Cell-Nonautonomous Effects of dFOXO/DAF-16 in Aging

Summary Drosophila melanogaster and Caenorhabditis elegans each carry a single representative of the Forkhead box O (FoxO) family of transcription factors, dFOXO and DAF-16, respectively. Both are required for lifespan extension by reduced insulin/Igf signaling, and their activation in key tissues can extend lifespan. Aging of these tissues may limit lifespan. Alternatively, FoxOs may promote longevity cell nonautonomously by signaling to themselves (FoxO to FoxO) or other factors (FoxO to other) in distal tissues. Here, we show that activation of dFOXO and DAF-16 in the gut/fat body does not require dfoxo/daf-16 elsewhere to extend lifespan. Rather, in Drosophila, activation of dFOXO in the gut/fat body or in neuroendocrine cells acts on other organs to promote healthy aging by signaling to other, as-yet-unidentified factors. Whereas FoxO-to-FoxO signaling appears to be required for metabolic homeostasis, our results pinpoint FoxO-to-other signaling as an important mechanism through which localized FoxO activity ameliorates aging.


INTRODUCTION
Forkhead box O (FoxO) transcription factors (TFs) are involved in a plethora of cellular processes to regulate whole-organism physiology and are major determinants of animal lifespan (Partridge and Brü ning, 2008;Salih and Brunet, 2008). Activation of FoxO-family TFs mediates the lifespan-extending effects of dampened insulin/insulin-like growth factor-like signaling (IIS) in both worms and flies (Kenyon et al., 1993;Slack et al., 2011;Yamamoto and Tatar, 2011). This evolutionary conservation appears to extend to humans, because certain genetic variants of Foxo3A are robustly associated with human longevity (Flachsbart et al., 2009;Kuningas et al., 2007;Willcox et al., 2008). Indeed, Forkhead-like TFs can even extend lifespan in a single-celled eukaryote, budding yeast (Postnikoff et al., 2012).
In Drosophila melanogaster, tissue-restricted activation of Drosophila foxo (dfoxo) is sufficient to extend lifespan (Demontis and Perrimon, 2010;Giannakou et al., 2004;Hwangbo et al., 2004). Such an increase in dfoxo activity confined to key tissues could promote whole-organism survival in two mutually compatible ways: cell autonomously and cell nonautonomously. The lifespan of the animal could be limited by pathology in a particular organ, so that cell-autonomous action of dfoxo in that organ alone could promote longevity (Rera et al., 2013). In addition, healthy aging may involve the coordinated action of multiple organ systems, with dfoxo in one organ altering whole-organism physiology through systemic changes (Demontis and Perrimon, 2010;Hwangbo et al., 2004;Rera et al., 2013). For example, adult-onset induction of dfoxo in the midgut and abdominal fat body (equivalent to mammalian liver and adipose) activates the transcription of Drosophila insulin-like peptide (dilp) 6 in the fat body, whereas in muscle dfoxo represses the activin ligand dowdle, and these endocrine signals have a distal effect on the median neurosecretory cells (mNSCs) in the brain, resulting in lowered DILP2 peptide in circulation (Bai et al., 2012(Bai et al., , 2013. Importantly, upregulation of dilp6 is required for the beneficial effect of dfoxo on lifespan (Bai et al., 2012). However, whether this requires dfoxo in tissues other than the ones producing the DILP6 signal remains unexamined.
