. H. Histone acetylation: molecular 18. Gronenberg, W. The trap-jaw mechanism in the association of DNA methylation with alternative mnemonics on the chromatin. Nature Rev. Neurosci. dacetine ants armigerum and splicing. 14, 97–111 (2013). sp. J. Exp. Biol. 199, 2021–2033 (1996). 39. Simola, D. F. et al. Social insect genomes exhibit 64. Hirano, Y. et al. Fasting launches CRTC to facilitate 19. Ehmer, B. & Gronenberg, W. Segregation of visual dramatic evolution in gene composition and regulation long-term memory formation in Drosophila. Science input to the mushroom bodies in the honeybee while preserving regulatory features linked to sociality. 339, 443–446 (2013). (Apis mellifera). J. Comp. Neurol. 451, 362–373 Genome Res. 23, 1235–1247 (2013). 65. Allis, C. D., Jenuwein, T., Reinberg, D. & Caparros, M. (2002). 40. Simola, D. F. et al. A chromatin link to caste identity in L. (eds) Epigenetics (Cold Spring Harbor Laboratory 20. Ehmer, B. & Gronenberg, W. Mushroom body the carpenter ant Camponotus floridanus. Genome Press, 2007). volumes and visual interneurons in ants: comparison Res. 23, 486–496 (2013). 66. Taylor, J. P. et al. Aberrant histone acetylation, altered between sexes and castes. J. Comp. Neurol. 469, This is the first study to analyse the role of histone transcription, and retinal degeneration in a Drosophila 198–213 (2004). modifications in eusocial insects, which reveals a model of polyglutamine disease are rescued by CREB- 21. Rossler, W. & Zube, C. Dual olfactory pathway potential regulatory role for H3K27ac and CBP in binding protein. Genes Dev. 17, 1463–1468 (2003). in Hymenoptera: evolutionary insights from sex and worker caste differentiation. 67. Kim, T. K. et al. Widespread transcription at neuronal comparative studies. Struct. Dev. 40, 41. Weiner, S. A. et al. A survey of DNA methylation activity-regulated enhancers. Nature 465, 182–187 349–357 (2011). across social insect species, life stages, and castes (2010). References 17–21 analyse brain morphology and reveals abundant and caste-associated methylation in 68. Ringrose, L. & Paro, R. Polycomb/Trithorax response neuronal connections, which are associated with a primitively social wasp. Naturwissenschaften 100, elements and epigenetic memory of cell identity. learning and behaviour in exemplary eusocial 795–799 (2013). Development 134, 223–232 (2007). insects. 42. Amarasinghe, H. E., Clayton, C. I. & Mallon, E. B. 69. Ferguson-Smith, A. C. Genomic imprinting: the 22. Nijhout, H. F. & Wheeler, D. E. Juvenile hormone and Methylation and worker reproduction in the bumble- emergence of an epigenetic paradigm. Nature Rev. the physiological basis of insect polymorphisms. bee (Bombus terrestris). Proc. R. Soc. B 281, Genet. 12, 565–575 (2011). Q. Rev. Biol. 57, 109–133 (1982). 20132502 (2014). 70. Cedar, H. & Bergman, Y. Linking DNA methylation and 23. Wheeler, D. E. Developmental and physiological 43. Robinson, G. E., Grozinger, C. M. & Whitfield, C. W. histone modification: patterns and paradigms. Nature determinants of caste in social Hymenoptera: Sociogenomics: social life in molecular terms. Nature Rev. Genet. 10, 295–304 (2009). evolutionary implications. Am. Naturalist 128, Rev. Genet. 6, 257–270 (2005). 71. Bird, A. DNA methylation patterns and epigenetic 13–34 (1986). 44. Hölldobler, B. & Wilson, E. O. The Ants (Belknap Press, memory. Genes Dev. 16, 6–21 (2002). 24. Seeley, T. D. Honeybee Democracy (Princeton Univ. 1990). 72. Lyko, F. & Maleszka, R. Insects as innovative models Press, 2010). 45. Liebig, J., Peeters, C., Oldham, N. J., Markstadter, C. for functional studies of DNA methylation. Trends 25. Liang, Z. S. et al. Molecular determinants of scouting & Hölldobler, B. Are variations in cuticular Genet. 