Mice in the study that were exposed to a brief dip in the levels of oxygen in the air that they were breathing adjusted more quickly to a new circadian rhythm than mice that received steady levels of oxygen, the researchers found.
In other words, the dip in oxygen levels seemed to help the animals adjust to the mouse equivalent of jet lag, according to the study, which was published today (Oct. 20) in the journal Cell Metabolism.” – Credit Sara G. Miller
Information and procedures carried out during the experiment are as followed:
O2 levels exhibit daily rhythms in blood and tissue of rodents
Physiological O2 rhythms reset clocks in cultured cells in HIF1α-dependent manner
Several core clock genes respond to changes in O2 levels in HIF1α-dependent fashion
Modulation of O2 levels accelerates the recovery of mice in a jet lag protocol
The mammalian circadian system consists of a master clock in the brain that synchronizes subsidiary oscillators in peripheral tissues. The master clock maintains phase coherence in peripheral cells through systemic cues such as feeding-fasting and temperature cycles. Here, we examined the role of oxygen as a resetting cue for circadian clocks. We continuously measured oxygen levels in living animals and detected daily rhythms in tissue oxygenation. Oxygen cycles, within the physiological range, were sufficient to synchronize cellular clocks in a HIF1α-dependent manner. Furthermore, several clock genes responded to changes in oxygen levels through HIF1α. Finally, we found that a moderate reduction in oxygen levels for a short period accelerates the adaptation of wild-type but not of HIF1α-deficient mice to the new time in a jet lag protocol. We conclude that oxygen, via HIF1α activation, is a resetting cue for circadian clocks and propose oxygen modulation as therapy for jet lag.
In mammals, a master circadian clock in the brain synchronizes subsidiary oscillators in peripheral tissues. The presence of cell autonomous oscillators in almost every cell in the body raises the question of how these oscillators are temporally coordinated. Because these cellular oscillators anticipate and function together in a proactive manner to environmental changes their temporal synchronization is critical (Schibler et al., 2015).
The molecular makeup for circadian rhythm generation is based on interlocked negative transcription-translation feedback loops (Feng and Lazar, 2012, Partch et al., 2014). The basic helix-loop-helix-PER-ARNT-SIM (bHLH-PAS) proteins BMAL1 and CLOCK heterodimerize and drive the expression of the Period (i.e.,Per1, Per2, and Per3) and Cryptochrome (i.e., Cry1 and Cry2) genes. Subsequently, PER and CRY proteins accumulate and repress the transcription of their own genes. An auxiliary essential feedback loop includes the orphan nuclear receptors of the REV-ERB and ROR families.
The quest for universal timing cues for peripheral clocks in mammals has yielded two principal entrainment signals: feeding and temperature. Both feeding rhythms and temperature cycles were shown to synchronize peripheral clocks and even uncouple them from the master clock in the brain (e.g., daytime restricted feeding) (Brown et al., 2002, Buhr et al., 2010, Damiola et al., 2000, Saini et al., 2012, Stokkan et al., 2001, Vollmers et al., 2009). Notably, both nutrient ingestion/processing and maintenance of body temperature are tightly linked to oxygen consumption. Furthermore, oxygen is readily available and recognized by most cells of the body, rendering it as an ideal signaling molecule. Hence, we posited that oxygen might function as a systemic timing cue for circadian clocks, a conjuncture that so far was never tested. For oxygen to function as a resetting cue for circadian oscillators, two requirements would be necessary: (1) it should exhibit daily rhythms and (2) peripheral oscillators must be sensitive to variation in oxygen levels within the physiological range.
We tested this hypothesis and identified daily rhythms in blood and tissue oxygen levels. These physiological oxygen rhythms were sufficient to synchronize clocks in cultured cells in a hypoxia-inducible factor 1α (HIF1α)-dependent manner. Furthermore, we found that moderate reduction in oxygen ambient levels, even for as short a period as 2 hr, accelerates the adaptation of wild-type, but not of HIF1α-deficient mice to the new time in a jet lag protocol. We suggest that oxygen functions as a prominent resetting cue for circadian clocks through HIF1α activation and propose oxygen modulation as a supportive therapy for jet lag.
