Anti-Aging

Circadian Clock Control by Polyamine Levels through a Mechanism that Declines with Age

Polyamines are essential polycations present in all living cells. Polyamine levels are maintained from the diet and de novo synthesis, and their decline with age is associated with various pathologies. Here we show that polyamine levels oscillate in a daily manner. Both clock- and feeding-dependent mechanisms regulate the daily accumulation of key enzymes in polyamine biosynthesis through rhythmic binding of BMAL1:CLOCK to conserved DNA elements. In turn, polyamines control the circadian period in cultured cells and animals by regulating the interaction between the core clock repressors PER2 and CRY1. Importantly, we found that the decline in polyamine levels with age in mice is associated with a longer circadian period that can be reversed upon polyamine supplementation in the diet. Our findings suggest a crosstalk between circadian clocks and polyamine biosynthesis and open new possibilities for nutritional interventions against the decay in clock’s function with age.

INTRODUCTION

Polyamines (i.e., putrescine, spermidine, and spermine) are essential ubiquitous polycations present in all living organisms. They participate in the regulation of various key cellular processes such as chromatin structure, gene transcription and translation, cell growth, and proliferation. Polyamines are believed to exert their effects through modulating protein-protein and protein-DNA interactions. Cellular polyamine levels are maintained within a narrow physiological range and are tightly regulated through uptake/secretion, de novo synthesis/catabolism, and inter-conversion. Altered polyamine metabolism is associated with various pathologies such as neurological abnormalities, malignancies, and aging (Moinard et al., 2005; Pegg, 2009). Circadian clocks orchestrate the daily oscillations in physiology and behavior. The mammalian circadian timing system consists of a central pacemaker in the brain that is entrained by light-dark cycles and synchronizes subsidiary oscillators in virtually all cells of the body in part by driving cyclic feeding behavior. The core clock circuitry relies on interlocked transcription-translation feedback loops. The transcription factors BMAL1 and CLOCK bind as heterodimers to E-box motifs present in the Period (Per) and Cryptochrome (Cry) genes and drive their transcription. Subsequently PER:CRY protein complexes accumulate and auto-repress BMAL1:CLOCK-mediated transcription. BMAL1 also activates Rev-erb transcription, which in turn suppresses Bmal1 expression. The coordinated activity of these feedback loops drives cyclic gene expression with an 24 hr period, both in cultured cells and living animals (Feng and Lazar, 2012; Mohawk et al., 2012). 

Circadian Clock Control by Polyamine Levels through a Mechanism that Declines with Age

Compelling evidence points toward an interplay between circadian rhythms and cellular metabolism (Asher and Sassone-Corsi, 2015; Asher and Schibler, 2011). Circadian clocks play a principal role in orchestrating the daily expression of regulators and enzymes involved in nutrient processing and energy homeostasis. Concomitantly, clocks are tightly coupled to cellular metabolism and respond to feeding cycles. However, the molecular mechanisms through which metabolism affects the clock’s function are largely unknown. Since polyamine levels are tightly regulated and polyamines facilitate protein-protein/ DNA interactions, which are fundamental for the clock’s function, we set out to examine: (1) whether circadian clocks regulate cellular polyamine levels, and (2) whether polyamines play a role in the function of the core clock, a conjecture that was so far never tested. In this study, we uncover a crosstalk between circadian clocks and polyamines. We show that both clock- and feeding-dependent mechanisms regulate the daily oscillations of key enzymes in polyamine biosynthesis and consequently polyamine levels. In turn, we found that polyamines participate in circadian period control in cultured cells and animals. At the molecular level, polyamines modulate the interaction between the repressor members of the core clock circuitry, PER2 and CRY1. Finally, we demonstrate that the decline in polyamine levels with age in mice is associated with a longer circadian period that can be reversed upon dietary polyamine supplementation.

