Cell Autonomy and Synchrony of Suprachiasmatic Nucleus Circadian Oscillators

THE CIRCADIAN NETWORK

Behavioral and physiological processes are coordinated by endogenous biological clocks, permitting organisms to synchronize to and anticipate changes in the environment. The circadian system serves this purpose by generating and sustaining 24h rhythms of biological processes and by synchronizing to external stimuli such as the solar day-night cycle. In mammals, circadian clocks exist throughout the body, in individual cells and organs, and these clocks are kept synchronized by a master pacemaker residing in the suprachiasmatic nucleus (SCN) of the hypothalamus [1, 2] (Glossary). The SCN, however, is potentially comprised of ~20,000 cell autonomous clocks of its own [3, 4]. Single neurons within the SCN exhibit independent rhythms of firing rate [4] and gene expression [5]. Advances in electrophysiological and imaging techniques, allowing long-term, single-cell recording of circadian rhythmicity, have improved our understanding of the role of both the cell autonomous clock and the SCN network in maintaining a stable circadian system. In addition, recent work has demonstrated that the connectivity between these individual SCN neurons is not only essential for proper timekeeping, but also imparts properties that distinguish the SCN as a master pacemaker of the circadian clockwork. Here, we focus on the network formed by the independent cellular oscillators of the SCN. We will review the evidence for multiple oscillators within the SCN, highlighting recent advances in our understanding of the complex organization and function of the cellular and network-level SCN clock. We will examine the importance of the SCN oscillator network for the robustness of the circadian system and discuss current views on the mechanism and function of coupling between SCN neurons.

THE SUPRACHIASMATIC NUCLEUS

The SCN is a heterogenous structure (Figure 1; see [6] for a comprehensive review). Subregions can be defined throughout the nucleus based on peptidergic content, anatomical location, and circadian parameters of the component cells. Traditionally, the SCN has been divided into two major subdivisions known as the dorsomedial (dmSCN) “shell” and the ventrolateral (vlSCN) “core” [3, 7]. The former contains cells expressing arginine vasopressin (AVP), while the latter is characterized by expression of vasoactive intestinal polypeptide (VIP), gastrin-releasing peptide (GRP) [3, 8], and neuromedin S (NMS) [9, 10] (Figure 1). There are numerous projections from the core to the shell, but fewer efferents from the dorsal shell innervate the ventral core [11]. Although they have served as a useful framework for exploring intercellular communication within the master pacemaker, the anatomically- and peptidergically-derived classifications of dm/vlSCN (or core/shell) do not begin to capture the complexity of the nucleus (see [12, 13] for further discussion). Variability exists along every axis of the SCN [14], demanding a rigorous examination of the function of individual SCN oscillators and subregions in the control of mammalian circadian rhythms.

Within most (if not all) mammalian cells, a transcriptional-translational feedback loop keeps circadian time (see [15, 16] for review; Figure 2). The basic-helix-loop-helix transcription factors CLOCK and BMAL1 accumulate and heterodimerize in the nucleus. Acting as a positive regulator, the CLOCK:BMAL1 complex activates transcription of Period (Per) and Cryptochrome (Cry) genes, leading to the subsequent accumulation of PER and CRY proteins. As their levels increase in the cytoplasm, PER and CRY dimerize, re-enter the nucleus, and the complex suppresses transcriptional activation by CLOCK and BMAL1. In addition to this core feedback loop, there are accessory loops in the molecular clock. The CLOCK:BMAL1 complex activates Rev-erbα. Rev-erbα is a transcriptional repressor of Bmal1, providing a mechanism by which buildup of BMAL1 protein can suppress further Bmal1 transcription. Eventually, PER and CRY proteins are degraded, at least in part through E3-ligase-mediated proteosomal degradation [17-21]. Casein kinase 1 (CK1) isoforms (α, δ and ε) contribute to protein turnover by phosphorylating PER, targeting it for degradation [22-26]. How the cell autonomous molecular clock drives biological and physiological circadian output rhythms is only poorly understood. However, it is clear that the coupling among oscillators within the SCN is vital to maintaining robust circadian rhythmicity.