The single Caenorhabditis elegans FoxO ortholog, DAF-16, can act both cell autonomously and cell nonautonomously to regulate gene expression (Libina et al., 2003;Murphy et al., 2007;Qi et al., 2012;Zhang et al., 2013). DAF-16 activity in one tissue can induce DAF-16 activity in another in a process of tissue entrainment mediated by altered expression of an insulin-like peptide (Murphy et al., 2007), which is highly reminiscent of the situation in the fly. For this reason, it has been widely believed that the fruit fly's dfoxo acts from specific cells to activate dFOXO in the whole animal in an instance of dfoxo-to-dfoxo signaling (Bai et al., 2012(Bai et al., , 2013Demontis and Perrimon, 2010;Hwangbo et al., 2004). However, the relevance of this tissue entrainment for Drosophila lifespan has not been experimentally tested. Indeed, there is a growing awareness that FoxOs in one tissue can also signal to other factors elsewhere, i.e., FoxO-to-other signaling. In the worm, DAF-16 activity in one tissue can elicit daf-16-independent responses in the receiving tissues (Qi et al., 2012;Zhang et al., 2013). The existence and relevance of dfoxo-to-other intertissue signaling is unexplored in Drosophila.
Here, we establish the relevance to aging of the cell-nonautonomous effects of dfoxo, differentiating between dfoxo-to-dfoxo and dfoxo-to-other signaling in adult Drosophila. We find that dfoxo-to-dfoxo signaling does not affect aging and confirm that the same is true of the worm daf-16. On the other hand, dfoxo in the gut and fat body can promote health of the neuromuscular system, possibly via transcriptional regulation of a secreted neuropeptide-like molecule, and dfoxo in mNSCs can extend lifespan. Both effects are independent of dfoxo's presence in other tissues, demonstrating the relevance of dfoxo-to-other signaling in Drosophila aging. At the same time, dfoxo-to-dfoxo signaling is required for the metabolic effects of localized dfoxo induction, showing that distinct physiological effects of tissue-restricted dfoxo activation are mediated by different signaling routes.

RESULTS
dfoxo-to-dfoxo Signaling in Drosophila Is Dispensable for Extension of Lifespan by Gut/Fat Body or mNSC dfoxo To examine whether activation of dFOXO in other tissues contributes to the lifespan-extending effects of induction of dfoxo in the adult gut and fat body, we generated strains where the tissue-restricted induction of dfoxo could be triggered by the RU486 inducer in either an otherwise wild-type or a dfoxo-null background (S 1 106>dfoxo or dfoxoD/D S 1 106>dfoxo). We used females, where the effects of dfoxo activation on aging are clearly observed (Giannakou et al., 2004). Because the lifespan effects of ectopic dfoxo expression can be conditional on the nutritional status of the animal (Bai et al., 2012;Min et al., 2008), we used a food with the optimal amount of dietary yeast (10% weight/volume) for lifespan under our laboratory conditions (Bass et al., 2007) and where expression of dfoxo targeted to adult gut and fat body robustly extends lifespan (Giannakou et al., 2008). Importantly, on this food, lifespan is maximized so that the effects of dfoxo can be studied as additional to the beneficial effects of the diet.
We found no detectable expression of dFOXO protein or of dfoxo transcript in the dfoxoD/D S 1 106>dfoxo females in the absence of the inducer ( Figures 1A and 1B). Feeding RU486 for 5 days resulted in equivalent increases in dfoxo transcript in S 1 106>dfoxo and dfoxoD/D S 1 106>dfoxo females ( Figure 1B; see Table 1 for detailed statistical analysis). The S 1 106 driver has been thoroughly characterized and, in the female fly, only drives expression in the gut and fat body (Poirier et al., 2008). To ensure the flies are experiencing the same nutritional conditions, we examined their feeding behavior with the proboscisextension assay (Wong et al., 2009) and found no significant differences ( Figure S1A).
We examined the effect on lifespan resulting from the presence of the inducer in the S 1 106>dfoxo and dfoxoD/D S 1 106>dfoxo lines in two sequential, independent, experimental trials ( Figure 1C), recording deaths of over 1,000 flies in total. The presence of RU486 from day 2 of adulthood extended the median lifespan of S 1 106>dfoxo females on average by 6% (log-rank test p < 0.05 for each trial; Figure 1C). The magnitude of the effect was less than previously reported (Giannakou et al., 2004) but is consistent with more recent work in our laboratory (Giannakou et al., 2008) and with six other independent trials performed in the course of the last 4 years (2008-2012, average median extension 5%; Figure S1B). The lifespan of dfoxo-null flies was also extended by a similar percentage (average 10%, log-rank test p < 0.05 for each trial; Figure 1C). Thus, the presence of dfoxo in the rest of the body is not required for the lifespan-extending effects of its induction in the gut/fat body.