RESULTS

Key Enzymes in Polyamine Metabolism and Polyamine Levels Oscillate in a Daily Manner

Mammals obtain polyamines by de novo synthesis and through uptake from the diet. The de novo polyamine biosynthesis pathway consists of several successive steps (Figure 1A) (Pegg, 2009). The first and rate-limiting step is the decarboxylation of ornithine to putrescine by ornithine decarboxylase (ODC). Subsequently, putrescine is converted to spermidine, and the latter is metabolized to spermine. These enzymatic reactions are carried out by the aminopropyltransferases, spermidine synthase (SRM) and spermine synthase (SMS), respectively. The aminopropyl group derives from the decarboxylation of s-adenosylmethionine by adenosylmethionine decarboxylase 1 (AMD1). In addition, cellular polyamine homeostasis is maintained through polyamine catabolism and inter-conversion (Casero and Pegg, 2009). These steps are mediated through the activity of several enzymes including spermidine/spermine-N1-acetyltransferase (SAT1), spermine oxidase (SMOX), and polyamine oxidase (PAOX). Since ODC is the rate-limiting enzyme in polyamine biosynthesis, its levels are tightly regulated. ODC is a short-lived protein that is targeted for degradation by antizyme (Az) through the 26S proteasome in a ubiquitin-independent manner (Kahana et al., 2005) or degraded by the 20S proteasome in an NQO1-regulated manner (Asher et al., 2005). Antizyme inhibitor (AzI) binds Az, preventing it from directing ODC for degradation (Kahana et al., 2005). First, we asked whether polyamine metabolism exhibit diurnal rhythmicity. Mice were sacrificed at 4 hr intervals throughout the day and liver transcript levels of polyamine metabolic enzymes and core clock genes were quantified (Figures 1B, S1A, and S1B).

 The majority of enzymes participating in polyamine anabolism were expressed in a daily manner. Odc, Srm, and Amd1 all reached their peak levels 16 zeitgeber time (ZT16) (Figure 1B). Notably, their daily expression profile mostly resembled that of Per2 (Figure S1B). The mRNA levels of AzI, which supports polyamine synthesis by stabilizing ODC, also exhibited shallow daily oscillations with peak levels ZT12 and trough levels ZT0 (Figure S1A). By contrast, the transcript levels of all tested enzymes participating in polyamine catabolism (i.e., Sat1, Smox, Paox), including Az, which targets ODC for degradation, were relatively constant throughout the day (Figure S1A). Our analysis evinced that enzymes participating in polyamine anabolism are expressed in a diurnal manner, whereas enzymes involved in polyamine catabolism are mostly constant throughout the day. To substantiate these findings, we examined the protein levels of ODC, the rate-limiting enzyme in polyamine synthesis, in livers from mice sacrificed throughout the day. In whole liver extracts, ODC was undetectable (data not shown), conceivably, due to its low abundance. Hence, we immunoprecipitated ODC from liver extracts and detected it by immunoblotting. ODC accumulated in a daily manner similar to its transcript levels (zenith and nadir levels ZT16 and ZT4, respectively) (Figure 1C). Next, we quantified the levels of the different polyamine species in mouse liver (Figure 1D). Putrescine, the product of ODC enzymatic activity cycled and reached its peak levels ZT16, in line with Odc mRNA and protein accumulation. Spermidine levels oscillated with a similar phase as putrescine but with shallow amplitude, whereas spermine levels were relatively constant throughout the day.

 Notably, analysis of serum putrescine and spermidine levels showed that putrescine levels are relatively constant throughout the day, while spermidine levels cycle with peak levels ZT16 (Figure S1C). Taken together, our data evinced that key enzymes in polyamine biosynthesis (i.e., Odc, Srm, and Amd1) and the polyamines putrescine and spermidine cycle throughout the day with peak levels during the night.

Clock- and Feeding-Dependent Mechanisms Regulate the Daily Expression of Key Enzymes in Polyamine Biosynthesis and Polyamine Accumulation 

Next, we examined whether the daily oscillations of key enzymes in polyamine biosynthesis and polyamine accumulation are dependent on a functional clock. We compared the daily expression profiles of polyamine metabolic enzymes in wild-type and Per1/2 null mice (Figures 2A and S2A). Per1/2 null mice exhibit arrhythmic behavior in constant darkness and their circadian clock genes’ expression is largely diminished (Figure S2B) (Zheng et al., 2001). Notably, in Per1/2 null mice, Odc, Srm, and Amd1 displayed relatively shallow daily oscillations compared to wild-type mice (Figure 2A). Furthermore, putrescine, spermidine, and spermine levels were fairly constant throughout the day (Figure 2B). These results suggested that PER1/2 and thus circadian clocks play a role in their rhythmic accumulation. Clock mutant mice (e.g., Per1/2 null) exhibit shallow feeding rhythms and consume relatively equal amount of food throughout the day (Adamovich et al., 2014).

