Daily temperature cycles prolong lifespan and have sex-specific effects on peripheral clock gene expression in Drosophila melanogaster

Biological processes that oscillate in phase with environmental variation increase the adaptability of an organism by anticipating predictable fluctuations in the environment (Schibler, 2017). Environmental variation at the daily level is anticipated by circadian rhythms in organisms. Those rhythms are generated by a set of highly conserved clock genes that are almost ubiquitously expressed in cells throughout the body [reviewed in Cox and Takahashi (2019) and Patke et al. (2020) for mammals and Drosophilamelanogaster, respectively]. The circadian system consists of a central pacemaker in the brain that is primarily entrained to the 24 h day/night cycle by light, and sets the timing of peripheral clocks throughout the body via a combination of internal entraining cues, including body temperature (Schibler, 2017).

Circadian rhythms are intrinsically linked with optimal physiological function because many physiological processes are under circadian control. Several diseases, including cancer (Chan and Lamia, 2020; Fu et al., 2002; Lee et al., 2010), metabolic syndrome (DiAngelo et al., 2011; Kohsaka et al., 2007; Sohail et al., 2015; Turek et al., 2005) and mental illness (Baird et al., 2012; Johansson et al., 2016), have been linked to disruptions of circadian machinery, especially via a dampening of amplitude and shifts in acrophase. Molecular pathways that underlie these conditions, such as Ras in cancer (Relógio et al., 2014; Williams et al., 2001), fat metabolism in obesity (Katewa et al., 2016) and tauopathy in neurological disease (Kim et al., 2018; Means et al., 2015), are also present and entwined with circadian machinery in D. melanogaster. In contrast, robust (i.e. high-amplitude) circadian rhythms are associated with resilience to shift work in humans (Reinberg et al., 1980), and better metabolic health (He et al., 2016; Katewa et al., 2016) and longevity (Hurd and Ralph, 1998; Katewa et al., 2016) in rodents and D. melanogaster. In addition, ageing is associated with a progressive weakening of circadian rhythmicity, an outcome that is highly conserved in mammals, fish and insects (Hood and Amir, 2017; Kunieda et al., 2006; Stankiewicz et al., 2017; Zhao et al., 2019). Therefore, the manipulation of cues that entrain the amplitude of clock gene expression could have beneficial effects in both mammals and insects.

Temperature is thought to be a universal entraining cue for clock gene expression in heterothermic insects (Glaser and Stanewsky, 2005) and even in homeothermic mammals (Buhr et al., 2010). Indeed, temperature fluctuations within the physiological range have been shown to modulate the amplitude of clock gene rhythms in vitro in mammalian tissue (Brown et al., 2002). In flies, cycling ambient temperatures similarly entrained clock gene rhythms (Boothroyd et al., 2007), synchronized activity rhythms (Busza et al., 2007; Sehadova et al., 2009) and led to faster rates of larval development compared with equivalent constant temperatures (Petavy et al., 2001). Thus, the effect of rhythms of temperature on circadian rhythms has important implications for human health, but efforts to investigate the role of temperature as an entraining cue in vivo have been impeded by the difficulty of manipulating the circadian rhythm of body temperature in homeotherms. To overcome this challenge, we previously manipulated energy intake in rats to elicit reliable and reproducible changes in the circadian rhythm of body temperature (Goh et al., 2016), aiming to establish whether there was an association between the body temperature rhythm and gene expression in peripheral tissues. An association was found, but it was not possible to conclude that the changes in gene expression were because of the altered body temperature rhythm, in part because nutritional signals, such as insulin, can also entrain the circadian clock (Crosby et al., 2019). Therefore, with the rat model, the effects of body temperature rhythms on clock gene expression cannot be disentangled from the effects of diet (Reinke and Asher, 2019).