27, 127–131 (2011). behavior in honey bees. Science 335, 1225–1228 hydrocarbons of queens and workers a reliable signal 73. Kronforst, M. R., Gilley, D. C., Strassmann, J. E. & (2012). of fertility in the ant Harpegnathos saltator? Proc. Queller, D. C. DNA methylation is widespread across 26. Robinson, G. E. Regulation of division of labor in Natl Acad. Sci. USA 97, 4124–4131 (2000). social Hymenoptera. Curr. Biol. 18, R287–R288 (2008). insect societies. Annu. Rev. Entomol. 37, 637–665 46. Grozinger, C. M., Sharabash, N. M., Whitfield, C. W. & 74. Wang, Y. et al. Functional CpG methylation system in a (1992). Robinson, G. E. -mediated gene expression social insect. Science 314, 645–647 (2006). 27. Herb, B. R. et al. Reversible switching between in the honey bee brain. Proc. Natl Acad. Sci. USA 100 This is the first report to show that a social insect epigenetic states in honeybee behavioral subcastes. (Suppl. 2), 14519–14525 (2003). has a fully functional methylation system. Nature Neurosci. 15, 1371–1373 (2012). 47. Kamakura, M. Royalactin induces queen differentiation 75. Weiner, S. A. & Toth, A. L. Epigenetics in social insects: This is a comparative analysis of DNA methylation in honeybees. Nature 473, 478–483 (2011). a new direction for understanding the evolution of profiles between honeybee nurses, foragers and 48. Liebig, J., Peeters, C. & Hölldobler, B. castes. Genet. Res. Int. 2012, 609810 (2012). reverted foragers; it shows, for the first time, the limits the number of reproductives in a ponerine ant. 76. Glastad, K. M., Hunt, B. G. & Goodisman, M. A. evidence of reversible epigenetic changes associated Proc. R. Soc. B 266, 1865–1870 (1999). Evidence of a conserved functional role for DNA with behavioural states in eusocial insects. 49. Endler, A. et al. Surface hydrocarbons of queen eggs methylation in termites. Insect Mol. Biol. 22, 28. Liebig, J., Hölldobler, B. & Peeters, C. Are ant workers regulate worker reproduction in a social insect. 143–154 (2013). capable of colony foundation? Naturwissenschaften Proc. Natl Acad. Sci. USA 101, 2945–2950 (2004). 77. Wang, X. et al. Function and evolution of DNA 85, 133–135 (1998). 50. Smith, A. A., Holldober, B. & Liebig, J. Cuticular methylation in Nasonia vitripennis. PLoS Genet. 9, 29. Penick, C. A., Liebig, J. & Brent, C. S. Reproduction, hydrocarbons reliably identify cheaters and allow e1003872 (2013). dominance, and caste: endocrine profiles of queens enforcement of altruism in a social insect. Curr. Biol. 78. Kucharski, R., Maleszka, J., Foret, S. & Maleszka, R. and workers of the ant Harpegnathos saltator. 19, 78–81 (2009). Nutritional control of reproductive status in honeybees J. Comp. Physiol. A Neuroethol Sens. Neural Behav. 51. Whitfield, C. W., Cziko, A. M. & Robinson, G. E. Gene via DNA methylation. Science 319, 1827–1830 (2008). Physiol. 197, 1063–1071 (2011). expression profiles in the brain predict behavior in This paper presents the first functional evidence of 30. Michener, C. D. The Bees of the World (Johns Hopkins individual honey bees. Science 302, 296–299 (2003). the regulatory role of DNA methylation in Univ. Press, 2000). This pioneering genome-wide analysis reveals that regulating caste fate in eusocial insects. 31. Honeybee Genome Sequencing, C. Insights into social gene expression profiles are closely assoicated with 79. Gazin, C., Wajapeyee, N., Gobeil, S., Virbasius, C. M. & insects from the genome of the honeybee Apis worker behaviours in honeybees. Green, M. R. An elaborate pathway required for mellifera. Nature 443, 931–949 (2006). 