Honey bee sociogenomics: a genome-scale perspective on bee social behavior and health
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Apidologie Review article
* INRA, DIB and Springer-Verlag France, 2013
DOI: 10.1007/s13592-013-0251-4
Honey bee sociogenomics: a genome-scale perspective
on bee social behavior and health
Adam G. DOLEZAL1 , Amy L. TOTH1,2
1
Department of Ecology, Evolution, and Organismal Biology, Iowa State University, Ames, IA, USA
2
Department of Entomology, Iowa State University, Ames, IA, USA
Received 3 July 2013 – Revised 9 October 2013 – Accepted 18 October 2013
Abstract – The biology of honey bees involves a host of developmental, behavioral, and physiological
components that allow thousands of individual bees to form complex social units. Fueled by a wealth of
information from new genomic technologies, a new approach, sociogenomics, uses a focus on the genome to
integrate the molecular underpinnings and ultimate explanations of social life. This approach has resulted in a
massive influx of data from the honey bee genome and transcriptome, a flurry of research activity, and new
insights into honey bee biology. Here, we provide an up-to-date review describing how the honey bee has been
successfully studied using this approach, highlighting how the integration of genomic information into honey
bee research has provided insights into worker division of labor, communication, caste differences and
development, evolution, and honey bee health. We also highlight how genomic studies in other eusocial insect
species have provided insights into social evolution via comparative analyses. These data have led to several
important new insights about how social behavior is organized on a genomic level, including (1) the fact that
gene expression is highly dynamic and responsive to the social environment, (2) that large-scale changes in
gene expression can contribute to caste and behavioral differences, (3) that transcriptional networks regulating
these behaviors can be related to previously established hormonal mechanisms, and (4) that some genes and
pathways retain conserved roles in behavior across contexts and social insect taxa.
genome / division of labor / behavioral maturation / caste / comparative genomics
1. INTRODUCTION approaches to bridge gaps between evolutionary
and mechanistic approaches to studying ani-
The social life of bees has been of intense mal societies began little more than a decade
interest to biologists and apiculturists for centu- ago.
ries. As such, there has been a wealth of studies This new approach, dubbed sociogenomics
on the evolution, behavior, colony organization, (Robinson 1999), proposed that the genome can
and development of honey bees and their form a centerpiece for linking different levels of
societies. These studies have spanned across analysis, allowing researchers to integrate the
levels of analysis, providing insight into both proximate causes of behavior, like gene expres-
the proximate and ultimate causes behind the sion and physiology, with more ultimate analy-
social complexity of honey bee society. ses, like behavioral ecology. By using the
However, specific focus on integrating these genome as a focal point, sociogenomics seeks
to provide a more comprehensive method for
understanding social life, from its evolution to
Corresponding author: A. Dolezal, its genetic regulation—and everywhere in be-
adolezal@gmail.com tween (Robinson et al. 2005). Since studies on
Manuscript editor: Stan Schneider honey bees have historically run the gamutA.G. Dolezal and A.L. Toth
across these levels of analysis, it is unsurprising disease and pathogen responses. Finally, while we
that bees have become an important model for focus specifically on how sociogenomics has
applying sociogenomic approaches. With the improved our understanding of bee biology,
increased scope and availability of genomic throughout the review, we highlight how a
tools (Table I), including the sequencing of the sociogenomic-minded exploration across social
complete honey bee genome (Weinstock et al. insect taxa has fueled comparisons for a better
2006), studies grounded in a genomic perspec- understanding of the evolution of eusociality.
tive have been successful in helping to tease
apart the different components that intersect to 2. WORKER DIVISION OF LABOR
build complex social organisms (Smith et al.
2008). 2.1. Foraging ontogeny
The sociogenomic approach spans across a
large swathe of applications and has been One of the most striking aspects of eusocial
defined broadly (Robinson et al. 2005). Here, insect societies, and honey bees in particular, is
we restrict our definition to research focusing the behavioral plasticity found within the
on large numbers of genes or on single genes worker caste. This flexibility takes the form of
that provide key insights into larger genetic temporal polyethism, in which workers transition
pathways, and exclude a rich and informative across different task repertoires as they age. After
literature on single genes (e.g., Ben-Shahar adult emergence, workers specialize on a variety
2005; Amdam et al. 2010) that are beyond the of in-nest tasks, such as brood care, and then
scope of this review. We focus our review on transition through stages of other tasks, such as
large-scale genomic or transcriptomic analy- nest maintenance and guarding, culminating in
ses, microarrays, or targeted studies that foraging behavior (Winston 1987). While this
explore or clarify genetic pathways consisting sequence of behavioral maturation occurs as a
of multiple genes. Furthermore, while we focus general pattern, workers exhibit a high level of
specifically on how the rise of sociogenomics flexibility in the rate of behavioral development,
has helped us understand the social life of bees, which allows individuals to respond to differing
this approach has been very successful in other colonial demands (Robinson 1992).
eusocial insects (Smith et al. 2008). In fact, one At this point in time, sociogenomics has been
important strength of sociogenomics is the more thoroughly applied to the study of worker
capability to make comparisons: the power to behavioral maturation than any other facet of
search for homologies in genomes across taxa honey bee biology and, therefore, stands as the
helps find clues to understand the evolution of best example of how successful this approach can
bee societies (Fischman et al. 2011). be. This is largely due to the strong background of
Here, we review how the use of sociogenomics literature and expertise spanning behavioral,
has advanced knowledge of many facets of honey genetic, neurobiological, and physiological
bee biology. We begin with worker temporal studies on temporal polyethism, which has allowed
polyethism; the behavioral transition from in-hive researchers to build a more comprehensive
to foraging tasks is arguably the best studied understanding of this system centering on
honey bee behavioral phenomenon using a investigation of the genome (Robinson 2002;
sociogenomic approach, and we use it as a Robinson et al. 2005; Smith et al. 2008) (Figure 1).
benchmark for comparison with other research Before sequencing of the honey bee genome,
foci. Then, we follow with descriptions of the most of the studies described as sociogenomic
progress made in understanding the evolution and were borne from the integration of behavioral,
regulation of caste differences, communication, neuronal, and physiological mechanisms with
and social immunity, and we also review how the genomic information generated from partial
approaches and methods pioneered with genome resources, which provided vastly more
sociogenomics have been applied to honey bee information than previous approaches that focusedTable I. How has the sequencing of the honey bee genome changed sociogenomic applications?