Flies lacking dfoxo have short lifespans ( Figure 1C), possibly due to developmental effects of the mutation (Slack et al., 2011), complicating the direct comparison between effects of RU486 in the two lines. Cox proportional hazards (CPH) is a survival analysis that allows for the significance of several covariates and their interactions to be examined. To establish whether there was any statistically significant difference in the response of dfoxo-null and wild-type flies to RU486, we combined the two experimental trials and analyzed the survival data using a mixed-effects Cox proportional hazards (MECPH) model (Table 1). Both RU486 (30% reduction in risk of death, p = 2 3 10 À4 ) and the presence of genomic dfoxo (p < 10 À15 ) had a significant effect on lifespan, but their interaction did not (p = 0.95). The absence of a significant interaction confirms that the effect of RU486 did not differ between the lines and hence that the presence of dfoxo elsewhere in the body does not affect the extension of lifespan by induction of dfoxo in the gut and fat body. Thus, tissue entrainment through dfoxo-to-dfoxo signaling is not required for longevity.
This result indicated that either dfoxo acts cell autonomously to extend lifespan or that dfoxo in one tissue activates dfoxo-independent longevity-assurance mechanisms in other tissues. The latter would occur through dFOXO-to-other signaling, as has been observed for DAF-16 (Qi et al., 2012;Zhang et al., 2013). To further test for dfoxo-to-other signaling, we manipulated the levels of dfoxo in cells whose prominent function is in adult endocrine signaling. mNSCs in the adult brain play an important role in aging by producing DILP2, DILP3, and DILP5 (Broughton et al., 2005) and possibly other endocrine signals. Expressing dfoxo specifically in the mNSC, using a dilp2-GAL4 driver, extended the lifespan of female flies in both wild-type and dfoxo nulls (p < 0.01 to either control in both backgrounds; Figure 1D; Table 1). CPH analysis found significant effects of the genomic dfoxo (p < 10 À15 ) and its induction in dilp2GAL4>dfoxo flies (50% reduction in risk of death, p = 2.2 3 10 À5 ) on lifespan but no evidence for a significant interaction between them (p = 0.33; Table 1). This confirmed that the effect on lifespan is independent of dfoxo in tissues other than the mNSCs. This longevity phenotype must represent a gain of function in the mNSC, because the ablation of mNSCs, representing a loss of function in these cells, requires dfoxo to extend lifespan (Slack et al., 2011). Indeed, we observed no significant changes in the mRNA levels of dilp2, dilp3, and dilp5 upon induction of dfoxo in mNSCs (Figures S1C and S1D). Furthermore, we found no changes in the mRNA levels of any dilps detectable in whole adults (dilp2 through to dilp7), including dilp6 ( Figure S1D), or their binding partner and regulator, Imp-L2 (Alic et al., 2011b) ( Figure S1E), upon activation of dfoxo in the mNSCs, confirming that dilp2GAL4 > dfoxo flies are not experiencing any alterations in systemic IIS activity. Because the principal role of these cells is in endocrine signaling, the physiological effects of dfoxo activation in the mNSCs are most likely to be mediated by dfoxo-toother signaling.