Drosophila melanogaster is an ideal animal model to test the effects of body temperature on clock gene rhythms and lifespan, because the potential entraining effects of diet can be eliminated. Because D. melanogaster is an ectotherm, it is possible to manipulate its body temperature by manipulating its ambient temperature (Goda and Hamada, 2019). In this study, we investigated the effect of daily cycling ambient temperature on the lifespan of male and female D. melanogaster. In addition, we aimed to determine whether exposure of the flies to cycling temperatures enhanced the amplitude of clock gene expression (i.e. Clk, cyc, per, tim, Pdp1, vri and cry). Given that over-expression of the clock genes tim and per in abdominal tissues, but not the brain, have been reported to extend lifespan in Drosophila (Katewa et al., 2016; Rakshit and Giebultowicz, 2013), we elected to separate the head and body in this study to gain some discrimination between the expression of clocks in different tissues. Because light is regarded as the strongest entrainer of circadian rhythms, we chose to expose the flies to synchronous light and temperature cycles, testing the hypothesis that the circadian rhythms would be enhanced when a temperature cycle is superimposed on the normal light:dark cycle. We also investigated the effect of cycling temperature on the expression of heat shock protein 83 (Hsp83), frost (Fst) and senescencemarkerprotein-30 (smp-30), which are involved in molecular pathways that may link temperature and lifespan. Heat shock proteins are highly conserved molecular chaperones that play an essential role in cellular responses to stress, making them a universal protective mechanism against cellular damage (Åkerfelt et al., 2010). Furthermore, induced and/or increased rhythmicity in stress response genes has been proposed as an adaptive response to oxidative stress and ageing in D. melanogaster (Kuintzle et al., 2017). Heat shock proteins are upregulated during thermal stress and elevated levels of expression have been linked to lifespan extension in D. melanogaster under transient heat and cold exposure (Hercus et al., 2003; Klepsatel et al., 2016; Le Bourg, 2007). In particular, Hsp83 is of interest because it is associated with temperature-mediated circadian entrainment and couples behavioural rhythms to clock gene rhythms in D. melanogaster (Goda et al., 2014; Hung et al., 2009). Likewise, Fst is a stress-response gene that is upregulated during cold exposure in D. melanogaster, and has been implicated in resilience to low temperature (Goto, 2001; Sinclair et al., 2007). Lastly, smp-30 is a highly conserved marker of senescence that is widely expressed (Ishigami and Maruyama, 2007; Scott and Bahnson, 2011). Mammalian studies have revealed that smp-30 has various anti-apoptotic, anti-oxidative and anti-inflammatory functions (Ishigami et al., 2002; Jung et al., 2015; Son et al., 2008). In D. melanogaster, the gene product of smp-30 is strongly upregulated during cold acclimation (Goto, 2000), and smp-30 may thereby play a role in temperature-mediated longevity in flies. We hypothesized that (1) flies housed under cycling temperature conditions (T_cyc) will have longer lifespans and increased levels of smp-30, and (2) T_cyc housing conditions will lead to higher amplitudes of clock gene rhythms and the temperature-responsive Hsp83 and Fst, particularly in aged flies.

Drosophila melanogaster Meigen 1830 from a stock population that was established in 2012 using wild-caught flies from Innisfail, Queensland, Australia (Dugand et al., 2018), were reared in a controlled temperature (CT) room at 25°C and 60% relative humidity. For experimentation, the flies were placed in temperature-controlled incubators set at 60% relative humidity (A1000 plant growth cabinet; Conviron, Winnipeg, Manitoba, Canada). The incubators were used to expose flies to two regimens of environmental temperature: a constant temperature condition (T_con) under which the incubator was kept at 25°C constantly, close to the normal preferred temperature of D. melanogaster (Hamada et al., 2008); and a cycling temperature condition (T_cyc) under which the incubator altered in a square wave cycle between 28°C during the light phase and 22°C during the dark phase (i.e. a 3°C amplitude oscillating around a daily mean of 25°C).

All the flies were raised on a standard cornmeal fly medium consisting of 1% (w/v) agar, 2% (w/v) brewer's yeast, 8.5% (w/v) sugar, 6% (w/v) cornmeal and 0.25% (v/v) Nipagin (H5501; Sigma-Aldrich, St Louis, MO, USA). In all experiments, flies were reared at a standard larval density (50 larvae per vial) in 50 ml plastic vials with 10 ml of food medium. Adult flies were maintained at a density of 20 flies per vial.