52. Toth, A. L. et al. Wasp gene expression supports an Ras-mediated epigenetic silencing. Nature 449, The first eusocial insect genome was analyzed in evolutionary link between maternal behavior and 1073–1077 (2007). honeybees. The honeybee genome, plus the eusociality. Science 318, 441–444 (2007). 80. Popkie, A. P. et al. Phosphatidylinositol 3‑kinase genomes of eight ant species, halictid bees and 53. Menzel, R. The honeybee as a model for (PI3K) signaling via glycogen synthase kinase‑3 dampwood termites (reference 32–36), laid a understanding the basis of cognition. Nature Rev. (Gsk‑3) regulates DNA methylation of imprinted loci. foundation for further epigenetic analyses. Neurosci. 13, 758–768 (2012). J. Biol. Chem. 285, 41337–41347 (2010). 32. Bonasio, R. et al. Genomic comparison of the ants 54. Evans, J. D. & Wheeler, D. E. Gene expression and the 81. Marks, H. et al. The transcriptional and epigenomic Camponotus floridanus and Harpegnathos saltator. evolution of insect polyphenisms. Bioessays 23, foundations of ground state pluripotency. Cell 149, Science 329, 1068–1071 (2010). 62–68 (2001). 590–604 (2012). 33. Gadau, J. et al. The genomic impact of 100 million 55. Borrelli, E., Nestler, E. J., Allis, C. D. & Sassone-Corsi, P. 82. Tee, W. W., Shen, S. S., Oksuz, O., Narendra, V. & years of social evolution in seven ant species. Trends Decoding the epigenetic language of neuronal Reinberg, D. Erk1/2 activity promotes chromatin Genet. 28, 14–21 (2012). plasticity. Neuron 60, 961–974 (2008). features and RNAPII phosphorylation at 34. Oxley, P. R. et al. The genome of the clonal raider ant 56. Kim, J. B. et al. Direct reprogramming of human neural developmental promoters in mouse ESCs. Cell 156, Cerapachys biroi. Curr. Biol. 24, 451–458 (2014). stem cells by OCT4. Nature 461, 649–643 (2009). 678–690 (2014). 35. Kocher, S. D. et al. The draft genome of a socially 57. Job, C. & Eberwine, J. Localization and translation of 83. Spannhoff, A. et al. Histone deacetylase inhibitor polymorphic halictid bee, Lasioglossum albipes. mRNA in dendrites and axons. Nature Rev. Neurosci. activity in royal jelly might facilitate caste switching in Genome Biol. 14, R142 (2013). 2, 889–898 (2001). bees. EMBO Rep. 12, 238–243 (2011). 36. Terrapon, N. et al. Molecular traces of alternative 58. True, H. L., Berlin, I. & Lindquist, S. L. Epigenetic 84. Smith, C. R. et al. Patterns of DNA methylation in social organization in a termite genome. Nature regulation of translation reveals hidden genetic development, division of labor and hybridization Commun. 5, 3636 (2014). variation to produce complex traits. Nature 431, in an ant with genetic caste determination. PLoS ONE 37. Lyko, F. et al. The honey bee epigenomes: differential 184–187 (2004). 7, e42433 (2012). methylation of brain DNA in queens and workers. 59. Sul, J. Y. et al. Transcriptome transfer produces a 85. Foret, S. et al. DNA methylation dynamics, metabolic PLoS Biol. 8, e1000506 (2010). predictable cellular phenotype. Proc. Natl Acad. Sci. fluxes, gene splicing, and alternative phenotypes This is the first brain methylome study in eusocial USA 106, 7624–7629 (2009). in honey bees. Proc. Natl Acad. Sci. USA 109, insects, which highlights the differentially 60. Page, R. E. & Fondrk, M. K. The effects of colony level 4968–4973 (2012). methylated genes between reproductive and selection on the social-organization of honey-bee 86. Lockett, G. A., Wilkes, F. & Maleszka, R. Brain non-reproductive castes, and the potential role (Apis-mellifera L) colonies - colony level components of plasticity, memory and neurological disorders: an of DNA methylation in modulating alternative pollen hoarding. Behav. Ecol. Sociobiol. 36, 135–144 epigenetic perspective. Neuroreport 21, 909–913 splicing. (1995). (2010).

