The present study systematically examined the entrainment of clock-controlled behavior to daily environmental temperature gradients. As a result, a number of key properties of circadian temperature entrainment were identified. Collectively, these properties, as described below, represent a circadian temperature entrainment mechanism that is optimized in its ability to detect the time-of-day information encoded in natural environmental temperature profiles.
First, daily temperature gradients show a remarkable efficiency at entraining behavioral and molecular daily rhythms in Drosophila. The minimal range identified here for efficient entrainment by temperature gradients (1.5 to 4°C) is similar to that previously reported for square wave temperature cycles (2 to 3°C; [19, 22]). Temperature input pathways, therefore, appear to be more than sensitive enough to respond to a wide variety of daily environmental temperature profiles.
Second, synchronization of circadian locomotor behavior is associated with detection of changes in relative temperature rather than absolute temperature. Circadian phase showed comparatively little dependence on absolute temperature, but was, instead, reliably predicted based on the phase of the relative temperature peak and the relative temperature of release. The use of a diverse panel of daily temperature wave forms and associated release temperatures in this study allowed this property of circadian temperature entrainment to be recognized. A crucial benefit of entrainment to relative temperature changes is that circadian clocks can effectively respond to daily temperature oscillations in the presence of seasonal (or other) changes in average temperature.
Third, the daily temperature peak as defined by the alignment of both ascending and descending daily temperature phases sets the circadian phase of entrainment. Peak circadian locomotor activity was strongly linked to the configuration of the ascending and descending temperature phases during entrainment. And this association was maintained irrespective of the phase of other features of the synchronizing daily temperature profile such as the trough. Again, analysis of a diverse panel of daily temperature wave forms in the present study was instrumental in allowing this property to be identified. A combined role of temperature increases and decreases in the circadian entrainment of eclosion in Drosophila pseudoobscura populations was proposed in an early study by Zimmerman et al. [23]. It is, therefore, very well possible that circadian temperature entrainment of locomotor activity and eclosion make use of the same mechanism.
Fourth, the relative temperature of release into free run modulates circadian phase. Whereas release into free running conditions at a temperature matching the entrainment peak has no apparent effect on circadian phase, progressive phase advances result from release at temperatures approaching the entrainment trough. Thus, changes in average daily temperature modulate rather than block entrainment to daily temperature gradients. This arrangement may allow for the simultaneous integration of both seasonal and daily information in generating rhythmic behavior. Seasonal modulation of circadian behavior occurs in part as a result of temperature sensitive splicing of the last intron of the clock gene period [16]. Indeed, lower temperatures directly induce more efficient splicing and, as a result, an advanced molecular and behavioral phase [17, 24]. Although temperature-dependent splicing of per may have contributed to the phase advances observed upon release at lower absolute temperatures this mechanism cannot fully account for the observed sensitivity to relative temperature changes. Behavioral phase advances were observed upon release at constant trough temperature even for mutants that block per splicing (see Additional file 6). Moreover, no obvious molecular phase advance in per transcript or protein accumulation was detected for wild-type flies released at low relative temperature (see Figure 9A and 9B, above).
Fifth, circadian phase responses to daily temperature profiles occur slowly and often require several days to be completed. Daily temperature gradients are weaker synchronizers of circadian behavior than light/dark cycles and have a more limited phase shifting potential per cycle. As a result, large temperature-dependent phase shifts are not triggered by single, anecdotal fluctuations in temperature, but rather by repeated exposure to convergent phase-shifting signals over the course of a number of days. The Drosophila clock is, therefore, tuned to repeated daily temperature components and able to resist dramatic short-term responses to irregular fluctuations in environmental temperature. Given the complexity and irregularity of natural environmental temperature profiles, this appears to be a very useful adaptation. The temperature-dependent phase responses described here are comparable to those reported earlier for square-wave temperature pulses and cycles [21, 25, 26]. For example, Boothroyd and colleagues found that it took flies 5 days to complete a 10-h phase advance in response to a square wave 18°C/25°C temperature cycle, whereas Busza et al. observed that 8-h phase advances and 8-h delays in response to a 20°C/29°C square wave were achieved over a time frame of more than 6 days or 5 days, respectively [21, 25].
Sixth, daily cycles in relative temperature can synchronize circadian behavior even when they are presented as part of more complex temperature profiles. Entrainment to daily temperature gradients persisted even in the context of continuous gradual temperature increases or decreases. To effectively detect the time-of-day information encoded in environmental temperature profiles circadian clocks have to be able to sense daily changes in relative temperature in the context of patterns that contain unrelated thermal signals, for example, due to weather-related developments. The present study shows that this is, indeed, the case in examples where continuous changes in absolute temperature are used as a back-drop for entrainment.
Seventh, naturally aligned daily temperature and light cycles provide cooperative circadian entrainment. Daily temperature gradient and light/dark entrainment reinforced each other if the phases of ascending and descending temperature were in their natural alignment with the light and dark phases, respectively. In nature, entrainment of the Drosophila circadian clock to temperature profiles occurs in the context of a daily light/dark cycle. To achieve the highest sensitivity towards environmental time-of-day information the circadian clock is expected to show light/dark-entrained and temperature-entrained phases that approximately coincide when these zeitgebers are in their natural alignment. The phase determinants of temperature and light entrainment both result from changes in solar irradiance and normally occur in succession during the second half of the day. Peak temperature, the main synchronizing signal for thermal entrainment (see above), generally falls in the afternoon several hours in advance of dusk [27], which is the major phase determinant of light entrainment (for example, [20, 28]). This natural association of temperature and light-dependent phase resetting likely facilitates cooperative entrainment. As shown here, separate or combined treatments with daily light/dark cycles and temperature gradients produce similar circadian behavioral phases when the phases of ascending and descending temperature match those of light and dark, respectively. In this arrangement, as in natural conditions, the daily temperature minimum coincides with the time just before dawn, while the temperature maximum is associated with a time point in the second half of the day.
