The SCN acts as an ensemble of individual oscillators

This post might be a bit over the top. For any and all background reading on the subject, make sure to visit http://circadiana.blogspot.com/, who has lots of useful information on the subject of the SCN, circadian rhythms and sleep. Without Bora's diligent and detailed work, I would not be able post this without giving more background on the subject. This is a portion of my thesis introduction, missing the references for which I hope you will excuse me.

Even unicellular organisms express circadian rhythms similar to mammals, demonstrating that these rhythms are generated at the cellular level (Kondo et al., 1993; Lin et al., 1999; Mittag and Wagner, 2003). In individual SCN cells, the expression of so-called clock proteins cycle in double autoregulatory feedback loops to regulate their own transcription in a circadian manner (recent reviews include: Reppert and Weaver, 2002; Panda et al., 2002b; Okamura, 2003). A circadian rhythm of action potential firing frequency can also be measured in many individual SCN cells when dispersed in cell culture (Welsh et al., 1995).
How then do these individual oscillators communicate with each other to construct the resulting circadian rhythm that controls organismal rhythms? In the mammalian SCN, the sum average of these individual oscillations can be measured in numerous ways. For instance, one can follow the peak and nadir of ensemble electrical excitability as expressed by the action potential firing frequency both in vivo and in the brain slice preparation in vitro (Inouye and Kawamura, 1979;Green and Gillette, 1982;Walsh et al., 1992). These studies demonstrate that the circadian rhythmicity of the individual oscillators as well as of the ensemble SCN is maintained in a brain slice preparation. A few models that help address the intercellular coupling that brings about this coordinated circadian rhythm within the SCN have been proposed. One study, using animals and SCN tissue cultures from animals expressing mutation-induced variations in circadian period length (Liu et al., 1997) found that though individual clock cells from dispersed SCN culture vary greatly in their period length, the average periods correlate well with the whole-animal period. Given that dispersed cells vary more than cells in organotypic culture, it was deduced that intra-SCN cell coupling is responsible for keeping SCN cells synchronized, even though each autonomous cellular period is determined by its own molecular clock. A model was proposed in which the circadian phases of core oscillators in the SCN are weakly coupled to one another and this core then recruits outlying cells to oscillate under the same period. Eventually, the majority of cells in the SCN oscillate with a period close to the average period expressed by the whole organism (Reppert and Weaver, 2002). More recently, a study examining acute primary brain slice cultures described the SCN as a network of at least three separate groups of oscillators, the phases of which distribute around the average phase of the entire network (Quintero et al., 2003). This phase heterogeneity, it was hypothesized, could arise from intercellular coupling. The relationship between these oscillators can be modulated by environmental input, a conclusion that is consistent with other models in which rhythmic and non-rhythmic regions of the SCN are delineated by the presence or absence of cells expressing calbindin and cells receiving retinal innervation (Antle et al., 2003). This last model depends upon rhythmic output neurons and nonrhythmic “gate” populations. The “gate” neurons help reset the circadian phase of the entire network, either following environmental input or by the output signals themselves, once a threshold of output activity has been achieved.
All three of these models incorporate the concept of intra-SCN neuronal coupling. In the first model, cellular coupling is of paramount importance to maintain the synchronized rhythmic oscillations of the SCN (Reppert and Weaver, 2002). The second model allows for cellular coupling and a range of sensitivity to the coupling signal to achieve the same end (Quintero, Kuhlman, and McMahon, 2003). And though the third model does not depend on coupling, it allows that cellular coupling could decrease the variability of period length among oscillators (Antle, et al., 2003).
What sorts of intercellular coupling mechanisms are available to SCN cells? Coupling leading to the expression of some circadian rhythms, such as hamster wheel-running activity, may be achieved by synapse-independent mechanisms. More specifically, SCN cells transplanted into SCN-lesioned arrhythmic animals restore the circadian rhythmicity of wheel-running activity even when the SCN cells were transplanted into the third ventricle (LeSauter et al., 1997) and when the SCN cells were dispersed and subsequently encased in an sylastic tube to inhibit synapse formation (Silver et al., 1996). Another indication of synapse-independent coupling is that blockade of intra-SCN synaptic neurotransmission for a period of days does not interfere with phase coherence, as shown by the faithful re-establishment of the appropriate phase following blockade relief. Several alternative coupling pathways have therefore been proposed, such as gap-junctions (Ueda and Ibata, 1989; Jiang et al., 1997b; Colwell, 2000; Jobst et al., 2004) and diffusible factors. Transforming growth factor-alpha (Kramer et al., 2001) and prokineticin 2 (Cheng et al., 2002) are two putative diffusible factors.
The discovery that diffusible factors influence circadian locomotor behavior, however, does not indicate that diffusible factors are the primary coupling mechanism of circadian rhythms, either between SCN neurons or for the coordination of rhythms to the various output target areas. To illustrate how diffusible factors and synaptic communication both play a role in coordinating and shaping circadian rhythms, consider what is known about a particular output target area of the SCN. Numerous projections from the SCN to target areas involved in neuroendocrine and autonomic control are believed to underlie organismal control of circadian rhythmicity (Reppert and Weaver, 2002). The classic example of this is the multi-synaptic pathway regulating melatonin release. GABA- and glutamatergic efferents from the SCN project to cells in the paraventricular nucleus (PVN), which, in turn, projects to the intermediolateral nucleus of the upper thoracic spinal cord. These, in turn, project via the superior cervical ganglia to the pineal gland, which is the site of melatonin production and release (Larsen et al., 1998). Another functional influence by the projection to the PVN is the circadian rhythmicity of adrenal steroids corticosterone and cortisol (Kalsbeek et al., 1996). An ablation-transplantation study showed that, in hamsters with restored circadian locomotor activity following transplantation, neither melatonin nor the adrenal steroids cycled in a circadian manner, suggesting that synaptic output from the SCN is required for these processes, even if locomotor activity can be restored without synaptic output (Meyer-Bernstein and Morin, 1996). These findings were strengthened by data showing that blockade of SCN – PVN synaptic transmission can induce daytime melatonin secretion (Kalsbeek et al., 2000). However, a recent study shows that arginine-vasopressin (AVP) can act as a diffusible factor from the SCN to generate a circadian rhythm in action potential firing frequency in the PVN recorded from hypothalamic slice cultures in which the SCN is excised (Tousson and Meissl, 2004). Yet the study also shows that AVP does not have the same effect in intact brain slices, indicating that under these conditions the neuronal (synaptic) signal overrides the humoral signal from the SCN.
Similarly in the SCN itself, diffusible humoral signals, or regularly released peptides, may serve to consolidate the broad circadian rhythms of SCN cells even in the absence of synaptic neurotransmission—as described above where synaptic transmission was blocked while the circadian clock continued to function. But, the network itself is a circuit of inhibitory synaptic connections that presumably functions as a more immediate cell-to-cell communication mechanism. This may influence other processes—for example action potential firing patterns—critical for the coordination between SCN neurons and the projection of a cohesive output pattern to target areas.