 Hence, to examine the responsiveness of Odc, Srm, and Amd1 expression to feeding, we analyzed the expression profiles of these enzymes and clock genes in wild-type mice fed either exclusively during the night, or solely during the day for 3 weeks. As expected, the expression pattern of clock genes in 3 weeks day-fed animals was completely inverted (Figures 2C and S2C) (Damiola et al., 2000). Per2 mRNA reached its maximal levels ZT4 in day-fed animals compared to ZT16 in mice fed either exclusively during the night or ad libitum (Figure 2). The expression profiles of Odc, Srm, and Amd1 resembled that of Per2 (Figure 2C). Thus, similarly to clock genes, principal enzymes in polyamine biosynthesis responded to changes in feeding time. However, since under these conditions both the molecular clock and feeding are in accordance and inverted, it is not possible to conclude whether rhythmic expression is driven by the circadian clock or rather responds directly to feeding. In an attempt to discriminate between these two possibilities, we examined the transcript profiles of Odc, Srm, Amd1, and clock genes during the first day of daytime feeding. During this time window Bmal1, Rev erba, and Dbp exhibited shallower oscillations that were slightly shifted compared to night-fed animals (Figure S2C). By contrast, Per2 mRNA exhibited two prominent peaks, one ZT16, similar to night-fed animals and an additional peak emerged ZT8 (Figure 2C). Conceivably the former is clock dependent, while the latter is driven by feeding. This is in line with previous reports classifying Per2 as a core clock component that is also systemically regulated, most likely by feeding (Kornmann et al., 2007). 

Remarkably, Odc, Srm, and Amd1 displayed a similar expression pattern as Per2 under all three tested conditions (Figure 2C). We concluded that both clock- and feeding-dependent mechanisms drive the daily expression of key enzymes in polyamine biosynthesis and polyamine accumulation.

BMAL1:CLOCK Bind to E-Box Motifs within the Odc Gene in a Circadian Manner

The resemblance in the expression profiles of Per2 and Odc in mouse liver under different feeding regimens (Figure 2) suggested that they might share similar mechanisms of gene expression regulation. BMAL1:CLOCK heterodimers bind to E-box motifs (CACGTG) present in the Per gene and drive its transcription (Mohawk et al., 2012). We identified three canonical E-box motifs in the Odc gene, two within its first intron (E-Box In1) and one in the 30 UTR (E-Box 30 UTR), (Figure 3A). The two E-box motifs within the first intron are conserved in mammals (Figure 3B) and were previously reported to participate in Odc expression by c-Myc (Bello-Fernandez et al., 1993). To determine whether BMAL1 binds these E-boxes, we performed chromatin immunoprecipitation (ChIP) experiments with BMAL1-specific antibodies throughout the day.  In agreement with previous studies, BMAL1 exhibited circadian binding to intron 2 but not to exon 4 of the Dbp gene, with maximal occupancy ZT8 and minimal ZT20 (Figure S3A), (Ripperger and Schibler, 2006). Importantly, our ChIP experiments revealed that BMAL1 binds the E-box elements in intron 1 but not in the 30 UTR region of the Odc gene. The binding of BMAL1 to intron 1 of Odc oscillated in a daily manner with peak and trough levels ZT8 and ZT20, respectively (Figure 3C). Similar results were obtained with CLOCK-specific antibodies (Figures 3D and S3B).

 These results suggested that BMAL1:CLOCK heterodimers bind to E-box motifs within the first intron of the Odc gene in a daily manner and hence are likely to drive the rhythmic transcription of Odc. 

