Ageing in a eusocial insect: molecular and physiological characteristics of life span plasticity in the honey bee


  • D. Münch,

    Corresponding author
    1. Department of Chemistry, Biotechnology and Food Science, Norwegian University of Life Sciences, P.O. Box 5003, N-1432 Aas, Norway; and
    Search for more papers by this author
  • G. V. Amdam,

    1. Department of Chemistry, Biotechnology and Food Science, Norwegian University of Life Sciences, P.O. Box 5003, N-1432 Aas, Norway; and
    2. School of Life Sciences, Arizona State University, P.O. Box 874501, Tempe, AZ 85287, USA
    Search for more papers by this author
  • F. Wolschin

    Corresponding author
    1. School of Life Sciences, Arizona State University, P.O. Box 874501, Tempe, AZ 85287, USA
    Search for more papers by this author

*Correspondence authors. E-mails:;


  • 1Commonly held views assume that ageing, or senescence, represents an inevitable, passive, and random decline in function that is strongly linked to chronological age. In recent years, genetic intervention of life span regulating pathways, for example, in Drosophila as well as case studies in non-classical animal models, have provided compelling evidence to challenge these views.
  • 2Rather than comprehensively revisiting studies on the established genetic model systems of ageing, we here focus on an alternative model organism with a wild type (unselected genotype) characterized by a unique diversity in longevity – the honey bee.
  • 3Honey bee (Apis mellifera) life span varies from a few weeks to more than 2 years. This plasticity is largely controlled by environmental factors. Thereby, although individuals are closely related genetically, distinct life histories can emerge as a function of social environmental change.
  • 4Another remarkable feature of the honey bee is the occurrence of reverted behavioural ontogeny in the worker (female helper) caste. This behavioural peculiarity is associated with alterations in somatic maintenance functions that are indicative of reverted senescence. Thus, although intraspecific variation in organismal life span is not uncommon, the honey bee holds great promise for gaining insights into regulatory pathways that can shape the time-course of ageing by delaying, halting or even reversing processes of senescence. These aspects provide the setting of our review.
  • 5We will highlight comparative findings from Drosophila melanogaster and Caenorhabditis elegans in particular, and focus on knowledge spanning from molecular- to behavioural-senescence to elucidate how the honey bee can contribute to novel insights into regulatory mechanisms that underlie plasticity and robustness or irreversibility in ageing.


When portrayed as an inevitable, passive and random decline of function, ageing is barely of interest to experimentalists. The insight that life span in flies, nematodes and mice can be predictably extended in a laboratory context, however, contributed greatly to reinvigorate ageing research (Shmookler Reis & Ebert 1996). The genetic analysis of long lived mutant strains (Lin, Seroude & Benzer 1998), molecular intervention in stress pathways (Mahler 2001) and dietary restriction (Piper, Skorupa & Partridge 2005) are now widely used tools in experimental gerontology and it is commonly accepted that ageing is a plastic or flexible process.

However, ageing is a fundamental process with great natural diversity, and by focusing only on a few animal models, candidate genes and experimental paradigms, one might risk oversimplifying the complexity of a phenomenon that is characterized by a plethora of dysfunctions. Most notably, an examination of life histories across the animal kingdom reveals large differences in life span between closely related species. For example, extreme cases are reported for nematode taxa, wherein life span can vary by a factor of 50 and more (Gems 2000, see also Ricklefs’ publication in the current issue). Intriguingly, even genetically closely related individuals of the same species might take very different life span trajectories. Thus, studying life span determinants in siblings with highly variable life span offers the unique opportunity to investigate how ageing and its molecular underpinnings are differentially regulated by non-genetic factors, such as social environmental changes.

Social insects – foremost honey bees and ants – are emerging as particularly rewarding systems in this regard. Their tremendously different ageing phenotypes, that occur naturally among individuals, outrivals all experimentally achieved live extensions by far (Keller & Jemielity 2006). Of the social insects, the honey bee (Apis mellifera) has the longest tradition in research (Winston 1987; Gould & Gould 1995). The honey bee has two alternative female castes (queens and workers) and a single male phenotype (the drone). Queen and worker fates are not genetically determined, and full sisters can belong to both castes. Yet, the highly reproductive queens are long-lived and can survive 2 years (Seeley 1978) or, in extreme cases up to 4 years (Botzina 1961; Page & Peng 2001). In contrast, the essentially sterile workers vary in life span from a few weeks to about 1 year, whereas the drone life span is only 4–5 weeks (Winston 1987). Thus, although queens and workers can be closely related genetically, distinct longevity phenotypes emerge between them. Within the worker caste, furthermore, variation in life span develops as a function of social environmental alterations. Throughout ontogeny, worker bees change between tasks in an orderly and usually age-dependent manner. Young workers typically perform within-nest activities like nursing larvae and after 2–3 weeks they make a transition to foraging duties collecting pollen (a protein source) and nectar (a carbohydrate source) for the colony (Winston 1987).

During favourable conditions with brood rearing, life spans of workers vary between 2 and 8 weeks. When workers switch from nest to foraging tasks, the behavioural transition is accompanied by a demographic shift due to a rapid increase in mortality as well as by manifold physiological changes (Robinson et al. 1992). Since the age at foraging onset is usually far more variable than the duration of the forager phase, the timing of this behavioural switch is the major determinant of a bee's overall life span. Moreover, during unfavourable periods when brood rearing and foraging ceases, a third worker sub-caste develops (diutinus or ‘winter’ bees). This sub-caste is characterized by an extreme life span potential of up to 1 year (Maurizio 1950). Thereby, sister worker bees can show a vast but naturally occurring variation in longevity, albeit sharing a highly similar genetic background. Worker bees, furthermore, facilitate studies that contrast the robustness of ageing processes to their plasticity, because they display a unique feature: behavioural reversion (Robinson et al. 1992). Remarkably, this naturally occurring ontogenetic feature is also inducible by experimental manipulations that change the large-scale demographic structure of a colony. In brief, a great number of generally older bees that were previously engaged in foraging tasks will return to nest duties (nursing) when all nurse bees are experimentally removed from the colony. This redirection of behavioural ontogeny affects several physiological markers of senescence (Amdam et al. 2005). The roles of chronological age and social task performance in ageing can therefore be largely decoupled in the honey bee, and thus they can be studied as independent factors.

Ageing and senescence are often used interchangeably and definitions thereof depend on the focus of the researcher and the object of study (Rose 1991). We use the term ageing, or senescence, to describe a progressive time-dependent decline in organismal functions on the molecular and behavioural level.

Ageing acts on all multicellular life. In depth genomic information (The Honey Bee Genome Consortium 2006) spurring a burst in the development of sophisticated molecular and biochemical tools, makes A. mellifera particularly suitable for research on the causal underpinnings of ageing in social insects. To highlight this potential for a broad readership, we here summarize and discuss recent studies on honey bees that examined irreversible but also plastic aspects of ageing at different systemic levels, ranging from molecular hallmarks of senescence to functional impairment.

Endocrine tissues and life-history integration

Our understanding of ageing has gained immensely from anatomical and biochemical work describing the phenomenology of senescence. However, another stimulating approach asks whether there are switches, be it molecular or environmental, that, at least in part, can trigger accelerated ageing, or, even more excitingly, that can postpone ageing. The bee's transition from nest to forager tasks is just such a major determinant of life span. Consequently, recent work in honey bees centres on the action of molecular pathways that accompany this behavioural change. The objective of these studies is to uncover key signals that pace the onset of foraging behaviour and, thereby, influence worker longevity.

In view of the general impact of hormones on insect growth, development and behaviour, it is not surprising that studies of endocrine gland function, hormonal feedback regulation and target tissue sensitivity for hormones are vibrant areas of gerontological research (Tatar, Bartke & Antebi 2003; Flatt, Tu & Tatar 2005; Kenyon 2005). Endocrine integration of longevity, life-history progression, and fertility has been studied in great detail in Drosophila (Tu, Flatt & Tatar 2006). Here, insulin-like peptides are released by the brain and regulate ageing, possibly by modulating downstream signals such as juvenile hormone (JH) and 20-hydroxyecdysone (20E) (Tatar et al. 2001; Tu & Tatar 2003). In support of this explanation, mutants of the ecdysone receptor (EcR) are long-lived, and decreased JH signalling extends life span in Drosophila (Tatar et al. 2001).

Insect JH is produced by the corpora allata, a set of paired endocrine glands behind the brain. Ecdysteroids are synthesized by the prothoracic gland in larvae, and by the gonads in adults. Functional roles of these endocrine tissues and their secretions historically have been mapped out in model systems such as the holometabolous moth (Manduca sexta) and the hemimetabolous locusts and cockroaches (e.g. Raikhel, Brown & Bellés 2005; Dubrovsky 2005). In contrast to Drosophila, these species have the advantages of large organismal size, though the lack of genomic information has been an obstacle to addressing composite questions that require a combination of genetic manipulation and an understanding of hormone titre dynamics at the individual level. The honey bee combines both, functional genomic potential and size.

As in Drosophila the bee's corpora allata complex and the ovary appear to play important roles in life-history regulation (Hartfelder et al. 2002; Page & Amdam 2007). In the case of the corpora allata, the gland's activity and resulting hormone titres were monitored throughout the behavioural transitions of workers. It was documented that, as a worker bee changes from nursing to foraging tasks, the activity of the corpora allata increases (Hartfelder 2000) and this elevated gland activity is linked to an increase in the circulating titre of JH. Conversely, behavioural reversal from foraging in the field to nursing in the nest, reduces gland activity and lowers the circulating hormone titre (Amdam et al. 2005).

Ovarian signalling in worker honey bees, on the other hand, is supported by several recent studies that tie social behaviour, endocrine feedback sensitivity, and longevity to variation in ovary size (Amdam et al. 2006, 2007; Page & Amdam 2007). Larger ovaries are linked to an earlier onset of foraging behaviour and thus a shorter life span. In contrast, smaller ovaries are associated with later foraging onset, and a longer life (Page & Amdam 2007). Interestingly, larger ovaries and thereby shorter life spans are also linked to a preference for pollen hoarding, which despite worker sterility represents an ancestral hallmark of reproductive behaviour (Amdam et al. 2006 and Y. Wang, G.V. Amdam, R.E. Page, unpublished data). Thus, the associations between ovary size, foraging onset and foraging preference fit well to data from Drosophila and Caenorhabditis elegans, in that they support an evolutionary conserved link between fertility and life span (Leroi 2001; Hsu, Murphy & Kenyon 2003; Partridge, Gems & Withers 2005; Flatt & Kawecki 2007 see also Bonduriansky et al. 2008).

Recently, ovarian factors were confirmed to affect corpora allata activity (Nilsen, Stay, Amdam, unpublished data), and the sensitivity of the regulatory feedback loop that determines the circulating JH titres of workers is conditional on ovary size, ovarian activity and yolk protein levels (Amdam et al. 2007). Thus, the emerging picture is that endocrine integration of worker life-history progression is governed by the corpora allata – ovarian axis (Nilsen et al. 2007). Yet, much work is needed to gain an in depth understanding of the causal connections between reproductive signalling systems and ageing in insects (Flatt et al. 2005, 2008). In line with this note of caution, the explanatory framework was recently expanded to include possible important roles of the fat body (analogous to vertebrate adipose and liver tissues). In Drosophila, expression levels of the transcription factor dFOXO in the fat body has major effects on longevity in adult female files (Hwangbo et al. 2004; Giannakou et al. 2007). In honey bee workers, the fat body produces vitellogenin (see below), which acts as an important endocrine factor (Nelson et al. 2007), and it transcribes mRNA for insulin-like peptides at rates that correlate with levels of vitellogenin and JH and, thus, with behavioural tasks (Nilsen et al. 2007).

In sum, just as the characterization of endocrinological pathways in Drosophila stimulates the understanding of their role in life span regulation, the recent demonstration of interlinkage between social traits, reproductive traits (ovary size, vitellogenin) and endocrinological signalling in honey bees suggests pleiotropic functions for hormones in the regulation of longevity. Further efforts should be directed towards understanding how the social context affects hormonal regulation, and most rewarding, towards finding the causal route of social environmental signals to the control of life span variability. Honey bee vitellogenin, a molecule which is not a classical hormone appears to be a major player in such cascades (see the next section).

Molecular regulation of honey bee ageing

Current knowledge of honey bee regulatory anatomy poses that JH interacts with vitellogenin. Vitellogenin is a nutritive yolk protein-produced and secreted by the fat body- and circulates in the haemolymph (Engels 1974; Engels & Fahrenhorst 1974; Pinto, Bitondi & Simões 2000). In recent years, accumulating evidence suggests that honey bee vitellogenin can influence biochemical processes, acts beyond its role in embryonic and adult nutrition (Seehuus et al. 2006b) and plays a major role in ageing (Omholt & Amdam 2004).

Vitellogenins in C. elegans (Murphy et al. 2003) and JH in Drosophila and mosquito (Tatar, Chien & Priest 2001; Tatar et al. 2003; Flatt et al. 2005; Brandt et al. 2005; Flatt & Kawecki 2007) are part of a regulatory system that has a positive effect on reproductive development and fertility, but a negative effect on life span. Yet, in the honey bee the relationship between JH and vitellogenin appears to contrast this norm: in queens and diutinus workers elevated vitellogenin levels are linked to low JH titres and correlate with extended life span (Fluri et al. 1981, 1982). In worker bees, the vitellogenin level is closely tied to the JH titre and to behaviour. High vitellogenin levels accompany low JH levels and nursing behaviour, while low vitellogenin levels are associated with high JH levels and may reinforce the forager behavioural state (Amdam & Omholt 2003). This correlation was hypothesized to explain negative effects that vitellogenin exerts on factors driving senescence (Omholt & Amdam 2004). For example, vitellogenin can act as a free radical scavenger to protect against oxidative stress (Seehuus et al. 2006b) and as a zinc carrier with a positive influence on maintenance of the cellular immune system (Amdam et al. 2004b).

Establishing RNA interference (RNAi) in the honey bee paved the way for functional testing of the proposed role of the vitellogenin gene in ageing regulation. Using this approach, a recent study assayed changes in behaviour and life span in response to vitellogenin knockdown (Nelson et al. 2007). Knockdowns started to forage earlier (precocious foraging) and exhibited a significantly shorter life span. However, the finding that vitellogenin controls multiple traits, not only foraging onset and thereby life span but also alters foraging preference (knockdowns bias their foraging efforts toward nectar), illustrates the key role vitellogenin holds in the regulatory networks that control life trajectories in the honey bee. This notion is further supported by RNAi studies confirming the regulatory interaction between vitellogenin and JH (Guidugli et al. 2005a; Amdam et al. 2007), and demonstrating that vitellogenin confers resistance to oxidative stress (Seehuus et al. 2006b). While there is now a wealth of evidence for a role of vitellogenin in honey bee life span regulation it is still unclear how the vitellogenin signal interacts with conserved pathways (i.e. pathways known to regulate ageing in other invertebrate and vertebrate models). In this regard, recent findings that point to associations of vitellogenin and JH with insulin/insulin-like growth factor signalling (IIS) and target of rapamycin (TOR) pathways (Page & Amdam 2007) can guide research aimed at understanding how conserved pathways may interact with unique regulatory features in order to orchestrate the great life span plasticity of honey bees (Fig. 1).

Figure 1.

The insulin/insulin-like growth factor signalling (IIS) pathway may influence ageing in honey bees through an interface with the feedback loop between juvenile hormone (JH) and vitellogenin. Conceptual drawing of a molecular network that includes honey bee IIS, JH and vitellogenin, and its role in ageing regulation. The model is based also on data from Drosophila and mosquito. Nurse bees (left) constitute a physiologically young phenotype, regardless of chronological age. It is likely that cues from the social environment, as well as nutritional status, mediate levels of insulin-like peptide 2 (ilp2) and TOR signalling. In nurse bees, resulting high titres of vitellogenin suppress synthesis of JH and the release of foraging behaviour, resulting in a slow-ageing phenotype. After a period of foraging activity (right), bees become physiologically old, irrespective of chronological age. Foraging behaviour is elicited when vitellogenin titres decline to release the inhibition of JH synthesis. High titres of JH, next, are correlated with increases in insulin-like peptide 1 (ilp1). As in solitary insects, JH, as well as ilp1, appear to be pro-ageing factors in the honey bee. Thus, foragers experience increased rates of senescence after the onset of foraging activity.

IIS and TOR signalling are undoubtedly two of the most important, highly conserved signal transduction pathways in eukaryotes and their impact on life span is well documented (Tatar et al. 2003; Kapahi & Zid 2004). Mutant or knockdown flies, mice, worms and humans for components of these pathways often are characterized by reduced size and fertility in combination with increased life span (Kimura et al. 1997; Abe et al. 1998; Böhni et al. 1999; Oldham & Hafen 2003; Kapahi et al. 2004; Kapahi & Zid 2004; Avruch et al. 2005; Suh et al. 2008). In mosquitoes, insulin can induce the expression of the vitellogenin gene, whereas RNAi knockdown of mosquito insulin receptor (InR), Akt (kinase involved in both pathways) and TOR inhibits insulin-induced vitellogenin expression (Roy, Hansen & Raikhel 2007).

Support for the connections between IIS, JH, vitellogenin, behaviour and life span in honey bees comes from observations of strains disruptively selected for foraging preference for either nectar or pollen (Page & Fondrk 1995). The strains represent remarkably different syndromes of physiology and behaviour that, again, include differences in levels of vitellogenin, JH, foraging onset and life span (Amdam et al. 2004a; Schulz et al. 2004; Page & Amdam 2007). These strains have been used to identify quantitative trait loci (QTLs) generically linked to the behavioural components of the syndromes. Noteworthy here is the fact, that the corresponding genome regions are highly enriched in IIS genes (Hunt et al. 2006). Based on these findings, accordingly, there are correlative data in support of connections between insulin and TOR signalling, JH and vitellogenin in honey bees (Ament et al. 2008).

In addition to potential roles in honey bee social regulation (Nelson et al. 2007; Nilsen et al. 2007), the IIS pathway appears to have been adapted to control the developmental differentiation of female worker and queen castes that also show dramatic differences in life expectancy. Recent data on RNAi-mediated knockdown of honey bee amChico (the insect homologue of insulin receptor 1 substrate, N. Mutti, F. Wolschin, A. Dolezal and G. V. Amdam, unpublished data) demonstrate that IIS is involved in regulation of body size and fertility, and similar results were obtained also for amTOR (Patel et al. 2007). These conserved roles of IIS and TOR gene action in honey bees support the idea of fundamental effects of the pathways also on somatic maintenance functions, including immunity (Saemann et al. 2007; Weichhart et al. 2007). While it would be premature to assume that IIS and TOR signalling affects life span interchangeably between honey bees, Drosophila and vertebrates, these pathways are shown to influence ageing in diverse organisms in a very similar fashion (Suh et al. 2008). A link between these systems and the honey bee vitellogenin–JH network is already evident. It remains now to be tested how vitellogenin, as a potent determinant of life span in honey bees, might has been exploited to affect IIS and TOR dependent cascades.

Biochemical changes during ageing in honey bees

This far, we have focused on signals that are regulators or determinants of longevity. Here, we shift to reviewing findings on the phenomenology of senescence and to discuss recent knowledge of the mechanisms that maintain organismal integrity.

Indicators of senescence are observed at different levels of cellular metabolism and are manifested as changes in metabolome (Kristal et al. 2007), transcriptome (Pletcher et al. 2002; McCarroll et al. 2004; Zerofsky et al. 2005) and proteome (Sharov & Schoneich 2007; Sowell et al. 2007). While metabolome information on C. elegans, D. melanogaster or A. mellifera, is currently scarce or not existent, the number of transcriptome and proteome studies on these species is rapidly growing, adding to our understanding of the ageing process (see below). In general, two different approaches can be distinguished for transcriptome as well as for proteome studies: whole-body approaches (Lund et al. 2002; McCarroll et al. 2004; Landis et al. 2004; Grotewiel et al. 2005) and studies focusing on specific tissues (Pletcher et al. 2002; Zhan et al. 2007). Both approaches are complementary and their overlap holds great promise for ageing research. Profiling the entire body allows assessment of major overall changes in the organism's physiology and has already contributed to fundamental insights into ageing (Lund et al. 2002; Landis et al. 2004; McCarroll et al. 2004; Grotewiel et al. 2005). Tissue-specific analysis, on the other hand, can enable more focused research and can reveal differences that are washed out by a whole-body approach. A common finding of whole-body studies focusing on transcript profiling is that mRNA coding for heat shock proteins, which are usually involved in protein folding, and for proteins/peptides with antimicrobial function are up-regulated during ageing (see also section on immunity below). On the other hand, mRNA coding for proteins involved in reproduction, for example, yolk- and fatty-acid-binding proteins, as well as for proteins involved in ATP-synthesis are down-regulated. Further refinement of these studies came from experiments using tissue-specific approaches (Corona et al. 2005; Zhan et al. 2007).

The yet most complete tissue-specific study on insects was performed by Zhan et al. They investigated the transcriptome of ageing Drosophila, describing changes over six time points in different tissues including brain, thoracic muscle, accessory glands, gut, testis, malpighian tubules and fat body (Zhan et al. 2007). The authors could show that several age-related changes are highly tissue-specific with < 10% transcript overlap between compartments, and that about half of the age-related genes are down-regulated while the other half is up-regulated. In effect, this work established that ageing is a highly complex and gradual process that affects all tissue types, albeit to different extents. The most affected transcripts were those coding for proteins involved in energy metabolism (muscle, brain, fat body), in protein degradation (brain, muscle), nutrient metabolism (gut, malpighian tubules, accessory glands), proteins involved in neurotransmitter release (brain), in stress resistance (accessory glands) and immune response (fat body), as well as cell-cycle related proteins (testis) and cytoskeletal proteins.

For the honey bee another recent paper describes changes abdominal, brain, and thoracic mRNA transcripts coding for antioxidant proteins with age. Corona et al. showed that, as long-lived queens age, there is a gradual decrease in the expression of putatively longevity-related mRNAs (Corona et al. 2005). In contrast, no pattern of decline in ageing worker bees was observed, and there was no marked expression difference between generally long-lived queens and short-lived workers. These findings indicate that, at the mRNA level, antioxidant enzymes are not necessarily correlated with longevity in the honey bee.

However, the abundance of mRNAs does not always translate into a corresponding abundance of proteins (Gygi et al. 1999; Hack 2004; Nie, Wu & Zhang 2006). Hence, these results need to be carefully interpreted. Protein and metabolite content can provide more accurate information on the active state of cells or tissues than transcript abundance. Accordingly, while scientists strive to integrate information on all biochemical levels (transcript, metabolite and protein), an increasing number of studies describe protein-level changes associated with ageing (Gafni 2004; Dremina, Sharov & Schoneich 2005). As of yet, few of these approaches focus on insects and available information is largely centred on D. melanogaster and A. mellifera (Sowell et al. 2007; Wolschin & Amdam 2007a). Sowell et al. examined protein changes in brains of adult flies at nine different time points. Similar to previous transcript studies they found heat shock proteins to be up-regulated and prophenoloxidase to be down-regulated with age (see also section on the immune system). Major protein groups that were affected include development and reproduction, general metabolism, and defence response.

In honey bee workers, it has been established that the abundance of proteins involved in glycolysis, ATP synthesis and free radical defence changes with the transition from nest to forager tasks (Schippers et al. 2006; Wolschin & Amdam 2007b). Foraging is typically performed by workers that are older than bees that perform tasks in the central nest (see Introduction). To decouple the effects of chronological age and social task on the proteome, therefore, a comparison was made between nest workers and forager bees before and after behavioural reversion. This refined approach allowed the study of proteomic differences in bees of different ages but that performed the same tasks (Wolschin & Amdam 2007a). The study demonstrates proteome changes with both age and behaviour visualized here by a principle component analysis (Fig. 2). It shows that workers at different behavioural stages (nest bees and foragers) and workers at the same stage yet of different age (nest bees before and after reversion) show distinct patterns of protein expression (Wolschin & Amdam 2007a).

Figure 2.

Principal components analysis (PCA) separating honey bees before and after behavioural reversion based on quantified proteins. Forager bees segregate from nurse bees (PC1, task-dependent separation), and nurse bees following reversion segregate from nurse bees before the reversion (PC2, age-dependent separation supported by nurse bees only). F, forager before the reversion; F_rev, forager after the reversion; N, nest bee before the reversion; N_rev, nest bee after the reversion. Figure redrawn from a PCA compiled through metagenealyse (

Although the effort to understand the roles of proteins in honey bee ageing is in its infancy, the studies outlined above provide an overview of age-related changes. In general, transcriptome and proteome directed studies can only provide correlates of the ageing process and conclusions drawn from these approaches have to be complemented by other assays. However, they have tremendous strength in providing information about which proteins and metabolic processes are affected, delivering larger scale data on candidates for further research.

Immunity, immunosenescence and recovery

Like vertebrates insects suffer from infectious diseases that are a prevalent threat, in particular to old individuals. Adult pathogenic diseases are considered to be a major cause of mortality in honey bees (Page & Peng 2001). Thus, defence mechanisms against infections likely are important determinants of honey bee survival. Remarkably, experiments on the honey bee immune system were the first to demonstrate how age-related deterioration on the cellular level reflects social environmental changes, such as nurse–forager transition and even reversion, that is, all the experimentally amenable life-history changes that make the honey bee system an outstanding model for studies of life span plasticity (Amdam et al. 2005).

Insects rely on innate immunity as a major line of defence against pathogens (Hoffmann 1995; Hoffmann, Reichhart & Hetru 1996). With the exception of the vertebrate's acquired immunity, that is, the antibody and T-cell response, there are many similarities in humoral and cell-mediated responses, both of which can be activated by microbial surface molecules (Martinelli & Reichhart 2005; Royet, Reichhart & Hoffmann 2005). Most of our current knowledge on insect innate immunity has been obtained from Drosophila, leading to the discovery of key processes like phagocytosis, destruction of microorganisms by antimicrobial peptides, free radicals, as well as RNAi-mediated inhibition of viruses (Zambon, Vakharia & Wu 2006; Wang et al. 2006).

The responses of the insect immune system are often induced by microbial peptidoglycans and signals are transmitted via nitric oxide and highly conserved signal transduction pathways including the Toll, IMD and Jak–Stat pathways (Lemaitre & Hoffmann 2007). As part of the innate immune system, proteolytic cascades that function in melanization and clotting are activated following an injury. Specialized cells in the haemolymph (haemocytes) attack and destroy invading microorganisms by phagocytosis, while, at the same time, antimicrobial peptides are produced by the fat body (cellular and humoral immune response, respectively). These peptides are released into the haemolymph and aid in the destruction of non-self cells that are recognized through protein–protein interactions (Hoffmann 1995). In Drosophila, the expression of antimicrobial peptides is modulated by the interplay of JH and ecdysone (T. Flatt, personal communication).

Less detailed information on the different pathways involved in pathogen defence is available for the honey bee, but current data based on behavioural reversion suggest that the regulatory control of honey bee immunity remains uniquely flexible during ageing (Amdam et al. 2005). Honey bee innate immunity was first defined by the number of functional haemocytes circulating in the haemolymph (Fluri et al. 1977), but later it was found that the humoral component also plays an important role in fighting bacterial pathogens (Josson et al. 1994).

Since the life expectancy of bees decreases substantially as soon as they start to forage, a likely functional correlate is an impaired immune response. Interestingly, the proportion of functional haemocytes, indeed, changes during honey bee worker ontogeny. The number of normal haemocytes is markedly reduced while the number of deformed or apoptotic (pycnotic) haemocytes is increased as nest bees convert to foraging behaviour (Wille & Rutz 1975). In other words, a change in task causes a depletion of functional immune cells. In addition, a reduction of haemolymph melanization capacity is associated with the onset of foraging activity. Melanization is a chemical reaction necessary for defence of pathogens and wound heeling. The reaction involves prophenoloxidase, which is synthesized and stored in haemocytes (Marmaras, Charalambidis & Zervas 1996; Soderhall & Cerenius 1998; Bedick et al. 2001; Zufelato et al. 2004). A reduced melanization response of foragers may therefore be linked to the lower number of normal haemocytes circulating in the haemolymph.

What directs the observed depletion of haemocytes, cells that are clearly beneficial for maintaining organismal integrity? The obvious key would be linking alterations of the immune system to signalling pathways involved in life span determination. Accordingly, it has been proposed that an interplay between JH and vitellogenin affects the number of normal haemocytes and thereby, the time-course of immunosenescence (Amdam et al. 2005). This proposition builds on the findings that diutinus bees have higher numbers of haemocytes, lower levels of JH and higher vitellogenin titres (Fluri et al. 1977). Also, when JH is injected into the haemolymph of worker bees, haemocytes are deformed (Rutz et al. 1974) and show a decrease in protein production that is characteristic of apoptotic cell death (Wille & Rutz 1975). In accord, JH injection is accompanied by decreased life expectancy and increased vulnerability to disease (Wille 1973).

Another major controlling factor in honey bee immunity might be the concentration of zinc in the haemolymph, since decreasing zinc levels also are associated with apoptosis of haemocytes (Amdam et al. 2004b). Interestingly, vitellogenin levels are closely correlated with haemolymph zinc levels and vitellogenin has been proposed to be the major zinc carrier in the haemolymph (Amdam et al. 2004b). Therefore, a link between JH, vitellogenin and zinc levels has been suggested (Amdam et al. 2004b, 2005), which may be responsible for changes in the honey bee immune response.

For antimicrobial peptides, recent studies describe the transcriptional and translational response of certain tissue types to a bacterial challenge and have found that in honey bees, as in other systems, antibacterial peptides are up-regulated following an infection. These studies mainly use profiling of transcripts (Evans 2004, 2006; Quinlan, Martin & Evans 2005) but a recent study also used a proteomics approach to detect changes in the honey bee brain after a bacterial challenge (Scharlaken et al. 2007). However, an increasing number of studies, including both transcript and proteome approaches, shows up-regulation of antimicrobial peptides with increasing age (Pletcher et al. 2002; Libert et al. 2006; Sowell et al. 2007; Zhan et al. 2007; Libert et al. 2008). It has been suggested that this pattern represents higher infection rates of older individuals, which is brought on by the decline of other components of the immune system, for example, prophenoloxidase and members of the JNK cascade (Pletcher et al. 2002; Zhan et al. 2007).

In agreement with these observations, ageing in humans, Drosophila, and other organisms is accompanied by a decline of the cellular immune system, which leads to a higher vulnerability of the body and thus to an increased risk of death (DeVeale et al. 2004; Zerofsky et al. 2005; Gruver, Hudson & Sempowski 2007). Further, recuperation of a defective immune system is usually only associated with recovery from infections as the decline of immune function is generally irreversible during ageing (Kay 1979; Gruver et al. 2007).

Yet, in the honey bee, when foragers are manipulated to revert back to nurse tasks, the situation is different. Rather than changing the pathological status, the experimenter changes the social environment and in turn intervenes in the behavioural, physiological and endocrine status of the organism. Thus, by applying this experimental paradigm it is possible to recover the number of normal haemocytes to levels that are indistinguishable from nest bees, hence from individuals that are typically younger (Amdam et al. 2005).

At present, it is unclear if the recovery of the cellular immune system is due to a production of new haemocytes or due to a recruitment of stationary cells that survived the nest bee to forager transition. It is also unclear how the interplay between JH and vitellogenin may affect such processes. The molecular mechanisms that underlie the reversibility of immunosenescence and honey bee ageing, thereby, clearly merit further investigation.

Oxidative stress and ageing

The free radical theory of ageing posits that reactive oxygen species (ROS) constitute a major driving force in ageing by introducing deleterious macromolecular damage (Harman 1956, 1981).

Specifically, ROS can oxidize the basic building blocks of life: proteins (Stadtman 1992; Sohal et al. 1993; Butterfield & Stadtman 1997), fats and DNA (Esterbauer & Cheeseman 1990; Moller & Loft 2004) to trigger effects as diverse as DNA damage, degradation of membrane proteins, disruption of signalling cascades and ultimately necrotic or apoptotic cell death (Richter et al. 1995; Dalle-Donne et al. 2006b; Muller et al. 2007). The electron transport chain of mitochondria is thought to be the major consumer of molecular oxygen and the principle site of ROS generation. A substantial fraction of molecular oxygen (0·4–4%) is only partially reduced and can thus generate the highly reactive ROS (Aguilaniu et al. 2005).

Several lines of evidence support ideas centred around the free radical theory of ageing. First, higher oxygen levels should increase the rate of ROS formation (Beckman & Ames 1998) and hence shorten life span. In accord, higher atmospheric oxygen levels accelerate ageing in C. elegans (Honda & Matsuo 1992) and in Drosophila (Baret et al. 1994). Second, application of antioxidants should extend life span. This was demonstrated for C. elegans, where life span was increased by 44% using a synthetic catalytic compound that mimics superoxide dismutase (SOD)–catalase activity (Melov et al. 2000). However, while a comparable study had provided evidence for a higher stress resistance induced by mimetics of natural antioxidants, it did not confirm a corresponding impact on life span for this treatment. Third, genetic manipulations that impair ROS formation by inhibiting the mitochondrial electron transport chain should increase life span. There is ample evidence for this prediction in C. elegans (Feng, Bussiere & Hekimi 2001; Muller et al. 2007). However, the life extending effect of partially inhibiting the mitochondrial respiratory chain appears to be confined to the nematode. In contrast, Drosophila and vertebrate models have a lower tolerance to extremely anaerobic conditions, so that similar manipulations would result in severe pathologies. In Drosophila, therefore, an alternative genetic approach is favoured to establish a causal link between ROS formation and life span. Increasing the level of ROS scavengers, as an example, seems like an obvious way to retard ageing. Indeed, over-expression of the genes encoding SOD1 and SOD2 has been shown to extend life span (Sun & Tower 1999; Sun et al. 2002). However, a great number of genetics studies focused on SOD or catalase activity instead of merely expression levels, and the inference on life extension is under discussion (Tatar 1999; Sohal 2002; Muller et al. 2007). Nevertheless, regardless of longevity effects, there is consensus that over-expression of antioxidant enzymes undeniably confers increased oxidative stress resistance in flies.

While some of the studies mentioned above draw a more complex picture of the link between free radicals and ageing, none of them dispelled the general assumptions posited by the free radical theory of ageing. Therefore, an increased level of oxidatively modified proteins is a widely accepted indicator of deleterious, senescence-associated molecular alterations (Levine & Stadtman 2001). One such alteration is the introduction of carbonyl groups into protein side chains by direct oxidation or by reaction with reactive oxygen intermediates (Stadtman 1990). In fact, the carbonyl content was shown to increase dramatically during the last third of life in several species, including humans (brain) (Smith et al. 1991) and the housefly (Sohal et al. 1993). Carbonylation apparently is irreversible and can only be removed by proteolytic degradation. Accordingly, an increase in cellular content of carbonylated protein is considered to be due to increasingly dysfunctional proteolytic pathways (Stadtman & Levine 2000; Shringarpure & Davies 2002) or due to accumulation of aggregates of heavily carbonylated proteins that are resistant to degradation (Dalle-Donne et al. 2006a).

Taken together, it is thus not surprising that the most widely used marker for global oxidative damage is protein carbonylation (Dalle-Donne et al. 2003). A common immunodetection protocol of carbonylated side chains makes use of the derivatization with 2,4-dinitrophenylhydrazine (DNPH) followed by the detection of the modified residue with DNP antibodies (Levine et al. 1990; Smith et al. 1998). Using this assay for immunohistochemistry and Western blot analysis, a correlation between ageing and the accumulation of carbonylated proteins was recently demonstrated in forager honey bee brains (Seehuus et al. 2006b). However, this prominent pattern of oxidative damage only occurred in the bees after an extended period of foraging activity. In all other behavioural classes, damage was at low levels independent of chronological age (Fig. 3). Thus, neither 8- and 20-day-old nurse bees nor 180-day-old diutinus bees showed elevated levels of carbonylation. These data support the notion that, also on the cellular level, the processes of senescence are not direct functions of chronological age in the honey bee.

Figure 3.

The level of oxidative stress damage in the bee brain is largely independent of chronological age. Immunolocalization of carbonylated proteins in the optic lobes of an 8 day-old nurse (a), a 180-day-old diutinus worker (b) and a 20-day-old forager bee (c). Immunopositive labelling, indicated by dark grey staining (arrows), is abundant only in the forager bee. In contrast, the diutinus bee that is chronologically older shows only low levels of carbonylation (modified from Seehuus et al. 2006a).

Two other studies on honey bees were aimed at explaining the striking discrepancy in task dependent longevity (i.e. for nurse or diutinus bee, vs. forager) and both provided evidence for vitellogenin conferring a lower susceptibility to oxidative stress. One line of evidence is the higher expression of vitellogenin in the exceptionally long-lived queens, which was documented for different compartments, that is, for the abdomen, the thorax and the head (Corona et al. 2007). While this study disclosed a correlation between vitellogenin expression and longevity, another study established a causal link between vitellogenin expression and oxidative stress resistance (Seehuus et al. 2006b). Using the RNAi technique, vitellogenin expression was reduced, which in turn lead to a higher mortality in bees that were injected with an oxidative stress inducing agent (paraquat). Moreover, the researchers showed that in comparison to other haemolymph proteins, paraquat preferentially induces carbonylation to vitellogenin, which is indicative of vitellogenin itself acting as an antioxidant. Though these data cannot yet explain the task dependent differences in carbonylation patterns of bee brains mentioned above.

To date, vitellogenin mRNA expression was confirmed for fat body cells (Corona et al. 2007). In principle, vitellogenin might be taken up by the brain through the haemolymph. This mechanism is well established for honey bee ovarian follicle cells that are immunopositive for vitellogenin (Guidugli et al. 2005b), but do not express vitellogenin themselves. The brain, and the insect's compound eye in particular, is an organ with an exceptionally high energy consumption (Laughlin, De Ruyter Van Steveninck & Anderson 1998). Foraging tasks require high visual activity, and thus an increase of mitochondrial ROS production in the forager brain is implied. Likely, this elevated oxidative load is paralleled by adaptations in alternative ROS scavenging pathways involving catalases, SODs or peroxidases.

Behaviour and ageing

In recent years Drosophila has become a model of choice for studying age-related behavioural deficits (Grotewiel et al. 2005). Senescence of locomotor activities has been reported for several performance tasks including negative geotaxis, a form of escape behaviour (Leffelaar & Grigliatti 1984; Gargano et al. 2005), phototaxis (Leffelaar & Grigliatti 1984; Simon, Liang & Krantz 2006), spontaneous locomotor activity (Minois, Khazaeli & Curtsinger 2001) and flight (Petrosyan, Hsieh & Saberi 2007). Part of the strength of the Drosophila system is the availability of several long-lived strains that can be used to resolve whether genetic predisposition for longevity can also alter behavioural senescence (Ganetzky & Flanagan 1978; Orr & Sohal 1994). Indeed, in mutants of a gene encoding the insulin receptor substrate, chico (Gargano et al. 2005) life span extension is accompanied by a delayed behavioural senescence. Nevertheless, the same behavioural aspect, that is, negative geotaxis, appeared to be unaltered in methuselah flies (Cook-Wiens & Grotewiel 2002). Thus, the authors conclude that life span extension does not necessarily confer protection from performance loss.

So far, only a few studies have attempted to assess relative changes across different performance tasks. Though, this integrative approach might reveal whether specific network functions are more prone to age-related damage than others. In Drosophila locomotion, geotaxis, and learning are declining progressively, beginning at 1–2 weeks of age. In contrast, electrical shock avoidance and the ability to escape free fall are preserved even in aged flies (Cook-Wiens & Grotewiel 2002; Simon et al. 2006). However, without proving structural data on damage of the neuronal substrate for a certain behavioural expression, these experiments can only indicate possible differences in the underlying circuits.

The honey bee, as a social insect, relies on a vast repertoire of sophisticated behavioural tools for communication. Flexibility and the ability to discriminate and memorize diverse odours, colours, visual patterns, textures and landmarks are prerequisites for efficient foraging. It is thus not surprising that the honey bee performs rather complex learning tasks very well and that test paradigms such as generalization, extinction learning and lateralization can be successfully applied. This portfolio includes paradigms that are commonly used in higher vertebrates (Stach, Benard & Giurfa 2004; Stollhoff, Menzel & Eisenhardt 2005; Letzkus et al. 2006). Although it has been suggested that cognition might only be applicable to humans, the senescence of cognition-like processes could also be studied in the honey bee (Menzel & Giurfa 2006). A recent study on workers demonstrates a functional decline in learning performance after an extended period of foraging activity (Behrends et al. 2007). Here, bees were trained to associate a specific odour (conditioned stimulus) with a sucrose reward (unconditioned stimulus). After several consecutive trials bees extend their proboscis, even before a reward is applied to the antennae. Bees of the same chronological age but engaged in different tasks were tested. The study showed that nurse bees performed equally well in learning independent of age. Yet in contrast, same-aged bees that were tested after different periods of foraging activity displayed significant differences in learning. Associative olfactory learning performance was specifically impaired in the bees that had foraged for more than 2 weeks (Fig. 4a). Thus, for the first time, this report provides evidence for a functional loss that exhibits a characteristic decoupling of chronological age from physiological age, thereby adding another dimension to previous findings at the cellular level (Seehuus, Krekling & Amdam 2006a). Also, it could be shown that bees that reverted from foraging to nest tasks were not impaired in olfactory learning. Task reversion, therefore, might protect from, or even compensate for, behavioural senescence. Interestingly, in the same study, no performance decline was found for sensory responsiveness, more precisely, for gustatory responsiveness (Fig. 4b). This again suggests that ageing might differentially affect neuronal circuitry (Simon et al. 2006).

Figure 4.

Task-dependent patterns of behavioural senescence in the honey bee. Whereas associative learning performance (a) decreases in bees with long foraging duration, the sensory responsiveness (sensory performance) of the corresponding animals remains intact (b). Left: bees were trained to associate an odour (conditioned stimulus, CS) with a sugar reward (unconditioned stimulus, US). Olfactory learning acquisition scores are a measure of the number of trials bees need to learn to associate CS and US, odour and sugar, respectively. (b) By applying sucrose solution of different concentrations to their antennae, bees were tested for gustatory sensitivity. No correlation between gustatory sensitivity and ageing was found. Bars indicate mean ± SE. Asterisks indicate significant differences; P < 0·05 (*) and P < 0·01(**) in two-tailed Mann–Whitney U tests. Modified from (Behrends et al. 2007).

The confirmation of a performance decline contrasts another study on honey bees, in which the authors failed to detect an age-dependent decline in learning performance (Rueppell et al. 2007). The obvious discrepancy between both studies might be an example of how pooling can yield contradictory or misleading interpretations. Whereas Rueppell et al. were using mixed populations of nurse and forager bees, only Behrends et al. controlled for task specificity. Thus, only the latter accounted for prior studies that indicate life span to be closely linked to the task a bee is engaged in (Page & Peng 2001; Amdam 2005; Seehuus et al. 2006b). Regardless of this shortcoming, one important conclusion of Rueppell et al. cannot be dispelled, that is the lack of functional senescence in survivors derived from a subpopulation with high mortality.

Neuronal correlates of senescence

Identifying the structural and molecular alterations within the central nervous system and within peripheral locomotor circuitries remains key to understanding behavioural senescence. Among the numerous hallmarks of age-related forms of neurodegeneration in vertebrates are: morphological changes, such as the reduction of spine number and dendritic complexity (Dickstein et al. 2007), shifts in synaptic transmission and neuronal excitability (Luebke et al. 2004; Disterhoft & Oh 2007), changes in apoptosis pathways (Bredesen, Rao & Mehlen 2006) and accumulation of protein aggregates (Hardy & Selkoe 2002). Since relatively little is known about these phenomena in invertebrates, we will highlight only a few studies that have promise towards increasing the understanding of how neurodegenerative processes can shape the course of behavioural senescence in the honey bee.

Several reports observed long-term morphological changes of neurons during adulthood in Drosophila (Technau 1984; Beramendi et al. 2007), in the honey bee (Farris, Robinson & Fahrbach 2001) and in Pheiodole dentata– an ant (Seid, Harris & Traniello 2005). Intrinsic neurons of the mushroom body, a structure involved in memory consolidation, have longer and more complex neurites in long-term forager bees than in short-term forager bees (Farris et al. 2001). In ants, the synapse and vesicle number of presynaptic boutons of mushroom body neurons increases with age (Seid, Harris & Traniello 2005). While the latter studies aimed at analysing experience dependent alterations, two other reports both observed two-phase structural dynamics by covering the full life span of Drosophila. During early adulthood the neurite number in mushroom body neurons increases (Technau 1984) but decreases again in older flies. Likewise, different morphological parameters of the neuromuscular junction only decrease during late adulthood while being maintained or increased during early adult life (Beramendi et al. 2007). Another recent finding in Drosophila may emerge as pivotal when addressing the link between structural loss and its molecular fundamentals (Martin-Pena et al. 2006). By counting synapses the authors could demonstrate that enhanced phosphatidylinositol-3-phosphate-kinase (PI3K) activity induces functional synapses in larval motoneurons as well as in brain projection neurons even of aged flies. The suggested neuroprotective function could, thus, make PI3K key in understanding the regulation of age-related neuronal dysfunction.

How ageing affects neuronal correlates of learning and memory is one of the most intriguing questions in ageing research. In honey bees as well as in Drosophila the diverse forms of memory, for example, long- and short-term memory, rely on the activation of different signalling cascades and can be selectively controlled by different behavioural tests and pharmacological treatments (Davis 1996, 2004; Tully et al. 1996; Muller 2000; Menzel 2001). Using behavioural test paradigms, two recent studies report that ageing, indeed, differentially affects different forms of memory in Drosophila (Tamura et al. 2003; Mery 2007). Another promising approach, however, is to track temporal and spatial changes of proteins involved in memory consolidation. Interestingly, a proteomics study on heads of ageing Drosophila revealed age-related changes for hundreds of proteins belonging to metabolism, primary development, reproduction and immune responses (Sowell et al. 2007). Another study based on transcript profiling of brain tissue obtained from ageing Drosophila showed that genes encoding for proteins involved in neurotransmitter release, protein degradation and energy production change during ageing (Zhan et al. 2007). While these studies revealed many interesting and important differences, they lack the direct association with learning and memory, two of the most important organismal functions heavily impaired by ageing processes.

We recently profiled essential proteins of the honey bee's central brain signal transduction pathways, some of them involved in learning and memory (Fig. 5). This proof-of-principle study shows that it is possible to monitor several proteins simultaneously, which are supposedly crucial to maintain neuronal connectivity and plasticity. Amongst others, these are protein kinase A (PKA), protein kinase C (PKC) (Davis et al. 1995; Menzel 2001), Leonardo (Broadie et al. 1997) and synapsin (Godenschwege et al. 2004; Michels et al. 2005). Moreover, the results indicate an age-dependent decrease in levels of some, but not all of these proteins. While, for example PKC and synapsin levels drop with foraging age, PKA levels remain constant. These observations are in good agreement with a previous report on nest bees of different ages (Humphries et al. 2003), and show that monitoring age-related decline of proteins involved in memory and behaviour is feasible.

Figure 5.

Selected proteins identified in the honey bee brain using a proteomics approach. For simplification, all enzymes are shown in one cell. GPCRK, g-protein coupled receptor kinase; GNBP, guanine nucleotide binding protein; GDPDI, GDP dissociation inhibitor; ArgK, arginine kinase; NDK-1, nucleoside diphosphate kinase; IP3R, inositol-triphosphate-receptor; Ppase, phosphatase.

Conclusions and future perspectives

To decipher how environmental factors modulate the molecular pathways of longevity regulation is a central question in ageing research. New insights are currently obtained through highly developed systems such as Drosophila and C. elegance, which provide a continuous flow of in depth information (Libert et al. 2007). Yet, in social hymenoptera cast evolution has shaped societies with discrete, albeit genetically interchangeable phenotypes that differ in life span by more than a factor of 10 – the most developed study system being the honey bee. Its potential as a major player in future biogerontological research was outlined not long ago (Page & Peng 2001; Omholt & Amdam 2004; Rueppell et al. 2007) and a steadily increasing number of studies has unveiled key signals of the molecular regulatory networks that influence ageing and describe senescence on levels ranging from tissue specific protein profiles and patterns of cellular immunity to learning impairment.

The plasticity of honey bee worker ageing is a naturally occurring phenomenon. However, its reliance on the social environment can be exploited as a powerful tool to experimentally manipulate life-history schedules. Apart from providing novel opportunities for expanding knowledge at this interface between ageing regulation and sociality, the honey bee gives ample options for comparative studies in which regulation of senescence is contrasted to solitary species. Evolution of advanced division of labour systems in the honey bee has been hypothesized to build on co-option of pleiotropic control pathways of reproductive physiology, behaviour and ageing that also are present in solitary insects due to shared ancestry (Amdam et al. 2004a, 2006). Here, the essentially sterile honey bee worker caste can serve as an exemplar to illustrate how IIS and TOR signalling pathways, when largely decoupled from direct reproduction, may evolve to relay social environmental changes and control plasticity in behaviour dependent patterns of ageing. The results presented here underline that studies aimed at understanding the connectivity of these regulatory networks are highly warranted.

Methods to further increase knowledge on molecular and structural aspects of honey bee ageing, such as gene silencing (Amdam et al. 2003; Guidugli et al. 2005a; Patel et al. 2007; Nelson et al. 2007), labelling of free radical damage (Seehuus et al. 2006b) and quantitative protein profiling (Schippers et al. 2006; Wolschin & Amdam 2007a,b) are already established. Similarly, available pharmacological and neurophysiological techniques enable the neural substrates of learning to be monitored and the different forms of memory consolidation to be dissected (Menzel, Leboulle & Eisenhardt 2006).

These prerequisites for an integrative approach to study ageing have already led to the intriguing data presented here and further provide a framework to stimulate multidisciplinary approaches. From this foundation, we believe, the honey bee can emerge as a model system to address some of the most startling questions in current ageing research.


This work was supported by the Norwegian Research Council with grant FUGE#175413 (to Daniel Münch and Gro V. Amdam), the National Institute of Aging NIA P01 AG22500 and the PEW Foundation (to Gro V. Amdam), and by a Feodor Lynen fellowship of the Alexander-von-Humboldt-Foundation, Bonn (to Florian Wolschin). We thank Thomas Flatt, Kate E. Ihle, Colin Brent, Navdeep Mutti, Nick Baker, Christina Tolfsen and Carsten Duch for helpful comments on the manuscript. Figure 1 was designed by Kate E. Ihle.