Application Pre-genome Post-genome Gain from genome References
Gene discovery e.g., Degenerate polymerase Computational analysis Whole genome instead Elsik, Mackey et al. (2007)
chain reaction (PCR) and of entire genome sequence of single gene, less
cloning on a gene-by-gene labor-intensive, allows
basis discovery of novel genes
Identification of e.g., Rapid amplification of Computational analysis of Whole genome instead Ament et al. (2012b)
regulatory sequences cDNA ends on a gene-by- entire genome sequence, of single gene, less
gene basis chromatin immunoprecipitation labor-intensive
and sequencing
Epigenetic profiling e.g., Methylation sensitive Computational analysis of Ability to identify actual Kronforst, Gilley et al.
(e.g., identifying amplified fragment length CpG observed/expected content genes and nucleotides (2008); Herb et al. (2012)
methylation) polymorphism of anonymous of genome, whole genome that are methylated,
methylated sites bisulfite sequencing whole genome coverage
instead of a small subset
Gene expression Quantitative PCR for select Whole-genome microarrays, Whole genome instead of Whitfield et al. (2002, 2003);
profiling genes, EST-based microarrays, RNA sequencing compared a fraction of the genome, Chen et al. (2012); Liang
screening ESTs for a small to whole genome RNA-Seq is less labor- et al. (2012)
number of differentially intensive and has a greater
expressed genes dynamic range
Sequence evolution Complex sequence of molecular Bioinformatic comparisons Whole genome instead of Hunt et al. (2010); Johnson
techniques, with the ability to across species that include select genes, enhanced and Tsutsui (2011)
Honey bee sociogenomics
target only a small number numerous genes and entire ability to uncover gene
of genes gene families losses and gains
Many studies on honey bee sociogenomics occurred before the sequencing of the honey bee genome and provided significant insights. However, the existence of an annotated
genome provides researchers with many powerful tools to investigate factors that would be much more difficult, or even impossible, without the ability to look at the whole genome.
This table summarizes how different applications have changed with the advent of the genomeA.G. Dolezal and A.L. Toth
Environment
Gene
Epigenetics
Expression
Phenotype
Regulatory Hereditary
elements Genetics
Figure 1. An integrative, sociogenomic approach as applied to the study of behavioral maturation into a forager
in worker honey bees. A sociogenomic approach to social behaviors has the potential to integrate many
different forms of genomic mechanisms that interact to affect phenotype. The best example of the use of
sociogenomics is the investigation of the underlying regulation of honey bee worker behavioral maturation, or
foraging onset. Integration of behavioral studies with a variety of genomic tools has provided insights into how
the environment (both external and social) and allelic variation in individuals affects gene expression,
regulatory elements, and epigenetics to form a network of effects that result in a specific behavioral phenotype.
Investigation of other phenotypes with a sociogenomic approach promises to reveal similar networks. Image
©Alex Wild, used by permission.
on one or a few genes (Whitfield et al. 2002). colonies were used to decouple the nurse-to-
These included expressed sequence tags forager transition from chronological age, revealing
(ESTs) and microarrays. For the honey bee, the changes in over 2,000 genes or about 40 % of genes
combination of these techniques, which provided assayed. Further, these gene expression profiles
partial sequence information for close to 50 % of can be used to predict the behavior of individual
the genes in the genome, allowed researchers to bees (Whitfield et al. 2003). In fact, many of the
identify thousands of genes of interest and then same molecular processes involved in the nurse-
screen them for expression levels (Whitfield et al. to-forager transition appear to be conserved in the
2002). Microarrays allowed for large-scale brain across species within the genus Apis, though
screenings for gene expression differences others, such as those involved in carbohydrate
between behavioral groups, helping to identify metabolism, circadian rhythm, and colony defense,
how differences in gene expression are related to differ between species (Sen Sarma et al. 2007). In
worker division of labor. Comparisons of the addition, a comparison of thousands of transcripts
brains of young nurses to old foragers (Kucharski from the brain and abdomen, across nine bee
and Maleszka 2002; Whitfield et al. 2003) found species, representing three origins of sociality,
clear differences in brain gene expression. Then, showed that genes involved in carbohydrate
behavioral manipulations using single-cohort metabolism are more rapidly evolving in eusocialHoney bee sociogenomics lineages, supporting arguments for the involve- in worker behavioral maturation (Ament et al. ment of metabolic pathways in social evolution 2008). (Woodard et al. 2011). Studies using the honey bee genome also After initial screenings identified correlations identified several transcription factors that between behavioral state and gene expression differ in expression between worker behavioral differences, more in-depth investigations further groups. Genes such as Creb, involved in neural fleshed out the relationship between worker plasticity in many animals (McClung and behavioral maturation and brain gene expres- Nestler 2008), and dorsal, which is involved sion. By integrating genomic analysis with in insect developmental patterning and immune experimental approaches, it has been possible response (Qiu et al. 1998), were identified as to build a more complete picture of the part of a transcriptional network that could interaction between physiology, environment, predict the expression of many other behavior- and