miR-132/212 Modulates Seasonal Adaptation and Dendritic Morphology of the Central Circadian Clock.

The central circadian pacemaker, the suprachiasmatic nucleus (SCN), encodes day length information by mechanisms that are not well understood. Here, we report that genetic ablation of miR-132/212 alters entrainment to different day lengths and non-24 hr day-night cycles, as well as photoperiodic regulation of Period2 expression in the SCN. SCN neurons from miR-132/212-deficient mice have significantly reduced dendritic spine density, along with altered methyl CpG-binding protein (MeCP2) rhythms. In Syrian hamsters, a model seasonal rodent, day length regulates spine density on SCN neurons in a melatonin-independent manner, as well as expression of miR-132, miR-212, and their direct target, MeCP2. Genetic disruption of Mecp2 fully restores the level of dendritic spines of miR-132/212-deficient SCN neurons. Our results reveal that, by regulating the dendritic structure of SCN neurons through a MeCP2-dependent mechanism, miR-132/212 affects the capacity of the SCN to encode seasonal time.

The suprachiasmatic nucleus (SCN) of the hypothalamus houses a central circadian pacemaker in mammals. The ~20,000 neurons in this bilateral structure coordinate internal daily rhythms in behavior and physiology with external cycles, the most predominant one being light availability due to the Earth’s rotational movement[1]. The so-called “molecular clock” is a ubiquitous machinery that sustains near 24-hour (circadian) rhythms in expression of “clock” genes via interlocking transcription and translation feedback loops (TTFLs). In the primary feedback loop, the positive limb, comprised of the transcription factors CLOCK and BMAL1, promotes the transcription of elements in the negative limb, the period and cryptochrome genes[2,3].

Although cells in the SCN can autonomously sustain molecular oscillations, to produce a robust, coherent output to peripheral clocks, they need to maintain synchrony at the tissue level: this intra-SCN synchrony is achieved through paracrine communication[4]. The neuronal population of the suprachiasmatic nucleus is predominantly GABAergic[5] and densely interconnected. Although it is heterogeneous in terms of the neuropeptides that are synthesized, there are two main anatomical and functional clusters: the “core” (ventrolateral region) and the “shell” (dorsomedial region)[6]. Neurons in the core express vasoactive intestinal polypeptide (VIP), and receive direct input from retinal ganglion cells[6]. Upon photic stimulation at critical time windows, core neurons quickly reset the phase of their molecular clock, which is essential for shifting behavioral cycles[7,8]. Neurons in the shell SCN secrete arginine vasopressin (AVP); unlike cells in the core, they take longer to re-adapt the phase of clock gene oscillations to changes in the external light/dark cycle[9].

In addition to maintaining 24-hour rhythms, the SCN can also encode variations in photoperiod or day length (i.e., a long day in the summer vs. a short day in the winter), allowing organisms to prepare for the environmental demands characteristic of each season throughout the year. The SCN relays photic information through a multisynaptic pathway to the pineal gland, which produces and secretes melatonin during the nighttime. This is required for physiological seasonal adaptation[10,11]. In photoperiodic mammals, distinct patterns of melatonin signaling acting in the pituitary gland and various hypothalamic nuclei allow for season-appropriate changes in appearance, reproductive physiology and metabolism[12-14]. Whether other mechanisms independent of melatonin signaling also contribute to seasonal changes in physiology and behaviour remains unclear. Mice of the C57BL/6 background exhibit photoperiod-dependent changes in circadian activity/rest cycles and SCN physiology despite their inability to produce melatonin[15]. This suggests that there may well be other mechanisms at play besides melatonin signaling that influence seasonal adaptation.

As is the case in other species, structural plasticity could also play a role in how the murine SCN network alters its properties to encode photoperiodic information. In Drosophila, seasonal adaptation requires axonal plasticity in brain clock neurons[40]. In seasonal songbirds, the higher vocal center in the brain undergoes remarkable morphological changes to enable song production, which is essential for breeding during the long photoperiod[41].

Seasonal time has been proposed to be a meta-property encoded within the network of circadian oscillators that comprise the SCN1[6]. Overall, under short days there is a higher degree of synchrony among SCN neurons, and under long days cell clusters are out-of-phase with each other[16]. This has been reported between the rostral and caudal SCN[17-20], and between the core and shell subcompartments[18,21,22]. VIP signaling appears to have a role in seasonal adaptation, as Vip−/− mice do not show photoperiod-dependent changes in SCN electrical activity[23]. Some electrophysiological mechanisms have been investigated in the context of seasonal adaptation. A switch in GABAergic transmission from inhibitory to excitatory, due to changes in the equilibrium potential of GABAergic currents, has been suggested to mediate adaptation to long photoperiods[24]. Moreover, Cl− transporter abundance and intracellular Cl− concentration can regulate the polarity and strength of GABAergic transmission. These processes were implicated in maintaining the phase disparity between the core and shell regions of the SCN under long days[25]. Additionally, changes in the properties of K+ currents have been shown in the SCN of long day-housed animals[26]. Beyond these studies, the mechanisms for photoperiodic plasticity in the SCN remain elusive.