Gut/Fat Body dfoxo Acts at a Distance Independently of dfoxo in Target Tissues To further investigate the role of dfoxo-to-other signaling in fly physiology, we examined the beneficial effects of gut/fat body induction of dfoxo on the neuromuscular system, an organ system distal to the site of dfoxo activation in our model. The ability of flies to climb a vertical surface is a suitable physiological measure of the performance of this organ system and is susceptible to aging (Cook-Wiens and Grotewiel, 2002). We scored the number of low, medium, and high climbers in three cohorts of $15 individuals of S 1 106>dfoxo or dfoxoD/D S 1 106>dfoxo genotype in the presence or absence of RU486 over $10 weeks.  Table 1 for statistical analysis of data in (B)-(E). Where used, box plots indicate median, first and third quartile, data range, and outliers.See also Figure S1.
Induction of dfoxo expression in the gut and fat body enhanced the climbing ability of female flies throughout their lifespan, observed as an increase in the proportion of high, or combined medium and high, climbers ( Figure 1E). This enhancement could be seen in both the wild-type and dfoxo-null backgrounds, revealing that it is independent of dfoxo in other tissues. Indeed, statistical analysis (mixed-effects ordinal logistic model, Table 1) confirmed that the effect of RU486 (p = 1.8 3 10 À3 ) and dfoxo (p = 5.4 3 10 À15 ) were both significant but that their interaction was not (p = 0.12). Hence, local action of dfoxo in the gut and fat body has a beneficial effect on the performance of a distal organ system. This could occur through systemic effects of healthy gut and fat body or through specific signaling events. In the latter case, its independence from dfoxo in the distal cells is again consistent with dfoxo-to-other signaling.

Gut/Fat Body dfoxo Regulates Expression of Nplp4
In order to trigger dfoxo-to-other signaling, the gut/fat body dfoxo may regulate the expression of a secreted factor other than dilp6. To identify such a factor, we determined the wholefly, genome-wide, transcriptional changes induced by RU486 in the S 1 106>dfoxo flies (Table S1; Figure 2A). We found that, besides the documented changes in dilp6 (Bai et al., 2012), induction of the gut/fat body dfoxo altered the mRNA levels of another gene encoding a signal peptide targeting its protein product for secretion, neuropeptide-like precursor 4 (Nplp4). The mature product of this gene is a YSY peptide of previously unknown function (Nä ssel and Winther, 2010). Quantitative PCR confirmed that activation of dfoxo led to repression of this gene ( Figure 2B). Hence, Nplp4 is a candidate for a secreted factor mediating dfoxo-to-other signaling. Interestingly, this gene was repressed in both heads and bodies of S 1 106>dfoxo females fed RU486 (p = 0.048 for RU486, p = 0.19 for body part:RU486 interaction; Figure 2B; Table 1), whereas, as expected, the induction of dfoxo itself was confined to the body, (p = 0.053 for body part:RU486 interaction; Figure 2C; Table 1), indicating Nplp4 responds to dFOXO both locally and distally.
Importance of dfoxo-to-dfoxo Signaling to Drosophila Metabolism Although dfoxo-to-dfoxo signaling is not required for lifespan extension, it may be required for other physiological changes in response to the activation of dFOXO in gut and fat body. To query the existence of these other physiological effects, we examined whether there are transcriptional changes in response to RU486 in S 1 106>dfoxo flies that do not occur in dfoxoD/D S 1 106>dfoxo females. We reasoned that the genes and processes that respond to RU486 in S 1 106>dfoxo flies but fail to do so in the dfoxoD/D S 1 106>dfoxo females may be regulated through dfoxo-to-dfoxo signaling.
Among the genes regulated by RU486 in S 1 106>dfoxo females, we identified all those for which the RU486-induced transcriptional change was altered by mutation of dfoxo by finding the genes whose transcript levels show a significant interaction between the presence of genomic dfoxo and its induction by RU486 in the relevant linear model ( Figure 2A; Table S1). The magnitude of fold-change for these genes was reduced on average in dfoxoD/D S 1 106>dfoxo compared to S 1 106>dfoxo females ( Figure 2A), indicating they require the presence of genomic dfoxo for correct expression. We confirmed the significance of this effect using a linear model (p = 1 3 10 À7 ; Figure 2A and the associated caption). Note that Nplp4 was equally repressed in dfoxoD/D S 1 106>dfoxo and S 1 106>dfoxo females (Table S1).