THE PHOTIC ENVIRONMENT AND SCN OSCILLATORS

COUPLING WITHIN THE SCN

In contrast to what is observed following jetlag or under conditions of forced desynchrony, the multiple oscillators of the SCN normally function in a unified manner. However, even in the coordinated, intact SCN, cellular oscillations are not completely synchronous. Individual neurons exhibit a range of phases [50]. Heterogeneity of phase can be observed along every anatomical dimension of the SCN [14]. There is a “wave” of oscillatory activity within the SCN [14, 36, 51]. Expression of Per genes and their protein products appears to spread through the nucleus, with the earliest peaking populations of neurons observed in the medial and dorsal SCN [14, 36, 51-53]. This pattern of PER expression within SCN subregions occurs in a consistent, stable manner, displaying both temporal and anatomical organization [14, 53] (Figure 3). The phase of subpopulations of SCN neurons does not necessarily correlate with the circadian period or peptidergic content of the cells [14]. The process by which this oscillatory pattern is organized within the SCN network is unknown. Determining the mechanism that establishes the phase of individual neurons and populations of neurons, and the hierarchy of SCN neuronal connectivity, will be a critical step towards understanding the SCN network. In-depth analysis of the spatiotemporal organization of the SCN [14, 53, 54] is likely to provide insight into regulation of the SCN itself and circadian outputs.

When SCN cells are dispersed in culture, the observed phases and periods are much more variable and the rhythms of individual neurons are significantly less reliable [4, 55-57], although synchrony can be observed when dispersed neurons are cultured at a high density [58]. These findings support the notion that the anatomical organization of the SCN and cell-to-cell communication permits coupling between autonomous cell oscillators to convey coordinated rhythms within the nucleus.

NETWORK PROPERTIES CONFER STABILITY AND ROBUSTNESS TO THE SCN AND THE CIRCADIAN SYSTEM

In addition to maintaining synchrony within SCN neuronal populations, there are two other important functions of SCN coupling which have come to light over the past several years: 1) providing robustness, precision, and stability to SCN rhythms; and, 2) generating and sustaining circadian rhythms.

Coupling of SCN neurons leads to robustness (strength and precision of rhythmicity, as well as resistance to perturbation) of the SCN itself and its output rhythms. As discussed above, the range of periods of individual, widely-dispersed cultured SCN neurons is greater and less stable than that of neurons recorded from whole SCN explants. This suggests that, in the intact nucleus, coupling between neurons serves not only to provide local synchrony, but also to influence the period of oscillations within individual cells and determine the period of the SCN as a whole. Indeed, it has been demonstrated that the period of locomotor behavior of hamsters can be estimated by averaging the period of neuronal firing rate of single cells within the SCN [59]. Mathematical modeling demonstrates that synchronization to the mean circadian period occurs once intercellular coupling strength reaches a certain threshold [59]. Thus, in widely dispersed culture systems, coupling strength is reduced resulting in only partial synchronization and a broad range of periods among individual cells [59]. The contribution of coupled oscillators to robustness of period is also observed in the mouse. There is less variability in the period of mouse (Mus musculus) locomotor activity and SCN explant Per1 expression than in cellular-level firing rate [55]. Even the most stable dispersed individual cells are only as precise as the least stable mouse locomotor rhythm or SCN explant [55]. Coupling of variable cellular oscillators results in the high amplitude, stable rhythms observed at the tissue, system, and organismal levels.

