A widely held view in ageing research is that the rate of ageing is influenced by the division of resources between growth rate and reproduction on one hand, and somatic maintenance on the other 1. This theory proposes that evolution sets the level of somatic maintenance to be just sufficient to prevent damage that would interfere with growth and reproduction but not waste energy that could be otherwise used to produce offspring or increase the rate of development. Thus, rather than being an active programme, ageing results from the eventual failure of the somatic maintenance set by this balance. However, the balance of resources between somatic maintenance and reproduction must also be responsive to the environment because external conditions affect growth and development and can vary considerably. Understanding this plasticity may help us understand how ageing is regulated and is subject to external interventions.

In the nematode Caenorhabditis elegans, the FOXO transcription factor DAF-16 plays a major role in mediating changes in resource allocation in response to environmental change. Firstly, DAF-16 is required for entry into the diapausal (developmental and growth arrested) dauer stage under conditions of starvation and overcrowding 2. Formation of this stress-resistant alternative to the normal third larval stage (L3) involves several morphogenetic and metabolic changes and upregulation of somatic maintenance genes. Dauer larvae can survive for several months until they find food, upon which they resume normal development into adults that have a normal lifespan 3.

Secondly, outside its role in dauer formation, DAF-16 activity is associated with increased stress resistance and survival and reduced reproductive fitness 4. Most studies of DAF-16 have focussed on its role in mediating the phenotypes of mutants of the C. elegans insulin/IGF like signalling (IIS) pathway. Activation of this pathway leads to the translocation of DAF-16 from the nucleus to the cytoplasm. When this pathway is disrupted by mutations such as reduction-of-function mutants of the daf-2 insulin-like receptor or the age-1 Class IA phosphoinositide 3-kinase (PI3K), the propensity to form dauers, stress resistance and adult lifespan are all increased as a result of DAF-16 activation 356. Studying gene expression in these mutants has led to the identification of many genes upregulated by DAF-16 789. These genes are involved in several processes that may increase lifespan, including those clearly involved in somatic maintenance such as heat shock proteins and antioxidant proteins. Consistent with the cost of DAF-16 activation, long-lived daf-2 animals show reduced fertility and cannot compete with the wild type 1011. Furthermore, overexpression of DAF-16 using transgenes slows development 1213 and daf-16 reduction of function by RNAi or mutation causes worms to reach reproductive adulthood slightly before wild type animals (4, unpublished observations), suggesting that even in the wild type, DAF-16 acts to slow development.

In contrast to daf-2 mutants, the long-lived age-1(hx546) mutant is able to compete with the wild type over several generations when the two strains are mixed under standard culture conditions 14. However, the age-1(hx546) mutant loses the ability to compete with the wild type if the mixed population is exposed to repeated cycles of starvation and feeding 14. These competition experiments are very sensitive to small changes in fitness and suggest that the age-1(hx546) mutation does not cause constitutive activation of DAF-16 but interferes with the ability of the worm to adapt to changing conditions. In agreement with this conclusion, a study of DAF-16 localisation reported no effect of the age-1(hx546) mutation under normal culture conditions 4. age-1(hx546) is a reduction-of-function mutant rather than a null mutation and its molecular nature is unknown 1516. Interestingly, only with age do age-1(hx546) mutants become resistant to hydrogen peroxide and show elevated levels of superoxide dismutase and catalase activities 1718. In summary, the age-1(hx546) mutation appears to affect the animal only during starvation and in ageing adults, so understanding how AGE-1 regulates DAF-16 during these conditions may reveal how this mutation causes increased longevity.

In this study, we focus on the starvation-induced first larval stage (L1) diapause [Additional File 1]. Like the dauer stage, the L1 diapause allows C. elegans to survive periods of long-term starvation. Starved L1s survive several weeks and time in this condition has no effect on subsequent adult lifespan 19. However, unlike the dauer diapause, there are no obvious morphological changes in the starved L1 that might protect the larvae from external stresses. We investigated how L1 larvae respond to oxidative stress and have shown that starvation induces resistance to H₂O₂ in a daf-16-dependent manner. Consistent with these results, DAF-16 was localised in cell nuclei after short-term starvation. However, after long-term starvation, DAF-16a::GFP translocated to the cytoplasm and this translocation was absent in the age-1(hx546) mutant. We show that a similar effect occurs in old worms as they become starved with age. These data suggest a novel mechanism underlying longevity in weak age-1 mutants and reveals an unexpected negative regulation of FOXO transcription by Class IA PI3Ks in response to starvation.