Examination of the Gene Ontology categories enriched in this group of genes revealed ''proteolysis'' as overrepresented (p = 3.1 3 10 À7 ; Table S1), hinting that protein metabolism may be regulated through a dfoxo-to-dfoxo signal. Pursuing this lead, we found that RU486 feeding triggered a small (12%) but significant reduction in total protein content of S 1 106>dfoxo females and that this effect was blocked by deletion of dfoxo ( Figure 2D; p = 3.1 3 10 À3 for RU486 by genotype interaction; Table 1). Similar significant changes were not observed in total triglyceride, total trehalose, or total glycogen content (Figure S2A). However, due to assay variability, we cannot discount possible subtle changes in these metabolites. On the other hand, total body mass followed closely the protein content ( Figure 2E; Table 1).
Surprisingly, both deletion of dfoxo and its induction in the gut and fat body reduced total protein content and fly weight. The two manipulations may act in different ways. The small size of dfoxo nulls is due to the developmental effects of the mutation (Slack et al., 2011) and, together with their reduced fecundity (Slack et al., 2011) ( Figure S2B), could explain the lowered body weight and protein content. On the other hand, S 1 106>dfoxo was induced in adulthood and had no effect on fecundity in either wild-type or dfoxo-null females (Figure S2B), but it had an effect on body weight and protein content. Hence, the latter two metabolic phenotypes of dFOXO activation in gut/fat body may depend on dfoxo-to-dfoxo signaling. However, because dfoxo was absent in all tissues throughout development, we cannot exclude the possibility that the inability of dfoxo nulls to respond to RU486, for these traits, is due to developmentally altered gut/fat body function. Note that the expression pattern of the proteolysis genes, which initially led us to this phenotype, cannot mechanistically explain the loss of protein in S 1 106>dfoxo females upon RU486 feeding, because these genes are repressed in this condition ( Figure S2C), but may rather be part of a homeostatic mechanism. Nevertheless, the results strongly indicate the effects on lifespan and metabolism of tissue-restricted activation of dFOXO can be separated by their requirement for dfoxo-to-other or dfoxo-to-dfoxo signaling.
Similar to Qi and coworkers, we used FUdR to reveal the lifespan extension caused by overexpression of daf-16 from the intestinal-specific ges-1 promoter in otherwise wild-type worms and asked whether this effect required the presence of daf-16 in other tissues ( Figure 3A). We found that intestinal activation of DAF-16 can extend lifespan of both wild-type and daf-16-deficient (daf-16(mu86)) worms fed HT115 bacteria (log-rank p < 0.05 in three out of five and three out of three assays, respectively; Figures 3A and S3A). We confirmed these findings using a second, independently derived transgene ( Figure S3B).
Upon further examination, we found that the ability of the intestinal daf-16 to extend wild-type lifespan was conditional on the food source and had a small but opposite effect when worms were fed on OP50 bacteria ( Figure S3C). This is similar to the effect of gut/fat body expression of dfoxo on Drosophila lifespan, which can depend on available nutrition (Bai et al., 2012;Min et al., 2008). Importantly, however, the ability of the intestinal daf-16 to extend lifespan in the absence of daf-16 elsewhere was observed under all conditions, including the worms fed OP50 (Figures 3 and S3A-S3C). MECPH analysis of the combined data obtained with one of the transgenes on HT115 bacteria (Figures 3 and S3A) confirmed that both the effects of daf-16 presence in the whole worm and its intestinal induction were significant (p < 10 À15 for both) and revealed a significant interaction of the two main effects (p = 2 3 10 À15 ; Table 1). Thus, intestinal daf-16 extended the lifespan of the mutant more than that of the wild-type worms (Figure 3), confirming that, as in the fly (Figure 1C) and during IIS dampening in the worm (Libina et al., 2003), tissue entrainment through daf-16-to-daf-16 signaling is not required for lifespan extension and, indeed, could even have the opposite effect.