Flies were reared at 25°C in the CT room containing the stock population. Adult flies were then allowed to mate for approximately 24 h after eclosion. The non-virgin adult flies were then anaesthetized using ice, separated by sex and placed in a vial containing 10 ml of food medium at a density of 20 flies per vial. These vials were placed in the incubators, randomly, at either T_con or T_cyc (N=100 flies per sex per treatment). Every 2–3 days, the number of deaths in each vial was scored when the flies were transferred to new vials containing fresh food medium. Flies that escaped during vial transfers were treated as censures.

For the lifespan assay, mated flies were anaesthetized using ice, separated by sex and placed into vials containing 10 ml of food medium at a density of 20 flies per vial. These vials were randomly assigned to either T_con or T_cyc in the incubators. Initial experiments began with 96 vials of flies for each sex (N=1920 flies per sex). The flies were transferred to vials containing fresh food medium every 2–3 days. After 14 days (an age that we refer to as ‘young’), one-eighth of the flies were transferred to a microfuge tube at each of four ZT points (ZT0=lights on, ZT6, ZT12 and ZT18), flash-frozen in liquid nitrogen, and stored at −80°C until RNA isolation. The other half of the flies were processed similarly after 40 days (an age that we refer to as ‘aged’). We processed four to six vials (biological replicates) per time point, per sex, per treatment.

Prior to RNA isolation, the frozen flies were vortexed in a TissueLyser LT (Qiagen, Victoria, Australia) and passed through a pre-chilled no. 25 standard mesh screen to separate the heads from the bodies. Total RNA was isolated separately from the heads and bodies using the TissueLyser LT and QiaZOL (Qiagen) as per the manufacturer's instructions. Reverse transcription was carried out at 42°C using murine Moloney leukemia virus reverse transcriptase (Promega, Sydney, Australia), 0.25 ng of random hexamers (Promega), and either 500 ng of total RNA from the heads, or 1000 ng of total RNA from the bodies. The resulting cDNA was purified using the QIAquick PCR purification kit (Qiagen).

The expression profile of seven clock genes (cyc, Clk, per, tim, Pdp1, vri and cry), as well as Hsp83, smp-30 and Fst, was measured. Real-time quantitative PCR (qPCR) using the SYBR Green PCR Kit (Qiagen) was carried out in duplicate in 10 μl reactions, containing 0.5 µmol l⁻¹ primers and 1 µl cDNA, on a CFX384 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). A temperature profile of 95°C for 10 min, followed by 40 cycles of 95°C for 1 s, 60°C for 1` 5 s and 72°C for 5 s, was used for amplification of DNA sequences. Reference genes (SdhA, Act5C, RpL32 and eEF1α2) were selected with the potential effects of ageing in mind (Ling and Salvaterra, 2011; Ponton et al., 2011). Relative quantification of gene expression was performed via interpolation against a seven-point standard curve of 10-fold serial dilutions of purified PCR product. The level of relative gene expression was then normalized against the geometric means of four reference genes (Vandesompele et al., 2002). To account for between-run variations, six samples, obtained from vials (20 flies per vial) of 5- to 10-day-old male (N=3) and female (N=3) flies, were included in each PCR run, and all values were normalized with reference to these six reference samples. The primer sequences, primer concentrations and PCR conditions that were used are provided in Table S1. Primer pairs were designed with Primer-BLAST (Ye et al., 2012) to include introns to avoid amplification from genomic DNA. All gene products were sequenced to confirm specificity of amplification. To compare the level of gene expression between ages and sexes, the mesor (daily mean) and amplitude of gene expression in the young male flies were set to a baseline of 1, and the data from the other groups (the aged males, young females and aged females) were expressed relative to that of the adult male flies. The acrophase of all groups were expressed as the number of hours advanced or delayed compared with that of the adult males.

