Dietary Restriction Extends the Lifespan of Circadian Mutants tim and per

Dietary restriction (DR), or decreased consumption of nutrients such as protein, extends lifespan in multiple model organisms, including Drosophila (Fontana and Partridge, 2015). In a recent article in Cell Metabolism, Katewa et al. (2016) presented evidence indicating that arrhythmic Drosophila females lacking the circadian regulators period (per) or timeless (tim) exhibit truncated lifespan extension in response to DR (Katewa et al., 2016). In Drosophila, Per and Tim proteins are critical regulators of circadian rhythm, or daily oscillations in activity and physiological function. In contrast with Katewa et al. (2016), we found that per and tim mutant females respond to DR with lifespan extension equivalent to controls. We investigated the discrepancy between these results, which could arise from differences between fly stocks, specific experimental conditions, or global environmental variations.

We measured the lifespan of per⁰¹ and tim⁰¹ mutants relative to their genetic controls on four diets with varying concentrations of yeast extract (YE): 0.01% (low), 0.5% (DR), 3% (standard), and 10% (high). Katewa et al. (2016) used 0.5% and 5% YE for DR and standard food, respectively. We found that female per and tim null mutants and their controls responded similarly to DR (Figures S1A and S1B). Both we and Katewa et al. (2016) used identical mutant alleles obtained from the same source lab (Krishnan et al., 2008; Rakshit et al., 2013). We confirmed that our mutants retained their respective mutations by Sanger sequencing (Figure S1C). Thus, loss of either per or tim function did not inhibit the expected longevity response to DR.

To further investigate the response to DR of arrhythmic flies, we tested two other methods of disrupting circadian regulation. First, Oregon R female flies (control lab strain) were made arrhythmic by constant light treatment, which destabilizes Tim and Per proteins (Naidoo et al., 1999). Second, we examined arrhythmic mutants lacking the circadian-regulating neurotransmitter pdf (Lin et al., 2004). Whereas Katewa et al. (2016) showed that constant light treatment inhibited DR-mediated lifespan extension, we found that control flies in constant light or the standard light/dark cycle showed similar DR-mediated longevity (Figure S1D). Equivalent DR-mediated lifespan extension was also observed in pdf mutants and their controls (Figure S1E). These results are consistent with our per⁰¹ and tim⁰¹ mutant studies and collectively demonstrate that loss of circadian regulation does not inhibit DR-mediated lifespan extension.

It was possible that our circadian mutants, though they retained the primary mutations, did not exhibit the metabolic phenotypes reported by Katewa et al., who proposed that altered lipid metabolism inhibits their response to DR (Katewa et al., 2016). To test whether our per⁰¹ mutant females had altered energy stores relative to controls, we first tested their survival under starvation conditions. Consistent with published data (Allen et al., 2016; Katewa et al., 2016), per⁰¹ mutants starved more quickly than controls on both standard and restricted diets, suggesting that per⁰¹ mutants indeed have lower lipid stores than controls (Figure S1F). We then measured lipid storage levels using thin-layer chromatography (Ulgherait et al., 2014). In agreement with published data (Allen et al., 2016; Katewa et al., 2016; Seay and Thummel, 2011), per⁰¹ females exhibited lower levels of lipid storage than controls (Figure S1G). Further supporting altered lipid metabolism, we also found that during starvation per⁰¹ mutants lost lipid stores more rapidly than controls (Figure S1G). Thus, these data support published results and demonstrate that our per⁰¹ mutants have altered lipid metabolism similar to that seen by Katewa et al. (2016). However, because DR-mediated lifespan extension is equivalent between our mutants and their controls, our results suggest that altered lipid metabolism is not sufficient to inhibit DR-mediated lifespan extension.

We next tested whether differences in food protocols affect DR-mediated lifespan extension. Our protocol (designated NY) and that of Katewa et al. (2016) (designated CA) differed in sugar composition, concentration of yeast extract (YE) for standard diet, and the type of food used during development (see Experimental Procedures and Katewa et al., 2016). To assess the effects of these differences, we measured lifespan in parallel using either our food protocol (NY) or the published protocol in Katewa et al. (2016) (CA). In both cases, DR-mediated lifespan extension was similar or greater in circadian mutant females compared to their controls (Figures S1H–S1K). Thus, differences in food protocols do not appear to account for contrasting longevity results between labs.