What might be the molecular mechanism underlying the circadian temperature synchronization described in this study? Analysis of temperature-entrained gene expression indicates that the phase relationship between behavioral circadian rhythms and molecular rhythms in the fly head is relatively stable and similar to that observed previously for light entrainment. Peak expression of CLK/CYC regulated transcripts (per, tim, vri) occurs shortly after peak temperature and PER and TIM protein expression show the expected ~6-h delay relative to their transcript levels [29]. These observations confirm and complement analyses from a previous study comparing the genome-wide transcript profiles in the adult head during and after treatment of wild-type and arrhythmic mutant flies with daily light/dark or square-wave temperature cycles [21]. In that study many circadian transcript profiles were identified that showed the same fixed relationship between light-entrained and temperature-entrained circadian phases, indicating that the same molecular clock mechanism is responsible for both light-entrained and temperature-entrained gene expression rhythms. Moreover, the expression levels of clock gene transcripts (per, tim, Clk, vri, Pdp1-ε, cry) and proteins (PER, TIM) were found to respond to environmental temperature changes in a clock-dependent manner [21, 30]. The identification here of the peak temperature phase as the main determinant of the temperature-entrained behavioral phase suggests that associated molecular events are responsible for temperature entrainment. Perhaps the two most relevant of these are the peak in transcript levels for CLK/CYC-regulated transcripts, including per and tim, and the accelerated accumulation of PER and TIM proteins. Temperature-dependent modulation of circadian phase has been related to changes in per transcript levels as well as PER and TIM protein levels [16, 21, 24, 26, 30–34].
As mentioned before, per transcript levels are known to be regulated by temperature-dependent splicing of the terminal intron of per and this, in turn, affects PER protein levels as well as locomotor activity in the evening [16, 24]. At lower constant temperatures splicing of this intron is enhanced, which leads to earlier accumulation of per mRNA and protein and advanced evening activity. This mechanism does, however, not appear to be required for synchronization of the clock to daily temperature cycles. Entrainment of circadian behavior to daily temperature cycles presented in constant darkness was largely unaffected in mutants that prevent splicing of the terminal per intron (Additional file 6). In addition, per splicing levels in wild-type flies did not show obvious daily changes in temperature cycles during constant light [26].
Complex formation of PER and TIM protein with CRYPTOCHROME (CRY) [34] and destabilization of TIM and PER protein [32] have been implicated in circadian phase resetting in response to heat shock treatment. However, these responses can be separated from entrainment to daily temperature cycles based on the requirement of the former for treatment with high temperatures (>34°C) and the involvement of CRY [32, 34] which are both dispensable for synchronization to daily temperature cycles [19, 21, 25, 26, 33, 34]. Moreover, heat shock treatment preferentially results in phase delays, whereas both phase advances and delays are readily observed in response to daily temperature cycles ([25, 34]; data not shown).
Changes in TIM protein levels in a subset of clock neurons (type 2 Dorsal Neurons, DN2s, and Lateral Posterior Neurons, LPNs, and, to a lesser extent, type 1 and 3 Dorsal Neurons, DN1s and DN3s) were reported in association with a shifted alignment of simultaneously applied light/dark and temperature cycles [31]. It is not yet clear, however, whether these changes could be responsible for the advanced onset of evening activity recorded when the thermophase and cryophase of a square-wave temperature cycle were centered around the lights-on and lights-off transitions, respectively [31]. Observations resulting from the combination of a light/dark cycle with an anti-phase daily temperature gradient described here, for wild-type, but not arrhythmic mutant flies, indicate that the advance in evening activity onset is a clock-dependent response rather than a startle response following a sudden temperature decrease. Apart from a possible temperature-specific role for the LPNs, the set of clock neurons responsible for circadian temperature entrainment mostly coincides with the set known to control circadian light entrainment [25]. In this context, the subtle changes in TIM expression in the DN1s, which have been reported to modulate evening activity in the context of light treatment [35–37], may merit further examination.
Thus, although there is precedent for the temperature-dependent regulation of circadian phase via modulation of either per transcript levels or PER and TIM protein levels, neither of these mechanisms has been conclusively shown to account for circadian synchronization to environmental temperature cycles. Further molecular studies tracking the dynamics of circadian transcript and protein expression as well as post-translational modifications during temperature cycle-dependent phase advances and delays are expected to provide insight into this issue.
In contrast to Drosophila and other poikilothermic organisms, homeothermic mammals such as humans do not have a body temperature that directly tracks environmental temperature. Nevertheless, there is evidence supporting the involvement of daily temperature cycles in the maintenance of mammalian circadian rhythms because body temperature itself not only exhibits daily clock-controlled rhythms but can also act as a synchronizing cue [38, 39]. Two pathways in particular have been proposed to be involved in mediating entrainment to mammalian body temperature cycles: (1) Heat Shock Factor 1 (HSF1)-dependent transcriptional regulation and (2) control of gene expression by cold-induced RNA binding proteins [39]. In this context, it is intriguing that rhythmic expression of HSF1-regulated heat shock proteins genes was not only observed in the mouse liver [40], but also in the heads of Drosophila exposed to a synchronizing daily temperature cycle of modest amplitude [21]. Additional research is needed to address the potentially conserved role of these genes in temperature entrainment.