The SCN oscillator network also conveys robustness against genetic mutations. Animals with chimeric SCN, comprised of both wildtype and circadian mutant (either the short period Tau mutation or the long period ClockΔ19 mutation) tissue, exhibit behavioral rhythms with components intermediate to, and sometimes dually expressing, the periods of the component cells [60, 61]. Neurons from the SCN of ClockΔ19 mice (which posses a dominant negative form of CLOCK, resulting in a lengthened circadian period and eventual arrhythmicity when housed in constant lighting conditions) maintain robust rhythmicity of firing rates in the intact nucleus [62]. However, in cell dispersals only a small percentage of SCN neurons from ClockΔ19 mice are able to maintain circadian firing rate rhythms [62]. Deletion of Per1 or Cry1 results in impaired rhythmicity in dissociated SCN neurons and fibroblasts, with many cells exhibiting no measurable rhythm [5]. The circadian rhythms of SCN explants and of behavior do not reflect this cell-autonomous loss of rhythmicity. SCN explants from mice null for either Per1 or Cry1 remain rhythmic, as does the locomotor activity of the animals [5]. This result suggests that coupling between (largely arrhythmic) individual neurons in the SCN slice (and in the mouse) are able to compensate for loss of cell autonomous rhythmicity [5] (Figure 4).

Recently, it was demonstrated that network properties of the SCN impart resistance to temperature resetting. Temperature is a powerful entraining cue for many species [63-65], but in mammals and other homeotherms the entraining effects of temperature are less robust, with temperature cycles serving as weak entraining signals (at best) in rodents [66-68] and primates [69]. Temperature entrainment of circadian gene expression rhythms has been observed in mammalian peripheral cells and tissues, but the SCN itself is less sensitive to temperature cues [70-72]. While mouse pituitary and lung explants entrain to daily temperature cycles, the SCN free-runs through a 24 h (12 h 36°C: 12 h 38.5°C) temperature cycle, seemingly uninfluenced by temperature cues [72]. When communication between SCN neurons is blocked with tetrodotoxin (TTX, which blocks voltage-gated sodium channels) or nimodipine (an L-type calcium channel blocker), however, temperature pulses synchronize and reset rhythms in individual cells resulting in a temperature-sensitive SCN [72] (Figure 5). Similarly, lung, but not SCN explants, entrain to temperature cycles of 20h (10h 35.5°C: 10h 37°C) and 28h (14h 35.5°C: 14h 37°C) [70]. Only large changes in temperature (≥ 6°C), cycling with a more moderate period (22h), are able to entrain PERIOD2 (PER2) rhythms in the SCN [70] (as measured in PER2∷LUC mice, which express PER2 fused with the luciferase [LUC] reporter [73]). These experimental findings have been supported by theoretical models demonstrating that coupling within an oscillator network (eg., the SCN) limits the range of entrainment [70].

Together, these data demonstrate that coupling within the mouse SCN conveys resistance to temperature entrainment and imparts robustness to circadian rhythmicity within the nucleus. Such resistance to temperature resetting may allow the SCN to control peripheral oscillators through body temperature rhythms while itself remaining unaffected by this rhythmic output [72]. This hypothesis needs to be directly tested on the background of a coupled and uncoupled SCN network. The data discussed above conflict with previous findings that the rat SCN can be phase-shifted [74] and entrained [75] by temperature. It is possible that the discrepancy between studies is due to a species difference in the sensitivity of the SCN to temperature signals. Such species differences in SCN resetting could result from underlying differences in SCN coupling strength or circadian amplitude (as seen with the ClockΔ19 mutation [76]). A more careful comparison of the effect of temperature on SCN explants from a variety of mammalian species is warranted. Interestingly, in one report of temperature sensitivity in the rat SCN, explants were taken from neonatal animals [75]. This raises the possibility that developmental changes in intercellular signaling may occur which confer temperature-resistance and other network attributes to the SCN. Investigating the developmental contributions to SCN function may reveal new properties of cellular communication within the oscillator network.