NATURE REVIEWS | GENETICS VOLUME 15 | OCTOBER 2014 | 687

© 2014 Macmillan Publishers Limited. All rights reserved REVIEWS

87. Li‑Byarlay, H. et al. RNA interference knockdown of 112. Queller, D. C. Theory of genomic imprinting conflict in 137. Linksvayer, T. A. in Encyclopedia of Animal Behavior DNA methyl-transferase 3 affects gene alternative social insects. BMC Evol. Biol. 3, 15 (2003). (eds Breed, M. D. & Moore, J.) 358–362 (Academic splicing in the honey bee. Proc. Natl Acad. Sci. USA 113. Linksvayer, T. A. & Wade, M. J. The evolutionary origin Press, 2010). 110, 12750–12755 (2013). and elaboration of sociality in the aculeate 138. LaPolla, J. S., Dlussky, G. M. & Perrichot, V. Ants and 88. Rinn, J. L. & Chang, H. Y. Genome regulation by long Hymenoptera: maternal effects, sib-social effects, and the fossil record. Annu. Rev. Entomol. 58, 609–630 noncoding RNAs. Annu. Rev. Biochem. 81, 145–166 heterochrony. Q. Rev. Biol. 80, 317–336 (2005). (2013). (2012). 114. Kronauer, D. J. C. Genomic imprinting and kinship in 139. Ferreira, P. G. et al. Transcriptome analyses of 89. Greenberg, J. K. et al. Behavioral plasticity in honey the social Hymenoptera: what are the predictions? primitively eusocial wasps reveal novel insights into bees is associated with differences in brain microRNA J. Theor. Biol. 254, 737–740 (2008). the evolution of sociality and the origin of alternative transcriptome. Genes Brain Behav. 11, 660–670 115. Drewell, R. A., Lo, N., Oxley, P. R. & Oldroyd, B. P. phenotypes. Genome Biol. 14, R20 (2013). (2012). Kin conflict in insect societies: a new epigenetic 140. Johnson, B. R. et al. Phylogenomics resolves 90. Cabili, M. N. et al. Integrative annotation of human perspective. Trends Ecol. Evol. 27, 367–373 (2012). evolutionary relationships among ants, bees, and large intergenic noncoding RNAs reveals global 116. Guzman-Novoa, E. et al. Paternal effects on the wasps. Curr. Biol. 23, 2058–2062 (2013). properties and specific subclasses. Genes Dev. 25, defensive behavior of honeybees. J. Hered. 96, 141. Schmidt, A. M., Linksvayer, T. A., Boomsma, J. J. & 1915–1927 (2011). 376–380 (2005). Pedersen, J. S. Queen-worker caste ratio depends on 91. Humann, F. C., Tiberio, G. J. & Hartfelder, K. 117. Libbrecht, R. & Keller, L. Genetic compatibility affects colony size in the pharaoh ant (Monomorium Sequence and expression characteristics of long division of labor in the Argentine ant Linepithema pharaonis). Insect Soc. 58, 139–144 (2011). noncoding RNAs in honey bee caste development — humile. Evolution 67, 517–524 (2013). 142. St Johnston, D. The art and design of genetic screens: potential novel regulators for transgressive ovary size. 118. Khila, A. & Abouheif, E. Reproductive constraint is a Drosophila melanogaster. Nature Rev. Genet. 3, PLoS ONE 8, e78915 (2013). developmental mechanism that maintains social 176–188 (2002). 92. Keller, L. Adaptation and the genetics of social harmony in advanced ant societies. Proc. Natl Acad. 143. Kile, B. T. & Hilton, D. J. The art and design of genetic behaviour. Phil. Trans. R. Soc. B 364, 3209–3216 Sci. USA 105, 17884–17889 (2008). screens: mouse. Nature Rev. Genet. 