Results and Discussion
Daily Rhythms in Blood and Tissue Oxygen Levels
To test whether there are daily variations in oxygen levels, we first monitored the oxygen consumption rate of mice using metabolic cages (Figure 1A). Oxygen consumption rate was elevated in the dark phase compared to the light phase. The increase in oxygen consumption during the night in mice coincided with their activity onset and food ingestion, both of which consume oxygen. This result incited us to examine whether there are daily changes in blood and tissue oxygen levels as well. We measured the oxygen levels in blood from mice using an oxygen optical fiber and found that they oscillate with zenith levels during the dark phase (Figure 1B). Concurrently, we employed a telemetric oxygen electrode device to continuously monitor oxygen levels in kidney of freely moving rats for several consecutive days (Figure 1C). Kidney oxygenation was rhythmic and reached its peak levels during the dark phase (Figures 1D and 1E). In line with previous reports, the mean daily O2 levels in kidney were ∼7% (6.94 ± 1.34%, mean ± SEM, and n = 5) (Carreau et al., 2011). The range of the rhythmic daily changes in kidney oxygenation was ∼3% O2 (Figure 1F). Taken together, our analyses identified daily oscillations in oxygen consumption and oxygen levels in blood and kidney with peak levels during the night.
HIF1α is a bHLH-PAS domain-containing transcription factor that responds to and participates in oxygen homeostasis (Majmundar et al., 2010). Under normoxia, HIF1α is rapidly degraded via the von Hippel-Lindau (VHL)-mediated ubiquitin-proteasome degradation pathway; however, once oxygen levels decrease, HIF1α degradation is inhibited and HIF1α accumulates. Hence, HIF1α protein levels are tightly regulated post-transcriptionally and inversely correspond to oxygen levels. Of note, HIF1α was previously reported to heterodimerize with other bHLH-PAS domain-containing proteins such as BMAL1 (Hogenesch et al., 1998). The above-described rhythms in oxygen levels prompted us to examine whether they incite daily changes in HIF1α levels. While Hif1α transcript levels were relatively constant throughout the day (Figure S1A), HIF1α nuclear protein levels exhibited daily rhythms with peak levels at zeitgeber time (ZT) ∼8 and ∼12 in mouse kidney (Figures 1G and S1B) and brain (Figures 1H and S1C), respectively. Thus, the accumulation of HIF1α in brain is ∼4 hr delayed compared to kidney and corresponds to the peak in REV-ERBα protein levels in both tissues.
Overall, we uncovered daily rhythms in blood and tissue oxygenation together with daily oscillations in HIF1α protein levels.