To test for possible secondary site mutations or genetic drift between the different stocks, we obtained per⁰¹ and tim⁰¹ mutants and their controls directly from Katewa et al. (2016) (CA) and measured lifespan using the CA food protocols. We independently replicated these studies at another site (FL) using the CA adult diets only. Similar to our studies with NY stocks, both NY and FL labs found that the CA stocks of per⁰¹ and tim⁰¹ mutants responded to DR with similar or greater lifespan extension than controls (Figures S1L–S1O). We also identified a difference in protocol regarding the use of live or deactivated yeast in the developmental medium for CA diet experiments (all other dietary components matched published CA protocols). We tested this with the CA stocks and found that the use of live yeast during development does not alter the response of per⁰¹ mutants to DR (Figure S1P).

To compare results between labs, we plotted median lifespan for CA stocks from the three different locations, including previously published data (Figure S1Q) showing that tim⁰¹ and per⁰¹ mutants had median lifespans an average of only 6.5 and 14 days longer on dietary restriction (DR) than on standard food. The median lifespans for their genetic controls were 20 and 22 days longer, respectively, on DR (see CA; Figure S1Q). In contrast, in NY and FL, DR-mediated lifespan extension of circadian mutants was similar to or greater than that observed for controls (Figure S1Q). In NY, DR extended median lifespan of tim⁰¹ and per⁰¹ mutants by 32 and 35 days, respectively, whereas DR extended median lifespan of their controls by 20 and 25 days, respectively. FL results were similar (Table S1). Relative changes are also presented as the percent increase in median lifespan mediated by DR over the standard diet (Figure S1R). Again, Katewa et al. (2016) showed that, compared to controls, DR-mediated lifespan extension was reduced in circadian mutants, whereas both NY and FL labs showed similar or greater lifespan extension in mutants. Because each lab used the same genetic stocks and adult food protocols (and NY matched developmental diet as well), these results suggest that neither differences in food protocol nor genetic differences account for the differences in lifespan responses to DR.

Taken together, our results suggest that environmental differences account for the differing responses of circadian mutants to DR. While aging and lifespan phenotypes can be affected by differences in food protocol and genetic background, here we found that differences in food protocol were not sufficient to account for the conflicting per and tim mutant phenotypes. Moreover, two independent labs observed similar or greater responses to DR in circadian mutants compared to controls using genetic isolates obtained from the original study (Katewa et al., 2016), indicating that genetic background is also an unlikely explanation for conflicting results. Because many environmental factors influence highly variable phenotypes such as lifespan, caution should be exercised in drawing conclusions based on quantitative effects of DR. Environmental factors that may impact aging studies include: temperature, humidity, light levels, and daily fluctuations of these parameters in incubators and in rooms where experimental flies are transferred to fresh food; the frequency of transfers; the frequency and method of scoring death; and the freshness of the food to which they are transferred. Most of these technical parameters were common between NY and FL labs (see Table S2). Specifying technical parameters should increase the reproducibility of results, and ruling out these influences will help to identify other environmental factors that impact lifespan and provide biological insight into mechanisms of aging.

Notably, the circadian mutants obtained from the original study initially exhibited a developmental delay and atypical mucoid, orange feces, suggestive of differences in associated microbiota between various lines. Both phenotypes largely disappeared within 2–3 generations. 16S sequencing of colonies obtained by plating gut microbiota of the CA stocks when they first arrived revealed two bacterial species not commonly found in Drosophila, which appear to cause the fecal orange color and mucoid texture: Methylobacterium extorquens (reddish bacteria) and Dermacoccus nishinomiyaensis (yellow mucoid bacteria). While we were not able to colonize our NY stocks or maintain the original CA stocks with these bacterial species to test their effect on DR, their prevalence in the original CA stocks likely reflects other differences in microbe association for otherwise genetically similar flies. Thus, our results suggest that differences in environmental conditions, affecting the composition of fly-associated microbes, may have significant effects on lifespan for specific genotypes. Given the relationship between circadian mutants, nutrient intake, and immune response (Allen et al., 2016), as well as the direct contribution of environmental microbes to fly nutrition (Yamada et al., 2015), future studies may uncover fundamental mechanisms linking microbial metabolism with circadian-regulated host physiology and lifespan.