Survival curves were plotted and compared using a log-rank (Mantel–Cox) test in Prism (version 7.03 for Windows; GraphPad Software, La Jolla, CA, USA). A Bonferroni corrected threshold (P=0.008) was used for the six pairwise comparisons between the four groups (male T_con, male T_cyc, female T_con, female T_cyc).

To analyse the profiles of gene expression, the expression datasets acquired from qPCR were first tested for rhythmicity using the R package ‘CircaCompare’ (version 3.6.2 for Windows), as described previously (Parsons et al., 2020). If gene expression was rhythmic (at P<0.05), then the estimates and standard errors for the mesor (defined as the mean value around which the daily rhythm oscillates), amplitude (defined as half of the distance between the peak and the nadir of an oscillation) and acrophase (defined as the time when the peak of an oscillation occurs), that describe each profile, were determined. The effect of age, sex and temperature treatment on the mesor, amplitude and acrophase of gene expression was analysed using three-way ANOVAs (with age, sex and culture temperature as sources of variation) in Prism (version 8.4.1 for Windows, GraphPad Software). When significant main effects or interactions were identified, CircaCompare was used to determine whether there was a significant difference (P<0.05) in the rhythmicity of the gene expression profiles of interest by pairwise comparison of the 95% confidence intervals. When one or more of the gene expression profiles was not rhythmic, three-way ANOVA could not be used, and thus two-way ANOVAs were used where possible. When there were multiple non-rhythmic profiles per gene that excluded the use of both three-way and two-way ANOVAs, pairwise comparisons were made using CircaCompare to detect differences in gene expression. All reported P-values are rounded to three significant figures.

The average ambient temperature, relative humidity and light intensity in the CT room and the incubators during a representative 10-week period of experimentation is given in Table 1. For the T_cyc incubators, it took approximately 2 h for the incubators to cool from 28 to 22°C, or to heat from 22 to 28°C, during the transition from the light phase to the dark phase, and vice versa, respectively.

The flies housed at T_cyc had a longer median survival (male: T_con=38 days, T_cyc=47 days; female: T_con=34 days, T_cyc=47 days) and 90% survival (male: T_con=47 days, T_cyc=54 days; female: T_con=47 days, T_cyc=57 days) than flies housed at T_con (P<0.0001 for both sexes) (Fig. 1). However, there was no difference in the maximum lifespan between T_con and T_cyc (60 days for all groups). The male and female flies had similar survival curves within each temperature profile (Bonferroni corrected threshold=0.008; T_con, P=0.02; T_cyc, P=0.92).

The circadian profiles of gene expression in D. melanogaster heads are shown in Fig. 2, while Fig. 3 shows the amplitude, mesor and acrophase that summarize the circadian expression of those genes. The associated P-values for all three-way and two-way ANOVAs are given in Table S2, and P-values for all pairwise comparisons are provided in Table S3.

In male flies, cyc expression developed rhythmicity when under T_cyc conditions. T_cyc also elicited an increase in amplitude in per expression in aged male flies (P<0.01). In aged female flies, the amplitudes of per, Pdp1, vri and cry were all higher under T_cyc compared with T_con (P<0.05; P<0.001 for cry). Interestingly, in young male flies, T_cyc conditions led to lower amplitudes in Clk, tim and Pdp1 (P<0.001), with a trend for a lower amplitude in vri (P=0.051), and a loss of rhythmicity in cry, when compared with T_con. Only Clk and tim in young female flies showed a similar decrease in amplitude when housed under T_cyc (P<0.001). None of these lowered amplitudes persisted in the aged flies, regardless of sex (P>0.05).

Compared with T_con, young male and young female flies under T_cyc conditions had a lower mesor of Clk, tim, Pdp1 and vri (P<0.01), which was not observed in the aged flies. Conversely, the aged male and female flies showed higher mesor for per and cry when under T_cyc compared with T_con (P<0.05).

Housing under T_cyc led to varying effects on acrophase. When compared with T_con, T_cyc led to a delay in the acrophase of per in all flies (P<0.05), and an advance in the acrophase of cry in young flies (P<0.01). In aged flies, however, T_cyc conditions led to a delay in the acrophase of Clk and vri in both sexes (P<0.01), as well as advances in the acrophase of tim and Pdp1 in females (P<0.05), but a delay in the acrophase of cry in males (P<0.05).