MicroRNAs (miRNAs or miRs) are short, non-coding RNAs that recognize elements within the 3′-untranslated regions (UTRs) of target mRNAs through base complementarity with their “seed sequence”, hindering translation and/or promoting transcript degradation. miRNAs have been increasingly recognized as regulators of circadian rhythms[27,28]. In regard to the mammalian central pacemaker, miR-132 and mir-219 have been examined before[29,30]. In our recent study[31], we investigated the role of the microRNA cluster miR-132/212. Although miR-132 and miR-212 are encoded in a single locus and their seed sequences are identical, their patterns of expression and putative target genes do not overlap entirely[32]. Previously, expression of miR-132 was shown to be light-responsive in the SCN, and to downregulate the behavioral phase-shifting response to acute photic stimulation by modulating the expression of genes implicated in chromatin remodeling and translational control[29,30]. However, in our study, a global deletion of the miR-132/212 cluster did not affect the behavioral response to acute photic stimulation under constant darkness, at nine different time points assessed throughout the circadian cycle[31]. The discrepancy between our previous investigations, where only levels of miR- 132 were tonically or transiently manipulated[29,30], and our recent study, where both miR-132 and miR-212 were genetically ablated, might indicate that miR-132 and miR-212 have different or opposing roles in regulating acute phase resetting of the clock. This question could be addressed by either deleting or transiently inhibiting miR- 212 alone without altering miR-132 expression. Since our study used a germline disruption of the miR-132/212 locus, an alternative explanation is that compensatory changes arising throughout development counteract the effects of miR-132/212 deletion on the phase shifting response. Using an inducible miR-132/212 knock-out model would help to clarify if this is the case.

Given that the expression of miR-132 and miR-212 are induced by neuronal activity[33,34], we hypothesized miR-132/212 ablation may affect “activity”-dependent plasticity of the circadian system, in particular in the context of exposure to different environmental light cycles. To address this, we examined the locomotor behavior of miR-132/212-deficient (miR-132/212) mice under long and short photoperiods as well as under non-24-hour cycles (T-cycles). miR-132/212 mice entrained better and more precisely to short days and short T-cycles than wild-type controls. Furthermore, a shortening of the behavioral period following exposure to a short T-cycle (also known as “after-effect”) was more pronounced in miR-132/212 mice compared to wild-type controls. To date, there is not a clear explanation for the persistent effects of T-cycles or photoperiod on the circadian clock, although some molecular events have been proposed. In one study, maternal exposure to T-cycles during pregnancy had long-lasting effects in the progeny, pointing to epigenetic mechanisms imprinting the central clock[35]. In hamsters, reversible methylation of the promoter region of Dio3, a gene encoding for a melatonin-dependent thyroid hormone enzyme, underlies reproductive activation under long days[36]. Two other studies analyzed DNA methylation programs in the SCN of animals adapted to long or short T-cycles[37,38]. Remarkably, changes in the DNA methylome were region-specific, and communication between the core and shell SCN was required to produce those changes[37,38]. The identities of those genes whose expression in the SCN is regulated by the photoperiod or T-cycle remain elusive, but are likely to reveal important insights on the cell-autonomous mechanisms that underlie the networklevel changes involved in this type of circadian plasticity. In our study, expression of the miR-132/212 target gene, MeCP2, was dysregulated in the SCN of miR-132/212 mice in a circadian- and photoperiod-dependent manner. The MeCP2 protein is capable of binding to methylated DNA, and we speculate that its association with methylated gene promoters may be important for regulating the gene expression programs underlying SCN network plasticity.

In another experiment, we found that long-term exposure to constant light had a milder period-lengthening effect on miR-132/212−/− mice than it did on wild-type animals. Disruption of synchrony among SCN neurons has been suggested to underlie the effects of constant light[39], although the mechanisms for this are not clear. In this scenario, SCN lacking miR-132/212 could be more resistant to desynchronization, leading to stronger coupling between clock neurons. This idea is supported by the higher amplitude of clock protein oscillations in miR-132/212 SCN under constant dark conditions compared to wild-type controls. We also examined PER2 expression throughout the rostral-caudal axis of the SCN after adapting mice to either short or long days. Circadian PER2 oscillations after adaptation to a summer-like photoperiod showed a widened peak, which was advanced in the caudal portion of the SCN in wild-type but not in miR132/212−/− mice. Under short days, PER2 rhythms had a narrow peak (compared to a 12h light:12h dark cycle) irrespective of genotype, although the amplitude was higher in miR-132/212 SCN relative to wild-type controls, another indication that intercellular synchrony may be greater in miR-132/212 animals. These results roughly correlate with the behavioral phenotypes of our knockout mice under short and long photoperiods, although a future study could address in more detail the progression of changes in PER2 rhythms during the process of photoperiodic adaptation.