Physiological Oxygen Rhythms Synchronize Circadian Clocks in Cultured Cells in HIF1α-Dependent Manner
We set out next to examine whether physiological rhythms in oxygen levels, namely 5% and 8% O2 nadir and zenith, respectively, can synchronize circadian clocks in a population of cultured cells. We made special efforts to minimize any potential confounding effects by: (1) refraining from using firefly luciferase based circadian reporters, as the luciferase enzymatic activity is sensitive to oxygen levels (Doran et al., 2011), (2) maintaining the incubator gas composition and temperature as constant as possible. To this aim, we used special chambers with CO2, O2, and temperature controls (see Experimental Procedures) (Figure 2A). Temperature, CO2, and O2 levels were continuously monitored throughout the experiment with a constant temperature of 37°C, 5% CO2, and O2levels as indicated. To generate rhythms in oxygen levels, O2 was replaced with the inherent gas nitrogen. We applied the following experimental scheme; 24 hr after cells (i.e., Hepa-1c1c7 or NIH 3T3) were seeded, they were exposed to three consecutive cycles of 12 hr 5%, 12 hr 8% O2 each (mimicking physiological O2 rhythms), and subsequently released to constant 8% O2 (free running). The control cells were maintained under constant 8% O2 throughout the entire experiment (Figure 2B). Samples were collected at 4 hr intervals in the course of the free running period (day 4), and the expression level of clock genes was determined. Due to the weak coupling between cells in culture, clocks in individual cells fail to maintain phase coherence with their neighboring cells after several days (Nagoshi et al., 2004). Thus, as expected, the transcript levels of the various clock genes were relatively constant throughout the circadian cycle in a cell population cultured under constant 8% O2 for 4 successive days. By contrast, cells that were exposed to physiological oxygen rhythms exhibited rhythmic expression of clock genes (Hepa-1c1c7 and NIH 3T3 cells; Figures 2C and S2A, respectively). Importantly, the phase relation, namely the relative peak time expression of the different clock genes, highly resembled the one obtained with dexamethasone pulse that is widely used to synchronize clocks in culture (Figure S2B) and in mice (Figure S1A). Our results indicated that physiological oxygen rhythms reset the molecular clock in cultured cells.
HIF1α not only responds to changes in oxygen levels, but also plays a critical role in oxygen homeostasis through gene expression regulation (Majmundar et al., 2010). We therefore examined the role of HIF1α in resetting the molecular clock upon oxygen rhythms. To this end, we employed Hif1α small interfering siRNA to specifically knock down Hif1α (Hif2α transcript levels were not affected by knock down of Hif1α). Remarkably, oxygen rhythms failed to elicit cyclic expression of clock genes in Hif1α deficient cells, and the expression levels ofRev-erbα, Rorα, Per1, Per2, Cry1, Cry2, and Dbp were a constant low compared to control cells (Figure 2D). We noticed a prominent effect for Hif1α knockdown on Cry2 and Rorα expression, in particular, as their transcript levels were substantially lower in Hif1α deficient cells already at CT0. Analysis of protein levels of several clock genes corroborated the gene expression data (Figure 2E). Notably, CLOCK protein levels were elevated in Hif1α deficient cells.
Thus, in Hif1α deficient cells, we failed to detect rhythmic expression of clock genes. This raised the possibility that HIF1α is either absolutely required for circadian rhythmicity, namely a core clock component, or specifically essential for clock resetting by oxygen rhythms. To distinguish between these two scenarios, we examined the requirement of HIF1α for resetting the clock by dexamethasone. Knock down of the Clock gene, a principal component of the core clock circuitry, completely abolished circadian rhythmicity (Figure S2B). By contrast, Hif1α knockdown had little effect on the rhythmic expression of clock genes in cells synchronized by dexamethasone (Figure 2F). The expression profiles of Clock, Bmal1, Rev-erbα, Cry1, and Dbp were very similar irrespectively of Hif1α levels, and the rhythmicity of Rorα, Per1, Per2, and Cry2was preserved, yet their expression levels were lower.
Taken together, our results evinced that HIF1α is specifically required for resetting the molecular clock by oxygen rhythms and not by dexamethasone, suggesting that HIF1α is not an integral component of the core clock circuitry, but rather functions upstream to the clock in response to changes in oxygen levels. We therefore concluded that HIF1α is the molecular link between oxygen and the circadian clock.