Prompted by the findings in the fly ( Figure 2D), we also examined if the induction of intestinal daf-16 in worms had an effect on their total protein content. We found that the intestinal daf-16 reduced whole-worm protein content (p < 10 À4 ; Figure 3B; Table 1) on HTT15 bacteria. In contrast to the fly, we found no evidence that this reduction is prevented by the absence of daf-16 in other tissues, and, in fact, found that mutation of daf-16 increases the overall protein content ( Figure 3B; Table 1). We obtained similar results with the second transgene (Figure S3D). Hence, this phenotype is mediated either cell autonomously by daf-16 in the intestine or through daf-16-to-other signaling. Thus, whereas the physiological effect appears conserved between the fly and worm, the way it is mediated differs.
It is also of note that, similar to the lifespan effect of intestinal daf-16 in an otherwise wild-type worm, we found this modulation of protein content conditional on the bacterial food source and neither the transgenes nor mutation of daf-16 had any significant effect when worms were fed OP50 bacteria (data not shown). For both lifespan and protein content, the alteration of phenotype between OP50 and HTT15 is reminiscent of the lifespan effects of certain sensory mutants in C. elegans (Maier et al., 2010) and suggests that intestinal DAF-16 plays a role in food perception.

DISCUSSION
In the fly, tissue-restricted dFOXO triggers endocrine factors to cause a drop in overall, systemic, IIS activity (Bai et al., 2012(Bai et al., , 2013Demontis and Perrimon, 2010;Hwangbo et al., 2004). Because insulin signals repress the activity of FoxOs (Brunet et al., 1999), this will result in body-wide activation of dFOXO (tissue entrainment), including further activation of dFOXO in the specific tissue (positive feedback). Our results show that the tissue entrainment is not required for the beneficial effects of dfoxo on lifespan or on healthspan. The regulation of systemic IIS by local dfoxo can still be relevant to lifespan as part of a positive feedback loop. For example, the upregulation of dilp6 by dFOXO in the fat body triggers a reduction in global IIS activity, and this, in turn, could be affecting lifespan by fine-tuning the activity of dFOXO in the fat body itself. daf-16 À2.3 3 10 À2 7.9 3 10 À3 <10 À4 intestinal daf-16 À3.7 3 10 À2 7.9 3 10 À3 6.2 3 10 À3 daf-16:intestinal daf-16 À9.3 3 10 À3 7.9 3 10 À3 0.24 a In all models, the effect of presence of dfoxo (daf-16) or its induction (overexpression) is examined; dilp2-GAL4 was used as reference for dilp2GAL4>dfoxo and UAS-dfoxo; ''body'' was used as reference for body versus head comparisons; '':'' indicates interaction term. b For mixed-effect linear models and linear models, the coefficient estimates either have no units because they are derived from the dfoxo/Act or Nplp4/ Act transcript ratios (for Figures 1B, 2B, and 2C) or are given in mg (for Figures 2D and 2E) or mg ( Figure 3B). For MECPH and CPH models, the coefficient estimate is the natural log of the hazard ratio, where a negative value indicates a beneficial effect on survival. For mixed-effect ordinal logistic, these are log-transformed odds of climbing high, where a negative value indicates a reduction in climbing ability.
Under certain experimental conditions, the lifespan effects of ectopic dfoxo expression can be conditional on the nutrients available to the animal (Bai et al., 2012;Min et al., 2008). Hence, tissue entrainment may also have conditional relevance. In addition, our results indicate that dfoxo-to-dfoxo signaling is required for the metabolic effects of localized dfoxo induction, namely a drop in protein content and fly weight, and further examination may reveal roles for dfoxo-to-dfoxo signals in yet other aspects of physiology.