genotype on worker behavioral maturation. ally linked genes (Weinstock et al. 2006; The brain gene expression differences found Chandrasekaran et al. 2011). Further genomic between nurses and foragers appear to be analyses of the brain showed that transcription influenced strongly by queen mandibular factors known to regulate development may be pheromone (QMP; Grozinger et al. 2003), also involved in behavioral maturation in honey which is produced by honey bee queens to bee workers (Sinha et al. 2006). In particular, regulate the behavior of workers (Winston and ultraspiracle (usp), a transcription factor linked Slessor 1998). Juvenile hormone (JH) signal- to JH signaling, appears to interact with other ing, long known to be an important regulator of transcription factors to help orchestrate a net- foraging onset (Robinson 1987; Sullivan et al. work of gene expression that occurs between 2000) had very large effects on gene expression, the brain and peripheral tissues to regulate leading to forager-like brain gene expression worker behavioral changes (Ament et al. even in bees reared in cages with no prior 2012b). DNA methylation, an epigenetic mod- foraging experience (Whitfield et al. 2006). ification to DNA (discussed in more detail in Further investigation showed how nutrition Section 4), also appears to have a role in worker and nutritional signaling pathways, specifically behavioral change; not only do nurse and insulin/insulin-like signaling (IIS), are in- forager bees differ in brain gene methylation volved in behavior and how other factors are patterns, but the methylation patterns are also affected downstream. In insects, IIS acts as a behaviorally reversible (Herb et al. 2012). key regulator of feeding behavior and metabo- Sociogenomic studies on division of labor in lism and also interacts with target of other eusocial species have also provided rapamycin (TOR) (Edgar 2006), another im- information on how common genetic toolkits portant metabolic pathway regulator, and JH could be used to build convergent social (Tu et al. 2005). Based on single-gene experi- behaviors. Microarray screening of the brains mental studies (Ben-Shahar 2005; Nelson et al. of Polistes metricus wasps showed that the gene 2007), a focus on the effects of these pathways expression profiles of foraging P. metricus had showed that changes in IIS and TOR affect significant overlap with the profiles of foraging behavioral maturation, and a reanalysis of honey bees, especially genes related to heat previous microarray data (Grozinger et al. stress, locomotion, and lipid metabolism 2003; Whitfield et al. 2006) showed differences (Toth et al. 2010). Further experiments showed in energy metabolism between nurses and that starved P. metricus workers had reduced lipid foragers (Ament et al. 2008, 2010). In fact, levels and increased foraging activity (Daugherty experimental perturbation of IIS causes changes et al. 2011), similar to the nutritional regulation of in the timing of foraging initiation, further foraging in the honey bee (Toth et al. 2005; Toth showing how IIS and its interaction with and Robinson 2005). These changes were accom- nutritional and metabolic pathways are involved panied by changes in brain gene expression that
A.G. Dolezal and A.L. Toth
significantly overlapped with changes found in selection regime produced divergent phenotypes
honey bees, including genes involved in insulin that were highly amenable to genetic mapping
and JH signaling. These two studies together studies. A series of analyses identified quanti-
provide some support for the idea that a common tative trait loci (QTLs) associated with pollen
genetic toolkit, centering on nutritional responses, hoarding, individual forager preferences (Hunt
could have a key role in the evolution of division et al. 1995), ovary size (Linksvayer et al. 2009;
of labor across eusocial insect lineages (Daugherty Graham et al. 2011), and JH responsiveness
et al. 2011). (Page et al. 2012) in these high-pollen-hoarding
As a whole, studies on behavioral maturation and low-pollen-hoarding strains. Furthermore,
in workers form the most complete picture of localization of these QTLs using the honey bee
how a sociogenomic approach can link many genome indicated that the QTLs contained
different factors to better explain a complex several genes related to IIS signaling, highlight-
biological process. By integrating the wealth of ing the potential importance of IIS signaling on
behavioral, physiological, and genetic informa- these behavioral phenotypes (Hunt et al. 2007).
tion with newly developed genomic tools, it has Microarray screening of the ovaries of high-
been possible to more thoroughly understand pollen-hoarding and low-pollen-hoarding strains
the molecular underpinnings of behavioral of bees further identified transcriptional differ-
phenotypes. Even before sequencing of the ences related to these behavioral phenotypes,
honey bee genome, microarray screens were particularly tyramine receptor (TYR) and a
able to identify how patterns in gene expression putative ecdysteroid hormone receptor (HR46)
differ as bees change behaviors (Whitfield et al. (Wang et al. 2012). Combined with the strong
2003) and in response to pheromonal stimuli body of experimental studies using single-gene
(Grozinger and Robinson 2002). Sequencing of and physiological approaches to bee behavior,
the honey bee genome has enabled more particularly those focusing on the yolk precur-
complete genomic analyses that include nearly sor vitellogenin, these studies have provided
all of the genes in the genome and has given many insights into the physiological regulation
researchers ready access to noncoding regions and possible evolutionary pathways to eusocial
of the honey bee genome. Table I provides a insect behaviors (reviewed by Page et al. 2012;
summary of pre-genome and post-genome ap- for further review, see Rueppell 2013).