The Effect of Oxygen Levels on Core Clock Gene Expression
We sought to identify potential downstream effector/s within the core clock circuitry that connect oxygen rhythms and HIF1α with circadian clock resetting. First, we examined the expression levels of clock genes in cultured cells throughout an oxygen cycle (Cue), namely 12 hr of 5% O2 followed by 12 hr of 8% O2 (Figure 3A). The transcript levels of Rev-erbα, Rorα, Per1, Per2, and Cry2were upregulated in response to a decrease in oxygen levels to 5% (Figure 3B).Cry1 levels were only induced once oxygen levels were restored to 8%, whereas the expression levels of Clock and Bmal1 were mostly unaffected. Next, we examined whether HIF1α is required for the induction of clock genes in the course of an oxygen cycle. We found that the induction of Cry2, Rorα, Per2,Cry1, and as expected Glut1, a known target of HIF1α (Chen et al., 2001), was blunted upon knock down of Hif1α (Figure 3C). Remarkably, the transcript levels of both Cry2 and Rorα were downregulated in the absence of HIF1α already under 8% O2 (i.e., ZT0) (Figure 3C) and responded to a decrease in oxygen levels in a dose-dependent and HIF1α-dependent manner (Figure S3A). Moreover, their transcript levels were sensitive to variations in oxygen levels within the physiological range, namely 3% O2 (Figure S3B). In comparison to different known HIF1α target genes, such as Glut1, Pdk1, and Ldha, both Cry2and Rorα appeared to be highly sensitive to small changes in O2 levels, similar to Glut1 (Figures S3A and S3B).
Bioinformatics analysis of the Cry2 gene revealed several hypoxia response element (HRE) and E-box motifs (Figure S3E); these elements are pertinent for HIF1α and BMAL1 binding, respectively. In line with previous genome-wide chromatin immunoprecipitation (ChIP) studies (Koike et al., 2012, Rey et al., 2011), our ChIP experiments showed rhythmic binding of BMAL1 specifically to a region within the Cry2 promoter that contains both E-box and HRE motifs (i.e., block D), with peak binding at ZT8 (Figure S3F). Likewise, we managed to ChIP BMAL1 specifically on this region in chromatin prepared from mouse kidney and cultured cells (Figures S3G–S3I). The presence of HRE prompted us to test the binding of HIF1α to the Cry2 gene, however, we failed to detect any specific binding (Figures S3F and S3G). It might be that HIF1α indeed binds to the Cry2gene, but we could not identify it either due to technical reasons or since it binds to other Cry2 genomic regions. Remarkably, although hypoxia induced the expression of Cry2, it did not affect the binding of BMAL1 to the Cry2 promoter (Figures S3H and S3I).
We reasoned that both Cry2 and Rorα are potential candidates that connect the oxygen-HIF1α axis and the circadian clock, as they readily responded to changes in oxygen levels in a HIF1α-dependent manner. To examine their potential role as the molecular link between the oxygen-HIF1α axis and the circadian clock, we tested whether they can phenocopy the effect of Hif2αknockdown on circadian clock resetting by oxygen rhythms. Knock down of Cry2primarily resulted in phase delay in the expression of clock genes in response to both oxygen rhythms and dexamethasone (Figures 3D and S3C, respectively). Knockdown of Rorα had some effect on the daily expression levels of several clock genes following oxygen rhythms and very little effect upon dexamethasone (Figures 3E and S3D, respectively). Yet, although both responded to changes in oxygen levels in a HIF1α-dependent manner, neither knock down of Cry2 nor ofRorα mimicked the effect of Hif1α knockdown. It is conceivable that resetting the circadian clock by the oxygen-HIF1α axis is mediated through the concerted action of several clock genes and potentially other factors yet to be identified.
Low Oxygen Pulse Accelerates the Adaptation of Mice in a Jet Lag Protocol
To examine the relevance of the above findings in the context of the whole animal, we examined the effect of moderate reduction in oxygen levels on the daily voluntary locomotor activity of mice under different light-dark regimens. Upon 12 hr light-dark cycles, mice exhibit robust rest-activity cycles, with activity onset every 24 hr once the light is turned off. This activity pattern is preserved in mice under constant dark, albeit with a slightly shorter period than 24 hr (Partch et al., 2014). We first examined whether cycles of 12 hr of 21% and 12 hr of 16% O2 in the subjective dark and light phase, respectively, will maintain animals on a precise 24 hr schedule in constant dark. The circadian period, however, was similar between mice that were housed under constant versus rhythmic oxygen levels (Figures S4A and S4B).