The amplitude of Hsp83 expression was higher in all flies under T_cyc, with a higher mesor being observed in only the aged males. The expression of Fst was rhythmic only in aged males under T_cyc, but the mesor of Fst was lower under T_cyc than T_con in females (P<0.05). In aged flies, the mesor of smp-30 was higher under T_cyc than T_con in both sexes (P<0.001), and the aged female flies under T_cyc rhythmicity of smp-30 expression. Notably, exposure to T_cyc rescued the age-related decrease of smp-30 in female flies, and reduced the decrease in male flies by 28%. No significant differences in acrophase were observed for Hsp83, Fst or smp-30.

The circadian profiles of gene expression in D. melanogaster bodies are shown in Fig. 4, while Fig. 5 shows the amplitude, mesor and acrophase that summarize the circadian expression of these genes. The associated P-values for all three-way and two-way ANOVAs are given in Table S4, and P-values for all pairwise comparisons are provided in Table S5.

T_cyc conditions led to sex-specific changes in clock gene amplitude. In young male flies, cyc was the only clock gene that showed an increase in amplitude or rhythmicity under T_cyc compared with T_con (P<0.01), whereas in young female flies, the amplitude of per and vri were higher under T_cyc compared with T_con (P<0.05). The amplitude of all clock genes was higher in aged male flies under T_cyc than under T_con (P<0.05), except for vri, where there was a trend for a higher amplitude (P=0.058). Furthermore, T_cyc completely rescued the age-related decrease in the amplitude of Clk, per, Pdp1 and vri expression, and lessened the degree of the age-related decrease of the remaining two clock genes, tim and cry, by 33 and 16%, respectively. In aged female flies, higher amplitudes under T_cyc were only observed in per and cry (P<0.05). Unlike the male flies, only the amplitude of cry in the female flies decreased with age when under T_con (P<0.001).

T_cyc conditions had no effect on the mesor of any gene in young male flies. In aged male flies, however, T_cyc conditions led to higher mesor of Clk, tim, Pdp1 and cry (P<0.01), thereby rescuing or reducing the age-related decrease in these four clock genes, but the mesor of cyc was lower under T_cyc than under T_con (P<0.05). The opposite effect was observed in the clock genes of aged females, where T_cyc exacerbated the decrease in the mesor of tim, Pdp1 and vri in females, and elicited an age-related decrease and increase in the mesor of per and cyc, respectively.

T_cyc conditions elicited few changes in acrophase in D. melanogaster bodies. Compared with T_con, T_cyc led to a delay in the acrophase of Clk in young and aged flies of both sexes, as well as a delay in the acrophase of vri in aged flies in both sexes (P<0.05). A small advance in the acrophase of tim was also observed in young female flies (P<0.05).

T_cyc conditions consistently elicited higher amplitudes of Hsp83 in young flies and in aged flies of both sexes. The profile of Fst gained rhythmicity under T_cyc in young males and in aged flies of both sexes. For smp-30, gain of rhythmicity was observed in the amplitude of young females and in aged males under T_cyc compared with T_con.

In both sexes, ageing under T_con led to a lower mesor for Hsp83 and smp-30 (P<0.01), but not Fst. The mesor of smp-30 decreased by 48% between young and aged males under T_cyc, compared with 78% under T_con. In female flies, ageing under T_cyc conditions completely rescued the age-related decrease in smp-30 mesor. Similarly, the age-related decrease in the mesor of Hsp83 that was observed under T_con (P<0.001) was reduced in male and female flies under T_cyc conditions. Conversely, T_cyc elicited a decrease in the mesor of Fst in both aged male and female flies (P<0.05).