proaches to studying the genetic basis of honey
bee biology. For example, full genome infor- 2.3. Guarding, undertaking, and scouting
mation has helped to clarify how transcription
factors and regulatory elements (Ament et al. While the sociogenomics of foraging onset is
2012a, b) and DNA methylation (Herb et al. the most investigated, other worker honey bee
2012) are involved in driving a host of pathway behaviors have also been studied with genomic
changes that influence behavioral maturation approaches. There are some important differ-
(Figure 1). ences with respect to the aforementioned work
on behavioral maturation in that some worker
2.2. Pollen-hoarding syndromes behaviors are more short-term responses to
colony needs and do not appear to involve
Genomic approaches have also been used to extensive shifts in gene expression. For exam-
study colony-level pollen-hoarding phenotypes ple, microarray comparisons of brain tissue
and the individual physiological and behavioral showed no significant differences in brain gene
differences that accompany them. Researchers expression of bees exhibiting guarding and
used selective breeding to produce bee strains undertaking behaviors, which occur for short
exhibiting opposite colony-level traits for pollen periods of time between nursing and foraging
storage, either high or low levels of pollen onset, even though these behaviors are clearly
hoarding (Page and Fondrk 1995). Thus, the discernible, indicating that distinct behavioralHoney bee sociogenomics
changes can occur even in the absence of large- 3. COMMUNICATION
scale transcriptional changes in the brain (Cash
et al. 2005). 3.1. Pheromones
Another important behavior, which only
some workers ever actually perform, is scouting The organization and maintenance of com-
behavior, which can take the form of worker plex social colonies requires intricate systems of
bees scouting for new food sources or new nest communication. In honey bees, the predominant
sites after a swarming event (Seeley 1985). A method of communication is through chemicals,
whole-genome microarray comparison of the mostly in the form of pheromones, which act as
brains of food scouts and non-scouting forager chemical signals by members of the hive to
bees revealed extensive differences in gene prevent intracolonial conflict and regulate be-
expression, most notably in genes involved in havioral plasticity. Primer pheromones affect
neurotransmitter systems known to be involved long-term physiological changes that result in
in novelty seeking behavior in other insect delayed behavioral responses, while releaser
species and humans, like catecholamine, gluta- pheromones act on more short-term processes
mate, and GABA signaling (Liang et al. 2012). to quickly change behavioral performance (Le
Conte and Hefetz 2008). Queens produce QMP,
a primer pheromone that prevents workers from
2.4. Aggression developing active ovaries and foraging onset
(Pankiw et al. 1998). Experimental manipula-
The genetic basis of yet another worker tion, followed by microarray screening of brain
behavior, defensiveness or aggressive behavior, gene expression, also showed that QMP treat-
has been a topic of active investigation. Crosses ment changes brain gene expression in over
of high-defensive-response Africanized honey 2,500 genes of laboratory bees and around 700
bees with low-defensive-response European genes in bees from field colonies, specifically
honey bees revealed several QTLs linked to activating genes associated with nursing behav-
increased defensiveness (sting-1, sting-2, and ior and repressing those associated with forag-
sting-3; Hunt et al. 1998). The use of genomic ing. In addition, transcription factor genes were
sequencing and linkage mapping on these QTLs affected at a higher proportion than other genes,
narrowed down the number of candidate genes suggesting that QMP may act by targeting
associated with defensive responsiveness, iden- transcription factors to initiate downstream
tifying orthologs of genes involved in nervous cascades of expression changes (Grozinger et
system development and activity and sensory al. 2003). Further, individual workers vary in
signaling (Hunt et al. 2007), though another their attraction to QMP, and their brain gene
study showed that potentially novel genes may expression reflects these differences. Analysis
also be involved (Lobo et al. 2003). The showed 960 differentially expressed brain tran-
molecular basis for increased aggressiveness in scripts between high response and low response
different contexts appears to utilize some con- to QMP bees, with particular differences in gene
served mechanisms, whether due to heredity networks related to neural network structure
(i.e., Africanized vs. European strains), age (Kocher et al. 2010a).
(young vs. old bees), or environment (exposure Alarm pheromone is a releaser pheromone
to alarm pheromone). The fact that similar that quickly stimulates an aggressive response
genes expressed in the brain influence aggres- in workers (Winston 1987). Even though re-
sive response due to these different influences sponse to alarm pheromone is very fast, gene
supports the argument that changes in the expression changes still occur in the brain,
regulation of gene expression via cis-regulatory particularly the immediate early gene c-Jun. In
mechanisms are at the heart of some forms of addition to affecting c-Jun, a transcription factor
behavioral diversity (Alaux et al. 2009c). involved in neural circuits, alarm pheromoneA.G. Dolezal and A.L. Toth also affects behavioral responses to subsequent In addition to studies in honey bees, the exposure long after the initial aggressive response power of a genomic approach to understanding has ceased (Alaux and Robinson 2007). Further- the molecular basis of chemical communication more, while alarm pheromone results in increased has been exemplified by findings made possible behavioral activity, it also causes downregulation by the recent sequencing of the Solenopsis of genes involved in brain metabolism, posing invicta fire ant genome. In fire ants, a single interesting questions regarding the relationship Mendelian factor in the form of different alleles between brain metabolic activity and overall at the Gp-9 locus determines if workers accept behavior. Genes modulated by alarm pheromone one or many queens. Gp-9 codes for an odorant- also show overlap with those that are upregulated binding protein, so it has been suggested that its in highly aggressive Africanized honey bees. The effects are due to modulation of pheromone fact that the same genes have effects on aggressive responses (Gotzek and Ross 2007). However, behaviors in different contexts suggests that alarm its effects are much more diverse, influencing a pheromone-regulated genes were likely involved number of different traits, including female size in the evolution of different aggressiveness and fecundity (Keller and Ross 1999; Gotzek phenotypes in honey bees (Alaux et al. 2009c). and Ross 2009). Facilitated by the sequencing Brood pheromone acts as both a primer of the genome (Wurm et al. 2011), a recent pheromone, acting in the long term to delay investigation of the genomic region where Gp-9 foraging in young bees, and as a releaser is located found that the Gp-9 allele is part of a pheromone, stimulating foraging in old bees heteromorphic chromosome, similar to a Y sex (Le Conte et al. 2001). Brain gene expression determination chromosome. Instead of deter- profiling reflects these effects, as brood phero- mining sex, these chromosomal differences help mone causes different effects on bees of maintain different intraspecific social pheno- different ages: in young workers, brood phero- types (Wang et al. 2013). While a similar social mone upregulates genes associated with nursing chromosome has not been identified in other and downregulates genes associated with forag- social insects, similar systems, increased access ing, and does the inverse in old bees, supporting to genomic tools can make novel discoveries the argument that pheromones affect behavior such as this possible. by mediation of gene expression, even in different contexts (Alaux et al. 2009b). 3.2. Dance language Bees perceive pheromonal signals through an incredibly sensitive and well-developed olfactory In addition to a complex system of chemical system. In insects, pheromones and other odorants communication, honey bees are well known for are carried to odorant receptors by odorant- their dance language, in which returning binding proteins or chemosensory proteins (Pelosi foragers use mechanical signals (i.e., sound, et al. 2005). The sequencing of the honey bee vibration, and tactile interaction) to communi- genome afforded an opportunity to explore the cate the location of food in the environment to full complement of ORs and OBPs in honey bees bees within the hive (Dyer 2002). and indicated that there has been an evolutionary Transcriptomic profiling of the nervous systems expansion of the number of olfactory proteins in of dancing foragers identified gene expression honey bees compared to other, nonsocial insects changes linked to dancing, with differences (Foret and Maleszka 2006). Interestingly, even particularly found in the mushroom bodies of though the antennae are the site of odorant the brain. Comparisons with dancing foragers sensation, both odorant-binding proteins (Foret from Apis florea and Apis dorsata identified and Maleszka 2006) and chemosensory proteins species-specific and species-consistent genes (Foret et al. 2007) are commonly expressed in related to dancing behavior. Further analysis other tissue, indicating their possible role in other of between-species differences, like those physiological functions. linked to motor control and metabolism, may
Honey bee sociogenomics
provide further insights into how genetic which differ in many more aspects than bees
differences between these species underlie the differentially performing vibrational signals. Inter-
differences in their dance language phenotypes estingly, some of the genes differentially expressed
(Sen Sarma et al. 2009). Furthermore, dancing in these signaling bees are those associated with
bees that perceive the location of food to be motor activities like locomotion courtship (Alaux
further away have different gene expression et al. 2009a).