Disruption of Polyamine Homeostasis Affects Circadian Rhythmicity

The daily regulation of polyamine biosynthesis prompted us to examine whether polyamines might in turn play a role in circadian rhythmicity. To maintain their polyamine levels within a narrow range, cells tightly regulate the influx of polyamines from the extracellular milieu (Casero and Marton, 2007). Indeed, polyamine supplementation (e.g., spermidine) to the growth medium of NIH 3T3 stably expressing a luciferase reporter gene under the control of the Per2 promoter (Per2-luciferase reporter) neither affected their intracellular polyamine levels nor significantly altered their circadian oscillations (Figures S4A–S4C). Hence, to attain elevated levels of intracellular polyamines and examine their effect on circadian rhythmicity, we employed NIH 3T3 cells stably overexpressing ODC (NIH 3T3-ODC) and NIH 3T3 cells overexpressing AzI (NIH 3T3-AzI), which inhibits ODC degradation. Cells were transduced with a Lenti-virus carrying the Per2-luciferase reporter. Both NIH 3T3-ODC and NIH 3T3-AzI exhibited poor circadian oscillations compared to NIH 3T3 mock transfected cells (Figure 4A). SDS-PAGE and immunoblot analysis confirmed the elevated levels of ODC and AzI protein in NIH 3T3-ODC cells and NIH 3T3-AzI cells, respectively (Figure 4B), and polyamine levels were elevated in these cells (Figure 4C).

 These results infer that polyamine homeostasis is critical for the clock’s function. Next, we performed the reciprocal experiment in which we examined whether polyamine depletion affects circadian oscillations in cultured cells. To substantially deplete cellular polyamines, we pretreated cells with a-difluoromethylornithine (DFMO), an irreversible inhibitor of ODC enzymatic activity (Pegg and Casero, 2011). An equal number of DFMO-treated and non-treated cells were seeded and monitored for their circadian rhythmicity and polyamine content. Bioluminescence recordings of the Per2-luciferase reporter revealed that DFMOtreated cells exhibit an 2 hr longer period (Figures 4D and 4E). As expected, polyamine levels were significantly reduced in DFMO-treated cells (Figure 4F). Similar results were obtained with NIH 3T3 cells stably expressing a luciferase reporter gene under the control of the Bmal1 promoter (Bmal1-luciferase reporter), (Figures S4D–S4F). To confirm that the effect of DFMO on the circadian period is due to polyamine-depletion, we supplemented DFMO-treated NIH 3T3 Per2-luciferase cells with ornithine, the precursor for polyamine biosynthesis, or with different polyamines. As predicted, addition of ornithine did not restore the circadian period, as ODC that converts ornithine to putrescine is inhibited by DFMO (Figures 4D and 4E). By contrast, addition of putrescine or spermidine, which bypasses ODC decarboxylase activity, was sufficient to restore the circadian period as in non-treated cells (Figures 4D and 4E). Concurrently, measurements of cellular polyamine content demonstrated the efficient uptake of polyamines by the depleted cells and the inter-conversion of the different polyamine species (Figure 4F). 

Similar results were obtained with primary tail fibroblast prepared from mice in which full-length PER2 protein was fused to LUCIFERASE (Figures S4G–S4I). To further examine the effect of polyamine depletion on circadian oscillations, we employed genetic approaches to downregulate ODC expression in cells, either by overexpression of Az, which targets ODC for degradation, or by knockdown of ODC. NIH 3T3 cells were transfected with the Per2-luciferase reporter together with an Az expression vector or an empty vector. Az overexpression was validated by SDS-PAGE and immunoblot (Figure 4I). The decrease in spermidine levels upon Az overexpression (Figure 4J) was accompanied by lengthening of the circadian period (Figures 4G and 4H). Similarly, knockdown of ODC, using Odc-specific siRNA in NIH 3T3 Bmal1-luciferase cells (Figure 4M), decreased cellular polyamine levels (Figure 4N) and resulted in a longer circadian period (Figures 4K and 4L). We conducted a set of control experiments to rule out the possibility that polyamine depletion affect the circadian period due to broad non-specific effects. First, DFMO treatment did not affect cell viability (Figure S5A) or the recordings of a luciferase reporter gene under the control of a CMV promoter (CMV-luciferase), (Figures S5B and S5C) excluding toxic or pleotropic effects. Second, in contrast to spermidine, neither K+ , Zn2+, nor Mg2+ restored the circadian period of DFMO-treated NIH 3T3 Per2-luciferase cells, excluding the possibility that polyamines affect circadian oscillations due to their ionic charge (Figures S5D and S5E).