An important consideration for our experiments is the difference in spatiotemporal dynamics between the rostralcaudal and the ventral-dorsal axes. In our experiments, we did not find consistent phase differences under the long photoperiod between the shell and core SCN, as has been reported by other groups[22]. The reason for this discrepancy is unclear, but it may be due to the light:dark (LD) cycle that we used in our study (16:8 LD, in hours), in contrast with the more extreme cycles under which ventral-dorsal phase differences were previously observed (i.e., 18:6, 20:4 and 22:2 LD)[22]. Although phase differences across both axes have been described in the context of photoperiodic adaptation, in recent years more emphasis has been given to the shell-core subdivision because of the functional implications of the peptidergic profiles of cells within each cluster. However, it is worth pointing out that the ventral-dorsal subdivision is most prominent in the central SCN, which contains both VIP and AVP neurons, whereas in the most rostral and caudal extremes the cells are predominantly AVPergic shell neurons. For most ex-vivo studies of SCN network properties, thin slices containing central SCN are generally used, unless otherwise specified. In our rhythmic profiles, we did not co-label PER2 with AVP or VIP; this may be important for further conclusions about the role of miR-132/212 in the spatiotemporal dynamics of clock protein expression within the SCN.

The miR-132/212 cluster has been implicated in regulation of neuronal morphology in the hippocampus and cortex[43-45]. In our study, we characterized dendritic spine density of SCN neurons from wild-type and miR- 132/212−/− mice maintained under different photoperiods. Relative to wild-type controls, we found a downregulation of spine abundance in miR-132/212−/− SCN at all time points and under all photoperiods examined. These data seem counterintuitive with our previous observation of enhanced intercellular synchrony in the miR-132/212 SCN. However, the network dynamics that maintain the organization and phase distribution of individual oscillators are just beginning to be unveiled[46]. Hence, the degree of structural connectivity might not necessarily translate to enhanced or diminished synchrony. In the future, this question might be examined in our model by using ex-vivo approaches with single-cell resolution. Interestingly, regardless of genotype, daylength altered the prevalence of different protrusion types. Under long days, we noted an increase in the number of spines and a decrease in varicose protrusions. When we analyzed SCN neuronal morphology in a seasonal rodent, the Syrian hamster, we found a similar effect of photoperiod on SCN spine density, namely, a reduction under short days when compared to long days. Importantly, this effect was independent of melatonin signaling, since it was still present in pinealectomized hamsters. We were able to correlate this morphological change with a strong suppression of miR-132 expression in short-day adapted hamsters compared to those housed under long days. These data suggest that the SCN can undergo structural changes that make its network flexible and adaptable to different photoperiods, and that miR-132/212 plays a role priming the SCN for seasonal changes in day length. Mice lacking miR-132/212 adapt more readily or more efficiently to short days, have difficulty entraining to long cycles, and resist the period-lengthening effects of constant light. Altered SCN connectivity may underlie all these phenotypes. It is worth noting that in our study we focused on two time points (middle of the day and middle of the night), hence we are unable to draw conclusions about the potential rhythmic changes in spine density in the SCN. Future investigations could examine this aspect, as well as other morphological parameters such as dendritic complexity and neurite length. The physiological implications of the structural plasticity that we observed in our study are also fertile ground for future research.

In terms of the molecular players that could potentially mediate the phenotypes of miR-132/212 mice, we focused primarily on MeCP2, a target gene for both microRNAs[30,47-49]. In our investigation, ablating MeCP2 expression in vivo and in vitro rescued the morphological phenotype of miR-132-212−/− SCN cells. The role of MeCP2 in dendritic structure is, by all accounts, complex. Analysis of neuronal morphology of MeCP2 mutant mouse lines have yielded contradictory results[50-53]. Some studies have found increased spinogenesis in the mutant mice whereas others have found the opposite. Effects seem to depend on gene dosage, developmental stage, and even brain region. Beyond the need for the spatiotemporal expression of MeCP2 to be tightly regulated, there is much to be learned about this gene in regard to neuronal morphology. A puzzling finding that emerged from our study is the SCN neuronal phenotype of mecp2+/− female mice. We found a considerable upregulation of spine density in these animals, regardless of their miR-132/212 status (−/− or +/+). Since Mecp2 is located on the X chromosome, mecp2 females exhibit a mosaic pattern of MeCP2 expression at the cellular level. Unfortunately, our technical approach to studying neuronal morphology did not allow us to distinguish MeCP2-expressing cells from those with null expression. Being able to discriminate between these two cell populations in MeCP2 heterozygous females would enable us to determine whether this dendritic phenotype was cell-autonomous or a consequence of altered SCN network connectivity in MeCP2 mutant animals.

In conclusion, our study found a novel role for the miR-132/212 cluster in seasonality of the SCN, and a new dimension of structural plasticity in the central circadian clock allowing for adaptation to environmental challenges.