Both DAF-16 in C. elegans and dFOXO in Drosophila can extend lifespan from the gut/fat body without being present in other tissues. The gut and/or fat body may represent the organs most vulnerable to aging, so that DAF-16/dFOXO directly prevents the otherwise lethal age-related pathologies in these organs. This, in turn, could have indirect benefits for other organs. Indeed, there is some evidence that the health of the Drosophila gut limits lifespan (Rera et al., 2013). Furthermore, DAF-16/dFOXO could regulate key metabolic genes in these tissues, such as lipases, fatty acid catabolic genes, and others, effecting a shift in energy utilization toward prolonged health and survival. However, dFOXO activity in other tissues, such as the muscle (Demontis and Perrimon, 2010) or the mNSC (Figure 1D), can also extend lifespan. Although it is conceivable that multiple tissues independently and simultaneously limit lifespan, in at least some of these interventions, the relevant effects must be cell nonautonomous.
DAF-16 in one tissue is known to trigger DAF-16-independent responses in other tissues (Qi et al., 2012;Zhang et al., 2013). In one case, this is mediated by induction of a transcriptional mediator, mdt-15, and is required in part for the beneficial effects of the intestinal activation of DAF-16 by daf-2(À) on whole-organism aging (Zhang et al., 2013). Our results indicate that dFOXO Red indicates the genes whose response to RU486 is significantly altered by genotype (significant genotype:RU486 interaction in the linear model). The red line is the regression line for these genes (slope = 0.49) and the black line is for the others (slope = 0.77). The significant difference in slope (p = 1 3 10 À7 ) indicates the genes marked in red are overall less responsive to RU486 in dfoxoD/D S 1 106>dfoxo than in S 1 106>dfoxo female flies. Gene lists are given in Table S1. (B) Nplp4 transcript levels (relative to Act and with body -RU486 set to 1) in bodies or heads of S 1 106>dfoxo female flies fed or not RU486. (C) dfoxo transcript levels (relative to Act and with body -RU486 set to 1) in bodies or heads of S 1 106>dfoxo female flies fed or not RU486. (D) Protein content of individual S 1 106>dfoxo or dfoxoD/D S 1 106>dfoxo female flies after 5-day feeding with RU486 or not. (E) Individual fly weight for the same conditions. Note the same color code is used in (B)-(E) and is given in (B). In (D) and (E), asterisk indicates significant difference at p < 0.05 by post hoc, pair-wise t test between À and + RU486 conditions. See Table 1 for statistical analysis of data in (B)-(E). See Figure S2 for further data.
may also initiate dfoxo-independent processes in the receiving tissues that counteract whole-organism aging. This is the most likely mechanism whereby its activity in the Drosophila mNSC can extend lifespan, and a similar mechanism may underlie the health benefits observed when it is induced in the gut and fat body. The search for the factors that mediate this effect of dFOXO at a distance is now of interest, and we identified Nplp4 as a candidate. The evolutionary persistence of this FoxO-to-other signaling between the fly and the worm strongly suggests that its relevance may extend to mammals.

Fly Husbandry and Experiments
All transgenes and the dfoxo mutant were backcrossed at least six times into the wild-type outbred Dahomey population carrying the w 1118 mutation and cured of Wolbachia infection and frequently outcrossed back into the wildtype population. The Dahomey stock was collected in 1970 in Dahomey (now Benin) and has been kept in population cages maintaining its lifespan and fecundity at levels similar to freshly caught stocks. The lines were maintained, and all experiments performed, at 25 C with 60% humidity and 12 hr:12 hr light:dark cycle on sugar-yeast-agar (1SYA) food (Bass et al., 2007). Experimental flies developed at standardized densities and oncemated females were sorted on day 2 of adulthood onto food containing 200 mM RU486 (Sigma) or control food as required (15 per vial for climbing assays, five for feeding, and ten for all others). Flies were harvested on day 7 for weight and metabolite measurements and protein and RNA analysis. Sample preparation and hybridizations to Dros2 Affymetrix arrays were performed and data analyzed with LIMMA, essentially as described elsewhere (Alic et al., 2011a). For further details, see Supplemental Experimental Procedures. Gene lists are given in Table S1.

Worm Husbandry and Experiments
Worms were maintained at 20 C unless otherwise indicated. Prior to experiments, animals were maintained at the permissive temperature and grown for at least one generation in the presence of food to assure full viability. Lifespan assays were performed on HT115 bacteria carrying empty pL4440 vector, or OP50 bacteria, in the presence of 10 mM FUdR. Worms were placed on these plates at the L4 stage and scored as dead or alive every 2-3 days. For further details, see Supplemental Experimental Procedures.

Statistical Analysis
Analyses were performed in JMP (SAS) or R. Further details are given in Table 1 and Supplemental Experimental Procedures. To determine difference in slopes of the regression lines between the two gene sets in Figure 2A, the linear model was fitted with RU486-induced response in dfoxoD/D S 1 106>dfoxo as the dependent variable and the response in S 1 106>dfoxo (continuous) and gene set (categorical) as explanatory variables, testing for the significance of the interaction term.

ACCESSION NUMBERS
The ArrayExpress accession number for the array data reported in this paper is E-MTAB-1232.

ACKNOWLEDGMENTS
Some worm strains were provided by the Caenorhabditis Genetics Center, which is funded by National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440). We acknowledge funding by the Wellcome Trust (to D.G. and L.P.) and Max Planck (to L.P.). This work is dedicated to the memory of Pedro Martinez.  Figure S3A for further lifespan trials and Table 1 for statistical analysis. See also Figures S3B and S3D for the effect of an independent transgene. See Figure S3C for the lifespan effects on     List of genes differentially expressed (10% FRD) in S106>dfoxo females with RU486 treatment, with indication of the ones that show significant interaction between genotype (S106>dfoxo or dfoxo null S106>dfoxo) and RU486 (10% FDR).
Biological process categories over-represented in genes with significant interaction term in list above.
Experimental flies developed at standardised densities and once-mated females were sorted on day two of adulthood onto food containing 200 µM RU486 (Sigma) or control food as required (15 per vial for climbing assays, 5 for feeding, 10 for all others). Note that the food from the exact same cook was used for the -and + RU486. Lifespan measurements were performed essentially as described (Giannakou et al., 2004 against a suitable standard. For glycogen measurements, a single female was homogenised into 100 !l of PBS 0.1% TritonX-100, and the glucose released by incubation with amyloglucosidase in the samples and glycogen standards determined with Infinity Glucose Reagent (Thermo Scientific). Flies were weighted singly or in pairs, and the two highest/lowest measurements for each condition discarded.
dfoxo mRNA was deemed detectable when the qPCR signal was above that ! observed in the absence of reverse transcriptase. mRNA levels of dilp2 through to 7 and Imp-L2 were quantified as described (Alic et al., 2011b;Broughton et al., 2005;Gronke et al., 2010). Sample preparation and hybridisations to Dros2 Affymetrix arrays were performed as described (Alic et al., 2011a). The microarray data were analysed in R. They were summarised and normalised using RMA, differential expression was assessed using Linear Models and the empirical Bayes moderated t-statistic implemented in LIMMA (Bolstad et al., 2003;Irizarry et al., 2003a;Irizarry et al., 2003b) and FDR was controlled using the described procedure (Benjamini and Hochberg, 1995).
Genes differentially expressed on RU486 feeding in S 1 106>dfoxo females were determined (at FDR=10%), and amongst these genes the ones with significant interaction between genotype and RU486 (FDR=10%) were determined by a Linear Model (full factorial design) fitted to all the data (the four conditions). Gene Ontology Enrichment was determined with David EASE (Dennis et al., 2003).