 Third, we examined whether other intermediate metabolites that are related to ornithine metabolism such as arginine, glutamine, or even GABA might be implicated in the effect of polyamine depletion on the circadian period. Unlike spermidine, none of these metabolites restored the circadian period length of DFMO-treated NIH 3T3 Per2-luciferase cells (Figures S5F and S5G). Taken together, based on pharmacological and genetic manipulations of polyamine levels in cultured cells, we concluded that low polyamine levels lengthen the circadian period, while highly elevated cellular polyamine levels impair circadian rhythmicity. 

Polyamine Depletion Affects Endogenous Circadian Gene Expression and Protein Accumulation

To corroborate the above-described findings, we monitored mRNA and protein expression levels of endogenous clock genes in untreated and polyamine depleted cells at 4 hr intervals for 2 consecutive days. In agreement with the data obtained using the different circadian reporters, the mRNA accumulation profile of several clock genes (i.e., Per2, Bmal1, Cry1, Rev erba) and output gene (i.e., Dbp) exhibited a longer period with delayed second and third peaks upon DFMO treatment (Figure 5A). We also observed more than 3-fold increase in the amplitude of the oscillations of Rev erba and Dbp upon polyamine depletion, whereas the amplitude of Per2, Bmal1, and Cry1 oscillations was unaffected. In accordance with Per2 mRNA expression, the peak in PER2 protein accumulation was also delayed upon polyamine depletion (Figure 5B). Hence, polyamine depletion lengthened the circadian period of cycling endogenous core clock and output genes.

Polyamines Modulate the Interaction between PER2 and CRY1

We wished to examine the molecular mechanisms through which polyamines exert their effect on circadian rhythmicity. In view of their role in regulation of protein-protein interactions, we set out to test the effect of polyamines on the interactions between clock proteins in living cells. We centered our analysis on two principal interactions within the core clock circuitry, namely BMAL1:CRY1 and PER2:CRY1, which participate in circadian period control. The former addresses the interaction between a repressor and an activator and the latter the binding within the repressor complex. To this aim, we employed a mammalian two-hybrid assay, in which BMAL1 or PER2 are fused to a DNA binding domain (BMAL1-BD and PER2-BD, respectively) and CRY1 to a transactivation domain (CRY1-AD). Binding of CRY1 to either BMAL1 or PER2 drives the expression of a Gal4-luciferase reporter and is detected by an increase in bioluminescence emitted from the cells (Langmesser et al., 2008). When polyamine-depleted NIH 3T3 cells were supplemented with putrescine, spermidine, or spermine, the interaction between PER2:CRY1 was increased (Figure 6A). The effect was dose dependent within a physiological concentration range (Casero and Marton, 2007) and specific to PER2:CRY1, as little effect on BMAL1:CRY1 interaction was observed (Figure 6A). Already concentrations as low as 1 mM of spermidine induced a considerable increase in PER2:CRY1 binding. 

As expected, addition of ornithine to DFMO-treated cells could not circumvent the inhibition of ODC-activity and as such did not enhance PER2:CRY1 interaction (Figure 6A). Taken together, these experiments suggested that polyamines specifically promote the interaction between PER2 and CRY1 in living cells with spermidine being the most potent. To further delineate the effect of polyamines on PER2:CRY1 interaction, we performed co-immunprecipitation experiments from cultured cells. PER2 and CRY1 were expressed in untreated or DFMO-treated 293HEK cells. In line with the above described experiments, PER2:CRY1 binding was reduced in polyamine-depleted cells (Figure 6B). Similar results were obtained with endogenous PER2 and CRY1 in untreated and polyamine-depleted NIH 3T3 cells throughout the circadian cycle (Figure S6A). To examine whether polyamine can regulate the binding in vitro, we expressed PER2 and CRY1 in DFMO-treated cells; cell extracts were prepared and mixed in the absence or presence of putrescine, spermidine, or spermine. Co-immunoprecipitatation experiments showed that spermidine promotes PER2:CRY1 binding both in NIH 3T3 and 293HEK cells (Figures 6C and S6B, respectively). Similar experiments were performed with liver nuclear extracts prepared from mice fed diet low in polyamines together with DFMO. Here again, PER2:CRY1 interaction was augmented in the presence of spermidine (Figure 6D). To specifically dissect the effect of polyamines on PER2, we performed a partial trypsin digestion assay with in vitro translated [35S]-labeled proteins (Figure S6C).

Share by: