Longevity in the Honeybee (Apis mellifera): Expression of Telomerase and Insulin Signaling Pathway Genes in Queen and Worker Bees

Author:  Eli Hornstein

Institution:  North Carolina School of Science and Mathematics
Date:  February 2008


A honeybee queen, the only fertile female in a colony, can live up to 47 times as long as her sterile female workers. This contradicts the conventional wisdom of a tradeoff between increased reproduction and improved health and longevity. Whereas most organisms have decreased longevity the more they reproduce, in eusocial insects, especially hymenopterans, this relation is reversed and the individuals that breed more (queens) have longer life-spans than non-reproductives (workers). This discrepancy leads to the question of what might be the physiological mechanism for individuals that reproduce more, outliving by many times individuals that do not reproduce. Longevity was examined at the genetic level in an attempt to find a correlation between reproduction and various factors of longevity. This study examined expression of two genes that are known to affect lifespan: Telomerase reverse transcriptase (Tert) and Forkhead Box subgroup O (FOXO). Three sets of experiments were conducted. First, Tert and FOXO expression was compared in queens and workers. Second, Tert expression was compared in workers that had been exposed to queen mandibular pheromone (QMP) or untreated controls (QMP dramatically alters bee physiology). Finally, Tert and FOXO expression were compared in mated and unmated queens. Quantitative real-time PCR was used to calculate relative expression. Queens produced more Tert and less FOXO than workers. Tert expression in workers was not affected by exposure to QMP. Virgin queens had greater expression of FOXO and lower expression of Tert relative to mated queens. Both FOXO and Tert expression seem to be linked to either longevity or reproductive potential. FOXO expression is lower in mated queens versus virgins, while Tert expression is higher in queens versus workers and in mated queens versus virgins. FOXO has also been correlated with longevity in QMP treated workers, possibly by slowing metabolic processes. Studies to confirm the activity of FOXO and Tert in longevity and reproduction are needed, including reducing expression levels with RNAi and monitoring the effects on longevity and reproduction.


Species that reproduce early and often, such as the mouse, have very short lifespans in comparison to species that reproduce late in life and infrequently, such as the elephant. The conventional wisdom is that there is a tradeoff between reproduction and longevity (Westendorp and Kirkwood 1998), but in eusocial insects such as honeybees, ants and termites the longevity of reproductive queens contradicts this principle (Hartmann and Heinze 2003). In eusocial insects, queens can live for years while workers live for weeks or months (a queen honeybee may live 47 times longer than nonbreeding workers) (Page and Peng 2001).

The exact cause of longevity in social insects is unknown but research on other organisms indicates that repair to the ends of chromosomes via the enzyme telomerase may increase lifespan (Blackburn 1990). Telomerase is a generic term that includes two genes: telomerase reverse transcriptase (Tert) and the telomerase RNA component (Terc). Tert codes for the enzyme that operates to maintain the loops of DNA that secure the ends of the chromosomes, known as telomeres. Terc is not translated and the RNA it transcribes serves as a template for the Tert polypetides. The specific function of Tert is to add canonical TTAGGG sequences via reverse transcription onto the ends of DNA sequences (Blackburn 1990). Honeybees have a telomerase very similar to that found in humans, but Tert may not be the only operator in longevity. For example, Drosophila does not use telomerase to repair its chromosome caps, and telomerase was only recently discovered in other arthropods (Osani et al. 2006).

Another candidate gene for longevity is FOXO. FOXO is a gene primarily related to the insulin receptor cascade, which regulates cell division. One of the main factors in longevity is rate of cell division, controlled in part by factors related to FOXO. If an organism has slower cell division rates, its longevity is increased because interphase is prolonged in the cell cycle, slowing cell growth and indirectly regulating division. In Drosophila melanogaster, the life spans of insulin receptor mutants were extended by more than 50% when FOXO was overexpressed (Hwangbo et al. 2004). FOXO has also been shown to be active in the mediation of cell death (Giannakou and Partridge 2004). Consequentially, FOXO may retard the signals that lead to cell death, thus prolonging the lifespan of an individual.

This study examined Tert and FOXO expression in honeybees to compare three factors that may be associated with longevity. The first factor was caste. Because queens live longer than workers, it is expected that queens would have higher expression of FOXO and Tert. The second factor was exposure to Queen Mandibular Pheromone (QMP). Previous experiments did not find any effect of QMP or starvation on FOXO levels, but did find a general correlation between QMP exposure and starvation resistance, which is related to longevity by the requirements of cellular upkeep present in both old age and starvation stress (Fischer and Grozinger, submitted). Because QMP has been shown to improve the functioning of one process related to longevity, Tert expression was compared in QMP treated (starvation resistant) and non-QMP treated worker bees to determine if Tert may be the mechanism for QMP-related starvation resistance. The third factor compared was reproductive status. Virgin and mated (both non-laying and laying) queens were compared in an attempt to link specific breeding status to longevity. Research indicates that only queens of eusocial insects that receive viable sperm maintain their unusual longevity (Schrempf et al. 2005). The comparison of queens that had not received sperm (virgins), received sperm but had not used it (non-laying), and those that had been actively reproducing (laying) should demonstrate transcriptional differences that were brought on by mating or reproduction. These studies can help determine if common molecular pathways associated with longevity in other animals have been co-opted to produce the unusually long lifespan observed in reproductive bees.


Study organism

The organism under investigation was the honeybee, Apis mellifera carnica. A. mellifera is a eusocial insect, meaning that it has division of labor within the hive, overlap of generations, and cooperative care of young. Honeybees are also haplo-diploid, meaning that queens can lay unfertilized eggs that develop into males, or fertilized eggs that develop into females (Winston, 1987). Queens and workers are produced from fertilized eggs and develop differently due to differences in larval nutrition starting 3 days after hatching (Winston, 1987).

The rearing, dissections, RNA extraction and cDNA synthesis were done by members of the Grozinger Lab. The queens (Apis mellifera carnica) were reared at the NCSU Bee Research Facility in Raleigh, NC. The source colony used for the experiment was headed by a queen (Glenn Apiaries, Fallbrook, CA) instrumentally inseminated with semen from a single drone. Queens were produced by grafting larvae and reared in queenless colonies for 7 days. Workers were obtained from the same source colony as the queen, to reduce genetic variation. The next four sections were performed previously in the Grozinger lab and stored cDNA was made available to me for my experiments.


Prior to emergence as adults, queen cells were placed with 35 one-day-old workers in Plexiglas cages. In total, 8 queens were used, each placed in a separate cage. Additionally, two cages of 35 one-day-old workers from the source colony were created. Food and water were changed daily. The two cages of workers and the ten cages with queens were collected by submerging in liquid nitrogen when the bees were 10 days old. RNA was extracted from the brains and used to produce cDNA from 8 queens and 8 workers, which was used in this experiment.

Queen Mandibular Pheromone

Cages of bees were treated with either 0.1 Queen equivalent of queen mandibular pheromone (QMP) in 1% isopropranol or a solvent control (1% isopropanol). After 3 days of treatment following cage set up, the bees were collected into liquid nitrogen. RNA was extracted from abdomens (dissected to remove the internal organs but retain the fat bodies) and used to produced cDNA from 8 treated workers and 8 nontreated workers, which was used in the experiment.

Reproductive Status

Queens were randomly assigned to a group (Virgin or naturally-mated), marked, and returned to the colony. Two days later, laying status of mated queens was determined (laying, non-laying) and all queens were collected on dry ice and stored at -80 C for processing. RNA was extracted from abdomens (dissected to remove the internal organs but retain the fat bodies) and used to produce cDNA from 8 treated workers and 8 nontreated workers, which was used in the experiment.

RNA extraction and cDNA synthesis

RNA was isolated from dissected brains (for the caste studies) or fat bodies (for the other two studies) using an RNeasy RNA extraction kit (Qiagen, Valencia CA), yielding 0.6-2 μg/sample. The cDNA was synthesized from 100 ng RNA using an Applied Biosciences cDNA synthesis kit.

Primers - All primers used were custom-designed by Primer Express software (Applied Bioscience) and purchased from Sigma, using A. m. gene sequences attained from the BLAST database. Quantitative real-time PCR - To perform quantitative real-time PCR (qRT-PCR) two µL samples of cDNA were loaded, into a 384 well plate with nine replicate wells per bee. All primers used were in the concentration 10 µM. A master mix was created for each gene by adding 1 µL each of forward and reverse primers, 1 µL dH2O, and 5 µL of SYBR-Green PCR master mix (Applied Biosystems, Foster City, CA) per 2 µL cDNA. The microcentrifuge tubes were spun briefly in a tabletop centrifuge to homogenize and were then placed back on ice. Using a multi-pipetter, 8 µL of the master mix was aliquoted per well for a total reaction volume of 10 µL. An extra block of wells was also loaded with only primer mix as a negative control. In total, each primer was used in 3 replicate cDNA wells.

Plates were sealed with a clear plastic adhesive sheet and placed in a centrifuge and spun for 5 minutes at 1000 rpm. The plate was placed in the quantitative real-time PCR (qRT-PCR) machine (ABI Prism 7900HT, Applied Biosystems). A new file was set up using the SDS qRT-PCR software with the temperature cycle of 50°C for 2 minutes, 95°C for 10 minutes, 95°C for 15 seconds, and 60°C for 1 minute. The last two steps were repeated 45 times. Although a standard PCR machine uses three individual steps for annealing and extension, the polymerase employed in this experiment operates at 60°C, allowing primer annealing and extension to happen in the same step, at 60°C.

Quantification was based on the number of PCR cycles (CT) required to cross a threshold of fluorescence intensity (ABI User Bulletin 2 described in Bloch et al. 2001). For each individual sample, the ratio of the expression level of the gene of interest to that of the control gene (eIF-S8) was calculated. The statistics were based on the difference between FOXO and eIF-S8 levels or the difference between Tert and eIF-S8 in the different samples being compared. The data were expressed as fold changes relative to the value of one data set (i.e., queens vs workers). Data were analyzed using JMP INTRO 5.0.1 from 2002.



A comparison of Tert and FOXO transcription levels (relative to the eIF-S8 control which stays constant) in queens and workers (fig. 1) found that queens had greater levels of Tert, than workers (T-test, N=16, p=0.02) but there was no significant difference in FOXO between queens and workers (T-test, N=16, p>.05). Workers produced significantly more FOXO than Tert (T-test, N=16, p less than 0.0001). However, it should be noted that differences in primer efficiency, rather than RNA levels, could account for differences when comparing Tert and FOXO without a standard curve.

QMP and starvation

There was no significant difference in Tert expression between worker bees exposed to QMP and bees not exposed to QMP (T-test, N=16, p>0.05,) (fig. 2). QMP treated bees had slightly lower levels of Tert mRNA than the non-treated bees, although the difference was not significant.

Reproductive Status

There were no detectable differences between laying and non-laying queens for either Tert or FOXO, so for statistical purposes they were grouped them together under the category of mated queens'. A comparison of mated queens (laying and non-laying), and virgin queens showed that mated queens had higher levels of Tert (T-test, N=14, p=0.0309) and virgin queens had higher levels of FOXO than mated queens (T-test, N=14, .0474). Virgin queens also produced much more FOXO than Tert (T-test, N=10, p less than 0.0001)(fig.3).


This study tested the hypothesis that queens, reproductive queens, and QMP-exposed starvation-resistant workers would show upregulation of Tert and/or FOXO related to increased longevity. Tert maintains chromosomal integrity, and so may influence lifespan through pure quantity, allowing cells to keep dividing longer, while FOXO, through a more complex process, can change metabolism, and prevent cells and organs from deteriorating. Lifespan in honeybees is an excellent model for study of these two genes, as discovery of factors in the exceptional reproduction-regulated longevity system of honeybees may lead to better understanding of the causes of organism-scale tradeoffs between longevity and reproduction. The ultimate goal was to determine if there is a specific correlation between FOXO or Tert in longevity, and to understand the genetic mechanism of longevity in an exceptional case.

The main finding was that Tert is regulated in bees by reproductive state and caste but not by QMP exposure. As indicated in Figures 1 and 3, the relative levels of Tert mRNA differed between workers and queens and between queens of different reproductive status. The difference between workers and queens can be expressed most simply as a 1.25 fold difference from workers to queens in Tert mRNA levels. This difference can be linked to reproduction or metabolism. Because fat body cDNA was used in the reproductive status analysis, and fat bodies do not have actively dividing cells, Tert does not have an obvious source in the fat body. However, the fat body in mated queens produces much higher levels of vitellogenin, an egg-yolk protein, resulting in greater activity of the fat body in general. In the caste experiment, because cDNA extracted from the brain was used to compare queens and workers, there also should not be any dividing cells in consideration. Genes associated with oxidative and metabolic processes have been found to be expressed in much higher levels in queen brains as opposed to worker brains. Because Tert expression correlates with changes in metabolic activity associated with reproduction, it is possible that there is a link between Tert and metabolism, as well as cell division.

Figure 1 - Comparison of relative Tert and FOXO levels in queens and workers. One standard error bar is shown.

Figure 1 - Comparison of relative Tert and FOXO levels in queens and workers. One standard error bar is shown.

Figure 3 - Comparison of Tert and FOXO levels in laying, non-laying, and virgin queens.

Figure 3 - Comparison of Tert and FOXO levels in laying, non-laying, and virgin queens.

Tert may also act as a tool in response to role shifts in honeybees, which are in turn mediated by metabolism-related genes. Amdam and Page (2005) found that while worker bees ordinarily go through a specific progression of roles, they can revert to previous roles if necessary for the hive, and can also enter a starvation-resistant state which increases lifespan tenfold. Tert may allow older bees to stay alive when hives are underpopulated, and may maintain cells during dormant periods.

Figure two indicates that QMP exposure is not responsible for changing levels of Tert mRNA. This is consistent with another study showing that QMP is not responsible for changing levels of FOXO (Fischer and Grozinger, in review). The Fischer and Grozinger study examined starvation resistance (quantified by survival time without food) and FOXO levels in QMP exposed and non-QMP exposed bees, and found no significant difference in FOXO levels between QMP exposed and non-QMP exposed worker bees. However, a correlation between QMP exposure and starvation resistance was shown, suggesting that other longevity-related genes may be linked to QMP. My results did not detect Tert as a link between QMP and longevity.

Figure 2 - Comparison of relative Tert levels in No Treatment (NT) and QMP treated worker bees.

Figure 2 - Comparison of relative Tert levels in No Treatment (NT) and QMP treated worker bees.

Differences were observed between mated and non-mated queens, but no differences were observed between laying and non-laying queens. It may be that seminal fluid acts in part to lower activation of FOXO, and other aging-related genes. Non-laying queens had not completely transitioned to the fully mated state, and thus were more like virgins, suggestive that mating alone is not sufficient, but the subsequent physiological changes. It is also by the same logic in the first graph that the lower levels of telomerase in virgins can be explained. Because the mated queens are producing eggs, there is more division that requires the activation of Tert. The virgin queens are not yet producing eggs, and therefore have lower levels of telomerase.

FOXO may act as a temporary means of longevity, as its expression is reduced in mated queens, while Tert trends to increase in mated queens. This suggests that different methods of longevity are emphasized by seminal fluid exposure, or other factors unique to mating. Hwangbo et al. (2004) found that FOXO activity increases lifespan in Drosophila, which confirms my results on a basic level. Hwangbo et al. (2004) also found that FOXO activity in the fat body was related to longevity, which was reflected in the honeybee results. The fat body, however, is not closely related to the act of reproduction, as my speculation would require.

Overall, this experiment provided a potential link between both Tert and FOXO in aging. Although informed speculations have been made on this link, further studies As a continuation, further experimentation would be done in the active bee season using a larger number of samples, with a focus on the queen/worker and reproductive status differences. A comparison of samples from brains, abdomens, and neutral sources (e.g. legs) would be done in the same context for both genes. By performing another experiment with standard-curve PCR, we would be able to compare FOXO and Tert to each other. RNAi could be used to lower FOXO and Tert levels in vitro and possibly in vivo, which would allow the observation of the effects of FOXO and Tert at both a cellular and organismal level. The impact of this research is related primarily to what causes aging in humans, and how to slow the process of aging. By learning more about the factors of decay related to Tert (cell division) and FOXO (insulin pathways) we can begin to devise a way to prevent the onset of aging on a cellular, rather than whole-organism level.


I would like to thank Dr. Christina Grozinger for allowing me to work in her lab, and for helping me immeasurably with experimental design, research and analysis. Also, I would like to especially thank Patrick Fischer for training me and guiding me on a day-to-day basis, and for spending many a late night slaving over 384 well plates. Sarah Ayroles, Freddie-Jeanne Richard, Brendon Fussnecker, Yongliang Fan, and the rest of the Grozinger lab were also incredibly helpful. Brian Wood encouraged me and always pushed me to think differently.


Amdam, GV, and RE Page Jr. (2005) Intergenerational transfers may have decoupled physiological and chronological age in a eusocial insect. Ageing Research Reviews 4: 398- 408.

Blackburn, EH (1990) Telomeres and their synthesis. Science 249: 489- 490.

Bloch, G et al. (2001) Behavioral rhythmicity, age, division of labor and period expression in the honey bee brain. Journal of Biological Rhythms 16: 444- 456.

Giannakou, ME and L Partridge (2004) The interactions between FOXO and SIRT1: tipping the balance towards survival. Trends in Cell Biology 14: 408.

Grozinger, CM et al. (2003) Pheromone-mediated gene expression in the honey bee brain. PNAS 2: 14159- 25.

Hartmann, A and J Heinze (2003) Lay eggs, live longer; division of labor and lifespan in a clonal ant species. Evolution 57: 2424- 2429.

Hwangbo, DS et al. (2004) Drosophila dFOXO controls lifespan and regulates insulin signaling in brain and fat body. Nature 249: 562- 566.

Osanai, M et al. (2006) Identification and characterization of the telomerase reverse transcriptase of Bombyx mori (silkworm) and Tribolium castaneum (flour beetle). Gene 376: 281- 289.


Page, RE Jr., and CY-S Peng (2001) Aging and development in social insects with emphasis on the honey bee, Apis mellifera L.. Experimental Gerontology 36: 695- 711.

Schrempf, A et al. (2005) Sexual cooperation: mating increases longevity in ant queens. Current Biology 15: 267- 270.

Schulz, DJ and GE Robinson (1999) Octopamine influences division of labor in honey bee colonies. Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology 187: 53- 61.

Westendorp, RG and TB Kirkwood (1998) Human longevity at the cost of reproductive success. Nature 396: 743- 746.

Whitfield, CW et al. (2003) Gene expression profiles in the brain predict behavior in individual honey bees. Science 302: 296- 299.

Winston, ML (1987) The Biology of the Honey Bee. Harvard University Press. Cambridge, MA.