We tested the effect of daily cycling temperature on lifespan and the profile of clock gene expression in young and aged D.melanogaster. We found that the superimposition of daily temperature cycles, on top of the light:dark cycle, increased longevity. Temperature cycling rescued the age-related decrease in amplitude of almost all clock genes that occurred in the bodies of male flies at T_con, an outcome that aligns with the hypothesis that an increase in the amplitude of molecular clock rhythms by manipulation of the strength of a temperature zeitgeber had beneficial effects on physiological function. The robust rhythms of Hsp83 and the rescue of smp-30 expression in the aged male and female flies under T_cyc potentially indicate that cycling temperatures elicit an increased capacity to cope with age-related oxidative stress.

An inverse relationship between constant ambient temperature and lifespan in ectotherms, including D. melanogaster, is well established (Klass, 1977; Liu and Walford, 1966; Miquel et al., 1976). Even in homeotherms such as mammals, a reduction in body temperature of 0.5°C extends lifespan, independently of caloric intake (Conti et al., 2006). One potential mechanism of ageing postulates that ageing is caused by the production of reactive oxidative species from oxidative metabolism (Harman, 1956; Liguori et al., 2018; Sohal and Weindruch, 1996). Therefore, temperature should have a direct effect on the rate of ageing via the Q₁₀ effect, which describes how the rate of chemical processes changes with temperature (IUPS Thermal Commission, 2001). Indeed, increased metabolism, and therefore increased production of reactive oxygen species (ROS) at higher temperature, accelerates ageing (Buttemer et al., 2010; Malek et al., 2004; Van Voorhies and Ward, 1999). Because the Q₁₀ relationship is exponential, the increase in metabolism from 25 to 28°C is larger than the decrease in metabolism from 25 to 22°C. Accordingly, the fact that we observed longer lifespan in the T_cyc flies argues that the beneficial effect of spending half the day at 22°C, or the effect of T_cyc as an additional zeitgeber that reinforces circadian physiology, or a combination of both, alleviated the deleterious effect of exposure to higher temperatures during the day. Previous findings that temperature cycles rescue behavioural rhythms in aged male and female Drosophila (Luo et al., 2012) and rescue per (but not tim) expression in aged male flies (Zhao et al., 2018), lend further support to the latter conclusion that the benefits of exposure to T_cyc strengthens circadian rhythms and outweighs the harmful effects of exposure to higher daytime temperature. Likewise, our finding that the expression of circadian systems was amplified under T_cyc does not appear to be a simple thermodynamic effect.

Our observation that the extension in lifespan under T_cyc conditions was associated with the maintenance of the amplitude of all clock genes (except in vri, where a similar trend was observed) in the bodies, but not the heads, of aged male flies supports the hypothesis that the maintenance of robust circadian rhythms via temperature cycles had beneficial effects. Our findings are broadly supported by previous studies showing that whole-body over-expression of cry increased lifespan, prevented the age-related dampening of activity rhythms in D. melanogaster, and was associated with an increase in amplitude of per, tim, Pdp1 and vri (but not Clk or cyc) (Rakshit and Giebultowicz, 2013). Similarly, reinforcing the circadian clock via the over-expression of tim, but not per, also led to longer lifespan (Katewa et al., 2016). Consistent with the general principle that peripheral clocks are more plastic than central clocks in mammals and D. melanogaster (Dibner et al., 2010; Xu et al., 2008), neuron-specific over-expression of cry and per in both of the aforementioned studies did not promote longevity (Katewa et al., 2016; Rakshit and Giebultowicz, 2013). In contrast, the over-expression of per in the fat body, gut and Malpighian tubules was sufficient to increase lifespan (Katewa et al., 2016).

We observed several sex differences in the mesor and amplitude of gene expression in D. melanogaster. First, age did not dampen the amplitude of clock gene expression in the bodies of female flies, which precluded a rescue of age-dampened rhythms by T_cyc similar to that observed in the male flies. Second, T_cyc elicited small but significant increases in the amplitude of Pdp1, vri and cry in the heads of the aged female flies, which were not observed in the aged male flies. Lastly, per expression, which is temperature driven in D. melanogaster, partially via temperature-dependent splicing of the 3′-UTR region (Goda et al., 2014; Majercak et al., 2004; Yoshii et al., 2007), was increased in the heads and bodies of both sexes under T_cyc, albeit to a greater extent in the males. The sex differences in the bodies may be explained by the ovaries in the females, which contribute approximately 70% of the RNA in the female abdomen, and do not appear to have oscillating levels of clock genes (Beaver et al., 2003; Hardin, 1994). Therefore, any temperature-driven changes in the rhythmicity of clock genes in the female body may have been masked by the non-cycling ovarian RNA.