INTERCELLULAR SIGNALING PLAYS A ROLE IN CIRCADIAN RHYTHM GENERATION

Beyond sustaining circadian rhythmicity, there is recent evidence that network properties of the SCN are capable of generating rhythmicity. Mathematical models have demonstrated that intercellular coupling can sustain rhythmicity in individual cellular oscillators, and experimental evidence supports this prediction [5, 77]. Culturing SCN neurons at very low densities results in a substantial decrease (as compared to explants or high density cell dispersals) in the number of neurons exhibiting sustained circadian rhythms of firing rate and PER2 expression [78], suggesting that coupling between SCN neurons is necessary to sustain rhythmicity in individual cells. The experiments in cells from Per1^-/- and Cry1^-/- mice demonstrate that even a few intrinsically rhythmic neurons are capable of producing circadian oscillations throughout the SCN, an effect which is also achieved through intercellular coupling [5, 62]. Surprisingly, rhythmic protein expression can also be observed from an SCN containing no intrinsically rhythmic neurons. Highly variable, rhythmic oscillations in PER2∷LUC have recently been observed from the SCN of Bmal1^-/- mice, which are behaviorally arrhythmic [79]. SCN explants from Bmal1^-/- mice display PER2∷LUC rhythms with periods ranging from 14-30h. While there is stochastic variability in the period length of oscillations from Bmal1^-/- SCN explants, these unstable rhythms persist for over 30 cycles in culture. Dissociated cells from the SCN of these mice possess no rhythmic PER2 expression, although PER2 levels fluctuate in a random manner. The rhythms observed in SCN explants, therefore, appear to be an emergent network property resulting from intercellular coupling on a background of molecular noise. Indeed, SCN-level oscillations are abolished by uncoupling SCN neurons with TTX [79]. Mathematical modeling provides theoretical confirmation that coupling among arrhythmic SCN neurons, which demonstrate fluctuating but not oscillating PER2 levels, can produce tissue-level rhythmicity [79]. This work demonstrates that the SCN network is capable of generating oscillations in the circadian range from a population of individually arrhythmic cells, illustrating a previously unappreciated level of robustness in the SCN system.

MECHANISMS OF INTERCELLULAR COMMUNICATION WITHIN THE SCN

The mechanisms by which individual cells in the SCN couple to produce stable, coherent rhythms are not fully understood. As discussed above, blocking sodium-dependent action potentials with TTX [50, 79] or isolating neurons within the SCN [78] leads to cellular desynchrony. This suggests that neurotransmission is vital to maintenance of circadian rhythmicity, although gap junctions may also play an important role in SCN coupling [80, 81].

UNRAVELING THE SCN NETWORK

Intercellular coupling of cell autonomous oscillators within the SCN renders synchrony and robustness to the master circadian pacemaker. Network level properties convey a stable circadian phase and period at the tissue level to a nucleus comprised of intrinsically variable neuronal rhythms. The ability of this complex network to generate rhythmicity upon a background of arrhythmicity [79] provides yet another piece of evidence for the importance and intricacy of coupling within the SCN.

While our knowledge of the role of networking in SCN function has advanced significantly in recent years, our grasp of the fundamental components of the network remains lacking (Box 1). Combining mathematical simulations of the oscillator network with experimental approaches [5, 70, 79, 110] will likely be necessary to unravel the complexity of SCN intercellular connectivity and the output of this multi-oscillator pacemaker. Moreover, the effect of silencing of individual neurochemical and anatomical components of the SCN network should be examined. In the fly (Drosophila melanogaster), studies using transgenic models and targeted ablation of specific populations of clock neurons have revealed mechanistic details of circadian network connectivity [111, 112]. Similar investigations should be undertaken in mammals, using emerging tools such as peptide-specific neurotoxins and cell-type specific cre-recombinase expressing mice. Deciphering the intercellular circadian code is an important step in advancing the study of the SCN network. Understanding the way in which balance is maintained between cell autonomy and synchrony within the SCN, and the manner in which this is translated into output signals, will provide insight into the means by which this nucleus coordinates a complicated web of circadian processes throughout the organism.