One of the key properties of the circadian clock is its phase resetting to a new lighting schedule. We therefore tested the effect of ambient oxygen levels on an experimental jet lag protocol of a 6 hr phase advance in the lighting schedule. We found that 12 hr of 16% O2 prior to the shift considerably accelerates the adaptation of mice to the new lighting schedule (Figures 4A and 4B ). Even a short pulse of 2 hr of 14% O2 following the shift in the lighting schedule was sufficient to shorten the adaptation time (Figures 4C and 4D). Next, we examined whether this effect, similar to the effect in cultured cells, is HIF1α dependent. Since Hif1α null homozygous mice are nonviable (Iyer et al., 1998), we testedHif1α heterozygous mice. Hif1α transcript and protein levels were decreased by ∼50% in Hif1α+/− compared to their wild-type littermates (Figures S4C and S4D).Hif1α deficient mice did not differ from their wild-type littermates in their circadian period (Figures S4E and S4F) or in their adjustment to a 6 hr shift in the lighting schedule (Figures 4E and 4F). However, in contrast to wild-type mice, neither 12 hr of 16% O2 nor 2 hr of 14% O2 accelerated the adaptation of HIF1α deficient mice to the new time, upon jet lag protocol (Figures 4G–4J), suggesting that HIF1α is required for oxygen resetting of circadian clocks in mice as well. In this conjuncture, we analyzed the expression levels of clock genes in whole brain samples in response to 2 hr of 14% O2 for wild-type and Hif1α deficient littermates (Figure S4G). We found a small, but statistical significant induction ofCry2 similar to Ldha, an HIF1α target, that was abrogated in Hif1α deficient mice. Notably, both Per1 and Cry1 transcript levels were downregulated in Hif1αdeficient mice irrespectively of oxygen levels.
We conclude that oxygen functions as a prominent resetting cue for circadian clocks in a HIF1α-dependent manner (Figure 4K) and propose oxygen modulation as a supportive therapy for jet lag. Notably, the DNA binding activity of NPAS2-BMAL1 heterodimers is regulated by carbon monoxide (Dioum et al., 2002). Thus, different gases (i.e., oxygen and carbon monoxide) appear to play a role in circadian clock control. It is plausible that impaired tissue oxygenation under pathological conditions such as cardiovascular diseases might also affect the clock function and consequently contribute to the pathophysiology of the disease. On a different note, in the majority of modern commuters airplanes, the cabin oxygen pressure corresponds to ∼16% O2 (Cottrell, 1988), and the aviation industry is investing substantial funds and efforts to improve and increase the cabin oxygen levels to 21% O2. This should be reconsidered in view of the beneficial effect of reduced oxygen levels in jet lag recovery that are reported here.
G.A. and Y.A. wrote the paper. G.A., Y.A., and B.L. designed experiments and conducted them together with M.G. and M.P.K.
We are grateful to Y. Kuperman for her help with the metabolic cages, R. Aviram and G. Manella for their help with the statistical analysis and graphical illustrations, E. Zelzer for kindly providing us with the Hif1α+/− mice, and all the members of our lab for their advice and valuable comments on the manuscript. Work in the G.A. laboratory was supported by the Israel Science Foundation (ISF 138/12), the European Research Council (ERC-2011 METACYCLES 310320), and Yeda Sela. G.A. is a recipient of the EMBO young investigator award and incumbent of the Pauline Recanati Career Development Chair. B.L. received a post-doctoral fellowship from the Feinberg Graduate School, Weizmann Institute of Science. Work in the M.P.K. laboratory was supported by the British Heart Foundation (No. FS/14/2/30630) and the European Union, Seventh Framework Programme, Marie Curie Actions (CARPEDIEM – No. 612280).