The beneficial effect of daily temperature cycles on lifespan may be partly attributable to an increase in the levels of smp-30 expression. The smp-30 gene is highly conserved between D. melanogaster and vertebrates and is widely expressed (Scott and Bahnson, 2011). Under T_con, our flies experienced an age-related decrease in smp-30 expression, as has been reported for mammalian SMP30 in aged rats, and its homologs in aged zebrafish (Fujisawa et al., 2011; Fujita et al., 1992). In contrast, the level of the smp-30 protein has been reported to increase with age in D. melanogaster, but the comparability of these results is confounded by differences in the light regimens used (Goto, 2000). In mammals, SMP30 deficiency leads to shortened lifespan (Ishigami et al., 2004) and higher levels of oxidative damage in the brain (Son et al., 2006), whereas the over-expression of SMP30 leads to lower levels of ROS (Handa et al., 2009; Son et al., 2008). Based on the oxidative stress theory of ageing (Harman, 1956), higher levels of smp-30 expression or activity may confer protection against oxidative insults in aged organisms.

In the present study, exposure to T_cyc led to robust expression and rhythmicity of Hsp83 in flies of both sexes, which contrasts with the low rhythmic or arrhythmic expression in fly heads and bodies under T_con. Taken together with the well-known relationship between the heat shock response and longevity in invertebrates (Baldi et al., 2017; Steinkraus et al., 2008; Tower, 2011; Zhao et al., 2005), the increased expression and rhythmicity of Hsp83 in aged flies in our study could have provided an increased capacity to respond to ageing-related cellular stresses (Kuintzle et al., 2017). In contrast, we found no effect of T_cyc on Fst expression. Thus, it appears that Hsp83, but not Fst, is implicated in mediating the physiological effects of temperature rhythms in our study.

As with many other species (Piper et al., 2011), dietary restriction protocols extend lifespan in D. melanogaster (Bass et al., 2007). Interestingly, energy intake and metabolism scale linearly in D. melanogaster over the ambient temperatures range from 13 to 33°C (Klepsatel et al., 2019). Furthermore, feeding behaviour in D. melanogaster is strongly rhythmic, and the majority of food intake occurs between ZT0 and ZT2 (Xu et al., 2008). Therefore, despite their longevity, the flies housed at T_cyc probably consumed more than flies at T_con, because they were at 28°C during this critical feeding window, while the flies at T_con were at 25°C during that window. The interaction between age, ambient temperature and food intake remains to be empirically verified, especially as the effects of ambient temperature and calorie restriction on lifespan in D. melanogaster appear to utilize overlapping pathways (e.g. 4E-BP and IGF-1) (Carvalho et al., 2017; Cintron-Colon et al., 2017).

The current study shows that temperature rhythms modulate lifespan in D. melanogaster, and data from the male flies support the notion that circadian amplitude may link temperature rhythms with longevity. The female flies also lived longer under T_cyc, but did not show as many increases in the amplitude of clock gene expression as the males, and further study is required to determine if any changes in the amplitudes were masked by the contribution of non-cycling RNA from the ovaries. It may also be that multiple pathways are present. The expression of smp-30 and Hsp83 are implicated as potential effectors of ambient temperature rhythms, and may confer increased resistance to age-related oxidative and inflammatory damage. Further study that discriminates the effects of T_cyc on sleep, activity, food intake and interaction with the light:dark cycle is required to disentangle the specific mechanisms that are responsible for modulating the various known pathways that modulate lifespan.

The authors thank the staff of the Plant Growth Facility at the University of Western Australia for providing and maintaining the temperature-controlled cabinets.