To investigate the function and regulation of DAF-16 under conditions in which C. elegans encounter starvation, we turned to the L1 diapause. This diapause state is entered once larvae hatch from their eggshells in the absence of food. These larvae arrest in development and are viable without food for weeks, re-entering normal development when food is encountered. Unlike the dauer diapause, the L1 diapause can be entered (and exited) by mutants lacking functional daf-16. This property makes the L1 diapause a physiologically relevant model for studying DAF-16 function and regulation in response to starvation and other stresses.

Most studies of daf-16 function have been carried out in the context of its role in the stress-resistance and extend lifespan of IIS mutants. Under standard growth conditions, mutation of daf-16 alone only slightly increases susceptibility to stress 2021. However, in this study we find that in the L1 diapause, mutation of daf-16 has a dramatic effect on the response of larvae to an acute treatment of hydrogen peroxide. For these experiments, we isolated L1s hatched in the absence of food in an aqueous phosphate buffer and exposed them to a range of concentrations of H₂O₂ for 45 minutes (see Methods). We then washed away the H₂O₂ and transferred the worms to agar plates containing bacterial food to reinitiate development and assess the long-term effects. The short-term effect of the treatment was to inhibit movement, but the major long-term effect was to cause an abnormal developmental arrest. For example, wild type larvae assessed two days after a treatment of 1 mM H₂O₂ remained permanently at a size corresponding to L2 or L3 stage and displayed very sluggish, uncoordinated movement. Starved L1s with loss-of-function mutations in daf-16 were very sensitive to H₂O₂ treatment: 0.25 mM H₂O₂ caused over 80% permanent arrest of daf-16(m27) or daf-16(mu86) mutants compared to no permanent arrest for the wild type (Figure 1B, F). The H₂O₂ sensitivity of daf-16 mutants was fully rescued by a transgene expressing a DAF-16a::GFP fusion protein 13 (Figure 1C), confirming the role of daf-16 in conferring H₂O₂ resistance.

To investigate the role of starvation, we tested the H₂O₂ resistance of L1s hatched in the presence of food on agar plates. In contrast to starved larvae, these fed larvae showed high sensitivity to 0.25 mM H₂O₂ (Figure ID) and it made little difference whether fed L1s were mutant for daf-16 or not (Figure 1E). These results demonstrate that daf-16-dependent protection of larvae from H₂O₂ depends on environmental conditions and is consistent with the hypothesis that DAF-16 is only activated when its transcriptional targets are needed for survival.

Mutation of the AGE-1 PI3K/Akt pathway affects daf-16 activity in dauer formation and lifespan. To test whether mutation of this pathway regulates the ability of daf-16 to protect starved L1s against H₂O₂, we quantified the resistance of mutants in the pathway using the assay described above. In order to assess how conditions and genotype affect H₂O₂ resistance, we used H₂O₂ concentrations that resulted in developmental arrest of a quantifiable proportion of the test population. For example, after treatment with 0.5 mM H₂O₂ approximately 30% of wild type larvae were permanently arrested at the L2/L3 stage, 20% were unaffected (Figure 1A, F and data not shown) and approximately 50% were temporarily delayed in development because they reached adulthood approximately one day later than did untreated controls.

Under these conditions, L1s with age-1(hx546) mutation showed increased resistance to H₂O₂ (Figure 1F). We next tested a gain-of-function mutation in the downstream kinase akt-1 22 and a reduction-of-function mutation in daf-18 23, the PIP₃ 3-phosphatase PTEN. These mutations activate the AGE-1 pathway and, in comparison to the wild type, they caused starved L1s to be more sensitive to H₂O₂ (Figure 1F). We also used a lower concentration of H₂O₂ (0.25 mM – which had no effect on the wild type) to differentiate these hypersensitive mutants further (Figure 1G). Thus, mutation of the AGE-1 pathway affects the ability of DAF-16 to protect worms from H₂O₂.

Although useful for identifying genes involved in a process, genetic analysis such as described above does not provide information on where, when or how genes are acting. To try to answer some of these questions, we investigated the subcellular localisation of DAF-16 because control of the presence of FOXO in the nucleus, where it acts as a transcription factor, is thought to be a major mechanism of its regulation. DAF-16 has been shown to translocate to the cell nucleus in response to heat stress, certain oxidative stresses and starvation 424. To investigate how starvation and oxidative stress influence DAF-16 activity in our system, we examined DAF-16 localisation in L1 larvae, using a strain previously characterised in a study by Lin et al. in 2001 13. This strain contained the transgene muIs71, an integrated DAF-16a::GFP construct that rescues daf-16 activity. In L1s hatched in the presence of food, we found that, as previously reported, DAF-16a::GFP was located in both the nucleus and the cytoplasm 13 (Figure 2A). However, in over 70% of L1s hatched in liquid culture and starved for one day, DAF-16a::GFP was localised solely to the nucleus (Figure 2B, E). This result is consistent with the idea that localisation of DAF-16 in the nucleus is the mechanism by which starvation confers oxidative stress resistance.

However, in L1s starved for two days or more, we found that DAF-16a::GFP translocates back to the cytoplasm and was distinctly excluded from the nucleus in over 70% of larvae (Figure 2C and 2E). This translocation of the DAF-16 transcription factor to the cytoplasm upon long-term starvation might be expected to lead to a decrease in the transcription of target genes and therefore an increased sensitivity to H₂O₂-induced arrest. However, when we tested this hypothesis using the H₂O₂ resistance assay described above, we found that after three days of starvation larvae displayed levels of resistance similar to if not greater than those of larvae starved for one day (Figure 3). This resistance was still dependent on daf-16 activity because daf-16 mutants were still hypersensitive to H₂O₂ after three days (Figure 3). These results suggest that even though DAF-16 translocates to the cytoplasm upon prolonged starvation, sufficient daf-16 dependent transcription and translation of somatic maintenance genes has already occurred to provide long-term stress resistance. Thus, one possibility is that translocation to the cytoplasm after long term starvation prevents excess expenditure of resources on somatic maintenance processes.

A likely mechanism for the translocation of DAF-16 from the nucleus to the cytoplasm is the activation of the AGE-1 pathway. To test this possibility, we examined DAF-16a::GFP in starved L1s with the age-1(hx546) mutation. In these larvae DAF-16a::GFP remained in the nucleus, both on day one and during subsequent days of starvation (Figure 2D, E). Continuous localisation of DAF-16 to the nucleus in the age-1(hx546) mutants may increase allocation of resources to stress resistance. This increased expenditure might explain why this mutant is unable to compete with the wild type during cycles of feeding and starvation 14. The age-1 dependent DAF-16 translocation we have discovered also suggests that long-term starvation activates AGE-1, which is a surprising conclusion considering current models proposing that Class IA PI3Ks are downstream targets of insulin-induced signalling in response to feeding. Recently a starvation-dependent, PI3K-dependent gene target has been identified in a human cancer cell line 25. One possibility is that long-term starvation relieves a nutrition-induced feedback mechanism that downregulates PI3K signalling in mammalian cells 26.

age-1-dependent DAF-16 translocation in response to prolonged starvation might be a consequence of endocrine stimulation or might occur through the upregulation of a cellular signal. One candidate for such a signal is H₂O₂ itself, which has been shown to act as an intracellular second messenger. This molecule can activate the PI3K/AKT pathway to cause FOXO3a to translocate to the cytoplasm of mammalian cells 27. To explore whether this response was conserved in C. elegans, we treated larvae starved for one day with increasing concentrations of H₂O₂. We found that H₂O₂, like long-term starvation, caused DAF-16a::GFP to translocate from the nucleus to the cytoplasm (Figure 4A, B). In a previous study, oxidative stress in the form of the superoxide producer juglone had been shown to cause DAF-16 to translocate to nucleus in fed L2 larvae 4. We found that, in L1s starved for one day, neither juglone nor paraquat, another superoxide producer, had any significant effect on DAF-16a::GFP localisation (data not shown), possibly because DAF-16 is already mainly nuclear in these larvae. Thus, H₂O₂ has a different effect from superoxide producers on DAF-16 localisation.

Importantly, H₂O₂-induced translocation of DAF-16a::GFP was ablated in the age-1(hx-546) mutant (Figure 4B), suggesting that AGE-1 mediates DAF-16 translocation to the cytoplasm upon H₂O₂ treatment. To test the involvement of AGE-1 further, we developed techniques to determine whether phosphatidylinositol (3,4,5) trisphosphate (PIP₃), the lipid product of AGE-1, is actually produced in C. elegans in response to H₂O₂. We found that, as in mammals, H₂O₂ did indeed stimulate PIP₃ production in starved C. elegans L1s (Figure 5A). No PIP₃ was produced in age-1(mg44) null mutants that were made viable by the daf-16(m27) loss-of-function mutation, demonstrating that PIP₃ generation requires the AGE-1 Class IA PI3K. Next, we tested the long-lived age-1(hx546) mutant for PIP₃ production. Despite the uncertainty about the molecular nature of this mutant, we found that no detectable PIP₃ was produced in age-1(hx546) larvae after H₂O₂ stimulation (Figure 5B). Thus, in starved L1s, the hx546 mutation blocks the ability of the AGE-1 Class IA PI3K to produce PIP₃ in response to H₂O₂.

Although it is technically difficult to test whether H₂O₂ is the endogenous signal that causes DAF-16 translocation to the cytoplasm after prolonged starvation, there have been several studies that make it a good candidate. Reactive oxygen species have been shown to increase upon serum or glucose starvation 2728. H₂O₂ can stimulate receptor tyrosine kinase signalling through its inhibitory effect on receptor tyrosine phosphatases. In addition, H₂O₂ can inhibit PTEN activity 2930, stimulate the association of mammalian Class IA PI3Ks with Ras 31 and activate Type I PIP kinase dependent PtdIns(3,4)P₂-5-kinase activity 32. All the above mechanisms could lead to H₂O₂-induced PIP₃ synthesis. Alternatively there could be a direct upregulation of IIS signalling in response to DAF-16 activation. A number of potential insulin-like ligands are upregulated in daf-2 mutants 79 and a recent study demonstrated that the Drosophila insulin receptor homologue was dramatically upregulated in flies overexpressing activated dFOXO 33.

As worms age, their pharyngeal pumping rate slowly reduces to zero 34. We wondered whether this reduction of food intake causes long-term starvation and therefore investigated whether DAF-16 translocates to the cytoplasm of cells in old adult worms. The data published by Henderson and Johnson using a DAF-16::GFP fusion transgene suggest that in adults under standard laboratory conditions without starvation, the nuclear localisation of DAF-16 is not sufficient to discern nuclei. However, in the strain used in their study, DAF-16::GFP is expressed at noticeably higher levels than the strains used in this study, and they recently been reported that their transgene contains a missense mutation in the DAF-16 sequence 35. Consistent with the initial description of the transgenes used in our study 13, we found that in reproductively active three day old adults raised at 20°C, DAF-16 was localised to both the cytoplasm and nuclei of cells with nuclei visible as distinct spots (Figure 6A).

However, as worms aged we found that nuclei became progressively less distinct, indicating a translocation to the cytoplasm (Figure 6C). In three day old worms carrying the age-1(hx546) mutation, DAF-16 localisation was similar to that seen in wild type worms of the same age, although nuclei were slightly more distinct than seen in the wild type (Figure 6B). However, in older age-1(hx546) mutants, DAF-16 became more nuclear and there was no translocation to the cytoplasm. Even in very old worms that were close to death as assessed by age, movement and appearance, cell nuclei were visible as distinct spots of GFP (Figure 6D, Table 1). Thus, there is an age-1-dependent translocation of DAF-16 to the cytoplasm in old worms. The lack of this age-dependent translocation in age-1(hx546) mutants may result in increased DAF-16 activity and hence explain the increased resistance to H₂O₂ and antioxidant activities observed in these mutants as they age 1718.

Mice lacking the adaptor protein p66Shc, which binds to receptor tyrosine kinases, have an extended lifespan 36. Interestingly, in cells derived from these mice, there is a defect in H₂O₂-induced translocation of FOXO3a to the cytoplasm 27. We have demonstrated a physiological whole-organism situation in which a FOXO transcription factor translocates to the cytoplasm and shown that this response is defective in worms carrying the age-1(hx546) mutation, which has a strong effect on DAF-16 localisation only after prolonged starvation or ageing. Like this mutant, the p66Shc mutant mouse has increased stress resistance without obvious deleterious effects 36. We propose a model whereby in these mutants, an evolutionarily-conserved response to prolonged starvation is traded for increased longevity.

Conclusion

DAF-16/FOXO is nuclear-localised and active in the starved L1 diapause to enhance stress resistance but translocates back to the cytoplasm after long-term starvation in a manner dependent on age-1/PI3K (Figure 7). We propose this response is a negative feedback mechanism that prevents overexpenditure on resources during starvation. A similar effect is observed in adult worms. Disruption of this translocation may explain how the increased longevity of the age-1(hx546) mutant has no detrimental effects apart from the cost of competiveness during cycles of starvation and feeding 14.

PI3K – phosphoinositide 3-kinase, PIP₃ – phosphatidylinositol (3,4,5) trisphosphate, WHB – worm HEPES buffer, PtdIns -phosphatidylinositol

DW conceived of, designed and performed all the experiments and drafted the manuscript. JRH helped develop methods to isolate and measure PIP₃ in C. elegans larvae. DG helped draft the manuscript. ND helped in the conception and design of the study. All authors have read and approved the manuscript.