6, 557–567 (2009). 119. Greer, E. L. et al. Transgenerational epigenetic (2005). 93. Wang, J. et al. A Y‑like social chromosome causes inheritance of longevity in Caenorhabditis elegans. 144. Schmidt, A. M., Linksvayer, T. A., Boomsma, J. J. & alternative colony organization in fire ants. Nature Nature 479, 365–371 (2011). Pedersen, J. S. No benefit in diversity? The effect of 493, 664–668 (2013). 120. Grossniklaus, U., Vielle-Calzada, J. P., Hoeppner, M. A. genetic variation on survival and disease resistance 94. Vecsey, C. G. et al. Histone deacetylase inhibitors & Gagliano, W. B. Maternal control of embryogenesis in a polygynous social insect. Ecol. Entomol. 36, enhance memory and synaptic plasticity via by MEDEA, a Polycomb group gene in Arabidopsis. 751–759 (2011). CREB:CBP-dependent transcriptional activation. Science 280, 446–450 (1998). 145. Baer, B. & Schmid-Hempel, P. The artificial J. Neurosci. 27, 6128–6140 (2007). 121. Snell-Rood, E. C. An overview of the evolutionary insemination of bumblebee queens. Insect Soc. 47, 95. Beshers, S. N. & Fewell, J. H. Models of division causes and consequences of behavioural plasticity. 183–187 (2000). of labor in social insects. Annu. Rev. Entomol. 46, Animal Behav. 85, 1004–1011 (2013). 146. Kocher, S. D., Tarpy, D. R. & Grozinger, C. M. 413–440 (2001). 122. Snell-Rood, E. C., Troth, A. & Moczek, A. P. The effects of mating and instrumental insemination 96. Wheeler, W. M. & Weber, N. A. Mosaics and Other DNA methylation as a mechanism of nutritional on queen honey bee flight behaviour and gene Anomalies Among Ants (Harvard Univ. Press, 1937). plasticity: limited support from horned beetles. expression. Insect Mol. Biol. 19, 153–162 (2010). 97. Kerr, W. E. Evolution of the mechanism of caste J. Exp. Zool. B Mol. Dev. Evol. 320, 22–34 (2013). 147. den Boer, S. P. A., Boomsma, J. J. & Baer, B. determination in the genus Melipona. Evolution 4, 123. Nelson, C. M., Ihle, K. E., Fondrk, M. K., Page, R. E. & A technique to artificially inseminate leafcutter ants. 7–13 (1950). Amdam, G. V. The gene vitellogenin has multiple Insect Soc. 60, 111–118 (2013). 98. Schwander, T., Lo, N., Beekman, M., Oldroyd, B. P. & coordinating effects on social organization. PLoS Biol. 148. Hartenstein, V. The neuroendocrine system of Keller, L. Nature versus nurture in social insect caste 5, e62 (2007). : a developmental and evolutionary differentiation. Trends Ecol. Evol. 25, 275–282 (2010). 124. Ratzka, C., Gross, R. & Feldhaar, H. Systemic gene perspective. J. Endocrinol. 190, 555–570 (2006). This review summarizes the evidence of knockdown in Camponotus floridanus workers by 149. Mutti, N. S. et al. IRS and TOR nutrient-signaling environmental and genetic effects on caste feeding of dsRNA. Insect Soc. 60, 475–484 (2013). pathways act via juvenile hormone to influence honey determination, and emphasizes the role of genetic 125. Hunt, J. H., Mutti, N. S., Havukainen, H., bee caste fate. J. Exp. Biol. 214, 3977–3984 (2011). variation on queen development in various eusocial Henshaw, M. T. & Amdam, G. V. Development of an 150. Tatar, M., Bartke, A. & Antebi, A. The endocrine insect species. RNA interference tool, characterization of its target, regulation of aging by insulin-like signals. Science 99. Hughes, W. O. H., Sumner, S., Van Borm, S. & and an ecological test of caste differentiation in the 299, 1346–1351 (2003). Boomsma, J. J. Worker caste polymorphism has a eusocial wasp polistes. PLoS ONE 6, e26641 (2011). 