profiles than those perceiving food to be
nearby, particularly in the mushroom bodies 4. CASTE POLYPHENISM
and optic lobes, with notable differences in
learning and memory systems (Sen Sarma et 4.1. Queen–worker developmental
al. 2010). In addition, differences in mushroom differentiation
body gene expression arise as bees accrue
foraging experience (Lutz et al. 2012), further In honey bees, the reproductive division of
indicating the importance of genomic changes labor between queens and workers is based on
in the regulation of behavioral plasticity, strict caste polyphenism, with extreme differences
especially in the brain. in physiology, morphology, and behavior between
Communication systems have also been impli- reproductive queens and functionally sterile
cated in the evolution and diversification of workers. The differences between the castes are
sociality in bees. By using next-generation se- determined due to differential feeding at critical
quencing for the rapid generation of transcriptomes stages during larval development; workers are fed
of nine different bee species, spanning three a restricted diet, while queens receive a diet richer
independent eusocial origins, researchers conduct- in royal jelly (Winston 1987). Changes in diet
ed genome-scale comparative analyses to deter- cause a cascade of changes in gene expression and
mine which genes show evidence of more rapid hormone signaling that result in the production of
rates of protein evolution and how these relate to different caste phenotypes. Larval consumption of
different levels of sociality. The results indicated a diet rich in royal jelly, and specifically the
that gland development genes were rapidly evolv- protein royalactin (Kamakura 2011), results in
ing in eusocial lineages including honey bees. This increased JH levels (Rembold 1987; Rachinsky
suggests that glandular structures and their chem- and Hartfelder 1990) which are involved in
ical products, likely used for increased social triggering queen development (Rembold et al.
communication, were targets of selection during 1974). Screening of whole-body gene expression
eusocial evolution (Fischman et al. 2011; Woodard showed that many of the genes overexpressed in
et al. 2011). In addition to the work on pheromones queen-destined larvae were linked to metabolism
and dance, there has been some inquiry into the and hormone responsiveness (Evans and Wheeler
sociogenomics of honey bee vibrational signals. 2001; Cristino et al. 2006; Barchuk et al. 2007). In
These signals are produced when some bees grasp particular, insulin receptor and insulin receptor
a nestmate and rapidly vibrate, resulting in the substrate, components of the IIS pathway, were
recipient bee changing its behavior in a context- overexpressed in queen-destined larvae (Wheeler
dependent manner (Schneider and Lewis 2004). et al. 2006) during times where JH content also
Using a microarray, researchers showed that brain rises (Rembold 1987). The identification of
gene expression differs in over 900 genes, with metabolic genes as possible modulators of JH
around half upregulated and half downregulated, in signaling and queen development led to further
bees that send these signals vs. those that do not. investigations, ultimately showing the importance
This is particularly notable because the number of the epidermal growth factor receptor pathway as a
differentially expressed genes linked to this vibra- modulator of queen–worker differentiation, trig-
tional signal is surprisingly substantial. For com- gered by the ingestion of royalactin (Kamakura
parison, around 1,300 genes are differentially 2011). In addition to the differences found during
expressed between young nurses and old foragers, development, genomic analyses have shownA.G. Dolezal and A.L. Toth
significant differences in gene expression between even though brain methylation does not differ in
adult queens and workers. Microarray screening newly emerged queens and workers (Herb et al.
of adult brains showed that approximately 2,000 2012). How does methylation affect expression?
genes are differentially expressed in the brains of Methylation does not appear to be closely tied to
queens and workers and over 200 of these are differential upregulation or downregulation of
expressed in a more queen-like manner in genes. Rather, methylation is often clustered in
reproductive workers, identifying a set of genes areas of genes where splicing occurs, suggesting
likely involved in reproductive activity, regardless that methylation may be involved in the regulation
of caste (Grozinger and Robinson 2007). Further of alternative splice variants (Flores et al. 2012;
investigation into the transcriptomic differences Foret et al. 2012). Also, methylation differences
between queen-destined and worker-destined lar- occurred on genes coding for some histones,
vae used next-generation RNA-Seq technology to proteins that are also epigenetically regulated
provide a more complete catalog of transcriptional and can affect chromatin structure and gene
differences than previously possible using ESTs expression (Lyko et al. 2010) and which may
or microarrays. These comparisons identified over have an important role in the regulation of bee
4,000 differentially expressed genes and clarified development (Dickman et al. 2013). Methylation
the dynamics of TOR expression, showing that differences are also associated with caste differ-
differences are greatest between queens and ences of other eusocial insect species, including
workers during the fourth (of five) larval instar several ant species (Bonasio et al. 2012; Smith et
(Chen et al. 2012). al. 2012) and Polistes wasps (Weiner et al. 2013),
Another aspect of honey bee development that indicating the possible importance of epigenetic
has been identified due to the expansion of modifications in the convergent evolution of
genomic tools is the importance of epigenetic eusocial societies.