151. Penick, C. A., Prager, S. S. & Liebig, J. Juvenile genetic basis in Acromyrmex leaf-cutting ants. 126. Zhou, X. G., Wheeler, M. M., Oi, F. M. & Scharf, M. E. hormone induces queen development in late-stage Proc. Natl Acad. Sci. USA 100, 9394–9397 (2003). RNA interference in the termite Reticulitermes larvae of the ant Harpegnathos saltator. J. Insect 100. Schwander, T. & Keller, L. Genetic compatibility affects flavipes through ingestion of double-stranded RNA. Physiol. 58, 1643–1649 (2012). queen and worker caste determination. Science 322, Insect Biochem. Mol. Biol. 38, 805–815 (2008). 152. Jindra, M., Palli, S. R. & Riddiford, L. M. The juvenile 552 (2008). 127. Filippakopoulos, P. et al. Selective inhibition of BET hormone signaling pathway in insect development. 101. Smith, C. R., Toth, A. L., Suarez, A. V. & bromodomains. Nature 468, 1067–1073 (2010). Annu. Rev. Entomol. 58, 181–204 (2013). Robinson, G. E. Genetic and genomic analyses of the 128. Schulte, C., Theilenberg, E., Muller-Borg, M., 153. Ament, S. A. et al. The transcription factor division of labour in insect societies. Nature Rev. Gempe, T. & Beye, M. Highly efficient integration and Ultraspiracle influences honey bee social behavior and Genet. 9, 735–748 (2008). expression of piggyBac-derived cassettes in the behavior-related gene expression. PLoS Genet. 8, This is a thorough review on the genes and honeybee (Apis mellifera). Proc. Natl Acad. Sci. USA e1002596 (2012). molecular pathways that are known to regulate 111, 9003–9008 (2014). 154. Page, R. E. Jr, Scheiner, R., Erber, J. & Amdam, G. V. caste determination and worker behavioural This study is the first to show transgenics in a The development and evolution of division of labor and transitions in honeybees and other eusocial insects. eusocial insect, which allows sophisticated genetic foraging specialization in a social insect (Apis mellifera 102. Feldhaar, H., Foitzik, S. & Heinze, J. Review. Lifelong manipulations to be carried out in the honeybee. L.). Curr. Top. Dev. Biol. 74, 253–286 (2006). commitment to the wrong partner: hybridization in 129. Patalano, S., Hore, T. A., Reik, W. & Sumner, S. 155. Page, R. E. Jr & Amdam, G. V. The making of a social ants. Phil. Trans. R. Soc. B 363, 2891–2899 (2008). Shifting behaviour: epigenetic reprogramming in insect: developmental architectures of social design. 103. Frohschammer, S. & Heinze, J. A heritable component eusocial insects. Curr. Opin. Cell Biol. 24, 367–373 Bioessays 29, 334–343 (2007). in sex ratio and caste determination in a (2012). 156. Libbrecht, R., Oxley, P. R., Kronauer, D. J. & Cardiocondyla ant. Front. Zool. 6, 27 (2009). 