effects caste determination. Epigenetic modifica-
tions occur when chemical modifications to DNA 4.2. Reproductive behavior
take place in response to an environmental
stimulus. Such modifications do not affect the Genomic tools have also been used to
DNA sequence, but cause structural changes in investigate the molecular underpinnings of
chromatin that can result in alterations in gene reproductive activation in both queens and
expression that may last across an individual’s workers. Since queen bees express extreme
lifetime or even across generations. One form of differences in behavior and physiology before
epigenetic modification is DNA methylation, in and after mating, they are an excellent model in
which methyl groups are attached to nucleotides, which to investigate the changes that occur with
usually CpG dinucleotides, and have the potential mating. Even though their mating biology is
to affect the expression of methylated sequences very different, changes in brain and ovary
(Bird 2007). In honey bees, dynamic “de novo” transcriptional profiles in honey bee queens
methylation is driven by DNA methyltransferase- overlapped with those observed in Drosophila
3 (DNMT3). Experimental silencing of DNMT3 melanogaster females, indicating that the regu-
in developing larvae prevents the attenuation of lation of post-mating behavior may be strongly
gene expression and mimics the response to a diet conserved across insect taxa (Kocher et al.
rich in royal jelly (Kucharski et al. 2008). 2008, 2010b). Though both involve individual
Furthermore, over 2,000 genes are differentially behavioral changes, there was no clear relation-
methylated in worker-destined larvae compared to ship between genes associated with queen
queen-destined larvae, with the majority being up- mating behavior and worker behavioral matura-
methylated in worker-destined larvae (Foret et al. tion (Kocher et al. 2008).
2012). While not as drastic, there are also “Anarchistic” bees are an unusual strain of
methylation differences in 550 genes in the brains honey bees where workers develop ovaries and
of adult queens and workers (Lyko et al. 2010), lay viable eggs, even in the presence of a layingHoney bee sociogenomics queen. This cheating behavior is partially nisms are involved in reproductive division of explained by four QTLs found in anarchistic labor in these species, while worker behavioral workers (Oxley et al. 2008). Screening of the regulation shows more conservation (Toth et al. heads and abdomens showed that wild-type 2010). Similarly, transcriptomic profiling of P. workers have more genes upregulated than canadensis queen and worker brains showed anarchistic workers, and it has thereby been little overlap with honey bee caste-specific hypothesized that egg laying may be the default genes (Ferreira et al. 2013). A comparison of and that normal workers upregulate genes that gene expression profiles between two species of “switch off” ovarian activation (Thompson et al. adult and pupal fire ant that analyzed the whole 2006). Anarchistic bees do show significant bodies of queens, workers, and males showed upregulation of some genes in the head and that, while gene expression differences occurred abdomen, particularly in vitellogenin, involved between queens and workers, the greatest in ovarian activation, and AdoHycase, which is interspecific gene expression differences were possibly involved in the regulation of DNA found between adult workers (Ometto et al. methylation (Thompson et al. 2008). Similarly, 2011). P. canadensis RNA-Seq data suggest that a genome-wide comparison of gene expression genes that are worker-biased in their expression in the whole bodies of workers showed that are more likely to be “novel,” with no homol- over 1,200 genes are differentially expressed in ogy to known sequences, further suggesting that normal workers vs. workers that became repro- molecular evolution occurs more rapidly in ductive due to queenlessness. Reproductive genes of importance to the worker caste workers overexpressed genes involved in repro- (Ferreira et al. 2013). ductive activation, compared to nonreproductive The use of comparative bioinformatic anal- workers, which exhibited increased expression ysis on existing datasets has become a useful of genes involved in flight metabolism and tool as genomic technologies have advanced foraging. Therefore, gene expression compari- and the amount of sequence data for honey bees sons indicated differences in reproductive acti- and other insects has drastically increased. This vation, as would be expected, but also approach can be exemplified by studies seeking differences in overall activity levels and behav- to better understand rates of gene evolution in ioral performance between these different phe- queens and workers. By honing in on previous- notypes (Cardoen et al. 2011). ly identified genes with worker-biased or queen- Comparisons of the genomics of caste and biased expression, it was possible to compare reproduction between honey bees and other rates of amino acid substitution of these genes eusocial insect species has also helped elucidate across honey bees and various nonsocial insects how different genomic components could be for which genomic sequence data were avail- involved in eusocial evolution. Identification of able. A comparison of queen-biased genes with gene expression profiles of whole bodies for worker-biased or non-biased genes showed that adult queens and workers of the paper wasp proteins associated with the queen caste had Polistes canadensis allowed interspecies com- evolved more rapidly than other proteins, parisons, identifying nine genes with conserved suggesting that selective pressure acted strongly caste function across species from four different on queen caste genes (Hunt et al. 2010). origins of eusociality, including bees, ants, and Another study, however, predicted that novel wasps (Sumner et al. 2006). A microarray study genes would be necessary for the evolution of of P. metricus brains further showed that, while complex social behaviors and, given that the wasps had different gene expression profiles majority of these behaviors occur in workers, based on their reproductive status, there was no worker-biased genes should be more likely to be significant overlap between wasp reproductive novel. Their analysis showed that, indeed, the genes and genes involved in honey bee caste worker caste expresses more genes specific to differences. This suggests that different mecha- social insect taxa than the queen caste. Howev-
A.G. Dolezal and A.L. Toth
er, while novel worker behaviors may have different genotypes. When high-hygiene and
arisen from novel genes, it is also possible that low-hygiene worker genotypes are mixed within
rapid evolutionary change still occurred in the the same colony, indirect social effects cause
queen caste, albeit acting upon ancestral genes behavioral changes. Specifically, low-hygiene
(Johnson and Tsutsui 2011). bees increase their hygienic behavior and
exhibit changes in brain gene expression when
5. HONEY BEE HEALTH mixed with high-hygiene nestmates (Gempe et
al. 2012). Therefore, while genotypic effects are
While the primary goal of sociogenomic clearly important, changes in the social envi-
studies is to provide a fuller understanding of ronment have strong potential to increase bee
the evolution and maintenance of sociality, the hygiene.
tools and approaches that were spawned from With sequencing of the honey bee genome,
sociogenomics have been applied to other broad-scale investigation of the genetic path-
questions in honey bee biology. In particular, ways involved in honey bee immune response
increased applications of genome-level investi- became more tractable. Comparisons of the
gation has helped provide a better understand- honey bee genome with that of Drosophila flies
ing of how a variety of factors influence honey and Anopheles mosquitoes first indicated that
bee health. Therefore, while the original intent honey bees, despite the higher pathogen risks
of sociogenomics was to investigate basic associated with colonial living, actually possess
questions in evolution, behavior, and physiolo- substantially fewer immunity-associated genes.
gy, this work has quickly provided key infor- This suggests that, among other possibilities,
mation for addressing applied questions. With selective pressure on disease prevention has
increasing worldwide concerns regarding polli- acted predominantly on social behavioral re-
nator health (Gallai et al. 2009), a better sponses to disease (such as hygienic behavior)
understanding of how honey bees respond to and not individual innate immune responses
health stresses on a genomic scale is of great (Evans et al. 2006). Despite this, the immune
utility. Stresses implicated in honey bee declines genes in honey bees show higher rates of
include pesticides (Mullin et al. 2010), nutrition evolution than those of Drosophila or nonim-
(Naug 2009), and disease. Honey bees suffer mune honey bee genes (Viljakainen et al. 2009).
from a number of diseases caused by bacteria, Further investigation showed that bacterial
viruses, and fungi (Evans and Schwarz 2011) immunostimulation of honey bees results in
and are affected by a number of pests, most changes in expression of hundreds of genes,
notably the Varroa destructor mite (Rosenkranz many of which are not normally associated with
et al. 2010). immune response. Changes in some of these,
Given the highly social nature of honey bee particularly those related to chemical signaling,
colonies, response to disease stress occurs at suggest that changes in expression of nonim-
both the individual and group levels. Individual mune response genes help to orchestrate behav-
bees respond to immune stress via cellular and ioral changes, such as increased grooming
humoral mechanisms, similar to other insects, (Wilson-Rich et al. 2009), that mitigate patho-
but colonies also exhibit social mechanisms gen risks (Richard et al. 2012).
(Wilson-Rich et al. 2009), such as high levels of This hypothesis was further supported
hygienic behavior (Rothenbuhler 1964), to through investigation of the effects of Varroa
prevent the spread of disease. Hygienic behav- infestation on gene expression. When Varroa
ior, characterized by uncapping of pupal cells infestation occurs, many gene expression
and the removal of diseased pupae, has been changes occur, but bees with naturally higher
linked to several QTLs (Oxley et al. 2010). tolerance to Varroa more highly express genes
Interestingly, hygienic behavior is also affected associated with olfaction and stimulus sensitiv-
by interactions between individual workers of ity, not immunity (Navajas et al. 2008). AHoney bee sociogenomics similar study comparing typical colonies with parasite interactions (Cornman et al. 2010), and those from another lineage of Varroa-resistant RNA deep sequencing has helped to identify bees, the Varroa-sensitive hygiene (VSH) strain, novel strains of bee viruses (Cornman et al. 2012, which have been selected for resistance to mite 2013). Metagenomic analyses, where researchers infestation via increased hygienic behavior, also screen diverse genetic material directly from the did not implicate immune genes in the resis- environment, have also been useful for under- tance to mite infestation. Furthermore, compar- standing bee health. A metagenomic survey of ison of VSH bees with the Varroa-resistant bees honey bees from colonies suffering from CCD from Navajas et al. (2008) showed little overlap helped to identify pathogens, specifically Israeli in gene expression between the two sources of acute paralysis virus, that were associated with Varroa-resistant bees. Instead, VSH bees had a that form of colony loss (Cox-Foster et al. 2007). similar profile to bees stimulated with brood Further, metagenomic screening of the microbiota pheromone, indicating a possible connection to of the honey bee gut suggests that a suite of brood care phenotypes, and with Africanized different bacteria in healthy bee guts have a role in honey bees, which are also very hygienic pathogen defense and nutrient utilization (Engel et towards Varroa (Le Conte et al. 2011). Another al. 2012). investigation compared Varroa effects on gene expression in Apis mellifera and the mite- 6. CONCLUSIONS resistant congener Apis cerana, finding signifi- cant differences in genes associated with me- As genomic tools have become available to tabolism and nerve signaling (Zhang et al. honey bee researchers, more facets of honey bee 2010). These studies indicate that there may be biology have been investigated using a several distinct genomic routes to behavioral sociogenomic approach. Worker behavioral mite resistance. maturation, the transition from nurse to forager, Further experimentation showed how genes has received the most attention. Integration of that are upregulated by pollen consumption, like genomic studies with the strong background those involved in protein metabolism, are knowledge of this system from behavioral, downregulated due to Varroa infestation (and ecological, physiological, and genetic studies the accompanying viruses that mites vector). has provided the most comprehensive charac- This provides insight into how mites may stress terization of any component of honey bee bees nutritionally and thus, supplies clues that biology (Figure 1). These investigations stand may be helpful in preventing some pest or as the best examples of how fruitful a pathogen effects (Alaux et al. 2011). Gut sociogenomic approach can be. In particular, microarray analysis of bees suffering from the work exploring how transcription factors are colony collapse disorder (CCD) identified a list involved in regulating large-scale gene expres- of 65 transcripts that may be markers for CCD, sion changes has helped to focus in on key as well as the increased presence of ribosomal transcriptional networks; this information can RNA fragments in CCD colonies, possibly due easily be buried in the data deluge from large- to increased viral infections (Johnson et al. scale transcriptomic analyses. By understanding 2009). how some transcription factors could act as Genomic methods have also been applied to genomic “hubs” to interact and control net- the pests, pathogens, and beneficial microorgan- works of gene expression, this approach has isms that affect honey bees. While not using the helped bring new understanding to how com- honey bee genome itself, these studies still show plex transcriptional networks regulate pheno- how the expansion of genomic approaches are types and how transcription factors could be at helping build a better understanding of honey bee the heart of a genetic toolkits that have been biology. Genomic tools developed for Varroa used by natural selection to build diverse provide useful methods for understanding host– phenotypes (Sinha et al. 2006; Chandrasekaran
A.G. Dolezal and A.L. Toth et al. 2011; Ament et al. 2012a, b). It is worth shared and novel genes and pathways that are noting, however, that most of these studies involved in the convergent evolution of similar provide predominantly correlative data (i.e., social traits and analyses can help identify genes between transcriptome and phenotype), and as that have evolved rapidly to build social such, functional analyses on these genes and phenotypes from solitary traits. pathways are still mostly lacking. Future work focusing on filling in these gaps will be vital for building a fuller understanding of the genomic Glossary of terms: Words rendered in bold aspects of honey bee biology. font in the body of the text are defined here. Further genomic investigations will also likely identify other players in honey bee social organi- Bioinformatics: The use of computational tech- zation. For example, the detection of microRNAs niques to manage and analyze large quantities in the honey bee genome has only recently begun of information from biological systems, pre- to reveal the importance of these small, noncoding dominantly genomic and transcriptomic data regions of RNA that regulate gene expression. (Hogeweg 2011). After computational identification (Weaver et al. 2007), experimental studies have begun to exam- Caste: Term used to describe a group of ine how microRNAs may have a role in behav- individuals in social insect colonies that spe- ioral maturation (Behura and Whitfield 2010; cializes, to some extent, in specific occupations Greenberg et al. 2012; Liu et al. 2012), showing as a result of division of labor. Social insect their potential importance in division of labor. castes can be associated with differences in age, Future studies on microRNAs will likely help to anatomy, and morphology. clarify their placement in the network of factors that affect honey bee behavior. cis-regulatory elements: A sequence of DNA As genomic tools become more advanced which, via the binding of transcription factors or and accessible, the study of more honey bee other proteins, regulates the expression of a gene phenotypes at the genomic level become more or genes on the same chromosome (Wray 2007). tractable, providing a clearer vision of honey bee behavior, communication, development, Division of labor: A social system in which and health. Future work will also likely provide individuals specialize in specific occupations. In a more complete picture of the chain from insect societies, queens mostly reproduce, where- genome to phenotype. Though the honey bee as workers engage in all tasks related to colony genome provided new insights into bee proteo- growth and development. Young workers tend to mics (Wolschin and Amdam 2007), most work in the nest, whereas older individuals forage studies assume that changes in mRNA expres- outside the nest. sion reliably represent changes in protein expression, which lies closer to the actual DNA methylation: A form of epigenetic modifi- phenotype. As proteomic techniques, like the cation in which methyl groups are attached to ability to perform large-scale proteomic analy- nucleotides, usually CpG dinucleotides, have the ses (Hernandez et al. 2012), continue to im- potential to affect the expression of methylated prove, the link between genes, proteins, and sequences (Bird 2007). organismal phenotype should become clearer. In addition, genomic studies are becoming Epigenetics: Environmental mediation of an increasingly possible in solitary Hymenoptera individual’s genome and/or its descendants, with- and other social insects, like wasps, ants, and out changes in DNA sequence, via mechanisms bees. Comparisons with these other species like DNA methylation and histone modification allow for insights for identification of both (Crews 2008).
Honey bee sociogenomics
Eusocial: Traditionally defined as social species Quantitative trait loci (QTL): Sections of
that show three features: extreme asymmetries DNA sequence (loci) that contain or are
in reproduction, with some individuals repro- linked to quantitative trait. QTLs can also
ducing a great deal and others little or not at be mapped to whole or partial genomes to
all; overlapping generations of adults in the further identify genes associated with the trait
nest; and cooperative care of offspring (Wilson of interest (Erickson et al. 2004).
1971).
Queen mandibular pheromone (QMP): Phero-
Expressed sequence tags (EST): ESTs are pro- mone produced by honey bee queens to regulate
duced by sequencing many clones from cDNA the behavior and reproductive physiology of
libraries; since the sequences from these cDNA workers (Winston and Slessor 1998).
libraries are originally derived from mRNA from
the organism of interest, ESTs provide important RNA-Seq: A form of transcriptomic profiling
information regarding what genes are being where high-throughput sequencing of all the cDNA
expressed (Gerhold and Caskey 1996). contained in a sample provides precise measure-
ments of gene expression (Wang et al. 2009).
Genome: The complete genetic code for an
organism. Single-cohort colonies: Behavioral manipulation
in which hives are created solely from young
Genetic toolkit: The concept that conserved genes workers. This modification of normal age demog-
and pathways have similar roles across a variety raphy results in newly formed colonies that lack
of taxa, helping to “build” different phenotypes foragers, and young workers subsequently transi-
from the same “tools,” resulting in diverse tion to foraging behaviors earlier than normal,
phenotypes regulated by similar factors (Toth allowing researchers to compare same-aged indi-
and Robinson 2007). viduals that perform different tasks (Nelson 1927;
Robinson et al. 1989).
Insulin/insulin-like signaling (IIS): Metabolic
pathway that acts as a key regulator of growth, Target of rapamycin (TOR): An important meta-
feeding behavior, and metabolism; in insects, it bolic regulator that interacts with the IIS pathway
also interacts with target of TOR and JH (Edgar (Tu et al. 2005).
2006, Tu et al. 2005).
Transcription factor: A protein that binds to a
Juvenile hormone (JH): Insect hormone involved regulatory DNA segment, regulating the transcrip-
in many behavioral and developmental processes, tion of specific target genes into mRNA.
including onset of foraging behavior in honey bees
(Hartfelder 2000).
Microarray: Technology that allow for the quan- Sociogénomique de l’abeille: une perspective à
tification of gene expression via the hybridization l’échelle génomique sur le comportement social et la
santé de l’abeille
of cDNA to complementary sequences on a chip
(Schena et al. 1995), used in conjunction with Génome / division du travail / maturation
ESTs to quantify known genes. comportementale / caste / génomique comparative
Honigbienen-Soziogenomik: Eine genomweite Sicht
MicroRNA: A small section of noncoding auf das Sozialverhalten und die Gesundheit von
Honigbienen
RNA that has transcriptional and posttransla-
tional effects on gene expression (Chen and Genom / Arbeitsteilung / altersbedingte
Rajewsky 2007). Verhaltensreifung / Kaste / vergleichende GenomikYou can also read