130. Rajakumar, R. et al. Ancestral developmental potential Keller, L. Ant genomics sheds light on the molecular 104. Robinson, G. E. & Page, R. E. Genetic determination of facilitates parallel evolution in ants. Science 335, regulation of social organization. Genome Biol. 14, guarding and undertaking in honeybee colonies. 79–82 (2012). 212 (2013). Nature 333, 356–358 (1988). 131. Bonabeau, E., Theraulaz, G. & Deneubourg, J. L. 157. Corona, M. et al. Vitellogenin underwent 105. Stuart, R. J. & Page, R. E. Genetic component to Quantitative study of the fixed threshold model for the subfunctionalization to acquire caste and division-of‑labor among workers of a Leptothoracine regulation of division of labour in insect societies. behavioral specific expression in the harvester ant. Naturwissenschaften 78, 375–377 (1991). Proc. R. Soc. B 263, 1565–1569 (1996). ant Pogonomyrmex barbatus. PLoS Genet. 9, 106. Rheindt, F. E., Strehl, C. P. & Gadau, J. A genetic 132. Lattorff, H. M. & Moritz, R. F. Genetic underpinnings e1003730 (2013). component in the determination of worker of division of labor in the honeybee (Apis mellifera). polymorphism in the Florida harvester ant Trends Genet. 29, 641–648 (2013). Acknowledgements Pogonomyrmex badius. Insect Soc. 52, 163–168 133. Bonasio, R. The role of chromatin and epigenetics in The authors thank the three anonymous reviewers and (2005). the polyphenisms of ant castes. Brief Funct. Genom. R. Graham for their insightful critiques and suggestions in 107. Anderson, K. E., Linksvayer, T. A. & Smith, C. R. 13, 235–245 (2014). improving an earlier version of this manuscript. This work has The causes and consequences of genetic caste 134. Duncan, E. J., Gluckman, P. D. & Dearden, P. K. been supported by a Howard Hughes Medical Institute determination in ants (Hymenoptera: Formicidae). Epigenetics, plasticity, and evolution: how do we link Collaborative Innovation Award (HCIA) #2009005 to D.R., Myrmecol. News 11, 119–132 (2008). epigenetic change to phenotype? J. Exp. Zool. Part B S.L.B. and J.L. 108. Huang, M. H., Wheeler, D. E. & Fjerdingstad, E. J. 322, 208–220 (2014). Mating system evolution and worker caste diversity in 135. Welch, M. & Lister, R. Epigenomics and the control of Competing interests statement Pheidole ants. Mol. Ecol. 22, 1998–2010 (2013). fate, form and function in social insects. Curr. Opin. The authors declare no competing interests. 109. Hamilton, W. D. The genetical evolution of social Insect Sci. 1, 31–38 (2014). behaviour. I. J. Theor. Biol. 7, 1–16 (1964). 136. Sasaki, T., Granovskiy, B., Mann, R. P., Sumpter, D. J. 110. Hamilton, W. D. The genetical evolution of social & Pratt, S. C. Ant colonies outperform individuals FURTHER INFORMATION behaviour. II. J. Theor. Biol. 7, 17–52 (1964). when a sensory discrimination task is difficult but not AntWeb: www.antweb.org 111. Haig, D. The kinship theory of genomic imprinting. when it is easy. Proc. Natl Acad. Sci. USA 110, ALL LINKS ARE ACTIVE IN THE ONLINE PDF Annu. Rev. Ecol. Syst. 31, 9–32 (2000). 13769–13773 (2013).

688 | OCTOBER 2014 | VOLUME 15 www.nature.com/reviews/genetics

© 2014 Macmillan Publishers Limited. All rights reserved